CHAPTER 8

CORRELATIONS WITH INDICATORS OF INTERMITTENT OR SEVERE SUN EXPOSURE

Results from epidemiologic studies of cutaneous malignant melanoma (CMM) have led to conclusions that ranged from those which implicate sunlight as a causal factor, (e.g., most of the latitude studies [Chapter 6], migration studies [Chapter 7]), to those which do not support sunlight as a causal factor (e.g., no overall elevated risk of melanoma observed among outdoor workers compared to office workers [Chapter 11]). In attempts to reconcile these differences, it has been suggested that the risk of CMM on sites not usually exposed to sun is increased by intermittent exposure to more intense sunlight, while chronic sun exposure at a relatively constant dose may have little effect or may be protective, due to preventive tanning of the skin. This hypothesis can be examined using different measures of intermittent exposure as surrogates for actual exposure. This chapter reviews studies of CMM which investigated intermittent exposures to sunlight as estimated by a history of sunny vacations, recreational activities, and history of severe sunburn (particularly in early life). Also included is a section which reviews studies of sunspots and seasonal differences in the incidence of CMM. Sunspots result in higher levels of UV radiation which reach the earth's surface in a cyclical pattern and may result in more severe UV exposures.

HISTORY OF SUNNY VACATIONS OR RECREATIONAL ACTIVITIES

A vacation in a sunny location may result in intermittent exposure to UV-B radiation at higher than usual levels. The levels of exposure can be variable but, in general, sunny vacations are chosen simply because of the increased sunlight in the vacation spot. Several studies have examined past history of sunny vacations using various methods to explore the role of sun exposure in the etiology of CMM.

Using data from the Swedish Cancer Registry for 1959 to 1968, Eklund and Malec (1978) found that CMM incidence increased with population density. In an attempt to explain this finding, the authors hypothesized that an increase in foreign travel as estimated by passport issuance, might explain the over-representation of CMM in the large cities because "foreign travel in Sweden generally means sunshine trips." A 3.7 percent increase in the mean frequency of annual passport issue between 1959 and 1968 corresponded to an increase in CMM incidence over the same time period. Both increases were especially high in the densely populated cities and counties. Regression analysis was performed using an exposure index based on annual UV radiation in the erythema-producing wavelengths and latitude. Results showed variations in CMM incidence between Sweden's major cities and some counties, but these differences were reduced when frequency of passport issuance was considered in the analysis. Thus, foreign travel as represented by passport issuance was seen as a possible explanation for the higher CMM incidence in the most populated areas of Sweden.

A Norwegian study (Klepp and Magnus 1979) investigated sunny vacations through self-administered questionnaires from 78 CMM cases and 131 unmatched controls who were other cancer patients. The study showed no differences between cases and controls in pigmentation characteristics (hair and eye color), in estimated time spent outdoors during leisure activities, or in degree of exposure to sunshine during occupational or leisure activities. There was, however, a borderline significant difference (p=0.05) between the proportion of cases (19.2 percent) and controls (9.2 percent) who had traveled to Southern Europe for sunbathing during the previous 5 years (estimated relative risk of 2.4). This study did not account, however, for socioeconomic status, a variable often related to CMM and the tendency to go on vacations.

A study of oral contraceptive use and CMM in England (Adam et al. 1981) included sun exposure factors to explore potential confounding of study results. The study included 169 women aged 15-49 with CMM registered during 1971 to 1976 at the Oxford and South Western cancer registries, and 507 female controls (three controls per case) matched by 5-year age group and marital status, drawn from the physicians' practice lists. The authors found no significant differences between cases and controls who "tanned themselves while on holiday abroad" (for legs, 78 percent cases and 73 percent controls, and for trunk, 70 percent cases and 67 percent controls). A higher proportion of cases than controls spent outdoor leisure time deliberately tanning their legs (77 percent vs. 69 percent) and trunk (64 percent vs. 53 percent), although these differences were not statistically significant.

