Arsenic exposure is a major public health problem affecting >100 million people worldwide (1
). Being a systemic toxin, arsenic has been associated with the risk of skin and several internal cancers (of lung, bladder, liver, and kidney), as well as cardiovascular, respiratory and neurological diseases (3
). Typically, these diseases appear after decades of chronic exposure (8
). A fraction of chronically exposed populations, usually within a few years of exposure, presents with classic arsenical skin lesions, characterized by hyperpigmentation of the skin with or without palmoplantar hyperkeratosis (9
). Researchers have long been considering skin lesions as a clinical marker of arsenic susceptibility. However, because of the lack of large-scale, individual-level longitudinal investigations, the natural history and sequelae of arsenic-related pathology in humans have largely been unknown. The issue of whether classic skin manifestations are predictors of subsequent internal cancers and other systemic diseases has been a burning research question in the field. In the current issue of the Journal
, Hsu et al. (10
) have addressed this question for the first time using longitudinal investigations involving a sizeable number of patients with arsenical skin lesions compiled by the authors from cross-sectional and case-control studies from the 1980s and 1990s in a heavily exposed area in southwest Taiwan.
The authors found that individuals with hyperkeratosis and/or skin cancer had elevated risks of lung and urothelial cancers after adjustments for arsenic exposure and relevant covariates. These elevated risks were evident among individuals exposed to low, medium, or high levels of arsenic. Lung cancer risk was elevated among both smokers and nonsmokers but was much more pronounced among smokers, leading to a significant multiplicative interaction between smoking and skin lesions. Lung cancer risks were evident among both <60- and ≥60-year age groups, but urothelial cancer risks were evident mainly among the <60-year age group. Individuals with hyperkeratosis and/or skin cancer also had unexpectedly high risks of prostate cancer, and participants with skin cancers had elevated risks of all internal cancers and upper gastrointestinal cancers. Finally, after adjustment for the presence of skin lesions or skin cancer, arsenic exposure itself was associated with only urothelial cancer, while cigarette smoking was associated with only lung cancer among these individuals.
The major strengths of the study include efficient use of the previously established and current research resources, prospective nature of the study, and careful considerations to analyses. Despite the strengths of this paper, several potential weaknesses are noted. The first is the potential for exposure misclassification. Cumulative arsenic exposure (CAE) measurements were based on the median arsenic concentrations in the wells of each village in which the subjects lived. In Taiwan and other areas, arsenic concentrations can vary greatly from well to well, even among wells within the same village. So, applying the same median concentration to every person in a particular village likely led to misclassification of arsenic exposure in some people. This could be important in this study because evidence suggests that arsenic-related skin lesions typically occur at high arsenic exposure levels (11
). Thus, it is possible that, compared with those without skin lesions, those with skin lesions who were classified as highly exposed (e.g., CAE: ≥20 mg/L × years) are more likely to be correctly classified, and those with skin lesions classified as having low exposure (e.g., CAE: <1.0 mg/L × years) are more likely to be incorrectly classified. A potential consequence of this is that, at a given measured CAE level, those with skin lesions could actually have higher true exposures (and thus higher true risks of arsenic-related internal cancers) than those without skin lesions. This could potentially bias the relative risks reported in this paper, although the extent of this bias is unclear. Another potential weakness is the small number of cancer cases, particularly for prostate cancer, where the total number of cases was only 9. This low number results in low statistical power and increases the possibility that some of the high relative risks reported for this cancer are due to chance or bias. For example, potential bias could occur if there are differences in prostate cancer diagnostic rates between those with and without skin lesions or skin cancer. This could occur if those with skin lesions or skin cancer tend to visit their medical providers more often because of their condition and therefore are more likely to be encouraged to have prostate cancer screening examinations. Other studies have linked arsenic to prostate cancer, but the more convincing studies have all come from the same region of Taiwan where this study was done (12
). As such, studies in other populations may be needed to help confirm that the links between arsenic ingestion and prostate cancer identified here truly represent causal effects. Other possible weaknesses in this study may relate to the fact that 3 separate previously recruited subcohorts were combined for these analyses and that arsenic exposure had ceased many years before the assessments for skin lesions and cancer begun, although it is not clear that either of these factors introduced major bias.
