This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
n i (r s , r d )/ O x (r s , r,)] exp . Here, x (r s , r,.) represents the excitation fluence detected at a fixed reference position, r,, in response to excitation from the source position, r s . Even though the excitation fluence, ^Trs, r,), is updated during the iteration, the Jacobian matrix can be directly calculated from Eq. (36), and
Near-Infrared Imaging
509
the change in <J> x (r s , rr) is small compared to that in <J> m (r s , r r ). As the source term of the emission diffusion equation is modified after each iteration [Eq. (22)], changes in the emission fluence are greater than that of the excitation fluence. The same consequence can be inferred from the integral equation [Eq. (36)]. Moreover, the phase of the emission fluence is greater than that of the excitation fluence and the normalized fluence, mAI>x, maintains a high phase contrast. Owing to noise and the ill condition of the Jacobian matrix and for inverting the systems of equations, updating can be accomplished using Newton's method [107] with Marquardt-Levenburg parameters A:
(37) Using excitation referencing at a single reference point, Lee and SevickMuraca [96] reconstructed an 8 X 4 X 8 cm3 phantom containing a 1 X 1 X 1 cm3 target with 100-fold greater ICG concentration, by using 8 excitation sources, twenty four detection fibers for collecting excitation light, and two reference detection fibers (one on either side of the reflectance and transillumination measurements) for collecting excitation light. Figure 28a is the original map containing two-dimensional slices of the three-dimentional geometry that demark the heterogeneity placement, while Fig. 28b is the three-dimensional reconstructed image. The results in Fig. 28 represent reconstructions based on emission FDPM measurements relative to excitation FDPM measurements at a fixed reference position: Ntziachristos and Weissleder [95] successfully reconstructed two fluorescent targets in a 2.5-cm diameter, 2.5-cm-long cylindrical vessel containing ICG and Cy5.5, and using CW emission measurements referenced to excitation measurements at each of the 36 detector fibers as a result of point excitation at 24 source fibers. The high density of measurements for reconstruction of the small simulated tissue volume is troublesome for validity of the diffusion equation used in the forward solver, but is similar to that demonstrated by Yang and coworkers [49] who reconstructed ICG and DTTCI in similarly sized phantoms and mice, presumably from absolute FDPM measurements at the emission wavelength alone. It is noteworthy that the studies of the reconstruction presented above assumed that the absorption and scattering properties were known a priori. However, using differential approaches coupled with Bayesian reconstruction approaches (see below), Eppstein and coworkers [108] demonstrated the insensitivity of reconstructions to changes in endogenous optical properties. Using a synthetic 256-cm3 volume containing 0.125-cm3 targets with 10:1 contrast in absorption owing to fluorophore and surrounded on four
510
Sevick-Muraca et al.
FIGURE 28 The reconstruction of piaxf using the excitation wave as a reference using the integral approach and Marquardt-Levenburg reconstruction. The image was required after 27 iterations with regularization parameters for AC ratio (ACR), AAC = 1.0, and for relative phase shift (RPS), A,, = 0.02 (a) optical property maps of true juaxf distribution (b) and reconstructed /uaxf distribution. Peak values of /uaxf reached 0.1205 cm 1 (c) Iteration vs. SSE.
Near-Infrared Imaging
511
sides with 68 sources and 408 detection fibers, Eppstein showed that when the absorption cross-section at the excitation wavelength, /zaxi, varied as much as 90% and was unmodeled while the scattering coefficient, ;ttsxl, varied 10% or less and was also unmodeled, the impact on the reconstruction was minimal or negligible. Similar results have recently been shown by Roy and Sevick-Muraca [ 109] who show unmodeled variations in all endogenous optical properties by as much as 50%, which did not impact reconstructions when emission FDPM measurements were individually self-referenced to excitation FDPM measurements, as was done with the CW measurements of Ntziachristos and Weissleder [95]. While it appears promising that fluorescence-enhanced optical tomography can be accomplished without much a priori information regarding the endogenous optical properties, these results are nonetheless on synthetic studies and need to be conducted on actual tissues of substantive and clinically relevant volumes for validation. 5.3
Differential Formulation of the Inverse Problem
A second approach of the full inverse imaging problem may be the differential formulation, but in reality, this time it is rewritten for measurement Z(r d > r s ), whether absolute, or relative to a reference measurement at the emission or excitation wavelength, or self-referenced relative to the excitation wavelength at each detector position, rd. We term this approach the differential formulation because a small change in the predicted measurements is directly expressed in terms of a small change in the optical properties, AX, using a Jacobian matrix, J, d(AZj)/3Xj. Considering the number of detectors to be M; then the error function is defined as the sum of square of errors between the measured and calculated values at detector i = 1..M:
F(X) = We refer to each f; as a residual and the gradients of the error function with respect to the property, X: VF(X) = 2JTf(X) |-
(39) M
V2F(X) = 2 JTJ + 2 L
i= l
-j
fi(X)V 2 f,(X)
(40) J
Consider the Taylor's expansion of function F around a small perturbation of optical properties, AX: F(X + AX) = F(X) + VF(X)-AX + - AX r - V 2 F(X)- A(X)
(41)
512
Sevick-Muraca et al.
which can be expressed as: F(X + AX) = F(X) + 2JTf(X) • AX JTJ +
+ 2 • AX
f,(X)V 2 f,(X) ' AX
(42)
and the function to be minimized, <J>(AX), can be explicitly written: 4>(AX) = F(X + AX) - F(X) = 2JTf(X) • AX + 2- AX T J T J +
L
t=i
ft(X)V2f}(X)
J
-AX
For first order Newton's methods, the term 2-AXT[E!l1 f i (X)V 2 f i (X)j • AX is neglected and the Gauss-Newton Method becomes one of minimizing, VO(AX) => 0 = JTJ • AX + J T f(X)
(44)
JTJ AX = -J T f(X)
(45)
The Levenberg-Marquardt method of optimization becomes [JTJ + A I ] - A X = -J T f(X)
(46)
The gradient based truncated Newton's method is based on retaining the second-order terms such that Eq. (43) becomes [89]: VO>(AX) => 0 = J T f(X) +
JTJ + L
fi(X)V 2 f,(X) • AX i= l
J
or, alternatively, VO(AX) => 0 = VF(X) + V2F(X) • AX
(48)
Typically, the first order Newton's methods are employed with the exception of the work by Roy [110]. In Newton's methods, it is assumed that A = J • AX and the solution is found using one of the several optimization approaches. The Jacobian can be computed either directly from the stiffness matrices of the finite element formulation or, simply but more computationally time consuming, from backward, forward, or central differencing approaches that compute the differences in the values of Z(r d , r s ) with small differences in the parameter to be updated, X(r s ). The Gauss-Newton and the Levenberg-Marquardt algorithms performed poorly in a large residual problem. Since the inverse is highly nonlinear and ill conditioned due to the error in measurement data, the residual at the solution will be large. It seems reasonable, therefore, to consider the truncated Newton method.
Near-Infrared Imaging
513
For the truncated Newton's method, the additional computational cost of computing the Hessian [associated with V2F(X)] is assisted by reverse automatic differentiation [89,110]. Using synthetic data, Roy has shown the feasibility for using the technique for three-dimensional reconstruction of lifetime, T, and absorption coefficient ^iaxf changes in frustrum and slab geometries from synthetic data containing noise that mimics experimental data [93].
5.4
Regularization and Other Approaches for Parameter Updating
In both the integral and differential formulations of the inverse problem, the tissue to be imaged must be mathematically discretized into a series of nodes or volume elements (voxels) in order to solve these inverse problems. The unknowns of the inverse problems then comprise the optical properties at each node or voxel. The final image resolution is naturally related to the density nodes or voxels. However, the dimensionality of the imaging problem is directly related to the number of nodes and can easily exceed 10,000 unknowns for a three-dimensional image. In a problem of this scale, the calculation of Jacobian matrices and matrix inversions involved in updating the optical property map are computationally intensive and contribute to the long computing times required to reconstruct the image. The instability arises because the measurement noise in the data or errors associated with the validity of the diffusion approximation can result in large errors in the reconstructed image. One of the greatest challenges associated with fluorescence-enhanced tomography is the propagation of error. In comparison with absorption imaging based on measurements of excitation light, fluorescence measurements have a reduced signal level and SNR. Lee and Sevick-Muraca [111] measured the SNR for single-pixel excitation and emission frequency domain measurements at 100 MHz to be 55 and 35 dB, respectively. In addition to the reduced signal, the noise floor of emission measurements can be expected to be elevated when excitation light leakage constitutes an increased proportion of the detected signal. Consequently, for emission tomography measurements, excitation light leakage is crucial and interference filters that attenuate excitation light four orders of magnitude (i.e., filters of optical density 4) may be clearly insufficient. Excitation light leakage will be a significant problem when emission measurements are conducted in tissue regions in which the target is absent and fluorescent contrast agents are not activated. Unfortunately, this type of error is not present in synthetic studies and is undoubtedly underestimated in the vast proportion of tomographic investigations to date.
514
5.4.1
Sevick-Muraca et al.
Regularization
Regularization is a mathematical tool used to stabilize the solution of the inverse problem and to make it more tolerant to measurement error. Regularization approaches will play an important role in the development of suitable algorithms for actual clinical screening. For example, when discretized the differential and integral general formulations result in a set of linear Newton's equations generally denoted by AY = Z, where Y are the unknown optical properties and Z are the measurements. This system is commonly solved in the least-squares sense where the object function Q = ||AY — Z||2 + A Y 2 is minimized, where A is called the regular!zation parameter. Minimization of this function results in Y = (A1 A + AI) 'ATZ. The regularization parameter is generally chosen either arbitrarily or by a Levenberg-Marquardt algorithm so that the object function is minimized [112]. Thus, the choice of regularization parameter is through a priori information and adds another degree of freedom to the inverse problem solution. While this section is not meant to be a mathematical treatise of inverse algorithm and regularization approaches, we nonetheless point out that in a recent work by Pogue and coworkers [100], a physically based rationale for empirically choosing a spatially varying regularization parameter is presented to improve image reconstruction. Bayesian Regularization. Eppstein and coworkers [87,88,91,108] use actual measurement error statistics to govern the choice of varying regularization parameters in their Kalman filter implementation to optical tomography. In their work, they developed novel Bayesian reconstruction technique, called APPRIZE (Automatic Progressive Parameter-Reducing Inverse Zonation and Estimation), specifically for groundwater problems and adapted them to fluorescence-enhanced optical tomography [104-106]. Unique components of the APPRIZE method are an approximate extended Kalman filter (AEKF), which employs measurement error and parameter uncertainty to regularize the inversion and compensate for spatial variability in SNR, and a unique approach to stabilize and accelerate convergence called data-driven zonation (DDZ). Using the notation [AX, f(X)] as described in Sec. 5.3, here the Newton's solution is formulated as [91]: AX = [[JT(Q + R) 'J + P;X'] ' • JT(Q + R ) - ' l - f ( X )
(49)
where Q is the system noise covariance which describes the inherent model mismatch between the forward model (the diffusion equation) and the actual physics of the problem; R is the covariance of the measurement error that is actually acquired in the measurement set; and Pxx is the recursively updated error covariance of the parameters, X, being estimated from the measurement error, f(X). The use of this spatially and dynamically variant co-
Near-Infrared Imaging
515
variance matrix results in the minimization of the variance of the estimated parameters taking into account the measurement and system error. The novel Bayesian minimal variance reconstruction algorithm compensates for the spatial variability in signal to noise ratio that must be expected to occur in actual NIR contrast-enhanced diagnostic medical imaging. Figure 29 illustrates the image reconstruction of 256-cm3 tissue-mimicking phantoms containing none (case 3), one (case 1), or two (case 2)
FIGURE 29 Image reconstruction with APPRIZE, a) The initial homogeneous estimate discretized onto the 9 x 1 7 x 1 7 grid used for the initial inversion iteration, and shown with the true locations of the 3 heterogeneities and the 50 detectors (small dots), b) Case 1: The reconstructed absorption due to the middle fluorescing heterogeneity, interpolated onto the 17 x 33 x 33 grid used for prediction, and shown with the locations of the 4 sources used (open circles), c) Case 2: The reconstructed absorption due to the top and bottom fluorescing heterogeneities shown with the locations of the 8 sources used (open circles), d) Case 3: The reconstructed absorption of a homogeneous phantom shown with the locations of the 4 sources used (open circles). Although the phantoms and reconstructions were actually 8 cm in the vertical dimension, only the center 4 vertical cm is shown here. (From Ref. 91.)
516
Sevick-Muraca et al.
1-cm3 heterogeneities with 50- to 100-fold greater concentration of ICG dye over background levels. The spatial parameter estimate of absorption owing to the dye was reconstructed from only 160 to 296 surface reference measurements of emission light at 830 nm (as described in Sec. 5.2.2) in response to incident 785-nm excitation light modulated at 100 MHz. Measurement error of acquired fluence at fluorescent emission wavelengths is shown to be highly variable. Another important feature of the Bayesian APPRIZE algorithm is the use of DZZ. With DDZ, spatially adjacent voxels with similarly updated estimates are identified through cluster analysis and merged into larger stochastic parameter "zones" via random field union [113]. Thus, as the iterative process proceeds, the number of unknown parameters, X, decreases dramatically, and the size, shape, value, and covariance of the different "parameter zones" are simultaneously determined in a data-driven fashion. Other approaches to reduce the dimensionality of the problems involve concurrent NIR optical imaging with MRI [10,114,115] and ultrasonography [116] to compartmentalize tissue volumes and to reduce the number of parameters to be recovered in the optical image reconstruction. 5.4.2
Simply Bounded Constrained Optimization
Imposing restrictions on the ill-posed problem can transform it to a wellposed problem as discussed above. Regularization is one method to reduce the ill posedness of the problem [117]. In the optical tomography problem, its solution, i.e., the optical properties of tissue, must satisfy certain constraints, and imposing these conditions in itself can regularize or stabilize the problem. Imposing these constraints explicitly restricts the solution sets and can restore uniqueness. Provencher and Vogel [117] have suggested two techniques: (1) prior
FIGURE 30 Three-dimensional reconstruction from simply bound truncated Newton's method. Actual distribution of fluorophore absorption coefficient of background tissue variability of endogenous (50%) and exogenous (500%) properties, (b) Reconstructed fluorophore absorption coefficient of background tissue variability of endogenous (50%) and exogenous (500%) properties using relative measurement of the emission fluence with respect to the excitation fluence at the same detector point, s = 0.0001. (c) Reconstructed fluorophore absorption coefficient of background tissue variability of endogenous (50%) and exogenous (500%) properties using relative measurement of the emission fluence with respect to the excitation fluence at the same detector point. (From Ref. 109.)
Near-Infrared Imaging
517
518
Sevick-Muraca et al.
knowledge and (2) parsimony for well posedness of the problem. The first condition requires that all prior physical knowledge about the solution be included in the model. The second condition protects against the introduction of nonphysical phenomena. Tikhonov and Arsenin [118] also suggested that, to obtain a unique and stable solution from the data, supplementary information should be used so that the inverse problem becomes well posed. The basic principle of using a priori knowledge of the properties of the inverse problem is to restrict the space of possible solutions so that the data uniquely determine a stable solution. In his work, Roy showed that the constrained optimization technique, which places simple bounds on a physical parameter to be estimated, may be more appropriate for solving the fluorescence-enhanced optical tomography problem [90]. Specifically, a range of fluorescent optical properties is physically defined for the problem and the recovered parameter, X, must always be positive. Specifically, he demonstrated use of the bounding parameter, £, not only as a means to regularize and accelerate convergence but as a means to set the level of optical property contrast that is to be reconstructed using referenced emission measurements [109]. Here the possible values of parameter estimates are stated to lie between upper and lower bounds. In the first pass of the iterative solution, the optical property map is recovered and parameter estimates that lie within the upper and lower bounds plus and minus a small bounding parameter, s, are recovered and held constant for the next iteration. In this manner, the number of unknowns decrease with iteration. Indeed, the value of the bounding parameter can be used to set the resolution and the performance of the tomographic image. For example, if the bounding parameter is large, then the tomographic image will "filter" out artifacts not associated with the target, whereas if the bounding parameter is small, then the tomographic image may sensitively capture artifacts and heterogeneity that is not necessarily associated with the target. Figure 30 illustrates the reconstruction using the simply bounded truncated Newton's method, which shows that as the bounding parameter is increased, the recovered image becomes less sensitive to the background "noise." 6.
SUMMARY: THE CHALLENGES FOR NIR FLUORESCENCE-ENHANCED IMAGING AND TOMOGRAPHY
In the earlier sections, an overview of the status of fluorescence-enhanced optical imaging was presented. The opportunity to develop an emissionbased tomographic imaging modality similar to that provided by nuclear imaging but without the use of radionucleotides is offered by NIR fluorescent agents. Yet the added challenge for NIR fluorescence-enhanced im-
Near-Infrared Imaging
519
aging over nuclear imaging is that, unlike nuclear techniques, an activating or excitation signal must first be delivered to the contrast agent before there is registration of the emission signal from the tissue. Preliminary data from animals (Table 1) and phantoms (not presented herein) suggest that penetration depth and sensitivity may very well be comparable to those of nuclear techniques. A side-by-side comparison of NIR fluorescence-enhanced imaging with nuclear imaging is needed before the comparative performance can be ascertained. Another opportunity for optical imaging is the potential for tomographic reconstruction and additional diagnostic information based on the fluorescence decay kinetics of smartly designed probes. Tomography of large tissue-simulating volume has been demonstrated from experimental data as well as synthetic data as reviewed herein (Table 2), albeit with the somewhat inconvenient point source and point detector geometries. The single point source and detector geometry is a throw-back to NIR optical tomography from endogenous contrast studies and may not be the appropriate geometry for fluorescence-enhanced optical imaging, especially when transillumination through large tissues is required. Nonetheless, the tomographic algorithms as reviewed in Sec. 5, are already established for these systems. The challenge for the future is to develop tomographic algorithms for illumination and detection that are clinically feasible and adaptable for hybrid, nuclear imaging. Although the area of NIR fluorescence-enhanced optical imaging is less than a decade old, the developments are apt to continue for the coming decade, hopefully resulting in an adjuvant tomographic imaging modality for nuclear imaging. ACKNOWLEDGMENTS The review is supported in part by the National Institutes of Health grants R01 CA67176 and R01CA88082 (P. I. Eppstein, University of Vermont) and the Texas Advanced Technology Research Program. REFERENCES 1.
2.
3.
DJ Hawrysz, EM Sevick-Muraca. Developments toward diagnostic breast cancer imaging using near-infrared optical measurements and fluorescent contrast agents. Neoplasia 2(5):388-417, 2002. BJ Tromberg, N Shah, R Lanning, A Cerussi, J Espinoza, T Phain, L Svaasand, J Butler. Non-invasive in vivo characterization of breast tumors using photon migration spectroscopy. Neoplasia 2:26-40, 2000. E Reddi, A Segalla, G Jori, PK Kerrigan, PA Lidell, AL Moore, TA Moore,
520
4. 5.
6.
7.
8.
9.
10.
11. 12.
13.
14. 15.
16.
17.
18.
Sevick-Muraca et al. D Gust. Carotenoporphyrins as selective photodiagnostic agents for tumors. Br J Cancer 69:40-45, 1994. T Takemura, S Nakajima, I Sakata. Tumor-localizing fluorescent diagnostic agents without phototoxicity. Photochem Photobiol 59:366-370, 1994. R Biolo, G Jori, JC Kennedy, P Nadeau, R Potteir, E Reddi, G Weagle. A comparison of fluorescence methods used in the pharmacokinetic studies of Zn(II) phthalocyanine in mice. Photochem Photobiol 53:113-118, 1991. RC Straight, RE Benner, RW McClane, PMNY Go, G Yoon, JA Dixon. Application of charge-coupled device technology for measurement of laser light and fluorescence distribution in tumors for photodynamic therapy. Photochem Photobiol 53:787-796, 1991. M Kohl, U Sukowski, B Ebert, J Neukammer, HH Rinneberg. Imaging of superficially growing tumors by delayed observation of laser-induced fluorescence. Proc SPIE 1881:206-221, 1993. R Cubeddu, G Canti, P Taroni, G Valentini. Time-gated fluorescence imaging for the diagnosis of tumors in a murine model. Photochem Photobiol 57:480485, 1993. R Cubeddu, G Canti, A Pifferi, P Taroni, G Valentini. Fluorescence lifetime imaging of experimental tumors in hematoporphyrin derivative-sensitized mice. Photochem Photobiol 66:229-236, 1997. V Ntziachristos, AG Yodh, M Schnall, B Chance. Concurrent MRI and diffuse optical tomography of the breast after indocyanine green enhancement. Proc Natl Acad Sci 97:2767-2772, 2000. JR Lakowicz. Principles of Fluorescence Spectroscopy. Plenum Press, New York, 1983. JP Houston, AB Thompson, M Gurfinkel, EM Sevick-Muraca. Sensitivity and penetration depth of NIR fluorescence-enhanced contrast imaging. Photochem Photobiol 2002 (submitted 07/02). AB Thompson, EM Sevick-Muraca. NIR fluorescence contrast enhanced imaging with ICCD homodyne detection: Measurement precision and accuracy. J Biomed Opt in press (2003). J Lakowicz, K Berndt. Lifetime-sensitive fluorescence imaging using an if phase-camera. Rev Sci Instrum 62:1727-1734, 1991. EM Sevick, JR Lakowicz, H Szmacinski, K Nowaczyk, M Johnson. Frequency-domain imaging of obscure absorbers: principles and applications. J Photochem Photobiol 16:169-185, 1992. JS Reynolds, TL Troy, EM Sevick-Muraca. Multipixel techniques for frequency-domain photon migration imaging. Biotechnol Progr 13(5):669-680, 1997. A Pelegrin, S Folli, F Buchegger, J-P Mach, G Wagnieres, H van den Bergh. Antibody-fluorescein conjugates for photoimmunodiagnosis of human colon carcinoma in nude mice. Cancer 67:2529-2537, 1991. S Folli, G Wagnieres, A Pelegrin. JM Calmes, D Braichotte, F Buchegger, Y Chalandon, N Hardman, DC Heusser, JC Givel, G Chapuis, A Chatelain, H van Den Bergh, JP Mach. Immunophotodiagnosis of colon carcinomas in
Near-Infrared Imaging
521
patients injected with fluoresceinated chimeric antibodies against carcinoembryonic antigen. Proc Natl Acad Sci 89:7973-7977, 1992. 19. S Folli, P Westermann, D Braichotte, A Pelegrin, G Wagnieres, H van den Bergh, JP Mach. Antibody-indocyanin conjugates for immunophotodetection of human squamous cell carcinoma in nude mice. Cancer Res 54:2643-2649, 1994. 20. MM Haglund, DW Hochman. AM Spence, MS Berger. Enhanced optical imaging of rat gliomas and tumor margins. Neurosurgery 35:930-939, 1994. 21. S Mordon, JM Devoisselle, V Maunoury. In vivo pH measurement and imaging of tumor tissue using a pH-sensitive fluorescent probe (5,6-carboxyfluorescein): instrumental and experimental studies. Photochem Photobiol 60: 274-279, 1994. 22. B Ballou, GW Fisher, AS Waggoner, DL Parkas, JM Reiland, R Jaffe, RB Mujumdar, SR Mujumdar, TR Hakala. Tumor labeling in vivo using cyanineconjugated monoclonal antibodies. Cancer Immunol Immunother 41:257263, 1995. 23. JM Devoisselle, S Soulie, SR Mordon, G Mestres, TMD Desmettre, H Maillols. Effect of indocyanine green formulation on blood clearance and in vivo fluorescence kinetic profile of skin. Proc SPIE 2627:100-108, 1995. 24. I Rokahr, S Andersson-Engels, S Svanberg, M-A D'Hallewin, L Baert, I Wang, K Svanberg. Optical detection of human urinary bladder carcinoma utilising tissue autofluorescence and protoporphyrin IX-induced fluorescence following low-dose ALA instillation. Proc SPIE 2627:2-12, 1995. 25. MM Haglund, MS Berger, DW Hochman. Enhanced optical imaging of human gliomas and tumor margins. Neurosurgery 38:308-316, 1996. 26. K Sakatani, M Kashiwasake-Jibu, Y Taka, S Wang, H Zuo, K Yamamoto, K Shimizu. Noninvasive optical imaging of the subarachnoid space and cerebrospinal fluid pathways based on near-infrared fluorescence. Journal of Neurosurgery 87:738-745, 1997. 27. D Neri. Targeting by affinity-matured recombinant antibody fragments of an angiogenisis-associated fibronectin isoform. Nat Biotechnol 15:1271-1275, 1997. 28. B Ballou, GW Fisher, J-S Deng, TR Hakala, M Srivastava, DL Farkas. Fluorochrome-labeled antibodies in vivo: assessment of tumor imaging using Cy3, Cy5, Cy5.5, and Cy7. Cancer Det Prev 22(3):251-257, 1998. 29. C Eker, S Montan, E Jaramillo, K Koizumi, C Rubio, S Andersson-Engels, K Svanberg, S Svanberg, P Slezak. Clinical spectral characterization of coIonic mucosal lesions using autofluorescence and d aminolevulinic acid sensitizations. Gut 44:511-518, 1999. 30. JS Reynolds, TL Troy, RH Mayer, AB Thompson, DJ Waters, KK Cornell, PW Snyder, EM Sevick-Muraca. Imaging of spontaneous canine mammary tumors using fluorescent contrast agents. Photochem Photobiol 70:87-94, 1999. 31. A Becker, K Licha, M Kresse, B Riefke, U Sukowski, B Ebert, H Rinneberg, W Semmler. Transferrin-mediated tumor delivery of contrast media for optical imaging and magnetic resonance imaging. Proc SPIE 3600:142-150, 1999.
522
Sevick-Muraca et al.
32.
R Weissleder, CH Tung, U Mahmood, A Bogdanov Jr. In vivo imaging of tumors with protease-activated near-infrared fluorescent probes. Nat Biotechnol 17:375-378, 1999. U Mahmood, C-H Tung. A Bogdanov, R Weissleder. Near-infrared optical imaging of protease activity for tumor detection. Radiology 21(3):866-870, 1999. A Becker, B Riefke, B Ebert, U Sukowski, H Rinneberg, W Semmler, K Licha. Macromolecular contrast agents for optical imaging of tumors: comparison of indotricarbocyanine-labeled human serum albumin and transferrin. Photochem Photobiol 72:234-241, 2000. M Gurfinkel, AB Thomson, W Ralston, TL Troy, AL Moore, TA Moore, JD Gust, D Tatman, JS Reynolds. B Muggenburg, K Nikula, R Pandey, RH Mayer, DJ Hawrysz, EM Sevick-Muraca. Pharmacokinetics of ICG and HPPH-car for the detection of normal and tumor tissue using fluorescence, near-infrared reflectance imaging: A case study. Photochem Photobiol 72:94102, 2000. K Licha, B Riefke, V Ntziachristos, A Becker, B Chance, W Semmler. Hydrophilic cyanine dyes as contrast agents for near-infrared tumor imaging: synthesis, photophysical properties and spectroscopic in vivo charactarization. Photochem Photobiol 72:392-398, 2000. M Yang, E Baranov, F Sun. X Li, L Li, S Hasegawa. M Bouvet, M AlTuwaijri, T Chishima, H Shimada, A Moossa, S Penman, R Hoffman. Wholebody optical imaging of green fluorescent protein-expressing tumors and metastases. PNAS 97:1206-1211, 2000. RM Hoffman. Visualization of GFP-expressing tumors and metastasis in vivo. Biotechniques 30:1016-1026, 2001. JE Bugaj, S Achilefu, RB Dorshow. R Rajagopalan. Novel fluorescent contrast agents for optical imaging of in vivo tumors based on a receptor-targeted dyepeptide conjugate platform. J Biomed Opt 6:122-133, 2001. S Achilefu. RB Dorshow, JE Bugaj, R Rajogopalan. Novel receptor-targeted fluorescent contrast agents for in vivo tumor imaging. Invest Radiol 35:479485, 2000. S Achilefu, JE Bugaj, RB Dorshow, HN Jimenez, R Rajagopalan. New approach to optical imaging of tumors. Proc SPIE 4259:110-114, 2001. S Achilefu, JE Bugaj, RB Dorshow, HN Jimenez, R Rajagopalan, RR Wilhelm, EG Webb, JL Erion. Site-specific tumor targeted fluorescent contrast agents. Proc SPIE 4156:69-78, 2001. A Becker, C Hessenius, K Licha, B Ebert, U Sukowski, W Semmler, B Wiedenmann C Grotzinger. Receptor-targeted optical imaging of tumors with near-infrared fluorescent ligands. Nat Biotechnol 19:327-331, 2001. C Bremen C Tung, R Weissleder. In vivo molecular target assessment of matrix metalloproteinase inhibition. Nat Med 7:743-748. 2001. B Ebert, U Sukowski, D Grosenick, H Wabnitz, KT Moesta, K Licha, A Becker, W Semmler, PM Schlag, H Rinneberg. Near-infrared fluorescent dyes for enhanced contrast in optical mammography: phantom experiments. J Biomed Opt 6(2): 134-140, 2001.
33.
34.
35.
36.
37.
38. 39.
40.
41. 42.
43.
44. 45.
Near-Infrared Imaging 46.
523
JC Finlay, DL Conover, EL Hull, TH Foster. Porphyrin bleaching and PDTinduced spectral changes are irradiance dependent in ALA-sensitized normal rat skin in vivo. Photochem Photobiol 73:54-63, 2001. 47. BW Rice, MD Cable, MB Nelson. In vivo imaging of light-emitting probes. J Biomed Opt 6(4):432-440, 2001. 48. NS Soukos, MR Hamblin, S Keel, RL Fabian, TF Deutsch, T Hasan. Epidermal growth factor receptor-targeted immunophotodiagnosis and photoimmunotherapy of oral precancer in vivo. Cancer Res 61:4490-4496, 2001. 49. Y Yang, N Iftimia, Y Xu, H Jiang. Frequency-domain fluorescent diffusion tomography of turbid media and in vivo tissues. Proc SPIE 4250:537-545, 2001. 50. CM Leevy, F Smith, J Longueville. Indocyanine green clearance as a test for hepatic function. Evaluation by dichromatic ear densitometry. JAMA 200: 236-240, 1967. 51. K Kogure, NJ David, U Yamanouchi, E Choromokos. Infrared absorption angiography of the fundus circulation. Arch Ophthalmol 83(2):209-214, 1970. 52. EM Sevick-Muraca, G Lopez, TL Troy, JS Reynolds, CL Hutchinson. Fluorescence and absorption contrast mechanisms for biomedical optical imaging using frequency-domain techniques. Photochem Photobiol 66:55-64, 1997. 53. ML Landsman, G Kwant, G Mook, WG Zijlstra. Light-absorbing properties, stability, and spectral stabilization of Indocyanine green. J Appl Physiol 40: 575-583, 1976. 54. S Mordon, JM Devoisselle, S Soulie-Begu, T Desmettre. Indocyanine green: physiochemical factors affecting its fluorescence in vivo. Microvas Res 55: 146-152, 1998. 55. KA Schafer, G Kelly, R Schrader, WC Griffith, BA Muggenburg, LA Tierney, JF Lechner, EB Janovitz, FF Hahn. A canine model of familial mammary gland neoplasia. Vet Pathol 35:168-177, 1998. 56. V Ntziachristos, X Ma, B Chance. Time-correlated single photon counting imager for simultaneous magnetic resonance and near-infrared mammography. Rev Sci Instrum 69:4221-4233, 1998. 57. KM McMaster, AE Giuliano, MI Ross, DS Reintgen, KK Hunt, VS Klimberg, PW Whitworth, LC Tafra, MJ Edwards. Sentinel lymph node biopsy for breast cancer—not yet the standard of care. N Engl J Med 339:990-995, 1998. 58. D Krag, D Weaver, T Ashikaga, F Moffat, S Klimberg, C Shriver, S Feldman, R Kusminsky, M Gadd, J Kuhn, S Harlow, P Beitsch, P Whitworth, R Foster Jr, K Dowlatshahi. The sentinel node in breast cancer—a multicenter validation study. N Engl J Med 339:941-946, 1998. 59. K Licha, A Becker, F Kratz, W Semmler. New contrast agents for optical imaging: acid-cleavable conjugates of cyanine dyes with biomolecules. Proc SPIE 3600:29-35, 1999. 60. MM Huber, AB Staubili, K Kustedjo, MHB Gray, J Shih, S Fraser, RE Jacobs, TJ Meade. Fluorescently detectable magnetic resonance imaging agents. Bioconj Chem 9:242-249, 1998. 61. X Li, B Chance, AG Yodh. Fluorescence heterogeneities in turbid media,
524
62.
63.
64.
65.
66.
67. 68. 69.
70.
71.
72.
73.
74. 75.
76.
77.
Sevick-Muraca et al. limits for detection, characterization, and comparison with absorption. Appl Opt 37:6833-6843, 1998. XD Zhu, S-P Wei, XW Guo. Imaging objects in tissue-like media with optical tagging and the diffuse photon differential transmittance. J Opt Soc Am 14: 300-305, 1997. RC Haskell, LO Svassand, T-T Tsay, T-C Feng, MS McAdams, BJ Tromberg. Boundary conditions for the diffusion equation in radiative transfer. J Opt Soc Am A 11:2727-2741, 1994. MS Patterson, B Chance, B Wilson. Time resolved reflectance and transmittance for the non-invasive measurement of tissue optical properties. Appl Opt 28:2331-2336, 1989. AH Hielscher, SL Jacques, L Wang, FK Tittel. The influence of boundary conditions on the accuracy of diffusion theory in time-resolved reflectance spectroscopy of biological tissues. Phys Med Biol 40:1957-1975, 1995. TJ Farrell, MS Patterson, B Wilson. A diffusion theory model of spatially resolved, steady-state diffuse reflectance for the noninvasive determination of tissue optical properties in vivo. Med Phys 9:879-888, 1992. EM Sevick-Muraca, CL Burch. Origin of phosphorescence re-emitted from tissues. Opt Lett 19:1928-1930, 1994. R Richards-Kortum, EM Sevick-Muraca. Quantitative optical spectroscopy for tissue diagnostics. Ann Rev Phys Chem 47:555-606, 1996. E Kuwana, EM Sevick-Muraca. Fluorescence lifetime spectroscopy in multiply scattering media with dyes exhibiting multi-exponential decays kinetics. Biophys J 83:1165-1176, 2002. V Chenomordik, D Hattery, I Gannot, AH Gandjbakhche. Inverse method 3-D reconstruction of localized in vivo fluorescence—application to Sjogren syndrome. IEEE J Sel Top Quant Electron 54:930-935, 1999. D Hattery, V Chernomordik, I Gannot, M Loew, AH Gandjbakhche. Fluorescence measurement of localized, deeply embedded physiological processes. Proc SPIE 3978:377-382, 2000. MS Patterson, BW Pogue. Mathematical model for time-resolved and frequency-domain fluorescence spectroscopy in biological tissue. Appl Opt 33: 1963-1964, 1994. CL Hutchinson, JR Lakov.'icz, EM Sevick-Muraca. Fluorescence lifetimebased sensing in tissues: a computational study. Biophys J 68:1574-1582, 1995. S Fantini, E Gratton. Fluorescence photon density waves in optically diffuse media. Optic Comm 173:73-79, 2000. RH Mayer, JS Reynolds, EM Sevick-Muraca. Measurement of fluorescence lifetime in scattering media using frequency-domain photon migration. Appl Opt 38:4930-4938, 1999. AE Cerussi, JS Maier, S Fantini, MA Franceschini, WW Mantulin, E Gratton. Experimental verification of a theory for time-resolved fluorescence spectroscopy of thick tissues. Appl Opt 36:116-124, 1997. A Chen. Effects of fluorescence and phosphorescence lifetime on frequency
Near-Infrared Imaging
78.
79.
80. 81. 82.
83.
84.
85. 86.
87. 88.
89.
90. 91.
92.
93.
525
domain optical contrast for biomedical optical imaging. MS thesis, Purdue University, West Lafayette, IN, 1997. MA O'Leary, DA Boas, B Chance, AG Yodh. Reradiation and imaging of diffuse photon density waves using fluorescent inhomogeneities. J Lumin 60, 61:281-286, 1994. J Wu, Y Wang, L Perleman, 1 Itzkan, RR Dasari, MS Feld. Time-resolved multichannel imaging of fluorescent objects embedded in turbid media. Opt Lett 20:489-491, 1995. J Chang, RL Barbour, H Graber, R Aronson. Fluorescence Optical Tomography. Proc SPIE 2570:59-72, 1995. MA O'Leary, DA Boas, XD Li, B Chance, AG Yodh. Fluorescence lifetime imaging in turbid media. Opt Lett 21(2):158-160, 1996. DY Paithankar, AU Chen, BW Pogue, MS Patterson, EM Sevick-Muraca. Imaging of fluorescent yield and lifetime from multiply scattered light reemitted from random media. Appl Opt 36:2260-2272, 1997. J Wu, L Perelman, RR Dasari, MS Feld. Fluorescence tomographic imaging in turbid media using early-arriving photons and Laplace transforms. Proc Natl Acad Sci 94:8783-8788, 1997. J Chang, HL Graber, RL Barbour. Improved reconstruction algorithm for luminescence when background luminophore is present. Appl Opt 37:35473552, 1998. H Jiang. Frequency-domain fluorescent diffusion tomography: a finite-element-based algorithm and simulations. Appl Opt 37(22):5337-5343, 1998. EL Hull, MG Nichols, TH Foster. Localization of luminescent inhomogeneities in turbid media with spatially resolved measurements of cw diffuse luminescence emittance. Appl Opt 37:2755-2765, 1998. MJ Eppstein, DE Dougherty, DJ Hawrysz, EM Sevick-Muraca. Three-dimensional optical tomography. Proc SPIE 3497:97-105, 1999. MJ Eppstein, DE Dougherty, TL Troy, EM Sevick-Muraca. Biomedical optical tomography using dynamic parameterization and Bayesian conditioning on photon migration measurements. Appl Opt 38:2138-2150, 1999. R Roy, EM Sevick-Muraca. Truncated Newton's optimization scheme for absorption and fluorescence optical tomography: Part II Reconstruction from synthetic measurements. Opt Exp 4:372-382, 1999. R Roy, EM Sevick-Muraca. Active constrained truncated Newton method for simple-bound optical tomography. J Opt Soc Am A 17(9): 1627-1641, 2000. MJ Eppstein, DJ Hawrysz, A Godavarty, EM Sevick-Muraca. Three-dimensional, near-infrared fluorescence tomography with Bayesian methodologies for image reconstruction from sparse and noisy data sets. Proc Natl Acad Sci 99:9619-9624, 2002. DJ Hawrysz, MJ Eppstein, J Lee, EM Sevick-Muraca. Error consideration in contrast-enhanced three-dimensional optical tomography. Opt Lett 26(10): 704-706, 2001 R Roy, EM Sevick-Muraca. Three-dimensional unconstrained and constrained image-reconstruction techniques applied to fluorescence, frequency-domain photon migration. Appl Opt 40 (13):2206-2215, 2001.
526 94.
Sevick-Muraca et al.
R Roy, EM Sevick-Muraca. A numerical study of gradient-based nonlinear optimization methods for contrast-enhanced optical tomography. Opt Exp 9(l):49-65, 2001. 95. V Ntziachristos, R Weissleder. Experimental three-dimensional fluorescence reconstruction of diffuse media by use of a normalized Born approximation. Opt Lett 26(12):893-895, 2001. 96. J Lee, EM Sevick-Muraca. 3-D fluorescence enhanced optical tomography using references frequency-domain photon migration measurements at emission and excitation measurements. J Opt Soc Am A 19:759-771, 2002. 97. EM Sevick-Muraca, DY Paithankar. Fluorescence imaging system and measurement. U.S. patent #5.865,754, issued 2/2/99. 98. JM Schmitt, A Knuttel, JR Knutson. Interference of diffusive light waves. J Opt Soc Amer A 9:1832-1843, 1992. 99. DY Paithankar, A Chen, EM Sevick-Muraca. Fluorescence yield and lifetime imaging in tissues and other scattering media. Proc Soc Photo-Opt lustrum Eng 2679:162-175. 1996. 100. B Pogue, T McBride, J Prewitt. U Osterberg, K Paulsen. Spatially varying regulari/ation improves diffuse optical tomography. Appl Opt 38:2950-2961. 1999. 101. MJ Holboke, AG Yodh. Parallel three-dimensional diffuse optical tomography. Biomedical Topical Meetings, OS A, Miami Beach, FL, 2000, pp 177-179. 102. M Xu, M Lax, RR Alfano. Time-resolved fourier diffuse optical tomography. Biomedical Topical Meetings, OSA, Miami Beach, FL, 2000, pp 345-347. 103. SR Arridge, JC Hebden. M Schweiger, FEW Schmidt, ME Fry, EMC Hillman, H Dehghani, DT Delpy. A method for three-dimensional time-resolved optical tomography. Int J Imaging Sys Technol 11:2-11, 2000. 104. MJ Eppstein, DE Dougherty. Optimal 3-D traveltime tomography. Geophysics 63:1053-1061. 1998. 105. MJ Eppstein, DE Dougherty. Efficient three-dimensional data inversion: soil characterization and moisture monitoring from cross-well ground penetrating radar at a Vermont test site. Wat Resources Res 34:1889-1900. 1998. 106. MJ Eppstein. DE Dougherty. Three-dimensional stochastic tomography with upscaling. U.S. patent application 09/110,506, (9 July 1998). 107. TJ Yorkey, JG Webster, WJ Tompkins. Comparing reconstruction algorithms for electrical impedance tomography. IEEE Trans Biomed Eng BME-34, 843— 852, 1987. 108. MJ Eppstein, DE Dougherty, DJ Hawrysz, EM Sevick-Muraca. Three-dimensional Bayesian optical image reconstruction with domain decomposition. IEEE Trans Med Imaging 20(3): 147-163, 2000. R Roy. A Godavarty, EM Sevick-Muraca. Fluorescence-enhanced, optical tomography using referenced measurements of heterogeneous media. IEEE Trans Med Imaging (in press). R Roy. EM Sevick-Muraca. Truncated Newton's optimization scheme for absorption and fluorescence optical tomography: part I theory and formulation. Opt Exp 4:353-371, 1999. J Lee. EM Sevick-Muraca. Fluorescence-enhanced absorption imaging using
Near-Infrared Imaging
112. 113. 114.
115.
116.
117.
118.
