A Prospective Study of Toenail Trace Element Levels and Risk of Skin Cancer

Nonmelanoma skin cancers (NMSC) are the most common malignancy in the United States. Basal cell carcinoma (BCC) comprises 80% of all NMSC, and squamous cell carcinoma (SCC) largely comprises the remaining 20% (1–3). Melanoma, although not as common as NMSC, is the most lethal form of all skin cancers and comprises the majority of skin cancer-related deaths (4). The incidence of BCC, SCC, and melanoma in the United States is on the rise and represents substantial morbidity with significant social and economic burden (3–8). Growing evidence supports the role of environmental and lifestyle factors, beyond ultraviolet radiation (UVR; refs. 3, 4, 9–12), in the development of skin cancer, including exposure to heavy metals such as arsenic (As; ref. 13).

Exposure to some of these metals, such as chromium (Cr), is considered a risk factor for numerous cancers and have been classified as a carcinogen by the International Agency for Research on Cancer (IARC; ref. 14). Exposure to trace elements, including Cr, iron (Fe), mercury (Hg), selenium (Se), and zinc (Zn), has been hypothesized to play a role in both carcinogenesis of the skin and prevention of skin cancer, but evidence is limited (15–22). A recent review examining published epidemiologic literature about trace element exposure and risk of skin cancer concluded that too little data existed to draw any conclusions about exposure to Cr, Fe, and Zn and risk of skin cancer (22). Although Se was hypothesized to lower skin cancer risk, the review suggested increased risk with NMSC. No epidemiologic studies exist on exposure to Hg and skin cancer risk (22).

To investigate the potential association between toenail levels of Cr, Fe, Hg, Se, and Zn and risk of skin cancer, we conducted a prospective study based on subsets of participants with existing toenail metal levels in two large US cohorts.

The study population comprises a subset of participants from the Nurses' Health Study (NHS) and Health Professionals Follow-up Study (HPFS). In 1976, NHS enrolled 121,700 female US registered nurses ages 30–55 years. In 1986, HPFS enrolled 51,529 male US health professionals (dentists, veterinarians, pharmacists, optometrists, osteopathic physicians, and podiatrists) ages 40–75 years. Both cohorts were followed with similar biennial questionnaires about medical history, lifestyle, and disease incidence. Follow-up rates generally exceed 90% in both cohorts. The study was approved by the human subject committees of Brigham and Women's Hospital and Harvard School of Public Health. Details of the two cohorts and skin cancer–related findings have been described elsewhere (23, 24).

Toenail clippings were provided by 33,737 (66% of HPFS) participants in 1987 and by 62,865 (52% of NHS) participants in 1982 to 1983. Demographics and lifestyle characteristics were similar among participants with and without toenail clippings (25). For the current analysis, we used existing data on toenail metal concentrations from nested case–control studies of incident cardiovascular disease (CVD; coronary heart disease and stroke) in both cohorts as well as additional toenail data sets of diabetes cases and controls from HPFS (25–28). We included incident CVD cases and their matched controls (matched on month of toenail sample return, age, race, and smoking status), and diabetes cases and their matched controls, with toenail metal levels. In HPFS, the CVD nested case–control studies excluded dentists due to possible high occupational Hg exposure (27). But the data set of diabetes cases and controls did not exclude dentists (28). We did not do any exclusion based on occupation in this study.

A total of 7,124 participants from NHS and 4,448 from HPFS with toenail metal concentrations and no prior history of cancer were available at toenail collection. Due to low skin cancer incidence among non-white participants (4), we excluded non-Caucasian participants. After these exclusions, 6,708 participants from NHS and 3,730 from HPFS remained in analysis. Toenail As levels were assessed only in a few hundred participants, and thus we did not assess As in the current study.

