J. Photochem. Photobid. B: Biol., 10 (1991) 303313
303
Bacteriochlorophyll-a as photosensitizer for photodynamic treatment of transplantable murine tumors Barbara W. Henderson+, Adam and Thomas J. Dougherty
B. Sumlin, Barbara
L. Owczarczak
Division of Radiation Biology, Roswell Park Cancer Institute, Buffalo, New York (U.S.A.) (Received September 27, 1990; accepted February 12, 1991)
Keyworu!.s. Photodynamic therapy, bacteriochlorophyll-a, a, transplantable mouse tumor photosensitization.
bacteriopheophytin-
Abstract Bacteriochlorophyll-a (bChla), which absorbs light of 780 nm wavelength, was tested for in vivo photodynamic activity in the SMT-F and RIF transplantable mouse tumor systems. High performance liquid chromatography (HPLC) analysis of tissue extracts showed that bChla was rapidly degraded in vivo to bacterlopheophytln-a (bPheoa) and other breakdown products. These were also photodynamically active, and tumor response could be achieved over a wavelength range of 660 to 780 nm, while tumor cure was restricted to wavelengths of 755 @Pheou) to 780 nm. A photosensitizing product absorbing at 660 run was also present in isolated tumor cells. Photodynamic cell kill of tumor cells isolated from tumors after bChlu accumulation in vivo, using 755 or 780 nm light in vitro, was exponential up to 20-40 J cm- ‘. Above this light dose little or no further damage could be achieved, which is an indication of the rapid photobleaching of these sensitizers. In vivo, vascular occlusion occurred readily if light treatment was delivered shortly after sensitizer administration, but was delayed if light treatment was carried out 24 h after injection. Although up to 70% of tumor cells were lethally damaged after completion of in vivo light treatment, concurrent severe vascular destruction seemed necessary for tumor cure. Normal tissue photosensitivity totally subsided within 5 days after sensitizer administration.
1. Introduction Photodynamic therapy (PDT), using the photosensitizer Photofrin II, is perceived to have at least three major limitations [ 1). (i) Photofrin II absorbs light of 630 nm wavelength, which lies well below the wavelength necessary for maximum tissue penetration (700-800 nm). ‘Author to whom correspondence should bc addressed.
loll-1344/91/$3.50
0 1991 -
Elsevier Sequoia, Lausannc
304
(ii) Photofrin II induces prolonged skin photosensitivity, the only major adverse effect known for this sensitizer. (iii) The currently available light delivery systems necessary to produce sufficiently powerful 630 nm light are expensive and difficult to maintain. Bacteriochlorophyll-a (bChla) has several photophysical and chemical characteristics which may help to overcome these limitations. It is a relatively efficient singlet oxygen producer (lOa yields between 0.17 and 0.60, depending on the solvent, have been reported) [2-41 and absorbs strongly (greater than 7 X lo4 M- ’ cm- ’ at 780 nm) [ 41 near the optimum wavelength for tissue penetration [ 51. Furthermore, diode laser technology is being developed to provide inexpensive and simple light delivery at this wavelength.
2. Materials
and methods
2.1. smitixer Bacteriochlorophyll-a was obtained from Porphyrin Products, Logan, UT or Sigma Chemical Co., St. Louis, MO. Several experiments were performed with material kindly supplied by Dr. T. G. Truscott of Paisley College of Technology, Renfrewshire, U.K. All samples used were analyzed by high performance liquid chromatography (HPLC) as described below and were at least 90% pure. Stock solutions were prepared in Hank’s balanced salt solution (HBSS) containing 10% Tween 80, and were further diluted 1:lO with HBSS before in vivo injection. 2.2. Tumor systems The SMT-F (spontaneous mammary carcinoma, fast growing) and RIF (radiation-induced fibrosarcoma) tumors were propagated according to established protocols in DBA/2Ha and C3H/HeJ mice respectively [6, 71. Prior to tumor inoculation, all hair was removed from the right flank of the animal by shaving and depilation. Tumors were used for experimentation when they had reached a surface diameter of 5-6 mm and a thickness of 2-3 mm. 2.3. Extraction and HPLC analysis Tissue samples of tumor and liver were homogenized under subdued light with a 1: 1 mixture of methyl alcohol and dichloromethane and filtered through glass wool; this was then washed several times with the same solvent mixture. Control experiments indicated that bChla was not degraded by this procedure. Two tumors were combined for each determination. HPLC analysis was carried out on all bChla samples as well as tissue extracts using a Spectra Physics (Mountain View, CA) model SP8700 chromatograph and a binary solvent system starting with 90% methanol-water and going to pure methanol over 30 min. A flow rate of 1.5 ml min- ’ was used with a reverse phase Whatman C-8 column. Detection was by absorption at 780 nm using a tunable wavelength detector (Spectroflow 757, also from Spectra Physics).
