Epidemiologic approaches to assessing human cancer risk from consuming aquatic food resources from chemically contaminated water.

Epidemiologic approaches to assessing human cancer risk from consuming fish from contaminated waters must confront the problems of long latency and rarity of the end point (cancer). The latency problem makes determination of diet history more difficult, while the low frequency of cancer as an end point reduces the statistical power of the study. These factors are discussed in relation to the study designs most commonly employed in epidemiology. It is suggested that the use of biomarkers for persistent chemicals may be useful to mitigate the difficulty of determining exposure, while the use of more prevalent and timely end points, such as carcinogen-DNA adducts or oncogene proteins, may make the latency and rarity problems more tractable.

This research group has previously reported on the potentiation of PDT by misonidazole, a bioreductive cytotoxin, in the treatment of Dunning prostate tumours . Subsequent investigations confirmed this finding for both well differentiated and anaplastic tumours and showed a PDT-dependent binding of tritiated misonidazole to both normal and tumour tissues (Hirsch et al., 1987). These findings are consistent with PDT-induced tissue hypoxia and subsequent cell killing by adjuvant bioreductive chemotherapy.
If PDT is to become a modality of therapy for solid tumours, non-invasive assays which can measure the extent of PDT-induced vascular shut-down could play a useful role in its development. We therefore undertook a study to investigate PDT-induced vascular shut-down in both well-perfused and poorly-perfused Dunning prostate tumours using scintigraphy with the radiopharmaceutical 99Technetium hexamethylpropyleneamine oxime (9Tc-HMPAO). 99Tc-HMPAO has been shown by Hammersley et al. (1987) to be a good marker of tumour perfusion in experimental tumours and was used by Rowell et al. (1989) to assess blood flow in human lung tumours. This non-invasive technique for monitoring PDT-induced vascular shut-down may have use in the clinic for estimating the extent and kinetics of perfusion shut-down. These parameters would be important for monitoring early treatment response and for scheduling adjuvant bioreductive chemotherapy targeted at naturallyoccurring and induced hypoxia.

Tumour model
The Dunning prostatic tumour model was initially isolated by Dr W.H. Dunning when it occurred spontaneously in a breeder rat (Dunning, 1963). Since initial isolation, several sublines have been developed and well characterised (Isaacs, 1987). The two sublines used in this study, the R3327-H and R3327-AT, represent the extremes of differentiation. The R3327-H tumour obtained from the Papanicolaou Institute (Miami, FL) is well differentiated, hormonally sensitive, diploid and slow growing with a doubling time of 12-20 days. The R3327-AT tumour obtained from Dr D. Coffey of John Hopkins University (Baltimore, MD), is anaplastic, hormonally insensitive, aneuploid, and rapidly growing with a doubling time of 2-3 days.
The tumours were maintained by serial passage in Fl males from matings of Fischer 344 (female) x Copenhagen 2331 (male) rats. The animals were bred by the University of Alberta Health Sciences Laboratory Animal Services from breeder stock obtained from Harlan Spargue, Dawley Inc (Indianapolis, IN) and were cared for in accordance with the guidelines of the Canadian Council on Animal Care. Donor tissue was selected from tumours that had been monitored for growth characteristics and histological appearance (grade) in order to avoid tumour drift. Using aseptic technique, selected tumour pieces of approximately 1 mm3 were implanted surgically into the left flanks of animals greater than 6 weeks of age, as previously described (Thorndyke et al., 1985). Following grafting, a latency period of 1-2 weeks for the anaplastic tumours and 4-5 months for the well differentiated tumour was observed. Tumour dimensions were measured with calipers in three mutually perpendicular directions while the animals were lightly anesthetised and tumour volumes calculated using the formula V = n/6 x D, x D2 x D3 as described elsewhere (Mador et al., 1982). Tumours were treated at volumes between 2-3 cm3 unless otherwise indicated.
