0031-8655190 $03.00+0.00 Copyright @ 1990 Pergamon Press plc
Photochemistry and Photobiology Val. 52, No. 4, pp. 819-823, 1990
Printed in Great Britain. All rights reserved
OZONE DEPLETION AND INCREASE IN ANNUAL CARCINOGENIC ULTRAVIOLET DOSE GERTKELFKENS*, FRANK R.
DE
GRUIJLand JANC.
VAN DER
LEUN
Institute of Dermatology, State University of Utrecht, P.O. Box 85500, NL-3508 GA Utrecht, The Netherlands (Received 14 March 1990; accepted 23 April 1990)
Abstract-An increase in skin cancer incidence due to an increase of solar ultraviolet (UV) radiation is one of the best quantitated effects of stratospheric ozone depletion. Until now, estimates of effective UV dosages could not be based on spectral data on carcinogenicity. Instead the spectral dependence of sunburn or mutations was used. These data contained little information on longwave ultraviolet radiation (UVA: 315-380 nm). Recently, in hairless mice, experimental data have become available on the carcinogenic effectiveness of the ultraviolet, including UVA. From these new data we can estimate the effect of ozone depletion on the ambient annual carcinogenic UV dose. We find that a 1% decrease in ozone yields a 1.56% increase in annual carcinogenic UV; this value is not strongly dependent on geographical latitude. From this result, combined with the dose-response relationship for UV carcinogenesis, we conclude that for a 1% decrease in total column atmospheric ozone an increase of 2.7% in non-melanoma skin cancer is to be expected.
UV dose, and (b) how is the increase in effective UV related to the ultimate response under consideration (carcinogenesis in our case). Here we will mainly deal with the first question because new relevant data have recently become available. The main problem in calculating an effective carcinogenic UV dose (UV,,.) is the wavelength dependence of the carcinogenic process. This wavelength dependence is usually described by action spectrum €(A), a function giving the relative effectiveness at each wavelength. As a first approximation we assume that the contribution at one wavelength is independent of that at other wavelengths. Then the effective irradiance can be computed by integrating the product of spectral irradiance I ( X ) and action spectrum over the wavelengths.
INTRODUCTION
The “ozone hole” over Antarctica and an increasing concern about environmental pollution in general have given new momentum to research on the effects of stratospheric ozone depletion. One of the best documented consequences of ozone depletion is the expected increase in skin cancer incidence, especially basal and squamous cell carcinoma. The solar ultraviolet-B (UVB, ultraviolet wavelengths between 280 and 315 nm)t, that is effectively attenuated by ozone is the most important factor in the etiology of these carcinomas (Fears et al., 1977; Urbach, 1978). In general, the assessment of the effects of an increased UV load carries two main questions: (a) how do we compute the increase in effective
I,, = SE(A)
*To whom correspondence should be addressed. tAbbreviations: BAF, biological amplification factor (percentage increase in skin cancer incidence for a lo/” increase in effective ultraviolet); BCC, basal cell carcinoma; Cole, action spectrum for photocarcinogenesis developed by Cole, Forbes and Davies in Philadelphia Skin and Cancer Hospital; GCS, Greens analytical representation for the erythema action spectrum published by Coblentz and Stair; IEC, erythema action spectrum published by the International Electrotechnical Committee; NMSC, nonmelanoma skin cancer; OAF, optical amplification factor (percentage increase in effective ultraviolet for a 1% decrease in ozone column thickness); PAR, erythema action spectrum published by Parrish; PTR, action spectrum for mutagenesis in E. coli published by Peak corrected for transmission through a human epidermis of 70 km thickness; SCC, squamous cell carcinoma; STSL, action spectrum for carcinogenesis in mice published by Sterenborg and Slaper; UV, ultraviolet (electromagnetic radiation with wavelengths in between 100 and 380 nm); UVA, ultraviolet-A (315-380 nm); UVB. ultraviolet-B (280-315 nm).
X
I(A) dA.
In the etiology of skin cancer the cumulative effective UV dose is the relevant quantity; this is the effective irradiance integrated over time. If we assume changes in skin dosages to be proportional to changes in annually available effective UV (Leach et al., 1978), the change in annual ambient UV dose is suitable to estimate the effect of ozone depletion. This assumption is not as selfevident as one might expect, but can be made plausible (de Gruijl and van der Leun, 1980). Until recently, for lack of an action spectrum for carcinogenesis the action spectrum for sunburn has often been used to estimate effective carcinogenic UV dose. However, during the last 5 years action spectra for carcinogenesis, in hairless mice SKHhrl (Cole et al., 1986; Sterenborg and van der Leun, 1987; Slaper, 1987) have become available. Although computations based on the sunburn action 819
GERTKELFKENS et a/.