527
frequency-domain photon migration: tolerance to measurement error. J Biomed Opt 6(l):58-67, 2000. SR Arridge. Optical tomography in medical imaging. Inverse Problems 15: R41-R93, 1999. MJ Eppstein, DE Dougherty. Simultaneous estimation of transmittivity values and zonation. Wat Resour Res 32:3321-3336, 1996. Y Pei, F-B Lin, RL Barbour. Modeling of sensitivity and resolution to an included object in homogeneous scattering media and in MRI-derived breast maps. J Biomed Opt 5:302-219, 1999. BW Pogue, KD Paulsen. High-resolution near-infrared tomographic imaging simulations of the rat cranium by use of a priori magnetic resonance imaging structural information. Opt Lett 23:1716-1718, 1998. MJ Holboke, BJ Tromberg, X Li, N Shah, J Fishkin, D Kidney, J Butler, B Chance, AG Yodh. Three-dimensional diffuse optical mammography with ultrasound localization in a human subject. J Biomed Opt 5:237-247, 2000. SW Provencher, RH Vogel. Regularization techniques for inverse problems in molecular biology. Numerical Treatment of Inverse Problems in Differential and Integral Equations. Boston: Birkhauser Press, 1983. AN Tikhonov, VY Arsenin. Solution of Ill-Posed Problems. Washington, DC: V. H. Winston & Sons, 1977.
15 Fluorescence in Photodynamic Therapy Dosimetry Brian C. Wilson Ontario Cancer Institute, University of Toronto, and Photonics Research Ontario, Toronto, Ontario, Canada Robert A. Weersink Photonics Research Ontario, Toronto, Ontario, Canada Lothar Lilge University of Toronto and Photonics Research Ontario, Toronto, Ontario, Canada
1.
INTRODUCTION
Photodynamic therapy (PDT) is a technique for treating a variety of malignant and nonmalignant conditions based on the use of light-activated drugs (photosensitizers). Typically, the photosensitizer is administered either systemically (intravenously or orally) or topically to the tissue to be treated. After allowing time for uptake of the photosensitizer to the target tissues or tissue structures, light of an appropriate wavelength to activate the drug is applied. This results in the photoproduction of one or more cytotoxic agents, leading to the intended cellular and tissue effects. For most photosensitizers used or under investigation clinically, it is likely that the main photophysical pathway is production of singlet oxygen, 'O2. Singlet oxygen is an excited form of oxygen that is highly reactive with biomolecules, leading typically 529
530
Wilson et al.
to oxidative damage. The Jablonski energy diagram for this so-called type II process is shown in Fig. 1. The absorption of a photon by the groundstate photosensitizer activates the molecule to the excited singlet state. This short-lived state (typically nanoseconds) may de-excite to the ground state either nonradiatively or by fluorescence emission, or may undergo a change of spin to the triplet state. This is relatively long lived, since decay to the ground state is a quantum mechanically forbidden transition. Energy exchange with ground-state oxygen is, however, an allowed transition, since 3 O2 is also a triplet state, and this excites the oxygen to 'O2. (Note that type I processes are those in which the reactive species are generated directly from the photosensitizer-excited singlet or triplet state and may or may not be oxygen dependent. They also lead to photosensitizer fluorescence from the singlet state.) Most of the current photosensitizers have some fluorescence emission. (The in vivo measurement of nonfluorescent drugs is discussed briefly later.) For the PDT efficiency to be as high as possible, a high triplet state quantum yield is desirable, which competes with the fluorescence quantum yield. However, using the fluorescence emission to monitor the photosensitizer does not require a high yield (e.g., a few percent), so that in practice a high triplet-state yield can be selected. Figure 2 shows the absorption and fluorescence emission spectra of some common PDT drugs. Note that the fluorescence excitation spectrum is generally very similar to the absorption spectrum, but that these may both change in biological media compared to the spectra in simple solutions, due to substrate (e.g., protein) binding. Shifts in the absorption peaks by several nanometers are common.
SINGLET STATES
"
BIOLOGICAL »-"-• ^
EFFECT
GROUND STATE
OXYGEN
FIGURE 1 Jablonski diagram for type II photodynamic reactions. The vertical axis indicates the energy for the distinct electronic states of the photosensitizer and oxygen molecules.
Photodynamic Therapy Dosimetry
531
O CD
•j=> Q) X O LU Q
400
600
500 600 Wavelength (nm)
700
800
(a)
FIGURE 2 (a) Absorption and fluorescence emission spectra of some common PDT photosensitizers. (b) Corresponding excitation-emission matrices.
Most PDT photosensitizers have spectra across the visible range, often with a strong Soret band in the UV-A/blue region. In order to obtain maximal light penetration in tissue for treating larger lesions (such as solid tumors), many newer PDT drugs have strong absorption bands in the far-red (650700 nm) and near-infrared (700-850 nm), and correspondingly their fluorescence is also at these long wavelengths. Figure 3 summarizes the classes of disease for which PDT is being investigated as a possible therapeutic method. Several of these treatments have regulatory approvals and are entering routine clinical practice. The potential applications are wide, reflecting the facts that (1) PDT can be applied to many body sites through the use of fiberoptic light delivery, as summarized schematically in Fig. 4, and (2) with different photosensitizers the mechanisms of action are varied, and include direct cell kill (by necrosis
Wilson et al.
532
Photofrin E •S c g ro
•3x ill
o in. cs
300
400 500 600 Emission (nm)
700
AIPcS4
o 10 c ,2 o
5 8
'Q x Hi
i 300
400 500 600 Emission (nm)
700
.m
(b) FIGURE 2
Continued
and/or apoptosis), vascular destruction, and immune stimulation or suppression. The photosensitizer fluorescence may be utilized in a number of ways, including: 1. 2. 3.
Lesion detection in vivo Lesion localization in vivo for therapy guidance, both PDT and surgical Quantification of the concentration of the photosensitizer in the target and other tissues in vivo for the purposes of a. Determining the drug pharmacokinetics in order to select the optimum time for light activation, or
Photodynamic Therapy Dosimetry In Vivo
photosensitizer
533
treatment light • solid tumors • dyslasias • papillomas • rheumatoid arthritis • age-related macular degeneration • cosmesis • psoriasis • endometrial ablation • localized infection • prophylaxis of arterial restenosis
Ex Vivo
• extracorporeal photophoresis • blood purging: HIV, hepatitis B, protozoa • bone marrow purging photosensitizer
I
FIGURE 3
treatment
Potential clinical applications of PDT.
b.
4.
5. 6.
Individualizing the light delivery to produce the required biological response ("dosimetry") Measurement of the photosensitizer photobleaching during PDT light activation, in order to estimate the effective PDT "dose" delivered Photosensitizer quantification in tissue samples (e.g., biopsies) ex vivo Determining the microdistribution of photosensitizer in tissues (ex vivo) or cells (in vitro) by fluorescence microscopy
Tissue autofluorescence may be used in addition or as an alternative to photosensitizer fluorescence for purposes 1, 2, and 4, or as a monitor of tissue response to treatment. Fluorescence imaging of other "contrast agents" may be of value for PDT in particular sites, such as the use of fluorescein angiography in PDT of age-related macular degeneration, a dis-
Wilson et al.
534
JLJJ
Superficial
Interstitial
Endoscopic
Whole Surface Intracavitory
FIGURE 4 Light irradiation methods for PDT, showing the wide variety of target geometries.
ease that leads to blindness due to growth of abnormal blood vessels in the choroid layer of the eye near the macula, which is the area of central acute vision. In this case, as illustrated in Fig. 5, fluorescein angiography is used both before treatment to show the area of choroidal neovasculature so as to target the treatment light and after treatment to evaluate the area and completeness of vascular closure. In this chapter, the emphasis will be on topics 2-6, since lesion detection and localization for treatment guidance are dealt with elsewhere, in the general context of fluorescence "contrast agents." It is worth noting, however, that many of the in vivo fluorescence-based methods and instruments that have been developed over the past 20 years have been strongly
Photodynamic Therapy Dosimetry
535
FIGURE 5 Fluorescein angiograms before and after PDT for age-related macular degeneration, showing closure of the area of abnormal neovasculature.
associated with the corresponding development of PDT as a treatment modality, since they provide complementary "search and destroy" capabilities. Thus, for example, autofluorescence bronchoscopy (Chapter 11) grew from the original endoscopic spectroscopy and imaging work with the photosensitizer hematoporphyrin derivative (HPD) and was originally aimed at localizing tumors in the lung for PDT treatment. (Interestingly, HPD itself was developed from hematoporphyrin in an attempt to improve its fluorescence properties and was only subsequently discovered to be an effective PDT drug.) Studies were performed with successively decreasing doses of HPD in order to reduce the associated skin photosensitization, and in the case of early bronchial cancer/dysplasia it was found that the detection accuracy was greatest with zero dose, i.e., using only the autofluorescence. However, fluorescence of HPD continues for the other objectives listed above, and many developments in fluorescence spectroscopy and imaging in vivo continue to be tightly linked to improving and monitoring PDT treatments, although they are also developing independently. In addition to PDT applications exploiting the photosensitizer fluorescence, fluorescence-based optical fiber probes are being developed for light fluence monitoring in tissue. In these, a small point-like volume of specific fluorophore is incorporated into an optical fiber and placed on or within the tissue. The PDT treatment light then excites the fluorescence, a fixed fraction of which is transmitted along the fiber and detected. Probes incorporating several different fluorophores at different positions along the fiber allow
536
Wilson et al.
multipoint measurements to be made simultaneously by spectral deconvolution. The concept can also be extended to tissue oxygen measurements using phosphorescence lifetime measurements. This technology and its applications will be discussed at the end of the chapter. 2. 2.1
PHOTOSENSITIZER QUANTIFICATION IN VIVO Relative Measurements
Figure 6 shows a simplified schematic diagram of noninvasive measurement of photosensitzer fluorescence in vivo. The excitation light from a light source (laser, or filtered lamp or light-emitting diode) at wavelength Aex is delivered to the tissue surface by an optical fiber (or fiber bundle). Then one or more collection fibers pick up some fraction of the light remitted through the tissue surface; this comprises elastically scattered light (diffuse reflectance), photosensitizer fluorescence (at wavelength A cm ), tissue autofluorescence, and inelastic (Raman) scattered light. Typically, the order of magnitude of total photons remitted for each of these components, relative to the incident light, is around 10 ', 10~ 4 , 10 \ and 10 7, respectively, a small fraction of which is collected by the detector fibers. The elastic scattering is removed by optical filtering, typically, placing a long-pass filter (>Aex) between the output end of the collection fibers and the photodetector. The Raman signal is usually negligible, whereas the photosensitizer fluorescence is observed superimposed on the tissue autofluorescence background. Figure 7 shows such fluorescence spectra, for different administered doses of a specific agent, in this case the prodrug aminolevulinic acid (ALA), m
source
f photodetector i
xpPH;!H tissue
FIGURE 6 Schematic of fiberoptic detection of fluorescence at a tissue surface.
Photodynamic Therapy Dosimetry
537
0.005
0.000
?„ (nm)
0.05
0-00
tQmg/kg@3hf
'
'
02
0.0 500 —-Barrett's
"-""I
(nm)
30mg/kg @ 6 Nr
600
700 Squarnous
FIGURE 7 In vivo fluorescence spectra in Barrett's and normal human esophagus, at different times post ALA administration and for different ALA doses. Individual spectra represent measurements at different points on the tissue and are normalized to the autofluorescence background at 500 nm. (See color insert.)
which induces increased biosynthesis of heme, the penultimate step of which is the fluorescent photosenisitizer protoporphyrin IX (PpIX). These data illustrate a number of points. First, the fluorescence intensity is highly variable from patient to patient and even point to point in the same patient for a given drug level and time point. This type of variability has been reported for other photosensitizers and is, of course, a major reason why quantitative measurements are required. Second, the drug fluorescence spectrum is known and can be easily identified on the autofluorescence background, at least for higher levels of drug; however, at low levels, the added drug spectrum may not be easily seen, especially if the autofluorescence itself is highly variable (which is certainly the case for diseased tissue, and indeed is the basis of autofluorescence-based diagnostics). Third, the increase in the PpIX fluorescence with ALA dose and the dependence on time after administration can be observed, at least in the average spectra of each tissue type. In order to obtain a quantitative estimate of the PpIX concentration, a fit may be made to the autofluorescence background: in the simplest case, by a linear interpolation between points just outside the main PpIX fluores-
Wilson et al.
538
cence peak, as illustrated in Fig. 8a, or by a polynomial fit, taking the (local) shape of the background into account (Fig. 8b). The residual photosensitizer signal, PSF*, is then proportional to the (average) tissue concentration, C ps . If all factors other than C—namely, the excitation light, excitation and detection geometry, fluorescence quantum yield, and tissue optical absorption and scattering coefficients at Aex and A ox —remain constant, then the FSF* values yield the relative drug concentrations. The assumption of these factors remaining constant usually holds best in following the photosensitizer pharmacokinetics in a given location in an individual patient. Even in this idealized case, however, complications arise. For example, photosensitizer aggregation/disaggregation depends in many cases on the local microenvironment (pH, polarity, etc.) and this alters the fluorescence yield. The yield can also change when the photosensitizer binds
550
600
650
700
750
Wavelength (run)
b
> CO
O
o C
O
550
600
650
700
750
Wavelength (nm) FIGURE 8 Subtraction of the tissue autofluorescence background to measure the photosensitizer level, (a) Linear background, (b) Polynomial fit to the background.
Photodynamic Therapy Dosimetry
539
to biomolecules in the tissue. Hematoporphyrin derivative (commercially, Photofrin, Axcan, Canada), for which most clinical approvals are in place at present, is a mixure of many porphyrin molecules, with different pharmacokinetics, binding, fluorescence yield, and photoactivity, so that even the relative fluorescence signal in an individual patient can be difficult to interpret in terms of how much photodyamically active drug is present in the tissue. Nevertheless, the relative PSF* measurements can provide useful guidance, even if not exact information, on, for example, the time to maximal photosensitizer uptake or time to maximal tumor-to-normal tissue ratio, and drug clearance in individual patients. 2.2
Absolute Measurements
In order to obtain the true value of the tissue concentration, C, the effects of the other factors must be taken into account. The most important of these is the effect of the tissue optical absorption coefficients, /Aa(Acx) and /A a (A em ), and transport scattering coefficients, /u-s(A ex ) and /u^(Aem), that attenuate the excitation and emission light. The general expression for the detected fluorescence signal, F*(Aex, Aem, r), measured at some point at radial separation, r, from an incident pencil excitation beam of intensity, I0, is F*(Aex, A em ) = Tr*(A ex x,
r')'E(A e m , r -
r')} dr'
(1)
where 17 and e are, respectively, the photosensitizer fluorescence quantum yield and extinction coefficient and (f> [W.crrT2] is the light fluence rate at point r' in the tissue. S is the sensitivity (responsivity) of the fluorescence collection/detection system, including the geometric efficiency for light entering the collection fiber (essentially given by the numerical aperture for a fiber in contact with the tissue, corrected for internal reflection at the tissuefair]-fiber interface) and the detector sensitivity at the emission wavelength. The term E(A cm , r — r') is the "escape function" that describes the probability that a fluorescent photon generated at point r' will reach the tissue surface (or measurement point within the tissue) at position r. If the photosensitizer concentration, C(r), is uniform, at least within the sampled tissue volume, then this can be taken outside the integral. The issues related to 17 and e have already been mentioned, and essentially these factors depend on the tissue microenvironment and drug binding. The strong dependence on the tissue optical absorption and scattering properties arises from the terms (f> and E. There have been a number of different approaches to handle the complexity of this dependence, as
540
Wilson et al.
discussed below. We have grouped these techniques into three broad categories: 1.
2.
3.
Calibration techniques, in which the fluorescence is measured as a function of known photosensitizer concentration in different tissues (or tissue-simulating phantoms) and used as a standard for comparison to determine the unknown concentration in vivo Model-dependent techniques in which attempts are made to determine the tissue optical properties and to apply a rigorous or approximate model of light propagation in tissue, essentially to derive the excitation fluence and fluorescence escape functions and, thereby, correct F* to yield F, and Model-independent methods, in which the measured fluorescence signal is (empirically) corrected by a separate measurement on the tissue (e.g., diffuse reflectance or tissue autofluorescence) or in which the excitation and detection geometry is selected to be independent of the optical properties.
There is only a very limited literature on each of these approaches, and each will be illustrated briefly. 2.3
Calibration Techniques
The simplest approach is to measure the photosensitizer fluorescence and then compare the signal with the fluorescence signal from tissue samples with known photosensitizer concentration and similar escape function (optical properties). An example of this type of measurement was that of Panjehpour et al. [1], who determined the correlation of photosensitizer fluorescence signal to the true in vivo concentration for several tissues, using phthalocyanine as the photosensitizer. A simple surface contact probe with a single excitation fiber surrounded by six fluorescence collection fibers was used for the measurements. The collection fibers delivered the collected light to a photodiode array spectrometer. The actual tissue concentration was determined by extraction assay (see below). For many tissues, a correlation was found between the measured fluorescence signal and the C value. However, the signal strength at constant C was strongly dependent on the tissue type because of different optical attenuation, as seen in Fig. 9a. In principle, these calibration curves can be used to estimate C, but this requires calibration for each tissue type. Durkin and Richards-Kortum [2] compared several methods of photosensitizer quantification, including two-flux Kubelka-Munk (KM) and partial least-squares (PLS) analysis. For the former, model-dependent method a derived transfer function was used to relate the fluorescence in dilute
18
o • * v
16 ro 141 c CD 12
liver tumor kidney spleen
o
4
6
CO
IS' CO
0 O
r:
64-
0
2
8
Concentration by Tissue Extraction (mg/kg)
(a)
Source (lx) F (AJ
R (AX)
1
—J Tissue — PR 2.5 O)
o liver ^ skin • muscle
2.0 03
QC
g = 1.0"ro "c o 0.5c o O 0.0 0 (b)
0.5
1.0
1.5
2,0
2.5
Concentration by Tissue Extraction
FIGURE 9 Correlation between measured fluorescence signal (F*) and the true concentration of photosensitizer, C, in different tissue for three different measurement techniques. Also shown are schematic representations of the principle of each technique, (a) Simple surface probe (Panjepour). (b) Fluorescence normalized to diffuse reflectance at fixed source-detector separations (Weersink). (c) Confocal probe (Pogue).
Wilson et al.
542
500
400-
liver muscle
o CO
g> 300 c/? 0
c 200J CD O
b 100 2
(c) FIGURE 9
4
Concentration by Tissue Extraction (mg/kg) Continued
solution to that in an optically turbid medium. PLS is one of a number of multivariant statistical techniques that are used widely in spectroscopic analysis, and here is considered as a "black box" calibration technique. It requires simulated or measured training data sets (here, fluorescence spectra) for a range of known C values. This set must encompass the range of possible variations in the spectra and expected photosensitizer concentrations. These two methods were compared in phantoms, where the prediction error was —5% for PLS and — 1 8 % for the KM model. To achieve this low prediction error in PLS, the input data set consisted of the combination of fluorescence spectra collected at two excitation wavelengths. The limitation of PLS is that building of an adequate training set requires an extensive set of measurements that encompasses all possible C values across the range of tissue types. This must be repeated for each photosensitizer. Furthermore, a wide spectral measurement range must be used to provide an adequate spectral "signature" in the tissues. Applying it to layered tissues or inhomoge-
Photodynamic Therapy Dosimetry
543
neous drug distributions may also present problems. Finally, although it can be very accurate, it provides no physical insight, so that it is difficult to assess what would be the effect of altering any of the measurement conditions. 2.4
Model-Dependent Techniques
To avoid the need for tissue-dependent calibration, assumptions can be made about the fluorescence escape function and excitation fluence distribution, based on a priori knowledge of the tissue optical properties. One approach is to collapse the escape function into 1-D by employing a broad-beam excitation with point collection and assume a single exponential dependence on depth. Another simplification is to assume that the optical properties at Aex and Aem are the same. Thus, Potter and Mang [3] postulated that the fluorescence signal was proportional to C*§2, where 8 is the penetration depth of light in the tissue at Aex = 630 nm. Separate measurements of 8 using diffuse reflectance were required using a second optical fiber probe. This was tested in vivo in an animal model with Photofrin. However, correcting the fluorescence signal by 82 still resulted in errors in the prediction of C. Profio et al. [4] used similar approximations but derived a relationship between the ratio of the fluorescence signal and the quantity (1 — y)/(l + y), where y is the diffuse reflectance. At Acx = 633 nm in tissue-simulating phantoms containing Photofrin, the prediction error for C was —20%, for a range of background absorption coefficients tissue/Ji and E, as summarized in earlier chapters. This requires some method of measuring the tissue optical properties /AU and /AS at both the excitation and emission wavelengths. This then must also be done noninvasively, such as by spatially or time-resolved diffuse reflectance spectroscopy [5]. This complete approach has not been reported. A compromise has been to incorporate the total diffuse reflectance spectrum, which is a measure of the transport aldebo and depends on /XU//AS, into the analysis of the measured fluorescence signal, as, for example, the work of Feld and colleagues [6,7], Patterson and Pogue [8], and Jacques and colleagues [9,10]. In most studies using this approach, the primary goal has been to determine the true (intrinsic) fluorescence spectral shape rather than absolute quantification of the fluorophore concentration. Feld's technique, explained in more detail in Chap. 4, performs well in recovering the intrinsic fluorescence shape, in comparison with Monte Carlo simulations and tests on ex vivo tissue samples. Patterson applied their fluorescence model to frequency domain measurements of liquid phantoms and was able to accurately predict phase, modulation, and DC fluorescence signals.
544
Wilson et al.
Gardener et al. [9] derived an analytical method for extracting the intrinsic fluorescence spectral shape and C value. A 1-D pathlength factor for the penetration of excitation light into the tissue and fluorescence escape was defined for a given geometry. This is a function of the diffuse reflectance, as measured at Acx and A em , as well as the penetration depth, <5. The method was tested on a set of tissue-simulating phantoms and yielded a prediction accuracy of about ±20%. The technique was limited to point excitation/area collection, with the source and collection detectors located far from the tissue surface. For other geometries, the pathlength factor needs to be reformulated. 2.5
Model-Independent Techniques
In these methods, no attempt is made to extract the tissue optical properties. Rather, (semi)empirical normalization factors are applied to compensate for differences in the optical properties between tissues, or the measurement geometry is made independent of these. Finlay et al. [11] were primarily interested in extracting the true intrinsic fluorescence spectra of PpIX and its photoproducts during PDT. They corrected the measured fluorescence spectrum, F*(A), by the measured diffuse reflectance spectrum, R(A), and the diffuse reflectance at the excitation wavelength. It was assumed that the light paths are the same for F and R at the same wavelength, so that the simple ratio F*(A)/R(A) recovers the intrinsic fluorescence spectrum shape, while the reflectance at the excitation wavelength corrects the signal intensity to recover C. The measurement probe consists of a ring of 24 fibers, with 8 fluorescence excitation fibers, 8 reflectance source fibers, and 8 detection fibers used to collect both F and R signals. This ring of fibers encircles a central fiber bundle to collect the diffuse reflectance at A ex . The true fluorescence spectrum was given by FT(A) =
R(A)
This functional relationship is a simplification of the relationship derived by Wu et al. [6] for recovery of the intrinsic spectral shape. The value of 0.8 for the power of the excitation reflectance signal was determined empirically. Measurements on scattering emulsion phantoms demonstrated that FT <* C and is insensitive to background tissue absorption. Absolute quantification requires only a further calibration on a single sample of known C. The probe worked well for the specified geometry, but the source-detector displacement (8 mm) may be too large for some applications, such as in endoscopy. Capulot and Mourant [12] use a contact probe in which excitation light was delivered by one optical fiber and the fluorescence and diffusely re-
Photodynamic Therapy Dosimetry
545
fleeted excitation light by a second fiber at radial distance p. They assumed that at small p the number of scattering events for both fluorescence and diffuse reflectance is the same. Consequently, the effect of tissue optical properties on the fluorescence signal can be reduced by normalizing to the reflection signal. Experiments were performed on scattering solutions over a range of collection distances, p = 0.24-3 mm, with a porphyrin fluorophore (excitation at 442 nm, fluorescence detection at 650-700 nm). Only a limited number of different absorption and scattering coefficients were used, so that the measurements were primarily made to check for linearity of the ratio with fluorophore concentration, which was best for p = 0.64 mm. Between solutions with different scattering properties, there was a 25% difference in the estimated value of C. Weersink et al. extended this approach by using two separate fibers for the fluorescence and diffuse reflectance signals [13]. Again, a contact probe was used with one fiber for delivery of excitation light, a second at PP to collect the fluorescence signal, and a third at pR to collect diffusely reflected excitation light. The ratio F(pp)/R(pR) was proportional to C in a given tissue, PF and PR were chosen to minimize the effects of tissue scattering and absorption, the optimal choices being pp = 0.65 mm and pR = 1.35 mm. This technique provided a prediction error of 15% over the ranges /A, = 0.5-2.0 and /Aa = 0.004-0.1 mm '. In vivo experiments on skin, muscle, and liver tissues demonstrated the utility of this technique. The fluorescence signal alone was strongly dependent on the tissue type, as expected, but using the ratio technique, a linear relationship was found between F/R and C (as determined by ex vivo assay), independent of the optical properties (Fig. 9b). As with many methods, external calibration on a phantom of known fluorophore concentration was required to provide the absolute C value, i.e., to give the slope of this curve. The technique samples an effective tissue volume of several cubic millimeters. The effect of layered tissue or inhomogeneous drug distribution are under investigation. Tissue autofluorescence has been investigated as a surrogate for the escape function by normalizing the photosensitizer fluorescence to the autofluorescence signal, assuming that this varies only with the optical properties between different tissues. This technique has typically used UV excitation to limit the optical penetration depth and hence the number of scattering events before the fluorescence escapes through the tissue surface. Doiron et al. [14] used a system designed specifically for the photosensitizer SnET2, with excitation at 442 nm (HeCd laser), with the autofluorescence and SnET2 fluorescence measured at 570 and 670 nm, respectively. The ratio auu,F*/psF* was compared with C determined by tissue extractions on several tissue types. In most cases, there was a strong correlation over the range C = 1-60 /tg/g, with prediction errors of about ±20%. However, the sensitiv-
546
Wilson et al.
ity at low concentrations (C < 0.1 yu-g/g) was limited by autofluorescence at the same wavelength as the photosensitizer fluorescence. Sinaasappel and Sterenborg [15] alleviated this problem by measuring the autofluorescence signal at the photosensitizer fluorescence band with a second excitation source that was off the photosensitizer excitation peak. They derived a function that relates C to the fluorescence signals such that the measured signal was independent of tissue optical properties and excitation/collection geometry. Four fluorescence signals were used, and the corrected signal takes the form Fj m * F; „
1 + aC
Y = —^ —* F,.m * F ,n 1 + bC
(3)
where i and j refer, respectively, to the excitation wavelengths on and off the photosensitizer excitation peak (405 and 435 nm for HpD) and m and n refer to emission wavelengths of the photosensitizer (>660 nm for HpD) and tissue autofluorescence (577 nm). A series of phantom measurements was used to simulate the detection of HpD fluorescence on an autofluorescence background. A significant improvement was found in the correlation between Y and C compared to F i-m and C for phantom solutions simulating weak autofluorescence. However, for phantoms simulating strong autofluorescence, the fluorescence signal was already strongly correlated with the concentration, and so the improvement using Y was not as appreciable. Also, Y was not independent of the strength of the background autofluorescence, with different calibration curves used for phantoms with different simulated autofluorescence signals. The signal Y was independent of background absorption at the excitation wavelengths (AY = ± 12% over 0.001 < /xa < 0.01 mm '). These autofluorescence-based techniques have a distinct advantage for imaging applications, since the processed signal is insensitive to the collection geometry and only limited spectral information is required. Measuring F* in a tissue volume in which there is minimal scattering can reduce the effects of varying tissue properties. Pogue et al. [16,17] used a confocal geometry to achieve this. This involved limiting the excitation and detection light to a very small subsurface volume. The fluorescence signals were found to be insensitive to local pigmentation (^J and nearly linear with C (Fig. 9c). However, there was still a strong dependence on the tissue scattering. A single, small-diameter delivery and collection fiber in contact with the tissue can be used to approximate the confocal geometry. However, this gives a very weak signal. Hence, multiple independent fibers can be used, separated by a distance large enough to avoid cross-talk (~0.51.0 mm) [17]. On phantoms with varying background absorption, the prediction error using a single large probe (1 mm diameter) and measuring only
Photodynamic Therapy Dosimetry
547
the fluorescence signal was 40%, while a multifiber probe (37 fibers each of 100 ^m diameter) had only a 10% prediction error. The fluorescence signal taken with the multifiber probes was still strongly dependent on the background scattering properties. However, taking the ratio of the fluorescence signal to reflectance of the excitation light reduced the prediction error to approximately 30% over the range of /u,s = 5-40 mm '. This technique measures C only near the probe tip, so that an interstitial probe has also been developed comprising 7 fibers with a total outside diameter of 0.3 mm. All of the techniques described above provide reasonable predictions within a limited set of tissue properties and measurement geometries. It is interesting to note that for accurately predicting C, calibration and modeling techniques have largely given way to model-independent techniques. Calibration techniques can provide excellent accuracy but require separate calibrations for each tissue type and measurement geometry. Modeling techniques have been very useful in extracting the true shape of the fluorescence spectrum. They also can provide some insight into the causes of the distortions in the fluorescence spectra. However, they have had limited success in accurately predicting C, due to errors introduced by approximations in the models and in estimates of the tissue optical properties, required as model inputs. When tissue optical properties are not explicitly used in the models, a separate measurement of the diffuse reflectance is required, introducing further measurement error and complicating the measurement process. The model-independent techniques are the easiest to use and the most accurate to date. They are geometry specific, with each geometry useful for particular applications, i.e., point surface probe measurements measuring tissue volumes, confocal techniques for measuring endothelial layers. All of these techniques require an external calibration to a sample with known C and approximately similar optical properties as the tissue samples being measured. 2.6
Layered Tissues
None of the above methods has been adapted to layered tissue. Recently, Farrell and colleagues [18,19] derived a diffusion theory model of spatially resolved fluorescence, F*(p), for two-layer tissues. This is an extension of earlier work on spatially resolved diffuse reflectance in homogeneous [20,21] and layered media [22,23]. The intention is to use the model to determine the concentration of photosensitizer localized in either the upper or the lower layer, or to derive the upper layer thickness, t. The diffusion theory predictions were compared to two-layer phantom measurements and Monte Carlo simulations. For photosensitizer localized in the top layer only, the shape of F*(p) was almost independent of t, but for drug in the low layer only, F*(p) varied significantly with the top layer thickness. In general, there was
548
Wilson et al.
excellent correspondence between the measured fluorescence, Monte Carlo simulations, and the diffusion model. However, to date, the evaluations have been done only for forward calculations, i.e., for predicting F(p) knowing Cupper* Ciowen and t. The more challenging inverse problem has not been reported. 2.7
Diffuse Reflectance Spectroscopy
Diffuse reflectance measurements [20-27] have also been used to measure in vivo concentrations, with the advantage that they can be used for photosensitizers wth little or no fluorescence. Spatially resolved diffuse reflectance spectroscopy [20,21,24] explicitly estimates the scattering and absorption spectra of the tissue, and C can then be estimated from the absorption spectrum, subtracting the tissue background spectrum. Layered tissue can affect the accuracy of this prediction, but light propagation models that account for the layering can improve the prediction of C in the lower layer [22,23]. An additional advantage of what is, in effect, in vivo absorption spectroscopy of the photosensitizer, is that it allows tracking of effects such as spectral changes due to disaggregation of the drug in vivo, which is very important for optimizing the drug-light time interval and the treatment wavelength. Determining these by fluorescence is not very straightforward because of the other factors that affect the fluorescence properties in vivo. Thus, reflectance spectroscopic results may be used to complement fluorescence information. In another diffuse reflectance technique, the source/collector fiber separation, p, may be chosen so that the pathlength of the diffusely reflected photons through the tissue is constant across a wide range of scattering values [25-27]. With such a constant optical pathlength, C can be determined from changes in the absorption spectrum seen as the photosensitizer is taken up in the tissue.
3.
PHOTOSENSITIZER PHOTOBLEACHING FOR PDT DOSIMETRY
The determination of the effective "dose" in PDT is complex in comparison with, say, radiation therapy because of the strong dependence on the local concentration and microdistribution of the photosensitizer, on the local light fluence and fluence rate, and on the local tissue oxygenation [28]. In addition, these factors are interdependent and may change dynamically during treatment. Thus, for example, a high initial concentration of photosensitizer increases the optical attenuation, thereby altering the light distribution. Through photobleaching, the photosensitizer absorption coefficient decreases
Photodynamic Therapy Dosimetry
549
during irradiation in a nonuniform way, with the greatest effect being closest to the tissue surface (external irradiation) or light source (interstitial). This then shifts the light distribution toward that in the absence of photosensitizer. The photosensitizer photobleaching may be oxygen and/or singlet oxygen dependent [29], and the local pO2 may change during treatment due to rapid vascular response of the tissue and/or photochemical depletion if the rate of conversion from 3O2 to 'O2 exceeds the replenishment rate through perfusion [30]. This conversion rate depends, in turn, on the local drug-light product, Much of the development of dosimetry techniques and devices for PDT has focused on making separate measurements of these factors, as in the case of measuring C discussed above. The measured or calculated values of C, <£, and pO2 are then combined through a photophysical/photobiological model. One alternative to this is to measure the 'O2 generation directly. In solution, this is commonly done by measuring its near-IR (1270 nm) luminescence. In tissue this has proved elusive because singlet oxygen is so reactive that its lifetime is greatly shortened in biological media, making the luminescence signal very weak. Recently, an extended near-IR PMT has become available, and we have achieved successful measurements of 'O2 luminescence from PDT drugs in cells and in vivo [31]. A technologically simpler and cheaper alternative at present is to monitor the photobleaching of the photosensitizer. The use of photobleaching as a dose metric is based on the idea that, as the photosensitizer goes through each photoactivation cycle (as per Fig. 1), there is a probability that it will be "destroyed," either directly or by indirect action of the singlet oxygen. The greater the level of photoactivation, the greater the photobleaching. It is then assumed that this also reflects a greater singlet oxygen production and, thereby, greater photodynamic effect. Four main issues arise. First, if the photosensitizer fluorescence is used to monitor the photobleaching, then the same problems of making quantitative noninvasive measurements are encountered as discussed above for determination of C. The photobleaching rate, (3 [cm2.J '], can be obtained from a series of relative measurements during irradiation, but to determine the absolute number of photosensitizer molecules photobleached per unit volume the absolute value of the concentration at the start of treatment, C0, is required. For firstorder kinetics, the concentration after a total fluence of O is given by
Coe-
(4)
In the case of indirect photobleaching, such as by 'O2, the kinetics are more
550
Wilson et al.
complex (second order if there is no other photobleaching mechanism), since the rate of 'O2 production depends on the instantaneous value of C. Thus: C(4>) = C () [1 + C () /3<£J '
(5)
Second, depending on the molecular changes to the photosensitizer, there may be changes in the fluorescence without an exactly corresponding change in the photodynamic efficacy of the drug. An example of this is where the photoproducts may be photosensitizing but have a different fluorescence spectrum from the parent photosensitizer, as in the case of ALAgenerated PplX where the original main single emission peak around 630 and 710 nm decreases with fluence but a new peak around 675 nm appears, indicating the appearance of a second photoactive compound [11]. Hence, simply monitoring the 700 nm fluorescence signal underestimates the amount of photoactive drug present and so underestimates the effective PDT dose. If these effects are known for a particular photosensitizer, as determined in solution or in preclinical cell or animal models, then they can be taken into account, although it may be necessary to monitor the fluorescence in multiple-wavelength bands or across the full emission spectrum to obtain adequate data on photobleaching and photoproducts. Third, and most difficult, the interpretation of the photobleaching as a unique reporter of the PDT dose may be an oversimplification. For example, with m-THPC it has been shown even in solution that not only is the photobleaching oxygen (and singlet oxygen) dependent but that this dependence changes with the tissue microenvironment and, especially, with binding of the photosensitizer to biomolecules such as proteins [29]. Thus, the photobleaching measurement depends on the microlocalization of the drug, which of course profoundly affects the photobiological damage. As a result, the photobleaching may by itself not provide sufficient information, and simultaneous measurement of the other factors, such as the local changes in pO2, may also be required. There is little information on this aspect of photobleaching for most of the PDT drugs under investigation. A final issue is that noninvasive or minimally invasive measurements of fluorescence will at best provide a measure of the local value of C averaged over some tissue volume. In practice, with the possible exception of the use of confocal microimaging in vivo, the microdistribution of drug will not be known. It has been observed that the variability in the microdistribution within a single tumor can be comparable to the variation in the volume value of C between subjects [32]. However, it has not been established, other than in specific instances, what, if any, is the relationship between the PDT response and the volume-average value of C. In some cases, the relationship seems to hold, but this may not be generally true. For example, in
Photodynamic Therapy Dosimetry
551
drugs that act primarily on the tissue vasculature, it may be the fraction in the vessels or bound to the vascular endothelial cells that determines outcome, not the total tissue content of photosensitizer. These complicating factors do not mean that photobleaching is of no value in PDT dosimetry but that much more detail work is required to understand the role of the effects in practice. For such studies to be possible, continuing development of minimally invasive techniques and instrumentation for monitoring the micro- and macrokinetics of photobleaching in vivo is essential, as are photophysical and photobiological models of the effects. Even if photobleaching proves to be of limited value as a quantitative measure of the PDT dose, it has a useful role in showing the uniformity and completeness of irradiation. For this purpose, fluorescence imaging is most appropriate, and Fig. 10 illustrates this for two different clinical situations. As discussed earlier, Farrell and colleagues [18,19] have developed a diffusion model of spatially resolved fluorescence for layered tissue. The model can be used to determine the depth of photobleaching in a nonlayered tissue with an initial uniform photosensitizer distribution. As light irradiation progresses, the photosensitizer is differentially photobleached in the top "layer" of tissue, and so the effective thickness of this layer increases. Changes in F*(p) can then be used to derive the effective thickness and, thereby, the depth of photobleaching. If the degree of photobleaching corresponds to depth of necrosis, then it provides a simple means of determining this. As an alternative dosimetry technique, it has been proposed [33] to use photobleaching of the tissue autofluorescence as a dose metric rather
FIGURE 10 Imaging of photosensitizer photobleaching during clinical PDT with Photofrin. (a) In esophageal tumor. (Courtesy of Dr. N. Marcon, Toronto, Canada.) (b) In the surgical resection in (residual) malignant brain tumor. (Courtesy of Dr. P. Muller and V. Yang, Toronto., Canada.) (See color insert.)
552
Wilson et al.
than that of the photosensitizer itself. This is based on the molecular or metabolic changes caused by photodamage. The specific endogenous fluorophore studied to date is NADH in RIF1 cells in vitro treated with BPD and in vivo on normal muscle tissue in a mouse model treated with BPD. The NADH fluorescence decreased relative to the total autofluorescence signal as treatment progressed. The NADH fluorescence lifetime did not change, indicating that the decreasing signal was due to a change in the concentration of NADH rather than a change in the nonradiative processes of the NADH. The authors speculate that the decreasing NADH concentration is a byproduct of the loss of mitochondrial function during treatment, which disrupts the production of NADH. The relationship between the change in NADH and the tissue response (e.g., necrosis) has yet to be established and is likely to be complex and tissue dependent. As in the case of measuring C, diffuse reflectance spectroscopy can be used for photobleaching measurements as an alternative or supplement to fluorescence photobleaching. However, the interpretation of the results may not be identical, since the mechanisms that cause reduced fluorescence are not necessarily identical to those that reduce the photosensitizer absorbance, depending on the particular photosensitizer. 4.
PHOTOSENSITIZER QUANTIFICATION IN TISSUE OR CELL SAMPLES
In preclinical studies in animal models and clinical treatments, it is often useful to measure the photosensitizer concentration in tissue samples (e.g., biopsies). This information may then be used for several purposes: (1) to determine the pharmacokinetics in individual cases where in vivo monitoring is not feasible for either technical reasons or clinical impracticality but where sequential tissue samples can be taken; (2) to give absolute normalization of sequential relative in vivo measurements, e.g., by making a series of measurements at different times (and locations) following administration and then taking a single tissue sample at the end, from which the absolute concentration can be determined and applied to the whole data set; and (3) to give the absolute value of C at the time of treatment as one of the independent outcome-determining factors in PDT dosimetry (and possibly during treatment to monitor photobleaching). For in vitro experiments, corresponding population-average measurements of the intracellular concentration of drug can be made from cell pellets obtained by centrifugation in order to construct dose-response curves. The methods below are essentially identical for cell pellets and tissue samples of the same weight. Several methods have been reported for such quantification, including absorption spectroscopy, (spectro)fluorimetry, high-pressure liquid chroma-
Photodynamic Therapy Dosimetry
553
tography, and radioscintigraphy. The essential factors are to ensure that the whole photosensitizer content of the tissue sample is assayed and that the assay technique indicates the relevant quantity in vivo that determines the PDT efficacy. The photosensitizer can be extracted from the tissue using multistage chemical processing with alcohols, ethers, or alkaline reagents. The challenge is to ensure that all, or at least a known and reproducible fraction, of the photosensitizer is extracted without chemical destruction of the photosensitizer. Alternatively, the tissue can be "solubilized," i.e., mechanically and enzymatically disaggregated into a colloidal suspension [34]. This ensures that there is no loss of material. The enzymatic digestion can also destroy the photosensitizer. The rate of destruction is different for each drug but predictable for most photosensitizers, so that the assay can be reliable if the timing and temperature of the solubilization are well controlled. Radiolabeling is the most sensitive assay, but it requires that a radiolabeled version of the photosensitizer is available that has the same pharmacokinetics as the unlabeled drug (usually the case if 3H or I4C labeling is done), that there is no free label, and that the labeling is stable in vivo. HPLC is particularly useful either if the photosensitizer is nonfluorescent or if it comprises multiple molecular species with potentially different kinetics and binding. Spectrofluorimetry has high sensitivity and can be used either with extracted photosensitizer or on solubilized tissue. Spectral scanning is preferable to simply measuring at a single emission peak in order to subtract the autofluorescence background. The calibration factor relating the signal to the true concentration must be determined independently for each tissue type and photosensitizer. This can be done by referencing against another calibrated technique such as radiolabeling, by generating a "standard curve" in which known concentrations of the photosensitizer are added to homogenized tissue and then extracted or solubilized, or by "spiking" the extracted or solubilized material with a known quantity of photosensitizer. The main factors to consider are (1) how best to reproduce the in vivo photosensitzer binding in the tissue, especially for drugs that have multiple components, and (2) how to account for photosensitizer aggregation that alters uptake and fluorescence yield. The lower detection limit depends on the mass of the tissue sample, the tissue type (because of the autofluorescence background and the loss of fluorescence signal due to light absorption), and the photosensitizer fluorescence yield and solubilization stability. For example, with 2-mg samples, concentrations as low as 0.01 jjug per g tissue can be reliably measured for Photofrin in brain, whereas concentrations of 2 ^tg/g are required for benzoporphyrin derivative in (highly pigmented) liver. For Photofrin, these values are about 5- to 10-fold lower if ether extraction is
Wilson et al.
