Cardnogenesfa vol.13 DO. 11 pp.2169-2174, 1992

The influence of ventral UVA exposure on subsequent tumorigenesis in mice by UVA or UVB irradiation

Gert Kelfkens, Frank R.de Grmjl and Jan C.van der Leun Institute of Dermatology, University of Utrecht, Utrecht, The Netherlands

Introduction For decades UV radiation (electromagnetic radiation with wavelengths from 100 to 400 nm) has been known to be an important factor in the induction of skin cancer (1-4). The immune suppressive action of UV radiation, discovered more recently, is probably an important step in this process. Fisher and Kripke (5) showed that UV-induced tumours were rejected upon transplantation to syngeneic mice, but that these tumours grew progressively in UV-exposed hosts. This systemic effect was not due to a generalized immune suppression but resulted from the development of specific T-suppressor lymphocytes (6,7). De Gruijl and van der Leun (8) reported that partial pre-exposure •Abbreviations: UVB, ultraviolet-B (280-315 nm); UVA, ultraviolet-A (315-400 nm). © Oxford University Press

Materials and methods Animals All experiments were performed with albino hairless mice (SKH:HRI) purchased from the Charles River Company, Germany. Male as well as female mice were used in approximately equal numbers. They entered me experiment at ages between 6 and 9 weeks. Maintenance Mice were housed separately in metal cages with 12 compartments. They had free access to tap water and standard mouse chow (RMH-B, Hope Farms, The Netherlands). Room temperature was stabilized at 25 ± 1°C. The room was illuminated in a 12 h day/night cycle with yellow light (Philips TL 40W716). These lamps do not emit any detectable UV radiation. Experimental

design

Two experiments were performed with identical ventral UVA pre-exposures, but with different subsequent dorsal exposure regimens (UVB or UVA). Experiment I. Two groups of 24 mice were randomly selected (with respect to sex, age and weight). In the first group the mice were daily ventrally exposed to UVA radiation for a period of 300 days. During this period the second group was kept under identical conditions, but left unexposed. Immediately following this 300 days, i.e. the pre-exposure period, both groups were subjected to daily dorsal UVB exposures until the end of the experiment (170 days of dorsal UVB exposure). Experiment 11. The ventral exposure regimen was identical to that in experiment I, but after trie first 300 days these m i c e were daily dorsally exposed to UVA radiation for a period of 325 days. One experimental group of 24 mice, kept under Identical conditions, was left unirradiated during the entire experiment, and served as an overall negative control group.

UV irradiation and dosimetry UVA. We used improved efficiency fluorescent UVA tubes (Philips TLK09,40W) as radiation source. To reduce the still significant UVB output, the lamps were filtered with a 10 mm thick window glass filter. The spectrum of the lamps before and after filtering, measured with a radiometer (Optronics, model 742) is plotted in Figure 1. The remnant UVB radiation is negligible for our purposes ( — 0.0001 % of the total UV output).

2169

Downloaded from http://carcin.oxfordjournals.org/ at University of California, San Diego on June 6, 2015

Exposure to ultraviolet-B radiation (UVB: 280-315 nm) can result in a decreased immune response. Tills immune suppression can be restricted to the exposed skin site (local immune suppression) but may also be systemic. To investigate whether ultraviolet-A radiation (UVA: 315-400 nm) could also exert such a systemic effect, we performed the present investigation. The study consisted of two parts. Experiment I: 24 albino hairless mice (SKH:HRI) were ventrally exposed to UVA radiation for 300 days (glass-filtered Philips TLK09 fluorescent tubes, daily dose: 350 kj/m 2 ), while 24 control mice were left unexposed. After this period the control animals were still tumour free, but 60% of the exposed animals had developed abdominal tumours. Subsequently ventral exposures were stopped and both groups were dorsally exposed to identical UVB regimens (Westinghouse FS40, daily dose: 900 J/m2). Experiment II: this was virtually the same as experiment I, but here the mice were dorsally exposed to UVA radiation (glass-filtered Philips TLK09, daily dose: 290 kj/m2) instead of UVB radiation. If we look at all tumours induced dorsally, we find no significant influence of preexposures to UVA radiation. This holds for dorsal UVB as well as for dorsal UVA exposures. In contrast to UVB, however, the UVA radiation induced many papillomas. Excluding the papillomas from the analysis we find that the induction of non-papillomas (mainly squamous cell carcinomas) under dorsal UVA exposure, is slightly enhanced in the ventrally pre-exposed group (difference significant at the P < 0.05 level). This suggests that UVA radiation induced only a weak systemic effect. Ventral UVA pre-exposure did not appear to affect dorsal skin irritation as expressed by scratch marks. The induction period for hyperkeratosis, however, was significantly shortened by the ventral UVA preexposure; this applied to dorsal UVB as well as dorsal UVA exposures.