Toenail concentrations of Cr, Fe, Hg, Se, and Zn were measured, while unaware of clinical data, using neutron activation analysis as previously described (25, 27, 28), at the University of Missouri Research Reactor Facility. Nail clipping samples from all toes were combined which, due to the elimination half-life of the metals, the growth rate of toenails, and the differential length of time (distance) from cuticle synthesis to time of clipping across the smallest to the largest toes, provided a time-integrated measure of metal exposure over approximately the previous year. Element determinations were performed in analytical batches between 2007 and 2011 for Se and 2009 and 2011 for Cr, Fe, Zn, and Hg. Potential laboratory drift was controlled by both standard comparison procedures for neutron activation analysis and repeated analysis of representative sample subsets. For quality control, toenail metal samples were co-analyzed with hair metal samples, which correlated well. Intra-assay coefficients of variability were 6.0% for Cr, 3.7% for Fe, 5.5% for Hg, 2.4% for Se, and 1.8% for Zn (25, 26, 29). In the NHS, Spearman correlation coefficients for the stability of toenail metals collected 6 years apart were 0.33 for Cr, 0.43 for Fe, 0.56 for Hg, 0.48 for Se, and 0.58 for Zn (P < 0.003 for all elements; ref. 30). Validation of these methods was reported for Se and Hg (30–34). Toenail Hg levels were correlated with fish intake in subsets of the cohort samples (30, 34, 35). Dentists had more than twice the level of Hg in toenails than non-dental health professionals (34), most likely due to occupational exposure.

Skin cancer diagnosis was asked in biennial questionnaires mailed to NHS and HPFS cohorts starting in 1984 (NHS) and 1986 (HPFS). Participants who reported melanoma or SCC were asked for permission to review their medical and pathology reports. Only confirmed invasive cases of melanoma and SCC were used. Medical records were not obtained for self-reported BCC; however, previous validation studies have demonstrated high validity of self-reported BCC in both cohorts, with 96% of cases in women and 84% of men confirmed by pathology records (36, 37). Differences in skin cancer ascertainment between BCC and SCC may have in part accounted for a markedly greater incidence of BCC than SCC.

We controlled for assay batch, and skin cancer risk factors including age (continuous), natural hair color (red, blonde, light brown, dark brown, black), number of moles on arms (0, 1–2, 3–5, ≥6), skin reaction to sun exposure for ≥2 hours as a child or adolescent (none/some redness, burn, painful burn/blisters), number of severe sunburns (0, 1–2, 3–5, ≥6), first-degree family history of melanoma (yes/no), history of physical examination (yes/no), and UVR exposure at residence. UVR exposure was measured as annual UV-B flux in mW/m², and was created by combining the latitude, altitude, and cloud cover of each participant's current state of residence, collected every 2 years. Biennial questionnaires updated body mass index (BMI; <25.0, 25.0–29.9, 30.0–34.9, ≥35.0 kg/m²), recreational physical activity (quintiles), smoking status (never, past, current), alcohol consumption (none, 0.1–4.9 g/d, 5–9.9 g/d, 10–19.9 g/d and 20+ g/d), and citrus consumption (quintiles). Additionally, we determined fish consumption using intake information on canned tuna, dark-meat fish, and other fish from food frequency questionnaires. Participants were asked how often on average they consumed the specified amount of fish during the past year. Intake was reported in nine categories: ≤1 time per month, 1–3 per month, 1 per week, 2–4 per week, 5–6 per week, 1 per day, 2–3 per day, 4–5 per day, and 6+ per day. Participant's fish intake was computed as the sum of these sources.

We evaluated associations of toenail metal concentrations with incident skin cancer using Cox proportional hazards models from the time of toenail sampling until skin cancer diagnosis, death, or end of follow-up (return date of the 2012 biennial questionnaire), whichever came first. Participants from NHS were followed from 1984 to 2012 and HPFS from 1986 to 2012. Metal concentrations were evaluated as quintiles for BCC and quartiles for SCC and melanoma (due to smaller sample sizes). We created quantiles based on distribution of toenail metals in women and men separately. We conducted a multivariable analysis adjusting for skin cancer risk factors. We calculated hazard ratios (HRs) for skin cancer according to metal concentrations. Participants were assigned the median value in their category of metals for trend tests as continuous variables. In additional multivariable analyses, we adjusted for BMI, physical activity, alcohol consumption, and smoking status. We conducted separate analyses for BCC, SCC, and melanoma. We performed study-specific analyses and calculated pooled HRs using a fixed-effects model. P values for heterogeneity were calculated using Q statistics.

We conducted several sensitivity analyses on BCC that (i) additionally adjusted for fish intake in the multivariable analysis of Hg levels, (ii) mutually adjusted for all metal levels simultaneously, (iii) mutually adjusted for Hg and Cr levels, and (iv) was among controls of the original nested case–control studies. We conducted analyses for Hg levels and BCC risk stratified by skin cancer risk factors and fish intake and calculated P values for interaction by these factors. We also conducted a stratified analysis by follow-up time period. SAS software version 9.2 (SAS Institute Inc.) was used for all statistical analyses. All statistical tests were two-tailed, and alpha criterion was set at < 0.05.

