In vivo determination of the absorption and scattering spectra of the human prostate during photodynamic therapy Jarod C. Finlay*, Timothy C Zhu, Andreea Dimofte, Diana Stripp, S. Bruce Malkowicz, Richard Whittington, Jeremy Miles, Eli Glatstein and Stephen M. Hahn Department of Radiation Oncology, University of Pennsylvania, 3400 Spruce St., 2 Donner Building, Philadelphia, PA USA 19104 ABSTRACT A continuing challenge in photodynamic therapy is the accurate in vivo determination of the optical properties of the tissue being treated. We have developed a method for characterizing the absorption and scattering spectra of prostate tissue undergoing PDT treatment. Our current prostate treatment protocol involves interstitial illumination of the organ via cylindrical diffusing optical fibers (CDFs) inserted into the prostate through clear catheters. We employ one of these catheters to insert an isotropic white light point source into the prostate. An isotropic detection fiber connected to a spectrograph is inserted into a second catheter a known distance away. The detector is moved along the catheter by a computer-controlled step motor, acquiring diffuse light spectra at 2 mm intervals along its path. We model the fluence rate as a function of wavelength and distance along the detector’s path using an infinite medium diffusion theory model whose free parameters are the absorption coefficient µa at each wavelength and two variables A and b which characterize the reduced scattering spectrum of the form µ’s = Aλ-b. We analyze our spectroscopic data using a nonlinear fitting algorithm to determine A, b, and µa at each wavelength independently; no prior knowledge of the absorption spectrum or of the sample’s constituent absorbers is required. We have tested this method in tissue simulating phantoms composed of intralipid and the photosensitizer motexafin lutetium (MLu). The MLu absorption spectrum recovered from the phantoms agrees with that measured in clear solution, and µa at the MLu absorption peak varies linearly with concentration. The µ’s spectrum reported by the fit is in agreement with the known scattering coefficient of intralipid. We have applied this algorithm to spectroscopic data from human patients sensitized with MLu (2 mg kg-1) acquired before and after PDT. Before PDT, the absorption spectra we measure include the characteristic MLu absorption peak. Using our phantom data as a calibration, we have determined the pre-treatment MLu concentration to be approximately 2 to 8 mg kg-1. After PDT, the concentration is reduced to 1 to 2.5 mg kg-1, an indication of photobleaching induced by irradiation. In addition, absorption features corresponding to the oxygenated and deoxygenated forms of hemoglobin indicate a reduction in tissue oxygenation during treatment.
Keywords: photodynamic therapy, motexafin lutetium, in-vivo light dosimetry, tissue optical properties, diffusion theory, diffuse reflectance.
*
132
[email protected]; phone (215) 615-3795; fax (215) 349-5978; http://www.xrt.upenn.edu/radiation_physics
Optical Methods for Tumor Treatment and Detection: Mechanisms and Techniques in Photodynamic Therapy XIII, edited by David Kessel, Proc. of SPIE Vol. 5315 (SPIE, Bellingham, WA, 2004) · 1605-7422/04/$15 · doi: 10.1117/12.528968
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
1
Introduction
The accurate in vivo determination of the optical properties of human tissues is an ongoing challenge in the optimization of photodynamic therapy (PDT). The distribution of treatment light within the target tissue is determined by the absorption and scattering properties of the tissue itself. Accurate characterization of the delivered light dose depends critically on these properties, especially for parts of the tissue remote from the light source. Measurement of tissue optical properties also has the potential to allow characterization of the distribution of sensitizer concentration and the oxygenation status of hemoglobin.1-3 In combination, these parameters have the potential to allow complete characterization of the factors that affect the efficacy of photodynamic treatment, namely the local light fluence rate and the local concentrations of sensitizer and oxygen. In addition, it is possible that the photodynamic degradation of the sensitizer as determined by the decrease in the sensitizer component of the absorption spectrum can provide an implicit measure of photodynamic dose.4 In this paper, we investigate the absorption and scattering spectra of human prostate tissue sensitized with the secondgeneration photosensitizer motexafin lutetium (MLu).5, 6 The absorption peak of MLu typically used for treatment is centered at 732 nm, a wavelength where the inherent absorption of tissue is relatively weak. While the shift to longer wavelength absorption in the second-generation sensitizers was primarily intended to increase the penetration of excitation light, it has the additional advantage from the point of view of absorption spectroscopy that the absorption arising from the sensitizer does not compete with a dominant background of hemoglobin absorption. On the other hand, the absorption band of MLu is narrow enough that it does not obscure the absorption features of hemoglobin. Under these circumstances, the sensitizer and oxy- and deoxyhemoglobin can be characterized independently on the basis of their absorption spectra. To obtain the optical properties of tissue, we use an analytic model of light propagation in turbid media. Several researchers have developed models compatible with CW measurements based on the diffusion approximation to the Boltzmann radiative transport equation.1, 7, 8 In this paper, we make use of the higher-order P3 approximation, which is valid over a wider range of optical properties and source-detector separations.9, 10 This approximation has been shown effective for the independent determination of the absorption and scattering coefficients of turbid media.11 We combine an infinite-medium P3 expression with a nonlinear fitting algorithm that assumes a reduced scattering spectrum of the form µ’s = Aλ-b. Constraining the reduced scattering spectrum to take this form increases the robustness of the algorithm and reduces the chances of crosstalk between the absorption and scattering coefficients.
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
133
2 2.1
Methods
Data acquisition
We have obtained diffuse fluence spectra using the instrument shown in figure 1. White light is provided by a tungsten lamp. Light from this source is coupled into a 200 micron fiber terminated in a spherical isotropic scattering tip. An identical isotropic scattering fiber is used to detect diffuse light. The measured signal is therefore proportional to the fluence rate in the medium Light collected by this fiber is imaged through a grating spectrograph onto a nitrogencooled CCD camera. Spectrograph & CCD
PC 1
Tungsten Lamp
PC 2
Figure 1: Experimental setup for spectroscopic measurements in turbid media. Computer 1 controls the motors, while computer 2 records the spectra acquired by the spectrograph and CCD.