Correspondence: R.B. Moore, Department of Surgery, Cross Cancer Institute, 11560 University Avenue, Edmonton, Alberta, Canada, T6G 1Z2. Received 1 May 1991; and in revised form 9 December 1991. Anesthetics All procedures were approved by the local animal care committee and were performed under general anesthesia. For short-term anesthesia (i.e., tumour measurement and tumour implantation) inhalation Halothane was employed. For anesthesia of intermediate duration, Ketamine (75 mg kg-') and Xylazene (7.5 mg kg-') were administered intraperitoneally.
For the prolonged anesthesia required for PDT and scintigraphy the animals were induced with Ketamine 12 mg kg-' and Xylazene 1.0 mg kg-' i.v. via an indwelling tail vein catheter and maintained with a constant infusion of Ketamine (25mgkg-'h-') and Xylazene (2.0mgkg-'h-'). The animals undergoing prolonged anesthesia were also given Atropine (0.05 mg kg-' i.m.) to dry up pulmonary secretions.

Photosensitiser
Photofrin II (Pll) was obtained from QuadraLogic Technologies (Vancouver, Canada) in frozen isotonic saline (2.5 mg ml-'). One bottle of this drug was thawed and dispensed into 2.5 ml aliquots and stored frozen, protected from light, until the time of use. PII was administered at 15 mg kg-' i.v. 2 h prilor to PDT in all studies except for the experiment which compared the magnitude of perfusion shut-down following light exposure at 2 or 24 h post-photosensitiser administration. In this experiment the new lyophilised Photofrin (MRWO #P89-0089) was used, also obtained from Quadra-Logic Technologies (Vancouver, Canada). This lyophilised preparation of drug was dissolved in 30 ml of sterile 5% dextrose and water to a final concentration of 2.5 mg ml-' and stored in frozen aliquots. It was administered at 7.5 mg kg-' because of its demonstrated increase potency in vitro (Chapman et al., 1991).
Interstitial phototherapy The light source used in these experiments was a Coherent CR-599 argon driven dye laser tuned to 630 nm with a maximum output of 4 Watts. The monochromatic light beam was split into eight beams of equal intensity and directed down 600 micron quartz optical fibres with 1.5 cm cylindrical diffusing tips . Interstitial phototherapy was carried out with seven of these fibres inserted into acetal plastic needles which were implanted into the tumour through a template, in an hexagonal pattern of equilateral triangles, 8 mm apart. A LaserguideTm power meter was used to assess the power output of each fibre and to confirm the wavelength which was set using a SPEXTm monochromator. The 8th fibre was placed in the power meter to continuously monitor the power output. The power output (dose rate) was kept constant at approximately 90 mW fibre and the time of exposure was varied to yield different light-doses as required.
Tumour temperature To prevent against hyperthermic effects, a 27 gauge copperconstain thermal couple (OmegaTm) was inserted into the tumour and temperatures were monitored during phototherapy to ensure that core temperature did not exceed 400C. External cooling with 70% isopropyl alcohol was employed if the intratumour temperature reached 40°C; however, this was a rare problem at a dose rate of 90 mW fibre. Scintigraphy 99"Tc-HMPAO was prepared from a kit (CeretecTm, Amersham International) by adding 5 ml of second elution 99mTc sodium pertechnetate (90 MBq ml-') to a lyophilised preparation of HMPAO in stannous chloride. Labelling efficiency of the lipid soluble complex was assayed by a two phase immiscible partition of 0.9% saline and ethyl acetate (Ballinger et al., 1988). A lipid phase which contained greater than 90% activity was considered acceptable and the preparation used within +h. One ml (90 MBq) of the radio-pharmaceutical was administered i.v. via an indwelling tail vein catheter and the catheter flushed with 1 ml of normal saline. After allowing a minimum of 5 min for distribution, static images were obtained by placing the animal both supine and prone on the the low-energy high-resolution collimator of a General Electric (StarportTm) 400 AC gammacamera. With a 129-151 KeV window, 10 min images were acquired on a Picker PCS 512 computer using a 128 by 128 matrix. Quantitative data on regional activity were obtained from the digitised image using the region of interest (ROI) program. When comparing ROIs, the pixel area was kept constant and the posterior and anterior activities were average to reduce subjective error and error resulting from overlying activity in other organs.