820
spectrum are likely to yield good approximations for the total UV load, it is of course better to base the calculations for the effective carcinogenic dose on these more appropriate action spectra. We present, therefore, analyses of the increase in annual effective UV caused by ozone depletion with the new action spectra and, for comparison, also with the formerly used ones. Results will be expressed in terms of the optical amplification factor (OAF) which gives the ratio between the percentage increase in effective dose and percentage decrease in ozone. MATERIALS AND METHODS
Computations. We developed a computer program to calculate the irradiance at the earth surface in dependence of the thickness of the ozone layer, for wavelengths in between 250 and 400 nm, solar zenith angles ranging from 0" to 90", altitudes up to 5 km, latitudes up to 65" and non-zero surface albedo. The procedure is based on the parameterization of UV fluxes by Schippnick and Green (1982). It starts from the extraterrestrial solar flux, as measured aboard the Nimbus 7 satellite (Schippnick and Green, 1982). Then attenuation by Rayleigh scattering, aerosol scattering, aerosol absorption and ozone absorption is accounted for. This yields, for a given solar zenith angle, the total (direct plus diffuse) irradiance at the earth surface. This irradiance is weighted with the action spectrum. The resulting effective irradiance is integrated over the chosen wavelength range (250-400 nm usually), and tabulated as a function of the zenith angle. Integration over the zenith angles as they occur during a year then produces the effective annual UV dose. Performing the computations for standard ozone conditions and for a 1% depleted ozone layer eventually yields the optical amplification factor. The computer program is written in Turbo-Pascal and operates on an IBM personal computer AT equipped with a mathematical co-processor (80287). The above described method of tabulating the effective irradiance as a function of the zenith angle greatly reduces computation time. However, to compute one OAF at a given latitude execution time is still about 45 min. In the computations presented here zero surface albedo, zero altitude and a standard stratospheric ozone concentration of 3.2 atm.mm are assumed. The latter is a simplification, as the ozone layer shows seasonal and latitudinal changes. Action spectra. The effect of ozone depletion was evaluated for six action spectra:
- The analytic representation which Green et al. (1974) developed for the erythema action spectrum published by Coblentz et al. (1934) (GCS) (Fig. 1). This action spectrum is strongly UVB dominated; wavelengths above 340 nm and below 280 nm are supposed to have zero effectiveness. - The action spectra for erythema as measured by Parrish et al. (1982) at 24 h after exposure (PAR) (Fig. 1). For this action spectrum contribution of the UVA is non-negligible. - The action spectrum proposed by the Commision Internationale de I'Eclairage for erythema evaluation. This action spectrum was constructed from the available erythema action spectra and was adopted by the International Electrotechnical Commitee as an action spectrum for risk to the skin in general in 1988 (IEC) (Fig. 1). - The action spectrum for mutagenicity in E. coli derived by Peak ef al. (1984), corrected for transmission through a 70 pm epidermis (PTR) (Fig. 2). Transmission data used were from Bruls and van
-
-
der Leun (1982). Under the assumption that mutations in the basal layer of the skin are the important events in carcinogenesis in human skin this should be a good action spectrum. The action spectrum for UV-carcinogenesis published by Cole ef al. (1986) (COLE) (Fig. 2). In this report the authors show the action spectrum for mouse edema (at 48 h after exposure) to be appropriate to describe photocarcinogenesis. We extended the action spectrum horizontally above 330 nm at the level of O.OOO1 (relative effectiveness), as suggested as an upper limit in the paper. The action spectrum published by Sterenborg and van der Leun (1987), which has been modified in the UVA by Slaper (1987) (STSL) (Fig. 2). This action spectrum is derived from numerous experiments on albino hairless mice; it is based on the UV dose needed to induce minimal tumors in 50% of the exposed mice. At present it is probably the most direct approximation of the action spectrum for UV-carcinogenesis available.
The action spectra described above are plotted in Figs. 1 (GCS, PAR, IEC) and 2 (COLE, STSL, PTR).
RESULTS
The effects of ozone depletion on the UV dose can be expressed in various ways. Basic information lies in the way the OAF depends on the sun's zenith angle. This zenith angle is determined by: geographical latitude, altitude, date and time of day. For risk assessments it is more useful to express the OAF for the effective annual dose, as a function of the geographical latitude. Results for three representative action spectra (CGS, STSL and IEC) are given in Figs. 3 and 4.
relative eficctivity -1EC
I
I
II
I I I I
250 I I - I - I
300
wav elcngth ( n rn) 350 400
Figure 1. Action spectra used to compute effective UV doses. (- - -) Coblentz-Stair modified according to Green (GCS). (.....) Parrish erythema action spectrum (PAR). (-) IEC action spectrum for risk assessments (IEC).
Ozone depletion and UV dose
82 1
optical amplification factor
relative effectivity
2.0-
-PT R ---COLE STSL
......
1.51
IA
A
A
" * A .
A
1.01
1
wavelength ( n m ) I 350 400
300
Figure 2 . Action spectra used to compute effective UV doses. (---) Cole mouse edema action spectrum (COLE). (.....) Sterenborg-Slaper carcinogenesis action spectrum (STSL). (-) Peaks action spectrum for mutagenesis corrected for transmission through a 70 pm epidermis (PTR). Optical amplification factor
8 A
I
A
8
a
8 A
8 A
* A A
I
01
Zenith angle (degrees) I_r_l-l_T_I-l-
0
20
40
60
"1 0-I
latitude ( d e g r e e s ) l__r_l-1-1-1_1
0
20
60
40
80
Figure 4. Optical amplification factor, based on the integrated annual effective UV (25WOO nm) dose, as a function of the geographical latitude. (M) Coblentz-Stair modified according to Green (GCS). (0)Sterenborg-Slaper carcinogenesis action spectrum (STSL). (A)IEC action spectrum for risk assessments (IEC).
in a higher OAF. For larger zenith angles however, this effect is counteracted by the increasing contribution of the UVA region. As the amount of UVA radiation is not much influenced by ozone depletion, a large relative UVA contribution tends to reduce the OAF (see discussion). The actual zenith angle at which the OAF starts to drop depends on the form of the action spectrum, mainly on the relative effectiveness of the UVA wavelengths. From Fig. 4 we conclude that, for latitudes between 0 and 65", the optical amplification factor is not strongly dependent on the latitude (variation less than 6% relative to the average). For this reason the mean OAF (between 0 and 65"), is a good parameter to describe the effects of ozone depletion. Mean values for the OAF (ranging from 0" to 65", with 5" increment) for all the action spectra under consideration are given in Table 1.
