Reduction of tumour oxygenation during and after photodynamic therapy in vivo: effects of fluence rate.

It has been proposed that the generation of O2 during photodynamic therapy (PDT) may lead to photochemical depletion of ambient tumour oxygen, thus causing acute hypoxia and limiting treatment effectiveness. We have studied the effects of fluence rate on pO2, in the murine RIF tumour during and after PDT using 5 mg kg(-1) Photofrin and fluence rates of 30, 75 or 150 mW cm(-2). Median pO2 before PDT ranged from 2.9 to 5.2 mmHg in three treatment groups. Within the first minute of illumination, median tumour pO2 decreased with all fluence rates to values between 0.7 and 1.1 mmHg. These effects were rapidly and completely reversible if illumination was interrupted. During prolonged illumination (20-50 J cm(-2)) pO2 recovered at the 30 mW cm(-2) fluence rate to a median value of 7.4 mmHg, but remained low at the 150 mW cm(-2) fluence rate (median pO2 1.7 mmHg). Fluence rate effects were not found after PDT, and at both 30 and 150 mW cm(-2) median tumour pO2 fell from control levels to 1.0-1.8 mmHg within 1-3 h after treatment conclusion. PDT with 100 J cm(-2) at 30 mW cm(-2) caused significantly (P = 0.0004) longer median tumour regrowth times than PDT at 150 mW cm(-2), indicating that lower fluence rate can improve PDT response. Vascular perfusion studies uncovered significant fluence rate-dependent differences in the responses of the normal and tumour vasculature. These data establish a direct relationship between tumour pO2, the fluence rate applied during PDT and treatment outcome. The findings are of immediate clinical relevance.

effectiveness, tissue pO2 may therefore be a determining factor.
In tumours, pre-existing regional hypoxia has been recognized as a barrier to effective treatment with other oxygen-dependent modalities, such as radiotherapy, and it may play a role in PDT as well (Fingar et al, 1992a). In addition, there exist two mechanisms by which PDT itself may induce tissue hypoxia and thus become self-limiting. Hypoxia may result from PDT-mediated damage to the microvasculature, which usually develops with time after PDT, but which can occur acutely during light delivery at high photosensitizer and light doses with certain photosensitizers (Henderson hypoxia is the consumption of 3°2 in the PDT-mediated formation and consumption of 102. Very rapid and substantial reductions in tissue oxygen tensions upon illumination of photosensitized tissue have been detected with transcutaneous oxygen electrodes by Tromberg et al (1990). Mathematical modelling by Foster et al (1991) has demonstrated that the rate of oxygen consumption during Photofrin PDT may be significant enough to move a fraction of the tumour into very low levels of oxygenation, outpacing the rate of oxygen diffusion from the capillaries and shrinking the radius of oxygenated tissue volume around them.
The rate of '02 generation is influenced by the fluence rate of illumination, suggesting that 302 consumption may in part be controlled by this treatment parameter. Studies in multicell tumour spheroids in vitro have revealed a strong relationship between the fluence rate, oxygen consumption (measured both by oxygen microelectrodes and modelled mathematically) and the fraction of PDT surviving cells (Foster et al, 1993;Nichols and Foster, 1994). The superior effectiveness of lower fluence rates in delaying tumour regrowth compared with treatments at higher fluence rates with the same overall fluence has been demonstrated by Gibson et al (1990) and Foster et al (1991). These results may be explained, at least theoretically, by more sustained tumour oxygenation throughout the course of treatment (Foster et al, 1991(Foster et al, , 1993Nichols and Foster, 1994).
This paper provides an experimental analysis of the relationships between fluence rate of light treatment, tumour oxygenation, vascular perfusion of tumour and normal skin, and tumour response.

MATERIALS AND METHODS Animals and tumour system
The RIF (radiation-induced fibrosarcoma) transplantable tumour, propagated in 8to 12-week-old female C3Hf/HeRos mice (obtained from the RPCI breeding facilities), was used for all experiments requiring tumours. All aspects of animal experimentation and husbandry were carried out in compliance with federal and state standards, and were approved by the Animal Care and Use Committee of RPCI. For tumour inoculation, 3 x 105 RIF cells, obtained from established tumours by enzyme digestion according to established procedures of successive in vitro and in vivo passage (Twentyman et al, 1980), were injected intradermally on the animals' right shoulder, the inoculation site having been shaved and depilated with Nair 24 h earlier. Animals were used 5-7 days after inoculation, when their tumours were about 4-6 mm in diameter and 2-4 mm in height. For microvascular perfusion studies of the skin, an area (1 cm in diameter) was also depilated on the left shoulder.