Tomography
In an attempt to interpret local failures following interstitial PDT, regional differences in perfusion of treated and nontreated tumours were measured. Since the resolution of single photon emission computed tomography (SPECT) was close to the dimensions of the tumours, the tumours were excised, solidified in a dry ice and 70% isopropyl alcohol slurry and serially sectioned into 2 mm slices. The slices were then placed, in proper orientation, directly on the low-energy high-resolution collimator of the gamma-camera, magnified x 2 and images acquired for 30 min on a 128 x 128 computer matrix to produce tomograms of the tumour. The computer images were enlarged to produce a pixel size representative of approximately 1 mm.
Isodose plots Light dosimetry in tumour used for tomographic studies were determined and compared with regional perfusion shut-down.
Light intensity measurements were obtained with a miniature light detector placed at known depths within the acetal plastic needles immediately prior to PDT of the tumour. The light intensity measurements for each tumour were interpolated to produce isodose plots in several planes perpendicular to the cylindrical diffusing fibres as described elsewhere (Arnfield et al., 1989).
Experimental protocol Animals were randomised to specific treatments when their tumours reached volumes of between 2-3 cm3. A lateral tail vein was cannulated percutaneously with a 24 gauge (Bax-terTm) teflon catheter. The catheter was capped with a PRNTm heparin lock and the tail was covered with a protective copper-tinned coaxial sheath. Each animal served as its own control by being imaged with 99mTc-HMPAO before and after PDT. There was no manipulation of the tumour (i.e. needle insertion) prior to PDT. Following pretreatment imaging, the tail vein catheter was injected with 0.3 ml of Heparin (100 units ml-') and the rat placed in a metabolic isolation cage. Eight half-lifes (T1 = 6.05 h) of isotope decay were allowed to pass prior to PDT and repeat imaging. Animals were treated with PDT at varying light-doses and imaged 1 h post-treatment, except for the time-dependency studies where the animals were imaged at various times post-PDT after a constant dose of 1,600J. Animals were then followed post-PDT for tumour growth, except for those sacrificed for tomographic studies. Control tumours that received no therapy were also included for comparison of normal growth rates.

Results
This study yielded both qualitative and quantitative data on PDT-induced vascular shut-down. Figure Ia shows a pre-PDT scintigram of a rat bearing the well differentiated R3327-H tumour in its left flank. This tumour is quite well perfused when compared to the essential organs, with a tumour/brain ratio of about 0.7. Figure lb is a post-PDT scintigram of the same animal 8 h following PDT at 1,600 J. There is virtually no perfusion to the tumour and perfusion to the surrounding normal tissue is also diminished. Likewise, Figure 2a is a pre-PDT scintigram of a rat bearing the anaplastic R3327-AT tumour in its left flank. This tumour is less well perfused relative to the R3327-H tumour. Figure 2b is a post-PDT scintigram of the same animal 1 h following PDT at 1,600 J. Again there is a perfusion deficit involving the treated tumour and surrounding normal tissue.
Quantitative data was obtained using the ROI program. Figure 3a and 3b demonstrate ROIs in which activities were measured to estimate perfusion of R3327-AT and R3327-H tumours, respectively. These tumour/brain ratios show that the well differentiated tumour, on average, is 1.77 x better perfused than the anaplastic tumour (Table I). Quantitative analyses of pre-and post-PDT scintigrams demonstrate a light-dose and time-dependent vascular shut-down for both the R3327-H and R3327-AT tumours. Figure 4 shows the light-dose-dependency of vascular shut-down to be similar for both tumours. Data points have been normalised to pretreatment perfusion levels which are significantly different for the two tumours (Table I). Light-dose-dependent vascular shut-down and tumour response in the R3327-AT tumour appear to correlate (Table II).
