Sensitivity to sunburn, along with tanning ability, has been used to develop a classification system of six skin phototypes. The most sensitive individuals (skin type I) develop a moderate to severe sunburn after a short term (an hour or less) exposure in the summer, rarely tan even after repeated exposure and generally have very fair, often freckled skin, red or blond hair, and blue eyes. The most resistant individuals (skin type VI) are darkly pigmented without exposure and become even more deeply pigmented upon exposure (Pathak et al. 1985). These classes have been widely used to classify individuals within a population according to skin type; from such efforts have come the conclusion that sensitivity to sunburn is a risk factor for skin cancer (discussed in more detail below).
These tumors share some traits and differ in others. Shared traits include an increased risk of developing these tumors with increased exposure to sunlight, and a pigment-related variation in susceptibility, with darker skinned races and individuals being at less risk of developing skin cancer for a given amount of exposure than those with lighter skin. As discussed in detail below, although these tumors share certain features, their biologic behavior and relationships to solar exposure are qualitatively and quantitatively different; thus, they are discussed individually.
Epidemiological studies are a primary source of information for assessing the relevance of solar UV exposure to the etiology of diseases such as skin cancer. Even though there has been considerable improvement in study design, epidemiologic findings in this research area still suffer from two principal flaws. First, because of the lack of an independent metric of exposure, e.g., a biomarker of cumulative UV exposure, epidemiologic studies currently base their exposure estimates on patient and control recall of exposure much of which occurred decades ago, thus are subject to ‘recall bias’. Second, there are inevitable correlations between what may be independent risk factors, e.g., sun sensitivity is directly related to the numbers of actual sunburns or the use of sunscreens. These in turn relate not only to intermittent exposure but also to sun seeking behavior that also increases the overall level of exposure.
The incidence rate of BCC among various white populations has been increasing recently, albeit more significantly in some locations than others. In Albuquerque, New Mexico, an evaluation by a major health care provider of the change in age-standardized rates of BCC between 1978 and 1991 found an increase of about 13% per year (Hoy, 1996)(SCC showed little change in rate.) In contrast, a survey done in Australia showed only about a 2% annual change between 1985 and 1990.
Because early studies in the U.S. showed that the majority of BCC (almost 80%) occurred on the most heavily exposed sites (head, neck and extremities), the risk of BCC was once thought to be directly related to life-time cumulative sunlight exposure (Scotto et al., 1981). However, recent epidemiologic data suggest that this conclusion is too simplistic. In two recent epidemiologic studies from Australia, Kricker and her colleagues (1995a, b), examined several interesting aspects of BCC. In the first report these authors found that the risks of BCC occurring on heavily exposed sites (head, neck and extremities) decreased with increasing total exposure, whereas the risks of BCC occurring on an intermittently exposed site (the trunk) showed the opposite pattern: increasing risk with increasing exposure. In the second report these authors concluded that intermittent exposure, especially in youth (between the ages of 15-19), may be important explaining BCC. These data are consistent with two hypotheses. Kricker et al.(1995a) suggest there is a plateau in BCC risk at higher levels of exposure, in accord with that postulated by others who had observed a similar pattern (Vitasa et al., 1990). Gallagher et al. (1995a) suggest that it is only childhood exposures that are important. Unfortunately distinguishing between these two hypotheses is difficult, since a majority of an individual’s life time UV dose is accumulated by age 18 (Stern et al. 1986), and operationally they are the same: one represents a time threshold and the other a dose threshold.
Kricker and her colleagues (1995a, b) also looked at BCC risk based on skin type and found differing dose response relationships for those who tan well as compared to those who tan poorly. For good tanners, the risk of BCC increased with increasing sun exposure, whereas for the poor tanners the risk was initially flat and then fell with increasing exposure. Assuming the plateau hypothesis is correct, these observations suggest that good tanners get a lower effective dose per hour of solar exposure (probably due to the effect of tanning and skin thickening) than the poor tanners and thus their keratinocytes achieve this plateau later in their lifetime. The presence of a plateau also suggests that in order to reduce risk, particularly in those who tan poorly, substantial reductions in exposure will be required.
