Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Chapter 5 DIRECT HUMAN HEALTH HAZARDS SUMMARY On the basis of current knowledge, we believe that ozone depletion and the resultant increase in W would not result in new health hazards, but would increase existing ones as described in the following sections. The W component of sunlight can cause direct damage to the skin, eyes, and immune system of humans. The W wave- lengths most affected by ozone concentration are essen- tially responsible for sunburn, an acute, inflammatory response of the skin. Although the exact targets and mechanisms for sunburn are not fully understood, enough is known about the doses required to predict the increased risk for any given increase in UV flux. For small increases in UV, simple sun avoidance measures would more than offset the increased risk of sunburn. Much less is known about the long-term effects of sunlight on skin. Chronic exposure to sunlight leads to degenerative changes in skin. However, because the effective wavelengths and the relationships between US dose and skin response are not known, the magnitude of the increased risk of degenerative changes that might accompany ozone depletion cannot be predicted. Epidemio- logical studies show that sunlight causes more than 90 percent of basal and squamous cell skin cancers and is a factor in melanoma. Experimental studies and theoretical considerations suggest that actually the wavelengths most affected by ozone (i.e., UV-B) cause basal and squamous cell tumors. Because the dosi~i~etry for humans is uncertain, only crude estimates can be made for the increased risk of these cancers as the result of any given increase in W flux. Techniques to measure individual UV exposures that either cause or prevent these cancers are still lacking. 75
76 Melanomas are undoubtedly related to sunlight, but the relationship is more complex and obscure than the relationship between sunlight and basal and squamous cell skin cancers. Consequently, the melanoma-sunlight relationship is more difficult to measure in epidemio- logical studies and to reproduce in animal experiments. The relationship of melanoma to W -B is even less clear, and there is no animal model in which this relationship can be explored. Current epidemiological data suggest that individual sensitivity to sun damage, exposure to sunlight in childhood, the relationship of childhood nevi to melanoma, and the association between sunlight and specific histological types of melanoma should be explored. In the eye, an acute painful irritation of the cornea, called photokeratitis, is caused by W-B. action spectrum for this effect is known, and the increased hazard for any given increase in UV flux predictable. The symptoms are easily prevented by avoiding or reducing exposure to sunlight. evidence that UV may be involved in the etiology of certain forms of cataracts, but the wavelengths most likely involved (W-A) are not those affected by ozone. Since the last NRC report (NRC 1979a), several new observations have heightened the awareness and broadened the understanding of the health hazards of human exposure to UV. These include a better understanding of the optical properties of skin and blood; the results of careful study of various exposure conditions that influence UV-induced skin cancer in laboratory animals; demonstration and quantification of two forms of DNA repair in viva in human skin (see Chapter 3); and documentation that the immune system of animals and humans is affected by UV irradiation of skin. UV affects the immune system in a variety of potentially important ways. For example, systemic W effects may well be a contributing factor to the efficacy of UV in inducing skin cancer. The is There is some ANATOMICAL AND OPTICAL PROPERTIES OF SKIN AND BLOOD Humans, like most life forms, live in a complicated, dependent relationship with the sun. All life derives its energy from the sun; photosynthesis drives almost all food chains, and the sun is the major source of heat. UV photochemistry in the skin is an obligate step in vitamin
77 D synthesis, and visible light photochemistry within the retina allows vision. On the other hand, the UV component of sunlight can injure or kill cells, including intact living human tissue. The organ most affected by UV is the skin. The optical properties of skin determine the amount of optical radiation reaching various depths in the tissue. Since the NRC (1976a, 1979a) reports, more accurate measurement techniques and useful optical models have made it possible to quantify, predict, and modify the optical properties of skin (Anderson and Parrish 1981, Wan et al. 1981). These advances may make it possible to localize important photobiologic chromophores (molecules or parts of molecules that absorb light), identify mechanisms of UV injury, and better quantify risks. When light enters the skin, a portion is scattered back to the environment, some is absorbed as it reaches various layers, and part is transmitted inward to successive layers of cells, until all the energy of the incident beam has been dissipated (Figure 5.1). The epidermis is a 100-micrometer (pm) sheet of cells that can be viewed as an unpolished optical absorption filter. A 10-pm layer of dead cells, protein, and Remittance Incident Radiation Stratu m Corneu m-~ (10 ,um) Epidermis_ (100 ,um) Dermis (3 mm) Regular Reflectance (=5%)~ Dermal / Epidermal Remittance / Remittance | ~°~,sorn';~_ / 1~; , ;,/ \_/ Blood Vessels FIGURE 5.1 Optical interactions of skin layers with W radiation (Parrish et al. 1978).
78 other biomolecules on the outermost surface of the epidermis is called the stratum corneum. Aromatic amino acids both free and in protein, urocanic acid, nucleic acids, and melanin are the major W-absorbing chromo- phores in the epidermis (Figure 5.2). The dermis is a l-millimeter (mm) to 4-mm layer of primarily collagenous connective tissue that provides much of the structural integrity of the skin. Optical scattering within the dermis largely determines the average pathlength and depth of penetration of various wavelengths of radiation. Dermal scattering is an inverse function of wavelength. The major pigments in the dermis include hemoglobin and bilirubin. In considering the effects of possible changes in the terrestrial solar spectrum resulting from ozone depletion, it is important to know the depths to which optical radiation penetrates human skin (Table 5.l). UV-B is strongly absorbed by the stratum corneum and by many 2or \1 1L 1.6 cn 1.2 A LL J i: ~ 0.8 o 0.4 _ O 200 I\ \ \ 1 \ \ Urocanic Acid - \ / DNA \ 1 `v, ' \ V ~ ~ ~ 3>~\ Dopa Melanin ~] ~ Tryptophan \ Tvrmcin. ~ _. \\\ 1 `_ 1 `~ 220 240 260 280 300 320 340 WAVELENGTH (nary) FIGURE 5.2 Optical absorption spectra of the major W-absorbing chromophores in the epidermis of human skin (concentrations in aqueous solution: urocanic acid 100,u molar, DNA 100 ,ug/ml, dope melanin 15 ,ug/ml, tryptophan 200,u molar, tyrosine 200 ,u molar) (Anderson and Parrish 1982~.
79 TABLE 5.1 Approximate Penetration Depths of Optical Radiation in Fair Caucasian Skin (,um) Depth to Which Following Percentages of Incident Energy Penetrate Wavelength (nary) 507O 37~c 1070 1% UV-C 250 1.4 2 4.6 9.2 280 1 1.5 3.5 7.0 UV-B 300461428 UV-A 3504060140280 Visible 4006090200400 450100150350690 5001602305301,100 6003805501,3002,500 7005207501,7003,500 Infrared 8008301,2002,8005,500 10001,1001,6003,7007,400 12001,5002,2005,10010,000 SOURCE: Modified from Anderson and Parrish (1971). molecules within living epidermal cells. About 50 percent of W -A penetrates fair Caucasian epidermis to be largely attenuated within the first 50 Am of the papillary dermis. The longer visible wavelengths penetrate much further. Biologically active UV reaches the level of cutaneous blood vessels. Endothelial cells and connective tissue elements may be directly affected by the radiation; blood cells, lymphatics, and humoral substances passing through the skin may be photochemically altered. The blood flow to the skin is 30 to 40 times greater than is necessary to supply nutrients and meet the metabolic needs of skin cells because it is primarily designed for heat regulation of the whole body. An equivalent of the entire blood volume may pass through the skin and be irradiated in a few minutes. The physiologic, pathologic, and possible therapeutic implications of this irradiation are just beginning to be understood.