Motorized positioners
Source
Detector
Scattering medium
Both fibers are inserted into the sample through clear catheters. The catheters are parallel and separated by 3 to 8 mm. Each fiber is connected to a computer-controlled step motor, allowing the source and detector fibers to be moved along their respective catheters independently. The characterization of optical properties requires a measurement of the fluence rate over a range of source-detector separations. This is accomplished by fixing the source at one location and moving the detector relative to it. Diffuse reflectance spectra are obtained at 2 mm intervals starting 4 mm on one side of the source and continuing to 16 mm beyond it on the other side. The origin of the detector position, i.e. the position which gives the shortest source-detector separation, is determined by a prior scan with a photodioide and is verified by checking the symmetry of the peak. The movement of the motors is controlled by software developed in our laboratory, which also triggers the collection of data by the CCD system. The minimum source-detector separation achievable with this system is determined by the separation between catheters, in this case 3 to 8 mm. As a result, no signal can be detected in samples whose absorption coefficient is greater than approximately 2 cm-1. In human tissue, this limit is exceeded at visible wavelengths sue to the absorption of light by hemoglobin. We therefore consider only data acquired at wavelengths longer than 650 nm in the analysis presented below. 2.2
Spectral analysis
The diffuse fluence spectra captured by the CCD are first divided by the spectra of the same source obtained in an integrating sphere. We have characterized the relationship between the input power and the resulting fluence rate inside
134
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
this particular integrating sphere. This division therefore gives a calibrated measure of the relative diffuse fluence rate of the sample, i.e .the ratio of the fluence rate to the power of the point source. We model the fluence rate in the sample using the P3 expression for fluence rate in an infinite medium developed by Hull and Foster11 based on the previous work of Star10 and Boas et al.9 The determination of the optical properties of the sample proceeds in two steps. First, we select a subset of the wavelengths at which the diffuse reflectance is measured. The data at these wavelengths are fit using a global non-linear fitting algorithm with a model that assumes a reduced scattering coefficient of the form µ’s = Aλ-b. This form of the reduced scattering spectrum, which is based on an approximation to the Mie scattering theory, has been widely used to model tissue scattering,1, 12, 13 and also approximates the reduced scattering spectrum of Lyposyn.14 The fitting algorithm’s free parameters are A, b, and the absorption coefficient at each wavelength. Nine wavelengths are usually sufficient to determine A and b accurately. In the second stage of fitting, the fluence rate profile at each wavelength is fit individually. The value of µ’s is fixed at the value determined in the first step, and only µa is varied. This fitting is repeated for each wavelength. The result of this fitting process is a complete absorption spectrum of the sample and two parameters A and b that characterize the scattering spectrum of the sample. Once the absorption spectrum is determined, the concentrations of absorbers with known spectra can be determined by linear fitting. To accomplish this fitting, we have implemented the singular value decomposition (SVD) algorithm of Press et al.15 implemented in Matlab (the MathWorks, Natick, MA). Following the method of Hull et al.,16 we have included a Fourier synthesis in the fitting to account for the presence of unknown absorbers. Our Fourier series is weighted exponentially to reduce high-frequency noise.17 The basis spectra used for MLu and ink are based on measurements in clear solutions. For in vivo fitting, we have adopted hemoglobin basis spectra from the work Wray et al.18 2.3
Phantom preparation
Tissue simulating phantoms were prepared from Liposyn and water. Liposyn has been shown to have a scattering spectrum similar to that of tissue.12, 14 The total lipid concentration in these phantoms was 0.5%, yielding a reduced scattering coefficient of approximately 6 cm-1 at 730 nm. MLu was added to the phantom in concentrations similar to that expected in tissue. Each phantom has a total volume of 2L, large enough to be considered infinite for the purposes of our analysis. The phantoms were housed in a large container whose sides were blackened to prevent the entrance of light from the outside and to reduce reflections. We present the results of two sets of phantoms, one with MLu as the only absorber, and another in which we also added black ink to simulate the presence of other absorbers in tissue. Like hemoglobin, the black ink used in these experiments exhibits relatively low absorption in the wavelengths near the MLu absorption peak, but dominates the absorption at shorter wavelengths. It therefore provides a reasonable test of the ability of our measurement and analysis system to determine the concentration of MLu spectroscopically. 2.4
In vivo data acquisition
The patients involved in this study were part of an ongoing clinical trial of PDT for the treatment of recurrent prostate cancer after radiation therapy. Prior to treatment, each patient’s prostate was sized by transrectal ultrasound. The prostate was divided into quadrants, and the placement of light sources to illuminate each quadrant was determined. The treatment was planned such that each quadrant received two or four treatment catheters and one measurement catheter. Each patient was given MLu intravenously at a dose and time predetermined by the dose-escalation schedule of the clinical protocol. The patient was then anesthetized, and the plastic catheters were inserted according to the predeveloped plan. Interstitial irradiation was performed with 732-nm light at a maximum fluence rate of 150 mW cm-2. Prior to and immediately following treatment, spectroscopic measurements where made in each quadrant. The white light source fiber was inserted into one of the treatment catheters, and the detector fiber was inserted into the measurement catheter. During treatment, cylindrical light sources were placed in the treatment catheters, and a single isotropic detector used for in vivo light dosimetry was inserted into each detection catheter. After treatment, these fibers were removed, and the spectroscopic measurements were repeated. The spectroscopic measurements therefore did not require the placement of additional catheters but rather made use of those catheters necessary for treatment.
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
135
3 3.1
Results
Verification in phantoms
In figure 2, we plot the fluence rate measured in a typical phantom as a function of source-detector separation at the nine wavelengths used in the initial fitting step to determine µ’s. The symbols indicate the measured data, corrected for instrument response. The lines indicate the best fit provided by the P3 model to the data.
3 443.6 489.4 534.8 579.7 624.2 668.3 712.0 755.3 798.1
2.5
Φ
2
nm nm nm nm nm nm nm nm nm
Figure 2: Fluence rate as a function of source-detector separation for the nine wavelengths used to determine the scattering parameters A and b. The solid lines indicate the best fits obtained using the P3 theory.
1.5
1
0.5
0 0.6
0.8
1
1.2 r (cm)
1.4
1.6
1.8
The measured and theoretical values agree over the full range of wavelengths and source-detector separations used in this study. The fits depicted here determine the values of A and b in the Mie-scattering model. It should be emphasized that only one value of A and one value of b are used for all wavelengths and source-detector separations, and that a single value of µa at each wavelength is sufficient to fit the data at all source-detector separations. λ = 640 nm 3
Data Fit data Best Fit
2.5
Φ
2
Figure 3: Fluence rate as a function of detector position measured at 640 nm in a phantom. The solid line indicates the best fit obtained by varying µa while maintaining the scattering spectrum detemined by the previous 9-wavelength fit. Only the data points indicated by the asterisk symbols (*) were included in the fit.
1.5
1
-2
136
0
2
4
6 8 z (mm)
10
12
14
16
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
Once the values of the scattering parameters are determined, a fit is performed at each wavelength individually to determine the value of µa. The results of a typical fit are shown in figure 3. To verify the accuracy of the global fitting algorithm, we have plotted the measured fluence spectra for one phantom as functions of wavelength for the sourcedetector separations included in the fit in figure 4. The agreement between the data (symbols) and the fit (solid lines) is good, indicating that the scattering parameters determined in the initial fitting step are applicable to the full range of wavelengths included in the fit. 1.7 mm 3.7 mm 5.7 mm 7.7 mm 9.7 mm 11.7 mm 13.7 mm 15.7 mm
3
2.5
Figure 4: Fluence rate as a function of wavelength for each of the source-detector separations included in the fit. The solid lines indicate the best fit resulting from the twostep fitting process, in which the scattering spectrum is confined to be of the form Aλ-b and µa is determined individually for each wavelength.
Φ/S
2
1.5
1
0.5 450
500
550
600
650 700 Wavelength (nm)
750
800
850
The absorption spectra obtained from a series of phantoms are shown in figure 5. The concentration of MLu used in each phantom is indicated in the legend. As expected, the absorption spectra are smooth and continuous and exhibit the characteristic 735-nm MLu absorption peak. The increase in absorption at shorter wavelengths is due to the presence of ink in the phantom. These absorption spectra are fit using the SVD fitting algorithm described above. Figure 6 depicts a typical SVD fit. The total fit matches the data accurately. The fit includes a Fourier series to account for the presence of unknown absorbers (not shown). In the cases presented here, the Fourier series contribution is lower than that of any of the basis spectra, indicating that our choice of basis spectra accurately reflects the true makeup of the phantoms and human prostate tissues.