A study of 595 melanoma case-control pairs in Western Canada by Elwood et al. (1985a) showed that substantial intermittent sun exposure (as assessed by vacation and recreation histories) was strongly associated with CMM. There was a significant (p<0.001) trend of increasing risk of CMM occurrence with the number of "sunny vacations" (defined as severe or more intense sun exposure than normal) per decade. A relative risk of 1.8 for CMM was also associated with a history of four or more sunny vacations per decade. This relative risk remained significant (RR=1.7, 95% C.I. 1.2-2.3) even after adjustment for pigmentation factors (hair color, skin color, history of freckles) and ethnic origin. Further analyses were conducted on the risk associated with certain activities and practices on sunny vacations. Sun exposure through beach and other light-clothing activities during sunny vacations was associated with a relative risk of 1.9 (95% C.I. 1.3-3.0) for 20 to 39 exposure hours per summer season compared with no summer vacation sun exposure after adjustment for pigmentary factors and ethnic origin. This risk level could be reached by 4 to 8 hours per day of sun exposure during a 1-week vacation. The relative risk for 40 or more hours of vacation sun exposure per summer season decreased slightly to 1.5 (95% C.I. 1.0-2.3), and was of borderline statistical significance. Socioeconomic factors were not controlled for in the above analyses, and may have led to biased results.

Holman et al. (1986) analyzed data from their matched case-control study of CMM in Western Australia to investigate the relationship of different histologic types of CMM with intermittent sun exposure, as measured by history of summer sun exposure (both recreational and total, which includes occupational) and clothing habits. Analyses were also performed to investigate variations in relationships of sun exposure and CMM by primary tumor site.

With the exception of HMFM, all histologic types of CMM appeared to be inversely associated with total summer outdoor exposure (none of the associations were statistically significant). Further analysis measured recreational exposure as a proportion of total outdoor time in summer to evaluate the concentration of sun exposure during days off work. This variable--recreational outdoor exposure proportion in summer (ROEP)--provided an index in which the range from 30 percent to 100 percent indicated an increasing concentration of outdoor (sun exposure) time during leisure days. The higher proportions in this index imply a "burst" of exposure over a relatively short part of the week as contrasted with low proportions (ROEP near 0 percent) for farmers who worked outdoors 7 days a week. The effects of ROEP were examined separately by age group (10-24 years, 25-39 years, 40 years and over) and by different time intervals prior to diagnosis of the case (0-4, 5-9, 10-19 and >20 years pre-diagnosis). Results showed little evidence of any association of CMM with ROEP after control for potential confounders (pigmentation characteristics, ethnic origin, and age at arrival in Australia). For SSM and NM, there was a suggestion of a dose-response gradient in ROEP at ages 10-24 years, but the trends were not statistically significant (p=0.148 for SSM and p=0.258 for NM). Additional analyses using the absolute average number of hours exposed per week or using the difference between average hours of recreational and work exposure as measurements of recreational exposure to summer sun showed no stronger evidence of an association between incidence of CMM and recreational sun exposure than that mentioned previously.

Analyses of specific outdoor recreational pastimes were conducted based on the 276 SSM case-control pairs. Both boating and fishing at least once per week during summer showed significantly increased risks for SSM when compared with those who never participated (boating: OR=2.43, 95% C.I. 1.10-5.39; fishing: OR=2.72, 95% C.I. 1.15-6.43). There was little evidence, however, for a relationship of SSM with swimming. Frequency of summer sunbathing at ages 15-24 years showed only a weak association with SSM (OR=1.26, 95% C.I. 0.78-2.05 for "less than one/week" and OR=1.32, 0.80-2.17 for "once or more per week"); however, when the analysis was confined to SSM occurring on the trunk, with "never sunbathing" as the reference group, the odds ratio for the higher frequency sunbathing group attained statistical significance with a p-value for trend 0.044, OR=1.20, 0.51-2.81 for less than once per week and OR of 2.55 (95% C.I. 1.05-6.19) for once or more per week. Again, socioeconomic factors were not controlled for in the analysis and may have led to biased results.

The types of bathing suit worn by women at ages 15-24 years (usually the period of most frequent sunbathing) and in the 10 years prior to case diagnosis were also examined. Results for the type of bathing suit worn showed a strong increase in risk of melanoma of the trunk with decreasing bathing suit coverage, even after control for potential confounders. Not all results reached statistically significant levels, as shown in Table 8-1, but increased ORs were associated with type of bathing suit style for CMM of the trunk (p-value for trend = 0.005 for women at ages 15-24 years and 0.006 for 0-9 years per diagnosis).