80 EFFECTS OTHER THAN CANCER Acute Responses of Normal Skin to W : Whole Organ Inflammation Many photochemical events are triggered by the absorption of UV and visible light by the variety of molecules within cells (see Chapter 3). Some of these alterations may have little consequence, whereas others may change cell function, cause cell death, or lead to the release of chemicals that affect adjacent cells or tissues. If there is sufficient damage to individual cells, the skin will react as a whole organ. Redness, swelling, heat, and pain appear after a latent period of several hours, and last for hours to days. The overall response of skin to UV is reparative and protective. The tender redness or erythema (commonly called sunburn) is the manifestation of UV-induced inflammation that has received the most attention. The presence and degree of UV-induced delayed erythema depends on the exposure and the wavelengths in the irradiating UV. The reciprocal of the lowest exposure required to induce erythema plotted against wavelength is the action spectrum (Figure 2.1). The 250-nm to 290-nm portion of this curve is the most erythemogenic waveband. Ozone depletion would have little effect on this waveband. Erythemal effectiveness falls by a factor of more than 1000 from 290 nm to 320 nm, the UV-B range. Over a wide range of intensities, both high-intensity radiation for a short time and low-intensity radiation for a long time produce the same response (erythema) as long as the same total dose is given; thus reciprocity holds. Because the action spectrum, dose-response curve, and intensity-time reciprocity relationship for sunburn are known, it is possible to calculate the decrease in time required to acquire a sunburn for any given ozone deple- tion. If UV-B increases by 10 percent, the decrease in time required to acquire a sunburn would be about 10 percent plus a small additional decrease in time because of the spectral shift to include shorter, more effective wavelengths. Erythema, however, is only one component of a complex tissue response. Recent studies (Parrish et al. 1981) have revealed other important components in this response involving a variety of kinds of skin cells, blood vessels, and circulating factors, each having its own thresholds, dose-response curves, and action spectra. Cell injury
81 and alterations of skin can occur without erythema. It has been shown that abnormal differentiation of keratino- cytes, DNA injury, and pigment production can occur at suberythemogenic UV doses. The chromophore, molecular mechanisms, and complex cascade of mediators and events are poorly understood and may vary with wavelength. Long-term Effects of UV on Skin: Solar Degeneration Chronic exposure to the sun causes a complex of changes in skin called actinic or solar degeneration. The skin appears thick and furrowed but may also have zones of thinned epidermis. Hyperpigmentation and hypopigmenta- tion, dilated blood vessels, and a leathery appearance are the other symptoms of the condition sometimes referred to as "sailor's skin'' or "farmer's skin" because excessive occupational exposure often causes these changes, espe- cially on the back of the neck. been called "premature aging," but there is no convincing evidence that the cellular mechanisms and connective tissue alterations are the same as those that occur in natural aging. The action spectrum for solar degenera- tion is not known, and therefore the potential effects of ozone depletion are not known. This condition has also Effects of UV on the Immune System diverse collection of The immune system is a complex and circulating and noncirculating cells in the body that provide protection against certain diseases and infec- tions. The system recognizes foreign molecules or cells and initiates complex reactions to dilute, reject, or counteract them. Recently, it has been discovered that W can alter the immune systems of animals and humans. For example, a mild sunburn results in the decreased viability and function of lymphocytes (circulating white blood cells) in humans for up to 24 hours (Morison et al. 1979), and in animals certain allergic reactions (Morison et al. 1981c), skin graft rejection (Morison et al. 1980), and other immune functions can be altered by giving other- wise tolerable doses of W to intact skin. Most of what is now known about this topic, which is termed photo- immunology, has been learned since the NRC (1979a) report.
82 The immune system is well represented in the skin by Langerhans cells in the epidermis, mast cells and lympho- cytes in the dermis, and other cellular elements perco- lating through the lymphatics and capillaries of the dermis. All these components of the immune system are therefore exposed to environmental light, and may be altered as a result of such exposure. The most detailed studies have been performed on experimental animals. Skin cancers induced in mice by UV-B radiation are highly antigenic, and many are rejected by an immunologic reaction even when transplanted into genetically similar mice, i.e., mice from the same highly inbred strain. These tumors, however, grow in immunosuppressed mice. The tumors that survive in the primary host do so because the UV irradiation has induced systemic, immunologic alterations that suppress specific immune responses. The mechanism, in part, involves the generation of regulatory, thymus-derived (T) suppressor cells in the lymphoid tissue of W-irradiated mice (Fisher and Kripke 1978, Spellman and Daynes 1978). Repeated exposure of mice to UV radiation induces a population of regulatory cells that prevent immunologic rejection of W-induced tumors. This effect has been demonstrated by cell transfer and reconstitution experiments (Fisher and Kripke 1978). The suppression is specific for W -induced tumors. W -irradiated mice also fail to respond to contact sensitizing antigens involved in allergic contact dermatitis and contact hypersensitivity. (The action spectrum for the inhibition of contact hypersensitivity is discussed in Chapter 3.) This represents a second systemic alteration in W-irradiated mice. It is also associated with antigen-specific T suppressor cells and is thought to be caused by a UV-induced alteration in the cells (Langerhans cells or macrophages) that present antigen to lymphocytes in the initiation of an immune response (Greene et al. 1979; Kripke 1980, 1981; Noonan et al. 1981b). Some evidence indicates that W can alter antigen presentation in a way that activates the suppressor cell pathway (Saucer et al. 1980, Towes et al. 1980), and thus affect the immune response. It does this by direct inter- action with antigen-processing cells. Suberythmogenic amounts of radiation are required for altering the function of Langerhans cells, and even the systemic alteration is produced by erythmogenic W exposures. Some of the effects of UV on immunologic pathways could
83 determine whether skin cancer develops or not (Fisher and Kripke 1981). Current efforts are directed toward understanding how UV alters antigen-presenting cells, defining the alteration in these cells that triggers suppressor cell production, and determining which immune responses are affected by the alteration. Several lines of evidence suggest that at least some of the above observations may apply to species other than mice. UV-B exposure can suppress immune responses in guinea pigs and rabbits, and both local and systemic suppression have been reported (Haniszko and Suskind 1963; Morison et al. 1980, 1981c). UV irradiation of guinea pigs, for example, results in a suppression of delayed hypersensitivity responses to contact allergens and injected hap/en-protein conjugates (Jessup et al. 1978. Morison et al. 1981c, Noonan et al. 1981b). In rabbits, the rejection of full-thickness skin grafts is delayed by treatment with oral methoxsalen and W-A radiation (a photochemotherapy for psoriasis) (Morison et al. 1980). The role of T suppressors in these phenomena is being investigated. There are a few reports of local suppression of contact hypersensitivity in human subjects following W exposure (Horowitz et al. 1974, O'Dell et al. 1980). There are increased UV-associated skin cancers in renal transplant patients (Marshall 1974). Although these cannot be ascribed to the immunosuppressive therapy with certainty, the observation is consistent with what would be expected if there were an immunological involve- ment in human photocarcinogenesis. The effects of UV-B radiation on Langerhans cells in human skin appear to be similar to those reported in rodents (Aberer et al. 1981). Human studies in this new area of research are less advanced from the viewpoint of pathophysiologic mechanisms than animal studies, but it has been firmly established that exposure to UV radiation does affect the immune function in humans. In normal human volunteers, single exposures to sunburn-causing doses of UV-B radiation (Morison et al. 1979) or oral methoxsalen and W -A radiation (Morison et al. 1981a) produce an alteration in the distribution and function of subpopulations of circulating lymphocytes. These effects are reversible within 48 to 72 hours. However, repeated exposure to such radiation may cause more long-lasting changes in lymphocyte viability and function (Morison et al. 1981b). The quantitative implications of the above observa- tions are not clear. It is possible that UV-induced
84 alterations of immune function are important in mediating the beneficial effects of W radiation in the treatment of skin disease and the harmful effects of such exposure, such as the development of skin cancer. Effects of W on the Eye Because solar W radiation is present during most of th daylight hours, the eye is exposed daily to some amount of solar W radiation throughout life. W-B is mostly absorbed within the cornea and conjunctive. The acute effects of excessive exposure to these wavelengths are primarily conjunctivitis and a corneal inflammation reaction known as photokeratitis. After W exposure, there is a period of latency varying more or less inversely with the amount of exposure. The latent period may be as short as 30 minutes or as long as 24 hours, but it is typically 6 to 16 hours. Photoconjunctivokeratitis causes the sensation of a foreign body or sand in the eye, varying degrees of excessive tearing, blinking, and intolerance of light. Corneal pain can be very severe, and the individual may be incapacitated for a period of time. These acute symptoms usually last from 6 to 24 hours, and almost all discomfort usually disappears within 48 hours. Rarely does exposure result in permanent damage. Unlike the skin, the ocular system does not develop tolerance or significant defenses against future W exposures. Anatomic conformations protect human eyes from acute overexposure to the W component of sunlight as do physiologic bright-light- avoidance responses when there is sufficient visible light to incite this protective response. There is epidemiological evidence that chronic exposure to sunlight may cause certain types of cataracts in humans and experimental evidence that W-induced photochemical changes (with and without the addition of exogenous photosensitizers) in the lens can cause cataracts. The action spectrum for these changes, however, appears to be in the W-A range and therefore would not be affected by ozone depletion.
85 CANCER EFFECTS Basal and Squamous Cell Skin Cancers What Was Known by 1979 Basal and squamous cell skin cancers constitute the most common malignancies in humans. These cancers are usually easily treatable but have a definite morbidity, cost, inconvenience, cosmetic liability, and mortality. The NRC Climatic Impact Committee (NRC 1975) and the Committee on Impacts of Stratospheric Change (NRC 1979a) were asked to predict whether increased exposure to W-B would be likely to increase basal and squamous cell skin cancer incidence rates. To do so, they had to make judgments on the basis of limited evidence from epidemiological studies, backed up by clinical and pathological observa- tions and the results of animal experiments, because human experimentation was out of the question. The same approach had to be used in 1964 to assess the relationship between cigarette smoking and lung cancer. The Surgeon General's report on smoking and health stated that deter- mination of whether the confirmed association between an event and a disease is causal is a matter of judgment that goes far beyond any statement of statistical proba- bility. It listed a number of criteria that must be used in assessing circumstantial evidence. These criteria included the consistency, strength, specificity, temporal relationship, and coherence of the association (Surgeon General 1964). Over the years, collective evidence has confirmed the existence of an association between basal and squamous cell skin cancers and sunlight. Epidemiological surveys have consistently identified an overwhelming predominance of these cancers in Caucasians, increasing mortality and incidence rates with decreasing latitude, higher rates of disease in outdoor than in indoor workers, and rates of disease increasing more rapidly at older ages. These associations have always been relatively strong and have been accepted as biologically rational: the skin is indeed exposed to the sun, and increased incidence rates with advancing age favor a sequence in which disease does not precede but follows exposure. The supporting clinical data have also been consistent in showing concentrations of these skin cancers among fair-complexioned individuals, particularly those with blue eyes who sunburn easily and have Celtic ancestors.