The study population mean (SD) age was 54.7(±6.2) years for women and 59.7(±9.1) years for men at time of toenail collection (Table 1). Toenail metal levels were higher in men than in women. Distribution of skin cancer risk factors and other lifestyle factors was generally similar among women and men, except women had a higher proportion of smokers than men. Men were older, had higher alcohol intake, and higher proportions of childhood blistering reactions to sun and 6+ blistering sunburns.

Participants with higher toenail Hg levels were more likely to consume fish, citrus, and alcohol, as well as more likely to exercise, and live in areas with greater UV exposure among both women and men (Table 2). Among men, participants with higher Hg levels were less likely to be smokers. The Spearman correlation coefficients between toenail Hg levels and fish intake were 0.34 (P < 0.0001) in women and 0.44 (P < 0.0001) in men.

During 26–28 years of follow-up, 2,433 BCC (1,553 women and 880 men), 334 SCC, and 130 melanoma cases were documented. Average follow-up times for diagnosis of BCC were 14 years in NHS and 12 years in HPFS. Table 3 presents the associations between toenail element concentrations and risk of BCC. Higher toenail Hg levels were associated with increased risk of BCC; the pooled multivariable HRs for the increasing quintiles of toenail Hg levels were 1.00 (reference), 1.00 (0.88–1.15), 1.13 (0.99–1.28), 1.27 (1.11–1.44), and 1.34 (1.18–1.52; P_trend < 0.0001). The positive association was consistent in women and men. Also, Cr was positively associated with BCC in women [top vs. bottom quintiles HR = 1.37 (95% CI, 1.17–1.60), P_trend < 0.0001[, but not in men. No associations were observed between other elements and BCC risk. The results remained similar when we additionally adjusted for BMI, physical activity, smoking status, alcohol consumption, and citrus intake [top vs. bottom quintiles HR for Hg = 1.27 (95% CI, 1.11–1.45)]. We conducted a multivariable analysis mutually adjusting for all of the metals simultaneously. The association between Hg levels and BCC risk remained similar and still significant [top vs. bottom quintiles HR for Hg = 1.36 (95% CI, 1.19–1.57)]. Similarly, the positive association between Cr levels and BCC risk in women remained significant [top vs. bottom quintiles HR for Cr = 1.33 (95% CI, 1.13–1.55) adjusted for BMI, physical activity, smoking status, alcohol consumption, and citrus intake, and 1.32 (95% CI, 1.12–1.56) adjusted for all metals simultaneously].

Because fish intake is the major source of Hg in the United States (38, 39), and correlated with Hg levels in the cohorts, we adjusted for fish intake in the analysis of Hg levels and BCC risk. The pooled multivariable HR for top versus bottom quintiles of Hg levels was slightly attenuated to 1.24 (95% CI, 1.08–1.41, P_trend = 0.003) after adjustment of fish intake. Fish intake was similarly positively associated with BCC risk; the pooled multivariable HRs for top versus bottom quintiles of fish intake were 1.33 (95% CI, 1.17–1.52, P_trend < 0.001). When Hg levels were adjusted for simultaneously, the HR for top versus bottom quintiles of fish was somewhat attenuated to 1.23 (95% CI, 1.07–1.42; P_trend = 0.004). We also adjusted for omega-3 fat intake, because fish is a major source of omega-3 fat. The pooled multivariable HR for top versus bottom quintiles of Hg levels remained essentially unchanged [HR = 1.30 (95% CI, 1.14–1.49), P_trend <0.0001] after adjustment of omega-3 fat intake.

We conducted stratified analyses of toenail Hg levels and BCC risk by fish consumption and known skin cancer risk factors including hair color, sunburns, skin reaction to sun exposure, and UV exposure at residence among women and men (Table 4). This was to explore whether Hg exposure might interact with any of the skin cancer risk factors, which may provide some biological explanations regarding how Hg could affect skin cancer. None of the P values for interaction were significant. Higher toenail levels of Hg were associated with BCC risk among those with low fish consumption [top vs. bottom quintiles HR = 1.46 (95% CI, 1.18–1.81), P_trend < 0.0001] and with brown/black hair [top vs. bottom quintiles HR = 1.37 (95% CI, 1.18–1.59), P_trend < 0.0001]. The associations were similar and significant by experience of severe sunburns, skin reaction to sun, and UV exposure. The results were similar in women and men (Supplementary Table S1). We also conducted a stratified analysis by follow-up time (up to 1996 vs. 1996–2012). The associations between Hg levels and BCC risk were similar by follow-up time [top vs. bottom quintiles HR = 1.43 (95% CI, 1.17–1.75) for up to 1996 and 1.27 (95% CI, 1.07–1.50) for 1996–2012]. We also conducted a sensitivity analysis among those who were controls of the original nested case–control studies; the association between Hg levels and BCC risk remained essentially unchanged [top vs bottom quintiles HR = 1.29 (95% CI, 1.07–1.55)]. Hg levels in the controls (0.51 μg/g in HPFS and 0.32 μg/g in NHS) were similar to those of non-controls (0.56 μg/g in HPFS and 0.31 μg/g in NHS).

