We sought to identify epidemiologic studies relevant to the research question: Which environmental trace element exposures are associated with skin cancer? Searches were performed in PubMed, Web of Science, and Embase (1972 –July 2018) with the terms “melanoma” OR “squamous cell carcinoma” OR “basal cell carcinoma” OR “keratinocyte carcinoma” OR “non-melanoma skin cancer” OR “skin cancer,” AND with “metals” OR “trace metals” OR “heavy metals” OR “minerals” OR “environmental exposure” OR “occupational exposure.” From this search we were able to select metals with existing literature relating to skin cancer. A secondary search included the above terms with “arsenic” OR “cadmium” OR “chromium” OR “copper” OR “iron” OR “selenium” OR “zinc” (presented in Tables 1–5). We also searched aluminum, beryllium, calcium, cobalt, lead, manganese, magnesium, mercury, and nickel given prior published possible associations with other cancers or role in normal skin development, homeostasis, and repair (12,13,15–19).

Studies reviewed reported exposure to one of the above-mentioned minerals in relation to risk of skin cancer in adult populations. All selected articles were original research, peer-reviewed, published in English, and specifically evaluated exposure to metal directly. If the full text of articles were unavailable, they were acknowledged in the text, but excluded in the tables. Only human epidemiologic studies were included. For example, nickel was excluded from the review since we found only non-human studies investigating nickel exposure and skin cancer (20–22). Only studies that explicitly investigated exposure to the mineral itself were included. For example, mercury was excluded from this review since it only has been studied indirectly with regard to occupations with possible exposure and risk of melanoma (23,24). Only minerals with literature suggesting a possible biological mechanism for risk of skin cancer were included. For example, lead was excluded given that there was no experimental data to suggest risk. Only one epidemiological study was found about lead exposure and skin cancer, and it was a case-control study examining toenail lead levels and melanoma risk, which reported no association (25). We found no studies on exposure to aluminum, beryllium, calcium, cobalt, manganese, and magnesium and risk of skin cancer. Thus, we excluded these elements. The elements we ultimately evaluated were arsenic, cadmium, chromium, copper, iron, selenium, and zinc.

Non-epidemiologic studies, including non-human experiments, were discussed in the text to supplement discussion of cancer risk. We included randomized controlled trials (RCTs), cohort, case-control, and cross-sectional studies. Ecological studies were discussed in text, but excluded from tables given the diminished quality of design with risk of data inaccuracy and difficulty to control for potential confounders among other limitations (26). Furthermore, ecological studies often investigated exposure to metals indirectly. However, some ecological studies were described in the text to help evaluate the totality of evidence. Details about study design, study population, exposure source, exposure measures, and results were recorded.

Included studies are shown in Tables 1–4, and briefly discussed in the text. For arsenic, zinc, and selenium, which had more than 5 studies available, flow charts of available studies were provided in supplementary material (Supplementary Figures 1–3). Based on quality of study design, more emphasis was placed on RCTs, followed by cohort studies, then case-control and cross-sectional studies, as the later studies are increasingly more prone to bias (26). This was also the order of discussion of the studies in text, and the order of listed studies in Tables 1–4. When multiple publications were available from the same population, we used the most recent publications and excluded earlier ones (25,27,28).

There are few studies that evaluate the association between arsenic exposure and melanoma (Table 1a). To our knowledge there are no RCTs that investigate arsenic exposure and risk of melanoma. A US cohort study found no association between exposure to arsenic-containing pesticides and melanoma (Table 1a) (53). Similarly, in a Danish cohort study, no association was found between exposure to arsenic in drinking water and risk for melanoma (Table 1a) (54). A US case-control study examined toenail arsenic exposure and melanoma using colorectal cancer patients as controls, and found an increased risk of melanoma with increasing toenail arsenic concentrations (Table 1a) (55). The association between arsenic exposure and melanoma risk needs to be evaluated in arsenic-endemic areas including Asian and Latin American countries (11,56). The effect of arsenic exposure on melanoma risk may be modified by genetic or constitutional factors, such as skin color and sun sensitivity, as Asian and Hispanic populations are more resistant than Caucasian populations to melanoma (52,57,58).

