THE WESTERN Journal of Medicine Refer to: Morton WE, Starr GF: Epidemiologic clues to the cause of melanoma. West J Med 131:263-269, Oct 1979

Epidemiologic Clues to the Cause of Melanoma WILLIAM E. MORTON, MD, DrPH, Portland, Oregon, and GRIER F. STARR, MD, Eugene, Oregon

An intensive search found that 146 cases of cutaneous or ocular melanoma had occurred among residents of Lane County, Oregon, from 1958 through 1972. Of these cases, 35 led to death. Countywide, increasing incidence rates were corroborated by increasing death rates for both sexes. Risk of disease was highest in a moist, flat residential area near the intersection of two rivers in the county's urban portion and in its agricultural area. The incidence pattern strongly suggested local cycles in subcounty units. Although overall melanoma risk was higher in urban areas, an apparent widespread rural epidemic was identified, beginning abruptly in 1965 and lasting several years. If similar geographic and temporal characteristics can be identified in other localities, the search for an etiologic agent could focus on a smaller range of possibilities. ALTHOUGH THE CAUSE of melanoma remains to be clarified, it probably is multifactorial in nature. Possible etiologic factors can be grouped into categories: microbiologic (viruses, mycotoxins), chemical (inorganic, organic), radiation (ionizing, perhaps nonionizing) and genetic abnormality (viral genomes, mutations), all of which may interact." 2 The heritable component (susceptibility) of melanoma causation is represented by the relative immunity enjoyed by darkly pigmented persons2 3 and by the occurrence of families with a strong multigenerational pattern of From the Environmental Medicine Division, University of Oregon Medical School, Portland (Dr. Morton), and the Department of Pathology, Sacred Heart Hospital, Eugene, Oregon (Dr. Starr). Supported by a grant from the Oregon State University, Environmental Health Science Center, headed by Dr. Virgil Freed. Submitted, revised, May 24, 1979. Reprint requests to: William E. Morton, MD, DrPH, Environmental Medicine Division, University of Oregon Medical School, Portland, OR 97201

involvement.4-6 The wide variations reported for the occurrence of melanoma in different parts of the world have drawn attention to the likelhood that environmental factors are important influences in its geographical distribution. There is general agreement that exposure to solar radiation is another important risk factor for melanoma incidence,238-" leading to an observed inverse correlation with latitude of residence. Of considerable concern are the reports of increasing incidence of melanoma and mortality in California and other states, Canada, western Europe and Australia.'2-'8 Lee and other investigators have ascribed this increase largely to a modem tendency to wear fewer clothes, thereby risking greater exposure to solar radiation,5"0""'7 although MacDonald has presented data from Texas that show higher melanoma rates in more THE WESTERN JOURNAL OF MEDICINE

263

CLUES TO THE CAUSE OF MELANOMA

humid communities with fewer sunny days.'8 Unfortunately, the designation sunny day is not an adequate measure of solar exposure, but adequate measurements are not generally available for intrastate comparisons. In Finland, Teppo and colleagues found that the typical north/sou'th gradient in melanoma incidence was obliterated if the data were adjusted for degree of urbanization.'9 It seems reasonable to entertain other hypotheses besides greater exposure to solar radiation to account for increased melanoma incidence. Suspicions of the potential role of chemical carcinogens are beginning to appear in the literature. Bahn and co-workers reported the development of melanoma in two men with occupational exposure to polychlorinated biphenyls;20'2' Albert and Puliafito reported two cases of ocular melanoma in a group of Ohio Valley chemical workers exposed to hydrazine and dimethylsulfate.22 Certain workers in steel and nonferrous metals industries also seemed at excessive risk.23 Therefore, increased melanoma risk in persons with greater outdoor exposure24'25 may be due to solar radiation exposure, to greater environmental chemical exposure or to an interaction between the two. Using 15-year mean annual mortality rates for Oregon, Figure 1 shows that high-risk counties are scattered almost randomly across the state

except for a cluster of counties in the southwestern corner. Because most of its population lives in the western third of Oregon, this southwestern cluster of counties with high melanoma mortality agrees with the national pattern of higher mortality in lower latitudes as reported by Mason and colleagues.26 However, we do not propose that latitude alone fully explains the high melanoma mortality in southwestern Oregon. The melanoma mortality for Lane County, Oregon's second largest metropolitan area, was entirely unremarkable and below the statewide average, but an opportunity to study the local distribution of this disease arose in the course of a study of skin cancer incidence in relation to water arsenic levels.27 The spatiotemporal characteristics of melanoma were sufficiently unique to warrant separate description.