We found similar direct associations between toenail Hg levels and risks of SCC [Table 5; pooled multivariable HR for top vs. bottom quartiles = 1.41 (95% CI, 1.03–1.94), P_trend = 0.04] and melanoma [Table 6; pooled multivariable HR for top versus bottom quartiles = 1.88 (95% CI, 1.12–3.16), P_trend = 0.02]. Fish intake had a nonsignificant, positive direction of association with SCC [pooled multivariable HR for top vs. bottom quartiles = 1.19 (95% CI, 0.88–1.61)] and melanoma [pooled multivariable HR for top versus bottom quartiles = 1.31 (95% CI, 0.76–2.26)]. The associations between toenail Cr levels and SCC risk was similar to those of BCC in women [Table 5; top vs. bottom quartiles HR = 1.41 (95% CI, 0.92–2.15)]. However, toenail Cr levels were not associated with melanoma risk in either women or men. Other elements were not associated with risks of SCC and melanoma in both women and men.

In this prospective study of toenail trace elements and skin cancer, we found a positive association between toenail Hg levels and risk of BCC. The association was consistent in women and men. We also found direct associations for SCC and melanoma, with limited statistical power. Other metals, except Cr in women, were not associated with skin cancer risk.

Hg, specifically as its organometallic species methylmercury, is a well-known nephro- and neurotoxin, and teratogen that primarily affects central nervous system development (38, 40–43). However, its association with cancer risk is less understood. Non-occupational exposure to Hg is predominantly through the consumption of seafood, particularly carnivorous fish, which contain methylmercury (38, 39). Although dental amalgams, which represent elemental (metallic) Hg, have been hypothesized to elevate Hg exposure (40, 44), previous evaluation in HPFS did not find that amalgam restoration was associated with toenail Hg levels (34).

Based on the ability of Hg to bind sulfhydryl groups, some have posited various molecular mechanisms for genotoxicity including generation of free radicals and oxidative stress, influence on DNA-repair mechanisms, inhibition of mitotic spindle formation, and direct interaction with DNA molecules (40, 45). Methylmercury chloride promoted renal tumors in animal studies (38, 46, 47). Human studies investigated methylmercury exposure and risk of cancer in Minamata, Japan, a site of methylmercury intoxication, and found increased mortalities due to liver cancer and leukemia (48–50). In a few small case–control studies conducted in Sweden on use of seed disinfection with organic Hg compounds, no associations were found with risk of soft-tissue sarcomas or lymphomas (51–54). Studies among mercury miners and chloralkali workers who were potentially exposed to inorganic Hg compounds have suggested potential increased risk of brain, renal, and lung cancers (38, 55–58). As such, the IARC classified methylmercury as a “potential” human carcinogen (group 2B), but metallic Hg and inorganic Hg compounds as not classifiable (group 3; ref. 38).

To our knowledge, our study was the first epidemiologic study of directly measured Hg exposure and risk of skin cancer (38). Some studies indirectly assessed Hg through possible occupational exposure and risk of melanoma. For example, dentists are exposed to Hg through dental amalgams. Three studies in Canada, Scandinavia, and Europe found that compared with the general population, dentists were at higher risk of melanoma with relative risks (RR) ranging 1.46–2.11 (59). A cohort study in Swedish women found that dental nurses were at higher risk of melanoma [RR = 1.71 (95% CI, 1.21–2.42)] than the whole population (19). That study also reported elevated melanoma risk among female hat makers [RR = 2.62 (95% CI, 1.41–4.88)] and described that mercuric nitrate was used in the production of felt hats. However, one US study found no association between dentists and melanoma risk (RR = 0.87; ref. 60).