The link between arsenic exposure and KC has been more extensively evaluated (11,43–45), though to our knowledge, there are no RCTs, and existing studies are largely ecological in design. The characteristics of arsenic-associated skin tumors include SCC in situ, SCCs, and BCCs (59–61). The first evidence of arsenic’s carcinogenic effects were among patients treated with arsenic-containing compounds for psoriasis, and then later in Germans exposed to arsenic containing pesticides (44,45). Eventually, several regions with high levels of arsenic contaminated drinking water revealed a dose-related relationship between arsenic exposure and KC (11). For example, in 1968, Tseng et al. conducted an ecologic analysis of arsenic-contaminated drinking water and KC prevalence among 40,421 residents in 37 villages of Taiwan’s Blackfoot disease endemic region and found an 8-fold difference in skin cancer prevalence between the highest level of arsenic exposure to the lowest, with an increasing trend in skin cancer prevalence from low to high (62). In a retrospective cohort study of Taiwan’s arsenic-endemic villages, skin cancer risk was related to the duration of living in the endemic area, duration of artesian well-water consumption, average concentration of arsenic in the drinking water, and an index for cumulative exposure to arsenic (Table 1b)(63).

The association between environmental arsenic exposure and KC has subsequently been reported in Asia, Eastern Europe, Latin America, and the US (36,42,56,64–67). Several ecological studies in endemic regions have found elevated standard mortality ratios (SMRs) of skin cancer among populations exposed to drinking water with high arsenic concentrations (11,37,68–73). In Chile, SMRs for KC have ranged from 3.2 (95% CI = 2.1 – 4.8) (73) to 7.7 (95% CI = 1.3 – 6.6) (37). In Taiwan, increased SMRs of skin cancer among people in arsenic-endemic areas have also been reported (68–72).

A summary of cohort, case-control, and cross-sectional studies of arsenic and skin cancer is in Table 1b. Multiple studies were conducted within Asian countries. Two hospital-based case-control studies in Taiwan revealed increased percentages of urinary methylarsonic acid (MMA), an organoarsenic compound commonly used in herbicides, and increased urinary levels of other arsenic species among patients with KC compared with controls (74,75). In the US, a population-based case-control study found no association between toenail arsenic levels and risk of SCC and BCC among residents in New Hampshire (76,77). In another population-based case-control study among residents in New Hampshire, a positive association was found between urinary measures of arsenic exposure and risk of SCC (77). A case-control study in Hungary, Romania, and Slovakia found a positive association between residential water arsenic concentration and BCC risk (78). Increased incidence and prevalence of KC has also been reported among residents of Wisconsin’s Fox River Valley, which contains arsenic-rich minerals in its bedrock layers (79,80), as well as in Eastern Europe (64), Mexico (81), and Vietnam (66). Taken together, epidemiologic studies from different geographical regions have consistently supported the positive association between arsenic exposure and KC risk. Studies which have evaluated BCC and SCC separately have reported similar positive associations.

Future directions for arsenic exposure and KC investigation include developing a better understanding of pathogenesis and genetic susceptibility. Individuals can vary in their susceptibility to arsenic toxicity (82). For example, only 15–20% of exposed individuals show evidence of arsenic induced skin damage (82). This variability may be influenced by a combination of inherited genetic factors and environmental and lifestyle factors. For example, the chromosomal region that contains the arsenite methyltransferase (AS3MT) gene is subject to multiple variants, which can affect an individual’s ability to metabolize and excrete arsenic (82–87). Variations in this gene could ultimately impact that individual’s susceptibly to arsenic exposure and degree of toxicity and carcinogenicity (82,86,87). As emerging data supports a role for genetic variation in arsenic metabolism, it may become a promising method of skin cancer risk evaluation.

The present epidemiological literature regarding cadmium exposure and risk of melanoma is limited. In an Austrian cohort study, metallothionein overexpression was a significant prognostic factor for primary melanoma patients (111). In an Italian case-control study examining trace elements in the toenails of 58 melanoma cases and 58 controls, higher levels of copper and lower levels of iron were found in patients with cutaneous melanoma, but no differences for cadmium (Table 2a) (25).

In experimental studies, metallothionein expression is associated with melanoma progression and has been suggested to be a poor prognostic indicator (109,112,113). In murine organ samples exposed to cadmium, melanoma cell invasion was enhanced through the induction of metallothioneins (110), suggesting a possible role for metallothioneins in malignancy and metastasis (110).

Despite cadmium being considered a co-mutagen with UVR (104), relatively little has been studied with regard to cadmium and KC. To our knowledge, there are no epidemiologic studies investigating cadmium and KC. Lansdown and Sampson administered percutaneous cadmium chloride solutions to shaved rats and found dermal hyperkeratosis and acanthosis and increased mitotic indices in epidermal cells (114), suggesting a possible interaction between cadmium and keratinocytes (108,115).

