Tucker, SB; Lamm, SH; Li, FX; Wilson, R; Byrd, DM; Lai, S; Tong, Y; Loo, L
Skin cancer and arsenic dermatoses (hyperkeratoses and dyspigmentation) have been observed in a three village area of Huhhot, Inner Mongolia, China, in 1991. Public health authorities have attributed the disease to arsenic in the drinking water based on their public health studies. They have conducted medical examinations of residents (n=3,228), obtained their well-use histories (n = 3179), and measured the arsenic content in local wells (n = 184). Other sources of arsenic have been sought but not found. This report presents the results of the analyses of their data. It is the first report examining the occurrence of skin cancer and arsenic dermatoses where exposure data is analyzed on an individual basis.
The analyses examine two arsenic exposure measures [peak arsenic concentration (PAC) exposure measure and a cumulative arsenic dosage (CAD) exposure measure] and four dermatological disorder outcome measures [hyperkeratoses, dyspigmentation, hyperkeratoses with dyspigmentation, and skin cancer]. Dose-response relationship analyses across exposure strata were conducted using frequency-weighted, simple-linear, hockey-stick, and most-likely estimate (MLE) models. Formal latency analyses were conducted using the MLE model.
The peak arsenic exposures ranged from undetectable (<10 ppb) to 2,000 ppb arsenic. 35% of the population had PAC measures below 50 ppb, 69% below 100 ppb, and 86% below 150 ppb. The cumulative arsenic dosages ranged from undetectable to 20372 ppb-years arsenic with 71% below 2,000 ppb-years exposure.
Eight skin cancer cases were identified as were 172 cases of hyperkeratoses, 121 cases of dyspigmentation, and 94 cases with both hyperkeratoses and dyspigmentation. All eight skin cancer cases occurred in individuals with both hyperkeratoses and dyspigmentation and with peak arsenic concentrations of 150 ppb or greater. Although cases of hyperkeratoses and of dyspigmentation occurred with peak arsenic concentrations under 50 ppb, they did not reach expected prevalences until higher peak arsenic concentrations.
These analyses reveal a statistically significant deficit of skin cancer among those with PAC exposures below 150 ppb, and particularly among those with PAC level between 50 ppb and 150 ppb. The dose-response curve for skin cancer is best described with respect to the peak arsenic concentration (PAC) by a frequency-weighted model with a threshold at or near 150 ppb arsenic or by a most likely estimate hockey-stick model with a threshold at 122 ppb arsenic. These results are consistent with the threshold-model analysis of the Taiwan data set that had showed a threshold at about 120 ppb. Issues of time consideration, latency, and misclassification have been considered, but do not at present appear to have markedly affected the analysis. A number of different approaches have been used to deal with confounding due to age, including the use of age-adjusted rates and of stratified analyses. This study should be replicated or expanded to further answer analytic questions.