The first evidence that inorganic arsenic was associated with prostate cancer in humans came from Taiwan in the late 1980s (Chen et al. 1988
; ). This was a follow-up study that focused on dose–response relationships between arsenic and cancer in a population exposed to high levels of arsenic in the drinking water from local artesian wells. The population studied was from the area of endemic “blackfoot” disease in southwest Taiwan, a disease involving the peripheral vascular dysfunction likely due, at least in part, to arsenic exposure (Chen et al. 1985
). Although the original study had not looked at cancer of the prostate (Chen et al. 1985
), the subsequent study found a remarkable association between arsenic exposure and prostate cancer mortality in this population (Chen et al. 1988
). In this regard, the age-standardized mortality from prostate cancer in the group exposed to the highest levels of arsenic in the drinking water (≥0.60 ppm) was nearly 6-fold greater than that of the general population in Taiwan. In addition, when drinking-water arsenic levels were stratified (< 0.30 ppm, 0.30–0.59 ppm and ≥0.60 ppm), a significant dose–response relationship occurred between arsenic level and age-adjusted prostate cancer mortality. The exposed population lived in a relatively small area and had similar lifestyles, diets, living conditions, and sociodemographic characteristics compared with those of nearby unaffected villages, prompting the authors to conclude that the striking differences in cancer mortality between these groups could be explained “solely by the difference in arsenic concentrations in drinking water” (Chen et al. 1988
Epidemiologic studies of arsenic exposure and prostate cancer in humans.
Prostate cancer is not always fatal, particularly in its early stages, and as the cause of death was determined in this study by death certificate (Chen et al. 1988
), it is likely that the rate of deaths would be much lower than the incidence of prostate cancers in this population. There were also large increases in mortality from liver, lung, skin, bladder, and kidney cancers in this population due to arsenic exposure that generally exceeded the rate of prostate cancer deaths (Chen et al. 1988
). Therefore, other cancers may have overshadowed relatively rare cancers of the prostate. Furthermore, prostate cancer is usually a disease of older men, and because arsenic is a very effective, multisite carcinogen, perhaps some of the most sensitive subjects may have died of other arsenic-induced cancers before the development of advanced and deadly prostate cancer. Indeed, prostate cancer is considered to have a relatively low case-fatality rate (IARC 2004
), making mortality as an end point potentially insensitive of actual disease status, at least in the early stages.
A follow-up study to those of Chen et al. (1985
concerning arsenic and cancer mortality used some of the same population at risk but added data from additional villages in the area of endemic blackfoot disease and specifically studied dose–response relationships (Wu et al. 1989
). In this study, the age-adjusted mortality for prostate cancer in the population exposed to the highest arsenic levels in the drinking water (≥0.60 ppm) was nearly 10-fold higher (9.18 deaths/100,000) than that at the lowest level (< 0.30 ppm; 0.95 deaths/100,000) of exposure. A clear dose–response relationship also occurred between arsenic exposure and prostate cancer mortality when drinking-water levels of arsenic were stratified (< 0.30, 0.30–0.59, and ≥0.60 ppm) in this study. These interpretations must be tempered by the small number of cancer deaths due to prostate cancer in this study, but, nonetheless, the findings are consistent with the prior work (Chen et al. 1988
). Exposure levels were determined by median village levels of arsenic in drinking-water wells, and, as such, may be subject to the “ecological fallacy” that the association observed at the village level may not hold at the individual level (Wu et al. 1989
). Even after considering this and other confounding factors, the authors felt that arsenic content should still be strongly suspected as the main cause of excess cancer deaths in this population (Wu et al. 1989
In subsequent work from Taiwan, the study population was expanded from the area of endemic blackfoot disease used in the first two studies (Chen et al. 1988
; Wu et al. 1989
) to a much more comprehensive study of all 314 precincts and townships in Taiwan as a whole. In all, 83,656 wells were tested for arsenic (Chen and Wang 1990
). Based on multiple regression analysis with adjustments for urbanization and age, mortality rates from cancer of the prostate again increased in correlation with increasing average drinking-water level of arsenic.
In an independent study of the area of endemic blackfoot disease in southwest Taiwan, Tsai et al. (1999)
computed age-adjusted standardized mortality ratios (SMRs) using death certificates with national reference rates. The SMR for prostate cancer in the arsenic-exposed population was 1.96, with a 95% confidence interval (CI) of 1.4–2.6, indicating a significant increase in the number of observed cases compared with the number of expected based on the national reference rates. The number of observed cases in this arsenic-exposed population was 48, and dose–response effects were not investigated.