SUNBURN IN EARLY LIFE

Elwood et al. (1984, 1985b), in a Canadian case-control study (595 matched pairs), assessed both the tendency to sunburn and history of sunburn. A specific question was asked about sunburn in childhood using gradations of 1, 2, and 3 for "rare, very mild, or no burn," "moderate or infrequent," and "severe or frequent burn", respectively. The authors found a significantly increased risk for sunburn in childhood (RR=1.9) although the risk became smaller and statistically nonsignificant after adjustment for pigmentation factors (hair, skin, and eye color). A history of frequent sunburn in childhood remained a significant risk factor along with pigmentation even after control for ethnic origin, but the authors concluded that "the risk is due to characteristics of pigmentation associated with poor sun tolerance" (Elwood et al. 1984).

Lew et al. (1983) found similar results in a Massachusetts case-control study (111 melanoma cases and 107 unmatched controls): risk factors with elevated ORs included "blistering from sunburns in adolescence" (OR=2.05, 95% C.I. 1.18-3.56) and "painful sunburn as a child" for both those who tanned well (OR=2.8, 1.3-6.3) and those who tanned poorly (OR=3.0, 0.9-9.8). A history of extended sunny vacations (30 days or more as a child) was also found to elevate CMM odds significantly (OR=2.5, 1.1-5.8). The authors concluded that the same etiology underlies each of these risk factors, i.e., "the degree of response to sun exposure," and that the nature of these traumatic exposures and sunny vacations in early life suggests that traumatic doses of sun may outweigh lifetime cumulative doses as a risk factor for melanoma. These findings should be interpreted with caution because there was no control for pigmentation factors or socioeconomic status, and the control selection was flawed (see Chapter 10). In addition, the potential for recall bias, whereby cases are more likely to remember early sunburn episodes than controls, should be considered.

Holman et al. (1986) analyzed sunburn histories of CMM cases and controls in Western Australia (507 pairs matched on age, sex, and area of residence) for age groups under 10 years and 15-24 years. No relationship between sunburn in childhood or early adulthood and any histologic type of melanoma was found after control for potential confounders such as chronic and acute skin reaction to sunlight, hair color, ethnic origin, and age at arrival in Australia. Compared with persons who reported no painful sunburns in childhood (under 10 years of age), the odds ratios for SSM were 1.06 (95% C.I. 0.65-1.75) for those who experienced painful sunburn up to four times and 1.11 (0.51-2.41) for those who reported five or more painful sunburns. For sunburn during ages 15-24 years, the corresponding odds ratios for SSM were 1.04 (0.66-1.66) for four or fewer sunburns and 0.98 (0.53-1.82) for five or more sunburns.

LIFETIME HISTORY OF SEVERE SUNBURN

The following section includes only those studies which reported sunburn history, i.e., remembered severe sunburn, whether in childhood, adolescence, or adulthood. The possibility of recall bias among CMM cases with respect to sunburn exposures should be considered in the interpretation of these studies.

MacKie and Aitchison (1982) conducted a case-control study of 113 age/sex-matched pairs (hospital-based controls) in the west of Scotland which also considered social class and skin type in the analysis. There was a significant difference (p<0.05) between the cases and controls in the history of severe or prolonged sunburn 5 years before diagnosis of CMM (i.e., blistering sunburn or erythema persisting 7 or more days after sun exposure). Overall, 56 percent of the melanoma patients had a history of severe sunburn compared with 22 percent of the controls (RR=2.8, 95% C.I. 1.1-7.4 referent group not specified by authors). When separated by sex, there were still significant differences (p<0.05) between the cases and controls with respect to history of severe or prolonged sunburn. The presence of recall bias among the melanoma cases could have biased the findings if cases were more likely to remember or overstate their history of sunburns.