305
2.4. In vivo photodynamic
therapy
treatment
Animals were given intravenous (i.v.) (tail vein) injections of bChla. At varying times thereafter they were restrained without anesthesia in specially designed holders, and tumors, normal back skin or hind feet were exposed to external light treatment. For light delivery, the following laser systems were used: a 5 or 20 W argon laser system (Spectra Physics) pumping a dye laser using the dyes DCM (Cooper Lasersonics, Inc., Palo Alto, CA) or LDS 751 (Exciton Chemical Co., Dayton, OH) and tuned to the desired wavelength by a birefringent filter; a 1.0 W diode laser (model SDL 820, Spectra Diode Labs, San Jose, CA) tuned to 780 nm by thermoelectric cooling. In each system, the output beam was split using a 50-50 beam splitter (Oriel Optical Corp., Stamford, CT) and coupled to 400 wrn quartz fiberoptic cables (Ensign-Bickford Optics Co., Avon, CT). Microlenses (focal length, 3.5 mm) were fitted to the end of each fiber to facilitate even light distribution throughout the treatment field. The power density of the delivered light was adjusted to 75 mW cmm2 (150 mW ce2 for foot response) at a spot size of 1 cm diameter, and was measured by a radiometer (Yellow Springs Instrument Co. Inc., Yellow Springs, OH). 2.5. Assessment
of tumor
response
and cure
Mice were examined daily for the first 14 days post-treatment and weekly thereafter for a total of 90 days. The disappearance of the tumor bulk was scored as a complete response. If no tumor regrowth had occurred by 90 days, animals were considered to be cured. 2.6. Assessment of vascular photosensitivity Damage to the microvasculature of the tumor-bearing or normal back skin of C3H/HeJ mice was assessed by fluorescence dye exclusion [8]. Briefly, immediately or at varying times after in vivo PDT treatment, animals were given iv. injections (tail vein) of 2.0 mg of fluorescein dye (10 mg ml- ’ of sodium fluorescein (Baker Chemical, Philipsberg, NJ) diluted in HBSS). Immediately thereafter the tumor and/or surrounding skin area was illuminated with UV light (Blak-Ray long-wave UV lamp, Ultra-Violet Products, Inc., San Gabriel, CA), and fluorescence distribution was visualized and recorded. 2.7. Assessment
of foot
response
The photosensitivity of mouse feet was studied in randomly bred Swiss mice. The right or left hind foot of the animals was treated at varying times after i.v. bChla injection with in vivo PDT at a specific wavelength (660, 755, 780 nm) as described above, or with a solar simulator (200 W xenon arc lamp, model 6137, Oriel Corp., Stratford, CT), equipped with an air mass 1 filter, at 90 mW cme2 for 30 min. The foot response was scored daily starting 24 h post-treatment according to the degree of damage (from 0 (no damage) to 2.0 (moderate dry desquamation and swelling of toes)).
2.8. In viva/in vitro cell survival assays RIF tumor cell survival was determined following either of two treatment protocols. (i) Sensitizer was injected into tumor-bearing animals and allowed to accumulate in the tumor for 2-24 h. Tumors were then excised, finely minced in HBSS and/or dispersed by an enzyme procedure [7]; 2 ml aliquots of the single cell suspension were transferred to wells of a 24 well tissue culture plate (Falcon@, Becton Dickinson, Lincoln Park, NJ) and exposed in vitro to light of the desired wavelength, delivered from the 5 W argon dye laser system at a power density of 50 mW cm- 2. Following light treatment, cells were transferred to 60 mm plastic culture plates for clonogenic assay. (ii) Sensitizer was injected into tumor-bearing animals, followed 2 or 24 h later by in viva light treatment at 780 nm as described above. Tumors were excised either immediately or at varying times after light treatment, finely minced, dispersed by an enzyme procedure and plated onto 60 mm plastic culture dishes for clonogenic assay. For clonogenic assay, which has been described in detail elsewhere [ 71, appropriate cell numbers were plated in minimum Eagle’s alpha medium with L-glutamine, ribonucleoside and deoxyribonucleoside, supplemented with 10% fetal bovine serum and antibiotics (all from Grand Island Biological Co., Grand Island, NY). 3. Results and discussion The structures of bChla and its demetallized form bacteriopheophytina (bPheoa) are shown in Fig. 1. The corresponding absorption spectra are presented in Fig. 2 (upper panel) together with the SMT-F tumor response as a function of light treatment wavelength and different treatment conditions (Fig. 2, lower panel). With light treatment carried out 2 h after bChla injection,
.Me
Bacteriochlorophyll-a Fig. 1. Chemical
structures
Eacteriopheaphytin-a of bChla
and bPheoa;
R = phytyl.
307
loo-
& t
80-
P
L-
F
60-
2%
40-
._ 8
20-
$F 00
,c;,, n-x-x--x*
:
:
0
:
X
,
,
,
X
:
:
500 540 580 620-&O -Go 740 780 820 860
Wavelength (nm) Fig. 2. Absorption spectra (in methanol-dichloromethane, 1:l) and SMT-F tumor response as a function of light wavelength. Upper panel: absorbance. Lower panel: tumor response, ten animals per group: X, 7 day tumor response, 5 mg kg- ‘, 270 J cm-‘, light treatment 2 h after drug injection; 0, 90 day tumor cure, 5 mg kg- ‘, 270 J cm-‘, light treatment 2 h after drug injection; Cl, 7 day tumor response, 10 mg kg-‘, 270 J cmm2, light treatment 24 h after drug injection.
complete tumor response lasting 7 days was achieved in 100% of animals over a wavelength range of 660-780 nm, whereas 90 day tumor cures occurred in 80% of mice, but only at wavelengths of 750-780 nm. The increased curability at this wavelength range may be related to the superior tissue penetration of light at these wavelengths [5], although no experiments were carried out to prove this point. If light treatment was given 24 h after drug injection, 7 day tumor response again spanned a wide wavelength range, but was maximal around 700 nm and decreased towards 780 nm. No tumor cures could be obtained under these conditions, even though twice the drug dose was used. No signs of drug (“dark”) toxicity, i.e. no signs of morbidity if the animals were kept in the dark, were observed up to the highest dose injected (35 mg kg-‘). The broad in vivoaction spectrum of bChla corresponds to the absorption spectra of bChla, bPheoa and further oxidation products (probably chlorophylllike (not shown)).