The temporal relationship of vascular shut-down was studied using both anaplastic and well differentiated tumours. Tomographic studies were performed with anaplastic tumours whose volumes were approximately 4cm3 so that boundaries and regional differences in vascular perfusion could be observed. A control tumour, receiving interstitial light only (no PIT), was included in each study. Both animals were treated with 2,400 J and received an equal dose of the radiopharmaceutical 99mTc-HMPAO (prepared from the same kit) 1 h post-PDT. Figure 6a shows intratumour distributions of radioactivity in a treated tumour (upper two rows) and in a control tumour (lower two rows). In the treated tumour a region of perfused tissue surrounds that adjacent to the illuminators and probably represents the limit of 'effective' 630 nm light exposure. Perfusion was lower in the periphery of the lased tumour compared to the non-lased tumour and this effect would become amplified at longer times, similar to results shown in Figure 5. Therefore a seven fibre icosohedral array with a 8 mm spacing will be limited to treating tumours of 2 -3 cm3 with attenuation and perfusion properties similar to the R3327-AT tumour. Indeed, if 991Tc-HMPAO perfusion can be considered a marker of biologically effective light penetration, then the regional limitation of perfusion shut-down should relate to the isodose plots of this tumour as shown in Figure 6b. The black line forming the ellipsoid around the isodose plots represents the tumour boundary as measured with calipers. The minimum dose required to produce acute vascular shut-down (1 h post-PDT) appears to correspond with the 35 J cm-2 isodose line which is about 3 mm from the surface of each illuminator.
Since these studies were designed to measure vascular damage, we administered the light-dose 2 h following administration of PIT when the drug concentration is maximal in the blood pool as demonstrated by Paramsothy et al. (1989). Clinical phototherapy however, is normally applied 24-72 h after photosensitiser administration in an attempt to exploit maximum tumour/normal tissue ratio of photosensitiser. We therefore, also measured vascular shut-down when phototherapy was applied 24 h following administration of PII. Perfusion shut-down with phototherapy at 2 h was, on average, 35% greater than that observed at 24 h following administration of PII (Table III). This difference in perfusion shut-down was significantly different for the two groups (Student's t-test, 0.02).

Discussion
Tumours with two extremes of tumour differentiation were chosen for these experiments because of their different proliferation kinetics and inherent perfusion which might influence PDT response. Previous experiments had shown that the R3327-AT tumour is more resistant to ionising radiation than its well differentiated counterpart, the R3327-H tumour (Mador et al., 1982;Thorndyke et al., 1985). The increased radioresistance of the R3327-AT tumour was attributed to a 15-25% hypoxic fraction, demonstrated by '4C-Misonidazole labelling. Little or no hypoxia was observed in well differentiated tumours of up to 3 cm3 with this hypoxic marker (Thorndyke et al., 1985).
In this current study, comparison of tumour/brain ratios of the perfusion radiopharmaceutical, 99mTc-HMPAO, revealed that the well differentiated tumour on average is 1.77 x better perfused than the anaplastic tumour. Since PDT is also dependent on oxygen for cell killing (Moan et al., 1985;Mitchell et al., 1985) it might be hypothesised that the anaplastic tumour would be more resistant to PDT. In fact, previous studies by Hirsch et al. (1987) had shown the R3327-H tumours to respond better to PDT than did the R3327-AT tumour at equal photofrin and light-doses. Our measurements of perfusion shut-down have failed to demonstrate a measurable difference in tumour response to PDT.
Both tumour sublines displayed a similar fractional shutdown of perfusion as a function of light-dose and time.