More extensive analyses of Western Canada study data on 595 age-, sex-, and residence-matched case-control pairs, by Elwood et al. (1985b), evaluated the relationship of CMM incidence to sunburn history at any age and vacation sunburn history for those with recorded vacations. There was an increasing trend for risk of CMM with increasing number and severity of sunburn episodes (p<=0.01) using a vacation sunburn score, and significantly raised odds ratios of CMM (p<=0.05) for each of the sunburn history categories considered separately (sunburn on vacations, sunburn in childhood, and history of severe sunburn causing pain or blistering for over 2 days) relative to those with no or mild sunburn. Vacation sunburn scores were analyzed with the variable "usual degree of suntan" in order to consider the separate effects of these variables, but each remained statistically significant after adjustment for the other, indicating that these two factors acted independently.

As in their earlier analysis of the sunburn in childhood data (Elwood et al. 1984), the authors evaluated whether the tendency to sunburn and sunburn history, as measured by vacation sunburn score, were independent risk factors for melanoma. The results of this analysis showed a weak association between CMM risk and vacation sunburn score after adjustment for the usual reaction to sun, i.e., ranging from "tan to burn" to "burn only." Multivariate analyses controlling for usual reaction to sun and other pigmentary factors (hair color, skin color, and freckles in adolescence) further weakened the association between CMM and vacation sunburns. The authors concluded that, while history of severe or frequent sunburn was associated with an increased risk of CMM, the tendency to burn easily and tan poorly was more strongly associated with CMM risk; thus, it appeared that it was factors associated with the tendency to burn rather than a positive history of sunburn which determined CMM risk. The authors concluded, however, that in contrast to sunburn history and vacation sunburn score, melanoma risk was increased by heavy vacation or recreational exposure to sunlight and that the increased CMM risk from these substantial intermittent sun exposures was independent of constitutional factors (pigmentation characteristics, reaction to sun, and number of nevi). This association is probably not due to the trauma of sunburn itself but to another characteristic of individual skin reaction, "related presumably to variations in melanocyte function, distribution or prevalence."

A case-control study (183 pairs matched by age, sex, and place of residence) in Queensland, Australia (Green et al. 1985) showed a positive trend of increasing CMM risk with an increasing number of severe sunburns (p<0.001). When the number of sunburns was grouped (0-1, 2-5, 6+) and an adjustment was made for the presence of pigmented nevi on the arms (the strongest risk factor determined in this study), and for age, the positive trend for CMM risk with number of sunburns remained significant (p<=0.05); the adjusted relative risk of melanoma for six or more severe sunburns was 2.4 (95% C.I. 1.0-6.1). The risk estimates were essentially unchanged by further control for other risk factors such as presence of other skin cancers, migrant status, and social class. The authors did not control for pigmentation variables such as hair or skin color or the tendency to sunburn; therefore, the results should be cautiously interpreted. Green et al. (1985) stated that "an experience of painful erythema indicates that acute high-dose UV has been delivered to the level of the melanocyte" and, because of this, the above-mentioned variables should not confound an association between CMM and severe sunburn (defined as sunburn with pain persisting longer than 48 hours, with or without blistering). The authors also concluded that the dose-response relationship obtained from their analysis supports a causal interpretation of an observed sunburn-cutaneous melanoma association. The authors believe that their results support an "intermittent episodes of acute UV exposure" theory, but may also fit a "cumulative" or dose-related theory in which the high UV dose has accumulated from multiple severe sunburn episodes.

Holman et al. (1986) analyzed sunburn histories from 507 CMM matched cases and controls (matched on sex, age, and residence) from Western Australia, ranking them according to increasing sunburn severity: "peeling sunburn," "painful sunburn (pain for 2 days or more)," and "blistering sunburn." After control for potential confounding factors (chronic and acute skin reaction to sunlight, hair color, ethnic origin, and age at arrival in Australia), only HMFM showed some association with the occurrence of severe sunburn (p-value for trend = 0.059). For nodular melanoma (NM), there was a significant trend in the opposite direction (p=0.010), giving the appearance of a protective effect. This result was based on small numbers (51 NM cases) and attributed to chance. For SSM and UCM, there was no association with past sunburn severity after control for confounding variables. Using data from the same study population, this group (Armstrong et al. 1986) was able to show that the prevalence odds ratio for nevi increased more or less linearly with increasing numbers of childhood sunburns up to the age of 10. The authors note that the relationship between nevi and sun exposure is a complex one and that their data are consistent with both total exposure and total outdoor exposure time in summer being important. In addition, the pattern of exposure may play a role.