Photosensitizers
Photofrin (QLT PhotoTherapeutics, Vancouver, BC, Canada) was used in all experiments. It was reconstituted in 5% dextrose in water (D5W) according to company instructions. Mice were injected i.v. (via tail vein) with 5 mg kg-' Photofrin 24 h before light exposure.

Determination of photosensitizer tumour concentrations
Tumour concentrations of Photofrin were determined 24 h after photosensitizer administration by fluorescence measurements after tissue solubilization as described (Bellnier et al, 1997). Fluorescence spectra of test samples were compared with a standard curve and gg of drug per g of tumour was calculated. The obtained values were used as input parameters for the computer simulation of 302 and 1O2distlibutions (see below).

In vivo photodynamic treatment
Twenty-four hours after photosensitizer administration, tumours and/or normal skin were exposed to light from a dye laser (Model 375, Spectra-Physics, Mountain View, CA, USA) pumped by an argon-ion laser (either Trimedyne Optilase, Santa Anna, CA, USA, or Model 171, Spectra-Physics). The dye laser was coupled to a 200-gm diameter quartz fibre fitted with a GRIN lens to provide uniform illumination. The light was set at a wavelength of 630 nm with a monochromator (DMC1-02, Optometrics USA, Ayer, MA, USA). Power output was measured with a thermal disc calorimeter (Model 210, Coherent, Auburn, CA, USA), and the power density was adjusted over a treatment field of 1 cm in diameter to yield fluence rates of 30, 75 or 150 mW cm-2. Total treatment fluences varied with the experimental end point to be measured and are given under Results.

Measurement of tumour P02
Intratumour pO2 was measured polarographically using the Eppendorf pO2 Histograph (Eppendorf Scientific, Madison, WI, USA) (Vaupel et al, 1991). Before and between measurements, the instrument was calibrated in 0.9% saline bubbled alternately with air and nitrogen to set 100% and 0% pO2 currents. Ambient air pressure and tumour temperature (measured using an Omega 30 gauge, '/2-inch needle thermocouple) were recorded and used to post-calibrate the data. For each experimental condition, the last track used for pO2 measurement was used for insertion of the thermocouple.
Tumours ranging in size from 45 to 145 mm3 were used for oxygen measurements; at that size, no spontaneous tumour necrosis is evident. Measurements were performed under sodium pentobarbital anaesthesia (66 mg kg-' i.p.) unless otherwise indicated. The 300-gm-diameter polarographic needle probe was aligned at the tumour surface after creation of a pinpoint hole, using a hypodermic needle, in the skin covering the tumour. The probe was advanced one step to ensure the tip was within the tumour, and automatic probe advancement was begun after the p02 values had stabilized. Probe advancement was set to a 0.7-mm forward step and a 0.3-mm retraction step for each reading. Probe track length and number of tracks measured were determined by tumour dimensions. Tracking was horizontal through the tumour, and care was taken to choose comparable track positions (tumour periphery vs centre) for control and experimental measurements. One to two tracks were measured for each measurement condition (see below), with a maximum of six tracks per tumour.
PDT treatments were carried out at fluence rates of 30, 75 or 150 mW cm-2. The following measurement protocols were used: (1) in each tumour, pO2 was measured immediately before illumination, within the first minute into illumination, and again during illumination when similar total light fluences in the range of 20-50 J cm-2 were reached; for measurements within the first minute the probe was aligned before illumination, while for measurements at the longer time intervals the probe was aligned during illumination; the fluence range of 20-50 J cm-2 was dictated by the time required to align the probe and carry out the PO2 measurements, i.e. measurements at 150 mW cm-2 were carried out between 2 and 6min into illumination, whereas measurements at 30 mW cm-2 were carried out between 20 and 26 min into illumination; (2) in each tumour, P02 was measured within the first minute of illumination, the light was removed and pO2 was again measured within the first minute after illumination; (3) in separate groups of tumours, pO2 was measured immediately and at 1, 3 and 5 h after a full treatment dose (100 J cm-2) of PDT. Appropriate control measurements were also performed on totally untreated tumours and tumours receiving light only, as well as on untreated tumours in mice with or without anaesthesia. In experiments that required prolonged illumination, mice were kept warm using a microwaveable hot pad. Five to ten animals were used per experimental group. Data are expressed as mean and median values, as well as percentages of values < 2 mmHg and . 5 mmHg. These parameters were chosen for analysis because they lie within the range of oxygen concentrations reported in the literature, in which 102 formation becomes limited by oxygen availability (Moan and Sommer, 1985;Foster and Nichols, 1995).