86 These clinical series have recorded another consistent finding, namely that most basal and squamous cell skin cancers occur on sites of the body habitually exposed to sunlight and often in the same tissue systems as sunburn. This overall picture was considerably strengthened by the finding that high rates of basal and squamous cell skin cancers are associated with defective DNA repair in patients with the inherited disease xeroderma pigmentosum (Kraemer 1980). This finding made a relationship between UV-B radiation and these cancers biologically plausible, on the basis of the knowledge that W-B can damage DNA in skin cells (see Chapter 3). There were very few hard data on the incidence of and mortality from skin cancers when the 1976 and 1979 reports of the Committee on Impacts of Stratospheric Change (NRC 1976a, 1979a) were prepared. The most solid evidence came from standardized measurements of skin cancer incidence (1971-1972 National Cancer Institute (NCI) survey) and of annual UV doses at four geographic locations in the United States with a range of W exposures (NRC 1975, Parrish et al. 1978, Scott and Straf 1977, Scotto et al. 1974). The estimated annual UV dose at ground level was made in two ways: (1) by readings of Robertson-Berger (R-B) meters, and (2) by calculation from known solar fluxes, ozone concentrations, and estimated cloud cover (NRC 1975). The first method of estimating is supposed to measure the accumulated dose of wavelengths in the erythema action spectrum, although from Figure 2.2 it can be seen that the R-B meter measures more W-A than is in the erythema spectrum. The second method makes a theo- retical estimate of dose corresponding to wavelengths included in the DNA action spectrum. It is obvious from the plots of the R-B response to various wavelengths of UV and the action of UV on DNA as a function of wavelength (Figure 2.2) that the two dose measurements are not the same. A change in ozone concentration will change the annual "DNA dose" much more than the "R-B dose" (see NRC 1979a, Figure D.5 and Table D.1). A 1 percent change in ozone will produce, at 40° north latitude, an approximate 2.3 percent change in the DNA-damaging dose, but only an approximate 0.8 percent change in the R-B dose (NRC 1979a). Both measures of UV dose are strong functions of latitude, increasing markedly as latitude decreases. The measurements of incidence and W exposure mentioned above were used to show that the reported statistical correla- tion between latitude and skin cancer incidence could, in
87 fact, be a correlation between the occurrence of skin cancer and the local annual exposure to UV-B. The measurements were used to make crude estimates of the increases in skin cancer incidence rate to be expected from various percentage decreases in ozone (NRC 1975, Scott and Straf 1977) or from the percentage increases in units of exposure as measured by an R-B meter (Scotto et al. 1974). With such limited, epidemiological findings, more weight than usual was given to clinical observations from both published and unpublished sources. The panels preparing the human health sections of the NRC (1976a) and (1979a) reports believed that death rates from skin cancer other than melanoma, based on this weighted evidence, were too low to have national significance, even though the cost of treatment and the disfigurement and morbidity resulting from these types of cancers were considerable. NRC (1979a) provided preliminary observations from ongoing NCI programs to relate UV (and by inference UV-B) to the incidence of basal and squamous cell skin cancers. The report provided a brief summary of the survey methods that are described in detail in NRC (1975). The informa- tion from the preliminary results available in 1979 confirmed earlier observations that Caucasians living in areas of high insolation do have higher rates of basal and squamous cell skin cancer than those living in areas of low insolation and that fair-skinned Caucasians, particularly those who sunburn easily or have limited ability to tan, are at measurably increased risk. The report concluded that in the United States, most basal and squamous cell cancer is found in these people. It also acknowledged that there are signs of an upward trend in the incidence rate or these cancers. Advances in Knowledge Since the NRC (1979a) report, new information from animal experiments has shed more light on the dose rate and the quantitative relation Experimental Photocarcinogenesis. ship between skin cancers other than melanoma and the biologically effective W wavelengths. Skin cancer has been induced in experimental laboratory animals by exposing them to wavelengths shorter than 320 nm. The UV wavebands most affected by alterations in ozone concen- trations (W -B) are carcinogenic in animal studies. Using
88 a xenon arc to represent extraterrestrial sunlight and a series of filters to simulate various atmospheric ozone concentrations, Forbes et al. (1980) showed that incre- mental additions of shorter wavelength W increased carcinogenic effects in hairless mice. Other factors have been studied. Freeman and Knox (1964) found that increased temperature at the time of U] exposure accelerated tumor production. Using environ- mental chambers to irradiate experimental animals, Owens et al. (1974) found that animals exposed to W and wind developed more tumors than animals receiving the same dose of W alone. In other animal groups, Owens et al. (1975) noted that animals maintained at high humidity developed tumors more rapidly than those maintained at low humidity. Mice are not perfect models for human photocarcino- genesis. The susceptibility to W -induced cancer varies with the strain. All known strains nave poor excision repair of DNA compared with people (see Chapter 3), and the optical properties of mouse skin differ from those of human skin. But many important models and concepts have resulted from the decades of work accumulated by using mouse models. Recent and ongoing studies have shown tha. the relationship between cumulative dose and cancer production is not simple. Under certain experimental conditions, the tumor yield can be increased by alteration of the exposure regime. The same total dose given at lower irradiance produces more tumors. Intermittent exposure may be more photocarcinogenic than the same dose given continuously, although the opposite holds for rats (Strickland et al. 1979). The susceptibility to UV-induced cancer seems to vary with age of the animal. W also alters the immune system of animals (see the discussion earlier in this chapter), which markedly influences the susceptibility to W-induced skin cancer All of these observations may eventually help us to better understand, predict, and make models of human photocarcinogenesis. Chemicals and W can interact in a variety of ways to affect tumor yield in skin. Either enhancement or inhibition of photocarcinogenesis may occur depending on the chemical carcinogen and the wavelength and dose of radiation used. Certain chemicals can promote but cannot initiate tumors. Antioxidants have been shown to either increase or decrease tumor yield under certain conditions (Forbes et al. 1981). Skin cancer can result from the interacting effects of a chemical and W irradiation at
89 doses at which neither agent alone is a primary carcinogen (Black et al. 1978). The combination of certain psoralens and W-A is an example. Not all chemicals that enhance photobiologic effects on cells or tissue are photocarcino- genic agents, but nevertheless as chemical pollution of our environment grows, chemical enhancement of photo . (The state of knowledge about UV interactions with other environ- mental stresses on ecosystems is discussed in Chapter 4.) carcinogenesis may be an increasing concern Epidemiology. This committee was given access to new epidemiological data, as well as to extended analysis of updated existing data. We were also provided with revised estimates of predicted increases in the incidence rates of skin cancer for a range of possible percentage reductions in stratospheric ozone concentration and/or changes in R-B meter units. The new information made available to us by the NCI came from three sets of data. The first consisted of counts of cases of newly diagnosed (nonrecurring) skin cancers from June 1, 1977, through May 31, 1978, among residents of Atlanta (Standard Metropolitan Statistical Area, SMSA), Georgia; Detroit (SMSA), Michigan; Minneapolis/St. Paul (SMSA), Minnesota; New Mexico; New Orleans (metropolitan area), Louisiana; San Francisco/Oakland (SMSA), California; King County, Washington (Seattle); and Utah. These eight geographic locations were chosen because they receive various intensities of solar radiation and most are participants in the Surveillance, Epidemiology, and End Results (SEER) program of the NCI. The second set of new data consisted of preliminary R-B meter measurements from meters installed in 1978 (Berger and Urbach 1982), which were used to estimate the W dose accumulated over one year in five geographic areas (1978 in Figure 5.3). When taken together with measurements from the meters installed in 1974, estimates of the annual accumulated W dose for all of the eight locations with new incidence data were available. The third set of data came from a telephone interview survey of patients with skin cancer and of general population controls at the same eight locations. The questionnaire was designed to obtain information on several host and environmental factors that may affect the risk of developing skin cancer. Descriptions of the questionnaire and the survey itself are in the literature (Scotto and Fraumeni, Jr. 1982). However, in the analyses by Scotto and co-workers at NCI of the relation between the incidence data and UV accumulated dose (as
go 350 300 X 250 LL ha he O 200 - ~: ~ 150 lo: O 100 (in 50 · Mauna Loa ~ 1 974 Esti mates · 1978 Estimates · El Paso New Orleans ~ Tal lahassee ~ Fort Worth ~ Albuquerque Atlanta S · Oakland Salt Lake City e Des Moines ~ Philadelphia ~ ~ Detroit Minneapolis 1 1 Bismarck e Seattle 15 20 25 30 35 40 DEG R EES NO RTH LATITUD E 45 50 FIGURE 5.3 Annual W measurements by latitude, 1974 and 1978. The W radi- ation index is total Robertson-lBerger meter counts over a one-year period multiplied by 10 - (preliminary monthly averages provided by Daniel Berger of Temple Uni- versity for the 1978 estimates). The meters read W-B between 290 rim and 320 nm, as well as some W-A. (Modified from Scotto et al. (1982~.) measured by an R-B meter), the incidence rates were not corrected for any confounding factors other than age and sex. The incidence data from the eight geographic locations provided age- and sex-specific rates for basal and squamous cell cancers, as well as rates of occurrence on different sites of the body. The UV accumulated dose (as measured by an R-B meter) clearly correlated with latitude (Scotto et al. 1982) (Figure 5.3). When age-adjusted rates were plotted against latitude, the incidence-of
91 skin cancer was inversely associated with latitude, whereas the incidence for all other cancers combined was not (Scotto et al. 1982) (Figure 5.4). In the southern part of the United States, the annual rates of skin cancer other than melanoma far exceeded the total annual rate for all other cancers. More detailed analysis showed that basal and squamous cell skin cancers were reported at earlier ages in the South. When the new annual sex- and age-adjusted incidence rates of these skin cancers were plotted against R-B meter measurements of UV, all incidence rates were found to be lower in 500 400 ~ - ~ ~ LIJ cr UJ 300 a o - J o ° 200 to to to LLJ LIJ 6 150 cr 100 30 34 · Skin Cancers Other Than Melanoma (1977-1978) All Other Cancers (1973-1976) ~3 O __ W __ lo____ O \ \ O \ O _ 0 cat ._ _ x ~_ \ c _ ~cot Z 0 _ _ 0 .R ~.m 0 t5 4 - LL J O C c ~a,, ~ c a) C A) . _ a O C 5 - ~_ al 4_ _ z ~ 6 1 1 1 1 - o 38 42 46 50 DEGREES NORTH LATITUDE FIGURE 5.4 Annual age-adjusted incidence rates (1970 U.S. standard) for basal and squamous cell skin cancer (1977-1978) and all other cancers (1973-1976) by latitude in the U.S. white population (Scotto et al. 19824.