SUNSPOTS AND SEASONAL INCIDENCE DIFFERENCES

Scotto and Nam (1980) analyzed monthly incidence of CMM from the Third National Cancer Survey (1969-1971) to test for seasonal patterns in incidence. There are monthly patterns in the amounts of solar radiation reaching the earth's surface, with the highest intensity UV-8 radiation occurring during the summer months. The authors examined seasonal patterns of CMM incidence for 2,167 white patients (998 males, 1,169 females) by sex, age, geographic region, and tumor site.

A highly significant seasonal pattern (p=0.00003) with a summertime peak was found for the total female CMM cases: over 20 percent of 811 cases were diagnosed during June and July, while less than 14 percent were diagnosed during the winter months of December and January. For males, no seasonal pattern of sustained peak or dip was found. Among males, the trunk was the most common tumor site (38 percent), while among females, it was the lower extremities (35 percent). Analysis of seasonal patterns of solar radiation by anatomic site showed two highly significant sites for females--upper extremities and lower extremities (p=0.0007 and p=0.0001, respectively), but none for males; upper extremities showed a tendency toward the seasonal pattern with summertime peak but this was not statistically significant (p=0.11). An attempt to reduce single-month disturbances in the patterns by grouping into 2-month periods still showed a significant pattern for females (p=0.003) but not for males (p=0.07).

The observations of seasonal patterns with summertime peaks in CMM incidence for females with tumors on upper or lower extremities may be due to the promotional effects of UV-B exposure, or it may be a result of greater awareness of skin changes and/or problems during the summer months when less clothing is worn.

Hinds et al. (1981) conducted similar analyses using 1960-1978 CMM incidence data for Caucasians in Hawaii. Based on 353 incident cases (males and females combined), the authors found significant seasonal patterns with summertime peaks for melanomas of all sites (p=0.018), and for those of the lower extremities (n=79, p=0.017). For head and neck melanomas, there seemed to be a similar seasonal pattern but it was of borderline statistical significance (p=0.069). Hinds et al. (1981) stated that, because there is little variation in the types of clothing worn throughout the year in Hawaii, it is unlikely that a seasonal pattern would be due to increased observation of the skin during the summer months. These authors concluded instead that their findings supported the hypothesis that solar (UV) radiation may be a short-term promoter of some malignant melanomas of the skin.

Houghton et al. (1978) used Connecticut Tumor Registry data on 2,983 CMM cases diagnosed during 1935 to 1974 to analyze CMM incidence rates. The rate per 100,000 rose from 1.1 to 6.2 over this 40-year period with evidence of cyclical patterns with 3 to 5 year peaks in incidence every 8 to 11 years. The secular increase in melanoma incidence was highly correlated (correlation coefficient = 0.9327, p<0.01) with three sunspot cycles of 8 to 11 years over a 33-year period. CMM incidence rose sharply at the peak of each sunspot cycle and the high rates persisted for 3 to 5 years before returning to a stable but increased rate. Controlling for the time effect on increasing rates did not alter the significant association between annual sunspot activity and CMM incidence rates in each of the subsequent 3 years.

Analyses of melanoma incidence from New York State and Finland also showed significant correlation with sunspot activity although data from Norway showed no significant results. Analysis of data from New York State (1950-1971) showed significant association between sunspots and CMM incidence in the 1 to 2 years subsequent to sunspot activity. In Finland, however, incidence rates were significantly correlated only with the years of sunspot activity and the first subsequent year; this decrease in lag period is postulated to be a result of Finland's higher latitude. Wigle (1978) reported similar cyclical variations in CMM incidence in Saskatchewan and Alberta, Canada; when CMM incidence rates were correlated with periods of sunspot activity, incidence rates were found to increase 0 to 2 years after the peak sunspot activity (0 to 1 in Saskatchewan and 2 in Alberta). After a review of these data, Houghton and Viola (1981) concluded that "rises in CMM incidence occur 0 to 3 years after sunspot peaks, suggesting that heavier exposures to UV radiation trigger the clinical appearance of melanoma."

FINDINGS

The following findings can be drawn from the review presented above:

8.1 A case-control study in Western Canada found evidence of increasing CMM risk with increasing number of "sunny vacations" taken, even after adjustment for pigmentation factors and ethnic origin. This study did not, however, control for socioeconomic factors which are likely to be associated both with CMM incidence and number of sunny vacations.