0.5
0.4 0.35 0.3 -1 µa (cm )
Figure 5: Absorption spectra of a sereies of Liposyn phantoms containing black ink and MLu. The MLu concentration is indicated in the legend, and increases from 0 (bottom curve) to 10 mg kg-1 (top curve)
0 1 1.5 2 3 4 5 7 10
0.45
0.25 0.2 0.15 0.1 0.05 0 650
700
750 Wavelength (nm)
800
850
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
137
Figure 6: SVD analysis of a typical phantom absorption spectrum. For clarity, only every 4th data point is shown. The solid line indicates the total fit, which includes a Fourier series designed to account for the presence of unknown absorbers.
Data Lutex Water Ink Total
0.09 0.08 0.07
-1 µa (cm )
0.06 0.05 0.04 0.03 0.02 0.01 0 -0.01 650
700
750 Wavelength (nm)
800
850
The amplitude of the MLu peak, as quantified by the SVD fitting algorithm, is shown in figure 7. As expected, the concentration of MLu determined by the fitting algorithm matches the known concentration in the phantom. The scattering parameters A and b are recovered independent of the concentration of MLu, and are in agreement with the known scattering properties of the sample (data not shown). In all but one case, the best fit concentration and the true concentration differ by less than 15% over the range from 0 to 10 mg kg-1. These results demonstrate that our methods are capable of quantifying the concentration of MLu in absorbing and scattering media in the presence of other absorbers over the range of MLu concentrations we expect to see in vivo.
Figure 7: Concentration of MLu determined from absorption spectra using the SVD fitting algorithm plotted as a function of the known concentration of MLu in two sets of phantoms. In both cases, the true concentration is recovered within 15% accuracy over the range from 0 to 10 mg kg-1.
10
Fit Lutex Concentration (mg kg-1)
9 8 7 6
No Ink With Ink Perfect Fit
5 4 3 2 1 0
0
1
2
3
4
5
6
7
8
9
10
True Lutex Concentration (mg kg-1)
3.2
Human prostate results:
Using the methods demonstrated above, we have determined the absorption and scattering spectra of MLu-sensitized prostate tissue in patients before and after irradiation with 732-nm light. A typical set of spectra are shown in figure 8(a) and (b). In each figure, the measured absorption spectrum is indicated by the data points. The components of the spectrum determined by the SVD algorithm are indicated by the dotted and dashed lines and their sum by the solid line. Prior to treatment (figure 8a), the MLu concentration was 7.5 mg kg-1, compared to 2.6 mg kg-1 after treatment. The changes in MLu concentration for the two patients for which data were acquired are summarized in table 1.
138
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
0.8
0.9
0.7
0.6
0.6
0.5 Lutex
0.5 -1 µa (cm )
-1 µa (cm )
b
0.8
a
0.7
0.4 Hb
0.3
0.4 0.3
HbO2 0.2
Hb
0.2 Lutex
0.1
0.1 HbO2
0
0 Fourier
-0.1 650
700
Fourier
750 800 Wavelength (nm)
850
900
-0.1 650
700
750 800 Wavelength (nm)
850
900
Figure 8: Absorption spectra obtained in a human prostate (a) before and (b) after PDT treatment. The contributions of MLu and oxy- and deoxyhemoglobin are indicated by the labeled curves. The fits also included the absorption of water (not shown). The Fourier series accounts for unknown absorbers and for inaccuracies in the absorption basis spectra.
Patient Number
MLu dose (mg kg-1)
Total Fluence (J cm-2 )
1 2 (RLQ) 2 (LUQ)
2 2 2
100 25 25
Drug-light interval (hours) 6 3 3
MLu concentration before PDT (mg kg-1) 2.2 7.5 8.4
MLu concentration after PDT (mg kg-1) 1.0* 2.6 2.4
Table 1: Concentrations of MLu measured in the prostate by absorption spectroscopy before and after PDT treatment. For patient 2, data were acquired in both the right lower quadrant (RLQ) and left upper quadrant (LUQ) of the prostate. The measurement indicated by the * was scaled to account for changes in source intensity by assuming that the absorption at 650 nm did not change.
For one patient, designated ‘2’ in table 1, we have obtained independent absorption spectra in two quadrants of the prostate, designated ‘RLQ’ and ‘LUQ’ for the right lower and left upper quadrants, respectively. Comparing these two quadrants gives a measure of the heterogeneity of response to PDT. The concentrations of MLu in the two quadrants agree to within 12%. Both of measurements in patient 2 detected approximately 4 times more MLu than was found in patient 1, likely a result of the difference in drug-light interval. Of particular interest from the point of view of PDT dosimetry is the PDT-induced destruction of the sesitizer, known as photobleaching. Self-sensitized photobleaching has been observed for a variety of sensitizers in a number of model systems.17, 19-22 It has been suggested that photobleaching can be used as a direct measure of photodynamic damage, allowing it to be used for PDT dosimetry.4 In both quadrants of patient 2’s prostate, 65% or more of the initial MLu is eliminated by photobleaching., compared with 44% for patient 1. In addition to quantifying the concentration of sensitizer, our measurements also give useful information about the hemoglobin status of the tissue being measured. The hemoglobin concentrations and saturations measured in these two patients before and after treatement are presented in table 2. Patient 1 exhibited an absorption spectrum dominated by deoxyhemoglobin both before and after treatment, indicating that the prostate was initially hypoxic in this case. This observation, coupled with the low concentration of hemoglobin, indicates that this prostate was poorly vascularized. It is possible that the lower concentration of sensitizer detected in this patient results partially from vascular deficiency. In patient 2, the hemoglobin satuaration was initially above 50% in both quadrants measured. After treatment, the saturation in both quadrants has decreased by approximately 20%. This decrease in oxygenation may result from photochemical oxygen consumption, an effect that has been measured in vivo for certain sensitizers. The fact that it
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
139
persists after significant photobleaching has occurred, and does not recover with the cessation of light delivery, however, indicates that the PDT-induced oxygen change we observe is likely the result of photodynamic damage to the vasculature, resulting in a decrease in the supply of oxygenated blood. Vascular damage has been reported for a number of sensitizers, and several treatment protocols have been designed specifically to target vessels.23 This interpretation is confirmed by the decrease in total hemoglobin concentration induced in both quadrants by PDT.
Hemoglobin concentration (µM) before PDT after PDT 29 26 150 128 280 89
Patient Number 1 2 (RLQ) 2 (LUQ)
µeff at 732 nm (cm-1)
Hemoglobin saturation before PDT 0 0.56 0.76
after PDT 0 0.31 0.52
before PDT 2.3 2.6 5.2
after PDT 3.3 4.8 6.1
Table 2: Comparison of the changes in hemoglobin concentration and oxygenation and effective attenuation coefficient among patients and between quadrants for Patent 2. In the case of Patient 1, the absorption spectra exhibited no discernable contribution from oxyhemoglobin before or after PDT.
It is likely that the differences in photobleaching we observe between patients correspond to a real difference in the photodynamic damage done to the target tissue. This conclusion is counterintuitive given the fact that patient 1 received 4 times more total fluence than patient 2 (see Table 1). However, the initial MLu concentration was 4 times smaller for Patient 1, offsetting the effect of the increased treatment fluence. In addition, photodynamic therapy derives its therapeutic effect from the generation and reaction of singlet oxygen in the target tissue. This requires that sufficient oxygen be present in the tissue throughout the course of treatment. In the case of Patient 1, the prostate was hypoxic before and after PDT, and presumably during treatment as well. We would therefore expect this prostate to receive a lower dose of singlet oxygen, and to exhibit less sensitizer photobleaching, than the well-oxygenated prostate of Patient 2, in agreement with the observations reported in Table 1.