of mice facilitated the induction of primary tumours on non preexposed sites, indicating a systemic effect of the UV irradiation. Most studies on the effect of UV on the immune system were restricted to ultraviolet-B radiation (UVB*: 280-315 nm). Data on the effect of ultraviolet-A radiation (UVA: 315-400 nm) on immune function are contradictory. In experimental animals neither an increased tumour susceptibility, nor a suppressed contact hypersensitivity could be detected following exposure to moderate UVA doses (9,10). Morison (11), however, found that large UVA doses enhanced susceptibility to transplanted UVBinduced tumours. Recent experiments suggest that immune modulation by UVA radiation may be important for humans, too. The number of suppressor T-lymphocytes increased after solarium exposure (12,13). UVA exposure resulted in a decreased activity of natural killer cells (14). These effects might result in increased tumour susceptibility after UVA exposure. In contrast, Hager et al. (15) stated in a preliminary report that UVA exposure results in enhanced cellular immunity. We performed the present experiments to investigate whether there is a discernible systemic effect of UVA radiation on subsequent UV tumorigenesis. The basic design of the investigation is comparable to the earlier ones (8) in which we measured a systemic effect of UVB radiation.

G.Ketfkens, F.R.de Gruijl and J.C.van der Leun relative spectral energy

Table I. Optimum log-normal parameters n (tm in days) and a for tumour induction; results are given for the four different tumour size categories recorded

1 -,

10"'-

Dorsal exp.

A''

1-

none

M 4.23

(69) a 0.22 ^4.28 (72) 0.29 ^5.36 (213) 0.59 M 5 . 1 8 (177) 0.94

UVA

itr4UVA

none UVA

/

300 325 350 wavefsngth (nm)

375

400

Only lamps with > 100 burning hours were used in order to attain a stable output. UVA irradiances were checked every fortnight with a Waldmann PUVA meter. An electronic dim circuit served to keep irradiance at the desired level. Ventral exposure During ventral exposure the cages were placed in a metal frame — 30 cm above a bank of six glass-filtered UVA tubes (Philips TLK0940W). The radiation reached the abdomen of the animals through the grated floor of the cage. To prevent the animals from climbing the metal grids on top of the cage and thus exposing their backs, these grids were replaced by closed, transparent Lexan sheets. As the mice never roll over while alive, their backs were effectively shielded from the UVA radiation. Only UVA reflected from the sides of the cage compartment and the Lexan top cover could reach the backs of the animals. This stray light level at the back of the animals was never more than 2.5% of the irradiance at the abdomen. Unfortunately, in this setup, the irradiance of the UVA radiation was reduced by mouse excrement falling on the window glass. To counteract this effect, a thin sheet of UVA transmitting plastic was placed over the window glass. Every morning at the start of the exposure this plastic sheet was replaced by a clean one. The variation in the irradiance over the daily exposure period (12 h) was thus limited to 20%. The total daily UV dose at the abdomen of the mice amounted to 350 kJ/m2 (radiation below 400 nm measured with Optronics radiometer model 742). The animals were exposed in this setup 7 days a week, 12 h a day during 300 days. The UV lamps were put on and off in the same cycle as the room illumination. Dorsal exposure UVA. For the dorsal UVA exposure the cages were placed in a fixed position under a bank of six filtered UVA tubes. Animals were exposed daily for 12 h. The daily UV dose on the back of the animals ( - 50% of the dose measured at the top of cage) was 290 kJ/m2 (radiation below 400 nm measured with an Optronics 742). Output was checked every fortnight with a UVA meter (Waldmann PUVA). An electronic dim circuit served to keep the irradiance at a constant level. UVB. For the dorsal UVB exposure the mice were exposed to fluorescent sunlamps (Westinghouse FS40 T12) placed over the cages. A coUimating structure between lamps and cages ensured that all the mice were exposed in the same configuration to radiation that was directed approximately straight downward. Although this coUimating structure results in an optimal irradiation geometry we had to omit h for the UVA exposures because it reduced the irradiance too much. Mice were exposed 7 days a week, 1.25 h a day. The UVB dose measured with a thermopile (Kipp El 1) amounted to 900 J/rrr (at the back of the animals). Output was checked every fortnight with a Robertson-Berger meter, and adjusted if required. For more details on the setup for the dorsal UVB exposure see (16). Animal observation The UVB-exposed animals were checked weekly. During the UVA exposures the animals were checked every fortnight. Apart from tumours all other abnormalities were recorded, such as redness, hyperkeratosis, scratch marks, bad condition of the mouse, etc. Tumours were categorized according to diameter in: all visible tumours, tumours with diameter a 1 mm, tumours with diameter £ 2 mm and tumours > 4 mm in diameter. The fraction of the skin area of the back and the flanks affected by scratching or hyperkeratosis was scored to get