Determination of tumour response
The animals were restrained, unanaesthetized, in specially designed holders and PDT was carried out at fluence rates of 30, 75 and 150 mW cm-2, with a total fluence of 75, 100 and 135 J cm-2 being delivered at each fluence rate. After treatment, tumour size was recorded 6 days a week for 30 days, and then weekly up to 90 days. Tumour size was measured in two orthogonal dimensions using an electronic calliper (Ultra-Cal Mark III, Fred V Fowler, Newton, MA, USA) and automatically entered into a spreadsheet (Excel, Microsoft, Redmond, WA, USA). Tumour volume was calculated with the formula V = (1 x w2)/2, and time to reach 400 mm3 was estimated using the natural log of the volume. Groups of eight to thirteen animals were evaluated. Results are presented as Kaplan-Meier curves in which the percentage of animals with tumour volumes less than 400 mm3 is plotted against the number of hours since treatment (GraphPad Prism, GraphPad Software, San Diego, CA, USA).
Fluorescein exclusion assay of skin microvascular perfusion This method for assessing microvascular perfusion of the normal skin before and after PDT has been described in detail earlier (Bellnier et al, 1995). Briefly, immediately after PDT light exposure and at 1, 3, 5 and 24 h thereafter, unanaesthetized animals were injected with 0.2 ml of 2 mg ml-' fluorescein (JT Baker Chemical, Phillipsburg, NJ, USA) in Hanks' balanced salt solution (HBSS) via the orbital plexus. Five to 8 min after fluorescein injection, fluorescein-specific fluorescence within and outside the treatment field (four readings each) was detected by a fluorometer. At least three mice were evaluated for each data point. Data were recorded as the ratio (treated/untreated sites) of the mean values of the measurements. The PDT treatment parameters for these experiments were a total fluence of 100 J cm-2 at fluence rates of 30, 75 and I00 mW cm-2.
'*RbCI uptake as a measure of tumour and skin microvascular perfusion This technique has been described by others (Lyng et al, 1992) and was used with slight modifications. Briefly, 86RbCl in 0.5 M (2-6 min) (P = 0.002) (P = 0.004) (P = 0.01) (P = 0.002) aTumour P02 was measured in three groups of animals, one for each fluence rate; changes in P02 were recorded for individual tumours from pre-illumination through illumination as described; data were calculated from the pooled results of all measurements for each condition; P-values were calculated for differences between the pre-illumination values and values of a given illumination condition using a paired test comparing changes within individual tumours.
British Journal of Cancer (1998) 77(9) . was diluted in HBSS to 125 jiCi ml-'. Mice were anaesthetized as described for pO2 measurements and injected with 0.2 ml (25 jiCi) of 86RbCl intraorbitally. This injection route yielded identical results to i.v. injection via tail vein (data not shown) and was adopted for ease and accuracy. Two minutes later, mice were sacrificed by cervical dislocation and their treated or untreated tumour and skin patch were excised and weighed. Activity was counted in a gamma counter (Cobra II Auto-Gamma Counter, Packard Instrument, Meriden, CT, USA). For tumour tissue the percentage of injected 86Rb taken up per g of tissue was used to express the data, with values corrected for mouse body weight. Because of the time-dependent development of oedema in the treated skin, and thus increase in the weight of the excised skin, weight was not used in the expression of the skin results. Instead, care was taken to keep the diameter of the excised skin patch as identical as possible. Values were only corrected for mouse body weight.