308
To trace the presence of bChla degradation products in the tissue, HPLC analysis of tumor extracts, prepared 2 and 24 h after bChla injection, was performed (Fig. 3); this showed that bPheoa was present 2 h after injection (together with several unidentified products), with levels rising with time after drug administration, It should be kept in mind that bPheoa has less than half the absorptivity of bChlu at 780 nm and that the concentrations of products absorbing at wavelengths other than the 780 nm monitored here are underestimated in these HPLC tracings. Liver extracts showed approximately 70% of injected material was metabolized to bPheou 2 h after injection. Tumor responses in the SMT-F and RIF tumors were closely comparable; bChlu showed poor selectivity in both systems, i.e. complete tumor response always included some necrosis of the normal surrounding skin, especially when treatment was carried out shortly after drug injection. To assess the photosensitizing potential of bChlu on the tumor cells directly, RIF tumor cells isolated after sensitizer accumulation in vivo were exposed to light in vitro and cell kill was determined as a measure of cellular photosensitivity (Fig. 4). The latter was found to increase with time after drug injection, as well as with injected drug dose. The photosensitizer appeared to be tightly bound to tumor cells, since even prolonged exposure to tissuedissociating enzymes did not change cellular photosensitivity as indicated by comparative experiments using only gentle cell dispersion in HBSS (data not shown). All cell survival curves were biphasic, showing breaks and leveling out between 20 and 40 J cmp2, which is consistent with the rapid photodestruction (photobleaching) of the sensitizer. Although the biphasic character of the survival curves could be explained by various other mechanisms such as uneven drug distribution etc., none of the many other, more stable sensitizers tested by us has shown such behavior at comparable survival levels; bChlu
SMT- F-Tumor
Time
(mid
Fig. 3. HPLC analysis of SMT-F tumor tissue extracts (1:l mixture of methyl alcohol dichloromethane); tissue extracted 2 h and 24 h after ixjection of 10 mg kg-’ bChla.
and
309
0,’0
I
I
IO
20
I
30
Joules/cm2
I
I
I
40
50
60
(780nm)
Fig. 4. Cytocidal effects on RIF tumor cells after exposure to bChla in viva, tumor ccl1 isolation and light treatment (780 nm) in vitro; cell survival was measured by clonogenic assay; each point represents the mean+ the standard error (SE) from three experiments, triplicate plates: 0, no drug control; 0, 5 mg kg-i, cell isolation 2 h post-lqjcction; 0, 5 mg kg-‘, cell isolation 24 h post-injection; A, 10 mg kg-i, celI isolation 24 h post-lqjection; A, 35 mg kg-‘, cell isolation 16 h post-injection.
“‘0
I
I
I
1
I
I
i0
20
30
40
50
I
60
Joules/cm* Fig. 5. Cytocidal effects on RIF tumor cells after exposure to bChla in vivo, tumor cell isolation (24 h after drug ir\jection, 10 mg kg-‘) and light treatment of varying wavelength in vitro; cell survival was measured by clonogenic assay; each point represents the mean + SE of three experiments, triplicate plates.
is among the most readily photo-oxidized of a series of non-metallo and metalloporphyrins, chlorins and bacteriochlorophyll-type compounds (91. In to demonstrate of photosensitizing of bChla in tumor 24 h after bChla in vivolin vitro experiments were carried out as described above, but also using the bPheoa absorption wavelength (755 nm) as well as 660 nm (Pig. 5). It should be kept in mind that the ‘Oa quantum yields for bChla and bPheou are very similar @Chlu = 0.6, bPheou = 0.75 [ 31). In all cases photodynamic cell kill could be achieved. Again, the survival curve for 755 nm light showed a break, whereas no such biphasic pattern was found at 660 nm, suggesting no or slower photobleaching of the sensitizing product absorbing at this wavelength.
310
Overall, direct tumor cell photosensitization was poor, falling far short of 1 log of cell kill at therapeutic drug doses over the light dose range tested, and being severely limited by the apparent rapid photobleaching process. Therefore, it appears that direct photodynamic effects on tumor cells can be ruled out as the sole mechanism for tumor eradication by bChla PDT. We therefore evaluated the vascular photosensitization induced by bChla in normal C3H/HeJ mouse back skin and back skin containing RIF tumors (Table 1). Prompt microvascular occlusion could be achieved in light treatment fields (with or without tumors present), if light treatment was carried out within 2 h after drug injection. If light exposure was delayed by 24 h, even a light dose of 270 J cm-’ (which causes complete tumor response in a large percentage of animals, but no tumor cure) did not occlude vessels within the time of light delivery. Vascular occlusion did, however, develop gradually between 2 and 4 h post-treatment. These differences in the expression of vascular photosensitivity can probably be attributed to different levels of circulating photosensitizer at the time of light treatment, a parameter which determines, to a high degree, vascular photosensitization through PDT [lo] with a number of different sensitizers. The fact that tumor cures could only be obtained if light treatment was carried out at short time intervals after drug injection implies that the more severe vascular effects are necessary for cure. This is consistent with our previous experience with these murine tumors and Photofrin II, where cure requires destruction of a margin of healthy tissue surrounding the tumor [ 81. Mechanisms for rapid and delayed vascular effects are currently under investigation. Some of the pattern of vascular occlusion is reflected in the tumor cell death kinetics following in vivo light treatment (Fig. 6), particularly treatment 24 h after drug injection. Here, the number of clonogenic cells per gram of tumor is reduced to between 30% and 50% of the control tumors (containing drug but not exposed to light) by the time light treatment is completed.
TABLE 1 Vascular shutdown following bChla Drug conditions
Light dose (J cm-‘)
Tie
after light treatment
Immediate
5 mg kg-’ (2 h post-iqjection) 10 mg kg-’ (24 h post-iqjection)
67 135 135 270
2h
4h
S
T
S
T
S
T
+/+ -
+ + -
-
+
+ +
+ +
S, skin; T, tumor. Assessed by fluorescein angiography of at least three animals for each treatment condition (780 nm; 75 mW cm-‘).
311
0.1
0.01
0.001
’ -1AO Light
I
,
I
I
I
I
I
I
1
2
3
4
5
6
7
Hours
Fig. 6. Reduction in tumor cell clonogenicity of RIF tumors as a function of time before and after bChla PDT in viva (270 J cm- ‘, 75 mW cm-‘, 780 run); 3-6 individual tumors per point; mean f SE: 0, 10 mg kg-‘, light 24 h post-irjection; 0, 5 mg kg-‘, light 2 h postinjection.