2170

> 4 mm

4.39 0.15 4.39 0.23 5.35 0.52 5.29 0.84

4.60 0.13 4.66 0.16 5.59 0.57 5.42 0.70

4.96 0.21 4.99 0.15 5.86 0.46 5.87 0.52

(81) (81) (210) (199)

(100) (105) (269) (226)

(143) (148) (352) (353)

prevalence (%) 99.

Fig. 1. Spectrum of the Philips TLK09 40W fluorescent tube, before ( - - ) and after ( ) filtering with 10 mm window glass.

> 2 mm

3

9584

75 -

16 iL

r

50-

25% ofthe skin area affected). Definitions and data analysis The first day of exposure was chosen as t = 0. If a tumour was observed for the first time at t = », and the previous checkup was at t = /,_, , then the induction time for this tumour was defined as (r;_, + rJ/2. Induction times for scratching and hyperkeratosis were defined in the same way. Group responses

Downloaded from http://carcin.oxfordjournals.org/ at University of California, San Diego on June 6, 2015

275

TLK09 -gfan fltered TLK09

> 1 mm

Influence of ventral UVA exposure on tumorigentsii were described by the prevalence (for tumours, scratching and hyperkeratosis): the fraction of affected mice. To correct the prevalences for animal deaths we employed an actuarial method described by Kaplan and Meier (17), later adapted to carcinogenesis by Pcto et al. (18). Starting from the tumour and death data this method computes the chance of tumour-free survival. One minus this chance then gives the prevalence in absence of animal deaths. To test whether the differences between the carcinogenic treatments (with or without pre-exposure in our case) were significant we employed a distribution-free method introduced by Peto et al. (18). This method compares the observed numbers of newly tumourbearing animals in each group, at a given checkup time, with the expected numbers if the treatments were alike. Making this comparison for several checkup times accumulates information on whether the hypothesis of no difference in efficacy should be rejected. For a concise quantitative description lognormal and Weibull distributions were fitted to the prevalence data, using a maximum likelihood method (16).

prevalence (%) 99-. 95-

75 50 2516

o

o/o

5-

10

50 100 time(day8)

200

500

prevalence (%) 99-, 95-

84

75-

Results No tumours were observed in the group that was left unexposed throughout the experiment. During the ventral pre-exposure three mice in each group had to be killed because of heavy scratching on the abdomen. At the end of this period 60% (12/21 in the first and 13/21 in the second group) of the animals were tumour bearing. By that time 23 animals survived in both non-pre-exposed groups without skin abnormalities. The subsequent dorsal exposures to UVB or UVA radiation induced tumours in all surviving mice. For all experimental groups the log-normal distribution described the data slightly better than the Weibull distribution. Therefore, we restrict ourselves to the log-normal description. Optimum values for the median tumour induction time (t^), the mean /t = ln^m), and standard deviation a, of the log-normal distribution, for tumours of different sizes induced by the different treatments are given in Table I. Kaplan-Meier plots of the tumour prevalences with the optimum log-normal curves are given in Figures 2 and 3. The prevalences of < 1 mm and < 4 mm diameter tumours under dorsal UVB exposure in the pre-exposed and non-pre-exposed groups are given in Figure 2. Figure 3 gives comparable data for dorsal UVA exposure. From Table I and Figure 2 it can be concluded that the UVA pre-exposure did not affect the induction of tumours by subsequent dorsal UVB irradiation. Comparison of the pre-exposed and the non-pre-exposed groups does not yield a significant difference at the 5 % significance level for any tumour category. Under dorsal UVA exposure, looking indiscriminately at all tumours, there is no significant difference between the UVA preexposed and the non-pre-exposed groups, at the P = 0.05 level. UV irradiation induced two tumour types: papillomas, macroscopically identifiable as exophytic growing tumours, often with a cauliflower-like surface; and non-papillomas (mainly squamous cell carcinomas and precursing actinic keratoses) with more endophytic growth patterns. UVB irradiation, in contrast to UVA irradiation, induced very few papillomas. In an earlier study (19) we found that the papillomas and the non-papillomas follow different induction kinetics, and that separation of these