Tumour and skin perfusion was determined during PDT at 30 and 150 mW cm-2 by injecting 86Rb after 1 min of light exposure and continuing with a further 2 min of light exposure before sacrificing the animal. Perfusion was also assessed immediately and 1, 5 and 24 h after PDT of 100 J cm-2 at fluence rates of 30 and 150 mW cm-2. Appropriate light-only and drug-only controls were included. At least three animals were used for each experimental condition.
Computer simulation of 302 and 102 distribution during PDT Computer software (PDT molecular oxygen-depletion model, PDT MODeM), developed by Henning et al (1995) based on previous mathematical modelling by Foster et al (1991), was used to visualize the effects of fluence rate on tumour 3°2 and 102 distribution throughout the intercapillary space during PDT and to predict a low fluence rate that might allow sustained tumour oxygenation in the RIF tumour model. In addition to the Photofrin-specific photophysical and photochemical reaction parameters (Foster et al, 1991), the following RIF tumour-specific parameters were entered: Photofrin tumour concentration, 11.2 jg g-' (experimentally determined as described above); vessel diameter, 7.55 jm (Fenton and Way, 1993); intercapillary distance, 240 jm (Fenton and Way, 1993); intracapillary HbO2 saturation, 30% (Rofstad et al, 1988).

Statistical considerations
For oxygen measurements, mean and median pO2 values were calculated using the program pO2 Pool (version 1.2), provided by  (Mantel, 1966) was performed in BMDP IL (1990) for analysis of time of tumour regrowth after treatment. All analyses were carried out using Instat (GraphPad Software, San Diego, CA, USA), unless otherwise indicated.

Simulated distribution of 302 and 102 during PDT
To facilitate the choice of appropriate experimental conditions for the biological end points, the effects of fluence rate on the distribution of 3°2 and '02 throughout the intercapillary space were computer simulated. Figure IA depicts the predicted steady-state distribution in 3°2 concentration in the RIF tumour during Photofrin PDT as a function of distance from a capillary during tumour illumination at 30, 75 or 150 mW cm-2. These steady-state values were reached within seconds of beginning illumination. According to Figure IA, the radius of well-oxygenated cells around the capillary can be expected to decrease with increasing fluence rate. Correspondingly, the radius of cells exposed to high levels of 102 can also be expected to shrink (Figure iB). It is noteworthy, however, that at high fluence rates the peak 102 concentrations in close proximity to the vessel will greatly exceed those at low fluence rates. As the model predicted that the fluence rate of 30 mW cm-2 would allow relatively sustained tumour oxygenation during PDT, the fluence rate of 30 mW cm-2 was chosen for our experimental studies. Fluence rates of 150 mW cm-2, the most often used clinical fluence rate, and 75 mW cm-2 were chosen because the model predicted marked, graded tumour oxygen depletion.
To test the relevance of the above predictions, direct measurements of intratumour oxygen partial pressures were carried out before and during Photofrin PDT. A number of control conditions were initially evaluated to determine the reliability of the method. No significant differences (all P-values 2 0.83) were detected between measurements in separate groups of untreated animals with or without anaesthesia (no anaesthesia: mean 11.4 mmHg, median 4.1 mmHg, < 2 mmHg 37%, 2 5 mmHg 44%, N = 9; anaesthesia: mean 11.5 mmHg, median 4.7 mmHg, < 2 mmHg 33%, . 5 mmHg 50%, N = 18). Therefore all subsequent measurements were carried out in anaesthetized animals. Light-only (no Photofrin) effects were evaluated sequentially in individual tumours (N = 9). No significant difference (P 2 0.22) was found between measurements before illumination (mean 8.3 mmHg, median 2.7 mmHg, <2 mmHg 34%, .5 mmHg 34%), those within the first minute of illumination (mean 8.5 mmHg, median 2.8 mmHg, < 2 mmHg 49%, 2 5 mmHg 43%) and those during the delivery of -20-50 J cm-2 at 150 mW cm-2 (mean 10.1 mmHg, median 2.4 mmHg, < 2 mmHg 46%, . 