8.2 Using a variable for recreational sun exposure, defined as the ratio of summer recreational sun exposure to total summer sun exposure, a Western Australia study found no significant association of this factor in any of the age groups examined. However, an increased SSM risk for some summer sun activities at early ages was observed for boating, fishing, and female sunbathing (on trunk only). For CMM of the trunk in women, particularly for SSM, risks increased with decrease in coverage by the bathing suit style worn at 15-24 years of age.

8.3 CMM risk is associated with a history of childhood sunburn and/or lifetime history of sunburn but this appears only to reflect an individual's pigmentary characteristics, particularly as they relate to poor sun tolerance.

8.4 A seasonal pattern with a summertime peak was found for CMM (female in U.S., both sexes in Hawaii); this may be related to a greater awareness of skin changes in the summer months.

8.5 Most studies which examined the relationship between CMM incidence rates and sunspot cycles found high correlations. Different studies have observed slightly different lag periods between peak sunspot activity and increased CMM rates.

REFERENCES

Adam, S.A., Sheaves, J.K., Wright, N.H., Mosser, G., Harris, R.W., and Vessey, M.P. A case-control study of the possible association between oral contraceptives and malignant melanoma. Br J Cancer 44:45-50 (1981).

Armstrong, B.K., de KLERK, N.H., and Holman, C.D.J. Etiology of common acquired melanocytic nevi: constitutional variables, sun exposure and diet. JNCI 77:329-335 (1986)

Eklund, G., and Malec, E. Sunlight and incidence of cutaneous malignant melanoma. Scand J of Plastic Reconstr Surg 12:231-241 (1978).

Elwood, J.M., Gallagher, R.P., Hill, G.B., Spinelli, J.J., Pearson, J.C.G., and Threlfall, W. Pigmentation and skin reaction to sun as risk factors for cutaneous melanoma: Western Canada melanoma study. Br Med J 288:99-102 (1984).

Elwood, J.M., Gallagher, R.P., Hill, G.B., and Pearson, J.C.G. Cutaneous melanoma in relation to intermittent and constant sun exposure--The Western Canada melanoma study. Int J Cancer 35:427-433 (1985a).

Elwood, J.M., Gallagher, R.P., Davison, J., and Hill, G.B. Sunburn, suntan end the risk of cutaneous malignant melanoma - The Western Canada melanoma study. Br J Cancer 51:543-549 (1985b).

Green, A., Siskind, V., Bain, C., and Alexander, J. Sunburn and malignant melanoma. Br J Cancer 51:393-397 (1985).

Hinds, M.W., Lee, J., and Kolonel, L.N. Seasonal patterns of skin melanoma incidence in Hawaii. Am J of Publ Hlth 71:496-499 (1981).

Holman, C.D.J., Armstrong, B.K., and Heenan, P.J. Relationship of cutaneous malignant melanoma to individual sunlight-exposure habits. JNCI 76:403-414 (1986).

Houghton, A.N., and Viola, M.V. Solar radiation and malignant melanoma of the skin. J Am Acad Dermatol 5:477-483 (1981).

Houghton, A., Munster, E.W., and Viola, M.V. Increased incidence of malignant melanoma after peaks of sunspot activity. Lancet April 8:759-760 (1978).

Klepp O., and Magnus, K. Some environmental and bodily characteristics of melanoma patients. A case-control study. Int J Cancer 23:482-486 (1979).

Lew, R.A., Sober, A.J., Cook, N., Marvell, R., and Fitzpatrick, T.B. Sun exposure habits in patients with cutaneous melanoma: A case control study. J Dermatol Surg Oncol 9:981-986 (1983).

MacKie, R.M., and Aitchinson, T. Severe sunburn and subsequent risk of primary cutaneous malignant melanoma in Scotland. Br J Cancer 46 955-960 (1982).

Scotto, J., and Nam, J. Skin melanoma and seasonal patterns. Am J Epidemiol 111(3):309-314 (1980).

Wigle, D.T. Malignant melanoma of skin and sunspot activity. Lancet 2:38 (1978).