3
Figure 9: Effective attenuation spectra corresponding to the absorption spectra shown in figure 8a, obtained prior to PDT.
2.8
-1 µeff (cm )
2.6
2.4
2.2
2
1.8
1.6 650
700
750 800 Wavelength (nm)
850
900
Finally, we consider the effective attenuation coefficient µeff, given by
3 µ a ( µ a + µ s′ ) . The rate of the exponential
decrease in fluence rate with distance from the source is determined by µeff. , making it an essential parameter in the design and monitoring of PDT treatment. Figure 9 illustrates the variation µeff with wavelength corresponding to the spectrum shown in figure 8a. This figure illustrates the advantage of using a sensitizer with a long-wavelength absorption peak; the value of µeff varies by nearly a factor of two between 650 and 800 nm, and increases even more
140
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
dramatically at shorter wavelengths. In table 2, we list the values of µeff at the treatment wavelength of 732 nm measured before and after PDT for each of the patients studied. The initial µeff differs by a factor of two between the two patients and even between two quadrants in one patient. In all three cases, PDT results in an increase in µeff, resulting primarily from an increase in scattering. It is possible that increased scattering indicates PDT-induces damage to scattering structures in the tissue or a biological response to photodynamic damage. The differences between quadrants and the changes induced in a single quadrant during PDT underscore the importance of real-time dosimetry.
4
Conclusions
We have demonstrated the accuracy and robustness of a technique for the measurement and analysis of absorption spectra in turbid media. Our method makes no assumptions concerning the absorption spectrum of the sample, allowing it to be sensitive to unknown absorbers and to changes in the spectra of known absorbers. The assumption of a known form for the reduced scattering spectrum makes the algorithm robust enough for clinical use, and eliminates the problem of crosstalk between absorption and scattering coefficients, which can reduce the accuracy of methods that determine µa and µs′ independently for each wavelength. The algorithm can accurately recover the absolute MLu concentration under optical conditions similar to those encountered in vivo. We have applied this method to the measurement and analysis of the absorption spectra of human prostate tissue before and after PDT, and have observed direct evidence of PDT-induced photobleaching of MLu in vivo. The measurement of multiple sites within a single prostate allows us to measure the heterogeneity of response to PDT. While the heterogeneity in photobleaching we observe is small, it may be significant enough to change the outcome of treatment if photobleaching is used for PDT dosimetry. This motivates the continued use of detectors in multiple quadrants rather than the generalization of a single measurement to the whole organ. Even greater variations were observed between patients, indicating that no single set of assumptions about optical properties and sensitizer concentrations can be generalized to all patients.
Acknowledgements This work was funded by Department of Defense grant DAMD17-03-1-0132 and by National Institutes of Health R21 grant CA88064-01.
References 1.
2.
3. 4. 5.
M. Solonenko, R. Cheung, T. M. Busch, A. Kachur, G. M. Griffin, T. Vulcan, T. C. Zhu, H.-W. Wang, S. M. Hahn, and A. G. Yodh, "In vivo reflectance measurements of optical properties, blood oxygenation and motexafin lutetium uptake in canine large bowels, kidneys and prostates," Phys. Med. Biol. 47, 857-873 (2002). M. S. Patterson, B. C. Wilson, J. W. Feather, D. M. Burns, and W. Pushka, "The measurement of dihematoporphyrin ether concentration in tissue by reflectance spectrophotometry," Photochem. Photobiol. 46, 337-343 (1987). E. L. Hull and T. H. Foster, "Noninvasive near-infrared hemoglobin spectroscopy for in vivo monitoring of tumor oxygenation and response to oxygen modifiers," Proc. SPIE 2979, 355-364 (1997). B. C. Wilson, M. S. Patterson, and L. Lilge, "Implicit and explicit dosimetry in photodynamic therapy: A new paradigm," Lasers Med. Sci. 12, 182-199 (1997). S. W. Young, K. Woodburn, M. Wright, T. D. Mody, W. C. Dow, F. Qing, J. Sessler, and R. A. Miller, "Lutetium texaphyrin (PCI-0123): a near-infrared, water-soluble photosensitizer," Photochem. Photobiol. 63, 892-897 (1996).
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
141
6. 7.
8.
9. 10. 11. 12. 13. 14. 15. 16. 17.
18.
19. 20. 21.
22.
23.
142
T. Mody, L. Fu, and J. Sessler, "Texaphyrins: synthesis and development of a novel class of therapeutic agents," Prog. Inorg. Chem. 49, 551-598 (2001). T. J. Farrell, M. S. Patterson, and 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, 879888 (1992). M. G. Nichols, E. L. Hull, and T. H. Foster, "Design and testing of a white-light, steady-state diffuse reflectance spectrometer for determination of optical properties of highly scattering systems," Appl. Opt. 36, 93-104 (1997). D. A. Boas, H. Lui, M. A. O'Leary, B. Chance, and A. G. Yodh, "Photon Migration within the P3 approximation," Proc. SPIE 2389, 240-247 (1995). W. M. Star, "Comparing the P3-approximation with diffusion theory and with Monte Carlo calculations of light propagation in a slab geometry," SPIE Institute Series IS5, 46-54 (1989). E. L. Hull and T. H. Foster, "Steady-state reflectance spectroscopy in the P3 approximation," JOSA A 18, 584599 (2001). J. R. Mourant, T. Fusilier, J. Boyer, T. M. Johnson, and I. J. Bigio, "Predictions and measurements of scattering and absorption over broad wavelength ranges in tissue phantoms," Appl. Opt. 36, 949-957 (1997). S. L. Jacques, "Origins of optical properties in the UVA, visible, and NIR regions," OSA Trends in Optics and Photonics on Advances in Optical Imaging and Photon Migration 2, 364-371 (1996). H. J. van Staveren, C. J. M. Moes, J. van Marle, S. A. Prahl, and M. J. C. van Gemert, "Light scattering in Intralipid-10 in the wavelength range of 400-1100 nm," Appl. Opt. 30, 4507-4515 (1991). W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, Numerical Recipes in C: The Art of Scientific Computing, Second ed. (Cambridge University Press, New York, 1992). E. L. Hull, M. G. Nichols, and T. H. Foster, "Quantitative broadband near-infrared spectroscopy of tissuesimulating phantoms containing erythrocytes," Phys. Med. Biol. 43, 3381-3404 (1998). J. C. Finlay, D. L. Conover, E. L. Hull, and T. H. Foster, "Porphyrin bleaching and PDT-induced spectral changes are irradiance dependent in ALA-sensitized normal rat skin in vivo," Photochem. Photobiol. 73, 54-63 (2001). S. Wray, M. Cope, D. T. Delpy, J. S. Wyatt, and E. O. R. Reynolds, "Characterization of the near-infrared absorption spectra of cytochrome aa3 and hemoglobin for the noninvasive monitoring of cerebral oxygenation," Biochim. Biophys. Acta 933, 184-192 (1988). W. R. Potter, T. S. Mang, and T. J. Dougherty, "The theory of photodynamic therapy dosimetry: Consequences of photo-destruction of sensitizer," Photochem. Photobiol. 46, 97-101 (1987). J. C. Finlay, S. Mitra, and T. H. Foster, "In vivo mTHPC photobleaching in normal rat skin exhibits unique irradiance-dependent features," Photochem. Photobiol. 75, 282-288 (2002). D. J. Robinson, H. S. de Bruijn, N. van der Veen, M. R. Stringer, S. B. Brown, and W. M. Star, "Fluorescence photobleaching of ALA-induced protoporphyrin IX during photodynamic therapy of normal hairless mouse skin: The effect of light dose and irradiance and the resulting biological effect," Photochem. Photobiol. 67, 140-149 (1998). M. Forrer, T. Glanzmann, D. Braichotte, G. Wagni\eres, H. van den Bergh, J. F. Savary, and P. Monnier, "In vivo measurement of fluorescence bleaching of meso-tetra hydroxy phenyl chlorin (mTHPC) in the esophagus and the oral cavity," Proc. SPIE 2627, 33-39 (1995). B. Krammer, "Vascular effects of photodynamic therapy," Anitcancer Res. 21, 4271-4278 (2001).