5 mmHg 41%). The number of previous tracks used within the limits of our experimental set-up did not seem to influence subsequent measurements within a given tumour. Effects of photosensitizer alone (no tumour illumination) were analysed through comparisons of tumour pO2 in mice receiving Photofrin (mean 10.2 mmHg, median 4.0 mmHg, < 2 mmHg 21%, 2 5 mmHg 44%, N = 32) with those who received nothing but anaesthesia (see above). No significant differences were found between these two groups (P 2 0.08). While, overall, the above control groups were quite consistent, variations in measurements among individual tumours were found. This appeared to be related to tumour geometry, with more compact, nodular tumours exhibiting lower intratumour pO2 than flat and thin tumours of similar volume. For this reason, fluence rate experiments were carried out with each animal/tumour serving as its own control, and data were evaluated in a paired manner. Table 1 summarizes the effects of illumination at different fluence rates on tumour oxygen pressure in Photofrin-sensitized tumours. When pO2 was measured within the first minute of light exposure, all three fluence rates produced similar, significant decreases in mean and median pO2 values, as well as in the percentage of values 2 5 mmHg, with a concomitant increase in values < 2 mmHg. When pO2 was measured again in the same tumours under illumination after an approximate fluence of 20-50 J cm-2 had been accumulated, the 30 mW cm-2 fluence rate emerged as markedly different from the higher fluence rates; all oxygenation parameters had returned and slightly surpassed the initial baseline values, with the percentage of values .5 mmHg actually being significantly greater than baseline. A smaller recovery was found with the 75 mW cm-2 fluence rate and no recovery at all with the 150 mW cm-2 fluence rate.
The reversibility of fluence rate-dependent oxygen depletion was tested in an additional group of animals. Figure 2 represents histograms derived from measurements in Photofrin-sensitized tumours under 150 mW cm-2 illumination, taken within the first minute of light ( Figure 2A) and again within 1 min after termination of light (Figure 2 B). Severe oxygen depletion under high fluence rate exposure was again seen. However, oxygenation was rapidly restored to levels similar to those before illumination (see Table 1) when illumination had ceased.
British Journal of Cancer (1998) 77(9) Analysis of the pooled results of tumour temperature measurements showed no consistent changes at any fluence rate, and no significant differences between pre-illumination conditions could be demonstrated by paired analysis of values. However, changes in temperature in individual tumours ranged widely, and both temperature increases and decreases were registered. Light of 150 mW cm-2 for 2 min, for example, resulted in average temperature changes from --20C to -+ 20C. Tumour P02 after PDT In a separate series of experiments, tumour pO2 was measured in a time-dependent manner after completion of light treatment (fluence 100 J cm-2) at 30 and 150 mW cm-2, with different groups of animals used for each measurement condition (Table 2). No significant differences from untreated controls were found in any of the oxygenation parameters immediately after cessation of light at either fluence rate. With time, values for mean and median pO2 and the percentage of values 2 5 mmHg decreased, while the percentage of values < 2 mmHg increased. Differences in the percentage of values < 2 mmHg and 2 5 mmHg were both significantly different from controls by 3 h post treatment, but not significantly different between the two fluence rates. Tumour response to PDT The possible treatment advantage to maintaining tumour pO2 by lowering the PDT fluence rate was examined through tumourresponse studies (Figure 3). In animals that received a total fluence of lOO J cm-2, the time of tumour regrowth to a volume of 400 mm3 was clearly fluence rate dependent, with the most rapid regrowth after the 150 mW cm-2 treatments and the most pronounced regrowth delay after 30 mW cm-2 PDT. Analysis of the median time of tumour regrowth to 400 mm3 showed the differences at these two fluence rates to be significant (P = 0.0004). Furthermore, while the 150 mW cm-2 fluence rate did not cure (O out of 14 animals) any of the mice treated with 100 J cm-2, cures were found at the lower fluence rates: 9% (1 out of 11 animals) at 75 mW cm-2 and 15% (2 out of 13 animals) at 30 mW cm-2. Even when a higher fluence was used (135 J cm-2), no cures (none out of eight animals) were found with 150 mW cm-2, while with 30 mW cm-2 cures (two out of eight animals) were still found at a fluence of 75 J cm-2.

Microvascular perfusion during and after PDT
In view of the predicated differences in peak 102 generation near the vessels as a function of fluence rate ( Figure 1B), fluence ratedependent changes in tumour and skin microvascular perfusion were examined.