92 locations with the lowest annual W dose (Scotto et al. 1981) (Figure 5.5). This was true for both basal and squamous cell cancers. The slope representing the correlation between accumulated W dose and incidence was steeper for squamous cell than for basal cell cancers for both sexes (Figure 5.5). The 1977-1978 incidence rates of both basal and squamous cell cancers increased steadily from younger to older age groups for both sexes, except for some leveling-off at extreme old age. These age-specific rates confirm repeated past observations that older people get more skin cancers than younger people. The higher frequency of incidence for older people is true in cohort as well as in cross-sectional analyses. In other words, the higher rate in older people appears to be a correct finding. The existence of these consistently higher rates of disease at advanced ages is believed to mean that the likelihood of skin cancer increases with accumulated W exposure (Fears et al. 1977). Mortality data have suggested that the incidence of the more invasive and lethal squamous cell cancers may increase with age more rapidly than the more common and less malignant basal cell type. Although only one out of five new cases of skin cancer other than melanoma is of the squamous cell type, squamous cell cancers cause four out of five nonmelanoma skin cancer deaths. As expected, the highest incidence rates of both basal and squamous cell cancers were for those on exposed areas of the face, head, and neck in both men and women (Figure 5.6). Overall, about 80 percent of the cancers began on the head and neck, 10 percent on the arms and hands, 6 to 7 percent on the trunk, and 2 to 4 percent on the legs and feet (Table 5.2). The very limited comparisons that the NCI staff could make between the 1971-1972 and the 1977-1978 incidence rates suggested to them a 15 percent to 20 percent increase in the number of basal cell cancers for both sexes (Scotto and Fraumeni, Jr. 1982) (Table S.3, Figure 5.5). Most of the additional tumors were found on the trunks (back) of males. While there was limited evidence of an overall increase in squamous cell cancers, there was a definite increase in the rate of squamous cell cancers on women's arms and hands. An association between skin cancer other than melanoma and W exposure in non-Caucasians had not been found until recently. The 1977-1978 survey found 68 black patients with basal cell or squamous cell skin cancers. In spite
93 100 >_ 50 CC 111 At o ~20 J o C o Lo O" 8 LL us a: c: ~ 500 CC fir As: ~ 200 u, o rid a, 1 00 us 50 a: LL a: 20 10 500 _ 200 _ ·~ -O. White Females 1 977-1 978 _ ~ 1971 -1972 1 977~1 978 Basal Cell _~ 1971-1972 ,~ 1 1 1 1 1 1 1 1 - - _~ Squamous Cell - 100 120 140 160 White Males Q71 1q79 . _ o - C ~ O ~ C , 4,, 1 1 1 1 1 1 1 1 1 100 120 140 160 180 ~ Squamous Cel I a . _ ~ c c 5, _ ~ ~ 1 80 200 - cr ~ - J3 c J 200 SOLAR UV RADIATION INDEX NOTE: For each sex and cell type, the model used to fit the data is IN incidence ~ ~ + Off, where F is the flux in Annual R-B meter counts x 10 4. FIGURE 5.5 Annual age-adjusted incidence rates for basal and squamous cell cancers among white females and males for two surveys, 1977-1978 (closed symbols) and 1971-1972 (open symbols), according to one year's W measurements at selected areas of the United States. The W radiation index is the total Robertson-Berger meter counts over a one year period multiplied by 10-4 . The meters read W-B between 290 rim and 320 nm, as well as some W-A. (Adapted from Scotto et al. (1981~.)
94 300 cry 250 6 UJ :~ 200 LL cat to to O 150 A ~ 100 LU At: o 100 <~ 80 ~60 to to to to A 40 LO ~ 20 Basal Cell Carcinoma Face, Head, Trunk and Neck C Male Femal e Total (all sites) 246.6 0 150.1 '~ ~2.5 4.0 1.5 1.2 13.6 Upper Lower Other Multiple Extremities Extremities Sites Sites Squamous Cell Carcinoma 46.1 Male Female Total (all sites) 65.4 1 3.7 O ~ 1 ~ Face, Head, Trunk Upper Lower Other and Neck Extremities Extremities Sites FIGURE 5.6 Annual age-adjusted incidence rates (1970 U.S. Standard) of basal and squamous cell cancers according to anatomic site and sex (U.S. white population, 1977-1978) (Scotto and Fraumeni, Jr. 1982~. 0.S 1~3 o.g 0.7 23.6 3.6 _~:: Multiple Sites
95 TABLE 5.2 Percentages of Basal and Squamous Cell Skin Cancers by Sex and Anatomic Site Among U.S. Whites, 1977-1978 Males Females Face, head, and neck 80 Upper extremities Trunk Lower extremitiesa Total 81 9 6 4 100100 11 7 2 aIncludes a small number of genital and unspecified tumors. SOURCE: NCI survey 1977-1978. TABLE 5.3 Annual Age-Adjusted Incidence Rates 0?er 100,000) for Basal and Squamous Cell Skin Cancers Among U.S. Whites by Cell Type and Sex, 1971-1972 and 1977-1978 1971-1972 NCI Survey 1977-1978 NCI Survey All Survey Areas Basal cell cancers Male 202.1246.6 Female 115.8150.1 Squamous cell cancers Male 65.565.4 Female 21.823.6 San Francisco-Oakland Basal cell cancers Male 197.9239.0 Female 117.2145.1 Squamous cell cancers Male 51.756.3 Female 15.818.4 Minneapolis-St. Paul Basal cell cancers Male 165.0213.1 Female 102.8144.0 Squamous cell cancers Male 36.536.6 Female 12.311.8 NOTE: The 1971-1972 survey included four areas, and the 1977-1978 survey included eight. However, only two locations, San Francisco-Oakland and Minneapolis-St. Paul, were common to both surveys. SOURCE: Scotto and Fraumeni, Jr. (1982).