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
Keywords: photodynamic therapy, motexafin lutetium, in-vivo light dosimetry, tissue optical properties, diffusion theory, diffuse reflectance.
*
132
[email protected]; phone (215) 615-3795; fax (215) 349-5978; http://www.xrt.upenn.edu/radiation_physics
Optical Methods for Tumor Treatment and Detection: Mechanisms and Techniques in Photodynamic Therapy XIII, edited by David Kessel, Proc. of SPIE Vol. 5315 (SPIE, Bellingham, WA, 2004) · 1605-7422/04/$15 · doi: 10.1117/12.528968
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
1
Introduction
The accurate in vivo determination of the optical properties of human tissues is an ongoing challenge in the optimization of photodynamic therapy (PDT). The distribution of treatment light within the target tissue is determined by the absorption and scattering properties of the tissue itself. Accurate characterization of the delivered light dose depends critically on these properties, especially for parts of the tissue remote from the light source. Measurement of tissue optical properties also has the potential to allow characterization of the distribution of sensitizer concentration and the oxygenation status of hemoglobin.1-3 In combination, these parameters have the potential to allow complete characterization of the factors that affect the efficacy of photodynamic treatment, namely the local light fluence rate and the local concentrations of sensitizer and oxygen. In addition, it is possible that the photodynamic degradation of the sensitizer as determined by the decrease in the sensitizer component of the absorption spectrum can provide an implicit measure of photodynamic dose.4 In this paper, we investigate the absorption and scattering spectra of human prostate tissue sensitized with the secondgeneration photosensitizer motexafin lutetium (MLu).5, 6 The absorption peak of MLu typically used for treatment is centered at 732 nm, a wavelength where the inherent absorption of tissue is relatively weak. While the shift to longer wavelength absorption in the second-generation sensitizers was primarily intended to increase the penetration of excitation light, it has the additional advantage from the point of view of absorption spectroscopy that the absorption arising from the sensitizer does not compete with a dominant background of hemoglobin absorption. On the other hand, the absorption band of MLu is narrow enough that it does not obscure the absorption features of hemoglobin. Under these circumstances, the sensitizer and oxy- and deoxyhemoglobin can be characterized independently on the basis of their absorption spectra. To obtain the optical properties of tissue, we use an analytic model of light propagation in turbid media. Several researchers have developed models compatible with CW measurements based on the diffusion approximation to the Boltzmann radiative transport equation.1, 7, 8 In this paper, we make use of the higher-order P3 approximation, which is valid over a wider range of optical properties and source-detector separations.9, 10 This approximation has been shown effective for the independent determination of the absorption and scattering coefficients of turbid media.11 We combine an infinite-medium P3 expression with a nonlinear fitting algorithm that assumes a reduced scattering spectrum of the form µ’s = Aλ-b. Constraining the reduced scattering spectrum to take this form increases the robustness of the algorithm and reduces the chances of crosstalk between the absorption and scattering coefficients.
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
133
2 2.1
Methods
Data acquisition
We have obtained diffuse fluence spectra using the instrument shown in figure 1. White light is provided by a tungsten lamp. Light from this source is coupled into a 200 micron fiber terminated in a spherical isotropic scattering tip. An identical isotropic scattering fiber is used to detect diffuse light. The measured signal is therefore proportional to the fluence rate in the medium Light collected by this fiber is imaged through a grating spectrograph onto a nitrogencooled CCD camera. Spectrograph & CCD
PC 1
Tungsten Lamp
PC 2
Figure 1: Experimental setup for spectroscopic measurements in turbid media. Computer 1 controls the motors, while computer 2 records the spectra acquired by the spectrograph and CCD.
Motorized positioners
Source
Detector
Scattering medium
Both fibers are inserted into the sample through clear catheters. The catheters are parallel and separated by 3 to 8 mm. Each fiber is connected to a computer-controlled step motor, allowing the source and detector fibers to be moved along their respective catheters independently. The characterization of optical properties requires a measurement of the fluence rate over a range of source-detector separations. This is accomplished by fixing the source at one location and moving the detector relative to it. Diffuse reflectance spectra are obtained at 2 mm intervals starting 4 mm on one side of the source and continuing to 16 mm beyond it on the other side. The origin of the detector position, i.e. the position which gives the shortest source-detector separation, is determined by a prior scan with a photodioide and is verified by checking the symmetry of the peak. The movement of the motors is controlled by software developed in our laboratory, which also triggers the collection of data by the CCD system. The minimum source-detector separation achievable with this system is determined by the separation between catheters, in this case 3 to 8 mm. As a result, no signal can be detected in samples whose absorption coefficient is greater than approximately 2 cm-1. In human tissue, this limit is exceeded at visible wavelengths sue to the absorption of light by hemoglobin. We therefore consider only data acquired at wavelengths longer than 650 nm in the analysis presented below. 2.2
Spectral analysis
The diffuse fluence spectra captured by the CCD are first divided by the spectra of the same source obtained in an integrating sphere. We have characterized the relationship between the input power and the resulting fluence rate inside
134
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
this particular integrating sphere. This division therefore gives a calibrated measure of the relative diffuse fluence rate of the sample, i.e .the ratio of the fluence rate to the power of the point source. We model the fluence rate in the sample using the P3 expression for fluence rate in an infinite medium developed by Hull and Foster11 based on the previous work of Star10 and Boas et al.9 The determination of the optical properties of the sample proceeds in two steps. First, we select a subset of the wavelengths at which the diffuse reflectance is measured. The data at these wavelengths are fit using a global non-linear fitting algorithm with a model that assumes a reduced scattering coefficient of the form µ’s = Aλ-b. This form of the reduced scattering spectrum, which is based on an approximation to the Mie scattering theory, has been widely used to model tissue scattering,1, 12, 13 and also approximates the reduced scattering spectrum of Lyposyn.14 The fitting algorithm’s free parameters are A, b, and the absorption coefficient at each wavelength. Nine wavelengths are usually sufficient to determine A and b accurately. In the second stage of fitting, the fluence rate profile at each wavelength is fit individually. The value of µ’s is fixed at the value determined in the first step, and only µa is varied. This fitting is repeated for each wavelength. The result of this fitting process is a complete absorption spectrum of the sample and two parameters A and b that characterize the scattering spectrum of the sample. Once the absorption spectrum is determined, the concentrations of absorbers with known spectra can be determined by linear fitting. To accomplish this fitting, we have implemented the singular value decomposition (SVD) algorithm of Press et al.15 implemented in Matlab (the MathWorks, Natick, MA). Following the method of Hull et al.,16 we have included a Fourier synthesis in the fitting to account for the presence of unknown absorbers. Our Fourier series is weighted exponentially to reduce high-frequency noise.17 The basis spectra used for MLu and ink are based on measurements in clear solutions. For in vivo fitting, we have adopted hemoglobin basis spectra from the work Wray et al.18 2.3