80
Figure 3. Optical amplification factor as a function of the solar zenith angle. (M) Coblentz-Stair modified according to Green (GCS). (0) Sterenborg-Slaper carcinogenesis action spectrum (STSL). (A)IEC action spectrum for risk assessments (IEC).
Figure 3 depicts the dependence of the OAF on zenith angle. For small zenith angles (high solar elevation) the OAF often increases with increasing zenith angle (GCS, STSL). This is due to the fact that for oblique angles the optical path of the radiation through the ozone layer increases. Consequently a 1%decrease in ozone layer thickness will cause a larger increase in effective UVB, resulting
Table 1 Action spectrum
OAF
Range
"% Eff. UVA"
GCS STSL PTR COLE PAR IEC
1.73 1.59 I 56
11.68-1.8 1 J [ 1.58-1.591
12 21 18 21 32
1.54
1.31 1.15
[I .541.60\ [ 1 s(k1.561 (1.25-1.331 [ 1 .O8-1.18]
32
*Computationof the effective UVA contribution for the various action spectra is based on the average annual solar spectrum at 30" Northern latitude.
GERTKELFKENSel al.
822 DISCUSSION
In the early seventies concern about ozone depletion arose for the first time. A lot of effort was spent on predicting the detrimental effects of an increased UV load on the human skin and later on the eye. Especially the question of how much incidence of skin cancer would increase with a possible ozone depletion attracted much attention. However, the action spectra used to estimate the increase in effective UV doses were spectra for the induction of erythema or for the induction of mutations in DNA (Green and Hedinger, 1978). These action spectra had negligible effectiveness in the UVA wavelength region. During the last decade the UVA region gained more and more attention, resulting in action spectra for UV-erythema and UV-carcinogenesis with appreciable contributions in the UVA. The amount of effective UVA influences the OAF, as can be visualized in the following way. Ozone depletion mainly leads to an increase in effective UVB, whereas the effective UVA remains practically unchanged. So the total OAF is determined by a changing UVB and an almost constant UVA fraction. An increasing UVA contribution will thus lower the OAF. This effect can be quantified by: OAF = OAFB + (OAFA - OAFB) * EFFA (2) where OAFA and OAFB are the optical amplification factors for the effective UVA and UVB respectively, and EFF, the fraction of the effectiveness contributed by UVA. As an example, if we extend the Parrish erythema action spectrum (Parrish et al., 1982) horizontally into the UVA at levels ranging from 0.05 to lo-’ (corresponding to relative UVA effectiveness ranging from 94 to 3%), the OAF will range from 0.15 to 1.76. This tendency of a decreasing OAF with increasing effectiveness in the UVA is clearly present in Table 1. From Table 1 it is also clear that the OAFS for the three action spectra most relevant for skin cancer (STSL, PTR and COLE) are very similar: 1.54-1.59. From the viewpoint of carcinogenesis the erythema1 action spectrum, GCS, produces an overestimated OAF, whereas the IEC clearly underestimates the OAF. The latter point is noteworthy, because this action spectrum is often used for risk assessments and safety regulations. In the case of ozone depletion, however, the increase in dose, and accompanying risk is definitely underestimated. For UV monitoring purposes so-called integrating Robertson-Berger meters have been stationed at several locations in the United States and Europe. Our computations based on the spectral sensitivity of this R-B meter yield OAFS of approx. 1.0, which implies little sensitivity to changes in stratospheric ozone. This can, to some extent, explain why the R-B network did not measure an increase in annual effective UV dose, corresponding to the slight decrease in stratospheric ozone measured over the
past decade (Lindley, 1988). However, the most important reason for this discrepancy is probably that, because the monitoring locations are in urban areas, the effect of a small decrease in stratospheric ozone is counteracted by an increase in air pollution, including tropospheric ozone. Moan el al. (1989) produced values around 1.0 for the OAF in northern Europe. For their computations they used a different mathematical model (radiation transport in a “standard atmosphere” divided in 2 km thick homogeneous layers), and, among others, a rather peculiar action spectrum for mutagenesis published by Jones et al. (1987). As a result their “average action spectrum” yields a high UVA contribution, as stated by the authors themselves. This is yet exaggerated by the fact that the ratio UVA/UVB increases for high latitudes. These factors may explain their low OAFS in comparison to the OAFS presented here. To get some insight in the effect of seasonal and latitudinal variations in ozone layer thickness on the OAF we computed OAFS for mean ozone concentrations ranging from 2.5 to 3.5 atm.mm at latitudes 0”, 30” and 60” using the Sterenborg-Slaper action spectrum. These OAFS ranged from 1.48 to 1.62, only a small variation from the value of 1.59 based on the calculations for 3.2 