This approaches the expected maximum for direct cell kill (Fig. 4). This is followed by a broad shoulder of about 3 h, during which no further cell kill takes place. After this interval, which corresponds to the interval to delayed vascular shut down, rapid cell death follows indicating the onset of tumor bed effects. Although the results following treatment 2 h after drug injection are not significantly different, tumor cell clonogenicity tends to fall off more sharply, beginning 1 h after light exposure, consistent with the earlier vascular effects. Nothing can be deduced from these data on tumor curability due to the sensitivity limitations of this assay. However, it is certainly dependent on the extent of normal tissue damage surrounding the tumor tissue. To evaluate the effects of bChla PDT on normal tissue (“skin”), the mouse foot response to either light exposure of a specific wavelength (780, 755, 660 run) or to simulated sunlight was tested as a function of time after drug injection. As described elsewhere, the mouse foot response assay seems to be particularly predictive for human skin photosensitivity [lo]. Although light exposure within 2 h caused moderately severe tissue damage (1.6, large epilation and moderate edema, Table 2), light exposure after 24 h caused only mild response (slight to moderate edema and erythema). It appeared that 755 nm light was most effective in eliciting normal skin response at
TABLE 2 Swiss mouse foot response and wavelength
to bCNa
PDT as a function
of time after sensitizer
Time between drug injectiona and light?’
Wavelength
2h
780
1.58 (0.50)
24 h
780 755 660
0.48 0.73 0.00
5d
780 755 660
0.00 0.00 0.00
(nm)
administration
Reaction score day l-2 after treatment (mean, SD)
(0.13) (0.10)
“5 mg kg-‘. b270 J cm-‘, 150 mW cm-‘. Ten animals per group.
24 h post-injection. No response could be elicited with 660 nm light at 24 h post-injection or by any of the wavelengths when light exposure was carried out 5 days after drug injection (Table 2). Solar simulation (exposure equivalent to 30 min of full sun at noon) produced no response when carried out 24 h after bChla injection of 10 mg kg-‘. In summary, bChla appears to be an effective, albeit rather poorly selective, photosensitizer of at least two transplantable mouse tumors. It shows a broad in vivo action spectrum which is consistent with the in vivo breadkdown of this material into the demetallized bPheoa, itself a photosensitizing compound, and into other unidentified materials. Some of these photosensitize near 660 nm and are probably oxidation products of bChla or bPheoa. This chemical instability makes precise pharmacokinetic evaluation, as well as evaluation of in vivo photobleaching rates, extremely difficult. Tumor cures require light exposure shortly after drug injection, while vascular photosensitivity is high, indicating the importance of tumor bed mechanisms in PDT tumor destruction with this sensitizer as observed for others, e.g. Photofrin II or mono-1-aspartyl chlorine e6 [8, 111. Photobleaching of the sensitizer is rapid and limits tumor cell kill. Effects were identical with 780 nm light delivery by a dye laser or by the simpler and more stable diode laser system, demonstrating the feasibility of diode laser application. Another major advantage of this sensitizer seems to be the total disappearance of normal tissue (“skin”) photosensitivity over a period of 5 days, which may be mainly due to the chemical instability of the material. Unfortunately, bChla is not readily commercially available at this time, a fact which has also limited this study, and production is expensive. These circumstances may well limit its broader scientific or clinical use.
313
Acknowledgments
The authors wish to thank Dr. George Truscott for kindly supplying several samples of sensitizer, and F. Scott Nowakowski and Michael Vertino for technical assistance. This investigation was supported by PHS Grants Nos. CA 42278 and CA 16717 awarded by the National Cancer Institute, DHHS.
References 1 T. J. Dougherty, Photodynamic therapy - new approaches, Se&n. Surg. Oncol., 5 (1989) 6-16. 2 A. A. Krasnovsky, Jr., Photoluminescence of singlet oxygen in pigment solutions, Pholocti. Photobiol., 29 (1979) 29-36. 3 A. A. Krasnovskii, Jr., I. V. Vichegzhanina, N. N. Drozdova and A. A. Krasnovskii, Generation and quenching of singlet molecular oxygen by bacteriochlorophyll and bactcriophcophytin a and b, Dokl. Akad. Nauk SSSR, 283 (1985) 474-477. 4 C. F. Borland, D. J. McGarvey, T. G. Truscott, R. J. Codgell and E. J. Land, Photophysical studies of bacterlochlorophyll a and bacteriopheophytin a - singlet oxygen generation, J. Photo&em. Photobiol. B: Biol., I (1987) 93-101. 5 B. C. Wilson, W. P. Jeeves and D. M. Lowe, In viva and post mortem measurements of the attenuation spectra of light in mammalian tissues, Photo&em. Photobiol., 42 (1985)
153-162. 6 T. J. Dougherty,
7
8
9 10
11
G. B. Grindey, R. Fiel, K. R. Welshaupt and D. G. Boyle, Photoradiation therapy II. Cure of animal tumors with hematoporphyrin and light, J. Natl. Cancer Inst., 55 (1975) 115-119. B. W. Henderson, S. M. Waldow, T. S. Mang, W. R. Potter, P. B. Malone and T. J. Dougherty, Tumor destruction and kinetics of tumor cell death in two experimental mouse tumors following photodynamic therapy, Cancer Res., 45 (1985) 572-576. V. H. Fingar and B. W. Henderson, Drug and light dose dependence of photodynamic therapy: a study of tumor and normal tissue response, Photo&em. Photobiol., 46 (1987) 837-841. A. A. Krasnovskii, Y. E. A. Venediktov and 0. M. Chernenko, Overkilling of singlet oxygen by the chlorophylls and porphyrins, Biophysics, 27 (1982) 1009-1016. B. W. Henderson, The significance of vascular photosensitization in photodynamic therapy, ln C. J. Gomer (ed.), Future Directions and Applications in Photodynumic Therapy, IS 6, SPIE Optical Engineering Press, Bellingham, WA, 1990, pp. 153-166. C. J. Gomer, A. Ferrario, M. Luna, N. Rucker and S. Wong, Basic mechanisms and subcellular targets related to PDT, in C. J. Gomer (ed.), Future Directions and Applications in Photodynamic Therapy, IS 6, SPIE Optical Engineering Press, Belllngham, WA, 1990, pp. 139-152.