After PDT (100 J cm-2) vascular perfusion in the tumour, as measured by 86Rb uptake, rapidly and markedly decreased with no significant differences between fluence rates (Figure 4). Perfusion remained low for at least 24 h, possibly showing a slight upward trend at that time point. Perfusion changes in the normal cutaneous microvasculature displayed a very different pattern. It has to be noted that the development of cutaneous oedema with time after PDT posed a problem for the usual presentation of the 86Rb uptake assay data, as the increase in skin weight artificially depressed the British Journal of Cancer (1998) 77 (9) measured values of the later time points when these were corrected for tissue weight. Tissue weight was therefore eliminated from the data calculation and care was taken to excise and evaluate identically sized skin samples ( Figure SB). As further validation of the 86Rb data, the fluorescein exclusion assay was also used to assess cutaneous perfusion changes ( Figure SA). Both approaches yielded similar results that demonstrated significant differences in the cutaneous vascular response to PDT as a function of fluence rate. Light delivery of 100 J cm-2 at 150 mW cm-2 showed a -50% reduction in perfusion by the end of light exposure, but perfusion returned to control levels within 1 h and remained there for the 24-h observation period. In contrast, the same fluence delivered at 30 mW cm-2 produced no acute reduction in perfusion, but perfusion decreased to very low levels over a 5-h post-PDT interval. At 24 h, skin appeared virtually unperfused in the fluorescein exclusion assay, but showed a slight recovery with the 86Rb assay. A fluence rate of 75 mW cm-2 (Figure SA) produced intermediate results.

DISCUSSION
This study confirms the concept of photochemical oxygen depletion during PDT, which was put forward by Tromberg et al (1990) and Foster et al (1991), and provides direct experimental evidence of PDT-induced oxygenation changes in a preclinical tumour system.
Our pre-light measurements of RIFpO2 (median 2.9-5.2 mmHg) agree well with studies by others (medians -1.2-4.0 mmHg) (Horsman et al, 1994;Honess et al, 1995;Kavanaugh et al, 1996) using the Eppendorf oxygen electrode system. These values were not significantly affected by the use of sodium pentobarbital anaesthesia either in this study or in others (Kavanaugh et al, 1996).
Measurements of tumour pO2 during PDT at various fluence rates (Table 1) revealed that lower fluence rate was associated with better sustained tumour oxygenation, in agreement with Foster's modelling of fluence rate-dependent photochemical oxygen depletion (Foster et al, 1991). The enhanced tumour pO2 during lower fluence rate PDT was also in overall agreement with the predictions of the PDT MODeM computer simulation. This program was developed as a research tool and is based on the simplest assumptions of oxygen demand and supply (Henning et al, 1995), i.e. it does not consider such potential complications as changes in fluence rate as light penetrates through tissue, photobleaching of the sensitizer, uneven photosensitizer distribution, changes in metabolic oxygen consumption and changes in vascular supply during PDT. Accordingly, this program was used simply as a guide to suggest fluence rates that would differentially affect tumourpO2 during PDT. The Eppendorf pO2 histograph was used to directly measure tumour pO2 during illumination at these fluence rates. Close examination of the Eppendorf pO2 measurements reveals that while low PDT fluence rate enhanced tumour oxygenation when a total fluence of 20-50 J cm-2had been delivered, with only 1 min of illumination all fluence rates resulted in a similar significant decrease in tumour pO2. Several explanations for these results are possible. (1) The absence of any significant difference in the uniformly low pO2 values of the earliest measurements for the three fluence rates may be a consequence of the limited sensitivity of the pO2 histograph at very low tissue oxygen tensions (Stone et al, 1993). (2) The observed decreases in tumour pO2 during PDT may be the result of vessel constriction and therefore reduced blood flow, whereby low fluence rate would cause reversible effects and high fluence rate would cause irreversible effects. However, our own data on tumour perfusion do not show any evidence of vessel constriction during light exposure, and several other facts contradict this possibility. Studies by Fingar et al (1992b) have found no acute vessel constriction associated with the low dose (5 mg kg-') of Photofrin used in this study.