atm.mm ozone. In summary, an overall value for the optical amplification factor of 1.56 for latitudes in between 0 and 65” appears to be the most realistic estimate for skin cancer. To compute the total increase in non-melanoma skin cancer incidence associated with this 1.56% increase in effective UV one has to know the dose-response relationship for carcinogenesis. This dose-response relationship can, in analogy to the OAF, be expressed in terms of the biological amplification factor (BAF) which gives the percentage increase in skin cancer incidence for a 1% increase in effective UV. This BAF depends on the dose-effect relationship for UV carcinogenesis which depends on the type of skin cancer. However, if the BAF is extracted from a combination of epidemiological data on skin cancer and climatological data on ambient UV, then the BAF will also depend on the action spectrum used to compute the effective UV dose (Green and Hedinger, 1978). With the Sterenborg-Slaper action spectrum and the epidemiological data published by Scotto et al. (1981), the BAF for squamous cell carcinoma (SCC) is about 2.5 and the BAF for basal cell carcinoma (BCC) 1.5. A lot of cancer registries do not discriminate between basal and squamous cell carcinoma, but only give data on non-melanoma skin cancer (NMSC) in general. In the United States BCC is about 4 times as abundant as SCC, consequently the BAF for non-melanoma skin cancer (NMSC) is approx. 1.7, i.e. the weighted mean of the contributing BAFs. The total amplification factor equals the product
Ozone depletion and UV dose
of OAF and BAF. So the total amplification factors are 4.0 and 2.4 for SCC and BCC respectively. The overall amplification factor for NMSC becomes 2.7. This means that, on the basis of the most recent data, every 1% decrease of total ozone column will ultimately lead to a 2.7% increase in non-melanoma skin cancer incidence. REFERENCES
Bruls, W. A. G. and 3. C. van der Leun (1982) The use of diffusers in the measurements of transmission of human epidermal layers. Photochem. Photobiol. 36, 709-713. Cohlentz, W. W.. R. Stair and J. M. Hogue (1934) The spectral erythemic reaction of untanned skin to ultraviolet radiation. J . Res. Nat. Bur. Stand. 8, 541-547. Cole, C. A,, P. D. Forbes and R. E. Davies (1986) An action spectrum for photocarcinogenesis. Photochem. Photobid. 43, 275-284. Fears, T. R., J . Scotto and M. A. Schneidermann (1977) Mathematical models of age and ultraviolet effects on the incidence of skin cancer among whites in the United States. A m J. Epidem. 105, 42s427. Green, A . E. S. and R. A. Hedinger (1978) Models relating ultraviolet light and non-melanoma skin cancer incidence. Photochem. Photobiol. 28, 283-291. Green, A. E. S., T. Mo and J. H. Miller (1974) A study of solar erythema radiation doses. Photochem. Photobiol. 20, 473-482. Gruijl, F. R. de and J. C. van der Leun (1980) A dose response model for skin cancer induction by chronic UV exposure of a human population. J . Theor. Bi d . 83, 487-504. Jones, C . A,, E. Huberman, M. L. Cunningham and M. J . Peak (1987) Mutagenesis and cytotoxicity in human epitheleal cells by far- and near-ultraviolet radiations: action spectra. Radiation Res. 110, 244-254.
823
Leach, J. F., V. E. McLeod, A. R. Pingstone, A. Davis and G . H. W. Deane (1978) Measurement of the ultraviolet doses received by office workers. Clin. Exp. Dermatol. 3, 77-79. Lindley, D. (1988) CFCs cause part of the global ozone decline. Nature (Lond.) 323, 293. Moan, J . , A. Dahlback, S. Larsen, T. Henriksen and K. Stamnes (1989) Ozone depletion and its consequences for the fluence of carcinogenic sunlight. Cancer Res. 49, 4247-4250. Parrish, J . A , , K. F. Jaenicke and R. R. Anderson (1982) Erythema and melanogenesis action spectra of normal human skin. Photochem. Photobiol. 36, 187-191. Peak, M. J . , J. G. Peak, M. P. Moehring and R. B. Webb (1984) Ultraviolet action spectra for DNA dimcr induction, lethality and mutagenesis in E. coli with emphasis on the UVB region. Photochem. Photohiol, 40, 613-620. Schippnick, P. F. and A. E. S. Green (1982) Analytical characterization of spectral actinic flux and spectral irradiance in the middle ultraviolet. Phorochem. Photobiol. 35, 89-101. Scotto, J . , T. R. Fears and J . F. Fraumeni (1981) Incidence of non-melanoma skin cancer in the United States. US Department of Health and Human Servies, NIH 82-2433. Slaper, H. (1987) Skin cancer and UV exposure: Investigations on the estimation of risks. Ph.D. Thesis, University of Utrecht, Utrecht, The Netherlands, pp. 147-154. Standardized IEC action spectrum (1988) IEC paper 3352-27. Sterenborg, H . J . C. M. and J . C. van der Leun (1987) Human Exposure to Ultraviolet Radiation Risks and Regulations (Edited by W. F. Passchier and B. F. M. Bosnjakovic), pp. 173-190. Elsevier Science, Amsterdam. Urbach, F. (1978) Evidence and epidemiology of ultraviolet-induced skin cancer in man. In International Conference on Ultraviolet Carcinogenesis (Edited by M. L. Kripke and E. R. Sass), Monograph 50, pp. 5-10. National Cancer Institute, Washington, DC.