303
Bacteriochlorophyll-a as photosensitizer for photodynamic treatment of transplantable murine tumors Barbara W. Henderson+, Adam and Thomas J. Dougherty
B. Sumlin, Barbara
L. Owczarczak
Division of Radiation Biology, Roswell Park Cancer Institute, Buffalo, New York (U.S.A.) (Received September 27, 1990; accepted February 12, 1991)
Keyworu!.s. Photodynamic therapy, bacteriochlorophyll-a, a, transplantable mouse tumor photosensitization.
bacteriopheophytin-
Abstract Bacteriochlorophyll-a (bChla), which absorbs light of 780 nm wavelength, was tested for in vivo photodynamic activity in the SMT-F and RIF transplantable mouse tumor systems. High performance liquid chromatography (HPLC) analysis of tissue extracts showed that bChla was rapidly degraded in vivo to bacterlopheophytln-a (bPheoa) and other breakdown products. These were also photodynamically active, and tumor response could be achieved over a wavelength range of 660 to 780 nm, while tumor cure was restricted to wavelengths of 755 @Pheou) to 780 nm. A photosensitizing product absorbing at 660 run was also present in isolated tumor cells. Photodynamic cell kill of tumor cells isolated from tumors after bChlu accumulation in vivo, using 755 or 780 nm light in vitro, was exponential up to 20-40 J cm- ‘. Above this light dose little or no further damage could be achieved, which is an indication of the rapid photobleaching of these sensitizers. In vivo, vascular occlusion occurred readily if light treatment was delivered shortly after sensitizer administration, but was delayed if light treatment was carried out 24 h after injection. Although up to 70% of tumor cells were lethally damaged after completion of in vivo light treatment, concurrent severe vascular destruction seemed necessary for tumor cure. Normal tissue photosensitivity totally subsided within 5 days after sensitizer administration.
1. Introduction Photodynamic therapy (PDT), using the photosensitizer Photofrin II, is perceived to have at least three major limitations [ 1). (i) Photofrin II absorbs light of 630 nm wavelength, which lies well below the wavelength necessary for maximum tissue penetration (700-800 nm). ‘Author to whom correspondence should bc addressed.
loll-1344/91/$3.50
0 1991 -
Elsevier Sequoia, Lausannc
304
(ii) Photofrin II induces prolonged skin photosensitivity, the only major adverse effect known for this sensitizer. (iii) The currently available light delivery systems necessary to produce sufficiently powerful 630 nm light are expensive and difficult to maintain. Bacteriochlorophyll-a (bChla) has several photophysical and chemical characteristics which may help to overcome these limitations. It is a relatively efficient singlet oxygen producer (lOa yields between 0.17 and 0.60, depending on the solvent, have been reported) [2-41 and absorbs strongly (greater than 7 X lo4 M- ’ cm- ’ at 780 nm) [ 41 near the optimum wavelength for tissue penetration [ 51. Furthermore, diode laser technology is being developed to provide inexpensive and simple light delivery at this wavelength.
2. Materials
and methods
2.1. smitixer Bacteriochlorophyll-a was obtained from Porphyrin Products, Logan, UT or Sigma Chemical Co., St. Louis, MO. Several experiments were performed with material kindly supplied by Dr. T. G. Truscott of Paisley College of Technology, Renfrewshire, U.K. All samples used were analyzed by high performance liquid chromatography (HPLC) as described below and were at least 90% pure. Stock solutions were prepared in Hank’s balanced salt solution (HBSS) containing 10% Tween 80, and were further diluted 1:lO with HBSS before in vivo injection. 2.2. Tumor systems The SMT-F (spontaneous mammary carcinoma, fast growing) and RIF (radiation-induced fibrosarcoma) tumors were propagated according to established protocols in DBA/2Ha and C3H/HeJ mice respectively [6, 71. Prior to tumor inoculation, all hair was removed from the right flank of the animal by shaving and depilation. Tumors were used for experimentation when they had reached a surface diameter of 5-6 mm and a thickness of 2-3 mm. 2.3. Extraction and HPLC analysis Tissue samples of tumor and liver were homogenized under subdued light with a 1: 1 mixture of methyl alcohol and dichloromethane and filtered through glass wool; this was then washed several times with the same solvent mixture. Control experiments indicated that bChla was not degraded by this procedure. Two tumors were combined for each determination. HPLC analysis was carried out on all bChla samples as well as tissue extracts using a Spectra Physics (Mountain View, CA) model SP8700 chromatograph and a binary solvent system starting with 90% methanol-water and going to pure methanol over 30 min. A flow rate of 1.5 ml min- ’ was used with a reverse phase Whatman C-8 column. Detection was by absorption at 780 nm using a tunable wavelength detector (Spectroflow 757, also from Spectra Physics).