Phantom preparation
Tissue simulating phantoms were prepared from Liposyn and water. Liposyn has been shown to have a scattering spectrum similar to that of tissue.12, 14 The total lipid concentration in these phantoms was 0.5%, yielding a reduced scattering coefficient of approximately 6 cm-1 at 730 nm. MLu was added to the phantom in concentrations similar to that expected in tissue. Each phantom has a total volume of 2L, large enough to be considered infinite for the purposes of our analysis. The phantoms were housed in a large container whose sides were blackened to prevent the entrance of light from the outside and to reduce reflections. We present the results of two sets of phantoms, one with MLu as the only absorber, and another in which we also added black ink to simulate the presence of other absorbers in tissue. Like hemoglobin, the black ink used in these experiments exhibits relatively low absorption in the wavelengths near the MLu absorption peak, but dominates the absorption at shorter wavelengths. It therefore provides a reasonable test of the ability of our measurement and analysis system to determine the concentration of MLu spectroscopically. 2.4
In vivo data acquisition
The patients involved in this study were part of an ongoing clinical trial of PDT for the treatment of recurrent prostate cancer after radiation therapy. Prior to treatment, each patient’s prostate was sized by transrectal ultrasound. The prostate was divided into quadrants, and the placement of light sources to illuminate each quadrant was determined. The treatment was planned such that each quadrant received two or four treatment catheters and one measurement catheter. Each patient was given MLu intravenously at a dose and time predetermined by the dose-escalation schedule of the clinical protocol. The patient was then anesthetized, and the plastic catheters were inserted according to the predeveloped plan. Interstitial irradiation was performed with 732-nm light at a maximum fluence rate of 150 mW cm-2. Prior to and immediately following treatment, spectroscopic measurements where made in each quadrant. The white light source fiber was inserted into one of the treatment catheters, and the detector fiber was inserted into the measurement catheter. During treatment, cylindrical light sources were placed in the treatment catheters, and a single isotropic detector used for in vivo light dosimetry was inserted into each detection catheter. After treatment, these fibers were removed, and the spectroscopic measurements were repeated. The spectroscopic measurements therefore did not require the placement of additional catheters but rather made use of those catheters necessary for treatment.
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
135
3 3.1
Results
Verification in phantoms
In figure 2, we plot the fluence rate measured in a typical phantom as a function of source-detector separation at the nine wavelengths used in the initial fitting step to determine µ’s. The symbols indicate the measured data, corrected for instrument response. The lines indicate the best fit provided by the P3 model to the data.
3 443.6 489.4 534.8 579.7 624.2 668.3 712.0 755.3 798.1
2.5
Φ
2
nm nm nm nm nm nm nm nm nm
Figure 2: Fluence rate as a function of source-detector separation for the nine wavelengths used to determine the scattering parameters A and b. The solid lines indicate the best fits obtained using the P3 theory.
1.5
1
0.5
0 0.6
0.8
1
1.2 r (cm)
1.4
1.6
1.8
The measured and theoretical values agree over the full range of wavelengths and source-detector separations used in this study. The fits depicted here determine the values of A and b in the Mie-scattering model. It should be emphasized that only one value of A and one value of b are used for all wavelengths and source-detector separations, and that a single value of µa at each wavelength is sufficient to fit the data at all source-detector separations. λ = 640 nm 3
Data Fit data Best Fit
2.5
Φ
2
Figure 3: Fluence rate as a function of detector position measured at 640 nm in a phantom. The solid line indicates the best fit obtained by varying µa while maintaining the scattering spectrum detemined by the previous 9-wavelength fit. Only the data points indicated by the asterisk symbols (*) were included in the fit.
1.5
1
-2
136
0
2
4
6 8 z (mm)
10
12
14
16
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
Once the values of the scattering parameters are determined, a fit is performed at each wavelength individually to determine the value of µa. The results of a typical fit are shown in figure 3. To verify the accuracy of the global fitting algorithm, we have plotted the measured fluence spectra for one phantom as functions of wavelength for the sourcedetector separations included in the fit in figure 4. The agreement between the data (symbols) and the fit (solid lines) is good, indicating that the scattering parameters determined in the initial fitting step are applicable to the full range of wavelengths included in the fit. 1.7 mm 3.7 mm 5.7 mm 7.7 mm 9.7 mm 11.7 mm 13.7 mm 15.7 mm
3
2.5
Figure 4: Fluence rate as a function of wavelength for each of the source-detector separations included in the fit. The solid lines indicate the best fit resulting from the twostep fitting process, in which the scattering spectrum is confined to be of the form Aλ-b and µa is determined individually for each wavelength.
Φ/S
2
1.5
1
0.5 450
500
550
600
650 700 Wavelength (nm)
750
800
850
The absorption spectra obtained from a series of phantoms are shown in figure 5. The concentration of MLu used in each phantom is indicated in the legend. As expected, the absorption spectra are smooth and continuous and exhibit the characteristic 735-nm MLu absorption peak. The increase in absorption at shorter wavelengths is due to the presence of ink in the phantom. These absorption spectra are fit using the SVD fitting algorithm described above. Figure 6 depicts a typical SVD fit. The total fit matches the data accurately. The fit includes a Fourier series to account for the presence of unknown absorbers (not shown). In the cases presented here, the Fourier series contribution is lower than that of any of the basis spectra, indicating that our choice of basis spectra accurately reflects the true makeup of the phantoms and human prostate tissues.
0.5
0.4 0.35 0.3 -1 µa (cm )
Figure 5: Absorption spectra of a sereies of Liposyn phantoms containing black ink and MLu. The MLu concentration is indicated in the legend, and increases from 0 (bottom curve) to 10 mg kg-1 (top curve)
0 1 1.5 2 3 4 5 7 10
0.45
0.25 0.2 0.15 0.1 0.05 0 650
700
750 Wavelength (nm)
800
850
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
137
Figure 6: SVD analysis of a typical phantom absorption spectrum. For clarity, only every 4th data point is shown. The solid line indicates the total fit, which includes a Fourier series designed to account for the presence of unknown absorbers.
Data Lutex Water Ink Total
0.09 0.08 0.07
-1 µa (cm )
0.06 0.05 0.04 0.03 0.02 0.01 0 -0.01 650
700
750 Wavelength (nm)
800
850
The amplitude of the MLu peak, as quantified by the SVD fitting algorithm, is shown in figure 7. As expected, the concentration of MLu determined by the fitting algorithm matches the known concentration in the phantom. The scattering parameters A and b are recovered independent of the concentration of MLu, and are in agreement with the known scattering properties of the sample (data not shown). In all but one case, the best fit concentration and the true concentration differ by less than 15% over the range from 0 to 10 mg kg-1. These results demonstrate that our methods are capable of quantifying the concentration of MLu in absorbing and scattering media in the presence of other absorbers over the range of MLu concentrations we expect to see in vivo.