Cr in its hexavalent state (VI) causes cancer of the lung and has been associated with cancer of the nose and nasal sinuses based on IARC (14). We are aware of only one epidemiologic study of Cr and skin cancer. A population-based case–control study in Italy examined toenail trace elements levels in 58 melanoma cases and 58 controls and found no association with toenail Cr levels (61). We found that increased toenail Cr levels were associated with BCC risk only among women independent of other metals including Hg. We do not have any biological explanations for the gender difference. There was no report of gender difference between Cr and other cancer sites in the IARC report (14). Given the paucity of data on Cr and skin cancer, further studies are needed to confirm our findings.

We found no association between Fe, Se, and Zn and risk of skin cancer for women or men. Although there are no epidemiologic studies on Fe exposure and skin cancer, Fe plays a key role in catalyzing redox reactions, which, in the presence of UVA radiation, can produce reactive oxygen species and play a role in UVA-mediated skin cell damage (62). There have been some epidemiologic studies of Se exposure using dietary, blood, and toenail levels of Se as well as supplemental Se and skin cancer (22). Some studies suggested a possible positive association between Se and SCC, but no association with risk of BCC or melanoma (21). A meta-analysis evaluating selenium supplementation and cancer risk found nonsignificant positive associations between selenium and NMSC with 4 studies [RR = 1.23 (95% CI, 0.73–2.08); ref. 21]. However, we found no association between toenail Se levels and any of the skin cancers. The epidemiologic studies on Zn and risk of skin cancer are limited, with case–control studies, and results that range from inverse to positive associations (22). Our findings using prospective data do not support any link between Zn exposure and skin cancer.

The prospective design, large number of BCC cases, evaluation in two independent cohorts, and ability to adjust for many skin cancer risk factors all represent major strengths of our study. Although we were limited by a single assessment of Hg exposure, toenail concentrations provide an excellent biomarker of integrated annual exposure. The reasonable correlation between concentrations in nails collected 6 years apart indicates that a single measurement represents exposure over longer periods (30). The findings were generally similar in two separate cohorts. However, these cohorts comprised of largely Caucasian and educated adults, potentially limiting generalizability of our findings. Still, the Hg levels were comparable to those from national data (27). We had limited sample size for SCC and, especially, melanoma, though similar directions of associations for metals and these skin cancers were found with BCC. SCC and melanoma cases were pathologically confirmed. BCC cases were documented by self-report, which could lead to a higher chance of misclassification compared with SCC or melanoma. Although we controlled for several risk factors of skin cancer, we cannot exclude residual confounding by those factors or other unknown factors. For example, we were unable to account for As exposure, which is a known NMSC carcinogen (13). However, in a small subset with toenail As levels (n = 202 NHS; n = 233 HPFS participants), the correlation between toenail Hg and As levels was low [Spearman r = 0.08 in NHS (P = 0.28) and 0.19 in HPFS (P = 0.005)]. The correlation between toenail Cr and As levels was similarly low. Therefore, although we cannot rule out the potential confounding by As exposure, it is likely to be minimal. Despite dentists comprising a large proportion of HPFS, we included a small number of dentists because we used existing data sets with toenail values that largely excluded dentists. Future evaluation of Hg levels and skin cancer risk among dentists would be of interest due to occupational Hg exposure (27).

In conclusion, our prospective data suggest that exposure to Hg, and potentially Cr, may increase skin cancer risk. Our results on toenail Hg levels and skin cancer risk were consistent by gender, type of skin cancer, and several skin cancer risk factors. Our findings need to be replicated in other populations and further investigated in terms of relevant biological mechanisms.

No potential conflicts of interest were disclosed.

All human studies have been performed in accordance with the ethical standards as laid down in the Declaration of Helsinki and reviewed and approved by an appropriate ethics committee.

Conception and design: N.H. Matthews, T. Li, A.A. Qureshi, E. Cho

Development of methodology: N.H. Matthews, T. Li, W.C. Willett, J.S. Morris, E. Cho

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.-Q. Li, W.C. Willett, M.J. Stampfer, J.S. Morris

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Koh, W.-Q. Li, T. Li, D.C. Christiani, J.S. Morris

Writing, review, and/or revision of the manuscript: N.H. Matthews, M. Koh, W.-Q. Li, W.C. Willett, M.J. Stampfer, D.C. Christiani, J.S. Morris, A.A. Qureshi, E. Cho

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Koh, A.A. Qureshi

Study supervision: N.H. Matthews, A.A. Qureshi, E. Cho

We would like to thank the participants and staff of the Nurses' Health Study and Health Professionals Follow-up Study for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data. This work was supported by the NIH (CA186107, CA167552, and CA198216) and by the Research Career Development Award of Dermatology Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.