50

2516 5-

10

50 100 Hmefdays)

200

500

Fig. 3. Kaplan-Meier plot giving the prevalences of tumour bearing mice in the ventrally pre-exposed (O—O) and non-pre-exposed ( A — A ) group, under dorsal UVA exposure, (a) The prevalence for just-observable tumours, (b) The prevalence for tumours of at least 4 mm in diameter.

populations gives additional information. Optimum values for the median tumour induction time (*„,), n = lnfr™), and standard deviation a, of log-normal distributions, fitted separately to the papilloma and non-papilloma prevalence data are given in Table n. Kaplan—Meier plots of the prevalences of ( 0.05). The non-papillomas (squamous cell carcinomas), however, occurred significantly faster in the pre-exposed group (P < 0.05). Under dorsal UVB exposure the standard deviatons in the nonpre-exposed group (0.13-0.22) were slightly larger than those found in a large series of UVB experiments (0.10—0.18) (16). In the groups with dorsal UVA exposures the as found were much larger than for other UVA experiments in our laboratory. Van Weelden et al. (20), and Slaper (21) reported a-values of 0.26 and 0.25 respectively, and our average is a = 0.54 (all tumours, in the non-pre-exposed group). In line with previous observations (19), excluding the papillomas substantially reduced the a to 0.22 (non-papillomas, no ventral pre-exposure), a value compatible with die a-values for UVB radiation in the present experiment. At the end of the ventral UVA exposure the tumour yield (the average number of tumours per survivor) amounted to 0.9 (42 surviving animals, 25 animals bearing a total of 38 tumours). After 100 days of dorsal UVB exposure the tumour yields in the 2171

Downloaded from http://carcin.oxfordjournals.org/ at University of California, San Diego on June 6, 2015

Histology At the end of the experiment, or if a mouse had to be killed for other reasons (heavy scratch marks, bad condition) tumours were sent in for histological examination. As characterization of small tumours is often undetermined (16), only tumours of diameters > 3 mm were sent in. From previous experiments (16) we have extensive histology of UVB-induced tumours. In the present study we focused, therefore, on the tumours induced by UVA radiation. In comparison to UVB radiation the number of tumours induced (yield) by UVA radiation is much smaller. For this reason only a limited number of UVA-induced tumours could be harvested. In total 51 tumours were excised, 13 from ventral UVA exposed skin, 14 from dorsal skin exposed to UVB, and 24 from dorsal skin exposed to UVA.

3

G.KeUkens, F.R.de Gnilj] and J.C.van der Leun

Table n . Optimum log-normal parameters /» (fm in days) and a, for the induction of papillomas and noTHpapillomas under dorsal UVA exposure Ventral exp.

Papillomas

Non-papillomas

None

H 5.53 (251) a 0.71 li 5.43 (228) a 1.14

5.78 (323) 0.22 5.63 (280) 0.50

UVA

Table m . Optimum log-normal parameters n (tm in days) and a for the induction of scratch marks and hyperkeratosis, under the various exposure regimens Dorsal exp.

Ventral exp.

Scratching

Hyperkeratosis

UVB

none

H 4.81 (123) a 0.63 li 5.09 (163) o0.85 li 4.42 (83) a 0.50 ;» 4.10 (64)

6.1 (445) 0.85 4.55 (94) 0.78 5.19(179) 0.69 3.94 (51) 0.47

UVA prevalence (%) 99-.

UVA

none

Q.

UVA

The influence of ventral UVA exposure on subsequent tumorigenesis in mice by UVA or UVB irradiation.

Exposure to ultraviolet-B radiation (UVB: 280-315 nm) can result in a decreased immune response. This immune suppression can be restricted to the expo...
563KB Sizes 0 Downloads 0 Views