96 of the small number of cases, the Scotto and Fraumeni, Jr. (1982) report a suggested latitude gradient among blacks for both cell types. Squamous cell cancers were more common than basal cell. In summary, the new information provided by analysis of the NCI population-based incidence data (corrected only for age and sex) confirmed and strengthened existing evidence in favor of a causal relationship between basal and squamous cell skin cancers and W. It also gave new insight into the magnitude of this problem. Basal and squamous cell skin cancers from W radiation have become so common that although the fatality rate is one death in every hundred cases, the overall national mortality figures actually resemble those for melanoma (Mason et al. 1975). The long-assumed protective effect of skin pigmentation was confirmed by the very small number of cases reported in blacks, and the suggested association (latitude gradient) of those cases with W exposure may be an important piece of confirmatory evidence. It was hoped that the data from the questionnaire survey could be used to assess the relative personal risk to individuals of developing basal and squamous cell cancers from their reported susceptibility character- istics, e.g., skin and eye color and ancestry. It was further hoped that, in this way, enough information could be gained for estimates to be made of the total amount of disease in a given community that could be attributed to individual susceptibility. This work has not yet been done. Early results of the questionnaire survey confirmed previous clinical observations that among people with skin cancer there is a higher occurrence of fair complexions, blue eyes, red or blond hair, and Scottish or Irish ancestors than among the general population. They also confirmed suspicions that differences in the frequencies of these particular characteristics in case-and-comparison groups were sufficiently variable across geographic areas to affect overall measures of the correlations between incidence and W exposure. However, this survey points out that at least three out of every ten individuals diagnosed as having basal or squamous cell skin cancers do not have fair complexions and that fewer than 50 percent have blue eyes. In other words, basal and squamous cell skin cancers are not confined to the fair-skinned, blue-eyed descendants of Scottish-Irish immigrants. This is important to remember in directing research into causes of skin cancer and in planning
97 measures for its control. For example, the most impor- tant marker may be a measure of sunburn response and the ability to tan. There are no studies collecting and measuring the total UV dose that an individual is naturally exposed to in a lifetime and correlating it with the subsequent incidence of skin cancer. Prospective studies needed to detect initiators of skin cancer in humans may be too . ; =] tar -m cone; for Thousands of participants would have to be monitored for decades. It may, however, be feasible to design studies of the promoting effects of US in human skin cancer other than melanoma. Studies of cohorts (groups born in a specific time interval and followed through life) of patients with psoriasis undergoing two different types of US phototherapy are contributing indirect but valuable information about skin cancer mechanisms. Groups of patients with psoriasis are being treated with photo- therapy using high-intensity W -B sources, mainly in the 290-nm to 300-nm part of the spectrum. This therapy would be expected to damage DNA in the same way as solar radiation of the same wavelengths. Although this form of phototherapy has been widely used for decades, it has only recently been used in high doses. Consequently, prospective studies of the long-term toxicity or skin cancer risk are just beginning. One retrospective study suggests that psoriatic patients who have had massive cumulative doses of UV-B and topical crude coal tar have -, . ~ ~ _ _ ~ ~ an increased incidence of skin cancer (Stern et al. 1982). Information is also becoming available from a relatively large cohort of psoriasis patients in 16 medical centers treated with photochemotherapy (oral methoxsalen and UV-A). Although this treatment damages DNA in a different way, the study raises some important questions about dose-response relationships, synergism, and cocarcinogenesis. For example, after four years of follow-up, rates of basal and squamous cell skin cancer in 1373 psoriasis patients treated with photochemotherapy using W -A (P WA) were 3 times higher than were expected on the basis of age, sex, and geographic-location incidence rate data. The proportion of squamous cell cancers was much-higher than average, and the additional cancers were mostly on parts of the body that are not normally exposed to sunlight but were exposed to photo- chemotherapy. A history of exposure to ionizing radiation greatly increased the risk of developing (particularly) squamous cell cancer (Stern et al. 1981). When measured
98 in treatment units, much more W-A exposure was needed to initiate squamous cell than basal cell cancers. Data from the four locations in the 1972 incidence survey were added to data from the eight 1978 survey locations (which gave 12 measurements from 10 locations). Scotto et al. (1981) plotted the age-adjusted incidence rates for basal and squamous cell cancers on a logarithmic scale against the annual W dose measured by an R-B meter (Figure 5.5). The incidence data are plotted on log- arithmic scales because of the observation that the 1971-1972 incidence data (and the more recent data) gave similar slopes on such a graph for all age groups and for males and females even though the absolute value of the incidence was appreciably greater for males than for females and higher age groups had a much greater incidence than lower age groups (Fears et al. 1976, NRC 1975). Such an exponential model implies that over the dose range considered there is no threshold for the population. The model was applied to the 1971-1972 data to obtain estimates of the relative increases in skin cancer incidence accompanying a relative decrease in ozone (calculated as a DNA dose) or as a relative increase in the R-B meter reading (NRC 1979a). A similar method of analysis was applied to the 1977-1978 data. For a 1 percent increase in the R-B meter reading, Scotto et al. (1982) predicted the incidence of skin cancer other than melanoma to increase from a low of 1.9 percent at high latitudes to a high of 2.9 percent at low latitudes. The overall estimate is an approximate 2.5 percent increase in skin cancer incidence for a 1 percent increase in the R-B meter reading (Scotto et al. 1982). Since a 1 percent decrease in ozone concentration corresponds to an approximate 0.8 percent increase in the R-B meter reading (NRC 1979a), these data imply that a 1 percent decrease in ozone concentration would result in an approximate 2 percent (0.8 x 2.5) increase in skin cancer other than melanoma among the white population of the United States. A correction needs to be made for statistical bias and for the variability in dose received among individuals of a particular population (Scott and Straf 1977). The estimated increase in squamous cell cancers due to increased W exposures is predicted to be greater than the increase in basal cell cancers (see Figure 5.5). For females the squamous cell values would be at least twice those for basal cell cancer, and for males at least 1.5 times (Scotto et al. 1981).
99 Percentage increases in various measures of skin cancer corresponding to various decreases in stratospheric ozone concentration have been predicted by the Panel to Review Statistics on Skin Cancer of the NRC Committee on National Statistics (NRC 1975, Appendix C). Using the simple model that the logarithm of the incidence is proportional to annual UV flux (NRC 1975, Appendix C), the panel has also estimated increases in skin cancer incidence from age- and sex-specific data from the 1977-1978 NCI survey (Scott 1981). The panel has made a number of refinements in the measures of UV used in their 1975 work. Theoretical computations have been done on more recent ozone data, with more wavelengths, and for . more areas. The panels estimates of annual UV dose reflect weights corresponding to the action spectrum for DNA damage (Figure 2.2) and are corrected for cloud cover and transmission through the skin. The relationship between the panel's values of W and those reported from readings of the R-B meter is very nearly linear. Thus either measure of UV will exhibit the approximately linear relationship between log incidence and UV, and, using the simple model, estimates of changes in log incidence corresponding to equivalent changes in the two measures of UV are found to be approxi- mately the same. However, the estimates of percentage increases in skin cancer incidence that correspond to a given percentage decrease in ozone depend on the resulting increase in the magnitude of the measure of UV used in the model, and thus depend on the locality, being greater at lower latitudes. Some predictions by the panel of percentage increases in skin cancer incidence in two localities for 5 percent and 10 percent reductions in stratospheric ozone are shown in Table 5.4. Those values are appreciably greater than those computed using R-B meter readings (Table 5.5). The differences between the two methods of calculation reviewed above are not the result of different epidemio- logical data, since they both use the same incidence rates, but of different ways of estimating the change in UV dose per unit change in ozone concentration. In addition, the NRC panel made corrections for variations in doses among individuals and in time (E. L. Scott, University of California, Berkeley, personal communication, 1982). All measures of annual light flux at the surface of the earth are functions of latitude. Visible light, WV-A, and UV-B increase with decreasing latitude. Assume
100 TABLE 5.4 Estimates, Derived from Calculations of the W Flux Corresponding to a DNA Action Spectrum, of Percentage Increases in Skin Cancer Incidence in the U.S. White Population Due to Reduction of Stratospheric Ozone Concentration (with 90 Percent Confidence Bounds) for Two Localities, by Sex and Cell Type Percentage Increase in Skin Cancer for Ozone Reduction of: SO Not Corrected Basal cell Bias Corrected 10% Not Corrected Bias Corrected Minneapolis-St. Paul Male 7.7 13.0 17.0 29 (2-14) (10-15) (11-25) (22-38) Female 5.8 9.8 12.3 21 (4-9) (8-12) (8-20) (16-28) Dallas-Ft. Worth Male 16.7 28 34 67 (10-25) (20-35) (21-56) (49-91) Female 11.3 20 26 47 (8-20) (15-31) (1842) (35-64) Squamous cell Minneapolis-St. Paul Male 12.7 24 29 54 (8- 19) (17-29) (19-42) (39-70) Female 12.0 21 27 50 (6-19) (16-27) (18-40) (37-65) Dallas-Ft. Worth Male 26 49 65 136 (1844) (36-63) (38-96) (95-190) Female 25 46 60 126 (16-41) (34-60) (37-90) (89-174) NOTE: The relation, log incidence = or +,BF + error, is assumed, where F is the annual flux weighted according to the DNA action spectrum in Figure 2.2. The values in the "Bias Corrected" columns are corrected for variations In doses among individuals and in time (Scott and Straff 1977). The correction is to multiply the estimate of ~ by 1 SOURCE: E. L. Scott, University of California, Berkeley, personal communication, 1982. for simplicity that they are all linear functions of latitude whose coefficients of proportionality depend on the wavelength region. In this special case, a 10 percent change in any wavelength region, such as the visible, as a result of a change in location in the United States, would correspond to a 10 percent change in .7.