554
used. These spectroflourimetric methods have excellent linearity over several orders of magnitude, as illustrated in Fig. 11. For nonfluorescent photosensitizers, absorption spectroscopy may be used as an optical assay method on the extracted or solubilized samples. In this case, the lower limit of the measurement is typically an order of magnitude higher than for fluorescence, or worse for highly pigmented tissues, due to the intrinsic tissue absorption background. The assay is most sensitive for photosensitizers with a strong absorption peak in the far-red/near-IR range, where the hemoglobin absorption is lower than at shorter wavelengths. 5.
FLUORESCENCE MICROSCOPY OF PHOTOSENSITIZER MICRODISTRIBUTIONS IN TISSUE OR CELL SAMPLES
Fluorescence microscopy has been widely used to study the microdistribution of PDT photosensitizers in cells and tissues. Qualitative imaging may be determined using conventional (epi)fluorescence microscopy, in which the whole specimen within the microscope field of view is excited and imaged simultaneously. More quantitative data can be obtained by confocal fluorescence microscopy (CFM), in which a laser beam excites the sample one point at a time and the image is formed by raster-scanning of the beam
1001
^c o 10
g -c 'ro a .t ^ c <ji
8 s 1 C CT3 O ^
O O
0.1(0
i-
0.01
0.01
0.1 1 10 100 Tissue Concentration (ug/g): Acid Extraction Method
FIGURE 11 Photosensitizer concentration in different tissues as measured by spectrofluorimetry after solubilization versus true concentration. (Adapted from Ref. 34.)
Photodynamic Therapy Dosimetry
555
FIGURE 12 Confocal fluorescence micrographs of ALA-induced PplX in gastrointestinal tissues, showing the specific microlocalization. (Courtesy of R. DaCosta and Dr. N. Marcon, Toronto, Canada.) (See color insert.)
(and or sample stage): Figure 12 shows examples of CFM in biopsied tissue samples. The signal is then proportional to the number of photosensitizer molecules within the excitation/detection voxel, apart from small background signals from fluorescence excited and detected above and below the focal plane and the contribution from autofluorescence. It is possible, in
556
Wilson et al.
principle, to make absolute measurements of the photosensitizer concentration distribution by calibrating the instrument with samples of similar optical absorption and scattering and containing known photosensitizer concentration, or by normalizing the images to the total concentration measured in a bulk sample of the tissue or call pellet (as above). However, this has not been reported much to date, and no reports achieving better than a factor of about 2 in accuracy are available. Care must be taken in handling the tissue or cell samples to avoid loss or redistribution of the photosensitizer and photobleaching. The latter can occur prior to imaging (commonly during sample sectioning from ambient light) or during the imaging from exposure to the microscope excitation light. Multiphoton CFM significantly reduces the photobleaching during imaging. 6.
FLUORESCENT OPTICAL FIBER FLUENCE RATE SENSORS
As discussed previously, the PDT efficacy is determined by the localized photosensitizer concentration, the availability of molecular oxygen, and the light fluence rate at the treatment wavelength. While reflectance probes can be used to ascertain the fluence rate close to the tissue surface, minimally invasive probes, such as optical fibers that can be inserted into the tissue, are often used. These probes need to have an isotropic response to correctly measure in an anisotropic light field. Isotropically responsing fibers were initially achieved by using a small spherical volume of a highly light-scattering material at the fiber tip. Multiple light scattering in this tip made the response, i.e., the fraction of the light that was transmitted along the fiber to a photodetector, largely independent of the direction of incidence of the light on the tip, except in the backward direction where the tip was attached to the fiber. However, these scattering probes must be at least 0.5 mm in diameter to achieve reasonable isotropy (±10-20%), since a minimal number of scattering events is required [35]. Fluorescent-based sensors, in which the scattering tip is replaced by a small volume of fluorophore (Fig. 13a), do not have this physical limitation because the fluorescence is intrinsically isotropic. Hence, if the shape of the fluorescent tip is optimized for isotropic response, scaling down in size does not change this but simply reduces the signal strength [36]. The principle is shown in Fig. 13a. Either organic fluorophores or solid-state crystals have been used in these probes. The latter have the advantage of no photobleaching, so that the responsivity (detected light/incident fluence) is independent of total fluence exposure. However, there are problems with production and mechanical stability of crystal probes. Some in vivo mea-
557
Photodynamic Therapy Dosimetry
0.!
•3
I.
0.01
CO
cp to
c 0.001
0 (b)
10 u eff
20
30
40
.-ii [cm'
FIGURE 13 Fluorescent optical fiber light fluence probes, (a) Principle showing single (upper) and multiple (lower) point sensors, (b) Investigational volume as a function of tissue optical properties. (Courtesy of N. Pomerleau, Toronto, Canada.)
surements have been performed [37,38] using a ruby microsphere, and other fluorescent crystals have been suggested [39]. These solid-state-based sensors also have fairly low fluorescent quantum yield (typically —5%) and, hence, low responsivity compared with sensors based on organic fluorophores. Various versions of fluence-rate sensors based on incorporating an organic fluorophore into a nonfluorescent matrix material have been proposed [36,39] and used in preclinical studies [37,40,41]. The fluorophore is selected to maximize the response, which is proportional to the product of its absorption coefficient at the PDT treatment wavelength and its fluorescent
558
Wilson et al.
quantum yield. The fluorescence signal is measured, after band pass filtering to block the scattered treatment light, by a standard diode and preamplifier. The matrix material is typically a polymer bonded to the distal end of the fiber, e.g., an epoxy, that is then machined to the correct shape. Monte Carlo simulations have been used to show that the physical dimensions (—200300 /xm diameter, 500 /urn length) of the fibers are much smaller than the effective ("investigational") tissue volume within which the light fluence is sampled, which depends on the tissue optical properties as shown in Fig. 13b. The responsivity of these sensors is typically —10 6 cm 2 . For fluoropores that are photobleached, the responsivity decreases with total fluence exposure, and so this must be monitored to maintain the probes in absolute calibration. Typically, the loss of signal is <5% per 100 J cm"2 exposure. The calibration can be carried out in a known light field: either in a parallel beam of known irradiance or in an integrating sphere. Since the investigational volumes of these probes have a finite size, their accuracy is biased depending on the local light fluence gradient in the tissue. The magnitude of this effect is dependent on the tissue optical properties and irradiation geometry. Thus, bias will occur if the probe is calibrated in an isotropic light field (e.g., in an integrating sphere). If the calibration is performed in an anisotropic fluence rate field, such as a liquid phantom with scattering and absorption coefficients similar to the tissue to be measured, then no bias is introduced. A recent development in this technology is a probe capable of measuring the fluence rate simultaneously at multiple points in the tissue. This is achieved by placing several sensors along the length of the optical fiber, each with a different fluorophore. The fluorophores are selected to be excited by the same (PDT treatment) wavelength, but to have distinctly different emission spectra. The total spectrum is then the sum of the individual spectra, each weighted by the local light fluence. This can be deconvolved, knowing the individual spectra, to obtain the separate fluence values. Thus, the fluence measurement position is "encoded" by the fluorescence spectra. A major advantage of these probes is that they enable volume surveillance of the fluence rate during PDT without the need to step the sensors through the tissue. The probes may be optimized by placing the fluorophore sensors with the highest responsivity at the positions of lowest anticipated fluence rate in the tissue in order to approximately equalize the signals from each sensor. An extension of this concept that is currently being investigated is to incorporate near-IR phosphors instead of (or in addition to) the fluorophores. The spectral deconvolution would be as for the fluence sensors, but now the phosphorescence lifetime is measured in each case. This determines the local
Photodynamic Therapy Dosimetry
559
tissue pO2 value, since oxygen quenches the phosphorescence and so reduces the lifetime in a deterministic way.
REFERENCES 1.
2.
3. 4. 5. 6. 7.
8.
9.
10.
11.
12.
13.
14.
15.
M Panjehpour, RE Sneed, DL Frazier, MA Barnhill, SF O'Brien, W Harb, BF Overholt. Quantification of phthalocyanine concentration in rat tissue using laser-induced fluorescence spectroscopy. Lasers Surg Med 13:23-30, 1993. AJ Durkin, R Richards-Kortum. Comparison of methods to determine chromophore concentrations from fluorescence spectra of turbid samples. Lasers Surg Med 19:75-89, 1996. WR Potter, TS Mang. Photofrin II levels by in vivo fluorescence photometry. Prog Clin Biol Res 170:177-186, 1984. AE Profio, S-S Xie, K-H Shu. Diagnosis of tumors by fluorescence: quantification of photosensitizer concentration. Proc SPIE 1203:12-18, 1990. MS Patterson. In: AJ Welch, MJC van Gemert, eds. Optical-Thermal Response of Laser-Irradiated Tissue. New York: Plenum Press, 1995, pp 333-364. JF Wu, S. Feld, P. Rava. Analytical model for extracting intrinsic fluorescence in turbid media. Appl Opt 32:3585-3595, 1993. MG Miiller, I Georgakoudi, Q Zhang, J Wu, MS Feld. Intrinsic fluorescence spectroscopy in turbid media: disentangling effects of scattering and absorption. Appl Opt 40:4633-4646, 2001. MS Patterson, BC Wilson. Mathematical model for time-resolved and frequency-domain fluorescence spectroscopy in biological tissues. Appl Opt 33: 1963-1973, 1994. CM Gardner, SL Jacques, AJ Welch. Fluorescence spectroscopy of tissue: recovery of intrinsic fluorescence from measured fluorescence. Appl Opt 35: 1780-1792, 1996. SL Jacques. Light distribution from point, line and plane sources for photochemical reactions and fluorescence in turbid biological tissues. Photochem Photobiol 67:23-32, 1998. JC Finlay, DL Conover, EL Hull, TH Foster. Porphyrin bleaching and PDTinduced spectral changes are irradiance dependent in ALA-sensitized normal rat skin in vivo. Photochem Photobiol 73:54-63, 2001. M Canpolat, JR Mourant. Monitoring photosensitizer concentration by use of a fiberoptic probe with a small source-detector separation. Appl Opt 39:65086514, 2000. RA Weersink, MS Patterson, K Diamond, S Silver, N Padgett. Noninvasive measurement of fluorophore concentration in turbid media using a simple fluorescence/reflectance ratio technique. Appl Opt 40:6389-6394, 2001. DR Doiron, JB Dunn, WL Mitchell, BK Dalton, GM Garbo, JA Warner. Fiber optic based fluorescence detection system for in vivo studies of exogenous chromophore pharmacokinetics. Proc SPIE 2396:312-322, 1995. M Sinaasappel, JCM Stenenborg. Quantification of the hematoporphyrin de-
560
16. 17. 18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
Wilson et al. rivative by fluorescence measurements using dual-wavelength excitation and dual-wavelength detection. Appl Opt 32:541-548, 1993. BW Pogue, T Hasan. Fluorophore quantification in tissue-simulating media with confocal detection. IEEE J Sel Top Quant Electron 2:959-963, 1996. BW Pogue, G Burke. Fiber-optic bundle design for quantitative fluorescence measurement from tissue. Appl Opt 37:7429-7436, 1998. TJ Farrell, RP Hawkes, MS Patterson, BC Wilson. Modeling of photosensitizer fluorescence emission and photobleaching for photodynamic therapy dosimetry. Appl Opt 37:7168-7183, 1998. DE Hyde, TJ Farrell, MS Patterson, BC Wilson. A diffusion theory model of spatially resolved fluorescence from depth-dependent fluorophore concentrations. Phys Med Biol 46:369-383, 2001. BC Wilson, TJ Farrell, MS Patterson. An optical fiber-based diffuse reflectance spectrometer for non-invasive investigation of photodynamic sensitizers in vivo. Proc SPIE 219-231, 1990. TJ Farrell, MS Patterson, B Wilson. A diffusion theory model of spatially resolved, steady-state diffuse reflectance for the noninvasive determination of tissue optical properties in vivo. Med Phys 19:879-888, 1992. G Alexandrakis, TJ Farrell, MS Patterson. Monte Carlo diffusion hybrid model for photon migration in a two-layer turbid medium in the frequency domain. Appl Opt 39:2235-2244, 2000. G Alexandrakis, TJ Farrell, MS Patterson. Accuracy of the diffusion approximation in determining the optical properties of a two-layer turbid medium. Appl Opt 37:7401-7409, 1998. RA Weersink, JE Hayward, KR Diamond, MS Patterson. Accuracy of noninvasive in vivo measurements of photosensitizer uptake based on a diffusion model of reflectance spectroscopy. Photochem Photobiol 66:326-335, 1997. JR Mourant, TM Johnson, G Los, IJ Bigio. Non-invasive measurement of chemotherapy drug concentrations in tissue: preliminary demonstrations of in vivo measurements. Phys Med Biol 44:1397-1417, 1999. JR Mourant, IJ Bigio, DA Jack, TM Johnson, HD Miller. Measuring absorption coefficients in small volumes of highly scattering media: source-detector separations for which path lengths do not depend on scattering properties. Appl Opt 36:5655-5661, 1997. JR Mourant, IJ Bigio, DA Jack, TM Johnson, HD Miller. Measurement of absorption coefficients in small volumes of highly scattering media. Proc SPIE 2979:294-299, 1997. BC Wilson, MS Patterson, L Lilge. Review Article: Implicit and explicit dosimetry in photodynamic therapy: a new paradigm. Lasers Med Sci 12:182199, 1997. BW Mcllroy. TS Mann, JS Dysart, BC Wilson. The effects of oxygenation and photosensitizer substrate binduing on the use of fluorescence photobleaching as a dose metric for photodynamic therapy. J Vib Spec 28:25-35, 2002. S Mitra. TH Foster. Photochemical oxygen consumption sensitized by a porphyrin phosphorescent probe in two model systems. Biophys J 78:2597-2605, 2000.
Photodynamic Therapy Dosimetry 31.
32.
33.
34.
35. 36. 37.
38.
39.
40.
41.
561
M Niedre, MS Patterson, BC Wilson. Direct near-infrared luminescence detection of singlet oxygen generated by photodynamic therapy in cells in vitro and tissues in vivo. Photochem Photobiol 75:382-391, 2002. BW Pogue, RD Braun, JL Lanzen, C Erickson, MW Dewhirst. Analysis of the heterogeneity of pO2 dynamics during photodynamic therapy with verteporfin. Photochem Photobiol 74:700-706, 2001. BW Pogue, JD Pitts, M-A Mycek, RD Sloboda, C Wilmot, M., JF Brandsema, JA O'Hara. In vivo NADH fluorescence monitoring as an assay for cellular damage in photodynamic therapy. Photochem Photobiol 74:817-824, 2001. L Lilge, C O'Carroll, BC Wilson. A solubilization technique for photosensitizer quantification in ex vivo tissue samples. J Photochem Photobiol B 39:229235, 1997. W Star. Light dosimetry in vivo. Phys Med Biol 42:763-787, 1997. L Lilge, T Haw, BC Wilson. Miniature isotropic optical fibre probes for quantitative light dosimetry in tissue. Phys Med Biol 38:215-230, 1993. L Lilge, K Molpus, T Hasan, BC Wilson. Light dosimetry for intraperitoneal photodynamic therapy in a murine xenograft model of human epithelial ovarian carcinoma. Photochem Photobiol 68:281-288, 1998. R Bays, G Wagnieres, D Robert, D Braichotte, JF Savary, P Monnier, H van den Bergh. Light dosimetry for photodynamic therapy in the esophagus. Lasers Surg Med 20:290-303, 1997. L Lilge, G Merberg, R Dacosta, BC Wilson. Nd3+ doped glass fluorescenttip fiber optical probes for quantitative fluence rate dosimetry in biological tissue. J Soc Opt Eng 2131:145-154, 1994. L Lilge, BC Wilson. Photodynamic therapy of intracranial tissues: A preclinical comparative study of four different photosensitizers. J Clin Laser Med Surg 16:81-91, 1998. L Lilge, MC Olivo, SW Schatz, JA MaGuire, MS Patterson, BC Wilson. The sensitivity of normal brain and intracranially implanted VX2 tumour to interstitial photodynamic therapy. Br J Cancer 73:332-343, 1996.
16 Controlled Drug Delivery in Photodynamic Therapy and Fluorescence-Based Diagnosis of Cancer Norbert Lange Universite de Lausanne, Lausanne, Switzerland
1.
INTRODUCTION
In 1998, the American Cancer Society estimated that over half a million Americans were expected to die from cancer. Nearly one-quarter of diseaseassociated mortality is related to cancer, making it the number two killer disease of Americans. With 800,000 cases, skin cancer is the most frequently diagnosed, including 40,000 cases of advanced malignant melanoma. Lung and colorectal cancer, followed by breast and uterine cancer in women and prostate cancer in men, are the leading cause of death in the United States. Statistically, in the United States one in four people will have some form of cancer at some stage in their lifetime. Thus, efficient management of these diseases remains one of the most challenging aspects of oncology care to date. 1.1 Conventional Treatment Strategies for Neoplastic Disease Conventional treatment strategies in oncology-related fields range from surgery to radiation therapy to chemotherapy. Each of these strategies has lim563
564
Lange
itations in providing the final goal: complete cure or at least satisfactory management of the disease that does not impede the quality of life. The greatest chance for complete cure is offered when therapy is performed at an early stage of disease progression. Surgical resection is often limited by the ability to identify and remove entirely the tumor and neighboring already transformed tissue areas. These remaining neoplastic cells may proliferate again and cause recurrence. Furthermore, surgery is often not effective against metastatic tumors that may have migrated from the site of the primary tumor. In some cases, effective removal of sufficient quantities of the surrounding healthy tissue ensuring the removal of all cancerous cells is limited. In the surgical removal of glioblastomas, for example, the surgeon would probably not be as aggressive as necessary in order to maintain as many functions as possible. Chemotherapy based on the systemic or regional administration of cytotoxic agents is often impeded by the lack of specificity of the drug for neoplastic tissue. Therefore, this therapy is often limited by systemic toxicity before truly therapeutic drug levels in the tumor can be achieved. Because chemotherapeutic drugs act on rapidly dividing cells, fast-growing cells of the intestinal lining and bone marrow can also be extensively damaged during treatment. Tumor cell heterogeneity and inhomogeneous blood vessel distribution in the tumor mass may result in cancerous cells that become resistant to a particular chemotherapeutic drug, leading to single- or multiple-drug resistance. On the other hand, radiation therapy can be specifically directed to the side of the tumor. It is based on the idea that noncancerous cells divide more slowly and have a better chance of DNA repair. However, this technique is limited by the potential damage to neighboring healthy tissue and, like surgery, depends on imaging devices allowing identification of the tumor site. None of these conventional therapies, whether alone or in combination, has provided complete cures for all types of cancer in all patients. 1.2
Photodynamic Therapy and Fluorescence Photodetection
An attractive and innovative alternative to these conventional therapies of cancer is based on the systemic or topical administration of a photoactive compound, which preferentially accumulates in neoplastic tissue. The interaction with light of an appropriate wavelength, results either in clear fluorescence demarcation or in destruction of tissue laden with the photoactive compound. Depending on the irradiation conditions and the properties of the photoactive substance, these modalities can be referred to as fluorescence photodetection (PD) [I] or photodynamic therapy (PDT) |2] (see Fig. 1),
Controlled Drug Delivery in Cancer
565
FIGURE 1 Schematic presentation of the principle of photodynamic therapy (PDT) and photodetection (PD). After topical or systemic administration of a photosensitizer, the dye accumulates preferentially within the tumor tissue. Upon illumination with light of an appropriate wavelength, the interaction of light with the fluorophore results in either the clear demarcation of tissue containing high amounts of the dye, or the initiation of processes that cause its destruction.
566
Lange
respectively. The recent regulatory approvals of Photofrin, benzoporphyrin derivative monoacid ring A (BPD-MA), and 5-aminolevulinic acid (ALA) as photosensitizing agents for the palliative treatment of cancers of the esophagus [3], the treatment of choroidal neovascularization (CNV) associated with age-related macular degeneration (AMD) [4,5], and the PDT of actinic keratosis (AK) [6,7] finally provided a broad acceptance of these methodologies in the medical community. On a molecular level, both PDT and PD are initiated by the excitation of the photoactive molecule to a short-lived, excited singlet state due to absorption of visible or near-infrared (near-IR) photons. Following excitation, fast radiationless relaxation processes populate the lowest excited singlet state S, (Fig. 2) of the molecule, from which it can return to its ground state S0 by radiationless processes or by emission a photon. In PD procedures this fluorescence is used to visualize diseased tissue areas with high amounts of the fluorophore using highly sensitive fluorescence imaging devices. Apart from fluorescence and radiationless deactivation, during its relaxation to the ground state, the excited molecule can undergo so-called intersystem crossing, ultimately yielding the lowest excited triplet state T,. The slower the decay of this triplet state, the more time the substance has to act on its environment via two principle mechanisms: 1.
2.
Type I photochemistry involves electron transfer between the photosensitizer and a nearby molecule, e.g., a membrane lipid or oxygen. After this redox reaction, the resultant substrate radical may react to generate peroxy radicals, subsequently undergoing radical chain reactions, or, in the case of oxygen, a reaction partner directly forms superoxide (O 2 ). In type II photochemistry, energy transfer occurs directly between the photosensitizer in its triplet state and ground state molecular oxygen, forming highly reactive singlet oxygen ('O 2 ).
Subsequently, both photochemical processes can trigger cascades of biochemical, biophysical, immunological, and physiological reactions, finally resulting in destruction of the irradiated tissue. The therapeutic value of PDT is influenced by complex interactions of multiple parameters such as photosensitizer concentration in tissue and its localization, photosensitizer selectivity, irradiation wavelength, fluence rate, tissue optical properties, tissue metabolic conditions, delay between drug administration and irradiation, and many other factors. However, the management of tumors using PDT can cause tumor ablation within a few days. On the whole, three types of mechanisms may be involved in the rapid response to PDT. First, PDT may directly destroy the malignant cells by necrotic or apoptotic mechanisms. Second, PDT may profoundly change the morphology of the tumor vascu-
Controlled Drug Delivery in Cancer
567
o
PpIX
Energy Transfer
3 + Absorption
4
Fluorescence
PDT
Effects
•Tumor Cell Destruction •Vascular Damage •Immune and Inflammatory Effects
300
400
500
600
700
800
FIGURE 2 Basic illustration of the photophysical and photobiological process associated with PDT and PD illustrated on the basis of the absorption and fluorescence emission spectrum of protoporphyrin IX. (For detailed description, see text.)
568
Lange
lature, including blood vessel occlusion, blood vessel stasis, and/or vascular leakage. Third, PDT can cause a modulation of the immune system by the release of cytokines or other inflammatory mediators from the treated tissue. Depending on the photosensitizer, overall light dose, and experimental protocol, several cell structures have been identified as targets for PDT, among which are microtubules, membrane organelles, especially lysosomes and mitochondria, plasma membrane, and nucleus. Note that the absence of only one of the components involved in PDT (i.e., photosensitizer, light, or oxygen) does not cause any phototherapeutic effects. Thus, accumulation of a photosensitizer in body areas, distinct from the treatment site, e.g., liver or kidneys, will not impair their functionality. On the other hand, hypoxic tissues, such as the cartilage, will in turn be preserved. Although considerable effort has been made to introduce this promising treatment modality into many oncological and nononcological fields, PDT and PD suffer, like any other treatment, from several drawbacks. Major drawbacks in this kind of therapy are the lack of chemical homogeneity and stability of the photosensitizer, skin phototoxicity induced by an enhanced retention of the compound in the skin, unfavorable physicochemical properties, and low selectivity with respect to the uptake and retention in normal versus tumor tissue. The latter condition, i.e., high ratios of tumor to normal tissue for the photoactive agent, avoids or minimizes photodamage in the peritumoral tissue. In order to further increase the impact of PDT and PD on the management of cancer, qualitative improvements must be made in (1) the definition of PDT and PD protocols that yield optimal results and assessment of the potential of PDT to compete with existing treatment strategies; (2) expansion of the scope of PDT for the management of other pathologies, including rheumatoid arthritis, restenosis of arteries after angioplasty, psoriasis, warts, viral or microbial infections, and blood banking; (3) enhancement of the efficacy of PDT and PD with respect to the properties of the photosensitizing agent. Table 1 summarizes the ideal properties of such photoactive agents used for management and the improved diagnosis of malignant and nonmalignant diseases. Since the toxic effect is induced by the local activation of the photosensitizer, while normal tissue not exposed to the light is spared, PDT can be considered to be selective per se. However, due to the drawbacks related to conventional photosensitizers as mentioned above, one of the essential goals in PDT and PD in order to improve the clinical utility of these methods remains enhancement of the ratio between the concentration of the photoactive agent in the tumor and the tissue from which the tumor originated. Simple derivatization of mostly porphyrinic macrocycles into more or less lipophilic/hydrophilic molecules as compared to the lead compound with more or less amphiphilic properties has not led to a major breakthrough
C
o tj
c
-o
0
S fO
O CO
'+3 CO
S
0
2
fD
'>
>
c
CO
2
CD T3 O
D
CD T3 O
4-1
a?
w
^
CD
~ •«=
CD
4-J
X
CD
Q) C
4-J
O SI Q.
O
SI
CL T3 C CO
CD O C CD O
> a
w
CO
0)
L.
111!
o
CD
.c
_D
I-
LL
o E
O
>-> C/) "'
—
CO
c Q_ if
"a o •M
LL
5
<
O
SI
CL
Q. CD O O
-a CD C/5
Q) CO 3 CD
D
^ "co
Photodynamic he
4-"
c/>
.Q D CO CD > "4-"
O CO
O O
4-1
S^ t! C
Illli
^
1 8 CD
£
co
2
II Hill!
si
Q_
H—
O V) CD
t CD Q. O CO
it
o
"co CD T3 — i— CD
CD
'55
o
CO
t a5
° °
^O < sI- O
2 Ql
x
LLI
CO
co
Q-
O 4-<
O SI CL.
'62 _0
o o
05
I
O
CO
o
CL
Q_
570
Lange
in this field. At present, renewed efforts to improve the selectivity of photosensitive compounds specifically address typical properties of neoplastic tissue, including vascular permeability; high cell proliferation rate; lower pH values; inefficient lymphatic drainage; and overexpression of antigen receptors, growth factor receptors, and cell adhesion molecules.
2.
RATIONAL ASPECTS FOR CONTROLLED DRUG DELIVERY
A primary problem in controlled drug delivery in PDT and PD of cancer is that in most instances the differences between normal cells of origin and malignant cells are subtle and difficult to quantify. Shared properties between normal and cancer cells have further confounded the development of more specific tumor contrast agents. In normal cells, growth is regulated through a complex web of physical and chemical signals. Depending on their origin, they can be considered to be those that are continuously renewing (bone marrow, intestine), those that proliferate slowly and may regenerate in response to damage (skin, liver, lung), and those that are relatively static. In adults, the rate of cell birth is maintained to be equal to the rate of cell death, whereas in children a slightly higher cell birth rate is necessary to allow development. Cancer can be defined as the emergence of cellular clusters in which cell growth regulation has gone awry. Any cell in the body has the potential to become cancerous if it receives genetic modulations, resulting from spontaneous replication errors, chemical carcinogens, radiation, or even viral infections. Mutated cells that have lost the ability of cell growth regulation will continue to divide, forming what is called the primary tumor, often originating from renewing tissues, such as the epithelium. As they begin to multiply, angiogenic factors, growth factors, and cytokines that stimulate the formation of new blood vessels are released and tumor-specific antigens are expressed. The tumor may be benign or it may begin to invade surrounding tissues, breaking through the basement membrane. Certain cells from these malignant tumors may further mutate to the extent of gaining the ability to leave the site of the primary tumor, enter the blood stream, and invade other tissues. The metabolism of such mutated cells may be altered with respect to their normal counterparts, and indeed considerable differences in oxygen consumption, iron content, and enzymatic activity have been observed between neoplastic and normal cells. Beyond direct injection into a tumor or otherwise localized "physical delivery" of photoactive substances, the ideal strategies for controlled drug delivery are those that allow for highly selective recognition of the neoplastic
Controlled Drug Delivery in Cancer
571
tissue. Based on the above-mentioned considerations, three basic strategies can be derived (Table 2): 1. 2. 3.
3.
Active targeting of a tumor-specific functions Passive targeting by modulation of the biodistribution properties of the photosensitizer or its formulation Using tumor cell-specific metabolic deficiencies for the targeting of neoplastic cells
ACTIVE TARGETING
As can be seen from Table 2, potential sites for active targeting of neoplastic tissue can be structural membrane proteins, receptors, glycoproteins, or lipoproteins associated with abnormal cell growth. Following the concept of "magic bullets," coined by the German bacteriologist Paul Ehrlich [8], such active targeting moieties include antibodies, antibody fragments, growth factor and hormone receptor ligands (proteins, peptides, and small synthetic organic molecules), and nutrient transporters. One of the most selective interactions between biological molecules known to science is characterized by antibody binding to specific antigens. Hence, it is not surprising that, in the search for magic bullets against cancer, antibodies that specifically target tumor-associated antigens have been sought. Monoclonal antibodies (MAbs), stemming from the immunoglobuliny (IgG) class, are the most commonly used carrier system for controlled drug delivery (see Fig. 3). The intact antibody can be described as a Y-shaped, symmetrical glycoprotein, composed of two identical heavy and light chains. The tips of the Y are the so-called antigen binding sites, which are variable. Complete IgGs have a molecular weight of about 150 kD. In antibody fragments (Fabs), the complement-activating Fc region and the socalled "hinge region" are removed (Fig. 3). Such molecules, either bivalent or monovalent, are of lower molecular weight and show advantages over intact IgG molecules in terms of their tissue penetration properties. Barriers to uniform antibody penetration have been suggested to be the endothelial layer and defense fibrous stroma associated with tumors, as well as the tight packing of tumor cells and the absence of lymphatic drainage. Furthermore, deletion of the Fc region, which is responsible for "effector functions" of the immunoglobulin molecule such as complement fixation, allergic responses, and killer T-cell activation, removes the most immunogenic portion of the IgG. Small conjugates made from these fragments usually have the advantage of faster localization in the target tissue and faster renal clearance. These are desirable properties for imaging and diagnostic purposes. And-
0 o c
0 0 0
0
0
O CO
O CO
O CD CO .>
CA
CA
CA
0
0
0
O CO
O CO
CA CA CA 0 0 0
CO
CJ
'o o
«
0
Q.
o> T3 °
CL <® O)
.— 0 > c/5
33 +-< 72. '*j '72 •*-> a aa 000 a aa
CA
0 0 CO 13 ^O Q. '^_, '^_i • • Q. Q. CA
co Q. CL >
O
o
f
c
0
0
0
CA .^
C35 O) O) Q. CO CO CO CO •—O i_ i_ i_ CT <4_
Vf_
-
-
H—
.i. 0 0 .c -^ •g cr cr .t; co
X. 2 J?
CA
v^ O) CA
a,
c CO
O O O c C L Q - O c r O O ' j r O c A
•j -^ -i: ^ c c o _j -43 -43 9- -r o.
CA
"CA o +j CA
o
w
O
en
| c o
£.£l
0 CD Q O)
0 o)
o«^ E
co ^
W
CA
1| (-0
s •>
O 5 NJ
< o cc m
c o o o M—
CA
O CL 0
4-1
CA
0
.c o
O 0
CO
o
O °
o c
o 0 a. CA
O
iL O
E
E
CM CO
0 -77
0 ^ 0 ««
=
Q-
c
O "
ro
CL
8
® 0
P
O
Q. CA
O
^
Controlled Drug Delivery in Cancer
nr
573
"r
DPhe-Cys-Tyr-DTrp-Lys-Thr-Cys-Thr-OH Somatostatin Specific Peptide (MW - 1.5 kD)
FIGURE 3 Schematic representation of several different targeting moieties based on the format of IgG. IgG consists of two identical heavy chains (blue, green) and two identical light chains (red, yellow). Both are organized as domains with about 110 amino acids in each domain, interconnected by disulfide bonds. The variable domains at the distal end of both arms of the Y-shaped IgG contain the antigen binding domain. Low molecular weight antibody fragments [F(ab')2, Fab, scFv] exhibit different pharmacokinetics and enhanced penetration properties with respect to the entire antibody. Recombinant scFv consists of the variable domains linked by flexible peptide linkers. Even small targeting moieties are represented by high-affinity peptides targeting specific cell surface receptors.
574
Lange
body fragments can be derived from selective cleavage of the whole antibody by pepsin and papain. Besides these antibody fragments, the most commonly used Fabs are recombinant single-chain antibody fragments (scFc), in which the resting fragments of the heavy and light chain are linked by flexible linkers [9,10). Such engineered proteins can also be humanized by replacing the murine framework regions with human framework regions. These carrying moieties, either MAbs, MAb fragments, or even smaller selective targeting peptides, can than be conjugated either directly to the photoactive substance or via intermediated linkers, including dextran [11,12], polyglutamic acid (PGA) [13,14], polyvinyl alcohols (PVA) [15], and polyethylene glycol (PEG) [16] (Fig. 4). In both cases of coupling procedure, the resulting dye-targeting moiety conjugate should fulfill several requirements with respect to its PDT and PD efficacy: • • •
The targeting unit will be selective for the tumor cells independent of their localization, with little or no recognition of healthy tissue. The conjugate will have the ability to bind to or be taken up by the targeted cells. The complex can carry the photoactive substance to the target cells without being disintegrated along the way.
FIGURE 4 Photosensitizers or fluorescent dyes covalently bound to targeting agents might help to improve PDT and PD of neoplastic tissue.
Controlled Drug Delivery in Cancer
• •
•
3.1
575
The photoactive substance will not or will only minimally impair the binding activity of the targeting unit. The targeting unit will not or will only minimally influence the biophysical properties of the photoactive substance, including fluorescence and singlet oxygen quantum yield. The conjugate will have the ability to be cleared from the body rapidly without impairing other body functions or inducing unwanted immunogenic responses.
Antibody-Mediated Drug Delivery in PDT and PD
Since Gold and Freedman discovered the presence of carcinoembyronic antigens (CEAs) [17,18] expressed in colon cancer tissue in 1966, it became an obvious call to involve agents specifically addressing such cellular targets in the management of cancer. However, only 17 years the pioneering work by Mew et al. [19] introduced the combination of photoactive substances and targeting moieties to the field of PDT and PD. Although it is widely accepted that the use of MAb as carrier moiety is one of the major breakthroughs further increasing the impact of PDT and PD in the management of cancer, most studies using this approach remained at the stage of in vitro testing of the immunoconjugates. In the few cases where in vitro and in vivo studies were performed with the same compound, the extrapolation of in vitro results to in vivo experiments was difficult to establish. In the first work with MAb-photosensitizer conjugates [19], hematoporphyrin (HP) was coupled to M-l, a MAb against the DBA/J2 myosarcoma. In this study, largely unpurified MAb, extracted from peritoneal ascites fluid, was conjugated via nonspecific carbdiimid coupling to give a 30:1 HP/Mab ratio. In vitro, this preparation, probably a mixture of HP and HP aggregates absorbed on MAb and mouse serum albumin, induced 95% killing of tumor cells. This was shown to be superior to controls treated with HP, Mab, or light alone. In vivo, however, the enhanced treatment efficacy was only minimal. In another in vitro study, carried out by the same group [20] using a similar HP-Mab conjugate directed against a leukemiaassociated antigen (CAMAL), phototoxicity increased by two orders of magnitude in comparison with free HP. In order to avoid an impairment of the antibody binding affinity and to enhance the molar photosensitizer/antibody ratio, early in the development of such compounds intermediary linkers were used to couple photoactive compounds to the targeting moiety [11]. Since then, photosensitizer/Mab ratios as high as 36:1 have been achieved [11]. However, most of the early works in this research field suffer from poor conjugate characterization and
576
Lange
purification. Furthermore, the charge of the complex may have a role with respect to the pharmacokinetics, biodistribution, binding activity, and internalization capacity of the complex [21]. In a recent work, del Governatore et al. addressed this specific problem in in vitro studies [22]. They coupled the photosensitizer chlorin e6 via a poly-L-lysine (pi) linker and a pl-succinyl linker (pl-sc) to a 17.1 MAb to give positively and negatively charged conjugates having a 4:5 (pi) and 8:9 (pl-sc) photosensitizer/MAb ratio, respectively. When cells from a human colorectal adenocarcinoma with a positive expression of antigens, recognized by the 17.1 MAb, were exposed to the conjugates, the uptake of the cationic (pi) form was found to be 38-fold enhanced as compared with the free photosensitizer, while only a 16-fold increase was observed using the anionic (pl-sc) conjugate. In subsequent PDT studies, the cationic photosensitizer-MAb conjugate was found to be more phototoxic than the anionic form. Compared with the free chlorin e6 and a nonspecific IgG-photosensitizer conjugate, the positively charged 17.1chlorin e6 conjugate was found to be 6.5 and 4.5 times more phototoxic, respectively, when cells were irradiated with a light dose of 3 J/cm2 at 660 nm. The same irradiation conditions resulted only in a 30% cell death when nontarget ovarian cancer cells (OVCAR-5) were exposed to the Mab 17.1photosensitizer conjugate, whereas for target HT29 cells 90% cell death was observed. Although enhanced photosensitizer/Mab [23] ratios are highly desirable to maximize the PDT efficacy of the immunoconjugate for PDT and for PD purposes, high photosensitizer/Mab ratios may result in physicochemical processes, such as self-quenching of the fluorescence, that counteract the use of coupling a photoactive compound to a targeting moiety. For example, when M-35 antibodies were directly conjugated to fluorescein via fluorescein isothiocyanate coupling, immunoconjugates with molar ratios between 4 and 24 were obtained [24]. In vitro characterization of these conjugates showed that highest fluorescence quantum yields were observed with conjugates of a molar photosensitizer/MAb ratio ranging from 10 to 14, whereas higher conjugation ratios showed decreased fluorescence, probably due to exciplex formation upon excitation. Furthermore, the binding activities of the MAb-fluorescein conjugates as a function of MAb/fluorescein ratio were investigated in vitro, using immobilized CEA. A small decrease in binding activity and increase in nonspecific binding was observed when CEA was exposed to Mabs labeled with more than 19 fluorescein molecules. Although the impact of high fluorescein loading on the biological activity was small in vitro, the picture changed completely when nude mice bearing CEA-expressing human colon carcinoma xenografts were injected with these conjugates. Twenty-four hours after administration, 38% and 43% of the injected dose (ID) was found per gram tumor, when unlabeled and
Controlled Drug Delivery in Cancer
577
fluorescein-labeled antibodies (1:4 ratio) were administrated, respectively. Only 31% ID per gram tumor was found in the tumors when higher Mab/ fluorescein ratios of 1:10 were used, further decreasing to 6% ID per gram tumor for 1:14 ratios. These results indicate that the in vivo evaluation of new conjugates should be performed in the early phase of the drug development process for immunophotodetection or immunopho to therapy. Thus, its not rational to design highly sophisticated conjugates with good photophysical properties and high binding activities similar to the native antibody in vitro, without optimization of their biodistribution, binding activities, and selectivity in vivo. Unfortunately, most studies addressing the biodistribution of conjugates of MAb-photosensitizer conjugates and their biological activities with respect to binding affinity and/or phototoxicity have been carried out are mostly in vitro [24-29]. However, in the few available in vivo studies, depending on the MAb-photosensitizer conjugate and the administration route, T/N ratios as high as 70:1 were reported (Table 3) [25]. Schultes et al. [26] reported a 60- to 130-fold increase in phototoxicity when phthalocyanines were coupled to a B43.13 antibody against the CA125 antigen expressed in abundance in ovarian cancer cells. No improvement in the in vitro phototoxic properties was reported when benzoporphyrin derivative (BPD) was coupled via a PVA linker to antibodies [30]. In vitro [28] as well as in vivo studies [31] have shown that the phototoxicity of MAb-photosensitizer conjugates can be reduced as compared with the free photosensitizer even if highly favorable T/N ratios were achieved. This can be attributed to ineffective binding [31] or poor internalization properties of some MAb-photosensitizer conjugates [28,32]. With respect to the biodistribution, there are at least four parameters that may affect the efficacy of the MAb-mediated PDT. These are: 1. 2. 3. 4.
A maximal concentration of the photoactive compound in the tumor A maximal T/N ratio Delivery of the photosensitizer to cellular or tumoral compartments, susceptible to PDT Homogeneous distribution of the targeting complex within the tumor tissue
Most authors focus on (1) and (2), while point (4) is insufficiently addressed in most papers on the improved drug delivery using MAbs as targeting moiety in PDT. Experimentally, the most convenient way to directly investigate the parameters listed above is by radiolabeling of the conjugate. Prior to conjugation to the photoactive substance, antibodies have to be labeled
v^
E
DC
'c _ CD
<^-
t—
CM
CO
o ID
r-
CO CM
CM co
CO
CM CO
a>
CM
oo
CO
CO
CM
C CD
S o
E
*- J2
"Z. ^
^
CD Q. X LU
H Q. • CD X 0 CD X "> CD ^ .^
C CO CD
D
__^
f
0 ^ o ° -r-
0 O
E 0
^ >— c 0 "^ ^ CO ^
0 — CM CD CM "-'
. . _0
CO "o m D CO >
C
0
'E
E
3
CJ
< °
o CD
o
0 T3 O
.C Q. O 0
>
a
> >
0
CO
0 _0
_0
•4-J
^
^
CC ^
^
O
-
"CD
^f
^ 0 ^ CO *". 0 ~
Z t- LO ro oo co r~
>
> >
>
>
CD
00 o o
0 o
0 o
CJ
CD =
E
. . _0
E
- E ' oo — ^
^^ _0
Je 3 ^r 3 'Jr ^ 15 /— 0 (— r- 0 § > _§ ^ _§ > o ro
e
1 OT CM 5 A CO O) CO ^ CD r- Z
CD CO
ir
^ i> ^
^
r- CO
Z
>
>
C CD
c
E 13 I
s
iZ
I
0
_i
0 CO
^ .C
LO CM
CD
|—
— > CD
<^ LU O
<^[
O
Lf) CM
r- O < CJ O CO
"c W
^ LU < ^ 0) . ~*f LU ^~i ^" ^l* CO ^i LO LU CO CO < CJ
CD _^ -0 V, **f 0 ^M» CD r— 111 0 LJJ ^ CJ CJ I
LT) CM
<£ O
"O 0 0 CD _O T3 CD DC
o
Q_ c/)
_0
T3
*^
o> c
+
1
1 +
1
1
+
+
4-
CO
c in Lf) CO
« ~O if,
co
|g
CM
U 0
<M
CO 00 ^" ^t CQ LU
'-P ^ <
LD
; 00 CO
^ UJ 4CD Lf) ^
r--
Lf5 CO
to co
O) CM <J ^~ D_
CM ^CD ^ ID O
o c: CQ co
H ^
c
Fluoresce
CD
CO "1 uj CD ^ "° LU Q
E o
Fluorochr
^
_f£ c
0 C
0
r^
'c 0
0
C
C
CD
c
£c
o —
CD
CJ
CD >
CO >• < >
0
|^^^ Q. S
O
D
CD
C
c
CO
^$ O 00 Q 0 >QL "0 — ^ QQ ^
"^
(j
CO DU X 9= 1< E
Fluoresce
o
T3 0 C 0 >
o
LU CD
Lf) CM
O
O 0 c 'c ro o _0 ro
0
- _C
'ECD 1
3 .Q 'i_
v
1 CD C
4—'
^O
+
_c
monoel
O
ZndDPhth
E
CO
0 .—
~ CD O
CD O
c _o
o CD E ^: 0 £ > o Hro > 0
-a
c ^ t-
D
T tetras
CD
.—
^f"
"CD CO O
O ±r
•4-1
5 o
uma
CD
c
Q.
