In this study we examined melanoma risk in relation to occupational exposure to pesticides among pesticide applicators in Iowa and North Carolina. The chemical subcohort approach, used in other reports, provides information on all (cancer) outcomes associated with a specific chemical and allows the AHS to provide dose–response information that may inform future risk assessments. The case–control approach used here allows us to consider all factors, not just chemicals, associated with a specific cancer such as melanoma. The AHS pesticide applicators were not shown to be at an increased risk of melanoma relative to populations of these two states (Alavanja et al. 2005
), but additional evaluations of melanoma are warranted in light of previous literature. Commonly reported risk factors such as sun sensitivity and sun exposure (Dennis et al. 2008
) were associated with melanoma in this cohort.
The strongest pesticide associations were with maneb/mancozeb (a dithiocarbamate fungicide) and parathion (an ethyl or methyl insecticide). In addition, dose–response relationships were seen for two (benomyl and carbaryl) of four different carbamate pesticides. Our carbaryl finding supports a previous report from a prospective analysis of carbaryl applicators in this cohort (Mahajan et al. 2007
) with 2 additional years of follow-up (36 additional cases among those completing the take-home questionnaires). Another previous report that focused on organochlorine insecticides within the cohort noted an association between melanoma and toxaphene for lifetime exposure but not for intensity-weighted lifetime days of exposure (Purdue et al. 2006
), a measure that takes into account factors such as protective clothing that may modify exposure. We did not see an association with melanoma cases diagnosed through 2005 and toxaphene for intensity-weighted lifetime days of exposure. The data suggested a possible association between melanoma and arsenical pesticides. Although arsenic exposure was limited, arsenical pesticides appeared to modify the effect of benomyl and maneb/mancozeb pesticides independent of age.
The hypothesis that melanoma may be related to pesticides stems from the relationships among epidermal melanocytes, nevi, and the development of melanoma. Dermatitis related to pesticide exposure was described in 1921 (McCord et al. 1921
). Other skin diseases or irritations related to pesticides have been reported, including a case report of erythema multiforme related to parathion (Spiewak 2001
). A review of 12 studies of farmers found that 8 showed an excess risk of melanoma (7 for other nonmelanoma skin cancers) (Blair and Zahm 1991
; Spiewak 2001
), but it is unclear how much of this is related to sun exposure compared with pesticides or other exposures. A study of white Ranch Hand Vietnam veterans found an increased risk of melanoma related to dioxin exposure and herbicide exposure (Akhtar et al. 2004
). An additional report of an increased standardized incidence ratio for melanoma among Pan Britannica Industry’s pesticide factory workers suggests that pesticides are related to the development of melanoma (Wilkinson et al. 1997
). A more recent study found an association with cutaneous melanoma and a longer duration of residential pesticide use (Fortes et al. 2007
). They found that the most common compounds for indoor pesticides used in these residents included pyrethroids and carbamates. Additional evidence has shown that pesticides, carbon tetrachloride, and formaldehyde are related to increased risk of intraocular melanoma (Holly et al. 1996
). We had too few cases of intraocular melanoma to examine this association.
We did not find other analytic studies that have reported an association with maneb/mancozeb or parathion and melanoma. In this large cohort of pesticide applicators, we only found about 7% of applicators had applied these pesticides; thus, the exposure rate in the general population is likely to be low. However, a study of banana plantation workers in Costa Rica reported an increased standardized incidence ratio for melanoma (Wesseling et al. 1996
). Chemicals used on bananas include maneb, mancozeb, and benomyl, along with dibromochloropropane, chlorothalonil, and formaldehyde (Wesseling et al. 1996
). They saw that the risk of melanoma also increased with the number of years of employment at banana plantations. This provides further evidence of the potential association between melanoma and maneb/mancozeb and benomyl. For parathion, we did not find any study directly linking it with melanoma. Nevertheless, a laboratory study of sunscreen found that those containing the physical ultraviolet absorbers titanium dioxide or zinc oxide enhance the transdermal absorption of parathion (Brand et al. 2003
). In our study, when we further adjusted levels of parathion associated with melanoma for sunscreen use, we found no differences in the ORs. However, applicators were not asked about the details on types of sunscreen used or frequency or duration of use.