101 TABLE 5.5 Estimates, Derived from Measurements of W Flux by a Robertson Berger Meter, of Percentage Increases In Skin Cancer Incidence In the U.S. White Population Due to Reduction of Stratospheric Ozone Concentration (with 95 Percent Confidence Bounds) for Two Localities, by Sex and Cell Type Percent Increase in Skin Cancer for Ozone Reduction of: 5% 1 0% Basal cell Minneapolis-St. Paul Male 5.6 11.2 (3-8) (5-1 7) Female 4.4 8.9 (2-7) (4-1 4) Dallas-Ft. Worth Male 8.4 16.9 (4-1 3) (8-26) Female 6.8 13.6 (3-1 1) (6-21) Squamous cell Minneapolis-St. Paul Male 8.7 17.4 (5-1 3) (9-26) Female 9.2 18.4 (4-14) (9-28) Dallas-Ft. Worth Male 13.3 27 (7-20) ( 1440) Female 14 28 (7-22) ( 1 343 ) NOTE: The model used to fit the data is log incidence = or + OF + error, where F is the flux in annual Robertson-Berger meter counts times 10-4. The values in the table were calculated using Scotto et al. (1981) estimates of percentage increases in incidence per 1 percent increase in R-B counts. These estimates were multiplied by 5 (or 10) percent increase in R-B counts times a radiation amplification factor of 0.8 (NRC 1 979a), which is approximately the percent increase in R-B counts per 1 percent decrease in ozone concentration. (The actual radiation amplification factor will be slightly larger as the reduction in ozone concentration gets larger.) No bias corrections were made as in Table 5.4. SOURCE: Derived from Scotto et al. (1981) and NRC (1979a).
102 W-B. But a change in stratospheric ozone concentration would change the UV-B component without changing the visible component of sunlight. Hence changes in flux computed from latitude dependencies cannot be used to predict increases in skin cancer, unless the wavelength region used for prediction corresponds to that for skin cancer induction. The epidemiological data, by them- selves, do not enable one to determine which wavelength region or regions are important in skin cancer incidence. Thus it is understandable that estimates derived from calculations of the UV flux corresponding to a DNA action spectrum and those from measurements of W flux by an R-B meter give different estimates of the predicted increase in skin cancer per unit decrease in ozone. To calculate such a prediction, an action spectrum must be assumed, and although the R-B meter is supposed to measure incident UV in the erythema action spectrum, in fact, as was noted earlier, it measures substantially more W-A (Figure 2.2), which is not sensitive to changes in ozone concentration. Moreover, at present, the best action spectrum to use for such a calculation is the DNA one, even though there are reservations about the roles played by combinations of wavelengths and immunological effects (see Chapter 3). Hence theoretically the best estimates are obtained by the calculations used in Table 5.4. A completely different way of estimating the impacts of ozone depletion on skin cancer other than melanoma is to assume a knowledge of the shape of the dose-response curves from animal data (de Gruijl and Van der Leun 1980, Rundel and Nachtwey 1978). The increase in skin cancer as a function of age at a particular location is assumed to follow such a dose-response curve, the dose increasing proportionately with age. Calculations based on such theories (de Gruijl and Van der Leun 1980) predict an approximate 5.5-fold increase in skin cancer per 1 percent decrease in ozone concentration, if one assumes that the 1 percent decrease in ozone corresponds to an approximate 2.3 percent increase in the UV that would damage DNA (NRC 1979a). Predictions by the NRC panel (Scott 1981) estimate the overall increase in basal cell skin cancer incidence per 1 percent decrease in ozone concentration to be between 2 percent and 5 percent depending on latitude. For squamous cell cancer the values are approximately double. The estimates by Rundel and Nachtwey (1978) and de Gruijl and Van der Leun (1980) are in agreement with the panel's. The uncertainties, at present, in such
103 values are approximately the same as the uncertainties in the predictions of expected decrease in ozone concentration. Experts agree that exposure to sunlight causes more than 90 percent of basal and squamous cell cancers in the United States. These estimates can be crudely tested in at least two ways. The first is to assume (a) that the incidence rate of basal and squamous cell cancers is the same in blacks and whites for all causes except sunlight, and (b) that in blacks these cancers are not caused by sunlight. For these crude calculations it is not necessary to take into account the probable differences in the relationship of basal versus squamous cell cancer to sunlight (Brodkin et al. 1969, Urbach et al. 1972). If these gross assumptions are accepted, the difference between white and black incidence rates would be an estimate of the rate of basal and squamous cell cancer in whites that is due to sunlight. For example, the annual age-adjusted incidence rate of both types of cancer in blacks is 3.4 per 100,000, compared with 232.6 in whites (Scotto et al. 1981). These figures indicate that 99 percent are related to sunlight. Even if it was assumed that all the cancers in blacks were squamous cell--since in blacks this type is relatively more common than basal cell--92 percent of squamous cell cancers in whites would still be related to sunlight. However, given that the incidence rates in blacks have some correlation with latitude, as was suggested earlier, these values would be overestimates. The second way to get a crude estimate of the n~r~ntaae of cancers caused by exposure to sunlight is ~ . . . .. ., . · ~ ~ w _a _ to assume that in whites the incidence rates on obey sites virtually never exposed to sunlight (genital areas) are base line measures of incidence from all other causes. For example, one tenth of 1 percent of basal cell cancer was found on the genital areas of men's bodies and 0.3 percent on women's (Scotto et al. 1981). The comparable figures were 0.8 percent and 2.8 percent for squamous cell cancers. With these base lines, more than 99 percent of basal cell cancers and more than 90 percent of squamous cell cancers would be attributed to UV irradiation.
104 Melanomas What Was Known by 1979 Melanoma is less common and more dangerous than other types of skin cancer. Data on mortality rates are shown in Table 5.6. In the NRC (1975) report, Environmental Impact of Stratospheric Flight, melanoma was accepted as a disease associated with exposure to sunlight and found on parts of the body exposed to sunlight although not concentrated on the sites that receive the highest intensity. There was a higher incidence at lower latitudes and among fair-skinned people. The reasons in favor of translating this association into one between melanoma and the W-B component of sunlight were summarized in NRC (1975) and repeated in NRC (1976a, 1979a). The argument was seen at that time to be clearly less substantial than that supporting the inferred relationship between WEB exposure and basal and squamous cell skin cancer. TABLE 5.6 Mortality of Melanoma by Country and Latitude for White Populations Latitude of Deaths from Country or Center of Melanoma per Geographical Area Population Million per Year South Island, New Zealand 45 S 8 North Island, New Zealand 39°S 12 Victoria 38 S 8 New South Wales 34°S 15 Cape of Good Hope 32 S 13 Natal 29 S 18 Transvaal and Orange Free State 27 S 10 Queensland 27 S 23 California 3 To N 12 Northeast United States 42 N 9 Italy 43o N 2 Switzerland 47o N 6 France 48 N 1 Canada 50° N 5 Netherlands 52 N 5 England s3o N 6 Eire 53°N 3 Scotland 56 N 4 Sweden 59° N 9 Norway 61 N 10 - SOURCE: Lancaster ( 19 56 ).