Methods For this study, the primary data sources, like those of the Third National Cancer Survey,28 were pathology files. All six hospitals in Lane County were served by a single group of pathologists whose files were scanned for cases of melanomas and other skin tumors for the years 1958 through 1972. We also sought Lane County cases from the pathologist in Coos Bay who periodically served coastal portions of Lane County, and from a mail-order private pathology laboratory

.5 Statewide Rates 1.5-2.1 High u M Very High Rates 2.2+ n ,

Alu/fnomoh

_

Figure 1.-Melanoma mortality in Oregon between 1958 and 1972. Mean annual age-standardized rates per 100,000.

264

OCTOBER 1979 * 131

* 4

CLUES TO THE CAUSE OF MELANOMA TABLE 1.-Trends in Melanoma Incidence and Mortality in Lane County, Oregon, Between 1958 and 1972* Incidence

Mortality

Both Males Females Sexes

Period

1958-1962 ... 1963-1967 ... 1968-1972 ... Total 1958-1972

3.2 5.6 7.7 5.7 (cases)[deaths](71)

Both Males Females Sexes

4.0 3.6 5.2 5.3 6.8 7.2 5.4 5.5 (75) (146)

0.9 0.0 2.5 0.5 2.4 1.8 1.9 0.9 [23] [12] *Mean annual age-standardized rates per 100,000.

0.4 1.4 2.1 1.4 [35]

TABLE 2.-Age Distribution by Gender of Melanoma Incidence and Mortality in Lane County, Oregon, Between 1958 and 1972* Incidence Males Females Both

Age (Years)

0-14 15-24 25-34 35-44 45-54 55-64 ..... 65-74 ..... 75-84 ..... 85 +. Rate for All Agest .. Total . .....

.....

.....

.....

.....

0.0 2.0 2.7 9.0 5.8 11.3 19.1 33.2 14.4

0.0 2.7 5.4 10.0 9.7 8.8 14.3 13.1 0.0

5.7 71

Mortality Males Females Both

0.0 2.3 4.0 9.5 7.7 10.2 16.6

0.0 0.0 0.0 3.0 2.6 4.3 2.9

22.0

23.4

2.8

0.0 0.0 0.0 2.4 1.6 3.9 3.5 11.7

6.0

0.0

0.0

0.0

5.5 5.4 75 146

1.9 23

0.9 12

1.4 35

0.0 0.0 0.0 1.8 0.7 3.6 3.9

ter 100,000.

*Mean annual age-specific crude rates p tStandardized to the age distribution of

population.

in Portland, from the Veteraris Administration Hospital tumor registry in Portl and and from the University of Oregon Medical S5 chool tumor registry in Portland. The large anci diverse medical community in Eugene has been a medical referral center for the surrounding larg ely rural counties for many years, so that very fey v persons go elsewhere for diagnosis and thera]py. Ocular melanomas were added to our dat:a because it did not seem logical to exclude a siIngle anatomic site of an uncommon disease. Mo rtality experience also was reviewed for Lane Cou[nty residents who died of cutaneous or ocular melanoma during these 15 years. For geographic analysis, cas;es were assigned to the census tract of residenc:e at the time of diagnosis or at time of death, re,spectively. Census tracts were grouped into geogiraphic regions or socioeconomic strata, the latte r according to a scoring system based on mediatn family income, proportion of families below t:he poverty level, adult education attainment and household crowding (code available on request) . For risk estimation we used mean annual rate's standardized by the direct method to the age dlistribution of the

TABLE 3.-Geographic Distribution by Gender of Melanoma Incidence and Mortality in Lane County, Oregon, Between 1958 and 1972* Geographic Region

Rural (Subtotal) . North ...... East ...... South ...... Midwest ..... Coast ...... Urban (Subtotal). NE Springfield SW Springfield S Eugene .... W Eugene .... NW Eugene .. N Eugene .... Core Eugene . Lane County Total ........