3 C
O
E
^ o Q. o CD t~ o
N
C
O
J3
_o
0
"c CO < CO CJ 03
<
Controlled Drug Delivery in Cancer
579
with i25I or I31 I using the iodogen or similar methods [24,25,28,29,33]. These techniques enable, for instance, the comparison of the binding activities of the MAb-photosensitizer conjugate with free MAbs, the determination of MAb/photosensitizer ratios, as well as the quantitative evaluation of the accumulation of the immunoconjugate in the target tissue. In experimental studies, most authors generally consider the tumor to muscle ratio. However, these ratios may not be relevant for potential clinical applications, since muscles are rarely the origin site or primary site of disseminated tumors. As mentioned above, antibodies have also been considered as carrier moieties for fluorochromes, enabling improved detection of early cancerous disease or superficial tumors using modern fluorescence imaging techniques. In such cases, several parameters should be taken into account with respect to the fluorescent label, including the detection and excitation efficacy and the native absorption, scattering, and fluorescence properties of the tissue [34]. One drawback of using fluorochromes for cancer detection is the relatively short mean pathway of light in tissue, which would give poor definition of deeply buried tumors. While the possibility to perform any kind of imaging could be reduced for such lesions, detection would still be possible through considerable thickness of tissue. First experiments with fluorescence-labeled antibodies were carried out using fluorescein as a fluorescent label. This specific dye has a high molar extinction coefficient at 490 nm, corresponding to one of the principle emission lines of an Ar-ion laser (488 nm). It has a high fluorescence quantum yield and a low toxicity in humans [35]. However, the excitation at 488 nm also produces a strong native autofluorescence in tissues [36], causing a significant background signal, which must then be subtracted using complex algorithms. Moreover, the fluorescence quantum yield of fluorescein is strongly pH sensitive having its maximum at pH 8, sharply decreasing below pH 6. In consequence, several research groups [25,33,37,38] started to use alternative fluorescence markers, thus circumventing the above-mentioned drawbacks. Several sulfoindocyanine dyes have advantageous optical properties, such as an excitation (absorption) maximum in the red and a fluorescence emission in the far red and near-infrared, where CCD cameras have spectral sensitivity peaking. At such wavelengths, the intrinsic tissue fluorescence is negligible and scattering of both excitation and emission light is minimal. Moreover, such dyes are insensitive to pH over a broad range, and self-quenching [24] is less important. Antibodies labeled with Cy5 were used for in vivo tumor detection using both mouse tumor models and human tumors xenografted in nude mice [25,33,37,38]. Folli et al. [25] tested the serum lifetime of iodinated and Cy5-labeled antibodies and found the Cy5 conjugates to have a similar circulation half-time to that of unconjugated antibodies. In contrast, and-
580
Lange
bodies labeled with high amounts of fluorescein were found to have dramatically shorter whole-body half-lives with a reduced ability to localize in tumors. Other derivatives, including Cy5.5 and Cy7, were tested for the efficacy of tumor visualization when coupled to specific monoclonal antibodies [37-39]. However, the presently available conjugates have shown significant limitations in vivo. Upon injection, some of these immunoconjugates were nonspecifically deposited in other organs [37-39]: First in the liver and spleen followed by an increasing fluorescence in the gall bladder [33,39], and second, in the intestines and colon, which were found to be weakly fluorescent for several days [39]. Despite the promising results obtained with conjugates between MAbs and photoactive substances both in vitro and in vivo, they have received only minimal clinical testing for PDT as well as for PD purposes [36,40,41]. Only three pilot studies have been carried out so far. Since then all efforts for clinical testing have been stopped, presumably due to several disadvantages of macromolecular carrier systems, including: 1. 2. 3. 4. 5.
Limited availability disposability of sufficient amounts of MAb Potential risk of systemic toxicity Complicated synthesis and purification procedures Loss of photochemical and/or biological activities Low bioavailability, caused by the large size of such conjugates barely penetrating biological barriers
In humans, a long time is required for a significant portion of circulating blood to access a small tumor, while the circulating half-time of antibodies is relatively short. Thus, the majority of the antibody will be cleared from the body before it has accessed the tumor tissue, and only small amounts of the conjugate will accumulate within the tumor when maintaining acceptably low exposure of patients to the drug. This problem becomes evident when comparing the data obtained in mice and in humans using practically the same MAb-fluorescein conjugates for photodetection purposes [24,36]. In a clinical pilot study for photodiagnostic purposes, a total of six patients were injected with 4.5-9 mg of MAb equivalents directed against CEA in colon cancer. The anti-CEA MAbs (CGP44290) were loaded with 10 and 14 molecules of fluorescein per radiolabeled MAb, respectively. In this study, photodiagnosis was performed 24 hr after injection ex vivo on surgical resections for all patients. In addition, one patient underwent an in vivo fluorescence diagnosis during rectosigmoidoscopy. All tumors were found to be positively stained by the targeting conjugates, but the fluorescence was found to be heterogeneously distributed within the tumor mass. Although 10 times more fluorescence was found in the tumor as compared to the healthy surrounding mucosa, only
Controlled Drug Delivery in Cancer
581
0.059% of the ID was found to accumulate per gram tumor. In contrast, 31 % ID/g tumor was found in mice bearing human colon carcinoma for the same MAb-fluorescein conjugate, indicating the huge difference between preclinical experiments performed on small animals and clinical studies in humans. In gynecology, Zn(II)phthalocyanine, coupled via an avidin-biotin complex to the OC125 antibody, was tested in women with ovarian cancer, positively proven for the expression of the CA125 antigen [40,41]. In these studies, the MAb-photosensitizer conjugate was applied 72 hr prior to irradiation of three women. In the first report [40], tumor devitalization was tested by histopathological analysis of biopsies taken 6 hr after irradiation in the course of the operation procedure. The first evidence of cell death was provided by the destruction of the mitochondrial membrane as confirmed by transmission electron microscopy. Furthermore, mitochondria swelling as well as an increased density of mitochondria was found in the irradiated areas. In a recent update of this study, a controlled trial involving 62 patients suffering from ovarian cancer was carried out [42]. All patients underwent surgical resection of the visual tumor nodules. After tumor debulking, 31 patients were intraperitoneally instilled with MAb-photosensitizer conjugates (2 mg B4313) 72 hr prior to operation and were locally irradiated with 5 J/cm2 at 675 nm. Subsequently, the course of these patients was compared to the control group treated by standard surgery. The survival was found to be 90% after 30 months in the PDT group, whereas only 78% of the control group survived up to this point. However, after 60 months both groups showed essentially the same survival rate (40% for the PDT group and 37% for the control group). Furthermore, 23% of women who underwent PDT showed subelius and 6.8% anastomosis following treatment. 3.2
Antibody Fragment-Mediated Drug Delivery in PDT and PD
In spite of their early promise, MAb conjugates were largely unsuccessful as targeting agents in chemo- as well as in photodynamic therapy. However, since it takes about one decade to develop a new therapeutic agent into a viable commercial product, the design of most antibodies was quite primitive and not conceived by modern molecular biology. Nevertheless, it must be considered that several intrinsic properties of intact antibodies will further delay their use in daily clinical practice. In IgG, the activity of antigen binding is determined by the sequences and conformation of the amino acids of the six complementary determining regions (CDRs) that are located on the light and heavy chains of the variable portion (Fv) of the antibody (see Fig. 3). Once Fv is bound to a specific
582
Lange
antigen, the sequences in the constant region (Fc) will trigger numerous immune functions. Due to their high molecular weight of approximately 150 kD, intact antibodies exhibit a long circulation half-life, which can lead to bone marrow exposure. Furthermore, interactions with Fc receptors located on normal tissue can alter the biodistribution of the MAb and potentially harm the patient. The most important drawback for the use of MAbs in PDT ad PD is, however, their poor bioavailability. Their large size limits the ability to diffuse from vasculature into a tumor, which can be attributed to the high hydrostatic pressure resulting from the disordered tumor blood vessels and the lack of draining lymphatics [43,44]. The resulting heterogeneous distribution of a photoactive substance within the target tissue will subsequently impede the efficacy of photodynamic treatment of the entire tumor mass because the range of action of [O2 was estimated to be in the order of 0.05 fjim [45]. However, instead of targeting tumor cells, one interesting approach worth noting in this context is the targeting of the tumor-associated vasculature. Because PDT is known to induce photothrombic effects, the targeting of vascular endothelial cells by MAb-photosensitizer conjugates with subsequent irradiation can result in occlusion of the blood vessel network. In consequence, tumor cells that depend on the nutrition and oxygen supplied by these vessels will become unable to survive. Another approach to overcoming the problem of poor bioavailability of intact immunoglobulins is to use small antibody fragments, thereby increasing their ability of tumor penetration and reducing the circulation halftime without losing the affinity for the targeted antigen. The smallest fragments, containing only the variable region of the light chain (VL) and the variable region of the heavy chain (VH), have a molecular weight of approximately 30 kD. In vivo, these molecules display rapid, biphasic pharmacokinetics with an equilibration phase of between 2 and 12 min and an elimination phase of typically 1.5-4 hr [46]. The most pertinent way to produce antibody fragments is to derive them from intact antibodies via pepsin cleavage to give bivalent antigen-binding fragments [F(ab') 2 ] (see Fig. 3). Hasan et al. [47] have investigated the influence of the charge on the intraperitoneal (IP) biodistribution of chlorin e6 coupled to a F(ab') 2 derived from the OC125 antibody recognizing the CA125 antigen expressed in 85% of nonmucinous epithelial ovarian cancer [48]. Comparable to similar work by the same group [22], chlorin e6 was specifically coupled via pi and pl-sc linkers to OC125-F(ab') 2 fragments. Subsequently, their binding activity and in vivo biodistribution was compared to that of the free F(ab') 2 . Moreover, the photosensitizer was compared to a nonrelevant rabbit Ig-photosensitizer conjugate, respectively. While nonspecific mouse IgG showed no specific binding in enzyme-linked immunosorbent assay (ELISA), cat-
Controlled Drug Delivery in Cancer
583
ionic F(ab')2/chlorin e6 conjugates did and the anionic form even increased binding activity in comparison with the free OC125 F(ab')2. After IP administration of these substances to tumor-bearing mice, free chlorin e6 showed the highest mean T/N ratio as compared with both the anionic and cationic F(ab')2-photosensitizer conjugates 24 hr after application. However, this high T/N ratio for the free photosensitizer is strongly influenced by the high T/N ratio observed in the kidneys, which are one of the major elimination routes for smaller proteins, thus showing rather small T/N ratios for the immunoconjugates. The highest tumor fluorescence was attained using the cationic F(ab')2-chlorin e6 conjugate. Fluorescence was found to be about three times higher than that of the anionic chlorin e6-F(ab')2 or the free chlorin e6. The results of this study indicate that, due to the high amounts of chlorin e6 delivered by the cationic immunoconjugate per gram of tumor, cationic charges stimulate the endocytosis and the lysosomal degradation of the OC125-pl F(ab')2 chlorin e6 conjugate. In the past, much of the screening for tumor-homing molecules was based on the immunization of mice and the screening for monoclonals in cultured pools of antibody-producing cells. In 1988, methodologies were developed to produce active antibody fragments in Escherichia coli [49,50]. The pioneering work of McCafferty and colleagues [51] a few years later allowed the development of large, combinatorial libraries of antibody fragments on the surface of bacteriophages for the screening of more efficient targeting moieties. In such, so-called phage display libraries, the interaction force between VL and VH is very low. Thus, engineering of a covalent link between VH and VL becomes necessary to obtain a stable molecule. The most common method for this purpose is to use a flexible linker of 15-20 residues to join the two domains, giving the resultant single-chain Fv fragment or scFv (see Fig. 3) [52]. Nowadays, these recombinant antibodies and antibody fragments represent over 30% of all biological proteins undergoing clinical trials. This boom is mostly due to modern phage display techniques. For the production of human antibody fragments, the VH and VL domain of human antibodies are polymerase chain reaction (PCR)-cloned in a filamentous phage DNA and displayed on the surface coat of phage. Then antigen-specific phage is selected by in vitro panning against the target antigen of interest. The advantage of such methodologies is in the isolation of high-affinity human antibodies to many different antigens within a few weeks. Selections performed in large libraries consisting of about 10'° different clones yielded scFv with KD as low as 2 X 10~ l() M [53]. In recent work, Ramjiawan and coworkers coupled Cy5 to a recombinant scFv expressed in E. coli for in vivo imaging of mice bearing subcutaneously human melanoma tumor cells [54]. The NovoMab-G2-scFv used in this work recognizes tumor antigens expressed in a variety of human
584
Lange
tumors, including melanoma, breast cancer, colon adenocarcinoma, squamous cell carcinoma, lung cancer, and prostate cancer. Whole-body imaging after illumination at 670 nm of mice injected with the scFv-Cy5 complex resulted in a strong fluorescence at the tumor site two hours after administration. Neither free Cy5 nor an irrelevant antibody fragment conjugated to Cy5 gave marked tumor fluorescence in this study. The strongest fluorescence surpassing the tumor fluorescence was observed in the kidneys in all cases. As expected for small scFv, the fluorescence was found to decrease rapidly, with a mean half-life of approximately 8 hr. Following the approach of targeting the tumor vasculature, the research group of Dario Neri has developed several recombinant scFv's directed against the extra domain B (ED-B) of the fibronectin isoform B (B-FN), which is found in abundance in vessels of neoplastic tissue [55,56]. Antibody fragments isolated from phage display libraries were labeled with different infrared dyes and tested in vitro and in vivo. Affinity measurements for nonlabeled scFv against immobilized fibronectin analogs using surface plasmon resonance techniques have shown KD values in the order of intact antibodies ( 1 . 1 nM for scFv(CGS-2) and 0.54 nM for scFV(L19) [57]). In some scFv, such high affinities were impeded after labeling with Cy7, presumably due to preferential labeling of amino groups in the antigen binding site [56]. Sulfoindocyanine dye labeled scFv's have been tested in several experimental animal models for angiogenic research, including tumor-bearing mice, the chick's chorioallantoic membrane model, and the rabbit's corneal pocket assay. In tumor-bearing mice, a strong and selective accumulation of the scFv(L19)-dye conjugate, but not of the scFv(HyHEL-10)-Cy7, was observed after IV administration (see Fig. 5). Using the scFv(CGS-2)/ Cy7 conjugate, tumor imaging was possible 10 min postinjection, lasting for at least 2 days [55]. For this conjugate, 2% ID/g tumor was localized in the tumor mass 24 hr after administration. Intact antibodies are polyvalent molecules. This provides a significant increase in functional affinity due to simultaneous binding to two or more target epitopes. As mentioned above, a major limitation of scFv molecules is their monovalent nature. Logically, one of the easiest approaches to further improve the binding affinity of scFv to its target antigen is the multimerization of the targeting moiety. Such an attempt will further reduce the elimination of the antibody fragment through renal routes [38,39,47,55,56]. Neri and colleagues addressed this specific problem by introducing an amphiphatic helix containing a cysteine residue at the C terminus to allow two scFv's for dermerization. While scFv(CSG-2) conjugates showed no increase in affinity, a second conjugate [scFv(CSG-l)] showed dramatically improved pharmacokinetic properties. This "diabody" exhibited good stability against
Controlled Drug Delivery in Cancer
585
FIGURE 5 Fluorescence images of nude mice, bearing a subcutaneously grafted tumor, recorded at three time points after injection of Cy7-labeled scFv(L19) and nonspecific scFv(HyHEL-IO). T = tumor site. (From Ref. 56.)
enzymatic proteolysis and low tumor uptake while retaining its binding activities to tumors. Pharmacokinetically, the targeting of neovascular structures is a "onecompartment" problem in which the main input parameters are vascular permeability and interstitial pressure. Furthermore, this approach overcomes the problem of producing specific targeting agents for each cancer, thereby generalizing the problem. However, tumor development is a highly complex process in which angiogenesis is generally preceded by an avascular phase, where the primary tumor is still localized. Superficial and microinvasive tumors as well as carcinoma in situ often lack the presence of a developed neovasculature. Thus, the targeting of angiogenesis-associated antigens by antibody fragment-photosensitizer conjugates can only help to manage or detect advanced tumors. Besides tumor development, angiogenesis is a process that is associated with many other human diseases, such as choroidal
586
Lange
neovascularization (CNV) associated with age-related macular degeneration, diabetic retinopathy, cardiovascular diseases, psoriasis, and rheumatoid arthritis. Thus, the targeting of neovessels in such disorders with scFv-photosensitizer conjugates and subsequent irradiation of the site of the lesion can improve current treatment strategies. Using their scFv(L19), Neri et al. recently demonstrated selective occlusion of neovessels in a rabbit cornea micropocket angiogenesis assay, when coupling chlorin e6 to this targeting moiety [58]. 3.3
Miscellaneous Receptor-Mediated Drug Delivery in PDT and PD
Due to the huge progress that has been made in biological and biochemical research related to cancer during the last decade, many of the processes in the development and growth of cancer cells as well as factors in the microenvironment that are required for tumor growth have been unraveled. This expansion of knowledge has identified a number of potential targets other than antigens for an improved drug delivery in the management of cancer. Worth mentioning in this context are those factors that are involved in angiogenesis, including cell adhesion molecules (CAMs) such as integrins [59-62] and growth factor receptors [63,64] such as vascular endothelial growth factor (VEGF) receptor [65,66] and basic fibroblast growth factor (bFGF) receptor [67]. Furthermore, receptors overexpressed in certain types of tumors, such as epidermal growth factor receptor (EGF-R) [68,69], estrogen receptor (ER) [70], and somatostatin receptors (sst) [71], have known internalizing ligands that can be used as targeting moiety for improved drug delivery. It is beyond the scope of this chapter to list all possible targets for cancer treatment. Instead, it will focus on those targets with potential use in PDT and PD. Recently, scientists involved in the field of biomedical optics have used new targeting agents to enhance the selectivity of photoactive compounds as required for PDT and PD. Although small, synthetic peptides have by far smaller affinities to receptors than antibodies or antibody fragments for their corresponding antigen, they have been proposed as targeting moieties. It has been demonstrated that such molecules, homing specifically on receptors and CAMs, can be conjugated to chelating agents without losing their affinity to the target molecule [72]. Some of the peptides consisting of fewer than 10 amino acids have shown significantly low IC50 values, e.g., octreotide (see Fig. 3) for the sst2 receptor (IC 5() = 0.4-2.1 nM) [73]. Small peptides have several advantages over large biomolecules, including ease of synthesis of a variety of compounds for combinatorial screening, reproducibility of high-purity substances, diffusiveness to solid tumors, absence of
Controlled Drug Delivery in Cancer
587
immunogenic responses, and the ability to incorporate pharmacokinetic modifying functional groups. Octeotride, a cyclic, eight-amino-acid peptide, that mimics somatostatin, has been recently coupled to an indocynanine green (ICG) analog for infrared fluorescence photodetection in rats bearing sst2 expressing pancreatic acinar carcinoma [74]. In this study, tumor visualization was already possible 10 min after intravenous administration of both free ICG and the peptide ICG-peptide conjugate. While ICG retention in the tumor was found to be only transitional, the octeotride-dye conjugate was observable for at least 24 hr in the target tissue. However, relatively high amounts of the targeting contrast agent were also observed in the pancreas and adrenal glands, known for increased sst2 expression, as well as in the kidneys, the main elimination channels for small peptides. A similar approach was followed by another research group with the objective of screening a large number of dye-labeled peptides in a highly parallel automatic nanosynthesis [75]. The methodology presented in this paper was applied in the preparation of different vasoactive intestinal peptide (VlP)-derived peptides and allows parallel screening of approximately 560 compounds. The corresponding receptors for this peptide have been detected in various tumors [76,77]. Cell surface receptors are proteins, which are typically composed of several binding domains, which contain at least one specific binding site. Besides their role in cell signaling processes, they are responsible for the delivery and uptake of cell substrates. Thus, receptors may be subdivided in two basic classes: (1) the "cargo"-type receptors, which deliver metabolic substrates, nutrients, and minerals to the cell; and (2) receptors engaged in signaling, triggering different intracellular events, when bound to a specific ligand. The latter class includes receptors with intrinsic tyrosine kinase activity, such as the EGF-R [68,69] and the insulin receptor. Receptors in this class respond to circulating stimuli to provoke the cell to undergo a wide array of shifts in metabolism that ultimately leads to growth and/or differentiation of individual cells. Gijsens et al. proposed using EGF as the targeting carrier for photosensitizer [78,79]. EGF-R was found in elevated levels in many tumors, such as squamous cell carcinoma, melanoma, glioma, bladder cancer, and breast cancer [68,69]. They coupled CM A via dextran, PVA, and human serum albumin (HSA) to EGF and investigated the influence of these linkers on the binding activity. Moreover, the cellular uptake and the production of reactive oxygen species (ROS) was studied and compared with the free CMA. Nonspecific coupling to the targeting unit presumably was the reason that the CMA-EGF conjugates showed smaller binding affinities in vitro as compared with free EGF. Using HSA as linker was found to impair the
588
Lange
binding to EGF-R-expressing cells less and to deliver five times more CMA to the intracellular space as compared with the free photosensitizer. The rationale of using insulin as a targeting unit for improved drug delivery of photosensitizers is the overexpression of the insulin receptor in human breast cancer. Up to 20 times higher insulin receptor expression in breast cancer cells was observed than in nonmalignant human breast cells [80,81]. Chlorin e6-insulin conjugates were reported to be approximately 100 times more phototoxic than free photosensitizer. Furthermore, lower light doses were needed with the targeting agent. This high phototoxicity was attributed to the favorable intracellular distribution of the photosensitizer following insulin-mediated endocytosis [82,83]. However, the preferential accumulation of the photosensitizer within the environment of the nucleus bear the potential risk of mutagenicity of the insulin-mediated PDT, since 'O2 may damage the DNA closely positioned near the nuclear membrane. Both approaches using tyrosine kinase-activating mediators focus on the targeted intracellular delivery of photoactive compounds, an important aspect with respect to the efficacy of the photodynamic treatment, which has been recently reviewed by Sobolev and colleagues [84]. However, these studies have not addressed the principle parameters of PDT which are probably more critical for PDT efficacy such as biodistribution, T/N ratios, and photosensitizer distribution within nodular tumors. The other category of cell surface receptors, the "cargo" receptor family, overexpressed in tumors and neoplastic tissue responds to the high metabolic activity of proliferating cells. Examples of these include the lowdensity lipoprotein (LDL) receptor and the transferrin (Tf) receptor, which serves as a docking site for iron-laden Tf [85,86]. The use of LDL as a drug delivery vehicle can be traced back to Gal et al. in the early 1980s [85]. The core of these blood plasma proteins consists of fatty acid-esterified cholesterol surrounded by a monolayer of lipids and free cholesterol for most tissues. Many neoplastic cells, including many of the most aggressive tumors, show enhanced uptake of LDL, which can presumably be attributed to the enhanced requirement for structural cholesterol and steroid-derived products in highly proliferative tissue [87]. In PDT, this specific property of rapidly dividing cells can be exploited by both direct coupling of a photosensitizer to LDLs or administration of liposomal or free photosensitizers, which upon injection become associated with LDL. The latter and mostly used approach, however, can be considered as passive targeting and will be discussed later in this chapter. The first indication of the targeted delivery of photosensitizer to tumor cells via (LDL)-(LDL-R) interaction was given, when BPD-MA was preincubated with lipoproteins and applied to tumor-bearing mice [88,89]. Fast pharmacokinetics were observed for all compounds in this study. As com-
Controlled Drug Delivery in Cancer
589
pared to the free BPD-MA, both HDL and LDL complexed BPD-MA was found to be more preferentially accumulated in the tumor and exhibited a high phototoxic potential 3 hr postadministration. Higher retention of the LDL precomplexed photosensitizer, as compared with the HDL analog, was concluded from PDT treatments 8 hr after administration, where only the LDL/BPD-MA showed significant photodamage [89]. In a more recent report, chlorin e6 was selectively conjugated via lysine-carbodiimide coupling to the apoprotein of LDL [90]. With an optimal chlorin e6-to-LDL ratio of 50:1, a four- to fivefold increase of photosensitizer uptake was observed in vitro, using a human retinoblastoma cell line. Phototoxicity studies in the same cell line showed an eightfold enhancement of cell killing using the LDL-chlorin e6 conjugate when compared with the free photosensitizer. Macrophages have also been found to accumulate LDL-photosensitizer conjugates to a high extent [91]. These cells act as mediators of foreignbody response to other cells that regulate the immune system. Consequently, they have been used for indirect targeted PDT and PD. Macrophages that have undergone phagocytosis can deliver selectively high amounts of LDLphotosensitizer conjugates to malignant cells via scavenger receptor mechanisms. Highly specific uptake of maleyated bovine serum albumin (BSA) was also observed in J774 murin macrophage-like cells [92]. 4.
PASSIVE DRUG DELIVERY IN PDT AND PD
In contrast to active tumor targeting, where the photoactive compound is covalently bound to a carrier substance that recognizes tumor-specific functions, in passive targeting only the pharmacokinetic properties of the photosensitizer or the corresponding formulation decide on the in vivo biodistribution upon systemic administration. Again, one of the principle targets for such passive drug delivery will be the intrinsic properties of the tumor-associated vasculature. Increasingly, unequivocal evidence has shown that solid tumors are characterized by hyperpermeability of the vasculature [43,44,93-97] and the immaturity of the lymph drainage [93,97]. Furthermore, blood flow velocity may be significantly slower in tumors in contrast with normal tissue. The tumor vascular system consists of both vessels from the preexisting organ or tissue and new vessels stimulated to grow by the release of angiogenic factors [43,44,98]. Among other things, the distribution of nutrients and drugs through the target tissue depends on the length, diameter, number, and geometry of the available blood vessels. The differences in permeability and lymphatic drainage in tumor tissue can produce significant tumor targeting of high molecular weight compounds and particulate carriers, which have a long plasma circulation half-life. These
590
Lange
observations led Maeda and coworkers to coin the term "enhanced permeability and retention" (EPR) for passive drug targeting of polymer conjugates (Fig. 6). Many macromolecules, such as albumin, antibodies, blood plasma proteins, and polymers, accumulate passively within solid tumors. Observations that liposomes and particles with a size of 200-600 nm [96], and many small differences of uptake of macromolecular weighting in the range between 10 and 800 kD are indicative that passive tumor targeting by the EPR effect can be accomplished using polymeric and micellar carriers. After IV administration, polymers that do not bind to blood plasma proteins display a plasma clearance that is primarily governed by the rate of liver uptake and clearance by the kidneys (see Fig. 6b). However, the biodistribution of photoactive substances essentially depends on their ability to bind to serum components, including lipoproteins and albumin. Pharmacological concepts provide several strategies that can considerably alter the pharmacokinetic properties of substances to be used for PDT and PD, including encapsulation in liposomes and derivatization of the lead structures resulting in more lipophilic, hydrophilic, or amphiphilic compounds. In the early 1980s, researchers began a more or less systematic search for improved tumor photosensitizers. Design criteria for such second-generation photosensitizers were built up with respect to purity, toxicity, stability, photophysical parameters, and solubility (Table 1). Presumably due to the high singlet oxygen quantum yield of the porphyrin skeleton, most of the second-generation photosensitizers described belong to this family (Fig. 7). For improved passive delivery of photoactive substances, the porphyrin system on its own provides 12 positions that can potentially be substituted by sulfonic acid, carboxylic acids, hydroxyl, quaternary ammonium salts, carbonyl substituents, and so forth to give a nearly countless number of possible derivatives of the lead structure. Furthermore, the porphyrin cycle can be oxidized, extended, and/or a central ion may be introduced to alter the pharmacological as well as the photophysical properties of the molecule. Similar derivatization possibilities apply for other singlet oxygen-generating compounds, including perylenequinones, triarylmethane dyes, phenothiazines, and hypericin derivatives. The resulting derivatives can be classified according to their solubility and charge into: 1. 2. 3.
Hydrophobic photosensitizers: reduced solubility in water and alcohol; no charged peripherical substituent Hydrophilic photosensitizers: soluble in water and alcohol; three or more charged peripherical substituents Amphiphilic photosensitizers: soluble in water and alcohols; two or fewer charged peripherical substituents
Controlled Drug Delivery in Cancer
591
This part of the chapter will focus only on such photosensitizers and corresponding formulations, which directly or indirectly tend to use the EPR effect for the delivery of photoactive compound to the target tissue. In particular, this applies to hydrophobic and, to a lesser extent, to amphiphilic photosensitizers which, due to their reduced water solubility, must be injected in a suitable delivery vehicle (Table 4). The porphyrin as well as the phthalocyanine skeletons are essentially hydrophobic. Consequently, derivatives of these two classes, having nonpolar substituents, have to be administered in a pharmaceutically acceptable form. The most common way for highly or moderately lipophilic compounds is to encapsulate the photosensitizer in liposomes or to prepare emulsions using emulsifiers such as ethoxy castor oil, Cremaphor-EL (CR), or polyoxyethylsorbitan monooleate (Tween) [99-125]. Liposomes are built up from lipids that contain two regions: (1) a polar hydrophilic head group (usually a phosphate group) and (2) a nonpolar, lipophilic tail, most likely consisting of a hydrocarbon chain with 14-18 carbon atoms. Highly hydrophobic compounds are generally carried in the lipid bilayer of liposomes, while hydrophilic drugs are entrapped inside the liposomal sphere. Jori and coworkers were the first to study the impact of liposomes on the biodistribution of Hp following IP administration [99]. The tumor uptake of Hp encapsulated in unilamellar liposomes of dipalmitoylphosphatidylcholine (DPPC) in MS-2 fibrosarcoma bearing mice was found to be slower as compared with the free drug. However, the final tumor concentration of liposomal-formulated Hp was approximately twice as high, while uptake in normal skin was lower than with the free Hp. Similar experimental conditions were used with another photosensitizing agent, Zn(II)phthalocyanine (ZnPC), by the same group [100]. Phthalocyanine and naphthalocyanine (Nc) dyes are highly hydrophobic and belong formally to the family of tetraazaporphyrins, with four benzenoid or naphthaloid rings attached to the tetrapyrrolic skeleton, respectively. In this study, maximal tumor concentrations of 0.6 ju-g/g tissue at 24 hr after IP application were reported, while for the same drug IV administration resulted in 0.3 /xg/g tumor tissue using approximately fourfold lower drug doses [101]. For IP administration, the maximal tumor/muscle ratio was found to be 7.5 at 24 hr postinjection, which was twice as high as for IV injection. For both administration routes, never more than 0.1 /Ltg/g tissue was observed in the skin between 1 and 168 hr postadministration. The highest T/N ratio with this photosensitizer was observed when ZnPc was delivered in a liposomal formulation consisting of a mixture of palmitoyloleoylphosphatylcholine (POPC) and dioleoylphosphatylcholine (9:1) instead of DPPC liposomes [102]. Approximately 10 times more of the photosensitizer was found in the tumor than in the normal tissue 48 hr after IV administration in Meth-A sarcoma-bearing mice.
592
Lange
Controlled Drug Delivery in Cancer
593
10
20
15 FIGURE 7 Schematic representation of the porphyrin skeleton, which is the basic structure for most photosensitizers.
However, using other rodent animal models for the investigation of the biodistribution of DPPC-delivered ZnPC qualifies the promising results observed in murine models. Only slight differences in fluorescence intensities in tumor tissue and blood vessels were reported for rats with isogeneic mammary carcinoma [103], and golden hamsters with chemically induced tumors showed tumor/muscle ratios of 1.29 at 72 hr after injection of 0.2 mg/kg body weight [104]. The second group of tetraazaporphyrins, Nc's, are even more hydrophobic than PC'S due to their additional four benzenoid substituents. However, the extension of the conjugated TT system results in a very strong absorption in the 750- to 800-nm region of the spectrum. One of the simplest of these compounds, Zn(II)naphthalocyanine (ZnNc), and three tetrasubstituted derivatives (see Table 4) formulated in DPPC liposomes were the subject of biodistribution and pharmacokinetic studies of Whorle et al. [105] and Shopova et al. [106]. Chemically induced rhabdomyosarcoma in golden hamsters accumulated about 0.7 /Ag/g tissue 24 hr postinjection and the corresponding tumor/muscle ratio was found to be 14. Male C57/BL mice implanted with Lewis lung carcinoma were used to compare the biodistribution of three ZnNc derivatives [106]. Following IP ad-
FIGURE 6 Schematic representation of models for accumulation of low, middle, and high molecular weight targeting agents at tumor and normal tissues after intravenous injection following the EPR concept.
Lange
IV* LL.
c o
ro
"5 E UJ
Z '"" .2 x 03 *~
CNi-CMCNr-CNCNCNCNCNCNCNO)
CDCOCOOO
CD t-
CD r-
CNI'3'
LOCOI^O^
CM C\i (<
IT) 00 CN IT) O) -^
«-OOO<-<-CN<-'-
^ T< <
<<
<-:
*- CD
o
E =
w
-o o
CD
<
> > > > > > > > > > > > Q .
> > > > > > > > >
E 0
o
_
Q. '~\
CD T3 0
-Q
^
T3 CD
CD ^
tj CD CD
"Q) Q
Qw \ 0 O O)
Ef
t i O O O O O O O O O O O O
CD CD CO C/5 3 3 o o
0 CD CD CD CO CO C/) CO 3 3 3 3
o 3
o o o
^
>
' c n n 3
cc^E^^IJE^cE^cE^E^^ieE <EScc^^2E«:EDCQ:;
**""~
•—
w
C D C D C D C D C D C D ( D C D C D C D 0 C D OOC/3C/3C/3C/5C/)C/)C/)C/)C/5C/)C/) 3 3 3 3 3 3 3 3 3 3 3 3
^
^E
LO
LO LT)
LO LO LO LO LO LO CO
LO
r--ococoLOcocr>oooococor^o
OLOCMLOOOOOO
O O ^ O O O O O O O O O O L O 00
L O O O O L O L O r - ^ - ^
c
o
Q.
E o
CJ CD .>
E ® GO
4-J
^
O 03
C/) ^
o
o
S>
^~ 6)
CO Q_
O5 LU
O Q
o ^
Q-D fl^
c
.^
0 Q
\U
o
CD CD
JJ
^CJ OCJ CJ CJ OCJ °r CJ 0 O 0 U ^Q_ D_ Q_ Q_ Q_ Q- Q_ TO Q- CD Q0 n D C ^ Q - Q C D - Q - C C Q - i r Q - D C O Q - O)Q- $ D - D C ^ CC Q- DC L U Q F Q O F C J Q U CJQQCJQQCJQOQCJQ-Q
(A
M— O
1
ti _5 —
?
3 o
Q.
H
c
.2
o
.2 DO
CD >
"D
'4-1
Tf
uj
m H
CJ CD
2
o Q-
X
00
o
0 < r
~O "O i o o io 3 3 CQOQC
X
< < 2
i i i 1 1 1 1 1 1 r" r~ (^ QJ [^ o i x x i r o i o oi o oi o oi o io ;rn : 3 3 3 3 3 3 3— QOQCQOQDQCQm 2
^goooooooooo |_
i
O
O
O
O
O
O
O
O
O
^»
c —
c £5
C3 '^
c ~
O(/3(/)cn-C^
^ ^
O
^ D _ C D C D C D C D C D C D C D C D C D 0 Q _
ooCDCUOOCDOCDCDCDOCDl
> O O O 0 0 ? C C
i^^^^^cococo
o <-
LO CO CO CO CO 0 0 0 0 o
0 <- *•* * 0 0 0 0
r^ oo
CM CO CM CM
o
c" c .E 6
o E
^ 5.
> ^co
o" CL
Q. CD
c
O
0
("•• CD ^t ^ <
*~ Z
<£ <£
LO ^ 00
z z z r^ co
t- LO
r» o o LO «- to
< z £2 £1
Q
CL Q)
CO
a 0 c
CD
-C
c o_ S CD
N Q_CJ
Q. a Q. a a
a > > > > >
> >
-Q E
.— •* CD ^f *™1
o E > >
°1
^ -^_ IM O £ c c co
2 co
Q o i-
CL
CQ CL o ^ . , 1—
a. O
Jp. O C r1-
N ^
3 "
-. CD CD C C —
.E
E D
o
0) CO
Q) CO
D
D
O 0
„ cB
CD
•(->
CO 3
CO E
0) CD
CD
CD
CO D
CO 3
CO D
CO D
O O 0 O
^ ^ 20 ro Is ^ s s ^^s
"o 'c >5 •£,o CD 0 c o ••• -—i CJ
Si >-D
s1
"CD — > O -C >" ^ "co
Is CO
•| g.
ss ss> CM CM CN LO r r LO r- r- CM t- CM 00 00 O O 0 0 O O o o
^. CD
%.
CO 0
"o
CJ
CD
tCD
CO
a. o c
CD
o
.c o
CO
c
o
o"- CO LO
•a m
Q_
^o
VL/
CO
+ O
0 0
0 a. a. CL Q- CC CL CL CL Q U Q Q Q
0 CL
U CL CL Q
CJ CL CL Q
o O
CJ U O U 0 CL CL _l CL CL CL
Q n Q QQ Q _i Q
Q- O
Q a_
(TJ
DC LU 0 CL
CD C CD
C
c
'E N
Q- .E 2 "5 c
Q
CD CD
^ i
CD" ^~ *J ^ c !5 H ^ "o. o o ra 3 X ''"' LU o a §E S- w o ro ^ '•M a a 2 ^ ">- o
gl <S
CD
w
CD O
O ^
^ CO 4—1
Q- c ? > CM" B l ^
§ Q.
S o '<£ c:
> CD X £
-i^ ^° .— JO > ^ >• ^ \ ° CD (D
£'•5 CJ fco '5 2 o >_o 2£ D
Q.
IT
"D < C 4--
^_
O O O) -C 00 ^^ — C Q. ^ ..- i_
O **-** r^j^ 5 2 CL 6 ^ -n S: CD C o (^ 0 o > '^ o -^ .c :- 0 cB n >- CD T3 "5
O ~
X
Q- !5 O *•• DC ^2
Q. 0
•?* 6 ^
0 0 0 O
Z Z z 'K n c c c c c 1- 1- N N N N N
zz
CD
0 0
CL CL c c
o o
CL CL C C
(0
uT uT
0 0 0 0 CL CL c c C C
CL a.
NNNNNNNN
.. -C O CO Q. CD
osome yiphos
Q" O CD L ,J_J
C CO CO V
c
>
=5 LUc"
"CD CD" a .c
JC Q.
—
c <^
CJ CD
Q. ^ W o
'c
fN ^ m •v 0
O)
Q -c _o
-f
N 2 CO
.2 cj •~ °r
E °
o
C
~^~ O) ^ D) O O
C>
£
— -S ^ a.