105 The argument was built on the facts that sunburn and melanoma are often found in the same tissue, melanomalike lesions can be induced by irradiation of chemically induced benign pigmented lesions in experimental animals, and individuals with xerodema pigmentosum have an extra- ordinarily high prevalence of melanoma (Kraemer 1980, Takebe et al. 1977). An additional argument was that one could read a pathogenetic relevance in the similarity of the erythema and DNA-damaging action spectra. As discussed in the next section, since 1976, the case for an association between W-B and melanoma has been weakened rather than strengthened by the results of additional clinical, pathological, and epidemiological studies. Furthermore (with the exception of a single animal), it has not been possible to use W -B alone to induce melanomas in experimental animals. The only statistical association that has been repeatedly found nationwide and worldwide is the one between melanoma incidence or mortality rates and latitude. Although widespread, the association is not totally consistent. There is still no clear evidence, although cohort analysis shows an increased incidence with age, that the latitude association is a dose-related relationship. There seems to be no doubt that Western countries have been living through a rapid increase in melanoma incidence (Houghton et al. 1980, Lee et al. 1979). Each successive cohort studied has had higher incidence rates. The epidemic has been affecting popula tions at many different latitudes with varying background incidence levels. ~ ~ ~ ~ ~_ ~_ The epidemiological picture ot higher Incidence rates In each successive birth cohort is reminiscent of the earlier lung cancer epidemic in those same countries, which resulted from the Progressive . . adoption of the habit ot cigarette smoking by more ana more members of each younger generation. In spite of grave reservations about the nature of the observed statistical association, NRC (1975, 1976a, 1979a) used the existing statistical association between either latitude or R-B meter readings and melanoma incidence or mortality to make predictions about the . . . likely increase in melanoma incidence, given future increases in UV-B exposure. It was clearly recognized that this decision was made without knowledge of the percentage of skin melanomas in the United States likely to be caused either wholly or in part by UV-B exposure and without good evidence pinpointing other factors that would be more powerful determinants of the future
106 incidence rates. It was argued that the predictions would be useful even if the association between latitude and melanoma incidence turned out to be indirect or an extremely remote index of the true causative factor. Lacking any epidemiological clues to major etiological factors other than sunlight, NRC (1976a, 1979a) tried to provide a behavioral explanation (changes in exposure patterns) for epidemiological inconsistencies on the basis of variability in personal susceptibility recognized in series of clinical observations. Following through on this line of reasoning the earlier studies had emphasized the need for new and more extensive data that would permit, for individuals with varying levels of innate susceptibility, analysis of measurements of exposure to sunlight. Advances in Knowledge Much new information about melanoma has been collected and published since 1979. Some of this information confirms the association between melanoma and latitude and levels of UV intensity. Much of it underscores and extends the inconsistent and sometimes paradoxical findings from past epidemiological studies, and some of it provides interesting new avenues for exploration. When the 1973-1976 incidence data from the NCI SEER program are plotted against the 1977-1978 NCI R-B meter measurements of accumulated dose in eight geographic locations, the results are consistent with those of earlier analyses and show a definite relationship between melanoma and measurements of annual solar W flux (Scotto et al. 1982) (Figure 5.7). The slope resulting from this statistical analysis is similar to that obtained for basal -elf cancers (Figure 5.5) and is virtually the same as the slopes previously developed from other bodies of data (Scott and Straf 1977, Scotto et al. 1982). Other newly published studies of incidence data again con- sistently report higher rates of melanoma at lower latitudes (Crombie 1979, Jensen and Bolander 1980, Malec and Eklund 1978). Several reported studies of trends in incidence have confirmed the continuation of the worldwide increase. Although the overall increase in the incidence of melanoma is virtually universal, the incidence on specific body sites has increased at various rates (Scotto et al., 1982). A report that melanoma of the eye has not
107 Skin Melanoma Wh its f emales ite Males 7 6 to 1 0 if: CC CC A: of en en 9 8 U] A: LL an j O ° ~ 5 _, _ ~ ~ O To 4 _ ~ ° ·L CC _ 3 ~ `,, L 100 o o o Detroit 1 1 1 1 120 1 40 . o a ~c <0 _ at ~. c c In 0 6 z 1 a x in 1 1 1 1 1 160 180 SOLAR UV RADIATION INDEX FIGURE 5.7 Annual age-adjusted incidence rates for skin melanoma (SEER data, 1973-1976) among white females (open symbols) and males (closed symbols), accord- ing to one year's W measurements at selected areas of the United States. The W radiation index is the total Robertson-Berger meter counts over a one-year period multiplied by 10-4. The meters read W-B between 290 rim and 320 nrn, as well as some W-A. (J. Scotto, National Cancer Institute, personal communication, 1981.) 200 importance, because this increased in incidence has some tumor occurs in the back of the eye, where UV does not penetrate, and is therefore unlikely to be associated with exposure to UV (Strickland and Lee 1981). A new series of reports on the occupational incidence of melanoma has provided very consistent information (Lee 1981), as has a second series of studies of the incidence of melanoma in immigrant and indigenous residents of Israel (Movshovitz and Modan 1973). Each series, however, provides information that has to be reconciled with that of the other. In the occupational series, four studies from different parts of the world and from very different latitudes failed to demonstrate any excess incidence of or mortality from melanoma among outdoor as compared with indoor workers of similar status. All four studies confirmed earlier reports of the correlation of increasing incidence with higher socioeconomic status.
108 In the second series, individuals born in high-incidence areas had higher incidence rates than did all immigrants. Immigrants from areas of lower incidence who had moved into areas of higher incidence assumed higher incidence rates, and their risk of developing melanoma increased with the number of years they had lived in their new and more dangerous locations. This is true for Israel (Anaise et al. 1978), Australia (Holman et al. 1980), and California (T. Mack, University of Southern California, personal communication, 1981). In the California study, California-born residents have the highest rates of melanoma incidence and immigrant Midwesterners who have moved to California have the lowest rates. The incidence rates for California residents born halfway between California and the Midwest fall somewhere in between. This ranking of incidence rates by place of birth among California residents does not hold for melanoma of the eye or for melanoma in parts of the body other than the skin. A report of a high incidence of melanoma among workers at the Lawrence Livermore National Laboratory may provide a unique opportunity to identify contributory, if not causative, etiological factors (Austin et al. 1981). Most other recent information concerns individual susceptibility and the etiology of specific histological types of melanoma, and precancerous conditions. Individual Susceptibility. Although Scandinavian populations have unusually high incidence rates of skin melanoma for their latitude of residence, it is now known that they also have high incidence rates of melanoma of the eye, which cannot readily be related to either sunlight or UV-B exposure (Strickland and Lee 1981). This combination suggests an underlying susceptibility to melanomas in general. There are a number of reports of higher incidence rates of melanomas in women during the later years of reproductive life in populations with different base rates of melanoma incidence and among different ethnic groups (Jensen and Bolander 1980, Lee and Storer 1980) The possibility of a specific hormonal component in the etiology of a certain proportion of these tumors is now being considered. Some early analysis of data collected at the University of Sydney suggests that in the women with higher incidence rates the ratio of superficial spreading melanoma to nodular and other histological types (Table 5.7) is higher than average. This increased .
109 o Cal 3 - o Ct no Ce :^ Ct .§ C) C) Ct Ct a, ¢ o o > Ct ~ V) ·C) V) ¢ .= ~ ~ Ct Cal o ho ,6 _ Cal , ~ 3 cat ·3 ~ - ~O D e 0 c ° ~ 0 · C, ~ a ~ ~_ ~ _ _ 5 C ^ _ G o _ C ~ _-= ~ 0 5 He in ~ a ~ C (~, V) s~>~ ~ e U) t ,, ~,. ~ 3 ~ o C ~·~- 8 o ~o ~ ~ U) ~ o Ct ~ C~ ~ ~-~ D ~O.= O ~ CS: ~ O L ~1 Ot ~O 1 C~ Ct ~ =0 Ctc~s ,:·_ ·- O ~ O- , ~ Ct ~ ~a~ s_ C) O ~ ~ ~=0 ¢ V) Ct ~ ~ Ct V, ~ ._ 4 - ,= O ,~, C~ ~ U~ Ct X ~ O C:' ~ a' . Ct ,~ s~ ~ 4- ~4 .- O =; o ,D V, 5 c~ Ct e~ Ct '~ O C~ ~ V) .9 C~ ~ U, Ct .= O cn ,~ O ~Ct X >,, ~ ,) · O ~ O ~Ct ,.= V) ~ ~ ~C~ ·C~ $- ~- 5 ~._ ;~ ~so _ >, ~: ~Z O C~ ~.. ·~ ~ ~ ~ o ~ c~ o~sa ~
110 ratio has also been found in other groups, for example, among men in the highest socioeconomic classes. It is also found on body sites with the highest rates of increasing incidence, namely, women's lower legs and men's trunks (backs) (McCarthy et al. 1980). These early findings could explain a number of other recent reports that note that the proportion of small and thin newly diagnosed melanomas is steadily increasing in populations with both very high and moderate incidence rates. One interpretation would be that the superficial spreading melanoma is the major cause of the increased rate in the groups identified above; another is that lesions are being detected at an earlier stage in those groups and are being treated while they are small and thin. Clinicians have always associated the lentigo melanoma with excessive exposure to W. Some clinicians believe that melanomas of this type are undercounted, particularly in the lower latitudes, where they are more common. It is the view of these clinicians that these cancers usually progress so slowly that they rarely reach a serious point during life, and thus remain undiagnosed. Without good diagnosis and reliable reporting, there can be no valid assessment of the distribution of these cancers over the body. Clinical observation would lead us to be believe that they virtually always appear on exposed areas of the body. If they are relatively rarely diagnosed and are not lethal, the published incidence rates of 6 percent to 10 percent among all melanomas must be taken as uncertain. If, for example, superficial spreading and nodular melanomas were not associated with exposure to UV and all incidence cases could be accurately counted, the proportion of lentigo melanomas among all melanomas would be higher in high-incidence areas, such as Texas or Australia, than in low-incidence areas. There are no available data with which to test this hypothesis (Lee and Strickland 1981). This is one example of the growing interest in specific histological types of melanoma. Much more detailed histological descriptions at the time of diagnosis are needed to provide the basis for pursuing this potentially fruitful research. Other preliminary data suggest that individuals with melanomas sunbathe less and use more sunscreens than do control subjects. Furthermore, they may have less residential, occupational, and recreational exposure to the sun. These findings seem to apply to both susceptible and less susceptible individuals (S. Graham, State