Photochemistry and Photobiology Val. 52, No. 4, pp. 819-823, 1990
Printed in Great Britain. All rights reserved
OZONE DEPLETION AND INCREASE IN ANNUAL CARCINOGENIC ULTRAVIOLET DOSE GERTKELFKENS*, FRANK R.
DE
GRUIJLand JANC.
VAN DER
LEUN
Institute of Dermatology, State University of Utrecht, P.O. Box 85500, NL-3508 GA Utrecht, The Netherlands (Received 14 March 1990; accepted 23 April 1990)
Abstract-An increase in skin cancer incidence due to an increase of solar ultraviolet (UV) radiation is one of the best quantitated effects of stratospheric ozone depletion. Until now, estimates of effective UV dosages could not be based on spectral data on carcinogenicity. Instead the spectral dependence of sunburn or mutations was used. These data contained little information on longwave ultraviolet radiation (UVA: 315-380 nm). Recently, in hairless mice, experimental data have become available on the carcinogenic effectiveness of the ultraviolet, including UVA. From these new data we can estimate the effect of ozone depletion on the ambient annual carcinogenic UV dose. We find that a 1% decrease in ozone yields a 1.56% increase in annual carcinogenic UV; this value is not strongly dependent on geographical latitude. From this result, combined with the dose-response relationship for UV carcinogenesis, we conclude that for a 1% decrease in total column atmospheric ozone an increase of 2.7% in non-melanoma skin cancer is to be expected.
UV dose, and (b) how is the increase in effective UV related to the ultimate response under consideration (carcinogenesis in our case). Here we will mainly deal with the first question because new relevant data have recently become available. The main problem in calculating an effective carcinogenic UV dose (UV,,.) is the wavelength dependence of the carcinogenic process. This wavelength dependence is usually described by action spectrum €(A), a function giving the relative effectiveness at each wavelength. As a first approximation we assume that the contribution at one wavelength is independent of that at other wavelengths. Then the effective irradiance can be computed by integrating the product of spectral irradiance I ( X ) and action spectrum over the wavelengths.
INTRODUCTION
The “ozone hole” over Antarctica and an increasing concern about environmental pollution in general have given new momentum to research on the effects of stratospheric ozone depletion. One of the best documented consequences of ozone depletion is the expected increase in skin cancer incidence, especially basal and squamous cell carcinoma. The solar ultraviolet-B (UVB, ultraviolet wavelengths between 280 and 315 nm)t, that is effectively attenuated by ozone is the most important factor in the etiology of these carcinomas (Fears et al., 1977; Urbach, 1978). In general, the assessment of the effects of an increased UV load carries two main questions: (a) how do we compute the increase in effective
I,, = SE(A)
*To whom correspondence should be addressed. tAbbreviations: BAF, biological amplification factor (percentage increase in skin cancer incidence for a lo/” increase in effective ultraviolet); BCC, basal cell carcinoma; Cole, action spectrum for photocarcinogenesis developed by Cole, Forbes and Davies in Philadelphia Skin and Cancer Hospital; GCS, Greens analytical representation for the erythema action spectrum published by Coblentz and Stair; IEC, erythema action spectrum published by the International Electrotechnical Committee; NMSC, nonmelanoma skin cancer; OAF, optical amplification factor (percentage increase in effective ultraviolet for a 1% decrease in ozone column thickness); PAR, erythema action spectrum published by Parrish; PTR, action spectrum for mutagenesis in E. coli published by Peak corrected for transmission through a human epidermis of 70 km thickness; SCC, squamous cell carcinoma; STSL, action spectrum for carcinogenesis in mice published by Sterenborg and Slaper; UV, ultraviolet (electromagnetic radiation with wavelengths in between 100 and 380 nm); UVA, ultraviolet-A (315-380 nm); UVB. ultraviolet-B (280-315 nm).
X
I(A) dA.
In the etiology of skin cancer the cumulative effective UV dose is the relevant quantity; this is the effective irradiance integrated over time. If we assume changes in skin dosages to be proportional to changes in annually available effective UV (Leach et al., 1978), the change in annual ambient UV dose is suitable to estimate the effect of ozone depletion. This assumption is not as selfevident as one might expect, but can be made plausible (de Gruijl and van der Leun, 1980). Until recently, for lack of an action spectrum for carcinogenesis the action spectrum for sunburn has often been used to estimate effective carcinogenic UV dose. However, during the last 5 years action spectra for carcinogenesis, in hairless mice SKHhrl (Cole et al., 1986; Sterenborg and van der Leun, 1987; Slaper, 1987) have become available. Although computations based on the sunburn action 819
GERTKELFKENS et a/.