Additionally, we found that, after 150 mW cm-2 PDT, tumour pO2 recovered within 1 min of terminating illumination (Figure 2). This observation is consistent with the concept of photochemical oxygen depletion and observations by Tromberg et al (1990), and does not agree with the finding by Fingar et al (1992b) of PDTinduced vascular constriction lasting for -1 h after treatment. (3) The observed recovery of tumour pO2 at low fluence rate with low-treatment fluence may be due to a decrease in metabolic oxygen consumption as a result of direct cell effects during PDT (Gibson et al, 1990), i.e. the sustained oxygenation during lowfluence-rate PDT would increase direct cell damage and thus decrease metabolic oxygen consumption. However, Foster et al (1991) have described any such decrease as negligible compared with photochemical oxygen depletion during PDT, thus this is an unlikely mechanism. (4) The fluence rate-dependent recovery of tumour oxygenation at lower fluence rates may be a consequence of photosensitizer photobleaching. Such a mechanism has been proposed by Foster and Nichols (1995) and Georgakoudi et al (1997), who have mathematically modelled and experimentally described the effects of photobleaching on the oxygen distribution in multicellular spheroids. According to this concept, the photosensitizer itself is destroyed via a self-sensitized singlet oxygen reaction with the photosensitizer ground state during illumination in the presence of molecular oxygen. This in turn decreases successive photochemical oxygen depletion and extends the radius of oxygen distribution within the tissue. The extreme, acute oxygen depletion during high-fluence-rate PDT would limit photobleaching, thus no recovery of tumour oxygenation would occur. The impact of fluence rate on photosensitizer photobleaching in vivo is currently under investigation.
Although the fluence rate of treatment did influence tumour oxygenation during PDT, it did not affect tumour pO2 status after PDT. At both high and low fluence rates tumour pO2 dropped steadily with time after treatment (Table 2), consistent with our earlier reports of increasing tumour hypoxia after PDT (Henderson and Fingar, 1987) and studies by others (Roberts et al, 1994;Chen et al, 1996). Consistent with the above observations, no differences were found in tumour perfusion after highor low-fluence-rate PDT. In both cases, tumour perfusion progressively decreased with time after treatment as has been described by others (Roberts et al, 1994;van Geel et al, 1994van Geel et al, , 1996. However, marked differences were found in cutaneous perfusion. The diverse effects on tumour and normal vasculature may be related to differences in vascular architecture, allowing normal and tumour vessels to respond differently to physical and chemical stimuli (Vaupel et al, 1989). Also, photosensitizer distribution in and around normal and tumour vessels may be different, therefore influencing local oxygen depletion during PDT and consequently vascular responses.
Despite similar post-PDT hypoxia, low-PDT fluence rate did enhance tumoricidal effects (Figure 3), as has been observed by others not only for Photofrin but also for other photosensitizers such as ALA-induced protoporphyrin IX, benzoporphyrin derivative (BPD) and tetra (m-hydroxyphenyl) chlorin (mTHPC) (Gibson et al, 1990;linuma et al, 1995;Blant et al, 1996). In the present study not only was the 30 mW cm-2 fluence rate superior to the higher fluence rates at equal total fluence (100 J cm-2), but even lower fluences produced more lasting tumour responses, implying markedly enhanced PDT efficiency.
At least two mechanisms may be responsible for the fluence rate-dependent differences in anti-tumour activity. First, the increased oxygen availability throughout the tumour tissue during illumination at low fluence rate seems to lead to a modest, but statistically significant increase in direct photodynamic tumour cell kill in this tumour model (Sitnik and Henderson, 1997). Similarly, increased fluence rate-dependent cytotoxicity was reported by Foster et al (1993) for multi-cell tumour spheroids in vitro. Second, the severe disruption of the microvascular perfusion of the tumour-surrounding skin at low-fluence-rate treatment may delay the re-supply of oxygen and nutrition to any PDT-surviving tumour cells and so retard tumour regrowth.
The observations reported here may be of significant clinical importance if it can be shown that lower fluence rate does not adversely affect therapeutic ratio. Preclinical studies examining the influence of fluence rate on PDT exposure of normal tissue, especially the skin, are currently underway, as is an extensive clinical study determining tumour pO2 during PDT in patients undergoing treatment of neoplastic skin lesions. ABBREVIATIONS HBSS, Hanks' balanced salt solution; PDT, photodynamic therapy; PDT MODeM, PDT molecular oxygen-depletion model; RIF, radiation-induced fibrosarcoma; RPCI, Roswell Park Cancer Institute.