0 ^
LO LO LO LO LO O CM «- CM CN CM CN CM LO O O O O O
h aJ f—
c
'•3 ? CD CD S 0) CD CD CD CO C/J CO CO CO CO CO D D D D 15 0 0 £ ° O 0 I 5 ^2
JSl Q
M—
'55 3
a CD
£
'_J O
LU
co fp "n C ~r CD 2 N 4-1
"?rt
'*""'
•—^ o .- c o •Q S £
cB S c > ^ co < O o 0 Z
^ C CL N
'o ™ .2 0
CD 2
J2 "a CL
596
Lange
ministration of 0.26 mg/kg body weight of each photoactive compound, tumor/skin ratios were similar for all sensitizers, ranging from 3.0 (ZnPc) to 4.0 (ZnMA4). This indicates that presumably the nature of the delivery vehicle rather than the nature of the photosensitizer influences the targeting properties of these drugs. The biodistribution of bis(diisobutyloctadecylsiloxy)silicon-2,3-naphthalocyanine (isoBosinc) encapsulated in DPPC liposomes was tested in three different animal models [107-109]. At comparable isoBosinc doses of 0.5 mg/kg body weight for both tumor-bearing mice models, controversial results were reported. While tumor/muscle ratios for female BALB7c mice with MS-2 fibrosarcoma were found to be as high as 17 [107], for C57BL/ 6 mice carrying a B16 melanoma neoplasm, this value was reported to be only 1.1 [109], although approximately the same amount of photosensitizer was found inside the tumor (—0.35 yu-g/g tissue). Applied in emulsions using Tween as surfactant, about 0.6 yu,g/g tumor tissue of isoBosinc was found 24 hr after injection [108]. However, using this drug formulation, skin uptake with 0.42 jjLg/g was very high. Interestingly, serum clearance was rapid in all three cases ranging from 85% [107] to 95% [109] at 24 hr postinjection. Several other hydrophobic and amphiphilic photoactive compounds incorporated in liposomes were studied with respect to their biodistribution and pharmacokinetics. Tumor/muscle ratios and tumor/skin ratios of 16.7 and 3.4, respectively, were found when tert-w-propylporphyvene encapsulated in DPPC liposomes was injected into female BALB/c mice bearing a MS-2 fibrosarcoma [110]. The highest tumor/muscle ratio was obtained with tetraphenylporphyrin (TPP) and the corresponding TPP conjugated to a carotenoid moiety injected as CR emulsion into tumor-bearing mice [111]. Twenty hours postinjection tumor/muscle ratios were 97 and 27 for TPP and the tetracarotenoid TPP derivative, respectively. Morgan et al. [112] compared the PDT efficacy of tin(IV)etiopurpurin (SnET2) delivered by DPPC liposomes or with CR emulsions. They found that using 1 mg/kg body weight both formulations provide a 100% cure in rats implanted with urothelial tumors. However, at lower doses SnET2 in CR was more effective than liposomal SnET2. The high PDT efficacy of CR-mediated drug delivery was also documented by Woodburn and colleagues [113]. An increased tumor uptake of ketochlorin, caused by a longer circulation time in the blood, was probably the reason for the higher efficacy of CR-mediated ketochlorin delivery as compared with the corresponding Tween emulsion. Very recently, the biodistribution of hypocrellin A formulated in liposomes was compared with a corresponding dimethylsulfoxide (DMSO)/saline system in S-180 sarcoma-bearing mice [114]. It was found that the photosensitizer retention in the tumor was strongly dependent on the delivery vehicle. While for the liposomal system maximal tumor/peritumor ratios
Controlled Drug Delivery in Cancer
597
peaked at 12 hr (—4.0), the maximal value for DMSO-delivered hypocrellin A was significantly lower (2.58) and peaked at 6 hr postinjection. Furthermore, PDT studies showed that the therapeutic outcome of liposomal hypocrellin A was higher than that of hypocrellin A in DMSO/saline. Presumably, one of the principle mechanisms for the improved tumor localization properties of photoactive substances delivered by liposomes is the altered repartition on serum lipoproteins upon administration. As mentioned above, neoplastic tissues express a particularly large number of membrane receptors for LDLs. In vitro and in vivo studies have clearly demonstrated the role of LDL receptor-mediated endocytosis in the improved intracellular delivery of photosensitizers (for a more detailed review, see Sobolev et al. [32]). It was shown that 75% of Hp encapsulated in DPPC liposomes became associated with heavy lipoproteins, whereas this portion was only 10% for PBS-delivered Hp [99]. Similar behavior was observed for BPD-MA in vitro [115]. Determination of the plasma distribution following incubation of human blood with liposomal or DMSO/saline-delivered BPD-MA showed that most liposomal BPD-MA (>90%) was associated with lipoproteins, whereas DMSO/saline BPD-MA was nearly evenly distributed between lipoproteins and albumin. The authors used these differences to explain that the slightly higher PDT efficacy of liposomal-delivered BPDMA as compared with the free photosensitizer. Zn(II)-tetradebenzobarrelenooctabutyloxyphthalocyanine [ZnPc(OBu)8], another photosensitizing compound, was shown to bind to a high percentage to LDLs (71%), when delivered in CR micelles to tumor-bearing mice. However, most of the works reported above rely on the use of rodent animal models. Although it is widely accepted that human LDL receptors are similar to mouse LDL receptors, a lower affinity for heterologous vs homologous LDL cannot be ruled out. Furthermore, pronounced interspecies differences in the blood plasma protein pattern might cause some misinterpretation with respect to the clinical use of liposomal-delivered PDT and PD. Passive drug delivery exploiting the EPR effect evidentially requires that the drug or drug formulation demonstrates the ability to circulate for a long time in the blood to provide sufficient drug accumulation within the neoplastic tissue. Besides encapsulation in liposomes, extensive research and development efforts have concentrated on the alteration of drug properties through drug-polymer conjugates. The most investigated class of polymers used for this purpose are PEG conjugates. The prevalence of PEG conjugates in pharmaceutical research is presumably due to the fact that PEG is nontoxic, soluble in water and most organic solvents, and available in a wide range of molecular weights (typically 2-20 kD). A 50-fold increase in blood circulation half-lives has been observed for PEG-conjugated peptides and proteins [126]. Furthermore, in PDT the enhanced water solubility of the
598
Lange
photosensitizer-PEG conjugates facilitates the IV administration, primarily of hydrophobic photosensitizers. Probably the most complete documentation of the impact of pegylation on the biodistribution, pharmacokinetics, and PDT efficacy exists about the photosensitizing agent 4,10,15,20-(mesotetrahydroxyphenyl)chlorin (mTHPC). This molecule, currently under clinical evaluation [127,128], is characterized by an extremely high absorption at 652 nm and high phototoxicity. However, tumor selectivity in humans has been shown to range between 15 for T3 invasive bronchial SCC and only 2 for carcinoma in situ [129,130]. In order to improve these poor tumor localization characteristics of mTHPC, the hydroxy residuals of the native photosensitizer were substituted with PEG 2000 or PEG 5000 to give water-soluble molecules with molecular weights of about 9 and 21 kD, respectively. Interestingly, in most in vivo studies performed on different animal models [131-137], the pegylated derivatives were compared with the native mTHPC. The (PEG 2000)4-mTHPC was studied on an ovarian cancer model [131], on rabbits inoculated with cottontail rabbit papilloma virus (CRVP) as well as on healthy dogs. Ronn and colleagues have determined the elimination half-life of (PEG 2000)4-mTHPC at about 121 hours to be significantly longer than that of free mTHPC (r\P_ - 27.5 hr) [136], irrespective of the drug dose used. In this study, the rabbit skin photosensitization was found to always be lower when using (PEG 2000)4-mTHPC than for mTHPC under similar conditions. Furthermore, (PEG 2000)4-mTHPC was reported to be safer than mTHPC, with respect to the damage produced on a healthy dog's larynx following irradiation. This was attributed to the low muscle content of (PEG 2000)4mTHPC as compared with mTHPC. Using an ovarian rat model, Hornung et al. [131] have shown extremely high T/N ratios, ranging between 21 (tumor/muscle) and 1.4 (tumor/ liver) 4 days after IV administration. These values are similar to those observed in nude mice grafted with a human colon carcinoma using the higher molecular weight PEG derivative [132]. However, probably due to the prolonged plasma half-life, tumor uptake was found to continuously increase up to 10 days postinjection, resulting in tumor to muscle ratios as high as 67 in the ovarian cancer model. Previously, Morlet et al. have shown a tumor/muscle ratio of 7:1 in human adenocarcinoma bearing mice 24 hr after IP injection of (PEG 2000)4-mTHPC [136J. In three successive papers, Ris et al. [133-135], working with (PEG 5000)4-mTHPC in nude mice bearing human mesothelioma, adenocarcinoma, and squamous cell carcinoma as well as healthy minipigs, showed effective photodynamic effects with this high molecular weight photosensitizer in comparative studies with mTHPC. In mesothelioma as well as in
Controlled Drug Delivery in Cancer
599
squamous cell carcinoma xenografts, (PEG 5000)4-mTHPC was reported to be superior to free mTHPC with respect to tumor necrosis depth. Furthermore, necrosis of healthy tissue in the bronchi and thoracic cavity of minipigs was significantly less with (PEG 5000)4-mTHPC than with mTHPC [134,135]. Skin photosensitization was not remarkable 12 hr after injection of (PEG 5000)4-mTHPC and irradiation with a light dose of 200 J/cm2 at 652 nm [134], and less damage to muscular structures was also observed in this study. This fact is probably the reason mice bearing human colon carcinoma died after irradiation with 20 J/cm2 when injected with mTHPC, whereas the same irradiation conditions were well tolerated after injection with (PEG 5000)4-mTHPC [132]. In this study, although less homogeneously distributed within the tumor tissue than mTHPC, (PEG 5000)4-mTHPC was reported to exhibit similar PDT efficacy with respect to delayed tumor growth. 5.
METABOLISM-MEDIATED DRUG DELIVERY IN PDT AND PD
Instead of targeting antigens or cell surface receptors, in a different kind of "active targeting" one can also exploit metabolism-related differences in neoplastic cells for the management or the diagnosis of cancer. Due to deficiencies in tumor-associated angiogenesis, large sections of sizable tumors can be left without adequate vascularization. The subsequent lack of nutrients and oxygen forces the cells to produce energy by glycolysis, resulting in overproduction of acidic byproducts. To maintain near-normal intracellular pH, cells actively transport protons to the extracellular space, which becomes acidified. Hence, pH-sensitive fluorescent dyes can be used to visualize solid tumors, producing high amounts of protons. Notably, fluorescein derivatives such as 5,6-carboxyfluorescein [138], 5'- (and 6'-) carboxyseminaphthofluorescein (C-SNAFL-1) [139], and 2',7'-bis(carboxyethyl)-5,6-carboxyfluorescein [140-142], having pKa values around 7, have been used to examine tumor pH in vivo. However, application of such pHsensitive dyes for the fluorescence photodetection of human disease remains unexplored. It is thought that some tumors cause a decrease in extracellular pH to retain secreted protease activity for basement membrane digestion as one of the initial steps for metastasis. This fact, together with the observation that in some tumors, especially those with fast-growing aggressive cells, specific enzymes are overexpressed, leads to a second approach in metabolism-related drug delivery. Upon administration, the drug carrying photoactive substances is distributed throughout the body. Predictable enzymatic reactions then convert the original form by removing some protective functions or
600
Lange
cleaving the photoactive compound to form a metabolite with significantly different photophysical properties. Cathepsin D is an abundant lysosomal aspartic protease distributed in various mammalian tissues. In most breast cancer tumors, cathepsin D is found at levels 2- to 50-fold greater than levels found in fibroblasts in normal mammary glands [143]. A drug containing an amino acid sequence that is recognized by cathepsin D can be cleaved at this position. Consequently, an active substance can be released. An example for a cathepsin D-sensitive spacer is the oligopeptide Gly-Pro-Ile-Cys-PhePhe-Arg-Leu-Gly [144]. Based on this motif, Weissleder and colleagues [145-147] have used a long-circulating, synthetic graft copolymer bearing Cy5.5 fluorochromes (see above) positioned on the cleavage substrate. Due to fluorescence quenching, when the copolymer is loaded with 11 -55 fluorescent entities, the reporter probe in its native state is nearly nonfluorescent but becomes brightly fluorescent when the fluorescent units are released by cathepsin D cleavage. After IV injection in mice bearing a BT-20 mammary adenocarcinoma, tumors became highly fluorescent within 12-48 hr, attaining a maximum at 24 hr postadministration. T/N ratios were highest for tumor/muscle (—120) and lowest for tumor/blood (~11) [147]. 5.1
Controlled Drug Delivery in ALA-Mediated PDT and PD
The most important approach in the PDT and PD of cancer using specific deficiencies of the metabolism of neoplastic cells is based on the systemic or topical administration of 5-aminolevulinic acid (ALA). ALA is not a photoactive substance by itself, but forms part of a substrate in the biosynthetic pathway of heme, the iron(II) complex of protoporphyrin IX (PpIX) (see Fig. 8). In contrast to heme, PpIX is a fluorescent molecule with a 'O2 quantum yield of approximately 0.5 [148], which makes it suitable for PDT and PD. Almost all nucleated cells in mammals exhibit the ability to produce PpIX. The route by which cells produce PpIX endogenously forms part of the overall scheme for the production of chemical energy. Nowadays the biosynthetic pathway of heme is relatively well understood and has been reviewed recently by Peng and coworkers [149]. In brief, the initial step in heme biosynthesis is the enzymatically catalyzed formation of ALA from glycine and succinyl-CoA inside the inner mitochondrial membrane. Following the entry of ALA into the cytosolic space, ALA dehydrase induces the condensation of two ALA molecules to form porphobilinogen (PEG). Subsequently, PEG deaminase (PBG-D) and uroporphyrinogen III cosynthase catalyze the cyclization of four PEG molecules to form the tetrapyrrolic skeleton. Finally, a series of decarboxylations and oxidations inside the cytoplasm as well as in the mitochondria have to take place before PpIX is
Controlled Drug Delivery in Cancer
IS II 0
>
w
r- +*
.E _CO
^ £ i3 03 - ro £
o 0 0 ^
.h "O T3 o
° o 0« ^
C
'-M
_J flj 0 ^ £
=^
C ja J=.
o o
o LU ^
03 O) C
m >w' J2 0) 'C
' 3 -C QJ (!) = O)--
x: 0 • ^1 o +^ o - - t
.-Q •-0 o c^'E
-
Q-
c
O
. c
w O
'*= "o °•CO ° 0 4-
•M _ -t-1 n C Q "
0 £ „ C
w
c = 2
I851 0 _r- . • 0 ~ CO > -
00
§
UJ
^0
O ii
® 'R rO? C ^ M-
OC
0
C
LU +3
oj X
-tr-l £2
-
CD -M D_ W
CD O
602
Lange
formed, by the removal of six hydrogens from the protoporphyrinogen IX, catalyzed by protoporphyrinogen oxidase, which is embedded in the inner mitochondrial membrane. Ferrochelatase, also located in the inner mitochondrial membrane, catalyzes the incorporation of iron into the PpIX cycle to form nonfluorescent heme. Heme biosynthesis is regulated by numerous control mechanisms, among which is the negative feedback control on the ALA synthase (ALA-S) activity as well as on the transcription, translocation, and transport of the enzyme into the mitochondria. By exogenously providing an excess of ALA one can bypass this regulatory step, thus allowing the production of heme and its various intermediates at rates that are primarily limited by the activity of enzymes involved in heme biosynthesis and the amount of intracellularly available ALA. Although PpIX formation is present in nearly all nucleated cells, its preferential accumulation and generation have been reported in tissues known for a high cellular turnover. Kinetically, the slowest step between the uptake of ALA and the formation of PpIX must be faster than the step by which PpIX is converted to heme in order to accumulate high amounts of PpIX in a cell exposed to ALA. At present, little is known about the exact mechanisms responsible for this difference in the response to exposure to ALA in normal and neoplastic cells. Experimental evidence has been found that in some tumors the ferrochelatase activity is lower, while those of ALA-D and PBG-D were higher than in normal cells [150,151]. Furthermore, Krieg et al. have shown that the iron content in urothelial cancer cell lines is significantly smaller than in normal cell lines [152]. However, besides a relatively good tumor selectivity, ALA has several supplementary advantages with respect to side effects that generally accompany PDT and PD. Although some patients suffer from mild, transient nausea and/or transient abnormalities of liver function after systemic application, it appears that at a dose lower than 60 mg/kg (oral) or 30 mg/kg (IV) no neurotoxic symptoms are observed. Moreover, after topical application of ALA, porphyrin plasma levels return to normal within 24 hr. Generally, ALA-mediated PpIX fluorescence cannot be detected 24 hr after topical application, significantly reducing the risk of severe skin photosensitization. These two advantages, i.e., high selectivity and absence of severe side effects, led to clinical trials of ALA-induced PpIX for phototherapeutic or photodiagnostic purposes in several medical fields, namely, dermatology [153], urology [154-156], gastroenterology [157], otorhinolaryngology [158,159], gynecology [160,161], neurosurgery [162], and cardiology [163]. The clinical use of ALA for PDT and PD has been discussed recently by Peng et al. [164] and Marcus et al. [165]. Despite promising progress in several medical domains, ALA-based treatments still lack the wide acceptance of the medical community and the approval of regulatory authorities.
Controlled Drug Delivery in Cancer
603
The only exception is the recent U.S. FDA approval of ALA-mediated PDT of actinic keratosis (AK). On the basis of some typical clinical aspects of this disease, one can explain some drawbacks that limit the use of ALA for PDT and PD for other pathologies: 1.
AK most commonly occurs in fair skin that has been exposed to sunlight. Due to exposure the direct, physical delivery of ALA, e.g., in oil-water emulsions, onto the diseased area is possible. Topical application is highly preferential to systemic administration because the drug concentration is increased in the target tissue and reduced in the surrounding tissue. However, local delivery requires a contact between the drug and the target tissue over considerable time periods, which could be difficult for other diseases, such as esophageal, gastric, or lung cancer. 2. AK is a flat, precancerous lesion that develops in approximately 20% of all cases into squamous cell carcinoma. For effective PDT treatment, a sufficiently homogeneous distribution of the photosensitizer in the entire neoplasm is desired. The low complete response (CR) rates (32-64%) in nodular basal cell carcinoma [164] are presumably due to the inhomogeneous distribution of resulting PpIX [166,167], while CR rates for ALA-PDT of superficial AK range between 81% and 100% [164]. 3. AK is most likely to appear on the face, ears, bald scalp, neck, back of hands and forearms, and lips. Such areas can be exposed to ALA for long periods in order to induce sufficiently high amounts of PpIX without impairing the patient's comfort. However, prolonged topical application times are either difficult (e.g., urology where long instillation times are not appreciated by the patients) or impossible (e.g., in gastroenterology). In summary, this modality seems to be limited by the amount of ALA that enters the target cells and penetrates into the interstitial space of the target tissue. Moreover, the resulting generation of PpIX is often too small, too slow, or too unevenly distributed to provide effective treatment or reliable fluorescence PD. This part of the chapter focuses on methods in ALA-based PDT and PD, adapting concepts of controlled drug delivery in order to overcome problems associated with the poor bioavailability of ALA. Considerable efforts have been made to circumvent these drawbacks by modifying ALA formulations [168—183], using physical methods [184-188], or derivatiza-
604
Lange
tion into more lipophilic prodrugs of ALA [189-207]. Table 5 summarizes these methods and relevant rational aspects for the corresponding concept. 5.1.1
Chemical Enhancement of ALA Drug Delivery
The simplest approach to enhancing the penetration depth of ALA and thus presumably the formation of PpIX is to embed DMSO in the corresponding formulation. DMSO is known as a potent skin penetration enhancer [208] and, at certain concentrations, an activator of various enzymes in the heme biosynthetic pathway [209,210]. Furthermore, DMSO has been found to stimulate cell differentiation [211]. Since the stratum corneum represents the major barrier of the epidermis (see below) to the penetration of topically applied compounds, the concept of adding 2-20% of DMSO (wt/wt) as a penetration enhancer has been widely accepted for ALA-based PDT in dermatology. Malik et al. found a significantly higher production of PpIX in mouse skin after application of ALA along with 2% DMSO as compared to ALA alone [168]. Recently, de Rosa and colleagues determined quantitatively the skin permeation of ALA in vitro as a function of DMSO concentration [169]. They found that the addition of 20% DMSO significantly increased the flux of ALA across the skin as compared with ALA or formulations containing only 10% DMSO. Furthermore, a 2.5-fold increase of the amount of formed PpIX was reported after in vivo application of oilin-water emulsions containing 10% ALA and 20% DMSO (wt/wt) as compared to the corresponding formulation without DMSO. However, it seems that this concept is not only applicable for dermatological purposes. An 40% increase in PpIX fluorescence intensity in ex-
TABLE 5 Concepts and Rational for Controlled Drug Delivery of 5-Aminolevulinic Acid Concept
Method
Rationale
Chemical enhancement
DMSO EDTA/DFO Liposomes 1,10-Phenatroline
Physical enhancement
Gels Iontophoresis Spray Derivatization
Enhanced penetration Chelating of iron Enhanced penetration Modulation of heme biosynthesis Prolonged contact Enhanced penetration Homogeneous distribution Enhanced penetration Increased uptake Selective cleavage
Prodrugs
Controlled Drug Delivery in Cancer
605
cised pig bladder mucosae was observed when DMSO was added to solutions containing 180 mM of ALA [170]. While DMSO seems to enhance the uptake of ALA and thus increases the PpIX formation by an improved delivery of the substantial substrate, a further increase of PpIX accumulation is expected to result from removal of iron ions, the key substrate for conversion of PpIX into heme. This can be achieved by the addition of ironchelating substances, such as ethylenediaminetetraacetic acid (EDTA), desferrioxamine (DFO), and 3-hydroxypyridin-4-ones (HPOs). It has been shown in vitro that the iron-specific DFO is superior to nonspecific EDTA at equal concentrations with respect to PpIX formation [171]. Depending on the ALA concentration, DFO was found to enhance the PpIX accumulation by a factor of up to 44 in V79 Chinese hamster lung fibroblasts when compared with ALA alone. A significant increase of PpIX fluorescence in mouse skin after the addition of EDTA was also observed by Malik et al. [168] and de Rosa et al. [169]. In agreement with Orenstein et al. [172], Warloe et al. [173] found a positive impact on the therapeutic outcome in nodular basal cell carcinomas when ALA was given along with EDTA/DMSO. In a large number of PDT treatments of basal cell carcinomas with ALA cream containing DMSO and EDTA, the CR rate was not improved in the case of superficial basal cell carcinomas but significantly increased in nodular lesions. However, the actual role of EDTA in clinical treatment is still under discussion. Again, the positive influence of iron-chelating agents cannot only be observed in dermatology. Using different iron-specific chelators, Marti et al. [170] and Chang et al. [174] found that PpIX production can be doubled in urothelial mucosa in vitro and in vivo after topical application of ALA solutions containing DFO and HPO, respectively. Interestingly, as found in vitro [175], 1,10-phenanthroline, a porphyrin biosynthesis modulator, has been shown to increase both the ALA-mediated PpIX accumulation and the subsequent photodestruction in BALB7c mice with solid Meth-A tumors [176]. 5.1.2
Physical Enhancement of ALA Drug Delivery
As mentioned earlier, in addition to chemical methods, physical methods can be used to enhance the penetration of ALA through biological barriers. Iontophoresis can be described as a process that facilitates the transport of ionic species, including charged molecules, by the application of electric current [212]. This technique uses a small electrical current to facilitate the drug transfer across the skin. In a typical setting, two electrolyte chambers with electrodes are placed on the surface and a constant current is applied. Among others, polarity, valency, and ionic mobility of the permeant as well as the composition of the delivery vehicle influence the efficacy of the ion-
606
Lange
tophoretic process. For a given drug (or formulation) the transport across the skin is directly proportional to the magnitude of the current, thus providing an efficient way to efficiently control the delivered amount of the drug. One can distinguish two different mechanisms, electromigration and electro-osmosis, independently contributing to the flux of the drug through the skin barrier. In electromigration the iontophoretic flux is caused by the electrical potential gradient applied to the charged species. On the other hand, electro-osmosis results from a net flow of small, positively charged species (mostly N a ' ) from the anode to the cathode colliding with the drug, thus literally carrying it through the skin. The latter mechanism becomes more and more important when the molecular weight of a positively charged molecule increases to 1000 or if the molecule is neutral. The transport of ALA, carrying a positive and negative charge at physiological pH, is probably slightly enhanced by this mechanism. Rhodes et al. [180] have used this technique to deliver different doses of ALA by applying different charges, ranging from 3-120 mC to a solution with 2% of ALA into the skin of healthy volunteers. They found that PpIX fluorescence correlated linearly with the applied ALA dose. Charges higher than 24 mC resulted in a delayed maximum of PpIX fluorescence intensity and prolonged presence of PpIX in the skin than lower charges, presumably due to an enhanced penetration. Interestingly, in this study, remarkably less ALA was used (2% [180] and 1% [181 ]) than in standard protocols with passively diffusing ALA [154-156,164]. Besides the enhanced penetration of ALA into the interstitial space of tissue, in PDT and PD of cancer it is often important to physically deliver the drug to a particular site of interest. In these cases, that drug has to be homogeneously distributed throughout the entire area and a good contact between the tissue and the drug must be warranted for sufficient time. However, in medical practice these two basic requirements are often difficult to accomplish due to the architecture of the organ in which PDT and/or PD will be performed. In gynecology, the prolonged retention of aqueous solutions of ALA after intravaginal delivery can be difficult due to the limited tightness of this organ. In other cases, such as gastroenterology or otorhinolaryngology, the esophagus or the bronchial tree is not easily accessible for topical drug administration in a minimally invasive way. Hillemanns et al. have added propylene glycol to an aqueous solution of ALA in order to increase the viscosity and to tighten the contact between the mucosa and the drug formulation for PD and PDT of cervical intraepithelial neoplasia (CIN) [182,183]. Although PDT after topical application of this ALA formulation has shown no improvement of CIN lesions in 7 women, PD together with spectroscopic measurements was reported to be more specific (75%) than standard colposcopy (specificity: 50%).
Controlled Drug Delivery in Cancer
607
Due to problems with respect to the controlled topical drug delivery of ALA, it has alternatively been applied orally. Although this method has been reported to be highly sensitive for the photodetection of Barrett's esophagus [213-215], the specificity was rather low (—30%). This fact is presumably due to nonspecific PpIX formation in underlying mucosal tissue, caused by the systemic uptake of ALA. Topical application would be highly favorable in such cases in order to release the drug only to superficial cells of the gastrointestinal mucosa. For this purpose, systems are needed that deliver ALA at a constant rate over an extended period to the esophageal wall. This can be realized by using bioadhesive release systems in which the swelling of a hydrogel controls the drug release. Patrice and colleagues have analyzed four different bioadhesive gels: Noveon AA-1, a poly(acrylic acid); keltron T, a polysaccharide; blanose; and lutron F127, a poly(oxyethylene)-poly(oxypropylene) copolymer, laden with different quantities of ALA with respect to their PpIX formation capacity in vivo [185]. Using both ALA-laden Lutrol and Noveon a clear PpIX buildup in murin gastric mucosa was demonstrated. For the latter a 92% increase in PpIX fluorescence intensity was observed as compared with aqueous solutions of ALA. This fact was attributed to longer transition times. Hence, the use of such bioadhesive gels might improve the photodetection of gastric cancerous and precancerous diseases due to higher PpIX fluorescence. Bioadhesive gels can be also useful in other medical domains, such as gynecology, because Noveon in particular has been developed for topical vaginal and oral delivery. Moreover, such gels are mostly transparent, thus interfering minimally with light used for photodiagnostic or phototherapeutic purposes. In contrast to the urinary bladder or the human skin, the tracheobronchial tree is not easily accessible to topical drug administration. Thus, for ALA-mediated fluorescence photodetection of lung cancer or laryngeal neoplasms researchers are confronted with totally different challenges. In this case, inhalation is the most attractive and least invasive for the delivery of ALA to the respiratory tract. Dry or liquid particles can be prepared and inhaled with the aid of dry-powder dispersers, liquid-aerosol generators, or nebulizers. Such devices produce particles that typically range in size from 1 jam to >10 jam. Depending on the diameter of the inhaled particles, they are trapped in the nasal passage, throat, larynx, bronchial walls, or alveoli. Particles greater than 10 jam in diameter are typically deposited in the nasal cavity; those of about 1-2 /jum in diameter will reach the alveoli. Although the alveoli provide a large surface area (80-140 m2) and thus are an attractive target for the systemic drug delivery of Pharmaceuticals, in the case of ALA-mediated PpIX a maximum of ALA should be deposited in the upper
608
Lange
tracheobronchial tract because lower areas are not accessible to endoscopic examination. Several parameters, including the physical device itself, the drug concentration and formulation, the volumetric flow rate, and the mass flow at the mouthpiece, will influence the pattern of the deposited drug dose and the distribution of the drug throughout the pulmonary system. Baumgartner and colleagues have tested two different inhalation devices with different deposition characteristics [184,186-188J. The commercially available PariBoy (PARI, Starnberg, Germany) giving an average droplet size of between 5 and 8 yum in diameter was reported to deliver only 20-26% of the inhaled solution to the upper part of the bronchial tree, whereas another device, especially designed for this purpose, deposited 7580% of the inhaled solution to this part of the lungs [184]. In a clinical phase I study, more than 100 patients were enrolled for the fluorescence photodetection of early cancerous lesions in the tracheobronchial tree [186,187]. For this purpose, ALA [10% to 4% (wt/wt)] was dissolved in an isotonic saline solution and patients were asked to inhale for 30-40 min. The fluorescence photodetection following inhalation of ALA solutions was reported to be very sensitive but the specificity was only 30-50%. However, this technique seems to be suboptimal with respect to the tissue characteristics in the upper respiratory tract for several reasons. The mucus, a highly lipophilic protective layer, represents a major barrier to hydrophilic substances. This can result in a shallow and inhomogeneous PpIX distribution, causing a high number of false-positive results. Usually, surfactants such as poly(oxyethylene)ethers or poly(glycol)ethers are added to prevent this effect. However, ALA is barely soluble in such organic solvents, which limits the use of surfactants. Furthermore, when ALA is dissolved in isotonic saline solution it results in a highly acidic solution, which may block PpIX biosynthesis [216,217] and cause coughing. The same group used a similar technique for the fluorescence photodetection of laryngeal neoplasms in 16 patients [188]. According to their preliminary results, a sensitivity of 95% and a specificity of 80% were established using this method. 5.1.3
Chemically Modified ALA for the Enhancement of ALA Drug Delivery
The most important progress in controlled drug delivery of ALA has recently been made by derivatization of ALA following the prodrug concept. According to Albert [218], a prodrug can be defined as a pharmaceutically inactive substance that undergoes spontaneous or enzymatic transformation, thus releasing the active drug. This approach aims at modifying the physicalchemical properties of a drug by covalently attaching a transport moiety,
Controlled Drug Delivery in Cancer
609
resulting in improved access to pharmacological targets. Prodrugs have been designed for several purposes, such as to overcome problems including incomplete absorption, too rapid clearance, or too rapid absorption as well as poor bioavailability. In order to understand the prodrug concept with respect to an enhanced delivery of compounds to epithelial tissues let us consider the skin as a model system. The limiting factor for the delivery of substances into or through the skin is the skin itself. Roughly, the skin can be subdivided into three principal layers: subcutaneous tissue, dermis, and epidermis. The latter represents the skin's major barrier function. This function is nearly entirely accomplished by the outermost few micrometers of the epidermis—the stratum corneum (SC). This extremely thin and least permeable skin layer is the ultimate stage of the epidermal differentiation process, forming several layers of dead cells (keratin-filled corneocytes) that are embedded in a lipid matrix. The rate-limiting step in epidermal drug delivery after topical application is the passive diffusion of drugs through the SC. In general, the ability of a molecule to cross biological barriers is one the most important parameters in drug delivery that governs the absorption, distribution, metabolism, and excretion of a drug. The drug has to partition between the mostly aqueous phase of biological fluids and lipid biomembranes, whose major constituents are phospholipids, cholesterol, sphingolipids, and glycolipids. All of these components are of amphiphilic nature. Thus, in order to successfully cross these barriers, any drug should have an appropriate balance between hydrophilic and lipophilic properties. The octanol/water partition coefficient P or, more likely, its logarithm (log P) is commonly used to describe the chemical's lipophilic properties. Although there is no general rule that can be applied across the vastly diverse drug molecules, some general considerations can be made. Within a homologous series of drug molecules, drug absorption usually increases as lipophilicity rises and is maintained at a plateau for a few units of log P after which there may be a steady decrease, resulting in a parabolic relation. A similar relationship has been found between log P and biological activity. In general, compounds with log P values below 0, i.e., hydrophilic, have a good water solubility but may have poor permeability, whereas drugs with a log P value far higher than 3 tend to favor absorption. Indeed, most of the disadvantages accompanying the topical use of ALA, especially with respect to its poor bioavailability, can be ascribed to the physical-chemical properties of this molecule itself. It can be calculated that under physiological conditions about 90% of the applied ALA will be present in its zwitterionic form, carrying a positive charge at the amino group and a negative charge at the deprotonized charboxyl group. Thus, the zwitterionic, hydrophilic ALA is far from being optimal with respect to the
610
Lange
above-mentioned conditions. Consequently, its penetration through the lipid bilayer of biological membranes and into the interstitial space of tissue is not favored, limiting the resulting PpIX to superficial cell layers with a high degree of heterogeneity. Furthermore, high doses and long application times are often necessary to induce sufficient amounts of PpIX to the target site. Chemically, the simplest way to control the lipophilicity of an acid is by replacing the negatively charged carboxylic function by esterification with an alkyl rest. Assuming that lipophilicity is one of the key parameters in bioavailability for ALA derivatives, this parameter can be varied, e.g., in the case of n-alkyl esters by varying the chain length of the corresponding alcohol. This occurs in positively charged, lipophilic compounds, which may be associated with the high negative charge on the surface of cells with a high membrane potential. Although most work on ALA derivatives has been carried out on ALA alkyl esters, some more exotic derivatives have been tested, mostly in cell culture [202,203]. Uehlinger et al. [192] first determined the log P values of a homologous series of ALA alkyl esters. In agreement with De Rosa et al. [169], they found a negative log P value for ALA (about — 1.5), which was slightly enhanced ( — 0.9) for the ALA methyl ester (m-ALA). All other derivatives were found to be lipophilic in nature. It has been shown that, by simple esterification, P can be varied by more than four orders of magnitude between ALA and ALA octyl ester (o-ALA). However, the magnitude of P is not the only important parameter in terms of bioavailability. Substances that are too lipophilic can be captured in the SC, thus losing their ability to induce PpIX synthesis within the target tissue. As mentioned above, for an optimized flux across the SC a balanced water and lipid solubility is required. Drug absorption can be described by Pick's first law of diffusion: dQ/dt = J dlug = ( P - D / h ) - A c - A wherein dQ/dt is the amount of drug diffused per time unit; J dlim the drug flux, P the partition coefficient, D the diffusion coefficient of the drug, h the pathlength of the barrier, and A the treated skin surface area. In order to enhance the drug delivery the manipulation of any of these parameters may influence the flux of the drug through the SC. However, it should be borne in mind that the alteration of one component might influence another factor in this equation. At the first view the increased partition coefficient of highly lipophilic ALA esters would increase J dlul , across the SC but can in turn also increase its affinity to the administration vehicle, e.g., in the case of a lipophilic water-in-oil emulsion. The overall effect in such a case might be balanced or even negative. Several authors have evaluated the influence of log P on the epidermal PpIX formation in living human and rat tissue in vitro [190] and in vivo
Controlled Drug Delivery in Cancer
611
[193-195]. Using nude mice, in one in vivo study the PpIX formation capacity of ALA hexyl ester (h-ALA) was compared with those of ALA under different experimental conditions [194]. In this study, the authors reported that PpIX formation after a short-term topical application of ALA was higher in comparison with h-ALA. However, comparing the applied drug doses on the basis of weight percentage (wt/wt) is somewhat misleading. Due to the considerably higher molecular weight of h-ALA (Mr = 251.7 g/mol) as compared with ALA (Mr = 167.5 g/mol), in terms of molality the effective, available drug content is significantly smaller in the case of h-ALA. Continuous application of ALA and h-ALA-containing creams over a period of 24 hr showed slightly higher PpIX levels for h-ALA than the corresponding ALA-induced PpIX for PpIX precursor contents higher than 0.5% (wt/wt). The authors found that when the mouse skin was tape-stripped prior to application of ALA or h-ALA, the maximal PpIX fluorescence intensity increased dramatically and reached the same level after ALA and hALA application. In this case, the maximal fluorescence intensity was reached earlier with h-ALA than with ALA using equal drug contents of 10% (wt/wt). From these observations, the authors concluded that the SC is a more effective barrier against h-ALA than for ALA. Using healthy mice, however, the principle problem, i.e., improved penetration into nodular lesions, was not addressed in this study. Dognitz et al. [195] have addressed this problem by determining the PpIX depth distribution following topical application of creams containing 20% of ALA (wt/wt) and 1 % of h-ALA (wt/wt), respectively, to humans with nodular basal cell carcinomas and squamous cell carcinomas. Under their experimental conditions, they found that h-ALA showed no significant advantages as a PpIX precursor as compared to ALA with respect to PpIX fluorescence intensity, penetration depth, and distribution. Interestingly, although they used about 30-fold lower concentrations of h-ALA than ALA and the SC was not removed prior to application, similar PpIX formation capacities were observed for both precursors. The only derivatives that have been successfully tested for improved PpIX formation after topical application to skin in vivo are ALA ethyl ester (e-ALA) and ALA propyl ester (pr-ALA) [197], which have shorter chained alkyl substituents and thus lower lipophilicity. Moreover, application of mALA and subsequent irradiation resulted in a stronger growth inhibition in WiDr tumor-bearing mice as compared with ALA [164]. Fritsch et al. [198] have compared the biodistribution of PpIX after topical application of ALA and m-ALA in human AK in vivo. Although PpIX formation was found to be more efficient when using ALA than m-ALA with respect to PpIX generation, the selectivity for the diseased tissue was greater for m-ALA. The impact of concentration and drug formulation on PpIX accumu-
612
Lange
lation and selectivity was recently studied by Batlle and colleagues [219] on chemically induced skin cancers in mice. In their study, the optimal delivery vehicle for the acid was a lotional formulation with 10% DMSO (wt/wt) containing 150 mg/mL of ALA, whereas for h-ALA a 30% (wt/wt) cream gave the best results. With both formulations similar PpIX levels were obtained in the tumor tissue. However, in terms of selectivity the h-ALA formulation showed several advantages over ALA since PpIX synthesis was found to be confined to the application site and significantly less PpIX was produced in neighboring healthy tissue. Again, these studies indicate the importance of a good balance between high P and retained water solubility, which should be adapted to the specific application. In dermatology, it seems that short-chain ALA alkyl esters have more favorable properties with respect to PpIX formation and distribution and PDT efficacy than long-chain ALA alkyl esters. As discussed before, besides penetration enhancers, iontophoresis can be used to pass substances across the SC with more ease. In particular, ALA esters carrying only one positive charge are supposed to be transported more efficiently by electromigration through the SC. It has been demonstrated that 50 times more m-ALA and 15 times more h-ALA can be transported through the SC of porcine skin than ALA by iontophoresis [196]. Gerscher et al. [193] have used iontophoretic methods for dosing ALA, ALA butyl ester (bALA), and h-ALA for PDT studies in the skin of healthy human volunteers. They observed, in agreement with Ref. 194, that h-ALA-induced PpIX peaked earlier in comparison with ALA-induced PpIX at comparable precursor concentrations. h-ALA was found to be slightly more phototoxic than ALA and more homogeneously distributed as determined by means of fluorescence microscopy. Besides an enhanced penetration of ALA derivatives into the target tissue, their uptake characteristics and their ability to be metabolized into PpIX play an important role if a particular prodrug can be considered as a potential alternative to replace ALA. In their first article on the clinical use of h-ALA for the PD of early human bladder cancer, Lange et al. [206] supposed that the ester has to be cleaved into its native compounds prior to entering the heme biosynthesis. There is experimental evidence that at least in part ALA esters must be cleaved to act as substrate for ALA dehydrase, the enzyme that converts two molecules of ALA to porphobilinogen. Taylor et al. found that no porphobilinogen synthesis was observed in vitro in the presence of ALA esters and ALA dehydrase [220]. However, the addition of ALA to these solutions resulted in formation of ester analogues of porphobilinogen, indicating that in the enzyme-catalyzed reaction an ALA ester may provide one substrate molecule for ALA dehydrase, the other being ALA itself.
Controlled Drug Delivery in Cancer
613
There are several indications that ALA primarily gains access to the intracellular space via active transport through the cellular membrane. Diand tripeptide transporters, such as the PEPT1 transporter, were identified as possible transporter system for the uptake of ALA into rat pancreatoma cells [200]. In addition, Doring et al. [221] have identified the PEPT2 transporter system for the transmembrane transport of ALA into the epithelial cells of kidneys. In spite of an evident resemblance beween ALA and yaminobutyric acid (GABA), they found that ALA but not GABA competes with the active uptake of PEPT1 and PEPT2 into the cell. However, using a human adenocarcinoma cell line (WiDr), it was shown that GABA and other structurally related compounds, such as taurine and /3-alanine, which are transported by system j3 transporters, effectively inhibit the uptake of ALA [199]. Furthermore, other amino acids, in particular those with polar, uncharged groups, were found to interfere with the uptake mechanism of ALA. In contrast to ALA, the transport of m-ALA was only slightly influenced by j8-alanine and no impact was reported in the case of h-ALA in the same study [199], suggesting a predominant diffusion of this highly lipophilic molecule. This assumption is further supported by the finding of Whitaker et al. [200] that the PEPT1 transporter actively carries ALA but not ALA esters through cellular membranes. However, other transporter systems could be used by ALA esters to gain access to the intracellular space. Nonpolar amino acids were found capable of inhibiting the uptake of m-ALA by about 60% [201]. On a cellular basis, the improvement of ALA-induced PpIX generation following the prodrug concept is dominated by two independent processes: (1) the rate of enhanced drug uptake; and (2) its rate of enzymatic conversion into PpIX. Applied at equimolar doses to intact and lysed T-cell lymphoma cells, the PpIX formation after exposure to ALA was compared with the PpIX formation after exposure to a homologous series of ALA alkyl esters [202]. In agreement with other authors [170,192,205], it was found that short-chained ALA alkyl esters led to less PpIX fluorescence than ALA or long-chained ALA alkyl esters in intact cells. In this report, ALA pentyl ester (p-ALA) was the most efficient derivative in respect of optimal PpIX formation. Prodrugs beyond the lipophilicity of p-ALA exhibited less PpIX fluorescence intensity. The authors attributed this fact to an increased capture of too lipophilic prodrugs within the cellular membrane. However, cell culture experiments of other authors have proved that, using ALA or one of its derivatives, a strong, biphasic, dose-dependent PpIX accumulation occurs [192,205,222,223]. PpIX generation was found to be positively correlated up to an optimal prodrug concentration copt, where highest PpIX fluorescence was observed, beyond which PpIX generation sharply decreased. The absolute value of copt varied with the type of the prodrug and the cell line.
614
Lange
This drop-down of PpIX generation at higher doses than copt was most likely explained by cytotoxic effects of the particular prodrug or its cleavage products [192,202]. Due to the high trans membrane penetration potential of several prodrugs, it can be assumed that the cytoplasm of cells is charged with high amounts of these derivatives. Subsequent to cleavage by nonspecific esterases, high amounts of ALA, toxic for the cell [171,192], are liberated. Hence, a good equilibrium among cell membrane penetration, saturation of heme biosynthesis, and cytotoxicity must be found for each application. In chemistry, methyl substituents are often used to protect the carboxylic function of organic acids against unwanted reactions. This is primarily due to the ease of cleavage of this ester function. Thus, it can be assumed that, once present intracellularly, m-ALA will be one of the most efficient prodrugs of a variety of ALA derivatives with respect to the rate of PpIX synthesis. This has been confirmed by Kloek et al. [202] on cell lysates exposed to equimolar concentrations of various ALA alkyl esters. They reported that the rate of enzymatic conversion of ALA alkyl esters to PpIX decreased with increasing chain length up to b-ALA. Higher ALA esters, with the exception of h-ALA, which showed a significantly enhanced activity, resulted in approximately the same PpIX levels as induced by b-ALA. The observations made in vitro have some important consequences with respect to their potential clinical application. First, short-chain ALA alkyl esters exhibit less PpIX formation than long-chain esters, presumably due to a less effective uptake. However, for the management of diseases with biological barriers for overly lipophilic compounds it is recommended that such compounds be considered also. Second, some cell lines seem to have certain specificity for the cleavage of particular prodrugs. Hence, the use of such derivatives might enhance the selectivity of ALA-induced PpIX for the treatment and identification of cells specifically expressing such esterases. Third, it appears that ALA esters with high membrane penetration potential enter the intracellular space without significant resistance as compared with ALA or short-chain ALA alkyl esters, thus rapidly providing a pool of PpIX precursors that might be sufficient to maintain PpIX synthesis over a long period. Based on the work of Kloek et al. [189], Lange et al. [207] tested this hypothesis on excised pig bladder mucosa in vitro. In their experiments, tissue samples were exposed to h-ALA solutions for different periods and the PpIX production was compared to continuous exposure to h-ALA at the same concentration. It was shown that application times as short as 5 min were sufficient to maintain an optimal PpIX accumulation for more than 3 hr. This observation might have an impact on several medical domains, such as gastroenterology, where, from a practical point of view, after topical administration the contact between drug and gastric mucosa can only be main-
Controlled Drug Delivery in Cancer
615
tained for several minutes. At that time, only two derivatives gained access to clinical evaluation. In dermatology, where poor penetration of ALA through the SC limited its application for the treatment of nodular basal cell carcinomas and squamous cell carcinomas, m-ALA was recently granted marketing authorization for the photodynamic management of AK and nodular basal cell carcinoma across Europe. The second derivative currently under clinical assessment in a phase III trial in Europe and the United States, h-ALA, was found to be suitable for the fluorescence photodetection of early human bladder cancer [206]. Currently, due to the significantly improved properties of h-ALA as compared with ALA, more than 120 patients have been examined using this PpIX precursor for the photodetection at Lausanne's CHUV hospital (P. Jichlinski, personal communication). Based on preclinical studies on excised pig bladder mucosae [170], which was used for (pro)drug screening, the authors selected h-ALA from a multitude of ALA alkyl esters with respect to its water-urine solubility and PpIX formation efficacy at low doses compared with ALA. They found that, in comparison with ALA, h-ALA accelerated and increased the PpIX synthesis by more than twofold at nearly 50 times lower concentration. Furthermore, as demonstrated by fluorescence microscopy, there was a deeper and more homogeneous penetration of the drug across the entire urothelium, which is promising with respect to the phototherapeutic management of urothelial lesions. The initial clinical study [206] compared the capacity of h-ALA to induce PpIX in tumors and early cancerous disease to ALA with respect to the precursor concentration, pharmacokinetics, and selectivity (see Fig. 9). For this purpose, 25 patients were instilled with either h-ALA or ALA under different experimental conditions. Quantitative measurements of the PpIX fluorescence in papillary tumors were used to evaluate the optimal conditions. Figure 10 demonstrates the advantageous use of h-ALA-induced PpIX for the fluorescence diagnosis of early stage urothelial neoplasm. The two pictures show a sequence of white-light (Fig. 10, top) and blue-light (Fig. 10, bottom) examinations of 8 mM h-ALA over a period of 1 hr. While white-light illumination shows no significant alterations of the bladder wall, the fluorescence mode spots a bright red fluorescence, indicating the position of a carcinoma in situ. The preliminary data reported in this pilot study indicate a high sensitivity and specificity of this method when comparing fluorescence with histopathological analysis, even though more than 20-fold lower drug doses were used as compared with ALA (see Fig. 9). The latter implies the possibility of a significant reduction in the cost of the drug. These promising results with respect to the high selectivity did not change significantly when instillation times as short as 30 min were used (A. Marti, unpublished results). It can be seen from Fig. 9 that much shorter instillation times can be used, when h-ALA is applied under optimal conditions to attain
616
Lange 200 180
h-ALA 8mM h-ALA 4mM ALA 180mM
160 140 03
120 CD
o c: CD O CO CD
100 80 60 40 20
0
1
2
3
4
5
6
time [h] FIGURE 9 Effect of instillation time and concentration on the relative intensity at 635 nm in papillary tumors of the human bladder (pTa G2). Using h-ALA instead of ALA results in higher fluorescence intensities in shorter time intervals while applying significantly reduced doses. (Reprinted from Ref. 207 by permission of Churchill Livingstone.)
a given fluorescence signal. The reduction of the instillation time drastically increases the comfort of the patient, makes outpatient treatment and examination in private urology offices feasible, and helps to reduce the costs of hospitalization. Moreover, fluorescence microscopic examinations on biopsies of papillary tumors have shown that PpIX was more homogeneously distributed and found in deeper tissue layers after instillation with h-ALA as compared with ALA [224]. Hence, one may expect h-ALA to be the superior drug for PDT. Due to the significant improvements attained in dermatology and urology by the use of prodrugs, such compounds can be expected to be suitable for other medical domains using ALA-mediated PDT and PD. It has been seen that the particular prodrug must be carefully adapted to the clinical requirements with respect to its lipophilicity and PpIX formation ability.