820
spectrum are likely to yield good approximations for the total UV load, it is of course better to base the calculations for the effective carcinogenic dose on these more appropriate action spectra. We present, therefore, analyses of the increase in annual effective UV caused by ozone depletion with the new action spectra and, for comparison, also with the formerly used ones. Results will be expressed in terms of the optical amplification factor (OAF) which gives the ratio between the percentage increase in effective dose and percentage decrease in ozone. MATERIALS AND METHODS
Computations. We developed a computer program to calculate the irradiance at the earth surface in dependence of the thickness of the ozone layer, for wavelengths in between 250 and 400 nm, solar zenith angles ranging from 0" to 90", altitudes up to 5 km, latitudes up to 65" and non-zero surface albedo. The procedure is based on the parameterization of UV fluxes by Schippnick and Green (1982). It starts from the extraterrestrial solar flux, as measured aboard the Nimbus 7 satellite (Schippnick and Green, 1982). Then attenuation by Rayleigh scattering, aerosol scattering, aerosol absorption and ozone absorption is accounted for. This yields, for a given solar zenith angle, the total (direct plus diffuse) irradiance at the earth surface. This irradiance is weighted with the action spectrum. The resulting effective irradiance is integrated over the chosen wavelength range (250-400 nm usually), and tabulated as a function of the zenith angle. Integration over the zenith angles as they occur during a year then produces the effective annual UV dose. Performing the computations for standard ozone conditions and for a 1% depleted ozone layer eventually yields the optical amplification factor. The computer program is written in Turbo-Pascal and operates on an IBM personal computer AT equipped with a mathematical co-processor (80287). The above described method of tabulating the effective irradiance as a function of the zenith angle greatly reduces computation time. However, to compute one OAF at a given latitude execution time is still about 45 min. In the computations presented here zero surface albedo, zero altitude and a standard stratospheric ozone concentration of 3.2 atm.mm are assumed. The latter is a simplification, as the ozone layer shows seasonal and latitudinal changes. Action spectra. The effect of ozone depletion was evaluated for six action spectra:
- The analytic representation which Green et al. (1974) developed for the erythema action spectrum published by Coblentz et al. (1934) (GCS) (Fig. 1). This action spectrum is strongly UVB dominated; wavelengths above 340 nm and below 280 nm are supposed to have zero effectiveness. - The action spectra for erythema as measured by Parrish et al. (1982) at 24 h after exposure (PAR) (Fig. 1). For this action spectrum contribution of the UVA is non-negligible. - The action spectrum proposed by the Commision Internationale de I'Eclairage for erythema evaluation. This action spectrum was constructed from the available erythema action spectra and was adopted by the International Electrotechnical Commitee as an action spectrum for risk to the skin in general in 1988 (IEC) (Fig. 1). - The action spectrum for mutagenicity in E. coli derived by Peak ef al. (1984), corrected for transmission through a 70 pm epidermis (PTR) (Fig. 2). Transmission data used were from Bruls and van
-
-
der Leun (1982). Under the assumption that mutations in the basal layer of the skin are the important events in carcinogenesis in human skin this should be a good action spectrum. The action spectrum for UV-carcinogenesis published by Cole ef al. (1986) (COLE) (Fig. 2). In this report the authors show the action spectrum for mouse edema (at 48 h after exposure) to be appropriate to describe photocarcinogenesis. We extended the action spectrum horizontally above 330 nm at the level of O.OOO1 (relative effectiveness), as suggested as an upper limit in the paper. The action spectrum published by Sterenborg and van der Leun (1987), which has been modified in the UVA by Slaper (1987) (STSL) (Fig. 2). This action spectrum is derived from numerous experiments on albino hairless mice; it is based on the UV dose needed to induce minimal tumors in 50% of the exposed mice. At present it is probably the most direct approximation of the action spectrum for UV-carcinogenesis available.
The action spectra described above are plotted in Figs. 1 (GCS, PAR, IEC) and 2 (COLE, STSL, PTR).
RESULTS
The effects of ozone depletion on the UV dose can be expressed in various ways. Basic information lies in the way the OAF depends on the sun's zenith angle. This zenith angle is determined by: geographical latitude, altitude, date and time of day. For risk assessments it is more useful to express the OAF for the effective annual dose, as a function of the geographical latitude. Results for three representative action spectra (CGS, STSL and IEC) are given in Figs. 3 and 4.
relative eficctivity -1EC
I
I
II
I I I I
250 I I - I - I
300
wav elcngth ( n rn) 350 400
Figure 1. Action spectra used to compute effective UV doses. (- - -) Coblentz-Stair modified according to Green (GCS). (.....) Parrish erythema action spectrum (PAR). (-) IEC action spectrum for risk assessments (IEC).
Ozone depletion and UV dose
82 1
optical amplification factor
relative effectivity
2.0-
-PT R ---COLE STSL
......
1.51
IA
A
A
" * A .
A
1.01
1
wavelength ( n m ) I 350 400
300
Figure 2 . Action spectra used to compute effective UV doses. (---) Cole mouse edema action spectrum (COLE). (.....) Sterenborg-Slaper carcinogenesis action spectrum (STSL). (-) Peaks action spectrum for mutagenesis corrected for transmission through a 70 pm epidermis (PTR). Optical amplification factor
8 A
I
A
8
a
8 A
8 A
* A A
I
01
Zenith angle (degrees) I_r_l-l_T_I-l-
0
20
40
60
"1 0-I
latitude ( d e g r e e s ) l__r_l-1-1-1_1
0
20
60
40
80
Figure 4. Optical amplification factor, based on the integrated annual effective UV (25WOO nm) dose, as a function of the geographical latitude. (M) Coblentz-Stair modified according to Green (GCS). (0)Sterenborg-Slaper carcinogenesis action spectrum (STSL). (A)IEC action spectrum for risk assessments (IEC).
in a higher OAF. For larger zenith angles however, this effect is counteracted by the increasing contribution of the UVA region. As the amount of UVA radiation is not much influenced by ozone depletion, a large relative UVA contribution tends to reduce the OAF (see discussion). The actual zenith angle at which the OAF starts to drop depends on the form of the action spectrum, mainly on the relative effectiveness of the UVA wavelengths. From Fig. 4 we conclude that, for latitudes between 0 and 65", the optical amplification factor is not strongly dependent on the latitude (variation less than 6% relative to the average). For this reason the mean OAF (between 0 and 65"), is a good parameter to describe the effects of ozone depletion. Mean values for the OAF (ranging from 0" to 65", with 5" increment) for all the action spectra under consideration are given in Table 1.