305
2.4. In vivo photodynamic
therapy
treatment
Animals were given intravenous (i.v.) (tail vein) injections of bChla. At varying times thereafter they were restrained without anesthesia in specially designed holders, and tumors, normal back skin or hind feet were exposed to external light treatment. For light delivery, the following laser systems were used: a 5 or 20 W argon laser system (Spectra Physics) pumping a dye laser using the dyes DCM (Cooper Lasersonics, Inc., Palo Alto, CA) or LDS 751 (Exciton Chemical Co., Dayton, OH) and tuned to the desired wavelength by a birefringent filter; a 1.0 W diode laser (model SDL 820, Spectra Diode Labs, San Jose, CA) tuned to 780 nm by thermoelectric cooling. In each system, the output beam was split using a 50-50 beam splitter (Oriel Optical Corp., Stamford, CT) and coupled to 400 wrn quartz fiberoptic cables (Ensign-Bickford Optics Co., Avon, CT). Microlenses (focal length, 3.5 mm) were fitted to the end of each fiber to facilitate even light distribution throughout the treatment field. The power density of the delivered light was adjusted to 75 mW cmm2 (150 mW ce2 for foot response) at a spot size of 1 cm diameter, and was measured by a radiometer (Yellow Springs Instrument Co. Inc., Yellow Springs, OH). 2.5. Assessment
of tumor
response
and cure
Mice were examined daily for the first 14 days post-treatment and weekly thereafter for a total of 90 days. The disappearance of the tumor bulk was scored as a complete response. If no tumor regrowth had occurred by 90 days, animals were considered to be cured. 2.6. Assessment of vascular photosensitivity Damage to the microvasculature of the tumor-bearing or normal back skin of C3H/HeJ mice was assessed by fluorescence dye exclusion [8]. Briefly, immediately or at varying times after in vivo PDT treatment, animals were given iv. injections (tail vein) of 2.0 mg of fluorescein dye (10 mg ml- ’ of sodium fluorescein (Baker Chemical, Philipsberg, NJ) diluted in HBSS). Immediately thereafter the tumor and/or surrounding skin area was illuminated with UV light (Blak-Ray long-wave UV lamp, Ultra-Violet Products, Inc., San Gabriel, CA), and fluorescence distribution was visualized and recorded. 2.7. Assessment
of foot
response
The photosensitivity of mouse feet was studied in randomly bred Swiss mice. The right or left hind foot of the animals was treated at varying times after i.v. bChla injection with in vivo PDT at a specific wavelength (660, 755, 780 nm) as described above, or with a solar simulator (200 W xenon arc lamp, model 6137, Oriel Corp., Stratford, CT), equipped with an air mass 1 filter, at 90 mW cme2 for 30 min. The foot response was scored daily starting 24 h post-treatment according to the degree of damage (from 0 (no damage) to 2.0 (moderate dry desquamation and swelling of toes)).
2.8. In viva/in vitro cell survival assays RIF tumor cell survival was determined following either of two treatment protocols. (i) Sensitizer was injected into tumor-bearing animals and allowed to accumulate in the tumor for 2-24 h. Tumors were then excised, finely minced in HBSS and/or dispersed by an enzyme procedure [7]; 2 ml aliquots of the single cell suspension were transferred to wells of a 24 well tissue culture plate (Falcon@, Becton Dickinson, Lincoln Park, NJ) and exposed in vitro to light of the desired wavelength, delivered from the 5 W argon dye laser system at a power density of 50 mW cm- 2. Following light treatment, cells were transferred to 60 mm plastic culture plates for clonogenic assay. (ii) Sensitizer was injected into tumor-bearing animals, followed 2 or 24 h later by in viva light treatment at 780 nm as described above. Tumors were excised either immediately or at varying times after light treatment, finely minced, dispersed by an enzyme procedure and plated onto 60 mm plastic culture dishes for clonogenic assay. For clonogenic assay, which has been described in detail elsewhere [ 71, appropriate cell numbers were plated in minimum Eagle’s alpha medium with L-glutamine, ribonucleoside and deoxyribonucleoside, supplemented with 10% fetal bovine serum and antibiotics (all from Grand Island Biological Co., Grand Island, NY). 3. Results and discussion The structures of bChla and its demetallized form bacteriopheophytina (bPheoa) are shown in Fig. 1. The corresponding absorption spectra are presented in Fig. 2 (upper panel) together with the SMT-F tumor response as a function of light treatment wavelength and different treatment conditions (Fig. 2, lower panel). With light treatment carried out 2 h after bChla injection,
.Me
Bacteriochlorophyll-a Fig. 1. Chemical
structures
Eacteriopheaphytin-a of bChla
and bPheoa;
R = phytyl.
307
loo-
& t
80-
P
L-
F
60-
2%
40-
._ 8
20-
$F 00
,c;,, n-x-x--x*
:
:
0
:
X
,
,
,
X
:
:
500 540 580 620-&O -Go 740 780 820 860
Wavelength (nm) Fig. 2. Absorption spectra (in methanol-dichloromethane, 1:l) and SMT-F tumor response as a function of light wavelength. Upper panel: absorbance. Lower panel: tumor response, ten animals per group: X, 7 day tumor response, 5 mg kg- ‘, 270 J cm-‘, light treatment 2 h after drug injection; 0, 90 day tumor cure, 5 mg kg- ‘, 270 J cm-‘, light treatment 2 h after drug injection; Cl, 7 day tumor response, 10 mg kg-‘, 270 J cmm2, light treatment 24 h after drug injection.
complete tumor response lasting 7 days was achieved in 100% of animals over a wavelength range of 660-780 nm, whereas 90 day tumor cures occurred in 80% of mice, but only at wavelengths of 750-780 nm. The increased curability at this wavelength range may be related to the superior tissue penetration of light at these wavelengths [5], although no experiments were carried out to prove this point. If light treatment was given 24 h after drug injection, 7 day tumor response again spanned a wide wavelength range, but was maximal around 700 nm and decreased towards 780 nm. No tumor cures could be obtained under these conditions, even though twice the drug dose was used. No signs of drug (“dark”) toxicity, i.e. no signs of morbidity if the animals were kept in the dark, were observed up to the highest dose injected (35 mg kg-‘). The broad in vivoaction spectrum of bChla corresponds to the absorption spectra of bChla, bPheoa and further oxidation products (probably chlorophylllike (not shown)).