Figure 7: Concentration of MLu determined from absorption spectra using the SVD fitting algorithm plotted as a function of the known concentration of MLu in two sets of phantoms. In both cases, the true concentration is recovered within 15% accuracy over the range from 0 to 10 mg kg-1.
10
Fit Lutex Concentration (mg kg-1)
9 8 7 6
No Ink With Ink Perfect Fit
5 4 3 2 1 0
0
1
2
3
4
5
6
7
8
9
10
True Lutex Concentration (mg kg-1)
3.2
Human prostate results:
Using the methods demonstrated above, we have determined the absorption and scattering spectra of MLu-sensitized prostate tissue in patients before and after irradiation with 732-nm light. A typical set of spectra are shown in figure 8(a) and (b). In each figure, the measured absorption spectrum is indicated by the data points. The components of the spectrum determined by the SVD algorithm are indicated by the dotted and dashed lines and their sum by the solid line. Prior to treatment (figure 8a), the MLu concentration was 7.5 mg kg-1, compared to 2.6 mg kg-1 after treatment. The changes in MLu concentration for the two patients for which data were acquired are summarized in table 1.
138
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
0.8
0.9
0.7
0.6
0.6
0.5 Lutex
0.5 -1 µa (cm )
-1 µa (cm )
b
0.8
a
0.7
0.4 Hb
0.3
0.4 0.3
HbO2 0.2
Hb
0.2 Lutex
0.1
0.1 HbO2
0
0 Fourier
-0.1 650
700
Fourier
750 800 Wavelength (nm)
850
900
-0.1 650
700
750 800 Wavelength (nm)
850
900
Figure 8: Absorption spectra obtained in a human prostate (a) before and (b) after PDT treatment. The contributions of MLu and oxy- and deoxyhemoglobin are indicated by the labeled curves. The fits also included the absorption of water (not shown). The Fourier series accounts for unknown absorbers and for inaccuracies in the absorption basis spectra.
Patient Number
MLu dose (mg kg-1)
Total Fluence (J cm-2 )
1 2 (RLQ) 2 (LUQ)
2 2 2
100 25 25
Drug-light interval (hours) 6 3 3
MLu concentration before PDT (mg kg-1) 2.2 7.5 8.4
MLu concentration after PDT (mg kg-1) 1.0* 2.6 2.4
Table 1: Concentrations of MLu measured in the prostate by absorption spectroscopy before and after PDT treatment. For patient 2, data were acquired in both the right lower quadrant (RLQ) and left upper quadrant (LUQ) of the prostate. The measurement indicated by the * was scaled to account for changes in source intensity by assuming that the absorption at 650 nm did not change.
For one patient, designated ‘2’ in table 1, we have obtained independent absorption spectra in two quadrants of the prostate, designated ‘RLQ’ and ‘LUQ’ for the right lower and left upper quadrants, respectively. Comparing these two quadrants gives a measure of the heterogeneity of response to PDT. The concentrations of MLu in the two quadrants agree to within 12%. Both of measurements in patient 2 detected approximately 4 times more MLu than was found in patient 1, likely a result of the difference in drug-light interval. Of particular interest from the point of view of PDT dosimetry is the PDT-induced destruction of the sesitizer, known as photobleaching. Self-sensitized photobleaching has been observed for a variety of sensitizers in a number of model systems.17, 19-22 It has been suggested that photobleaching can be used as a direct measure of photodynamic damage, allowing it to be used for PDT dosimetry.4 In both quadrants of patient 2’s prostate, 65% or more of the initial MLu is eliminated by photobleaching., compared with 44% for patient 1. In addition to quantifying the concentration of sensitizer, our measurements also give useful information about the hemoglobin status of the tissue being measured. The hemoglobin concentrations and saturations measured in these two patients before and after treatement are presented in table 2. Patient 1 exhibited an absorption spectrum dominated by deoxyhemoglobin both before and after treatment, indicating that the prostate was initially hypoxic in this case. This observation, coupled with the low concentration of hemoglobin, indicates that this prostate was poorly vascularized. It is possible that the lower concentration of sensitizer detected in this patient results partially from vascular deficiency. In patient 2, the hemoglobin satuaration was initially above 50% in both quadrants measured. After treatment, the saturation in both quadrants has decreased by approximately 20%. This decrease in oxygenation may result from photochemical oxygen consumption, an effect that has been measured in vivo for certain sensitizers. The fact that it
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
139
persists after significant photobleaching has occurred, and does not recover with the cessation of light delivery, however, indicates that the PDT-induced oxygen change we observe is likely the result of photodynamic damage to the vasculature, resulting in a decrease in the supply of oxygenated blood. Vascular damage has been reported for a number of sensitizers, and several treatment protocols have been designed specifically to target vessels.23 This interpretation is confirmed by the decrease in total hemoglobin concentration induced in both quadrants by PDT.
Hemoglobin concentration (µM) before PDT after PDT 29 26 150 128 280 89
Patient Number 1 2 (RLQ) 2 (LUQ)
µeff at 732 nm (cm-1)
Hemoglobin saturation before PDT 0 0.56 0.76
after PDT 0 0.31 0.52
before PDT 2.3 2.6 5.2
after PDT 3.3 4.8 6.1
Table 2: Comparison of the changes in hemoglobin concentration and oxygenation and effective attenuation coefficient among patients and between quadrants for Patent 2. In the case of Patient 1, the absorption spectra exhibited no discernable contribution from oxyhemoglobin before or after PDT.
It is likely that the differences in photobleaching we observe between patients correspond to a real difference in the photodynamic damage done to the target tissue. This conclusion is counterintuitive given the fact that patient 1 received 4 times more total fluence than patient 2 (see Table 1). However, the initial MLu concentration was 4 times smaller for Patient 1, offsetting the effect of the increased treatment fluence. In addition, photodynamic therapy derives its therapeutic effect from the generation and reaction of singlet oxygen in the target tissue. This requires that sufficient oxygen be present in the tissue throughout the course of treatment. In the case of Patient 1, the prostate was hypoxic before and after PDT, and presumably during treatment as well. We would therefore expect this prostate to receive a lower dose of singlet oxygen, and to exhibit less sensitizer photobleaching, than the well-oxygenated prostate of Patient 2, in agreement with the observations reported in Table 1.
3
Figure 9: Effective attenuation spectra corresponding to the absorption spectra shown in figure 8a, obtained prior to PDT.
2.8
-1 µeff (cm )
2.6
2.4
2.2
2
1.8
1.6 650
700
750 800 Wavelength (nm)
850
900
Finally, we consider the effective attenuation coefficient µeff, given by
3 µ a ( µ a + µ s′ ) . The rate of the exponential
decrease in fluence rate with distance from the source is determined by µeff. , making it an essential parameter in the design and monitoring of PDT treatment. Figure 9 illustrates the variation µeff with wavelength corresponding to the spectrum shown in figure 8a. This figure illustrates the advantage of using a sensitizer with a long-wavelength absorption peak; the value of µeff varies by nearly a factor of two between 650 and 800 nm, and increases even more
140
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
dramatically at shorter wavelengths. In table 2, we list the values of µeff at the treatment wavelength of 732 nm measured before and after PDT for each of the patients studied. The initial µeff differs by a factor of two between the two patients and even between two quadrants in one patient. In all three cases, PDT results in an increase in µeff, resulting primarily from an increase in scattering. It is possible that increased scattering indicates PDT-induces damage to scattering structures in the tissue or a biological response to photodynamic damage. The differences between quadrants and the changes induced in a single quadrant during PDT underscore the importance of real-time dosimetry.