Overall, the findings from this study provide one more line of evidence that individuals who were once chronically exposed to arsenic are at higher risk of internal cancers decades after the cessation of exposure. This type of information is important in planning medical care in areas where arsenic exposures have been high. This includes very large populations in Bangladesh, West Bengal, Chile, Argentina, Taiwan, the United States, and many other countries worldwide that have or have had elevated levels of arsenic in their water (13
). This study is also important in that it highlights the fact that susceptibility to arsenic-related disease varies greatly from person to person. Although some of the factors responsible for these differences in susceptibility have been identified, including diet (14
), genetics (15
), or coexposures like smoking (17
), many of the differences in susceptibility remain unexplained. Currently, many regulatory policies aimed at preventing arsenic-related diseases are based on risk estimates from studies in the general population, not on risks that may occur in susceptible groups (19
). Further research aimed at identifying those factors that confer susceptibility will aid in helping to develop policies that not only protect the “average healthy” adult, but also protect people who may be particularly susceptible to arsenic. These prevention efforts are important in not only highly exposed areas like Bangladesh or West Bengal, India, but also countries like the United States, where many people are exposed through the use of arsenic-contaminated private wells (20
) or through the consumption of foods that have recently been found to sometimes contain high concentrations of arsenic, for example, rice, milk, and organic foods using rice syrups (21
As mentioned above, arsenic exposure is associated with cardiovascular, respiratory, and other noncancer diseases (4
). Many of these diseases lead to premature death, especially in resource-poor settings where clinical and preventive care for these otherwise manageable diseases might be lacking (23
). Because these noncancer outcomes are much more prevalent than lung and urothelial cancers, future studies need to investigate if arsenical skin lesions are also predictive of cardiovascular and respiratory outcomes. Evaluation of the biological basis of the association of arsenical skin lesions with internal cancers and potentially noncancer outcomes is also important. If shared biomarkers (including genetic factors) that are predictive of both arsenic-related skin lesions and cancer/noncancer outcomes can be identified, then appropriate interventions could be designed among exposed populations even prior to the development of skin lesions, which is a much earlier event than the systemic diseases. In their conclusion, the authors state, “It is urgently recommended that individuals at risk with arsenic-induced hyperkeratosis and/or skin cancer should stop smoking and avoid exposure to any other lung carcinogen” (10, p. 211). Although we agree with this, it must be remembered that arsenic-related disease and mortality also occur in people without skin lesions. In fact, the majority of arsenic-related internal cancers, cardiovascular and respiratory diseases, and associated mortality happen among individuals without skin lesions or skin cancer. As such, the recommendations about avoidance of smoking and other carcinogens should not pertain just to arsenic-exposed individuals with skin lesions but, rather, they should apply to anyone with a history of chronic arsenic exposure.
In conclusion, within the constraints of the study settings, Hsu et al. (10
) addressed a difficult research question with careful rigor and reported results that highlight yet another dimension of the health consequences of arsenic exposure even decades after the cessation of exposure. The study findings not only reveal an avenue of risk assessment and prevention of internal cancers among populations with chronic arsenic exposure but also underscore the need for similar investigations for cardiovascular, respiratory, and other arsenic-related long-term health consequences. Given the emerging evidence of arsenic exposure from dietary sources beyond contaminated drinking water and occupational and environmental settings, and also because the vast majority of diseases and deaths among exposed populations do not show classic dermatological manifestations, larger and more comprehensive investigations of the health effects of arsenic exposure, especially at lower doses, are needed. In parallel, because the risk of known arsenic-related health outcomes remains elevated decades after exposure, cessation research toward identification of early clinical and biological markers of long-term risk as well as avenues for prevention, in addition to policy actions for exposure reductions, is warranted.