Only three epidemiological studies were found investigating iron exposure and risk of melanoma. A case-control study in the US investigated dietary intake of various vitamins and minerals, and found a non-significant inverse trend toward reduced risk of melanoma with increased dietary iron intake (Table 2b) (141). A case-control study in Australia evaluating nutrient intake also found an inverse association between dietary iron intake and risk of melanoma ( Table 2b) (142). Furthermore, an inverse association between toenail iron concentrations and melanoma risk was observed in an Italian case-control study (Table 2b) (25). These epidemiological studies contrast with experimental studies suggesting a possible protective role for iron in melanoma development (132). Further studies on the possible relation between reduced iron status and melanoma etiology are necessary.

There are no epidemiologic studies investigating iron exposure and KC. In a study measuring levels of iron, copper, and zinc in the skin with noninvasive diagnostic x-ray spectrometry, all three elements were increased in both BCCs and SCCs compared with skin of healthy controls (143). In another histochemical examination of invasive BCCs and SCCs, only copper, not iron or zinc, were detected (144). In untransformed HaCaT and transformed A431 human keratinocytes, co-exposure with arsenic and iron was found to synergistically promote malignant transformation of untransformed keratinocytes, and progression of transformed keratinocytes (145). Despite possible associations between iron and UVA-induced skin damage, further studies are needed to elucidate a relation between iron and KC.

To our knowledge, there are only two small population-based studies on environmental copper exposure and risk of melanoma. An Italian case-control study found increased risk of melanoma with higher toenail copper levels (Table 2c) (25). Another case-control study in Spain found no association between serum copper levels and melanoma (Table 2c) (157). Further studies are needed to better elucidate a potential connection between copper consumption and melanoma.

The epidemiological literature on copper and KC in humans is limited. In a case-control study, copper levels were examined in 46 patients with BCCs and controls, and no difference was found in dietary consumption of zinc or copper (Table 2d) (158). In another case-control study, ceruloplasmin, a major copper carrying protein in the blood, was noted to be decreased in patients with AKs and BCCs compared with controls (159). Further epidemiologic studies are needed to better elucidate the potential relationship between copper and KC.

Given the potential role of copper in tumorigenesis, as seen in other cancer patients, one would expect increased risk of skin cancer with increased copper levels (153). Studies measuring levels of copper in BCCs and SCCs both noninvasively and using histochemical techniques have detected increased amounts of copper compared with uninvolved skin or skin of healthy controls (143,144). A murine study noted incidental development of SCCs at or near sites of nickel-copper alloy ear tags in 8.8% of mice compared with 0% in the untagged ears, suggesting that chronic topical exposure to one or both of the metals is carcinogenic (21). Some studies have also suggested that a lower level of copper may confer a risk of KC, particularly lower levels of copper as a cofactor for antioxidant enzymes. Immunohistochemical stains of skin cancer biopsies have demonstrated reduced levels of CuZnSOD in AKs and SCCs, but increased levels in BCCs (160). Others have also found lower levels of CuZnSOD in SCCs and BCCs and surrounding tissues compared with younger-aged control skin (161). In a murine study, promotion and progression of papillomas, keratoacanthomas, and SCCs were found to be inhibited by pretreating with copper(II) (3,5-diisopropylsalicylate) 2, a superoxide dismutase agent with copper as a cofactor (162).

There are six epidemiologic studies on environmental trace zinc exposure and risk of melanoma. Some studies have found an inverse association between zinc exposure and risk of melanoma. In a US ecological study using state-averaged cancer mortality rate data for Caucasian Americans during 1970–94, indices for dietary zinc were found to be inversely correlated with melanoma mortality rate (183). In a population-based case-control study in the Czech Republic, lower serum zinc concentrations were found among subjects with melanoma (Table 3a) (184). In another case-control study, an inverse association was found between dietary zinc intake and risk of melanoma in Australians (Table 3a) (142).

Conversely, there are studies that have found positive associations between zinc exposure and risk of melanoma. Two hospital-based case-control studies found increased serum zinc concentrations among melanoma patients (157,185). Another hospital-based case-control study found increased zinc concentrations in melanoma lesions compared to uninvolved skin of cases and skin of healthy controls (Table 3a) (143). There are also studies that have found no significant associations between zinc intake and melanoma (Table 3a) (25).