Controlled Drug Delivery in Cancer
617
FIGURE 10 Sequence of a white light (top) and a blue light (bottom) examination 1 hr after instillation of 8 mM h-ALA. White light illumination shows a nearly normal appearance of the bladder wall, whereas fluorescence photodetection of the same area clearly indicates the site of a carcinoma in situ, not visible under white light. (See color insert.)
618
Lange
Therefore, it is important to find valid preclinical tests that make it possible to predict whether or not ALA derivatives are useful for a particular indication. Furthermore, physiological parameters must be taken into account. For the intravenous application of simple ALA alkyl esters, when cleaved by nonspecific esterases in human blood will not be advantageous over ALA. For such purposes, more exotic prodrugs that are more stable against hydrolysis have to be used. However, cell culture experiments have shown that such compounds also hardly enter the heme biosynthetic pathway. In a second scenario, due to the acidic conditions of the stomach after oral administration of ALA esters, for example, will be hydrolyzed. Consequently, no improvement can be expected to result from the use of these substances for this administration route. However, in this context it must be mentioned that cleavage of ALA esters may be an advantage for several applications, such as PDT and PD of Barrett's esophagus. In this particular case, topical application might be preferable to systemic application in order to improve the specificity of this method. The fast uptake of long-chain ALA esters could be sufficient to provide high prodrug doses to the gastric mucosa during the passage of liquid solutions through the esophagus. Furthermore, since these esters have to be applied at lower doses to induce high amounts of PpIX and will subsequently be cleaved into ALA and the free alcohol, side effects reported after oral administration of too high ALA doses (>60 mg/kg), such as nausea and hypertension, can be reduced. Besides PD and PDT of Barrett's esophagus, the PpIX-mediated PD of early human lung cancer seems to be one of the potential application fields of ALA derivatives. The physicochemical properties of lipophilic prodrugs might also be favorable for the PD of neoplasms in the respiratory tract for several reasons. Although these prodrugs have not been shown to be superior to ALA with respect to the PpIX formation rate in excised paranasal sheep mucosae, they can be applied at significantly lower doses [207]. Furthermore, the protective mechanism of the lungs is based on evacuation of mucus by ciliated cells. Mucus, which is produced by goblet cells, submucosal glands, and epithelial cells, consists of two different layers. The upper (highly lipophilic) layer, which determines viscoelastic properties of the mucus, contains macromolecules that are in part polymerized. This layer is supported by another layer, consisting of water and ions. In contrast to ALA, ALA esters will be perfectly dissolved in both layers of the mucus as well as in surfactants given along with the inhalant, which might improve the spatial drug distribution throughout the organ. 6.
CONCLUSIONS
Over the last few decades, photodynamic therapy and fluorescence photodetection have become an important part of the multidisciplinary approach to an improved management of cancer.
Controlled Drug Delivery in Cancer
619
Due to several intrinsic drawbacks exhibited by conventional photosensitizers, huge efforts have been undertaken during recent years to increase the impact of these modern treatment modalities by introducing concepts of controlled drug delivery in this field. For example, recent molecular biological research has provided an extremely high number of new potential targets for photosensitizer/targeting unit conjugates. This approach of combining well-developed techniques in biomedical optics and a large number of known photoactive compounds with methods of biochemistry and molecular biology provides two major advantages: 1.
2.
The targeting unit, specific to receptors and antigens expressed in the target tissue, makes the choice of the photoactive compound widely independent of its intrinsic pharmacokinetic properties. Thus, the photoactive unit can be chosen with respect to its spectroscopic properties, i.e., absorption and fluorescence maxima, and fluorescence and singlet oxygen quantum yield. The supplementary selectivity provided by the way the light is delivered to the tumor will decrease the risk of cross-reactions and severe side effects potentially present when targeting moieties are charged with cytotoxic preloads.
Furthermore, other concepts in controlled drug delivery, such as passive targeting of the tumor vasculature or metabolism-mediated drug delivery, have been proved to be useful and to improve the efficiency of standard delivery methods when carefully adapted to the specific problem. The future will show in which way new scientific results, such as the outcome of the human genome project and the development of large and more sophisticated combinatorial targeting libraries (e.g. phage display libraries), will have an impact on this promising treatment alternative. ACKNOWLEDGMENTS The author thanks the Ernst Schering Research Foundation for their financial support. The author also thanks Hubert van Bergh for his confidence and support during recent years as well as for proofreading this manuscript. REFERENCES 1.
2.
GA Wagnieres, WM Star, BC Wilson. In vivo fluorescence spectroscopy and imaging for oncological applications. Photochem Photobiol 68:603-632, 1998. TJ Dougherty. Photosensitizers, therapy and detection of malignant tumors. Photochem Photobiol 45:879-889, 1987.
620 3.
4.
5.
6.
7.
8. 9.
10.
11.
12.
13.
14.
15.
Lange T Okunaka, H Kato, C Conaka, H Yamamoto, A Bonaminio, ML Eckhauser. Photodynamic therapy of esophageal carcinoma. Surg Endos 4:150-153, 1990. U Schmidt-Erfurth, JW Miller, M Sickenberg, H Laqua, I Barbazetto, S Gragoudas, L Zografos, B Piguet, CJ Pournaras, G Donati, AM Lane, R Birngruber, H van den Bergh, A Strong, U Manjuris, T Gray, M Fsadni, NM Bressler, ES Gragoudas. Photodynamic therapy with verteporfin for choroidal neovascularization caused by age-related macular degeneration. Arch Ophthalmol 117:1177-1187, 1999. U Schmidt-Erfurth, J Miller, M Sickenberg, A Bunse, H Laqua, E Gragoudas, L Zografos, R Birngruber, H van den Bergh, A Strong, U Manjuris, M Fsadni, AM Lane, B Piguet, NM Bressler. Photodynamic therapy of subfoveal choroidal neovascularization—clinical and angiographic examples. Greafes Arch Clin Exp Ophthalmol 236:365-374, 1998. JC Kennedy, SL Marcus, RH Pottier. Photodynamic therapy (PDT) and photodiagnosis (PD) using endogenous photosensitization induced by 5-aminolevulinic acid (ALA): mechanisms and clinical results. J Clin Laser Med Surg 14:289-304, 1996. RM Szeimies. S Karrer, A Sauerwald, M Landthaler. Photodynamic therapy with topical application of 5-aminolevulinic acid in the treatment of actinic keratoses: an initial clinical study. Dermatology 192:246-251, 1996. P Ehrlich. Collected Studies on Immunity. New York: John Wiley & Sons, 1906. M Whitlow, D Filpula, ML Rollence, SL Feng, JF Wood. Multivalent Fvs: characterization of single-chain Fv oligomers and preparation of a bispecific Fv. Protein Eng 7:1017-1026, 1994. A Pliicktriun. Mono- and bivalent antibody fragments produced in Escherichia coli: enigneering, folding and antigen binding. Immunol Rev 130:151-188, 1992. AR Oseroff, D Ohuaha, T Hasan, T Brommer, ML Yarmush. Antibody-targeted photolysis: selective photodestruction of human T-cell leukemia cells using monoclonal antibody chlorin e6 conjugates. Proc Natl Acad Sci USA 83:8744-8748, 1986. SL Rakestraw, RG Tompkins, ML Yarmush. Antibody-targeted photolysis: in vitro studies with Sn(IV)chlorin e6 covalently bound to monoclonal antibodies using a modified dextran carrier. Proc Natl Acad Sci USA 87:4217-4221, 1990. T Hasan, A Lin, D Yarmush, A Oseroff, M Yarmush. Monoclonal antibodychromophore conjugates as selective phototoxins. J Controlled Release 10: 107-117, 1989. T Hasan, CW Lin, A Lin. Laser-induced selective cytotoxicity using monoclonal antibody-chromophore conjugates. Prog Clin Biol Res 288:471-477, 1989. N Davis, D Liu, AK Jain, SY Jiang, F Jiang, A Richter, JG Levy. Modified polyvinyl alcohol benzoporphyrin derivative conjugates as phototoxic agents. Photochem Photobiol 57:641-647, 1993.
Controlled Drug Delivery in Cancer
621
16. N Brasseur, R Ouellet, C La Madeleine, JE van Lier. Water soluble aluminium phthalocyanine-polymer conjugates for PDT: photodynamic activities and pharmacokinetics in tumor bearing mice. Br J Cancer (in press). 17. P Gold, SO Freedman. Demonstration of tumor-specific antigens in human colonic carcinomata by immunological tolerance and absorption techniques. J Exp Med 121:439, 1965. 18. P Gold, SO Freedman. Specific carcinoembyronic antigens of the human digestive system. J Exp Med 122:467, 1965. 19. D Mew, CK Wat, GH TowersH, JG Levy. Photoimmunotherapy: treatment of animal tumors with tumor-specific monoclonal antibody-hematoporphyrin conjugates. J Immunol 130:1473-1477, 1983. 20. D Mew, V Lum, CK Wat, GH Towers, CH Sun, RJ Walter, W Wright, MW Berns, JG Levy. Ability of specific monoclonal antibodies and conventional antisera conjugated to hematoporphyrin to label and kill selected cell lines subsequent to light activation. Cancer Res 45:4380-4386, 1985. 21. MA Slinkin, C Curtet, A Faivre-Chauvet, C Sai-Maurel, JF Gestin, VP Torchilin, JF Chatal. Biodistribution of anti-CEA F(ab')2 fragments conjugated with chelating polymers: influence of conjugate electron charge on tumor uptake and blood clearance. Nucl Med Biol 20:443-452, 1993. 22. M Del Governatore, MR Hamblin, EE Piccinini, G Ugolini, T Hasan. Targeted photodestruction of human colon cancer cells using charged 17.1 A chlorin e6 immunoconjugates. Br J Cancer 82:56-64, 2000. 23. FN Jiang, AM Richter, AK Jain, JG Levy, C Smits C. Biodistribution of a benzoporphyrin derivative-monoclonal antibody conjugate in A549-tumorbearing nude mice. Biotechnol Ther 4:43-61, 1993. 24. A Pelegrin, S Folli, F Buchegger, JP Mach, G Wagnieres, H van den Bergh. Antibody-fluorescein conjugates for photoimmunodiagnosis of human colon carcinoma in nude mice. Cancer 67:2529-2537, 1991. 25. S Folli, P Westermann, D Braichotte, A Pelegrin, G Wagnieres, H van den Bergh, JP Mach. Antibody-indocyanin conjugates for immunophotodetection of human squamous cell carcinoma in nude mice. Cancer Res 54:2643-2649, 1994. 26. BC Schultes, SB Spaniol, S Schmidt, W Ertmer, HJ Biersack, D Krebs. Photodynamic laser therapy with antibody-coupled phthalocyanine. Proc SPIE 2078:148-157, 1994. 27. MB Vrouenraets, GW Visser, C Loup, B Meunier, M Stigter, H Oppelaar, FA Stewart, GB Snow, GA van Dongen. Targeting of a hydrophilic photosensitizer by use of internalizing monoclonal antibodies: a new possibility for use in photodynamic therapy. Int J Cancer 88:108-114, 2000. 28. MB Vrouenraets, GW Visser, FA Stewart, M Stigter, H Oppelaar, PE Postmus, GB Snow, GA van Dongen. Development of meta-tetrahydroxyphenylchlorinmonoclonal antibody conjugates for photoimmunotherapy. Cancer Res 59: 1505-1513, 1999. 29. M Carcenac, C Larroque, R Langlois, JE van Lier, JC Artus, A Pelegrin. Preparation, phototoxicity and biodistribution studies of anti-carcinoem-
622
30.
31.
32. 33.
34.
35. 36.
37.
38.
39.
40.
41. 42.
43.
Lange bryonic antigen monoclonal antibody-phthalocyanine conjugates. Photochem Photobiol 70:930-936, 1999. KJ Steele, D Liw, N Davis, H Deal, J Levy. The preparation and application of porphyrin monoclonal antibodies for cancer therapy. Proc SPIE 1065:7379, 1989. BA Goff, U Hermanto, J Rumbaugh, J Blake, M Bamberg, T Hasan. Photoimmunotherapy and biodistribution with an OC125-chlorin immunoconjugate in an in vivo murine ovarian cancer model. Br J Cancer 70:474-480, 1994. AS Sobolev, DA Jans, AA Rosenkranz. Targeted intracellular delivery of photosensitizers. Prog Biophys Mol Biol 73:5190, 2000. CA Vogel. MC Galmiche, P Westermann, LQ Sun, A Pelegrin, S Folli, A Bischof Delaloye, DO Slosman, JP Mach, F Buchegger. Carcinoembryonic antigen expression, antibody localisation and immunophotodetection of human colon cancer liver metastases in nude mice: a model for radioimmunotherapy. Int J Cancer 67:294-302, 1996. GA Wagnieres, WM Star, BC Wilson. In vivo fluorescence spectroscopy and imaging for oncological applications. Photochem Photobiol 68:603-632, 1998. JK Dart, RJ Marsh, A Garner, RJ Cooling. Fluorescence angiography of anterior uveal melanocytic tumours. Br J Ophthalmol 72:326-337, 1988. S Folli, G Wagnieres, A Pelegrin, JM Calmes, D Braichotte, F Buchegger, Y Chalandon, N Hardman, C Heusser, JC Givel, et al. Immunophotodiagnosis of colon carcinomas in patients injected with fluoresceinated chimeric antibodies against carcinoembryonic antigen. Proc Natl Acad Sci USA 89:79737977, 1992. B Ballon, GW Fisher, AS Waggoner. DL Farkas, JM Reiland, R Jaffe, RB Mujumdar, SR Mujumdar, TR Hakala. Tumor labeling in vivo using cyanineconjugated monoclonal antibodies. Cancer Immunol Immunother 257-263, 1995. B Ballou, G Fisher, AS Waggoner, DL Farkas, JM Reiland, TR Hakala. Tumor labeling in vivo using Cy5-monoclonal antibodies. Proc Am Assoc Cancer Res 35:504, 1994. B Ballou, GW Fisher, JS Deng, TR Hakala, M Srivastava, DL Farkas. Cyanine fluorochrome-labeled antibodies in vivo: assessment of tumor imaging using Cy3, Cy5, Cy5.5, and Cy7. Cancer Detect Prev 22:251-257, 1998. S Schmidt. U Wagner. P Oehr, D Krebs. fClinical use of photodynamic therapy in gynecologic tumor patients—antibody-targeted photodynamic laser therapy as a new oncologic treatment procedure]. Zentralbl Gynakol 114:307311. German 1992. S Schmidt. Antibody-targeted photodynamic therapy. Hybridoma 12:539541. 1993. S Schmidt, U Wagner. PDT for cytoreduction in cases of ovarian cancer. In: P Wyss, Y Tadir, BJ Tromberg, Haller, eds. Photomedicine in Gynecology and Reproduction. Basel: Karger, 2000, pp 296-301. RK Jain. LT Baxter. Mechanisms of heterogeneous distribution of monoclonal
Controlled Drug Delivery in Cancer
44. 45.
46. 47.
48.
49. 50. 51.
52.
53.
54.
55.
56.
57.
58.
623
antibodies and other macromolecules in tumors: significance of elevated interstitial pressure. Cancer Res 48:7022-7032, 1988. RK Jain. Transport of molecules across tumor vasculature. Cancer Metastasis Rev 6:559-593, 1987. J Moan, K Berg. The photodegradation of porphyrins in cells can be used to estimate the lifetime of singlet oxygen. Photochem Photobiol 53:549-553, 1991. GP Adams. Improving the tumor specificity and retention of antibody-based molecules. In Vivo 12:11-22, 1998. LR Duska, MR Hamblin, MP Bamberg, T Hasan. Biodistribution of charged F(ab')2 photoimmunoconjugates in a xenograft model of ovarian cancer. Br J Cancer 75:837-844, 1997. BY Karlan, W Amin, SE Casper, BA Littlefield. Hormonal regulation of CA125 tumor marker expression in human ovarian carcinoma cells: inhibition by glucocorticoids. Cancer Res. 48:3502-3506, 1988. A Skerra, A Pluckthun. Assembly of a functional immunoglobulin Fv fragment in Escherichia coli. Science. 240:1038-1041, 1988. M Better, CP Chang, RR Robinson, AH Horwitz. Escherichia coli secretion of an active chimeric antibody fragment. Science 240:1041-1043, 1988. J McCafferty, AD Griffiths, G Winter, DJ Chiswell. Phage antibodies: filamentous phage displaying antibody variable domains. Nature 1990 Dec 6; 348(6301):552-554, 1990. RE Bird, KD Hardman, JW Jacobson, S Johnson, BM Kaufman, SM Lee, T Lee, SH Pope, GS Riordan, M Whitlow. Single-chain antigen-binding proteins. Science 242:423-426, 1988. MD Sheets, P Amersdorfer, R Finnern, P Sargent, E Lindquist, R Schier, G Hemingsen, C Wong, JC Gerhart, JD Marks, E Lindqvist. Efficient construction of a large nonimmune phage antibody library: the production of highaffinity human single-chain antibodies to protein antigens. Proc Natl Acad Sci USA 95:6157-6162, 1998. B Ramjiawan, P Maiti, A Aftanas, H Kaplan, D Fast, HH Mantsch, M Jackson. Noninvasive localization of tumors by immunofluorescence imaging using a single chain antibody with broad cancer specificity. Cancer 89:1134-1144, 2000. D Neri, B Carnemolla, A Nissim, A Leprini, G Querze, E Balza, A Pini, L Tarli, C Halin, P Neri, L Zardi, G Winter. Targeting by affinity-matured recombinant antibody fragments of an angiogenesis associated fibronectin isoform. Nat Biotechnol 15:1271-1275, 1997. M Birchler, G Neri, L Tarli, C Halin, F Viti, D Neri. Infrared photodetection for the in vivo localisation of phage-derived antibodies directed against angiogenic markers. J Immunol Meth 231:239-248, 1999. L Tarli, E Balza, F Viti, L Borsi, P Castellani, D Berndorff, L Dinkelborg, D Neri, L Zardi. A high-affinity human antibody that targets tumoral blood vessels. Blood 94:192-198, 1999. M Birchler, F Viti, L Zardi, B Spiess, D Neri. Selective targeting and pho-
624
59. 60.
61.
62.
63.
64.
65.
66.
67. 68.
69.
70.
71. 72.
73.
Lange tocoagulation of ocular angiogenesis mediated by a phage-derived human antibody fragment. Nature Biotechnol 17:984-988, 1999. PC Brooks, RA Clark, DA Cheresh. Requirement of vascular integrin alpha v beta 3 for angiogenesis. Science 264:569-571, 1994. M Friedlander, PC Brooks, RW Shaffer, CM Kincaid, JA Varner, DA Cheresh. Definition of two angiogenic pathways by distinct alpha v integrins. Science 270:1500-1502, 1995. M Friedlander, CL Theesfeld, M Sugita, M Fruttiger, MA Thomas, S Chang, DA Cheresh. Involvement of integrins alpha v beta 3 and alpha v beta 5 in ocular neovascular diseases. Proc Natl Acad Sci USA 93:9764-9769, 1996. J Luna, T Tobe, SA Mousa, TM Reilly, PA Campochiaro. Antagonists of integrin alpha v beta 3 inhibit retinal neovascularization in a murine model. Lab Invest 75:563-573, 1996. DJ Dumont, TP Yamaguchi, RA Conlon, J Rossant, ML Breitman. Tek, a novel tyrosine kinase gene located on mouse chromosome 4, is expressed in endothelial cells and their presumptive precursors. Oncogene 7:1471-1480, 1992. DJ Dumont, GH Fong, MC Puri, G Gradwohl, K Alitalo, ML Breitman. Vascularization of the mouse embryo: a study of flk-1, tek, tie, and vascular endothelial growth factor expression during development. Dev Dyn 203:8092, 1995. X Yi, N Ogata, M Komada, C Yamamoto, K Takahashi, K Omori, M Uyama. Vascular endothelial growth factor expression in choroidal neovascularization in rats. Graefes Arch Clin Exp Ophthalmo] 235:313-319, 1997. EA Pierce, RL Avery. ED Foley, LP Aiello, LE Smith. Vascular endothelial growth factor/vascular permeability factor expression in a mouse model of retinal neovascularization. Proc Natl Acad Sci USA 92:905-909, 1995. PA D'Amore. Mechanisms of retinal and choroidal neovascularization. Invest Ophthalmo] Vis Sci 35:3974-3979, 1994. H Modjtahedi. S Eccles, J Sandle, G Box, J Titley, C Dean. Differentiation or immune destruction: two pathways for therapy of squamous cell carcinomas with antibodies to the epidermal growth factor receptor. Cancer Res 54: 1695-1701, 1994. BR Voldborg, L Damstrup, M Spang-Thomsen, HS Poulsen. Epidermal growth factor receptor (EGFR) and EGFR mutations, function and possible role in clinical trials. Ann Oncol 8:1197-1206, 1997. AT Ferguson, RG Lapidus, NE Davidson. The regulation of estrogen receptor expression and function in human breast cancer. Cancer Treat Res 94:255278, 1998. T Reisine, GI Bell. Molecular biology of somatostatin receptors. Endocr Rev 16:427-442, 1995. M de Jong, WA Breeman, WH Bakker, PP Kooij, BF Bernard, LJ Hofland, TJ Visser, A Srinivasan, MA Schmidt, JL Erion, JE Bugaj, HR Macke, EP Krenning. Comparison of (11 l)In-labeled somatostatin analogues for tumor scintigraphy and radionuclide therapy. Cancer Res 58:437-441, 1998. YC Patel, CB Srikant. Subtype selectivity of peptide analogs for all five
Controlled Drug Delivery in Cancer
74.
75.
76.
77.
78.
79.
80.
81. 82.
83.
84. 85.
86.
87.
625
cloned human somatostatin receptors (hsstr 1-5). Endocrinology 135:28142817, 1994. S Achilefu, RB Dorshow, JE Bugaj, R Rajagopalan. Novel receptor-targeted fluorescent contrast agents for in vivo tumor imaging. Invest Radiol 35:479485, 2000. K Licha, S Bhargava, C Rheinlander, A Becker, J Schneider-Mergener, R Volkmer-Engert. Highly parallel nano-synthesis of cleavage peptide-dye conjugates on cellulose membranes. Tetrahedron Letters 41:1171-1715, 2000. N Lamharzi, AV Schally, M Koppan, K Groot. Growth hormone-releasing hormone antagonist MZ-5-156 inhibits growth of DU-145 human androgenindependent prostate carcinoma in nude mice and suppresses the levels and mRNA expression of insulin-like growth factor II in tumors. Proc Natl Acad Sci USA 95:8864-8868, 1998. JC Reubi. In vitro identification of vasoactive intestinal peptide receptors in human tumors: implications for tumor imaging. J Nucl Med 36:1846-1853, 1995. A Gijsens, L Missiaen, W Merlevede, P de Witte. Epidermal growth factormediated targeting of chlorin e6 selectively potentiates its photodynamic activity. Cancer Res 60:2197-2202, 2000. A Gijsens, P De Witte. Photocytotoxic action of EGF-PVA-Sn(IV)chlorin e6 and EGF-dextran-Sn(IV)chlorin e6 internalizable conjugates on A431 cells. Int J Oncol 13:1171-1177, 1998. G Milazzo, F Giorgino, G Damante, C Sung, MR Stampfer, R Vigneri, ID Goldfine, A Belfiore. Insulin receptor expression and function in human breast cancer cell lines. Cancer Res 52:3924-3930, 1992. V Papa, A Belfiore. Insulin receptors in breast cancer: biological and clinical role. J Endocrinol Invest 19:324-333, 1996. TV Akhlynina, AA Rosenkranz, DA Jans, AS Sobolev AS. Insulin-mediated intracellular targeting enhances the photodynamic activity of chlorin e6. Cancer Res 55:1014-1019, 1995. TV Akhlynina, DA Jans, AA Rosenkranz, NV Statsyuk, IY Balashova, G Toth, I Pavo, AB Rubin, AS Sobolev. Nuclear targeting of chlorin e6 enhances its photosensitizing activity. J Biol Chem 1997 272:20328-20331, 1997. AS Sobolev, DA Jans, AA Rosenkranz. Targeted intracellular delivery of photosensitizers. Prog Biophys Mol Biol 73:51-90, 2000. D Gal, M Ohashi, PC MacDonald, HJ Buchsbaum, ER Simpson. Low-density lipoprotein as a potential vehicle for chemotherapeutic agents and radionucleotides in the management of gynecologic neoplasms. Am J Obstet Gynecol 139:877-885, 1981. A Becker, B Riefke, B Ebert, U Sukowski, H Rinneberg, W Semmler, K Licha K. Macromolecular contrast agents for optical imaging of tumors: comparison of indotricarbocyanine-labeled human serum albumin and transferrin. Photochem Photobiol 72:234-241, 2000. M Lenz, WP Miehe, F Vahrenwald, G Bruchelt, P Schweizer, R Girgert. Cholesterol based antineoplastic strategies. Anticancer Res 17:1143-1146, 1997.
626
Lange
88.
BA Allison, PH Pritchard, AM Richter, JG Levy. The plasma distribution of benzoporphyrin derivative and the effects of plasma lipoproteins on its biodistribution. Photochem Photobiol 52:501-507, 1990. BA Allison, E Waterfield, AM Richter, JG Levy. The effects of plasma lipoproteins on in vitro tumor cell killing and in vivo tumor photosensitization with benzoporphyrin derivative. Photochem Photobiol 54:709-715, 1991. U Schmidt-Erfurth, H Diddens, R Birngruber, T Hasan. Photodynamic targeting of human retinoblastoma cells using covalent low-density lipoprotein conjugates. Br J Cancer 75:54-61, 1997. MR Hamblin, EL Newman. Photosensitizer targeting in photodynamic therapy. II. Conjugates of haematoporphyrin with serum lipoproteins. J Photochem Photobiol B 26:147-157. 1994. MR Hamblin, JL Miller, B Ortel. Scavenger-receptor targeted photodynamic therapy. Photochem Photobiol 72:533-540, 2000. Y Matsumura, H Maeda. A new concept for macromolecular therapeutics in cancer chemotherapy: mechanism of tumoritropic accumulation of proteins and the antitumor agent smancs. Cancer Res 46:6387-6392, 1986. H Maeda, J Wu, T Sawa, Y Matsumura, K Hori. Tumor vascular permeability and the EPR effect in macromolecular therapeutics: a review. J Controlled Release 65:271-284, 2000. H Maeda, LW Seymour, Y Miyamoto. Conjugates of anticancer agents and polymers: advantages of macromolecular therapeutics in vivo. Bioconjug Chem 3:351-362, 1992. F Yuan, M Dellian, D Fukumura, M Leunig, DA Berk, VP Torchilin, RK Jain. Vascular permeability in a human tumor xenograft: molecular size dependence and cutoff size. Cancer Res 55:3752-3756, 1995. H Maeda, Y Matsumura. CRC Crit Rev Ther Drug Carrier Syst 6:193-210, 1989. J Folkman. In: PM Mendelson, MAP Howley, LA Liotta, eds. The Molecular Basis of Cancer. Philadelphia: WB Saunders, 1995. G Jori. E Reddi, I Cozzani, L Tomio. Controlled targeting of different subcellular sites by porphyrins in tumour-bearing mice. Br J Cancer 53:615-621, 1986. E Reddi, G Lo Castro, R Biolo, G Jori. Pharmacokinetic studies with zinc(II)phthalocyanine in tumour-bearing mice. Br J Cancer 56:597-600, 1987. E Reddi, C Zhou, R Biolo, E Menegaldo, G Jori. Liposome- or LDL-administered Zn (Il)-phthalocyanine as a photodynamic agent for tumours. I. Pharmacokinetic properties and phototherapeutic efficiency. Br J Cancer 61:407411, 1990. K Schieweck. HG Capraro, U Isele, P van Hogevest, M Ochsner, T Maurer, E Batt. CGP 55847, liposome-delivered zinc(II)phthalocyanine as a phototherapeutic agent for tumors. Proc SPIE 2078:148-157, 1994. HL van Leengoed, V Cuomo, AA Versteeg, N van der Veen, G Jori, WM Star. In vivo fluorescence and photodynamic activity of zinc phthalocyanine administered in liposomes. Br J Cancer 69:840-845, 1994. M Shopova. V Mantareva, K Krastev, D Hadjiolov, A Milev, K Spirov, G
89.
90.
91.
92. 93.
94.
95.
96.
97. 98. 99.
100. 101.
102.
103.
104.
Controlled Drug Delivery in Cancer
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
115.
116.
627
Jori, F Ricchelli. Comparative pharmacokinetic and photodynamic studies with zinc(II) phthalocyanine in hamsters bearing an induced or transplanted rhabdomyosarcoma. J Photochem Photobiol B 16:83-89, 1992. D Wohrle, M Shopova, S Muller, AD Milev, VN Mantareva, KK Krastev. Liposome-delivered Zn(II)-2,3-naphthalocyanines as potential sensitizers for PDT: synthesis, photochemical, pharmacokinetic and phototherapeutic studies. J Photochem Photobiol B 21:155-165, 1993. M Shopova, D Wohrle, N Stoichkova, A Milev, V Mantareva, S Muller, K Kassabov, K Georgiev. Hydrophobic Zn(II)-naphthalocyanines as photodynamic therapy agents for Lewis lung carcinoma. J Photochem Photobiol B 23:35-42, 1994. V Cuomo, G Jori, B Rihter, ME Kenney, MA Rodgers. Liposome-delivered Si(IV)-naphthalocyanine as a photodynamic sensitiser for experimental tumours: pharmacokinetic and phototherapeutic studies. Br J Cancer 62:966970, 1990. MM Zuk, BD Rihter, ME Kenney, MA Rodgers, M Kreimer-Birnbaum. Pharmacokinetic and tissue distribution studies of the photosensitizer bis(di-isobutyl octadecylsiloxy)silicon 2,3-naphthalocyanine (isoBOSINC) in normal and tumor-bearing rats. Photochem Photobiol 1994 Jan;59(l):66-72, 1994. R Biolo, G Jori, M Soncin, R Pratesi, U Vanni, B Rihter, ME Kenney, MA Rodgers. Photodynamic therapy of B16 pigmented melanoma with liposomedelivered Si(IV)-naphthalocyanine. Photochem Photobiol 59:362-365, 1994. M Guardiano, R Biolo, G Jori. K Schaffner. Tetra-n-propylporphycene as a tumour localizer: pharmacokinetic and phototherapeutic studies in mice. Cancer Lett 44:1-6, 1989. E Reddi, A Segalla, G Jori, PK Kerrigan, PA Liddell, AL Moore, TA Moore, D Gust. Carotenoporphyrins as selective photodiagnostic agents for tumours. Br J Cancer 69:40-45, 1994. AR Morgan, GM Garbo, RW Keck, SH Selman. New photosensitizers for photodynamic therapy: combined effect of metallopurpurin derivatives and light on transplantable bladder tumors. Cancer Res 48:194-198, 1988. K Woodburn, CK Chang, S Lee, B Henderson, D Kessel. Biodistribution and PDT efficacy of a ketochlorin photosensitizer as a function of the delivery vehicle. Photochem Photobiol 60:154-159, 1994. ZJ Wang, YY He, CG Huang, JS Huang, YC Huang, JY An, Y Gu, LJ Jiang. Pharmacokinetics, tissue distribution and photodynamic therapy efficacy of liposomal-delivered hypocrellin A, a potential photosensitizer for tumor therapy. Photochem Photobiol 70:773-780, 1999. AM Richter. E Waterfield, AK Jain, AJ Canaan, BA Allison, JG Levy. Liposomal delivery of a photosensitizer, benzoporphyrin derivative monoacid ring A (BPD), to tumor tissue in a mouse tumor model. Photochem Photobiol 57:1000-1006, 1993. M Soncin, L Polo, E Reddi, G Jori, BD Rihter, ME Kenney, MA Rodgers. Unusually high affinity of Zn(II)-tetradibenzobarrelenooctabutoxyphthalocyanine for low-density lipoproteins in a tumor-bearing mouse. Photochem Photobiol 61:310-312, 1995.
628
Lange
117.
E Reddi, S Cernuschi, R Biolo, E Menegaldo, G Jori. Liposome orLDL administered ZN(II)phthalocyanine as a photodynamic agent for tumors III. Effect of cholesterol on pharmacokinetics and phototherapeutic properties. Laser Med Sci 5:339-343, 1990. U Isle, K Schieweck, R Kessler, P van Hoogevest, HG Capraro. Pharmacokinetics and body distribution of liposomal zinc phthalocyanine in tumor bearing mice: influence of aggregation state, particle size, and composition. J Pharm Sci 84:166-173, 1995. V Cuomo, G Jori, B Rihter, ME Kenney, MA Rodgers. Tumour-localising and photosensitizing properties of liposome-delivered Ge(IV)-octabutoxyphthalocyanine. Br J Cancer 64:93-95, 1991. M Soncin, L Polo, E Reddi, G Jori, ME Kenney, G Cheng, MA Rodgers. Effect of axial ligation and delivery system on the tumour-localising and photosensitising properties of Ge(IV)-octabutoxy-phthalocyanines. Br J Cancer 71:727-732, 1995. A Segalla, C Milanesi, G Jori, HG Capraro, U Isele, K Schieweck. CGP 55398, a liposomal Ge(IV) phthalocyanine bearing two axially ligated cholesterol moieties: a new potential agent for photodynamic therapy of tumours. Br J Cancer 69:817-825, 1994. RW Boyle, J Rousseau, SV Kudrevich, M Obochi, JE van Lier. Hexadecafluorinated zinc phthalocyanine: photodynamic properties against the EMT-6 tumour in mice and pharmacokinetics using 65Zn as a radiotracer. Br J Cancer 73:49-53, 1996. E Allemann, N Brasseur, O Benrezzak, J Rousseau, SV Kudrevich, RW Boyle, JC Leroux, R Gurny, JE Van Lier. PEG-coated poly(lactic acid) nanoparticles for the delivery of hexadecafluoro zinc phthalocyanine to EMT6 mouse mammary tumours. J Pharm Pharmacol 47:382-387, 1995. N Brasseur, TL Nguyen, R Langlois, R Ouellet, S Marengo, D Houde, JE van Lier. Synthesis and photodynamic activities of silicon 2,3-naphthalocyanine derivatives. J Med Chem 1994 Feb 4;37(3):415-420, 1994. DS Lawrence, SL Gibson, ML Nguyen, KR Whittemore, DG Whitten, R Hilf. Photosensitization and tissue distribution studies of the picket fence porphyrin, 3,1-TPro, a candidate for photodynamic therapy. Photochem Photobiol 61:90-98, 1995. C Delgado, GE Francis, D Fisher. The uses and properties of PEG-linked proteins. Crit Rev Ther Drug Carrier Syst 9:249-304, 1992. P Mlkvy, H Messmann, J Regula, M Conio, M Pauer, GE Millson, AJ MacRobert, SG Bown. Photodynamic therapy for gastrointestinal tumors using three photosensitizers—ALA induced PPIX, Photofrin and MTHPC. A pilot study. Neoplasma 45:157-161, 1998. P Grosjean, JF Savary, J Mizeret, G Wagnieres, A Woodtli, JF Theumann, C Fontolliet, H Van den Bergh, P Monnier. Photodynamic therapy for cancer of the upper aerodigestive tract using tetra(m-hydroxyphenyl)chlorin. J Clin Laser Med Surg 14:281-287, 1996. D Braichotte, JF Savary, T Glanzmann, P Westermann, S Folli, G Wagnieres, P Monnier, H Van den Bergh. Clinical pharmacokinetic studies of tetra(meta-
118.
119.
120.
121.
122.
123.
124.
125.
126. 127.
128.
129.
Controlled Drug Delivery in Cancer
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
629
hydroxyphenyl)chlorin in squamous cell carcinoma by fluorescence spectroscopy at 2 wavelengths. Int J Cancer 63:198-204, 1995. T Glanzmann, C Hadjur, M Zellweger, P Grosiean, M Forrer, JP Ballini, P Monnier, H van den Bergh, CK Lim, G Wagnieres. Pharmacokinetics of tetra(m-hydroxyphenyl)chlorin in human plasma and individualized light dosimetry in photodynamic therapy. Photochem Photobiol 1998 May;67(5):596602, 1998. R Hornung, MK Fehr, J Monti-Frayne, TB Krasieva, BJ Tromberg, MW Berns, Y Tadir. Highly selective targeting of ovarian cancer with the photosensitizer PEG-m-THPC in a rat model. Photochem Photobiol 70:624-629, 1999. P Westerman, T Glanzmann, S Andrejevic, DR Braichotte, M Forrer, GA Wagnieres, P Monnier, H van den Bergh, JP Mach, S Folli. Long circulating half-life and high tumor selectivity of the photosensitizer meta-tetrahydroxyphenylchlorin conjugated to olyethylene glycol in nude mice grafted with a human colon carcinoma. Int J Cancer 76:842-850, 1998. HB Ris, V Im Hof, CM Stewart, D Mettler, HJ Altermatt. Endobronchial photodynamic therapy: comparison of mTHPC and polyethylene glycol-derived mTHPC on human tumor xenografts and tumor-free bronchi of minipigs. Lasers Surg Med 23:25-32, 1998. HB Ris, A Giger, VI Hof, D Mettler, JC Stewart, U Althaus, HJ Altermatt. Experimental assessment of photodynamic therapy with chlorins for malignant mesothelioma. Eur J Cardiothorac Surg 12:542-548, 1997. HB Ris, T Krueger, A Giger, CK Lim, JC Stewart, U Althaus, HJ Altermatt. Photodynamic therapy with mTHPC and polyethylene glycol-derived mTHPC: a comparative study on human tumour xenografts. Br J Cancer 79: 1061-1066, 1999. L Morlet, V Vonarx, MT Foultier, A Gouyette, C Stewart, P Lenz, T Patrice. In vitro and in vivo spectrofluorometry of a water-soluble meta-(tetrahydroxyphenyl)chlorin (m-THPC) derivative. J Photochem Photobiol B 39:249257, 1997. AM Ronn, J Batti, CJ Lee, D Yoo, ME Siegel, M Nouri, LA Lofgren, BM Steinberg. Comparative biodistribution of meta-Tetra(Hydroxyphenyl) chlorin in multiple species: clinical implications for photodynamic therapy. Lasers Surg Med 20:437-442, 1997. S Mordon, JM Devoisselle, V Maunoury. In vivo pH measurement and imaging of tumor tissue using a pH-sensitive fluorescent probe (5,6-carboxyfluorescein): instrumental and experimental studies. Photochem Photobiol 60: 274-279, 1994. S Mordon, JM Devoisselle, S Soulie. Fluorescence spectroscopy of pH in vivo using a dual-emission fluorophore (C-SNAFL-1). J Photochem Photobiol B 28(l):19-23, 1995. X Marechal, S Mordon, JM Devoisselle, S Begu, B Guery, R Neviere, B Buys, G Dhelin, JC Lesage, D Mathieu, C Chopin. In vivo application of intestinal pH measurement using 2',7'-bis(carboxyethyl)-5,6-carboxyfluores-
630
141.
142.
143. 144.
145.
146.
147.
148. 149.
150.
151.
152.
153. 154.
155.