308
To trace the presence of bChla degradation products in the tissue, HPLC analysis of tumor extracts, prepared 2 and 24 h after bChla injection, was performed (Fig. 3); this showed that bPheoa was present 2 h after injection (together with several unidentified products), with levels rising with time after drug administration, It should be kept in mind that bPheoa has less than half the absorptivity of bChlu at 780 nm and that the concentrations of products absorbing at wavelengths other than the 780 nm monitored here are underestimated in these HPLC tracings. Liver extracts showed approximately 70% of injected material was metabolized to bPheou 2 h after injection. Tumor responses in the SMT-F and RIF tumors were closely comparable; bChlu showed poor selectivity in both systems, i.e. complete tumor response always included some necrosis of the normal surrounding skin, especially when treatment was carried out shortly after drug injection. To assess the photosensitizing potential of bChlu on the tumor cells directly, RIF tumor cells isolated after sensitizer accumulation in vivo were exposed to light in vitro and cell kill was determined as a measure of cellular photosensitivity (Fig. 4). The latter was found to increase with time after drug injection, as well as with injected drug dose. The photosensitizer appeared to be tightly bound to tumor cells, since even prolonged exposure to tissuedissociating enzymes did not change cellular photosensitivity as indicated by comparative experiments using only gentle cell dispersion in HBSS (data not shown). All cell survival curves were biphasic, showing breaks and leveling out between 20 and 40 J cmp2, which is consistent with the rapid photodestruction (photobleaching) of the sensitizer. Although the biphasic character of the survival curves could be explained by various other mechanisms such as uneven drug distribution etc., none of the many other, more stable sensitizers tested by us has shown such behavior at comparable survival levels; bChlu
SMT- F-Tumor
Time
(mid
Fig. 3. HPLC analysis of SMT-F tumor tissue extracts (1:l mixture of methyl alcohol dichloromethane); tissue extracted 2 h and 24 h after ixjection of 10 mg kg-’ bChla.
and
309
0,’0
I
I
IO
20
I
30
Joules/cm2
I
I
I
40
50
60
(780nm)
Fig. 4. Cytocidal effects on RIF tumor cells after exposure to bChla in viva, tumor ccl1 isolation and light treatment (780 nm) in vitro; cell survival was measured by clonogenic assay; each point represents the mean+ the standard error (SE) from three experiments, triplicate plates: 0, no drug control; 0, 5 mg kg-i, cell isolation 2 h post-lqjcction; 0, 5 mg kg-‘, cell isolation 24 h post-injection; A, 10 mg kg-i, celI isolation 24 h post-lqjection; A, 35 mg kg-‘, cell isolation 16 h post-injection.
“‘0
I
I
I
1
I
I
i0
20
30
40
50
I
60
Joules/cm* Fig. 5. Cytocidal effects on RIF tumor cells after exposure to bChla in vivo, tumor cell isolation (24 h after drug ir\jection, 10 mg kg-‘) and light treatment of varying wavelength in vitro; cell survival was measured by clonogenic assay; each point represents the mean + SE of three experiments, triplicate plates.
is among the most readily photo-oxidized of a series of non-metallo and metalloporphyrins, chlorins and bacteriochlorophyll-type compounds (91. In to demonstrate of photosensitizing of bChla in tumor 24 h after bChla in vivolin vitro experiments were carried out as described above, but also using the bPheoa absorption wavelength (755 nm) as well as 660 nm (Pig. 5). It should be kept in mind that the ‘Oa quantum yields for bChla and bPheou are very similar @Chlu = 0.6, bPheou = 0.75 [ 31). In all cases photodynamic cell kill could be achieved. Again, the survival curve for 755 nm light showed a break, whereas no such biphasic pattern was found at 660 nm, suggesting no or slower photobleaching of the sensitizing product absorbing at this wavelength.
310
Overall, direct tumor cell photosensitization was poor, falling far short of 1 log of cell kill at therapeutic drug doses over the light dose range tested, and being severely limited by the apparent rapid photobleaching process. Therefore, it appears that direct photodynamic effects on tumor cells can be ruled out as the sole mechanism for tumor eradication by bChla PDT. We therefore evaluated the vascular photosensitization induced by bChla in normal C3H/HeJ mouse back skin and back skin containing RIF tumors (Table 1). Prompt microvascular occlusion could be achieved in light treatment fields (with or without tumors present), if light treatment was carried out within 2 h after drug injection. If light exposure was delayed by 24 h, even a light dose of 270 J cm-’ (which causes complete tumor response in a large percentage of animals, but no tumor cure) did not occlude vessels within the time of light delivery. Vascular occlusion did, however, develop gradually between 2 and 4 h post-treatment. These differences in the expression of vascular photosensitivity can probably be attributed to different levels of circulating photosensitizer at the time of light treatment, a parameter which determines, to a high degree, vascular photosensitization through PDT [lo] with a number of different sensitizers. The fact that tumor cures could only be obtained if light treatment was carried out at short time intervals after drug injection implies that the more severe vascular effects are necessary for cure. This is consistent with our previous experience with these murine tumors and Photofrin II, where cure requires destruction of a margin of healthy tissue surrounding the tumor [ 81. Mechanisms for rapid and delayed vascular effects are currently under investigation. Some of the pattern of vascular occlusion is reflected in the tumor cell death kinetics following in vivo light treatment (Fig. 6), particularly treatment 24 h after drug injection. Here, the number of clonogenic cells per gram of tumor is reduced to between 30% and 50% of the control tumors (containing drug but not exposed to light) by the time light treatment is completed.
TABLE 1 Vascular shutdown following bChla Drug conditions
Light dose (J cm-‘)
Tie
after light treatment
Immediate
5 mg kg-’ (2 h post-iqjection) 10 mg kg-’ (24 h post-iqjection)
67 135 135 270
2h
4h
S
T
S
T
S
T
+/+ -
+ + -
-
+
+ +
+ +
S, skin; T, tumor. Assessed by fluorescein angiography of at least three animals for each treatment condition (780 nm; 75 mW cm-‘).
311
0.1
0.01
0.001
’ -1AO Light
I
,
I
I
I
I
I
I
1
2
3
4
5
6
7
Hours
Fig. 6. Reduction in tumor cell clonogenicity of RIF tumors as a function of time before and after bChla PDT in viva (270 J cm- ‘, 75 mW cm-‘, 780 run); 3-6 individual tumors per point; mean f SE: 0, 10 mg kg-‘, light 24 h post-irjection; 0, 5 mg kg-‘, light 2 h postinjection.