4
Conclusions
We have demonstrated the accuracy and robustness of a technique for the measurement and analysis of absorption spectra in turbid media. Our method makes no assumptions concerning the absorption spectrum of the sample, allowing it to be sensitive to unknown absorbers and to changes in the spectra of known absorbers. The assumption of a known form for the reduced scattering spectrum makes the algorithm robust enough for clinical use, and eliminates the problem of crosstalk between absorption and scattering coefficients, which can reduce the accuracy of methods that determine µa and µs′ independently for each wavelength. The algorithm can accurately recover the absolute MLu concentration under optical conditions similar to those encountered in vivo. We have applied this method to the measurement and analysis of the absorption spectra of human prostate tissue before and after PDT, and have observed direct evidence of PDT-induced photobleaching of MLu in vivo. The measurement of multiple sites within a single prostate allows us to measure the heterogeneity of response to PDT. While the heterogeneity in photobleaching we observe is small, it may be significant enough to change the outcome of treatment if photobleaching is used for PDT dosimetry. This motivates the continued use of detectors in multiple quadrants rather than the generalization of a single measurement to the whole organ. Even greater variations were observed between patients, indicating that no single set of assumptions about optical properties and sensitizer concentrations can be generalized to all patients.
Acknowledgements This work was funded by Department of Defense grant DAMD17-03-1-0132 and by National Institutes of Health R21 grant CA88064-01.
References 1.
2.
3. 4. 5.
M. Solonenko, R. Cheung, T. M. Busch, A. Kachur, G. M. Griffin, T. Vulcan, T. C. Zhu, H.-W. Wang, S. M. Hahn, and A. G. Yodh, "In vivo reflectance measurements of optical properties, blood oxygenation and motexafin lutetium uptake in canine large bowels, kidneys and prostates," Phys. Med. Biol. 47, 857-873 (2002). M. S. Patterson, B. C. Wilson, J. W. Feather, D. M. Burns, and W. Pushka, "The measurement of dihematoporphyrin ether concentration in tissue by reflectance spectrophotometry," Photochem. Photobiol. 46, 337-343 (1987). E. L. Hull and T. H. Foster, "Noninvasive near-infrared hemoglobin spectroscopy for in vivo monitoring of tumor oxygenation and response to oxygen modifiers," Proc. SPIE 2979, 355-364 (1997). B. C. Wilson, M. S. Patterson, and L. Lilge, "Implicit and explicit dosimetry in photodynamic therapy: A new paradigm," Lasers Med. Sci. 12, 182-199 (1997). S. W. Young, K. Woodburn, M. Wright, T. D. Mody, W. C. Dow, F. Qing, J. Sessler, and R. A. Miller, "Lutetium texaphyrin (PCI-0123): a near-infrared, water-soluble photosensitizer," Photochem. Photobiol. 63, 892-897 (1996).
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
141
6. 7.
8.
9. 10. 11. 12. 13. 14. 15. 16. 17.
18.
19. 20. 21.
22.
23.
142
T. Mody, L. Fu, and J. Sessler, "Texaphyrins: synthesis and development of a novel class of therapeutic agents," Prog. Inorg. Chem. 49, 551-598 (2001). T. J. Farrell, M. S. Patterson, and 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, 879888 (1992). M. G. Nichols, E. L. Hull, and T. H. Foster, "Design and testing of a white-light, steady-state diffuse reflectance spectrometer for determination of optical properties of highly scattering systems," Appl. Opt. 36, 93-104 (1997). D. A. Boas, H. Lui, M. A. O'Leary, B. Chance, and A. G. Yodh, "Photon Migration within the P3 approximation," Proc. SPIE 2389, 240-247 (1995). W. M. Star, "Comparing the P3-approximation with diffusion theory and with Monte Carlo calculations of light propagation in a slab geometry," SPIE Institute Series IS5, 46-54 (1989). E. L. Hull and T. H. Foster, "Steady-state reflectance spectroscopy in the P3 approximation," JOSA A 18, 584599 (2001). J. R. Mourant, T. Fusilier, J. Boyer, T. M. Johnson, and I. J. Bigio, "Predictions and measurements of scattering and absorption over broad wavelength ranges in tissue phantoms," Appl. Opt. 36, 949-957 (1997). S. L. Jacques, "Origins of optical properties in the UVA, visible, and NIR regions," OSA Trends in Optics and Photonics on Advances in Optical Imaging and Photon Migration 2, 364-371 (1996). H. J. van Staveren, C. J. M. Moes, J. van Marle, S. A. Prahl, and M. J. C. van Gemert, "Light scattering in Intralipid-10 in the wavelength range of 400-1100 nm," Appl. Opt. 30, 4507-4515 (1991). W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, Numerical Recipes in C: The Art of Scientific Computing, Second ed. (Cambridge University Press, New York, 1992). E. L. Hull, M. G. Nichols, and T. H. Foster, "Quantitative broadband near-infrared spectroscopy of tissuesimulating phantoms containing erythrocytes," Phys. Med. Biol. 43, 3381-3404 (1998). J. C. Finlay, D. L. Conover, E. L. Hull, and T. H. Foster, "Porphyrin bleaching and PDT-induced spectral changes are irradiance dependent in ALA-sensitized normal rat skin in vivo," Photochem. Photobiol. 73, 54-63 (2001). S. Wray, M. Cope, D. T. Delpy, J. S. Wyatt, and E. O. R. Reynolds, "Characterization of the near-infrared absorption spectra of cytochrome aa3 and hemoglobin for the noninvasive monitoring of cerebral oxygenation," Biochim. Biophys. Acta 933, 184-192 (1988). W. R. Potter, T. S. Mang, and T. J. Dougherty, "The theory of photodynamic therapy dosimetry: Consequences of photo-destruction of sensitizer," Photochem. Photobiol. 46, 97-101 (1987). J. C. Finlay, S. Mitra, and T. H. Foster, "In vivo mTHPC photobleaching in normal rat skin exhibits unique irradiance-dependent features," Photochem. Photobiol. 75, 282-288 (2002). D. J. Robinson, H. S. de Bruijn, N. van der Veen, M. R. Stringer, S. B. Brown, and W. M. Star, "Fluorescence photobleaching of ALA-induced protoporphyrin IX during photodynamic therapy of normal hairless mouse skin: The effect of light dose and irradiance and the resulting biological effect," Photochem. Photobiol. 67, 140-149 (1998). M. Forrer, T. Glanzmann, D. Braichotte, G. Wagni\eres, H. van den Bergh, J. F. Savary, and P. Monnier, "In vivo measurement of fluorescence bleaching of meso-tetra hydroxy phenyl chlorin (mTHPC) in the esophagus and the oral cavity," Proc. SPIE 2627, 33-39 (1995). B. Krammer, "Vascular effects of photodynamic therapy," Anitcancer Res. 21, 4271-4278 (2001).
Proc. of SPIE Vol. 5315
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/11/2015 Terms of Use: http://spiedl.org/terms