Incidence Males Females Both

5.9 3.5 4.8 9.2 4.7 6.8 5.6 3.5 4.6 4.5 2.0 3.3 7.4 2.8 5.2 4.1 6.8 5.4 5.4 6.4 5.9 4.4 10.6t 7.4 5.9 4.6 5.2 5.3 4.7 5.0 4.0 8.3 6.2 6.7 5.4 6.0 9.0 14.3* 11.8t

Mortality Males Females Both

3.6

3.3

3.4

2.2 3.1 1.7 2.9 1.4 1.7 1.7 0.0 3.7 0.5 2.7 2.4 2.1 1.5

5.7

5.4

5.5

1.9

0.7 0.0 1.0 0.0 0.0 3.4t 1.0 4.8t 0.0 1.0 1.3 0.7 1.6 0.6

1.4 1.6 1.3 1.4 0.7 2.5 1.3 2.3 1.7 0.8 2.0 1.2 1.8 1.0

0.9

1.4

*Mean annual age-standardized rates per 100,000. tP ' 05 Significance of difference from respective county rate.

population of the United States in 1970. For statistical judgment of comparisons of regional rates with the countrywide rate we used the statistic Z, equal to the rate difference (XoX0 ) divided by the square root of Xo(B X0) (1 /No + 1/NJ 1/T; where B is the rate base used, X1 is the regional rate being tested, X0 is the respective countywide rate, N1 and No are the mean or median populations from which X1 and X0 were calculated, and T is the number of years included in the mean annual rate. -

Results Table 1 shows that during this 15-year period, 146 new cases of melanoma were diagnosed in Lane County (5.5 per 100,000 per year), and 35 deaths resulted from melanoma (1.4 per 100,000 per year). According to these data, both sexes showed increasing incidence and mortality, with the exception of male mortality between 1968 and 1972. Men had consistently greater mortality and often had higher incidence rates as well. The total melanoma incidence rates in Lane County for 1968 to 1972 were higher than the 1969 to 1971 Third National Cancer Survey's cutaneous melanoma incidence rates by 67.4 percent for men and 54.5 percent for women,28 an excess that cannot be accounted for by the 4 percent of Lane County cases that were ocular melanomas. There were no melanoma cases or deaths among nonwhites, who make up only a small fraction of the population (0.72 percent THE WESTERN JOURNAL OF MEDICINE

265

CLUES TO THE CAUSE OF MELANOMA A

All Lane County 5.5 |mHigh Rates 70-10 4 E Very High Rates 10.5+

TABLE 4.-Socioeconomic Distribution by Gender of Melanoma Incidence and Mortality in Lane County, Oregon, Between 1958 and 1972* Incidence Males Females Both

Socioeconomic

Scoret

High 20-39

Eugene

Springfield

8.8 4.3 6.5 10.2 3.8 7.2 4.9 4.9 6.7 3.5 2.8 1.9 Total ......... 5.7 5.4

40-49 50-59 60-69 70-79 Low 80-89

....

....

....

....

....

....

6.4 8.5 5.5 4.9 5.1 2.5 5.5

Mortality Males Females Both

0.0 1.2 1.0 2.7 2.3 0.0 1.9

2.2 1.4 0.7 0.8 1.0 0.0 0.9

1.2 1.3 0.8 1.7 1.6 0.0 1.4

*Mean annual age-standardized rates per 100,000. tSum of unweighted scores for 1970 census variables: * family income * percent of families with income below poverty level * adult education attainment * household crowding

rates for women and for both sexes combined,

and because northeast Springfield had significant elevations of both incidence and mortality among Figure 2.-Melanoma incidence in Lane County, Oregon, between 1958 and 1972. Mean annual age-standardized rates per 100,000.

in 1960 and 1.45 percent in 1970), so that racial distribution patterns did not affect our observations. Table 2 illustrates the effect of age. Both sexes showed bimodal age distributions for both melanoma indices, with secondary peaks for the 35 to 44 year age stratum, but with maximum rates at ages 75 to 84 years for men and 65 to 74 years for women. Incidence rates were slightly higher among women than men at ages 15 to 54 years and higher among men than women 55 years old or older. The greater total male mortality was due almost entirely to a very high death rate in men aged 75 to 84 years (seven deaths). Age at diagnosis ranged from 15 to 89 years, while age at death varied from 35 to 83 years. Table 3 presents the geographic regional distribution of melanoma occurrence and mortality. The range of incidence rates was greater for women (2.0 to 14.3) than for men (3.6 to 9.2), which reflects classification by place of residence. This geographical distribution would be expected to be more accurate for women than for men on the average (due to lesser influence of occupational confounding) and is compatible with the hypothesis that place of residence influences melanoma risk. The contiguous urban regions of north Eugene and northeast Springfield seemed to warrant suspicion of high melanoma risk because north Eugene had significantly elevated incidence 266

OCTOBER 1979 * 131

* 4

women.