In line with earlier studies, the multicenter Helios study in southern Europe showed that the risk of SCC goes up with lifetime exposure (Rosso et al., 1996). A smaller, but very thorough, Canadian study (Gallagher et al., 1995b) on 180 SCC patients treated in 1983 and 1984, did not find any significant increase in risk related to lifetime exposure, but rather, increased risk was associated with chronic occupational exposure over the decade prior to diagnosis of the tumor. As the authors point out, these findings may be hampered by random errors in recall and by a correction for body area exposed. Besides introducing a new source of error, the correction for exposed body area may not be appropriate for SCC since risk is primarily determined by the genotoxic damage at the site of occurrence. That persistent UV exposure in the final stages of tumor development is important can be inferred from the observation that avoiding sun exposure, including through the use of sunscreens, reduces formation of actinic keratoses, precursor lesions of SCC (Thompson et al. 1993). The Canadian study also found an increase in risk in relation to regular overexposure (sunburns) during childhood. The latter correlation with childhood sunburns has also been found in other studies (Kricker, 1992;1992, Grodstein, et al., 1995). However, as pointed out above and by Armstrong and his colleagues (1997), this may primarily imply that a high level of childhood exposure, rather than sunburns per se, is the dominant risk factor. Considering the presently available epidemiological data, it appears prudent to conclude that UV exposure contributes to SCC risk both in early and late stages of tumor development.
Brash (1991) and his colleagues (Ziegler et al., 1993) found UV-related mutations in the p53 tumor suppressor gene in the majority of SCC as well as BCC. In addition, this group found that about 60 % of actinic keratosis (AK, precursor lesions of SCC) bear such mutations (Ziegler et al., 1994). These mutations were also found in experimentally UV-B-induced murine skin cancers, first in low percentages (Kress et al., 1992) but later in the majority (Kanjilal et al., 1993;1993, Dumaz et al., 1997). In samples from normal skin of the shoulders of Australians who had skin cancers removed, cells with a specific UV-related p53 mutation could be detected, while they were virtually absent in unexposed buttock skin (Nakazawa et al., 1994). In mouse experiments, microscopic clusters of cells with high levels of mutant p53 protein were observed long before the UV-induced macroscopic skin tumors appeared (Berg et al., 1996). Such microscopic clusters have also been found in human skin from surgical resections (Ren et al., 1996;1996, Brash et al., 1997). Hence, it appears that mutations in the p53 gene occur at a very early stage of the development of SCC (in contrast to other tumors, like colon cancer, where it marks a late conversion to malignancy).
In addition to earlier experiments in which SCC were induced in hairless mouse skin by chronic UV-B exposure, it has recently been reported that SCC can be similarly induced in normal human skin grafted onto immune deficient mice (Nomura et al., 1997). These experiments clearly show that the human skin taken out of its own environment can develop AK and SCC within 1 to 2 years of daily UV-B exposure, although the yield of frank SCC was low. This may provide an important new model to investigate quantitative and wavelength relationships between skin cancer induction and UV exposure.
As with BCC, the risk of CM does not appear to be directly linked to cumulative, lifetime UV exposure. CM frequently occurs at anatomical locations that are not the most heavily sun-exposed. Furthermore, as noted in a recent review of almost 30 epidemiologic studies, an increased risk of melanoma is associated principally with an increase in intense exposures of the intermittent type, e.g., such as those received by areas exposed only during outdoor recreational activities (Elwood and Jopson, 1997).
There is also considerable evidence that exposures in early childhood area may be important. A number of epidemiologic studies have also found higher melanoma risk with increasing sun exposure in individuals who lived in sunny areas during their childhood (Holman et al., 1984; Autier and Dore; 1998). The latter authors also found that sun exposure during childhood may in some instances constitute a significant risk factor for melanoma only if there is substantial sun exposure during adult life, i.e., that childhood and adult exposures act interdependently.
An additional risk factor for CM revealed by epidemiologic studies is the presence of one or more forms of pigmented lesions on the skin - either freckles or moles (which are also known as melanocytic nevi). In a number of studies, prevalence of nevi was the single most important determinant of melanoma risk, a finding that was later confirmed by clinical studies (Gallagher and McLean, 1995). Pathological examination of melanomas frequently reveals histologic evidence of a preexisting nevus (USEPA, 1987), leading to the suspicion that at least some nevi may be precursor lesions for melanoma. Since nevi exist in a gradient of premalignancy, with the common melanocytic nevus being the most benign and the dysplastic nevus being the least benign, some effort has been spent in evaluating the relative risk associated with various kinds of nevi. In a number of well-conducted epidemiologic studies, it has been found that having a large number of moles is associated with a higher risk of melanoma (Bataille et al., 1998; Berwick, 1998).
As shown in Fig. 2.3, a recently described
action spectrum for melanoma in fish (Setlow et al., 1993), unlike that
developed in mice for SCC (deGruijl et al., 1993) (and the rest of the
action spectra in Fig. 2.3, appears to have a strong UV-A dependence. Work
in the South American opossum provides some support to the notion that
exposure to UV-A may be important to CM. In that animal model UV-A treatments
alone induced melanocytic hyperplasia, a precursor to melanoma in these
animals (Ley, 1997). To date, however, these lesions have not progressed
to malignant melanoma. These findings suggest that the etiology of melanoma
is probably complex and likely involves a multistep process of both UV-B
and UV-A induced changes in a variety of different molecules (Longstreth