80
Figure 3. Optical amplification factor as a function of the solar zenith angle. (M) Coblentz-Stair modified according to Green (GCS). (0) Sterenborg-Slaper carcinogenesis action spectrum (STSL). (A)IEC action spectrum for risk assessments (IEC).
Figure 3 depicts the dependence of the OAF on zenith angle. For small zenith angles (high solar elevation) the OAF often increases with increasing zenith angle (GCS, STSL). This is due to the fact that for oblique angles the optical path of the radiation through the ozone layer increases. Consequently a 1%decrease in ozone layer thickness will cause a larger increase in effective UVB, resulting
Table 1 Action spectrum
OAF
Range
"% Eff. UVA"
GCS STSL PTR COLE PAR IEC
1.73 1.59 I 56
11.68-1.8 1 J [ 1.58-1.591
12 21 18 21 32
1.54
1.31 1.15
[I .541.60\ [ 1 s(k1.561 (1.25-1.331 [ 1 .O8-1.18]
32
*Computationof the effective UVA contribution for the various action spectra is based on the average annual solar spectrum at 30" Northern latitude.
GERTKELFKENSel al.
822 DISCUSSION
In the early seventies concern about ozone depletion arose for the first time. A lot of effort was spent on predicting the detrimental effects of an increased UV load on the human skin and later on the eye. Especially the question of how much incidence of skin cancer would increase with a possible ozone depletion attracted much attention. However, the action spectra used to estimate the increase in effective UV doses were spectra for the induction of erythema or for the induction of mutations in DNA (Green and Hedinger, 1978). These action spectra had negligible effectiveness in the UVA wavelength region. During the last decade the UVA region gained more and more attention, resulting in action spectra for UV-erythema and UV-carcinogenesis with appreciable contributions in the UVA. The amount of effective UVA influences the OAF, as can be visualized in the following way. Ozone depletion mainly leads to an increase in effective UVB, whereas the effective UVA remains practically unchanged. So the total OAF is determined by a changing UVB and an almost constant UVA fraction. An increasing UVA contribution will thus lower the OAF. This effect can be quantified by: OAF = OAFB + (OAFA - OAFB) * EFFA (2) where OAFA and OAFB are the optical amplification factors for the effective UVA and UVB respectively, and EFF, the fraction of the effectiveness contributed by UVA. As an example, if we extend the Parrish erythema action spectrum (Parrish et al., 1982) horizontally into the UVA at levels ranging from 0.05 to lo-’ (corresponding to relative UVA effectiveness ranging from 94 to 3%), the OAF will range from 0.15 to 1.76. This tendency of a decreasing OAF with increasing effectiveness in the UVA is clearly present in Table 1. From Table 1 it is also clear that the OAFS for the three action spectra most relevant for skin cancer (STSL, PTR and COLE) are very similar: 1.54-1.59. From the viewpoint of carcinogenesis the erythema1 action spectrum, GCS, produces an overestimated OAF, whereas the IEC clearly underestimates the OAF. The latter point is noteworthy, because this action spectrum is often used for risk assessments and safety regulations. In the case of ozone depletion, however, the increase in dose, and accompanying risk is definitely underestimated. For UV monitoring purposes so-called integrating Robertson-Berger meters have been stationed at several locations in the United States and Europe. Our computations based on the spectral sensitivity of this R-B meter yield OAFS of approx. 1.0, which implies little sensitivity to changes in stratospheric ozone. This can, to some extent, explain why the R-B network did not measure an increase in annual effective UV dose, corresponding to the slight decrease in stratospheric ozone measured over the
past decade (Lindley, 1988). However, the most important reason for this discrepancy is probably that, because the monitoring locations are in urban areas, the effect of a small decrease in stratospheric ozone is counteracted by an increase in air pollution, including tropospheric ozone. Moan el al. (1989) produced values around 1.0 for the OAF in northern Europe. For their computations they used a different mathematical model (radiation transport in a “standard atmosphere” divided in 2 km thick homogeneous layers), and, among others, a rather peculiar action spectrum for mutagenesis published by Jones et al. (1987). As a result their “average action spectrum” yields a high UVA contribution, as stated by the authors themselves. This is yet exaggerated by the fact that the ratio UVA/UVB increases for high latitudes. These factors may explain their low OAFS in comparison to the OAFS presented here. To get some insight in the effect of seasonal and latitudinal variations in ozone layer thickness on the OAF we computed OAFS for mean ozone concentrations ranging from 2.5 to 3.5 atm.mm at latitudes 0”, 30” and 60” using the Sterenborg-Slaper action spectrum. These OAFS ranged from 1.48 to 1.62, only a small variation from the value of 1.59 based on the calculations for 3.2 