This approaches the expected maximum for direct cell kill (Fig. 4). This is followed by a broad shoulder of about 3 h, during which no further cell kill takes place. After this interval, which corresponds to the interval to delayed vascular shut down, rapid cell death follows indicating the onset of tumor bed effects. Although the results following treatment 2 h after drug injection are not significantly different, tumor cell clonogenicity tends to fall off more sharply, beginning 1 h after light exposure, consistent with the earlier vascular effects. Nothing can be deduced from these data on tumor curability due to the sensitivity limitations of this assay. However, it is certainly dependent on the extent of normal tissue damage surrounding the tumor tissue. To evaluate the effects of bChla PDT on normal tissue (“skin”), the mouse foot response to either light exposure of a specific wavelength (780, 755, 660 run) or to simulated sunlight was tested as a function of time after drug injection. As described elsewhere, the mouse foot response assay seems to be particularly predictive for human skin photosensitivity [lo]. Although light exposure within 2 h caused moderately severe tissue damage (1.6, large epilation and moderate edema, Table 2), light exposure after 24 h caused only mild response (slight to moderate edema and erythema). It appeared that 755 nm light was most effective in eliciting normal skin response at
TABLE 2 Swiss mouse foot response and wavelength
to bCNa
PDT as a function
of time after sensitizer
Time between drug injectiona and light?’
Wavelength
2h
780
1.58 (0.50)
24 h
780 755 660
0.48 0.73 0.00
5d
780 755 660
0.00 0.00 0.00
(nm)
administration
Reaction score day l-2 after treatment (mean, SD)
(0.13) (0.10)
“5 mg kg-‘. b270 J cm-‘, 150 mW cm-‘. Ten animals per group.
24 h post-injection. No response could be elicited with 660 nm light at 24 h post-injection or by any of the wavelengths when light exposure was carried out 5 days after drug injection (Table 2). Solar simulation (exposure equivalent to 30 min of full sun at noon) produced no response when carried out 24 h after bChla injection of 10 mg kg-‘. In summary, bChla appears to be an effective, albeit rather poorly selective, photosensitizer of at least two transplantable mouse tumors. It shows a broad in vivo action spectrum which is consistent with the in vivo breadkdown of this material into the demetallized bPheoa, itself a photosensitizing compound, and into other unidentified materials. Some of these photosensitize near 660 nm and are probably oxidation products of bChla or bPheoa. This chemical instability makes precise pharmacokinetic evaluation, as well as evaluation of in vivo photobleaching rates, extremely difficult. Tumor cures require light exposure shortly after drug injection, while vascular photosensitivity is high, indicating the importance of tumor bed mechanisms in PDT tumor destruction with this sensitizer as observed for others, e.g. Photofrin II or mono-1-aspartyl chlorine e6 [8, 111. Photobleaching of the sensitizer is rapid and limits tumor cell kill. Effects were identical with 780 nm light delivery by a dye laser or by the simpler and more stable diode laser system, demonstrating the feasibility of diode laser application. Another major advantage of this sensitizer seems to be the total disappearance of normal tissue (“skin”) photosensitivity over a period of 5 days, which may be mainly due to the chemical instability of the material. Unfortunately, bChla is not readily commercially available at this time, a fact which has also limited this study, and production is expensive. These circumstances may well limit its broader scientific or clinical use.
313
Acknowledgments
The authors wish to thank Dr. George Truscott for kindly supplying several samples of sensitizer, and F. Scott Nowakowski and Michael Vertino for technical assistance. This investigation was supported by PHS Grants Nos. CA 42278 and CA 16717 awarded by the National Cancer Institute, DHHS.
References 1 T. J. Dougherty, Photodynamic therapy - new approaches, Se&n. Surg. Oncol., 5 (1989) 6-16. 2 A. A. Krasnovsky, Jr., Photoluminescence of singlet oxygen in pigment solutions, Pholocti. Photobiol., 29 (1979) 29-36. 3 A. A. Krasnovskii, Jr., I. V. Vichegzhanina, N. N. Drozdova and A. A. Krasnovskii, Generation and quenching of singlet molecular oxygen by bacteriochlorophyll and bactcriophcophytin a and b, Dokl. Akad. Nauk SSSR, 283 (1985) 474-477. 4 C. F. Borland, D. J. McGarvey, T. G. Truscott, R. J. Codgell and E. J. Land, Photophysical studies of bacterlochlorophyll a and bacteriopheophytin a - singlet oxygen generation, J. Photo&em. Photobiol. B: Biol., I (1987) 93-101. 5 B. C. Wilson, W. P. Jeeves and D. M. Lowe, In viva and post mortem measurements of the attenuation spectra of light in mammalian tissues, Photo&em. Photobiol., 42 (1985)
153-162. 6 T. J. Dougherty,
7
8
9 10
11
G. B. Grindey, R. Fiel, K. R. Welshaupt and D. G. Boyle, Photoradiation therapy II. Cure of animal tumors with hematoporphyrin and light, J. Natl. Cancer Inst., 55 (1975) 115-119. B. W. Henderson, S. M. Waldow, T. S. Mang, W. R. Potter, P. B. Malone and T. J. Dougherty, Tumor destruction and kinetics of tumor cell death in two experimental mouse tumors following photodynamic therapy, Cancer Res., 45 (1985) 572-576. V. H. Fingar and B. W. Henderson, Drug and light dose dependence of photodynamic therapy: a study of tumor and normal tissue response, Photo&em. Photobiol., 46 (1987) 837-841. A. A. Krasnovskii, Y. E. A. Venediktov and 0. M. Chernenko, Overkilling of singlet oxygen by the chlorophylls and porphyrins, Biophysics, 27 (1982) 1009-1016. B. W. Henderson, The significance of vascular photosensitization in photodynamic therapy, ln C. J. Gomer (ed.), Future Directions and Applications in Photodynumic Therapy, IS 6, SPIE Optical Engineering Press, Bellingham, WA, 1990, pp. 153-166. C. J. Gomer, A. Ferrario, M. Luna, N. Rucker and S. Wong, Basic mechanisms and subcellular targets related to PDT, in C. J. Gomer (ed.), Future Directions and Applications in Photodynamic Therapy, IS 6, SPIE Optical Engineering Press, Belllngham, WA, 1990, pp. 139-152.