Lange cein (BCECF) fluorescence imaging. Photochem Photobiol 70:813-819, 1999. GR Martin, RK Jain. Noninvasive measurement of interstitial pH profiles in normal and neoplastic tissue using fluorescence ratio imaging microscopy. Cancer Res 54:5670-5674, 1994. G Helmlinger, F Yuan, M Dellian, RK Jain. Interstitial pH and pO2 gradients in solid tumors in vivo: high-resolution measurements reveal a lack of correlation. Nat Med 3:177-182, 1997. WP Ren, BF Sloane. Cathepsins D and B in breast cancer. Cancer Treat Res 83:325-352, 1996. SV Gulnik, LI Suvorov, P Majer, J Collins, BP Kane, DG Johnson, JW Erickson. Design of sensitive fluorogenic substrates for human cathepsin D. FEBS Lett 413:379-384, 1997. R Weissleder, CH Tung, U Mahmood, A Bogdanov. In vivo imaging of tumors with protease-activated near-infrared fluorescent probes. Nat Biotechnol 17: 375-378, 1999. U Mahmood, CH Tung, A Bogdanov, R Weissleder. Near-infrared optical imaging of protease activity for tumor detection. Radiology 213:866-870, 1999. CH Tung, U Mahmood, S Bredow, R Weissleder. In vivo imaging of proteolytic enzyme activity using a novel molecular reporter. Cancer Res 60: 4953-4958, 2000. RW Redmond, JN Gamlin. A compilation of singlet oxygen yields from biologically relevant molecules. Photochem Photobiol 70:391-475, 1999. Q Peng, K Berg, J Moan, M Kongshaug, JM Nesland. 5-Aminolevulinic acidbased photodynamic therapy: principles and experimental research. Photochem Photobiol 65:235-251, 1997. M Kondo, N Hirota, T Takaoka, M Kajiwara. Heme-biosynthetic enzyme activities and porphyrin accumulation in normal liver and hepatoma cell lines of rat. Cell Biol Toxicol 9:95-105, 1993. NM Navone, CF Polo, AL Frisardi, NE Andrade, AM Battle. Heme biosynthesis in human breast cancer—mimetic "in vitro" studies and some heme enzymic activity levels. Int J Biochem 22:1407-1411, 1990. RC Krieg, S Fickweiler, OS Wolfbeis, R Knuechel. Cell-type specific protoporphyrin IX metabolism in human bladder cancer in vitro. Photochem Photobiol 72:226-233, 2000. C Fritsch, G Goerz, T Ruzicka. Photodynamic therapy in dermatology. Arch Dermatol 134:207-214, 1998. P Jichlinski. M Forrer, J Mizeret, T Glanzmann, D Braichotte, G Wagnieres, G Zimmer, L Guillou, F Schmidlin, P Graber, H van den Bergh, HJ Leisinger. Clinical evaluation of a method for detecting superficial surgical transitional cell carcinoma of the bladder by light-induced fluorescence of protoporphyrin IX following the topical application of 5-aminolevulinic acid: preliminary results. Lasers Surg Med 402-408, 1997. M Kriegmair, R Baumgartner, R Knuechel, P Steinbach, A Ehsan, W Lumper, F Hofstadter. A Hofstetter. Fluorescence photodetection of neoplastic uro-
Controlled Drug Delivery in Cancer
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
166.
167.
168.
631
thelial lesions following intravesical instillation of 5-aminolevulinic acid. Urology 44:836-841, 1994. M Kriegmair, R Baumgartner, W Lumper, R Waidelich, A Hofstetter. Early clinical experience with 5-aminolevulinic acid for the photodynamic therapy of superficial bladder cancer. Br J Urol 77:667-671, 1996. L Gossner, M Stolte, R Sroka, K Rick, A May, EG Hahn, C Ell. Photodynamic ablation of high-grade dysplasia and early cancer in Barrett's esophagus by means of 5-aminolevulinic acid. Gastroenterology 114:448-455, 1998. M Mehlmann, CS Betz, H Stepp, S Arbogast, R Baumgartner, G Grevers, A Leunig, Fluorescence staining of laryngeal neoplasms after topical application of 5-aminolevulinic acid: preliminary results. Laser Surg Med 25:414420, 1999. KF Fan, C Hopper, PM Speight, G Buonaccorsi, AJ MacRobert, SG Bown. Photodynamic therapy using 5-aminolevulinic acid for premalignant and malignant lesions of the oral cavity. Cancer 78:1374-1383, 1996. P Wyss, M Fehr, H Van den Bergh, U Haller. Feasibility of photodynamic endometrial ablation without anesthesia. Int J Gynaecol Obstet 60:287-288, 1998. P Hillemanns, M Korell, M Schmitt-Sody, R Baumgartner, W Beyer, R Kimmig, M Untch, H Hepp. Photodynamic therapy in women with cervical intraepithelial neoplasia using topically applied 5-aminolevulinic acid. Int J Cancer 81:34-38, 1999. W Stummer, S Stocker, S Wagner, H Stepp, C Fritsch, C Goetz, AE Goetz, R Kiefmann, HJ Reulen. Intraoperative detection of malignant gliomas by 5aminolevulinic acid-induced porphyrin fluorescence. Neurosurgery 42:518525, 1998. MP Jenkins, GA Buonaccorsi, R Mansfield, CC Bishop, SG Bown, JR McEwan. Reduction in the response to coronary and iliac artery injury with photodynamic therapy using 5-aminolaevulinic acid. Cardiovasc Res 45:478485, 2000. Q Peng, T Warloe, K Berg, J Moan, M Kongshaug, KE Giercksky, JM Nesland. 5-Aminolevulinic acid-based photodynamic therapy. Clinical research and future challenges. Cancer 79:2282-2308, 1997. SL Marcus, RS Sobel, AL Golub, RL Carroll, S Lundahl, DG Shulman. Photodynamic therapy (PDT) and photodiagnosis (PD) using endogenous photosensitization induced by 5-aminolevulinic acid (ALA): current clinical and development status. J Clin Laser Med Surg 14:59-66, 1996. A Martin, WD Tope, JM Grevelink, JC Starr, JL Fewkes, TJ Flotte, TF Deutsch, RR Anderson. Lack of selectivity of protoporphyrin IX fluorescence for basal cell carcinoma after topical application of 5-aminolevulinic acid: implications for photodynamic treatment. Arch Dermatol Res 287:665-674, 1995. Q Peng, T Warloe, J Moan, H Heyerdahl, HB Steen, JM Nesland, KE Giercksky. Distribution of 5-aminolevulinic acid-induced porphyrins in noduloulcerative basal cell carcinoma. Photochem Photobiol 62:906-913, 1995. Z Malik, G Kostenich, L Roitman, B Ehrenberg, A Orenstein. Topical appli-
632
Lange
cation of 5-aminolevulinic acid, DMSO and EDTA: protoporphyrin IX accumulation in skin and tumours of mice. J Photochem Photobiol B 28:213218, 1995. 169. FS De Rosa, JM Marchetti, JA Thomazini, AC Tedesco, MVLP Bentley. A vehicle for photodynamic therapy of skin cancer: influence of dimethylsulphoxide on 5-aminolevulinic acid in vitro cutaneous permeation and in vivo protoporphyrin IX accumulation determined by confocal microscopy. J Controlled Release 65:359-366, 2000. 170. A Marti, N Lange. H van den Bergh, D Sedmera, P Jichlinski, P Kucera. Optimisation of the formation and distribution of protoporphyrin IX in the urothelium: an in vitro approach. J Urol 162:546-552, 1999. 171. K Berg, H Anholt, O Bech, J Moan. The influence of iron chelators on the accumulation of protoporphyrin IX in 5-aminolaevulinic acid-treated cells. Br J Cancer 74:688-697, 1996. 172. A Orenstein, G Kostenich, H Tsur, L Roitman, B Ehrenberg, Z Malik. Photodynamic therapy of human skin tumors using topical application of 5-aminolevulinic acid, DMSO and EDTA. Proc SPIE 2325:100-105, 1995. 173. T Warloe, Q Peng, H Heyerdahl, J Moan, HB Steen, KE Giercksky. Photodynamic therapy with 5-aminolevulinic acid-induced porphyrins and DMSO/ EDTA for basal cell carcinoma. Proc SPIE 2371:226-235, 1995. 174. SC Chang, AJ MacRobert, JB Porter, SG Bown. The efficacy of an iron chelator (CP94) in increasing cellular protoporphyrin IX following intravesical 5-aminolaevulinic acid administration: an in vivo study. J Photochem Photobiol B 38:114-122, 1997. 175. N Rebeiz, S Arkins, CA Rebeiz, J Simon, JF Zachary, KW Kelley. Induction of tumor necrosis by defta-aminolevulinic acid and 1,10-phenanthroline photodynamic therapy. Cancer Res 56:339-344, 1996. 176. N Rebeiz, CC Rebeiz, S Arkins, KW Kelley, CA Rebeiz. Photodestruction of tumor cells by induction of endogenous accumulation of protoporphyrin IX: enhancement by 1,10-phenanthroline. Photochem Photobiol 55:431-435, 1992. 177. H Fukuda, S Paredes, AM Batlle. Tumor-localizing properties of porphyrins. In vitro studies using the porphyrin precursor, aminolevulinic acid, in free and liposome encapsulated forms. Drug Des Deliv 5:133-139, 1989. 178. H Fukuda, S Paredes, AM Batlle. Tumour-localizing properties of porphyrins. In vivo studies using free and liposome encapsulated aminolevulinic acid. Comp Biochem Physiol B 102:433-436, 1992. 179. AF Hurlimann, G Hanggi, RG Panizzon. Photodynamic therapy of superficial basal cell carcinomas using topical 5-aminolevulinic acid in a nanocolloid lotion. Dermatology 197:248-254, 1998. 180. LE Rhodes, MM Tsoukas, RR Anderson, N Kollias. lontophoretic delivery of ALA provides a quantitative model for ALA pharmacokinetics and PpIX phototoxicity in human skin. J Invest Dermatol 108:87-91, 1997. 181. A Stenzl, I Eder, H Kostron, H Klocker, G Bartsch. Electromotive diffusion (EMD) and photodynamic therapy with delta-aminolaevulinic acid (delta-
Controlled Drug Delivery in Cancer
182.
183.
184.
185.
186.
187.
188.
189. 190.
191.
192.
193.
194.
633
ALA) for superficial bladder cancer. J Photochem Photobiol B 36:233-236, 1996. P Hillemanns, H Weingandt, R Baumgartner, J Diebold, W Xiang, H Stepp. Photodetection of cervical intraepithelial neoplasia using 5-aminolevulinic acid-induced porphyrin fluorescence. Cancer 88:2275-2282, 2000. P Hillemanns, M Korell, M Schmitt-Sody, R Baumgartner, W Beyer, R Kimmig, M Untch, H Hepp. Photodynamic therapy in women with cervical intraepithelial neoplasia using topically applied 5-aminolevulinic acid. Int J Cancer 81:34-38, 1999. A Sibourd-Baudry. Photodetection de cancer bronchique a 1'aide de 1'acide aminolevulinique: pharmacocinetique et microscopie de fluorescence. Diploma Thesis. Swiss Federal Institute of Technology, Lausanne, 1998. V Vonarx, S Eleouet, J Carre, P loss, A Gouyette, AM Leray, C Merle, Y Lajat, T Patrice. Potential efficacy of a delta 5-aminolevulinic acid bioadhesive gel formulation for the photodynamic treatment of lesions of the gastrointestinal tract in mice. J Pharm Pharmacol 49:652-656, 1997. R Baumgartner, RM Huber, H Schulz, H Stepp, K Rick, F Gamarra, A Leberig, C Roth. Inhalation of 5-aminolevulinic acid: a new technique for fluorescence detection of early stage lung cancer. J Photochem Photobiol B 36: 169-174, 1996. RM Huber, F Gamarra, H Hautmann, K Haussinger, D Wagner, M Castro, R Baumgartner. 5-aminolevulinic acid (ALA) for the fluorescence detection of bronchial tumors. Diagn Ther Endosc 5:113-118, 1999. M Mehlmann, CS Betz, H Stepp, S Arbogast, R Baumgartner, G Grevers, A Leunig. Fluorescence staining of laryngeal neoplasms after topical application of 5-aminolevulinic acid: preliminary results. Lasers Surg Med 25:414420, 1999. J Kloek, GM Beijersbergen van Henegouwen. Prodrugs of 5-aminolevulinic acid for photodynamic therapy. Photochem Photobiol 64:994-1000, 1996. A Casas, AM Batlle, AR Butler, D Robertson, EH Brown, A MacRobert, PA Riley. Comparative effect of ALA derivatives on protoporphyrin IX production in human and rat skin organ cultures. Br J Cancer 80:1525-1532, 1999. Y Berger, A Greppi, O Siri, R Neier, L Juillerat-Jeanneret. Ethylene glycol and amino acid derivatives of 5-aminolevulinic acid as new photosensitizing precursors of protoporphyrin IX in cells. J Med Chem 43:4738-4746, 2000. P Uehlinger, M Zellweger, G Wagnieres, L Juillerat-Jeanneret, H van den Bergh, N Lange. 5-Aminolevulinic acid and its derivatives: physical chemical properties and protoporphyrin IX formation in cultured cells. J Photochem Photobiol B 54:72-80, 2000. S Gerscher, JP Connelly, J Griffiths, SB Brown, AJ MacRobert, G Wong, LE Rhodes. Comparison of the pharmacokinetics and phototoxicity of protoporphyrin IX metabolized from 5-aminolevulinic acid and two derivatives in human skin in vivo. Photochem Photobiol 72:569-574, 2000. JT van den Akker, V lani, WM Star, HJ Sterenborg, J Moan. Topical application of 5-aminolevulinic acid hexyl ester and 5-aminolevulinic acid to nor-
634
195.
196.
197.
198.
199.
200.
201.
202.
203.
204.
205.
206.
207.
Lange mal nude mouse skin: differences in protoporphyrin IX fluorescence kinetics and the role of the stratum corneum. Photochem Photobiol 72:681-689, 2000. N Dogniz, D Salomon, JP Ballini, T Gabrecht, N Lange, H van den Bergh, G Wagneres. Comparison of ALA and ALA hexylester-induced PplX depth distribution in human skin carcinoma (in preparation). Lopez RFV, Vitoria M Bentley LB, Charro MB, Salomon D, van den Bergh H, Lange N, Guy RH. lontophoretic delivery of an homologous series of ionized 5-aminolevulinic acid esters for future application in the photodynamic therapy of skin cancer (accepted for publication in Photochem Photobiol). Q Peng, J Moan, T Warloe, V lani, HB Steen, A Bjorseth, JM Nesland. Buildup of esterified aminolevulinic-acid-derivative-induced porphyrin fluorescence in normal mouse skin. J Photochem Photobiol B 34:95-96, 1996. C Fritsch. B Homey, W Stahl, P Lehmann, T Ruzicka, H Sies. Preferential relative porphyrin enrichment in solar keratoses upon topical application of delta-aminolevulinic acid methylester. Photochem Photobiol 68:218-221, 1998. E Rud. O Gederaas, A Hogset, K Berg. 5-aminolevulinic acid, but not 5aminolevulinic acid esters, is transported into adenocarcinoma cells by system BETA transporters. Photochem Photobiol 71:640-647, 2000. CJ Whitaker, SH Battah, MJ Forsyth, C Edwards, RW Boyle, EK Matthews. Photosensitization of pancreatic tumour cells by delta-aminolaevulinic acid esters. Anticancer Drug Des 15:161-70, 2000. OA Gederaas, A Holroyd, SB Brown, D Vernon, J Moan, K Berg. 5-Aminolevulinic acid methyl ester transport on amino acid carriers in a human colon adenocarcinoma cell line. Photochem Photobiol 73:164-169, 2001. J Kloek, W Akkermans, GM Beijersbergen van Henegouwen. Derivatives of 5-aminolevulinic acid for photodynamic therapy: enzymatic conversion into protoporphyrin. Photochem Photobiol 67:150-159, 1998. S Eleouet, N Rousset, J Carre, L Bourre, V Vonarx, Y Lajat, GM Beijersbergen van Henegouwen, T Patrice. In vitro fluorescence, toxicity and phototoxicity induced by delta-aminolevulinic acid (ALA) or ALA-esters. Photochem Photobiol 71:447-454, 2000. R Washbrook, PA Riley. Comparison of delta-aminolaevulinic acid and its methyl ester as an inducer of porphyrin synthesis in cultured cells. Br J Cancer 75:1417-1420, 1997. JM Gaullier, K Berg, Q Peng, A Anholt, PK Selbo, LW Ma, J Moan. Use of 5-aminolevulinic acid esters to improve photodynamic therapy on cells in culture. Cancer Res 57:1481-1486, 1997. N Lange. P Jichlinski, M Zellweger, M Forrer, A Marti, L Guillou, P Kucera, G Wagnieres, H van den Bergh. Photodetection of early human bladder cancer based on the fluorescence of 5-aminolaevulinic acid hexylester-induced protoporphyrin. IX: A pilot study. Br J Cancer 80:185-193, 1997. N Lange, L Vaucher, A Marti, AL Etter, P Gerber, H van den Bergh, P Jichlinski, P Kucera. Routine experimental system for defining conditions used in photodynamic therapy and fluorescence photodetection of (Non-)neoplastic epithelia. J Biomed Opt 62:151-159, 2001.
Controlled Drug Delivery in Cancer 208. 209.
210.
211. 212. 213.
214.
215.
216.
217.
218. 219.
220. 221.
222.
223.
224.
635
BW Barry. Dermatologic Formulation: Percutaneous Absorption, Drugs and Pharmaceutical Science. Vol. 18. New York: Marcel Dekker, 1983. LH Conder, SI Woodard, HA Dailey. Multiple mechanisms for the regulation of haem synthesis during erythroid cell differentiation. Possible role for coproporphyrinogen oxidase. Biochem J 275:321-326, 1991. Z Malik, M Djaldetti. 5-Aminolevulinic acid stimulation of porphyrin and hemoglobin synthesis by uninduced Friend erythroleukemic cells. Cell Differ 8:223-233, 1979. SW Jacob, R Herschler. Pharmacology of DMSO. Cryobiology 23:14-27, 1986. AK Banga, S Bose, TK Ghosh. Iontophoresis and electroporation: comparisons and contrasts. Int J Pharm 179:1-19, 1999. P Hinnen, FW de Rooij, ML van Velthuysen, A Edixhoven, R van Hillegersberg, HW Tilanus, JH Wilson, PD Siersema. Biochemical basis of 5-aminolaevulinic acid-induced protoporphyrin IX accumulation: a study in patients with (pre)malignant lesions of the oesophagus. Br J Cancer 78:679-682, 1998. H Messmann, R Knuchel, W Baumler, A Holstege, J Scholmerich. Endoscopic fluorescence detection of dysplasia in patients with Barrett's esophagus, ulcerative colitis, or adenomatous polyps after 5-aminolevulinic acid-induced protoporphyrin IX sensitization. Gastrointest Endosc 49:97-101, 1999. H Messmann. 5-Aminolevulinic acid-induced protoporphyrin IX for the detection of gastrointestinal dysplasia. Gastrointest Endosc Clin N Am 10:497512, 2000. C Fuchs, R Riesenberg, J Siegert, R Baumgartner. pH-Dependent formation of 5-aminolaevulinic acid-induced protoporphyrin IX in fibrosarcoma cells. J Photochem Photobiol B 40:49-54, 1997. L Wyld, MW Reed, NJ Brown. The influence of hypoxia and pH on aminolaevulinic acid-induced photodynamic therapy in bladder cancer cells in vitro. Br J Cancer 77:1621-1627, 1998. A Albert. Chemical aspects of selective toxicity. Nature 182:421-423, 1958. A Casas, C Perotti, H Fukuda, L Rogers, A Butler, A Batlle. Hexyl-aminolevulinic acid induces selective porphyrin synthesis in tumours. Abstracts of the 9th ESP Meeting, Lillehammer, 2001, Abstract 557. EL Taylor, DI Verdon, SB Brown. Abstracts of the 9th ESP Meeting, Lillehammer, 2001, Abstract 551. F Doring, J Walter, J Will, M Focking, M Boll, S Amasheh, W Clauss, H Daniel. Delta-aminolevulinic acid transport by intestinal and renal peptide transporters and its physiological and clinical implications. J Clin Invest 101: 2761-2767, 1998. FM Rossi, DL Campbell, RH Pettier, JC Kennedy, EF Dickson. In vitro studies on the potential use of 5-aminolaevulinic acid-mediated photodynamic therapy for gynaecological tumours. Br J Cancer 74:881-887, 1996. SL Gibson, JJ Havens, TH Foster, R Hilf. Time-dependent intracellular accumulation of delta-aminolevulinic acid, induction of porphyrin synthesis and subsequent phototoxicity. Photochem Photobiol 65:416-421, 1997. A Marti, P Jichlinski, N Lange, JP Ballini, P Kucera (in preparation).
17 Tissue Oxygen Measurements Using Phosphorescence Quenching David F. Wilson and Sergei A. Vinogradov University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A.
1.
BACKGROUND
Oxygen is consumed in large amounts to provide the metabolic energy required for the survival of aerobic organisms. It is also a reactant in many essential catabolic and anabolic reactions. In more complex aerobic organisms, such as mammals, there are intricate cardiovascular systems to assure that oxygen and other metabolites are delivered in sufficient amounts and at appropriate concentrations to each part of the body. It is, therefore, important to know the concentration of oxygen that is required for its many biological reactions as well as the concentrations present at the reaction sites. To obtain this information it is necessary to have a method(s) for measuring oxygen that can be used under the conditions that exist in vivo. Only through accurate measurement of oxygen in situ is it possible to gain an understanding of the role of oxygen in metabolism. This is particularly important in complex organisms, such as in mammals. In mammals, alterations in either the metabolism of individual cells or the cardiovascular system can result in tissue pathology. Oxygen-dependent quenching of phosphorescence or simply phosphorescence quenching is an optical method capable of quantifying oxygen over a wide range of conditions in biological systems [1,2]. It offers important 637
638
Wilson and Vinogradov
advantages over the other available oxygen measurement methods. For example, it is a noninvasive method and allows for repetitive measurements of oxygen in living tissue. Although phosphorescence quenching is a relatively new technique for measuring oxygen, it has been evolving rapidly, and its areas of application are steadily expanding. In this chapter, we try to provide a basis for the understanding of the use of oxygen-dependent quenching of phosphorescence. A brief description of the "state of the art" instruments and phosphorescent dyes and their capabilities, short summaries of areas of application that are proving particularly fruitful, and a prospectus of the future developments in the technology are presented below.
2.
PHOSPHORESCENCE QUENCHING: A BASIS OF THE METHOD
Phosphorescence is the emission from the triplet excited state. The effectiveness of quenching of phosphorescence by oxygen is a function of the frequency of collision between the triplet state molecules and molecular oxygen [3]. This fact provides the basis for using phosphorescence as a tool for measuring oxygen concentration. For in vivo tissue measurements, watersoluble phosphorescent dyes (phosphors) are introduced directly into the blood [1,2,4,5], whereas in some other applications it may be more advantageous to use polymer-supported phosphors. In this chapter we will focus on the applications that employ soluble phosphors. The reader is directed to a recent review [6] for the latest information on the polymer-supported oxygen sensors. The phosphor, introduced to the media of interest, is excited by a photon (h^ cxcil ) and returns to the ground state either with emission of light (hf phos — phosphorescence) or by transferring energy to other molecules (quenchers) in the environment (Fig. 1). The rate of the phosphorescence decay depends on the concentration of the quencher molecules in the solution. In systems where oxygen (O2) is the primary quencher (as it is in most biological systems), the measured phosphorescence lifetime may be converted to oxygen pressure using the Stern-Volmer relationship:
7 = 7= ! + K t | T 0 -p0 2
(1)
where I0 and r() are the phosphorescence intensity and the lifetime in the absence of oxygen; I and r are the intensity and the lifetime, respectively, at an oxygen pressure of pO2; and Kq is the rate constant of the reaction between the phosphor in the triplet state and molecular oxygen. The value
Phosphorescence Quenching
639
+ hv.phos
FIGURE 1 Kinetic scheme of phosphorescence and quenching. Pso, phosphor in the ground singlet state; PTi, phosphor in the excited (shown with dot) triplet state; h^excit, excitation photon; h^phos, phosphorescence; Kp, rate of the phosphorescence decay in the absence of quencher, which is proportional to the reciprocal of the lifetime TO; Kq, oxygen quenching constant. When quenching occurs, the quencher molecule O2 is promoted to the excited state (shown with dot).
of Kq is a function of the diffusion constants of phosphor and oxygen, temperature, and phosphor microenvironment. The Stern-Volmer relationship holds as long as the concentration of the quencher is much higher than that of the triplet excited state molecules. This requirement holds for all of the common applications, in which the concentration of oxygen is generally above 1CT8 M and the concentration of excited triplet molecules is generally less than 10 14 M.
3.
MEASURING PHOSPHORESCENCE LIFETIMES
There are two methods for measuring excited state lifetimes, namely, the frequency domain method and the time domain method [7]. Both methods have been used for the purpose of measuring phosphorescence lifetimes [811] and for oxygen measurements in vivo in particular [12—15]. In the frequency domain method, the phosphor is excited using a sinusoidally modulated light, and the lifetime is determined from the delay between the excitation and the emitted phosphorescence. In the time domain method, the phosphor is excited by a fast (with respect to the phosphorescence lifetime) flash of light. The phosphorescence lifetime is determined by fitting the resulting intensity decay with an exponential function. Below we will review the instruments for measuring phosphorescence lifetimes in frequency domain (3.1) and in time domain (3.2). Also, we will briefly describe the instruments for measuring lifetime distributions (3.3) and the instruments for obtaining two-dimensional lifetime images (3.4).
640
3.1
Wilson and Vinogradov
Instruments for Measuring Phosphorescence Lifetimes in Frequency Domain
In frequency domain, the sample is excited with modulated monochromatic light. The excitation light absorbed by the phosphor is emitted as phosphorescence, but after a certain delay period. As a result, phosphorescence emission is modulated with the same frequency but delayed in time (phase shifted) with respect to the excitation light. After optical filtering, the emitted light is detected by a sensitive measuring device, such as a photomultiplier. The detector signal is digitized at a frequency at least twice the highest frequency used in the measurements and analyzed to give the phase shift <£, which is used to calculate the phosphorescence lifetime r:
where 01 = 2trf and f is the modulation frequency. In practice, it is often advantageous to make lifetime measurements in a so-called phase mode. In this mode, not the modulation frequency f but a phase angle is preselected and held constant throughout the measurement range. For Pd-ra^w-tetra-/:>-carboxyphenylporphyrin (PdTCPP) bound to bovine serum albumin (BSA), currently the most widely used phosphor, r() at 38°C is 646 /xsec and the lifetime rat air saturation is 16 /msec [16,17]. The important range of oxygen concentrations in biological systems extends from zero to approximately 280 /xM. The Stern-Volmer relationship (1) and Eq. (2) show that in order to maintain a constant value of the phase shift c/> during a set of measurements in which there are significant changes in oxygen concentration, it is necessary to vary the modulation frequencies from 100 Hz to about 10,000 Hz. Instruments are available that can measure phosphorescence lifetimes either at a given fixed frequency or by automatically adjusting the frequency from 100 Hz to more than 20.000 Hz as needed to obtain the selected phase shift (/> [18]. Two important advantages of the frequency domain method are: (1) frequency lock amplification minimizes signals of the frequencies other than that used for the modulation; (2) interference from ambient light is minimized because only signals modulated at the same frequency as the excitation light are amplified. The main components of instruments that measure lifetime in frequency domain are discussed below. These include (1) modulated sources of monochromatic light; (2) emission detectors; (3) digital data analysis system for determining emission phase shift and calculating phosphorescence lifetime.
Phosphorescence Quenching
1.
2.
3.
641
Generation of sinusoidally modulated light using light emitting diodes. In the frequency domain instruments developed and used in our laboratory, light-emitting diodes (LEDs) are used as excitation sources. A sine wave of the desired frequency is generated by a digital signal processor (DSP) and passed through the smoothing filters to provide control of the current in the LED driving circuit. The technology has reached the point where LEDs can provide monochromatic light of sufficiently high intensity while at the same time offering excellent long-term stability. LEDs with light output of 2 mW optical power or greater are available at many wavelengths, including 450 nm, 524 nm, and 635 nm, which are of particular interest for excitation of the porphyrinbased phosphors. The light output of green and blue LEDs depends nonlinearly on the driving current. Therefore, the driving current needs to be corrected to provide an accurate sinusoidal light output. Measuring phosphorescence emission. The light emitted from the sample is collected and carried to the detector by a light guide, where it is passed through optical filters to remove the residual excitation light. The analog signal from the detector is amplified by AC-coupled operational amplifiers and delivered to an analogto-digital converter (ADC) to convert the signal into digital form for further processing. Data analysis in the frequency domain. The array of numbers (data vector) is fitted to a function F(t), which describes the dependence of the emission intensity on time t at a given modulation frequency f: F(t) = B + A X sin(27rf X t - 0)
(3)
where A is the amplitude of the emission and B is the baseline term. The x2 fitting allows for the simultaneous determination of the emission amplitude A and the angular functions of the phase angle (cos
Instruments for Measuring Phosphorescence Lifetimes in Time Domain
The literature has several reports of the instruments for measuring phosphorescence lifetimes in time domain (see, for examples [12,13]). More extensive literature exists on the instrumentation for time domain fluorescence lifetime measurements (see, for example, [7]), in which essentially the same
642
Wilson and Vinogradov
principles are used but the components of instruments are significantly different. In a time domain experiment, the sample is excited with a flash of light, which decays much faster than phosphorescence (usually 1-5 /.isec at half-maximal intensity). The excitation light is optically filtered and delivered to a sample either directly or via a light guide. A fraction of the absorbed light is emitted as phosphorescence. The emitted light is collected, optically filtered to minimize the back-scattered excitation light, and delivered to a measuring device. The intensity of the emitted light rises as the convolution of the flash with the phosphorescence decay and then decays back to the baseline. Since the flash is very fast in comparison with the phosphorescence lifetime, after a delay of about 15-20 /JLSQC the detected signal can be safely considered undisturbed phosphorescence. The signal is digitized with frequency sufficient to allow for the decay to be accurately analyzed. If the decay is single exponential, then only a few data points is required to determine the phosphorescence lifetime. These can be collected using simple, variable-width gates, which allow for the collection of samples at different times after the excitation flash. The analysis in this case is a simple linear %2 fitting of the data point logarithms. However, there can be complicating features, such as relatively high contributions of excitation light and/or fluorescence. In addition, the phosphorescence decay itself can be a composition of signals with different lifetimes (see, for example, [19]). These can be resolved only by digitizing with a frequency high enough to resolve the fastest of the contributing signals. In phosphorescence lifetime measurements, digitization at about 10 MHz is usually adequate for accurately determining the fastest component(s), e.g., the excitation light/fluorescence. 3.3
Instruments for Determining Phosphorescence Lifetime Distributions and Converting Them into Oxygen Histograms
When phosphor is contained in a medium with a heterogeneous distribution of oxygen concentrations, the phosphorescence from a sample consists of the signals with different lifetimes, which are all summed together in a multiexponential decay I(t). The plot of the relative intensities of single exponential components against the corresponding lifetimes is called a lifetime distribution P(r). Each component of the distribution P(r) corresponds to the phosphor in a microscopic volume with its own oxygen concentration. When phosphor is dissolved in the blood, the phosphorescence arises from the molecules contained in the vessels throughout the microvasculature, including arterioles, capillaries, and venuoles. Each of these vessel types
643
Phosphorescence Quenching
Oxygen
Q2
20
40 60 80 Q = pO2 (Torr)
Intesnity
Y2 -V1
100
Lifetime
V1 •
-V2highpO 2 (Q 2 50
100 t (usec)
150
FIGURE 2 Hypothetical volume of tissue, containing only two blood vessels V1 and V2 with average oxygen concentrations QI and Q2, respectively (Q2 > QJ.
contains blood with different mean oxygen pressures. Under normal conditions, the local oxygen pressures range from about 80 mm Hg down to near 0 mm Hg. Thus, blood in tissue presents an important example of a system that has a distribution of the phosphorescence lifetimes P(r), caused by heterogeneous distribution of quencher concentrations. Let's describe this distribution using function i^(Q), where Q is the concentration (or partial pressure) of oxygen. A trivial example illustrating the relationship between the oxygen distribution and the phosphorescence lifetime distribution is shown in Fig. 2. The illuminated volume in this case contains just two blood vessels VI and V2 with average oxygen concentrations Ql and Q2, respectively (Q2 > Ql), and the size of vessel VI is twice the size of V2. The distribution of oxygen concentrations #(Q) in such a system consists of two sharp peaks with integral intensities proportional to the sizes of the vessels. The corresponding lifetime distribution P(r) also has two peaks with the same ratio of integrals, but with different widths. This is due to the nonlinearity of Q —> r coordinate transformation, defined by the Stern-Volmer relationship: r0KQQ
(4)
644
Wilson and Vinogradov
The phosphorescence decay originating from such a sample will contain basically two components, one with shorter lifetime and higher intensity (VI), and another with longer lifetime and lower intensity (V2). Reconstruction of the lifetime distribution from time-resolved data presents an interesting and challenging problem. Mathematically this is a subset of the more general problem of Laplace transform inversion [20], relevant to many areas of medical imaging. One approach to this problem is to a priori assume the distribution shape (see, for example [21]), e.g., Gaussian, Lorentzian, etc., and then find its parameters using x2 fitting of the experimental time-resolved data. However, more general, robust, and unbiased approach is based on so-called regularized inversion methods [22], e.g., maximum entropy method (MEM) [23-25] and exponential series method (ESM) [26-28]. In general, the accuracy of the recovered distribution is determined by the noise present in the data. A simple algorithm has been developed [19] specifically to deconvolute phosphorescence decays into the lifetime histograms. The quality of the phosphorescence time-resolved data obtained using standard time domain phosphorometers [29] has been sufficient for calculating lifetime distributions using this algorithm. More recently, the same algorithm was used to implement the MEM for real-time analysis of the phosphorescence data in frequency domain [30]. The latter method has been incorporated into the software for a new frequency domain phosphorometer [31,32]. The instrument/algorithm combination will enable researchers to monitor the changes in mean tissue oxygen pressures as well as changes in the oxygen distributions (see [33] for preliminary results). 3.4
Instruments for Imaging Phosphorescence Lifetime/Oxygen Distributions in Two Dimensions
Phosphorescence quenching is an optical method and therefore phosphorescence intensity can be imaged in the same manner as fluorescence intensity [34-37]. Imaging has been used to show qualitative changes in tissue oxygenation in many tissues in vivo, including brain [35], carotid body [38], heart [39,40], and intestine [36], as well as in rat heart [37], and liver [34], and cat carotid body [41,42] in vitro. Phosphorescence intensity imaging offers high temporal resolution (>30 frames/sec), but accurate determination of oxygen pressures is possible only by measuring the phosphorescence lifetimes. Phosphorescence lifetime imaging methods have been reported for both time and frequency domains [43-46]. The relatively long lifetimes of phosphorescence, compared with those of fluorescence (microseconds vs nanoseconds), make phosphorescence lifetime imaging more practical. In order to obtain a phosphorescence lifetime image, photomultiplier or other
645
Phosphorescence Quenching
point detector is replaced with infrared-sensitive, intensified, gated CCD camera. In time domain measurements, the intensifier, which controls the sensitivity of the camera, has to be switched on or off in less than 1 ynsec, whereas in frequency domain the intensifier has to be modulated at audiofrequencies (100-10,000 Hz). A scheme of the phosphorescence lifetime imaging experiment is shown in Fig. 3. 4.
PHOSPHORS FOR OXYGEN MEASUREMENTS
Many compounds are phosphorescent, but only few are suitable for oxygen measurements. Considerations important for selecting optimal phosphors are (1) high quantum efficiency of phosphorescence; (2) strong absorption of light in the wavelength region optimal for the excitation; (3) phosphorescence lifetimes and oxygen quenching constants appropriate for the measurements; (4) the possibility to chemically modify the phosphor in order to give it physical properties (hydrophilicity, molecular size, etc.) optimal for the chosen application. The phosphorescent or, more generally, luminescent compounds used for oxygen sensing can be divided into two main groups: a-diimine complexes of Ru(II) [47] (complexes of Au(III) [48], Os(II) [49], Ir(III) [50], and Re(II) [51] are also investigated), and Pt(II) and Pd(II) complexes of porphyrins [1,52,53] and other related tetrapyrrolic macrocycles [54]. The major difference between the compounds in these two groups is their lifetimes, which in deoxygenated solutions are in the order of hundreds of nanoseconds for Ru(II) and Os(II) complexes and tens to hundreds of mi-
IMAGE
L
J
oo
CCD camera
Power Supply
Flash lamp
Data station (computer with frame grabber and timer board)
Excitation Phosphorescence
Studied object (area of tissue with differently oxygenated blood vessels)
FIGURE 3
Phosphorescence lifetime imagining in two dimensions.
646
Wilson and Vinogradov
croseconds for Pt(II) and Pd(II) porphyrins. The use of group VIII a-diimine complexes for oxygen sensing was recently reviewed in great detail [6j. Here we will focus on the use of metalloporphyrins, which in our opinion are much more applicable to oxygen measurements in biological systems. Upon excitation with visible light, porphyrins are promoted to excited singlet states. The presence of a heavy metal atom (Pt or Pd) significantly increases the probability of a spin-forbidden transition to a nearby located triplet state—a process known as intersystem crossing. Emission from the triplet-state, phosphorescence, is significantly prolonged in time, increasing the probability of collisions between the triplet state and other molecules. Oxygen in the ground state is also a triplet and can very effectively exchange energy with a triplet state porphyrin and since in biological systems O 2 is by far the most abundant quencher, metalloporphyrins as probes are extremely selective to oxygen. In a wide range of concentrations, the dependence of phosphorescence lifetime on oxygen concentration, or partial pressure, is given by the Stern-Volmer relationship ( 1 ) . The phosphorescence quantum yields are generally higher for Pt (2030%) [3] than for Pd porphyrins (8-12%) 13,531, although the latter are still considered powerful phosphors. Phosphorescence lifetimes of Pd porphyrins in deoxygenated solutions are substantially longer (100-2000 /asec) [1,3,19] than those of the Pt complexes (10-100 yusec) (see, for example, [3]). Nevertheless, the lifetimes of both groups of compounds are still long enough to allow for effective collisional quenching by O 2 . In general, the photophysical properties of Pt and the Pd porphyrins are very well suited for O2 sensing. Porphyrin dianions are tettr/dentate ligands. Unlike a-diimines in the case of Ru(II)-chelates, they form very stable complexes with Pt(II) and Pd(II). For example, Pd porphyrins can be chlorosulfonated without any detectable loss of metal by treating them with hot chlorosulfonic acid [53,55]. Therefore, release of a free metal in vivo can only occur if the porphyrin skeleton itself is destroyed. This implies that if a porphyrin is encapsulated in a protective shell, which prohibits access of large molecules (proteins), the possibility of the metal release is practically eliminated. On the other hand, porphyrins can be readily derivatized by many developed synthetic methods. Functional groups placed on the porphyrin periphery provide anchor points for the covalent attachment of protective ligands. 4.1
Phosphors Absorbing in the Visible Region of the Spectrum
Pd and Pt porphyrins have strong absorption bands in the near-UV region of the spectrum (400 nm to 430 nm) and weaker absorption bands in the
Phosphorescence Quenching
647
green region (510-550 nm) [52]. These phosphors are very effective for measuring oxygen in solutions in vitro and in the surface layer of tissue in vivo. However, in tissue, excitation light at these wavelengths cannot penetrate to a great depth because of the strong absorption by natural chromophors such as hemoglobin, myoglobin, cytochromes, flavoproteins, noniron-heme proteins, etc. In the absence of readily available phosphors absorbing in the near-IR, Pd-meso-tetra-(4-carboxyphenyl)porphyrin (PdTCPP) is the currently most extensively used phosphor. One disadvantage of PdTCPP is that it needs to be prebound to albumin, which provides the surrounding environment for the chromophor and the necessary water solubility. When bound to albumin, PdTCPP is suitable for in vivo measurements quenching constant Kq and lifetime TO. To eliminate the necessity of using albumin, a new generation of completely synthetic phosphors is being developed [56,57] (see below). These are Pd porphyrins incorporated inside dendritic cages. One such phosphor, based on the same PdTCPP, is called Oxyphor R2. This is a water-soluble compound, bearing 16 carboxyl groups on the periphery. At neutral or nearneutral pH, the carboxyls are ionized, and Oxyphor R2 is a polyanion with total charge of —16. Although the phosphorescence of Oxyphor R2 is still influenced by albumin in the blood, it is readily soluble in physiological saline, and no albumin is needed to prepare solutions for injection into the blood. In the presence of albumin, the calibration constants of Oxyphor R2 are similar to those of PdTCPP [56,58], but the high solubility in water makes it much easier to use and minimizes the possibility of adverse reactions. 4.2
Phosphors Absorbing in the Near-Infrared Region of the Spectrum
There has been an extensive effort to identify/synthesize phosphors that both absorb and emit in the near-IR region of the spectrum (620 nm to about 1000 nm) where the absorption of the natural pigments of tissue is minimal. The search has focused on compounds with extended aromatic TT systems, such as the extended porphyrins [59] and/or partially oxidized porphyrins [54]. The most successful to date are the Pd tetrabenzoporphyrins, which possess strong absorption bands at about 440 nm and at 620-640 nm and emission near 800 nm [59-61]. Pd tetra-(4-carboxyphenyl)-tetrabenzoporphyrin (PdTCPTBP) has been synthesized [62,63] and characterized as a phosphor for oxygen measurements. The properties of this phosphor are similar to those of PdTCPP in the sense that PdTCPTBP binds to albumin with high affinity and this binding decreases the oxygen quenching constant. The polyglutamic dendritic derivative (Oxyphor G2) of the basic phosphor
648
Wilson and Vinogradov
PdTCPTBP has also been synthesized [64] and shown to be an effective phosphor for measurements in vivo. Oxyphor G2, like Oxyphor R2, is a highly water-soluble material. It's quenching constant is lowered significantly in solutions containing 1-2% BSA, and has a value of about 200 Hg mm' 1 sec"1. 5.