atm.mm ozone. In summary, an overall value for the optical amplification factor of 1.56 for latitudes in between 0 and 65” appears to be the most realistic estimate for skin cancer. To compute the total increase in non-melanoma skin cancer incidence associated with this 1.56% increase in effective UV one has to know the dose-response relationship for carcinogenesis. This dose-response relationship can, in analogy to the OAF, be expressed in terms of the biological amplification factor (BAF) which gives the percentage increase in skin cancer incidence for a 1% increase in effective UV. This BAF depends on the dose-effect relationship for UV carcinogenesis which depends on the type of skin cancer. However, if the BAF is extracted from a combination of epidemiological data on skin cancer and climatological data on ambient UV, then the BAF will also depend on the action spectrum used to compute the effective UV dose (Green and Hedinger, 1978). With the Sterenborg-Slaper action spectrum and the epidemiological data published by Scotto et al. (1981), the BAF for squamous cell carcinoma (SCC) is about 2.5 and the BAF for basal cell carcinoma (BCC) 1.5. A lot of cancer registries do not discriminate between basal and squamous cell carcinoma, but only give data on non-melanoma skin cancer (NMSC) in general. In the United States BCC is about 4 times as abundant as SCC, consequently the BAF for non-melanoma skin cancer (NMSC) is approx. 1.7, i.e. the weighted mean of the contributing BAFs. The total amplification factor equals the product
Ozone depletion and UV dose
of OAF and BAF. So the total amplification factors are 4.0 and 2.4 for SCC and BCC respectively. The overall amplification factor for NMSC becomes 2.7. This means that, on the basis of the most recent data, every 1% decrease of total ozone column will ultimately lead to a 2.7% increase in non-melanoma skin cancer incidence. REFERENCES
Bruls, W. A. G. and 3. C. van der Leun (1982) The use of diffusers in the measurements of transmission of human epidermal layers. Photochem. Photobiol. 36, 709-713. Cohlentz, W. W.. R. Stair and J. M. Hogue (1934) The spectral erythemic reaction of untanned skin to ultraviolet radiation. J . Res. Nat. Bur. Stand. 8, 541-547. Cole, C. A,, P. D. Forbes and R. E. Davies (1986) An action spectrum for photocarcinogenesis. Photochem. Photobid. 43, 275-284. Fears, T. R., J . Scotto and M. A. Schneidermann (1977) Mathematical models of age and ultraviolet effects on the incidence of skin cancer among whites in the United States. A m J. Epidem. 105, 42s427. Green, A . E. S. and R. A. Hedinger (1978) Models relating ultraviolet light and non-melanoma skin cancer incidence. Photochem. Photobiol. 28, 283-291. Green, A. E. S., T. Mo and J. H. Miller (1974) A study of solar erythema radiation doses. Photochem. Photobiol. 20, 473-482. Gruijl, F. R. de and J. C. van der Leun (1980) A dose response model for skin cancer induction by chronic UV exposure of a human population. J . Theor. Bi d . 83, 487-504. Jones, C . A,, E. Huberman, M. L. Cunningham and M. J . Peak (1987) Mutagenesis and cytotoxicity in human epitheleal cells by far- and near-ultraviolet radiations: action spectra. Radiation Res. 110, 244-254.
823
Leach, J. F., V. E. McLeod, A. R. Pingstone, A. Davis and G . H. W. Deane (1978) Measurement of the ultraviolet doses received by office workers. Clin. Exp. Dermatol. 3, 77-79. Lindley, D. (1988) CFCs cause part of the global ozone decline. Nature (Lond.) 323, 293. Moan, J . , A. Dahlback, S. Larsen, T. Henriksen and K. Stamnes (1989) Ozone depletion and its consequences for the fluence of carcinogenic sunlight. Cancer Res. 49, 4247-4250. Parrish, J . A , , K. F. Jaenicke and R. R. Anderson (1982) Erythema and melanogenesis action spectra of normal human skin. Photochem. Photobiol. 36, 187-191. Peak, M. J . , J. G. Peak, M. P. Moehring and R. B. Webb (1984) Ultraviolet action spectra for DNA dimcr induction, lethality and mutagenesis in E. coli with emphasis on the UVB region. Photochem. Photohiol, 40, 613-620. Schippnick, P. F. and A. E. S. Green (1982) Analytical characterization of spectral actinic flux and spectral irradiance in the middle ultraviolet. Phorochem. Photobiol. 35, 89-101. Scotto, J . , T. R. Fears and J . F. Fraumeni (1981) Incidence of non-melanoma skin cancer in the United States. US Department of Health and Human Servies, NIH 82-2433. Slaper, H. (1987) Skin cancer and UV exposure: Investigations on the estimation of risks. Ph.D. Thesis, University of Utrecht, Utrecht, The Netherlands, pp. 147-154. Standardized IEC action spectrum (1988) IEC paper 3352-27. Sterenborg, H . J . C. M. and J . C. van der Leun (1987) Human Exposure to Ultraviolet Radiation Risks and Regulations (Edited by W. F. Passchier and B. F. M. Bosnjakovic), pp. 173-190. Elsevier Science, Amsterdam. Urbach, F. (1978) Evidence and epidemiology of ultraviolet-induced skin cancer in man. In International Conference on Ultraviolet Carcinogenesis (Edited by M. L. Kripke and E. R. Sass), Monograph 50, pp. 5-10. National Cancer Institute, Washington, DC.