111 University of New York at Buffalo, personal communication, 1981). There is considerable evidence of familial concentration of melanoma (S. Graham, State University of New York at Buffalo, personal communication, 1981). In a recent study of 214 patients with melanomas, an appropri- ate group of controls, and family members of both groups, it was found that family members of patients with melanomas have high relative risks that are of the order of eightfold for all first-degree relatives (parents, offspring, and siblings) and twelvefold for parent- offspring pairs (Duggleby et al. 1981). These very high risks in family members could be consistent with other studies reported below. It is perhaps important to mention that this high relative risk found in relatives of the individuals first identified as having melanoma (which may be of great importance in clarifying the etiology of melanoma) occurs in relatively few instances and can account for only a few among all cases of melanoma. Precursor Lesions. There have been a number of recent reports of precursor dysplastic nevi (Clark et al. 1978; Elder et al. 1980, 1981; Reimer et al. 1978; Wiskemann 1977). Dysplastic nevi are usually large irregular moles on the skin that exhibit evidence of abnormal histological development (dysplasia). It was first believed that these lesions were always part of a familial condition called B-K Mole Syndrome (Clark et al. 1978; Green et al. 1978, 1980). However, it is now believed that there are both familial and sporadic dysplastic nevus syndromes and that the progression from a typical (i.e., histologically normal) nevus to a melanoma is analogous to the progression in the cervix from normal endothelial cells to squamous cell carcinoma in situ. It is also believed that on the skin, as on the cervix or in the bronchi, dysplasia is likely to occur in multiple sites. A body of histopathological, clinical, and biochemical evidence is being accumulated to explore this hypothesis, and some tentative results from laboratory experiments suggest that the fibroblast cells of patients with dysplastic nevi and hereditary cutaneous melanoma are peculiarly sensitive to W radiation (Smith et al., in press). During the past 5 years, there has been an increased number of laboratory and case control studies of individual human beings and population-based incidence
112 studies. Preliminary results are available from very few. Other investigators expect that their results are likely to strengthen evidence favoring individual types of susceptibility and to emphasize the need to analyze melanoma incidence rates by histological type (Sober et al. 1979). There does not seem to be any reason to expect strengthening of evidence in support of a hypo- thesis that lengthy accumulation of exposures to W radiation per se is the overriding or even one of the most important causes of melanoma other than lentigo maligna melanoma. There is an increasing number of individually inconclusive reports that all suggest that a history of acute exposures such as sunburn or marked skin sensitivity to sun exposure may be particularly important (Beitner et al. 1981, Jung et al. 1981, Paffenbarger et al. 1978, Sober et al. 1979). In light of the inconsis- tent and inconclusive state of knowledge about a possible dose-response relationship between melanoma and W. we are unwilling to make quantitative estimates of the effects of reduced concentrations of atmospheric ozone on the incidence of melanoma. ~,£ ., ~ ~ _.-~ PROTECTION AGAINST DAMAGE FROM SUNLIGHT Most of the direct human health hazards predicted to result from a depletion of stratospheric ozone concentra- tion, and a consequent increase in solar W. stem from exposure of the skin--increased incidence of sunburn, solar degeneration, skin cancer, and immune system effects. All skin is not equally susceptible to UV damage, however. There are two principal intrinsic barriers to UV. One is the stratum corneum on the outermost surface of the skin, which absorbs the most biologically active wavelengths of UV. This layer is approximately the same in all individuals and can be thickened as a reparative response to W injury of skin. The other physiologic, chemical, and optical protector against UV is a pigment called melanin, which is produced by cells in the epidermis called melanocytes. This pig The production of ants pigment is increased after sun exposure (tanning). The base line amount of melanin and the cape-city to increase melanin production are genetically determined. White persons have much less melanin than blacks. Caucasians have different levels of melanin in their skin In general, those with the least base line pigment have . ment gives skin its brownish color. .
113 the least capacity for tanning. These individuals are the most susceptible to sun damage of all kinds. The base line pigmentation of very dark skinned races protects against UV-B radiation 30 times better, and of moderately dark-skinned races 3 to 5 times better, than that of fair Caucasians. The range of base line pigmentation, and the capacity for tanning (i.e., for increasing melanin production), in fair Caucasians has been arbitrarily divided into four categories (see NRC 1979a, Appendix H), depending on the person's assessment of his or her own propensity to sunburn (relative absence of base line melanin) and ability to tan. Information is obtained by asking a standardized question about response to sun exposure. This method, called "skin typing," has proved to be a useful shorthand for categorizing persons in terms of responses to phototherapy, sunscreen testing, and clinical surveys. It is, however, not quantitative and is subject to cohort and interviewer bias. It simply predicts photobiologic response on the basis of the subject's memory of past photobiologic response. Two additional excellent barriers against UV are . ~ ~ ~_ ~l~=n~--mh~mi~ I S available. Were are now ex~ll=~ ~ =~ A.. ~ that when applied to the skin absorb W before it reaches viable cells. They provide a wide range of added protec- tion that can reach a factor of more than 10. This means that if it normally requires 25 minutes of sun exposure at noon in June to cause minimal sunburn in a fair person, a sunscreen with a protection factor of 10 would change the requirement to 250 minutes. This large amount of protection is more than enough to cover the UV increases likely to result from possible ozone depletion. Screening provided by protein, melanin, and topically applied sun- screens is most likely additive (Hawk and Parrish 1982). The other means of protection is the most effective. Avoidance of sunlight between 11:00 a.m. and 2:00 p.m. greatly reduces the exposure of skin to UV-B. Even modest changes in human behavior can decrease solar W exposure by factors that are much greater than the least conserva- tive factors estimated for ozone-related increases in W. Finally, the possible anticarcinogenic effects of $-carotene and synthetic retinols are being explored, but the roles of these compounds are complicated and controversial at this time.
114 RESEARCH RECOMMENDATIONS The following list of research recommendations is not exhaustive but has been limited to those issues that should receive attention first. Two of the several direct human health hazards that might be expected to result from an increase in the intensity of solar UV radiation should be emphasized in future research: immune system effects and skin cancer. The list is not organized according to priority. 1. Photoimmunology is a new and important area of research. It appears that erythmogenic (sunburn-causing) UV exposures can cause systemic alterations in the immune systems of animals and humans. The implications of these findings for understanding the pathogenesis of skin cancer and certain other diseases must be investigated. The identification of common mechanisms would be an important contribution. As an initial step, studies to determine the magnitude of UV-B effects on the human immune system, the dose-response relationships, and the effective wave- lengths should be vigorously pursued. 2. The use of animal models to study W-induced skin cancer (experimental photocarcinogenesis) has proved valuable in understanding the role of UV in the develop- ment of human skin cancer other than melanoma. Further animal studies are needed to understand interactions among parameters such as intermittent exposures, different wave- lengths, dose rates, and agents that modify cellular responses to W irradiation. 3. An animal model for light-induced melanoma must be discovered before it will be possible to determine if a reduction in stratospheric ozone concentration will cause an increased incidence of melanoma in humans. Dose-response relationships and effective wavelengths should be determined. 4. Prospective studies of patients undergoing various forms of phototherapy and photochemotherapy could be helpful in obtaining quantitative information about the relationship of certain UV wavebands to human skin cancer. 5. Epidemiological studies of skin cancer incidence and mortality rates have supplied valuable evidence confirming the existence of an association between basal and squamous cell skin cancers and sunlight. As basal and squamous cell skin cancers are not routinely reported to cancer registries, it will be necessary to maintain routine surveillance by periodic surveys during the next
115 50 years. These incidence surveys should be at intervals no longer than 10 years and should collect data that can be subjected to cohort as well as cross-sectional analysis. 6. In addition to (5) above, epidemiological research on skin cancer other than melanoma should concentrate on retrospective and prospective studies of individual human beings. The latter will need some simple measures of effective individual exposure to W -B to correlate with incidence and/or documentation of complete protection from UV exposure to correlate with prevention of skin cancer. 7. Information obtained since 1979 makes it clear that the etiology of malignant melanoma is even more complex than previously believed. A number of risk factors are involved, and, in addition, there are various subtypes of melanoma. In order to determine the associa- tion between UV and melanoma, it is essential to determine the incidence of and latitude dependence of the various melanoma subtypes. To do this, careful epidemiological studies based on reliable clinical and much more detailed histological descriptions at the time of diagnosis are needed. 8. Epidemiological studies of individual human beings and their effective exposures are essential in learning more about the etiology of melanoma. These studies should include some that focus on the experience of children, some that explore associations between the development of nevi and the sensitivity to sunlight exposure, and some that explore the protective aspects of exposure to wavelengths other than W -B.