Figure 2 shows the distribution by individual census tracts for melanoma incidence in both sexes combined, which was very high (10.5 or more) in a group of five census tracts in north

and northwest Eugene. The peak rate of 12.0 occurred in two adjacent census tracts, the larger of which lies at the confluence of the McKenzie and Willamette Rivers. In the urban tracts, those with merely high rates (7.0 to 10.4) tended to cluster in the flat, moist bottom land adjacent to the peak-risk tracts, while rural high-risk census tracts included the major part of the agricultural land in Lane County bordering on Benton County (whose high-risk status was evident in Figure 1). According to the Sanitation Division of the Lane County Health Department, the high-risk melanoma region along the northern edge of both Eugene and Springfield has also been the region where the most active mosquito control efforts in the county have been required. From 1950 to 1963 DDT was the primary insecticide; since 1963 malathion and fenthion have been used. One might suspect DDT exposure as potentially important, given a recent report of excessive melanoma risk after exposure to polychlorinated biphenyls.20,21 The overall socioeconomic distributions of melanoma incidence and deaths are given in Table 4. Mortality in women showed the strongest tendency for a direct association with socioeconomic status. The same tendency for a positive correlation with socioeconomic status appeared in the incidence data, where it was manifested by

CLUES TO THE CAUSE OF MELANOMA

interpreted as a predominately rural epidemic in which the urban west, northwest and north Eugene regions also participated. Just before the 1965 outbreak, melanoma incidence had declined in most parts of the county, producing a strong impression of cyclical incidence pattern. Although division of the total time span into four shorter periods lessened the likelihood of recognizing significant local deviations from the respective countywide incidence rates, scattered significant deviations were evident (Table 5). In only one region, north Eugene, did the local rate consistently exceed the countywide rate during all four time periods, corroborating the earlier impressions about this region given in Table 3 and Figure 2. The rural north showed two incidence peaks of about the same magnitude occurring about a decade apart. Other small outbreaks took place in Springfield between 1962 and 1964, and on the coast between 1969 and 1972. After studying the geographic and temporal variations observed in melanoma incidence, it was natural to wonder if there existed a seasonal pattern of occurrence, as approximated by the date of diagnosis. Table 6 shows the months of diagnosis by year and by gender. From 1958 to 1963 there appeared to be some grouping of cases during the sunny months of June through October; however, the subsequent years showed a general year-round pattern of occurrence. Overall, there were four to nine new cases of melanoma in men in any one month, and peak months were June and August; in women there were between one and ten cases appearing in any one month, and peak months were January and May. Therefore, there was a difference between sexes but no simple pattern for either. We need to compare

both sexes, although less consistently. The effect of socioeconomic status on melanoma distribution seemed to be relatively weaker and more inconsistent than the geographic pattern, and neither effect was influenced by the proportion of darkskinned persons in the population (as judged by proportion of blacks by census tract). In Table 5 the melanoma cases were regrouped into four sets of rates rather than three because this temporal grouping showed more distinctly the onset of a rural epidemic of melanoma incidence that began in 1965. This phenomenon could not be accounted for by any change in diagnostic criteria or change in membership of the pathology group, nor by change in hospital service pattern of the pathologists, so that it is TABLE 5.-Trends in Melanoma Incidence in Geographic Regions of Lane County, Oregon, Between 1958 and 1972* Geographic Region

Rural (Subtotal) ..... North ............ East .............. South ............. Midwest .......... Coast ............. Urban (Subtotal) .... NE Springfield .... SW Springfield .... S Eugene ......... W Eugene ......... NW Eugene ....... N Eugene ......... Core Eugene ...... Lane County Total ...

19581961

19621964

19651968

19691972

2.2 10.2§ 1.7 1.5 0.0 0.0 4.2 4.5 3.3 5.1 3.8 0.0 8.3 4.5

0.6 0.0 0.0 0.0 3.1 0.0 3.8 12.4t 5.9 3.5 2.3 0.0 5.9 0.0 2.7

6.4 5.8 8.2 4.1 6.5 6.5 5.5 4.7 1.3 3.1 7.9

7.5 10.6 6.6 4.6 7.9 12.2

3.5

6.9 5.7 7.8 6.8 5.8 5.1 15.6t 4.5 7.1

11.0§ 9.3 4.4 5.8

*Mean annual rates per 100,000. tP

Epidemiologic clues to the cause of melanoma.

THE WESTERN Journal of Medicine Refer to: Morton WE, Starr GF: Epidemiologic clues to the cause of melanoma. West J Med 131:263-269, Oct 1979 Epidemi...
1MB Sizes 0 Downloads 0 Views