The tomographic studies showed that the penetration of (effective' light required to produce acute vascular shut-down after a total dose of 2,400 J is limited to tumour zones within 3 mm from each illuminator. When compared with the light dosimetry, this zone of perfusion shut-down corresponds to a threshold dose of about 35 J cm-2. The light-doses delivered to the effected tumour tissue boundary ranged from 100 to 35 J cm2 as a consequence of the rapid fall off in light intensity resulting from spacial dilution and tissue attenuation, as well as, the limited resolution of the assay. Other factors which can contribute to the heterogeneity of this biological effect are non-uniform PII distribution and variable tissue oxygenation resulting from regional differences in tumour perfusion. Our  produce permanent vascular collapse with a PII dose of 10 mg kg-'. Hilf et al. (1987) using 31P-NMR andStar et al. (1986) using sandwich observation chambers have reported higher values of 54Jcm-2 and 72 J cm2, respectively, to produce permanent vascular changes. This experimental agreement on 'effective' light-dose for PDT response is surprisingly good when one considers that diverse biological systems, with differing vascularisation, were utilised in the different studies. c Although near complete vascular shut-down at 1 h post-°P DT (2,400 J total dose) was limited to a light penetration of about 3 mm, a significant decrease in perfusion was noted beyond this distance. This could theoretically be the result of lower light fluence or the diffusion of vasoactive substances a) a.   . released by PDT (Fingar et al., 1990;Henderson et al., 1989).
Whatever the cause of this moderate tumour ischemia, it could become a target for bioreductive chemotherapy for potentiation of PDT response Henry et al., 1989). The overall significance of these nuclear medicine studies is that 9'Tc-HMPAO might be usefully employed in clinical situations to monitor the degree and extent of PDT induced vascular shut-doWn using SPECT. Since tumour response is difficult to predict because of intra-and inter-tumour heterogeneity in the distributions of PII, activating light and oxygen, a non-invasive measure of tumour response would assist in prescribing adequate treatment. The time-dependence of vascular shut-down in this study, revealed a maximal effect about 8 h after PDT. An interval of this duration would allow for post-PDT adjuvant chemotherapy prior to maximal shut-down, thus ensuring drug delivery and avoiding interaction with the photo-oxidation process. This interval prior to the perfusion nadir differs from the studies of Wieman et al. (1988) and Tromberg et al. (1990) who report maximal shut-down within 5 min. Star et al. (1986) however, reports a nadir occurring several hours and up to a day post-PDT at light fluence of > 70 J cm-2. The delayed nadir is consistent with cell swelling and interstitial edema resulting in further perfusion impairment. The solid black line surrounding the isodose plots represents the tumour boundary.

Conclusion
Our observations are consistent with other studies investigating vascular effects of PDT using different techniques such as: radiolabelled microspheres (Selman et al., 1984), sandwich observation chambers (Star et al., 1986), NMR spectroscopy (Hilf et al., 1987) and dye exclusion (Fingar et al., 1987). Together these studies demonstrate that infarction is an early, dose-dependent phenomenon that occurs in both welland poorly-differentiated tumours, as well as, in normal tissue and probably accounts for the majority of tumour cell L-.V-L.A _4 K-A-4 6---J r---I I 4-1 I kill observed following in vivo PDT. Our data demonstrate the utility of a non-invasive assay that can provide both morphological and functional information by a clinically acceptable technique. SPECT imaging of 'mTc-HMPAO would be particularly useful for monitoring the degree and volume of PDT induced perfusion shut-down. However, improved photosensitisers which can be activated by longer wavelengths (more penetrating) of light are needed to facilitate treating tumours of clinically detectable volumes.
Financial support to conduct this research was provided by the Alberta Cancer Board, The Northern Alberta Urology Foundation and the Alberta Heritage Foundation for Medical Research. We would like to acknowledge the technical assistance of Mr Bert Meeker and Mr Kevin Small. The skillful assistance of Gina Kennedy, Barbara Haagen, Cindy Johns and Frank LoCicero in preparing this manuscript is appreciated.