There is limited epidemiologic data regarding trace environmental zinc exposure and KC. In the same US ecological study investigating zinc and melanoma mortality, zinc and state-averaged KC mortality rate data was examined, and the dietary zinc index was also found to be inversely correlated with KC (183). In a case-control study, zinc levels were examined in patients with BCCs and cancer-free controls, and no difference in dietary consumption of zinc or copper was found between both groups (Table 3b) (158). Further studies are needed to better elucidate a potential connection between zinc exposure and KC.

Experimental study results are also mixed. As discussed with copper, immunohistochemical stains of AK, SCC, and BCC biopsies have shown reduced levels of CuZnSOD compared to skin of controls (161). Conversely, increased levels of zinc in BCCs and SCCs compared with skin of healthy controls has been demonstrated using noninvasive diagnostic x-ray spectrometry (143). In another histochemical examination of invasive BCCs and SCCs, zinc was not detected (144).

There have been 7 epidemiological studies of selenium and melanoma (Table 4a). In a US double-blind randomized placebo controlled trial among those with a history of cutaneous BCC or SCC, 200 μg/d of selenium supplementation was not effective in reducing melanoma risk (28). In two US prospective studies, no association was found between either toenail selenium concentrations or self-reported selenium supplement use and melanoma (209,210). Two case control studies similarly revealed no association between plasma and toenail selenium concentrations and melanoma (25,211,212). Conversely, some studies have found an association between selenium exposure and melanoma, though these studies in comparison to RCTs and cohort studies are more prone to bias given limitations in study design. An Italian prospective study found that exposure to tap water with high selenium levels was associated with melanoma risk (Table 4a) (213). In a case-control study, increased concentrations of plasma selenium were associated with increased risk of melanoma among an Italian population (214). In the same study, toenail and dietary selenium exhibited no evidence of a relation with melanoma risk; this difference could have been in part due to differences in specific selenium compounds (Table 4a) (214).

Based on these studies, selenium has not shown any beneficial role against melanoma risk. A few studies suggested potential adverse effects of selenium. In a murine study, a dose-dependent difference was found with selenium and melanoma development, with moderate dosage increasing tumor growth, and high dosage effectively treating and preventing recurrence of fully malignant tumors (215). In vitro studies have shown selenium inducing dose-dependent apoptosis in human A375 melanoma cell lines by inducing mitochondria-mediated oxidative stress (216). Taken together, these studies demonstrate the need to further investigate the exposure classification of selenium biomarkers, and metabolism of selenium to elucidate the potential relation between selenium exposure and melanoma risk.

There are multiple epidemiologic studies investigating selenium exposure and risk of KC, including RCTs and prospective cohort studies. A double-blind RCT investigated whether 200 μg/d selenium as selenized yeast could prevent KC among BCC and SCC patients from the Eastern US (28). They found that selenium supplementation in fact elevated risk for SCC (relative risk [RR] = 1.25, 95% CI = 1.03 – 1.51) and total KC (RR = 1.17, 95% CI = 1.02 – 1.34), but not BCC (Table 4b) (28). A sub-study of the trial then tested 400 ug/d of selenium supplementation and found no effect, while those who continued to receive 200 μg/d of selenium maintained a higher risk of SCC (RR = 1.88, 95% CI = 1.28 – 2.79) and KC (RR = 1.50, 95% CI = 1.13–2.04) (Table 4b)(217). In a small trial among 184 French organ graft recipients, 200 μg/d selenium-supplementation had no effect on skin cancer (218). Case-control or cohort studies in the UK or US have not found an association between dietary, serum, or supplemental selenium and BCC (158,212,218–223) or SCC (212,220,223,224) (Table 4b). A meta-analysis evaluating selenium supplementation and cancer risk found non-significant positive associations between selenium and KC with 4 included studies (RR=1.23 [95% CI 0.73–2.08]) (207). In summary, the effect of selenium exposure on risk of KC is inconclusive despite relatively large numbers of existing epidemiological studies, while there is some suggestion of positive association with SCC risk.

The suggested positive association contradicts some experimental studies. Selenomethionine, a selenium organic compound, when applied topically for two weeks at increasing concentrations was effective in protecting against acute UV damage to the skin (225). In human keratinocytes, p53 activation was significantly diminished when incubated in selenomethionine both pre and post UVR irradiation (226). In another in vitro study with human keratinocytes exposed to UVR, a reduction in apoptosis was found by 71% when cells were incubated with selenomethionine or sodium selenite (227). A similar reduction in apoptosis had been noted in prior studies (228). Given selenium’s increasingly popular role as a dietary supplement (207) it is important to better understand the relation of this element to skin cancer.