The role of drinking-water arsenic in prostate cancer mortality has also been studied in a U.S. population (Lewis et al. 1999
). Mortality was assessed in a retrospective cohort of Millard County, Utah, residents along with drinking-water arsenic exposure levels that accounted for residence time in the study area. The cohort consisted of 2,073 members with at least 20 years of exposure history and was assembled through membership records of the Church of Jesus Christ of Latter-day Saints. Arsenic exposure was stratified into low (< 1,000 ppb-years), medium (1,000–4,999 ppb-years) and high (≥5,000 ppb-years) levels (Lewis et al. 1999
). Without considering specific arsenic exposure levels, the overall SMR for prostate cancer mortality was significantly elevated in the cohort (1.45; 95% CI, 1.07–1.91, based on 50 deaths) compared with that of Utah white males. The authors indicate that SMR analysis hinted at a dose–response relationship when based on low (SMR = 1.07), medium [1.70 (significantly elevated)] and high (1.65) arsenic exposure (Lewis et al. 1999
In a study from Australia, geographic areas with soil arsenic > 100 mg/kg and/or drinking-water concentrations > 0.01 mg/L were selected and related to cancer incidence (Hinwood et al. 1999
). Standardized incidence ratios (SIRs) were generated for 22 areas of elevated arsenic exposure in Victoria and compared with all Victorian cancer rates as a baseline. For all areas with any elevated arsenic (soil or water or both), the SIR was significantly increased for prostate cancer (1.14; 95% CI, 1.05–1.23). Exposure was also stratified as only high soil or only high water arsenic (low) or both high soil and high water arsenic (high). When arsenic exposure was stratified by exposure type (i.e., high water only, high soil only, high water/high soil), the SIR for prostate cancer remained significantly elevated (1.20; 95% CI, 1.06–1.36), in the high water/high soil category. Dose–response analysis was performed on data stratified based on water content of arsenic as low (< 0.01 mg/L), medium (0.01–0.1 mg/L), high (0.1–0.2 mg/L), and very high (> 0.2 mg/L) levels. No linear dose response was detected for prostate cancer incidence using this water stratification, but based on graphical presentation, the SIRs for the high and very high categories appeared elevated (95% CIs did not include 1.0). The study included 619 cases of prostate cancer. The authors make the point that of those targets expected a priori
from other studies, only prostate cancer was significantly elevated.
In a population of male copper foundry workers industrially exposed to arsenic as well as other metals, a correlative survey of plasma neoplastic biomarkers was conducted (Szymanska-Chabowska et al. 2004
). A strong positive correlation occurred between urinary arsenic concentration and serum prostate-specific antigen (PSA). PSA is a well-established biomarker for prostate cancer that is considered a mainstay of early prostate cancer detection. The exposure to other metals complicates interpretation of this study, but the correlation between arsenic in the urine and circulating PSA was robust. In this regard, tumors arising from human prostate epithelial cells transformed by inorganic arsenic in vitro
also show a remarkable overexpression of PSA (Achanzar et al. 2002
The results of various positive studies of prostate cancer and arsenic exposure were considered as a whole by the IARC (2004)
. The specific conclusion was that “data from southwest Taiwan indicate a consistent pattern of increased mortality from prostate cancer in areas with high contamination by arsenic, and there is evidence of a dose-related effect” (IARC 2004
). Although the prostate was not specifically mentioned as a human target site in the final evaluation of the monograph, the implications of the text are clear and, at least in part, are supported by the data from the United States and Australia, which make it less likely that the Taiwanese are uniquely sensitive. Whatever the conclusion, the available evidence indicates an obvious need for additional studies of arsenic as a human prostatic carcinogen.
As a potential complicating factor in dose–response analysis, evidence indicates that arsenic can adversely affect testicular function in animals, even at levels near the range for some human exposure situations. This includes loss of testicular weight, diminished sperm count, and decreased 17β-hydroxysteroid dehydrogenase (17β-HSD) activity in mice chronically given 4 ppm arsenic in the drinking-water (Pant et al. 2004
). In this regard, 17β-HSD is an enzyme important in production of testosterone from its immediate precursors, such as androstenedione. Similarly, in rats chronic oral arsenic exposure decreases testicular weight, sperm count, testicular 17β-HSD activity, and plasma and testicular testosterone concentrations (Jana et al. 2006
). Prostate cancer, particularly in its early stages, is dependent typically on circulating androgens and will regress with orchiectomy and/or antiandrogen therapy, two strategies commonly used in prostate cancer treatment (Kyprianou and Isaacs 1988
). Thus, if higher doses of arsenic similarly suppressed testosterone production in humans, this could complicate the dose–response analysis by potentially diminishing carcinogenic response at higher doses. There is no direct evidence of this in humans, however.