A link between arsenic and cancers of the bladder and lung and nonmelanoma skin cancer is well established. An association between arsenic and melanoma has only been reported in one other study to date, with an OR of 2.1 (95% CI, 1.4–3.3) for the highest quartile of toenail arsenic content (Beane Freeman et al. 2004
). Our data support the possible association between melanoma and arsenic that is not explained by age, but the data are limited by the rarity of exposure and lack of assessment of frequency or duration of exposure. The mechanistic pathways of arsenical carcinogenesis may include oxidative stress (An et al. 2004
; Shi et al. 2004
), ultraviolet enhanced mutagenicity (Chen et al. 2006
; Rossman 2003
), and genotoxicity or altered DNA repair (Huang et al. 1995
; Kochhar et al. 1996
; Mahata et al. 2003
). Arsenic may also work by an epigenetic mechanism that changes the function of the DNA without affecting the normal DNA sequence. Although many arsenical compounds have been discontinued in the United States, arsenical pesticides are still widely available in some countries, and some farms have leftover supplies that continue to represent some potential risk (Reigart and Roberts 1999
). Several studies of humans have shown an association between nonmelanoma skin cancer and heavy arsenic exposure via drugs, drinking water with a high arsenic content, or the occupational environment (Chen et al. 1985
; Guo et al. 2001
; Hsueh et al. 1995
; International Agency for Research on Cancer 1998
; Karagas et al. 2001
; Pesch et al. 2002
; Tseng 1977
). Most published studies examining arsenic exposure and skin cancer risk originate from Taiwan, Bangladesh, or China. Among these studies, only one specifically mentioned examining melanoma and did not find an association (Guo et al. 2001
); however, melanoma is rare in Chinese populations. An interaction has been demonstrated in one cross-sectional study where the risk of skin lesions associated with various levels of arsenic exposure was greater in those with excessive sun exposure (Chen et al. 2006
). We did not see an interaction with sun exposure in our data, but we had limited power to examine this.
The AHS has several strengths, including a prospective design, comprehensive pesticide exposure assessment, completeness of follow-up, and high participation rates. Previous analyses have shown that AHS applicators completing the take-home questionnaire were similar to those who completed only the enrollment questionnaire, with the exception that those completing the take-home questionnaire tended to be older (Tarone et al. 1997
). Our analyses (data not shown) of melanoma in association with pesticides detailed on the enrollment questionnaire showed magnitudes for all who completed the enrollment questionnaire similar to those when such analyses were restricted to subjects who also completed the take-home questionnaire. A comparison of the incident cutaneous melanoma cases reported in the overall cohort (those completing the enrollment questionnaire) and those who completed the enrollment and the take-home questionnaire showed similar distributions by histologic site and body site (Dennis et al. 2008
). Additionally analyses showed ORs similar to other studies for known sun sensitivity risk factors for melanoma. A limitation of this study was the small number of subjects who applied some of the pesticides, thus limiting the power of some analyses at this time.
Sun exposure, perhaps the strongest risk factor for melanoma, is difficult to capture via questionnaire. Because farmers spend a great deal of time in the sun, we cannot rule out the possibility that these pesticides-specific results are driven by sun exposure. However, results deferred for pesticides within a specific class, and within the limits of small numbers, were similar in Iowa and North Carolina. Furthermore adjusting for owning the farm or farm size (which might affect time outdoors) did not alter these findings. In addition, we had insufficient information on lifelong crop patterns to assess confounding by other factors potentially related to growing orchard fruits where arsenical pesticides were historically used. Finally, multiple comparisons may be an issue because we initially evaluated 50 pesticides. However, we initially focused on associations at the ≥ 0.01 significance level in the crude analyses (data not shown) and considered biologic plausibility. These results should also be interpreted with regard to their consistency with other studies.