OXYGEN MEASUREMENTS IN VITRO USING PHOSPHORESCENCE QUENCHING
5.1
Advantages of the Phosphorescence Quenching Method
Oxygen-dependent quenching of phosphorescence is a very effective method for measuring oxygen in solutions. In this section we will summarize the reasons for which this method is particularly valuable for specific applications in vitro. 5.1.1
Noninvasiveness of the Measurements in Closed Chambers
Phosphorescence can be measured through the wall of a vessel, eliminating the possibility of oxygen leaking from the external environment into the sample solution. A sample solution containing the phosphor can be sealed in a transparent vessel, made of oxygen-impermeable material, and oxygen measured through the vessel wall. 5.1.2
High Temporal Resolution
It is possible to determine phosphorescence lifetime in time domain using data obtained just from a single decay. For example, for a lifetime of 400 /zsec, the decay measurement requires only about 2 msec. Thus, it is theoretically possible to make repetitive measurements separated by as little as 2-3 msec. At the present time, reported measurements have not exceeded about 15/sec in either time domain [5,66,67] or in frequency domain [31]. These temporal resolutions are excellent but still are well below the theoretical limit. 5.1.3
High Accuracy at Low Oxygen Pressures
Phosphorescence quenching method is highly sensitive at low oxygen pressures. Values of oxygen quenching constant Kq can be as high as several thousand mm Hg"' sec ', and the lifetimes at zero oxygen, r,,, of some Pd porphyrins can be longer than 2 msec [3]. From the Stern-Volmer equation ( I ) , if the quenching constants were 5,000 mm Hg ' sec'1 and r() were 2 msec, the phosphorescence lifetime r would decrease to \T (from 2 msec to
Phosphorescence Quenching
649
1 msec) when the oxygen pressure increased from zero to 0.02 mm Hg. Using such a phosphor, oxygen could be easily measured to less than 0.0005 mm Hg. 5.1.4
Possibility to Measure Oxygen in Small Samples
Oxygen measurements can be readily made in volumes as small as 20-40 /uiL [65,92,97], making phosphorescence quenching one of the most sensitive methods available. It is feasible, for example, to measure the respiratory rate of a single mammalian cell, although such measurements have yet to be reported. It is also possible to make measurements in flowing streams of fluids in tubes with internal diameters of less than 20 5.1.5
Possibility to Measure Oxygen Concentration Gradients
The capacity to measure oxygen distributions is a unique strength of this method. If a phosphor is uniformly distributed in a homogeneous medium, such as is the case of phosphor in solution, phosphorescence lifetime distributions can be used to determine oxygen gradients that occur in that medium. For example, a layer of growing cells attached to the flat surface of a tissue culture plate and covered by a thin layer of unstirred growth medium will generate a linear gradient in oxygen concentration from the cell surface to the interface of the growth medium with air. At steady state, oxygen flux by diffusion down the concentration gradient is proportional to the rate of oxygen consumption by the cells. In all of the elementary volumes with different oxygen concentrations, the phosphor is present in equal concentrations. The difference in oxygen concentrations results in the difference of the phosphorescence lifetimes. The lifetime distribution [31,32], therefore, allows calculation of the oxygen concentration gradient. The rate of oxygen consumption by the cells can then be derived from the oxygen gradient, given the diffusion constant of oxygen in the medium. 5.2
Selected Applications of Phosphorescence Quenching In Vitro
In the earliest applications of the phosphorescence quenching method, time domain instruments were used to measure the oxygen dependence of mitochondrial and cellular respiration [2,66,67]. The oxygen concentrations were measured in stirred suspensions of mitochondria and cells closed in glass chambers. The instruments, which were specifically designed for rapid measurements, allowed for collection of up to 10 oxygen readings per second. As a result, it was possible to accurately measure depletion of oxygen at low concentrations (from 0 to 50 /xM), when all sources of oxygen were decreased to insignificant levels. The time course of oxygen depletion and
650
Wilson and Vinogradov
the value of the oxygen pressure at 50% of maximal consumption rate (P50) could be accurately determined even when this value was less than about 2 mm Hg. A major effort has been devoted to the development of phosphor containing oxygen-permeable membranes, which would create an optical variant of oxygen electrode [68-73]. The advantages of such optical sensors are (1) lack of a requirement for a direct coupling between the sensor and the rest of the measuring apparatus (the phosphorescence from the membrane can be measured through the space); (2) very low (insignificant) oxygen consumption during the measurement; and (3) high sensitivity at low oxygen concentrations (see [6] for a review on polymer supported oxygen sensors).
6. 6.1
OXYGEN MEASUREMENTS IN VIVO Phosphors for Use In Vivo
Phosphors for oxygen measurements in vivo have some requirements in addition to those listed in Sec. 4. Metalloporphyrins in general have low solubility in aqueous media and tend to associate with various macromolecules (proteins, DNA, etc.) present in biological systems. If a porphyrin binds to any of a variety of different sites on the proteins/lipids, etc., its microenvironment changes, which causes changes in the values of the lifetime at zero oxygen r () and/or of the oxygen quenching constant K q . This results in heterogeneity of calibration constants and negates the primary strength of the phosphorescence quenching technique. In order to provide phosphors with homogeneous microenvironments that are not altered in the presence of biological macromolecules, efforts are being made to encapsulate porphyrins within molecular structures that would shield them from the surrounding medium. Such superstructures are designed (I) to control the access of oxygen to the encapsulated metalloporphryin, (2) to make the entire phosphor molecule water soluble, (3) to make the phosphor inert to the enzymatic components of blood in order to eliminate all possible toxicity effects, and (4) to keep the molecular size and surface functionalities such that the phosphor is excreted from the blood by kidney filtration. Simple inorganic ionic groups, such as sulfonato groups [1,37,53], have been tried as hydrophilic modifiers of Pd porphyrins. Although porphyrins synthesized by this approach are water soluble and show strong phosphorescence, their oxygen quenching constants (K q ) are in the range of 2000-5000 mm Hg ' sec '. This is higher than is optimal for measurements in the physiological range (0-100 mm Hg). Such high quenching constants indicate that the porphyrinic fragments are highly exposed to oxygen.
651
Phosphorescence Quenching
A protective environment around the porphyrin can, in theory, be provided by attaching large hydrophilic ligands. Different polymers such as polyethylene glycols (PEGs), amino-PEGs, polyvinyl alcohols, carbohydrates, etc., were used to modify Pd porphyrins [74,75]. The resulting porphyrins were highly water soluble but still had excessively high quenching constants (Vinogradov and Wilson, unpublished results). More successful constructs were recently suggested [56,57], in which porphyrins are encapsulated inside dendrimers. Dendrimers are highly branched polymers with uniform molecular weight and well-defined spatial characteristics. It is becoming increasingly popular to use dendrimers for encapsulation of photoactive functionalities [76]. In the first model series of dendritic phosphors, PdTCPP served as a core fragment, while the dendrons were built of L-glutamates. The obtained dendrimer-metalloporphyrins were described by the general formula PdPorphGluNOH, where N = 1-4 (dendrimer generation) and Glu N OH is a glutamic layer (Fig. 4). All polyglutamic Pd-porphyrins have strong phosphorescence with A mux = 690 nm and lifetimes in the range of 0.5-1 msec in deoxygenated water solutions. The phosphorescence is effectively quenched by oxygen, but in water solutions, dendritic "shells" indeed provide protection to the porphyrin core. This was evidenced by a decrease in the values of the oxygen quenching constants with an increase in the dendrimer size. It was also observed that dendrimer-porphyrins had decreased sensitivity to changes in pH, ionic strength, and solute content of the medium. It can be reasonably expected that the next few years will see an ever-ex-
dendrimer generation
terminal group
PdPorphGluNOR
Dendrimer Generation
0
894
2 3 4
1410 2442 4506 8634
1
FIGURE 4
MW
Dendritic polyglutamic Pd porphyrins.
Number of Carboxyls
Number of Glutamates
4 8 16 32 64
0 4 12 28 60
652
Wilson and Vinogradov
panding selection of phosphors with widely "tunable" quenching constants and other relevant physical properties. 6.2
Oxygen Measurements In Vivo: General Properties and Advantages
There is a wide range of normal and pathological conditions for which it is important to know the level of tissue oxygenation. To the present time, oxygen measurements using phosphorescence quenching have been restricted to the research using animal models and to only a few of the many possible applications. Experimental uses of the different types of instruments overlap, in part because the instrument selection was decided by availability rather than by choice. Below we will list applications reported in the literature, the types of instruments used, and the unique aspects of the data obtained. The advantages of the different types of instrumentation for measurements in vivo will be discussed below. There is no evidence for toxicity of the phosphors that are currently in use. For some of the currently used Pd porphyrin-based phosphors, albumin is added in excess over the phosphor in vitro, usually to 1% or 2% by weight in the solution. The albumin has a site(s) that binds the porphyrins with high affinity [2,4,66,67]. This site provides a local lipophilic environment for the porphyrin whereas the rest of the albumin molecule provides the water solubility. The microenvironment provided by albumin protects porphyrin from other molecules in the medium, including oxygen [4,5]. The oxygen quenching constant of Pd porphyrin bound to albumin is only about one-eighth of that corresponding to the same Pd porphyrin simply dissolved in an aqueous buffer [4]. Although there is an excess of albumin in the blood, the porphyrin still has to be prebound to albumin before it can be injected. This can result in pathological reaction to the albumin and/or to minor (in amount) contaminants in the albumin. New water-soluble phosphors, such as Oxyphor R2 (see Sec. 4.1), can be used albumin free, thus avoiding the potential toxicity effects. Preliminary toxicity studies have been made and renal excretion measured for Oxyphor R2 [58J. Even when Oxyphor R2 is injected in amounts 3 and 9 times that needed for phosphorescence imaging, there was no toxicity observed, as evidenced by no alterations in animal behavior or histopathological examination of the tissues 1 and 10 days after the injections. Current studies indicate that it will be possible to prepare synthetic phosphors with microenvironments that will make their phosphorescence characteristics independent of the contents of the medium over the wide range of conditions (pH, ionic strength, other solutes, etc). In addition, the rate of excretion by the kidneys will be selectable, i.e., minutes, hours, or days,
Phosphorescence Quenching
653
depending on the experimental or clinical application. The quenching characteristics of the phosphors designed for in vivo use will be unaffected by proteins in the blood, and the calibrations of Kq and TO in vitro will be fully valid for the studies in vivo. The method is highly specific to oxygen. When phosphors are used that bind to albumin and/or have protective shells, oxygen is generally found to be the only agent in blood that affects the phosphorescence lifetime. If, however, additional agents, such as fluorescent dyes or nuclear magnetic resonance contrast agents, are added to the blood, the possibility of interactions exists and appropriate tests must be made. Each new phosphor also needs to be checked for possible quenching by natural paramagnetic agents such as nitric oxide and organic free radicals, although these are usually present in much lower concentrations than oxygen and are generally less efficient quenchers. Phosphorescence lifetime measurements are not affected by the presence of other chromophors, such as hemoglobin, as long as their absorption does not change in synchrony with phosphorescence decay—a very unlikely event. Other advantages of oxygen measurement by phosphorescence quenching include high accuracy throughout the physiological range, noninvasiveness, high temporal resolution, absolute calibration, and relative ease of design. 6.3
Selected Applications of Phosphorescence Quenching In Vivo
To measure the oxygen levels in a given tissue, it is important to use the variant of technology best suited to the tissue characteristics. For example, oxygen levels in the retina of the eye are best measured by phosphorescence imaging. The retina is a thin object (100-200 /Jim), which is structurally integrated with other important structures, such as the optic nerve head. Vessels draining the optic nerve head cross the surface of the retina, adding another level of complexity to the object. Only with imaging can the different structures be identified and their oxygen levels selectively determined by measuring in appropriate regions [77]. Measurements by point systems (i.e., light guide phosphorometers) can provide better time resolution and are generally more sensitive, but require that imaging be used to select the point of measurement. Oxygen imaging by phosphorescence quenching has been made in the retina of cats [44,78] and pigs [79], whereas point measurements have been used to measure in the anterior chamber [80] and beneath contact lenses placed on the surfaces of rabbit eyes [15]. Oxygen levels in tumors and other solid tissues have been effectively monitored by phosphorescence quenching either using imaging or by point
654
Wilson and Vinogradov
measurements [81-87]. The phosphorescence quenching method has been extensively applied in studies of oxygen delivery by the microcirculation [88-100]. The oxygen concentrations within single microvessels in vivo have been measured by phosphorescence quenching [100]. Phosphors have been reported to leak from the vessels of some tumors, entering into the interstitial space. The resulting phosphorescence has been used to estimate the oxygen gradient from the vessel into the surrounding tissue [101]. When using near-IR phosphors, the points of introducing the excitation light and for collecting the phosphorescence can be separated by a significant thickness (cm) of tissue [45]. Oxygen histograms provide a measure of the degree of hypoxia in the sampled tissue and an accurate assessment of the relative tissue volumes corresponding to each oxygen pressure. Near-IR measurements of phosphorescence in tissue will almost certainly be extended to determining oxygen distributions in three dimensions in the future, but at the moment such imaging systems remain under development. Point measurements are also appropriate for measuring the time course of changes in tissue oxygenation introduced by altered breathing, perfusion, and so forth, since they can be carried out repetitively with high temporal resolution [102-106]. Current instrumentation is capable of determining average oxygen pressures several times per second or oxygen histograms at intervals of 1 sec or slightly longer. These measurement rates are sufficient for most physiologically important changes. "Optical sectioning" can be used to obtain information about the levels of oxygenation at different depths in the tissue. This approach makes use of the fact that the depth of light penetration into tissue is very dependent on the wavelength of the light. Thus, using excitation light of different wavelengths, the measured phosphorescence lifetimes arise from the phosphors at different depths in the tissue. From the differences it is possible to qualitatively estimate the oxygenation at different depths in the tissue. Phosphors are available that absorb light of wavelengths from the ultraviolet to the near-IR. When making surface measurements in which the emission is collected from the same side as the excitation light is applied, the intensity of phosphorescence emitted from different regions declines rapidly with depth of the region, primarily due to attenuation of the excitation light by absorption and scattering. Both scattering and absorption decrease with increase in the wavelength of light. Thus, the average depth of origin of the phosphorescence is very small in the near-UV (<50 jum) but increases to a few hundred micrometers for green light (500-540 nm) and to about 1 mm for near-IR light (630-860 nm). Excitation at the different absorption bands of Pd porphyrins (419 and 524 nm) have been used to show that the oxygen levels in the retina are much lower than those in the underlying choroid layer of the eye. The light with wavelength of 419 nm excites phosphor
Phosphorescence Quenching
655
primarily in the retina, whereas 524-nm light elicits phosphorescence from both the retina and the underlying choroid. Thus, phosphorescent images of the eye, obtained using blue excitation (419 nm), showed that the mean oxygen levels in the retina are of about 35 mm Hg. The phosphorescence decays excited with green light were biphasic. The lifetime of the longer phase of the decay was close to that observed when blue excitation was used, whereas the shorter lifetime indicated much higher oxygen levels (60— 90 mm Hg). The latter is consistent with the much more extensive vasculature in the choroid layer [44]. This optical sectioning approach has been extended to study the oxygenation of kidney in the rat [107]. Two phosphors with different absorption and emission spectra were injected into the blood in the same animal, and oxygen was imaged using excitation at 524 nm (Oxyphor R2) and 635 nm (Green 2W) [45]. The resulting oxygen pressure maps with green excitation (Oxyphor R2) showed lower oxygen pressures than did those with red excitation (Green 2W), indicating that oxygen levels in tissue were significantly higher at greater depths (about 200-500 //jn versus 1-2 mm). 7.
FUTURE TECHNOLOGY DEVELOPMENTS AND APPLICATIONS: A PROSPECTUS
Phosphorescence quenching is revolutionizing the field of oxygen measurements, in some cases markedly extending the realm in which the measurements can be made, and in others increasing the accuracy and resolution of the measurements. Despite what has been accomplished, however, the technology is in its infancy and rapid further development can be expected in the next few years. In addition to continuing improvement in all aspects of individual oxygen measurements, there will be several qualitative advances. These will include (1) time and frequency domain phosphorometers that routinely measure oxygen histograms in samples with heterogeneous distributions of oxygen (such as tissue); (2) real-time (video rate) systems for imaging phosphorescence lifetimes (oxygen), thus allowing direct visualization of changes in oxygen in samples, including tissue, as they occur; (3) systems for imaging oxygen in three dimensions; and (4) new phosphors with a wide range of custom-designed properties, including some optimized for use in vivo. Phosphorescence quenching could also become a clinical method for assessing vascular integrity and oxygen delivery to tissue. As such it would be an important diagnostic tool for the detection and quantification of the extent of a wide range of diseases as well as for evaluating the extent of response to therapy. These would include essentially all vascular diseases, such as those associated with diabetes, major diseases of the eye, tumor
656
Wilson and Vinogradov
detection and treatment, cerebral and cardiac vascular events. However, the clinical potential depends on phosphors being approved for clinical application, and that will require extensive additional testing.
REFERENCES 1. 2.
3.
4.
5.
6. 7. 8.
9.
10.
11.
12.
13.
JM Vanderkooi, DF Wilson. A new method for measuring oxygen concentration in biological systems. Adv Exp Med Biol 200:189-193, 1986. DF Wilson, JM Vanderkooi, TJ Green, G Maniara, SP DeFeo, DC Bloomgarden. A versatile and sensitive method for measuring oxygen. Adv Exp Med Biol 215:71-77, 1987. JB Callis, JM Knowles, M Gouterman. Porphyrins. XXVI. Triplet excimer quenching of free base, zinc, palladium, and platinum complexes. J Phys Chem 77:154-156, 1973. JM Vanderkooi, G Maniara, TJ Green, DF Wilson. An optical method for measurement of dioxygen concentration based upon quenching of phosphorescence. J Biol Chem 262:5476-5482, 1987. DF Wilson, WL Rumsey, TJ Green, JM Vanderkooi. The oxygen dependence of mitochondrial oxidative phosphorylation measured by a new optical method for measuring oxygen concentration. J Biol Chem 263:2712-2718, 1988. JN Demas, BA DeGraff, PB Coleman. Oxygen sensors based on luminescence quenching. Anal Chem 71:793A, 1999. JR Lakowicz. Principles of Fluorescence Spectroscopy, 2nd ed. New York: Plenum Press, 1999. WR Ware, LJ Doemeny, TL Nemzek. Deconvolution of fluorescence and phosphorescence decay curves. A least squares method. J Chem Phys 77: 2038-2048, 1973. PM Gewehr, DT Delpy. Optical oxygen sensor based on phosphorescence lifetime quenching and employing a polymer immobilised metalloporphyrin probe. Part 1. Theory and instrumentation. Med Biol Eng Comp 31:2-10, 1993. JR Alcala, C Yu, GJ Yeh. Digital phosphorimeter with frequency domain signal processing: application to real-time fiber-optic oxygen sensing. Rev Sci Instrum 64:1554-1560. 1993. SC Liao, Z Xu, JA Izatt, JR Alcala. Real-time frequency domain temperature and oxygen sensor with a single optical fiber. IEEE Trans Biomed Eng 44: 1114-1121, 1997. TJ Green, DF Wilson, JM Vanderkooi, SP DeFeo. Phosphorimeters for analysis of decay profiles and real time monitoring of exponential decay and oxygen concentrations. Anal Biochem 174:73-79, 1988. M Pawlowski, DF Wilson. Monitoring of the oxygen pressure in the blood of live animals using the oxygen dependent quenching of phosphorescence. Adv Exp Med Biol 316:179-185, 1992.
Phosphorescence Quenching 14. 15.
16.
17.
18. 19.
20.
21.
22. 23. 24. 25.
26. 27.
28.
29. 30.
31.
657
IP Torres-Filho, M Intaglietta. Microvessel PO2 measurements by phosphorescence decay method. Am J Physiol 265:H1434-H1438, 1993. DM Harvitt, JA Bonanno. Direct noninvasive measurement of tear oxygen tension beneath gas-permeable contact lenses in rabbits. Invest Ophthal Vis Sci 37:1026-1036, 1996. LW Lo, CJ Koch, DF Wilson. Calibration of oxygen-dependent quenching of the phosphorescence of Pd-meso-tetra (4-carboxyphenyl) porphine: a phosphor with general application for measuring oxygen concentration in biological systems. Anal Biochem 236:153-160, 1996. M Sinaasappel, C Ince. Calibration of Pd-porphyrin phosphorescence for oxygen concentration measurements in vivo. J Appl Phys 81:2297-2303, 1996. Oxygen Enterprises Ltd.: www.oxygenent.net SA Vinogradov, DF Wilson. Phosphorescence lifetime analysis with a quadratic programming algorithm for determining quencher distributions in heterogeneous systems. Biophys J 67:2048-2059, 1994. JG McWhirter, ER Pike. On the numerical inversion of the Laplace transform and similar Fredholm integral equations of the first kind. J Phys 11:17291745, 1978. AS Golub, AS Popel, L Zheng, RN Pittman. Analysis of phosphorescence in heterogeneous systems using distributions of quencher concentration. Biophys J 73:452-465, 1997. AN Tikhonov. Solutions of Ill-Posed Problems. London: John-Wiley & Sons, 1977. J Skilling, RK Bryan. Maximum entropy image reconstruction: general algorithm. Mon Not R Astron Soc 211:111-124, 1984. SF Gull, J Skilling. Maximum entropy method in image processing. IEEE Proc 131:646-659, 1984. AK Livesey, JC Brochon. Analyzing the distribution of decay constants in pulse-fluorometry using maximum entropy method. Biophys J 52:693-706, 1987. DR James, WR Ware. Recovery of underlying distributions of lifetimes from fluorescence decay data. Chem Phys Lett 126:7-11, 1986. A Semiarczuk, BD Wagner, WR Ware. Comparison of the maximum entropy method and exponential series method for the recovery of distributions of lifetimes from fluorescence lifetime data. J Phys Chem 94:1661-1666, 1990. R Ware. Recovery of fluorescence lifetime distributions in heterogeneous systems. In: V. Ramamurthy, ed. Photochemistry in Organized and Constrained Media. New York: VCH, 1991. OxySpot-1, Medical Systems Inc., Greenvale, NY. SA Vinogradov, DF Wilson. Recursive maximum entropy algorithm and its application to the luminescence lifetime distribution recovery. Appl Spectrosc 54:849-855, 2000. SA Vinogradov, MA Fernandez-Seara, BW Dugan, DF Wilson. A method for measuring oxygen distributions in tissue using frequency domain phosphorometry. Comp Biochem Physiol A, 132:147-152, 2002
658
Wilson and Vinogradov
32.
SA Vinogradov, MA Fernandez-Seara, BW Dugan, DF Wilson. Frequency domain instrument for measuring phosphorescence lifetime distributions in heterogeneous samples. Rev Sci Instrum 72:3396-3406, 2001. DF Wilson, SA Vinogradov, BW Dugan, D Biruski, L Waldron, SA Evans. Tumor oxygenation measured using new frequency domain phosphorometers. Comp Biochem Physiol 132:153-159, 2002. WL Rumsey, JM Vanderkooi, DF Wilson. Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue. Science 241:1649-1651, 1988. DF Wilson, A Pastuszko, R Schneiderman, JE DiGiacomo, M Pawlowski, M Delivoria-Papadopoulos. Effect of hyperventilation on oxygenation of the brain cortex of neonates. Adv Exp Med Biol 316:341-346, 1992. C Ince, JP van der Sluijs, M Sinaasappel, JA Avontuur, JM Coremans, HA Bruining. Intestinal ischemia during hypoxia and experimental sepsis as observed by NADH videofluorimetry and quenching of Pd-porphine phosphorescence. Adv Exp Med Biol 361:105-110, 1994. CH Barlow, DA Rorvik, JJ Kelly. Imaging epicardial oxygen. Ann Biomed Eng 26:76-85. 1998. S Lahiri, WL Rumsey, DF Wilson, R Iturriaga. Contribution of in vivo microvascular PO 2 in the cat carotid body chemotransduction. J Appl Physiol 75:1035-1043, 1993. WL Rumsey, M Pawlowski, N Lejavardi, DF Wilson. Oxygen pressure distribution in the heart in vivo and evaluation of the ischemic "border zone." Am J Physiol 266:H1676-H1680, 1994. WL Rumsey. M Pawlowski, N Lejavardi, DF Wilson. Measurement of oxygen pressure in the heart in vivo using phosphorescence quenching. Adv Exp Med Biol 361:93-97, 1994. WL Rumsey. R Iturriaga, D Spergel. S Lahiri, DF Wilson. Optical measurements of the dependence of chemoreception on oxygen pressure in the cat carotid body. Am J Physiol 261:C614-C622, 1991. WL Rumsey, R Iturriaga, DF Wilson, S Lahiri, D Spergel. Phosphorescence and fluorescence imaging: new tools for the study of carotid body function. New York: Plenum Press, 1990, pp 73-79. M Pawlowski, DF Wilson. Imaging oxygen pressure in tissue in vivo by phosphorescence decay. Adv Exp Med Biol 361:83-91, 1994. RD Shonat. DF Wilson, CE Riva. Oxygen distribution in the retinal and choroidal vessels of the cat as measured by a new phosphorescence imaging method. Appl Opt 31:3711-3718. 1992. SA Vinogradov, L-W Lo, WT Jenkins, SM Evans, C Koch, DF Wilson. Noninvasive imaging of the distribution of oxygen in tissue in vivo using near infra-red phosphors. Biophys J 70:1609-1617, 1996. RD Shonat. ES Wachman, W Niu, AP Koretsky, DL Farkas. Near-simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope. Biophys J 73:1223-1231, 1997. JN Demas, BA DeGraff. Applications of luminescent transition metal com-
33.
34.
35.
36.
37. 38.
39.
40.
41.
42.
43. 44.
45.
46.
47.
Phosphorescence Quenching
48.
49.
50.
51.
52. 53.
54.
55.
56.
57.
58. 59. 60.
61. 62. 63.
659
plexes to sensor technology and molecular probes. J Chem Educ 74:690-696, 1997. A Mills, A Lepre, BRC Theobald, E Slade, BA Murrer. Use of luminescent gold compounds in the design of thin-film oxygen sensors. Anal Chem 69: 2842-2847, 1997. W Xu, KA Kneas, JN Demas, BA DeGraff. Oxygen sensors based on luminescence quenching of metal complexes: Os(II) complexes suitable for laser diode excitation. Anal Chem 68:2605-2609, 1996. G Di Marco, M Lanza, M Antonio, I Stefio, C Di Petro, G Romeo, S Campagna. Luminescent mononuclear and dinuclear Ir(III) complexes immobilized in a polymeric matrix as solid-state oxygen sensors. Anal Chem 70: 5019-5023, 1998. L Sacksteder, JN Demas, BA DeGraff. Design of oxygen sensors based on quenching of luminescent metal complexes: effect of ligand size on heterogeneity. Anal Chem 65:3480-3483, 1993. D Eastwood, M Gouterman. Porphyrins. XVIII. Luminescence of (Co), (Ni), Pd, Pt complexes. J Mol Spectrosc 35:359-375, 1970. SA Vinogradov, DF Wilson. Metallotetrabenzoporphyrins. New phosphorescent probes for oxygen measurements. J Chem Soc Perkin Trans II 103-111, 1994. DB Papkovsky, GV Ponomarev, W Trettnak, P O'Leary. Phosphorescent complexes of porphyrin-ketones: optical properties and application to oxygen sensing. Anal Chem 67:4112-4117, 1995. AM Gonsalves, RAW Johnstone, MM Pereira, AMP Desantana, AC Serra, AFFN Sobral, PA Stocks. New procedures for the synthesis and analysis of 5,10,15,20-tetrakis(sulphophenyl)porphyrins and derivatives through chlorosulphonation. Heterocycles 43:829-838, 1996. SA Vinogradov, DF Wilson. "Dendritic" porphyrins—new protected phosphors for oxygen measurements in vivo. Adv Exp Med Biol 428:657-662, 1997. SA Vinogradov, LW Lo, DF Wilson. Dendritic polyglutamic porphyrins: probing porphyrin protection by oxygen-dependent quenching of phosphorescence. Chem Eur J 5:1338-1347, 1999. LW Lo, SA Vinogradov, CJ Koch, DF Wilson. A new, water soluble, phosphor for oxygen measurements in vivo. Adv Exp Med Biol 428:651-656, 1997. SA Vinogradov, DF Wilson. Extended porphyrins. New IR phosphors for oxygen measurements. Adv Exp Med Biol 411:597-603, 1997. SA Vinogradov, DF Wilson. Metallotetrabenzoporphyrins. New phosphorescent probes for oxygen measurements. J Chem Soc Perkin Trans II 2:103111, 1995. O Finikova, A Cheprakov, I Beletskaya, S Vinogradov. An expedient synthesis of substituted tetraaryltetrabenzoporphyrins. Chem Commun 261-262, 2001. SA Vinogradov, DF Wilson. Palladium catalyzed carbonylation of Br-substituted porphyrins. Tetrahedron Lett 39:8935-8938, 1998. AV Khoroshutin, SA Vinogradov, DF Wilson. Synthesis of halogenated tetra-
660
64.
65.
66. 67.
68. 69.
70.
71.
72.
73.
74. 75.
76.
77.
78.
Wilson and Vinogradov benzoporphyrins: a route to functionalized tetrabenzoporphyrin derivatives. Abstr. Pap. ACS, 215: 49-ORGN, U16-U16 Part 2, 1998. SA Vinogradov, DF Wilson. Dendrimers with tetrabenzoporphyrin core: novel near-IR phosphors for O2 measurements. Abstr. Pap. ACS, 217: 50-INOR, U1122-U1122 Part 1, 1999. RD Shonat, PC Johnson. Oxygen tension gradients and heterogeneity in venous microcirculation: a phosphorescence quenching study. Am J Phys 272: H2233-H2240, 1997. M Robiolio, WR Rumsey, DF Wilson. Oxygen diffusion and mitochondrial respiration in neuroblastoma cells. Am J Physiol 256:C1207-C1213, 1989. WL Rumsey, C Schlosser, EM Nuutinen, M Robiolio, DF Wilson. Cellular energetics and the oxygen dependence of respiration in cardiac myocytes isolated from adult rat. J Biol Chem 265:15392-15399, 1990. G-E Khalil, M Gouterman. Method for measuring oxygen concentration. US Patent 4,810,655, March 7, 1989. PM Gewehr, DT Delpy. Optical oxygen sensor based on phosphorescence lifetime quenching and employing a polymer immobilised metalloporphyrin probe. Part 2. Sensor membranes and results. Med Biol Eng Comp 31:1121, 1993. PM Gewehr, DT Delpy. Analysis of non-linearity of optical oxygen sensors based upon phosphorescence lifetime quenching. Med Biol Eng Comp 32: 659-664, 1994. SC Liao, Z Xu, JA Izatt, JR Alcala. Real-time frequency domain temperature and oxygen sensor with a single optical fiber. IEEE Trans Biomed Eng 44: 1114-1121, 1997. DB Papkovskii, AI Yaropolov, AP Savitskii, J Olah, VD Rumyantseva, AF Mironov, IV Troyanovskii, NA Sadovskii. Fiber-optic oxygen sensor based on phosphorescence quenching. Biomed Sci 2:536-539, 1991. I Lahdesmaki, LD Scampavia, C Beeson, J Ruzicka. Detection of oxygen consumption of cultured adherent cells by bead injection spectroscopy. Anal Chem 71:5248-5252, 1999. SA Vinogradov, DF Wilson. Extended porphyrins—new IR phosphors for oxygen measurements. Adv Exp Med Biol 411:597-603, 1997. DB Papkovsky, AN Ovchinnikov, GV Ponomarev, T Korpela. Complexes of the phosphorescent porphyrin dyes with hydrophilic carriers: optical properties and sensing applications. Anal Lett 30:699-716, 1997. See, for example: M Fischer, F Vogtle. Dendrimers: from design to application —a progress report. Angew Chem Int Ed 38:885-905, 1999; DK Smith, F Diederich. Functional dendrimers: unique biological mimics. Chem Eur J 4: 1353-1361. 1998. SD Cranstoun, CE Riva, JL Munoz, CJ Pournaras. Continuous measurements of intravascular pO 2 in the pig optic nerve head. Klin Monatsblatter Augenheilkunde 210:313-315, 1997. RD Shonat, DF Wilson, CE Riva, SD Cranstoun. Effect of acute increases in intraocular pressure on intravascular optic nerve head oxygen tension in cats. Invest Ophthal Vis Sci 33:3174-3180, 1992.
Phosphorescence Quenching 79.
80.
81.
82. 83.
84.
85.
86.
87.
88.
89.
90.
91.
92. 93.
661
S Blumenroder, AJ Augustin, FH Koch. The influence of intraocular pressure and systemic oxygen tension on the intravascular pO2 of the pig retina as measured with phosphorescence imaging. Surv Ophthal 42:8118-8126, 1997. JW McLaren, 8 Dinslage, JP Dillon, JE Roberts, RF Brubaker. Measuring oxygen tension in the anterior chamber of rabbits [published erratum appears in Invest Ophthal Vis Sci 1999 Nov;40(12):2779]. Invest Ophthal Vis Sci 39: 1899-1909, 1998. DF Wilson, GJ Cerniglia. Localization of tumors and evaluation of their state of oxygenation by phosphorescence imaging. Cancer Res 52:3988-3993, 1992. DF Wilson, GJ Cerniglia. Oxygenation of tumors as evaluated by phosphorescence imaging. Adv Exp Med Biol 345:539-547, 1994. SM Evans, CJ Koch, KM Laughlin, WT Jenkins, T Van Winkle, DF Wilson. Tamoxifen induces hypoxia in MCF-7 xenografts. Cancer Res 57:5155-5161, 1997. MW Dewhirst, ET Ong, RD Braun, B Smith, B Klitzman, SM Evans, D Wilson. Quantification of longitudinal tissue pO2 gradients in window chamber tumors: impact on tumour hypoxia. Br J Cancer 79:1717-1722, 1999. R Linberg, CD Conover, KL Shum, RG Shorn Increased tissue oxygenation and enhanced radiation sensitivity of solid tumors in rodents following polyethylene glycol conjugated bovine hemoglobin administration. In Vivo 12: 167-173, 1998. CD Conover, R Linberg, KL Shum, RG Shorn The ability of polyethylene glycol conjugated bovine hemoglobin (PEG-Hb) to adequately deliver oxygen in both exchange transfusion and top-loaded rat models. Artif Cells Blood Subst Immob Biotechnol 27:93-107, 1999. G Helmlinger, F Yuan, M Dellian, RK Jain. Interstitial pH and pO2 gradients in solid tumors in vivo: high-resolution measurements reveal a lack of correlation. Nat Med 3:177-182, 1997. RD Shonat, KN Richmond, PC Johnson. Phosphorescence quenching and the microcirculation: an automated, multipoint oxygen tension measuring instrument. Rev Sci Instrum 66:5075-5084, 1995. DG Buerk, AG Tsai, M Intaglietta, PC Johnson. Comparing tissue PO2 measurements by recessed microelectrode and phosphorescence quenching. Adv Exp Med Biol 454:367-374, 1998. DG Buerk, AG Tsai, M Intaglietta, PC Johnson. In vivo tissue pO2 measurements in hamster skinfold by recessed pO2 microelectrodes and phosphorescence quenching are in agreement. Microcirculation 5:219-225, 1998. D Erni, H Sakai, AG Tsai, A Banic, GH Sigurdsson, M Intaglietta. Haemodynamics and oxygen tension in the microcirculation of ischaemic skin flaps after neural blockade and haemodilution. Br J Plast Surg 52:565-572, 1999. M Intaglietta, PC Johnson, RM Winslow. Microvascular and tissue oxygen distribution. Cardiovasc Res 32:632-643, 1996. H Kerger, DJ Saltzman, MD Menger, K Messmer, M Intaglietta. Systemic and subcutaneous microvascular Po2 dissociation during 4-h hemorrhagic shock in conscious hamsters. Am J Physiol 270:H827-H836, 1996.
662
Wilson and Vinogradov
94.
H Kerger, DJ Saltzman, A Gonzales, AG Tsai, K van Ackern, RM Winslow, M Intaglietta. Microvascular oxygen delivery and interstitial oxygenation during sodium pentobarbital anesthesia. Anesthesiology 86:372-386, 1997. DC Poole, PD Wagner, DF Wilson. Diaphragm microvascular plasma pO2 measured in vivo. J Appl Physiol 79:2050-2057, 1995. KN Richmond. RD Shonat, RM Lynch, PC Johnson. Critical pO2 of skeletal muscle in vivo. Am J Physiol 277:H183 1-H1840, 1999. IP Torres Filho, M Intaglietta. Microvessel PO2 measurements by phosphorescence decay method. Am J Physiol 265:H1434-H1438, 1993. IP Torres Filho, H Kerger, M Intaglietta. pCK measurements in arteriolar networks. Microvasc Res 51:202-212, 1996. AG Tsai, B Friesenecker, MC Mazzoni, H Kerger, DG Buerk, PC Johnson, M Intaglietta. Microvascular and tissue oxygen gradients in the rat mesentery. Proc Natl Acad Sci USA 95:6590-6595. 1998. L Zheng, AS Golub, RN Pittman. Determination of PO2 and its heterogeneity in single capillaries. Am J Physiol 271 :H365-H372, 1996. IP Torres Filho, M Leunig, F Yuan, M Intaglietta, RK Jain. Noninvasive measurement of microvascular and interstitial oxygen profiles in a human tumor in SCID mice. Proc Natl Acad Sci USA 91:2081-2085, 1994. O Tammela, A Pastuszko, NS Lajevardi, M Delivoria-Papadopoulos, DF Wilson. Activity of tyrosine hydroxylase in the striatum of newborn piglets in response to hypocapnic hypoxia. J Neurochem 60:1399-1406, 1993. OK Tammela, N Lajevardi, CC Huang, DF Wilson, M Delivoria-Papadopoulos, A Pastuszko. The effects of induced apneic episodes on cerebral cortical oxygenation in newborn piglets. Brain Res 741:160-165, 1996. D Song, M Olano. DF Wilson, A Pastuszko, O Tammela, K Nho, RG Shorr. Comparison of the efficacy of blood and polyethylene glycol-hemoglobin in recovery of newborn piglets from hemorrhagic hypotension: effect on blood pressure, cortical oxygen, and extracellular dopamine in the brain. Transfusion 35:552-558, 1995. D Song, J Marczis, M Olano, AG Kovach, D Wilson, A Pastuszko. Effect of hemorrhagic hypotension on cortical oxygen pressure and striatal extracellular dopamine in cat brain. Neurochem Res 22:1111-1117, 1997. M Olano. D Song, S Murphy, DF Wilson, A Pastuszko. Relationships of dopamine, cortical oxygen pressure, and hydroxyl radicals in brain of newborn piglets during hypoxia and posthypoxic recovery. J Neurochem 65: 1205-1212, 1995. WR Rumsey, B Abbott, LW Lo, SA Vinogradov, DF Wilson. Imaging of oxygen in the surface and deep areas of the kidney. Adv Exp Med Biol 411: 591-596. 1997.
95. 96. 97. 98. 99.
100. 101.
102.
103.
104.
105.
106.
107.
Index
Absorption, 2-7, 33, 64, 73, 84, 119, 145, 153, 157, 183-184, 225, 237, 243, 404, 647, 654 Amino acids, 214, 240 Aminolevulinic acid (ALA), 364, 475, 536, 600-619 Angiogenesis, 249, 436-438, 446 Antibody, 470, 483, 571, 588 Apoptosis, 435-436 Arterial, 398-402, 409, 414, 415-425 Autofluorescence, 58, 62, 153-154, 186-187, 192-193, 195-196, 203, 226-232, 240-247, 315317, 333-338, 340-344, 363368, 379-385, 533, 545, 546 Basal cell carcinoma, 344, 347, 350357, 563 Benzoporphyrin derivative (BPD-MA), 4, 5, 14, 165-168, 551, 554, 597 Carcinogenesis, 249 Cervical cancer, 257, 265-273 Cervix, 255-257 Collagen, 187, 195, 217, 364, 423, 426 Computation, 35, 502-519
Confocal microscopy, 143, 145, 154155, 161-167, 182-184, 223, 240 Continuous wave (CW), 34, 41, 50, 54, 220, 369, 454-455, 490-494 Contrast agents, 385-386, 467-489, 645-647, 650 Cytometry, 197 Diffusion, 22, 29-58, 37, 410, 492498, 548 Diffusion equation, 36, 41, 50, 54, 85, 492-498 Dysplasia, 257, 265-273, 378-385 Elastin, 187, 195, 217, 364, 398 Endogenous, 144, 158, 186-187, 214217, 226-232, 237-261, 363368, 409-412 Energy transfer, 25-27, 214 Epithelium, 248-257 Escape function, 39, 57, 327-331 Exogenous, 58, 144, 187-188, 363364, 467-489
FAD, 240, 247 Finite difference, 35, 55, 157, 245 663
Index
664
Fluorescein, 139, 160-163, 165,468469 Fluorescence lifetime. 15, 17-20, 37, 40, 153, 211-232, 373, 375, 402, 403. 408, 4 1 1 , 415, 423, 449450, 551 Fluorophores, 15, 17, 34, 37, 65, 144145, 151-154, 158-159, 163, 167-168, 170-171, 173, 214217, 280-289, 409-411, 415, 422-423, 559 Fourier transform, 47, 49 Frequency domain, 30, 41, 46, 220222, 403, 421, 458-467, 495496, 639-641 Green fluorescent protein (GFP), 144, 164-166, 188, 431-435, 472 Green's function. 47, 54 Hypoxia, 439, 654 Imaging, 149, 165, 170, 213-214, 296, 370-373, 445, 448, 499-518, 617 Indocyanine green (ICG), 469-473, 475-483 Internal conversion, 5-6 Intersystem crossing, 7, 646 Inverse problem, 57, 448, 499-518 Luciferase, 473 Lung cancer, 361-396 Lysosomes, 247 Metastasis, 215, 275-278, 438 Microscopy, 143-173, 181-204.211232, 554 Mitochondria. 193, 216-217, 247, 601 Mode-locked laser, 182. 184 Monte Carlo, 35, 62, 158, 190, 258, 557 NAD(P)H, 110, 128, 133-135, 186187, 193. 215-217. 226-228, 239, 240, 247. 257, 364, 551
Near infrared (NIR), 30, 157,173, 227228, 445-519, 647, 654 Nucleus, 228 Optical fiber, 37, 451-452, 556 Optical sectioning, 146-148, 182, 223226, 654 Oxygen, 27, 129, 213, 216, 528, 643, 655
Pathology, 181, 338-339 Peptide, 484-487 Phantoms, 249-252 Phosphorescence, 637, 645-647, 650 Photobleaching, 145, 154, 160, 165, 167, 184, 338-339, 533, 549 Photodamage, 161, 167, 173, 184, 191192, 227-228, 551 Photodynamic, 58, 385, 529, 564-570 Photosensitizer, 24, 468-473, 474-475, 529. 564-570, 591-599 Point spread function (PSF), 65, 145, 150, 164, 170, 189, 190, 225226 Porphyrins, 243, 474-475, 591-599, 646, 650-651 Proteins, 240 Protoporphyrin IX (PPIX), 364, 537, 600-619 Quantification, 536, 553 Quantum yield, 12-17, 34, 37, 63, 75, 154, 158-159, 162-163, 165166, 212, 238, 449-450, 539 Quenching, 22-25, 637, 653 Radiative transport, 34, 46 Reconstruction, 112-113, 203, 325333, 448, 499-518 Reflectance, 30, 48, 57, 84, 112-113, 144, 147, 150, 154, 159, 163, 170, 183-184, 508 Regularization, 513 Rhodamine, 164
665
Index Scattering, 30, 34, 43, 63, 116-119, 145, 157, 189-191, 225, 230, 238, 237, 245, 404, 420, 654 Sectioning, 144, 169, 170-171, 182184, 190, 654 Skin cancer, 343-346, 350-357, 563, 603 Spectroscopy, 2, 10-12, 129, 133, 200, 247, 291, 341-342, 366-377, 401, 402-409, 414, 418, 419, 553 Time-domain, 19, 20, 30, 157, 217220, 222, 402-404, 417, 421, 455-458, 494-495, 641-642 Time-resolved, 19, 20, 34, 54, 217-222, 369, 373, 375, 397, 402-411,
414, 418, 422, 415-421, 448, 499-518 Tissue, 154-155, 159, 161, 173, 230232, 403-409 Tomography, 226, 496-519 Toxicity, 385, 652 Turbid media, 61-106, 143, 158, 223226, 447 Two-photon microscopy, 181-209, 224 Vasculature, 446, 642 Video rate, 154, 158-159, 160-169, 170-172, 194-196, 231-232, 372-373, 375-377 Wide-field imaging, 224-225
