Page 10: Guidelines for Canadian Drinking Water Quality: Guideline Technical Document – Arsenic

9.0 Health effects

9.1 Effects in humans

The acute toxicity of the various forms and valences of arsenic in humans is predominantly a function of their rate of removal from the body. Metallic arsenic (0 valence) is not absorbed from the stomach and as such does not have any adverse effect. Some arsenic compounds, such as the volatile arsenine (AsH3), are not present in food or water. Additionally, some organic arsenic compounds have little or no toxicity or are rapidly eliminated from the body in the urine. Lethal doses for the most common arsenic compounds (AsH3, As2O3, As2O5, MMAV, and DMAV) in humans range from 1.5 mg/kg bw (As2O3) to 500 mg/kg bw (DMAV) (Buchet and Lauwerys, 1982). AsH3, As2O3, and As2O5 are gaseous forms of arsenic found in air, and MMA and DMA are organic forms of arsenic found in water.

Symptoms of acute arsenic intoxication associated with the ingestion of well water containing arsenic at 1.2 and 21.0 mg/L have been reported (Feinglass, 1973; Wagner et al., 1979). Early clinical symptoms of acute arsenic intoxication include abdominal pain and vomiting, diarrhoea, pain to the extremities and muscles, and weakness with flushing of the skin. These symptoms are often followed by numbness and tingling of the extremities, muscular cramping, and the appearance of a papular erythematous rash 2 weeks later (Murphy et al., 1981). A month later, symptoms may include burning paraesthesias of the extremities, palmoplantar hyperkeratosis, Mee's lines on fingernails, and progressive deterioration in motor and sensory responses (Fennell and Stacy, 1981; Murphy et al., 1981; Wesbey and Kunis, 1981).

Signs of chronic arsenicalism, including pigmentation and development of keratoses, peripheral neuropathy, skin cancer, peripheral vascular disease, hypertensive heart disease, cancers of internal organs (bladder, kidney, liver, and lung), alterations in gastrointestinal function (non-cirrhotic hypertension), and an increased risk of mortality resulting from diabetes, have been observed in populations ingesting arsenic-contaminated drinking water in southwestern Taiwan (Chen et al., 1985, 1992; Wu et al., 1989), Bangladesh (Smith et al., 2000), Chile (Borgono and Greiber, 1971; Zaldívar, 1980; Zaldívar and Ghai, 1980), India (Mandal et al., 1998), the United States (Valentine et al., 1982; U.S. NRC, 1999; U.S. EPA, 2001a), Mexico (Cebrian et al., 1983), and Canada (Hindmarsh et al., 1977). Dermal lesions, such as hyperpigmentation, warts, and hyperkeratosis of the palms and soles, are the most commonly observed symptoms in 70-kg adults after 5-15 years of exposure equivalent to 700 µg/day or within 6 months to 3 years at exposures equivalent to 2800 µg/day (U.S. EPA, 2001a).Footnote 2

Numerous adverse effects, particularly among children, have been associated with the consumption of arsenic-contaminated water in Antofagasta, Chile (mean arsenic concentration 0.6 mg/L). Effects on the skin (leukomelanoderma, hyperkeratosis), respiratory system (chronic coryza, cough, bronchopulmonary diseases), cardiovascular system (myocardial infarction, peripheral vascular disorders such as ischaemia of the tongue, Raynaud's phenomenon, acrocyanosis), and digestive system (abdominal pain, chronic diarrhoea) were observed in children under 16 years of age (Zaldívar, 1980; Zaldívar and Ghai, 1980). The prevalence of these symptoms decreased after the installation of a water treatment plant in 1972 (mean arsenic concentration 0.08 mg/L); however, prevalence rates were still higher than those of the control population (Zaldívar and Ghai, 1980). Dermal lesions in young people ingesting drinking water containing high arsenic concentrations have also been reported elsewhere (Tseng et al., 1968; Cebrian et al., 1983).

The largest epidemiological study on arsenic to date was conducted in a limited area of southwestern Taiwan (an area well known for its high incidence of blackfoot disease). This data set has been analysed by numerous authors (e.g., Tseng, 1977; Chen et al. 1985, 1992; Wu et al. 1989; U.S. NRC, 1999, 2001) for assessing the health effects of arsenic through ingestion of arsenic-contaminated drinking water. Tseng (1977) divided a population of 40 421 into three groups based on the arsenic content of their well water (high =0.60 mg/L, medium 0.30-0.59 mg/L, and low 0.01-0.29 mg/L). There was a clear dose-response relationship between exposure to arsenic and the frequency of dermal lesions, "blackfoot disease" (a severe peripheral vascular disorder) (Yu et al., 1984), and skin cancer. Despite certain methodological weaknesses in this early study, it is now widely accepted that exposure to high concentrations of arsenic is a cause of peripheral vascular disease. Blackfoot disease is now sometimes used as an indicator of exposure to arsenic (U.S. EPA, 2001b).

More epidemiological evidence for an association between the incidence of various cancers of the internal organs and the ingestion of arsenic-contaminated water comes from a study conducted in a limited area of southwest Taiwan. In this study, standardized mortality ratios (SMRs) for cancers of the bladder, kidney, skin, lung, liver, and colon were significantly elevated in the area of arsenic contamination. The SMRs for bladder, kidney, skin, lung, and liver cancer also correlated well with the prevalence rate for blackfoot disease (Chen et al., 1986). In an additional case-control study of 69 bladder, 76 lung, and 59 liver cancer mortality cases as well as 368 community controls matched for age and sex, the odds ratios of developing bladder, lung, and liver cancers for those who had used artesian well water for 40 or more years were 3.90, 3.39, and 2.67, respectively, compared with those who had never used artesian well water. Dose-response relationships were observed for all three cancer types by duration of exposure, and the odds ratios were not changed significantly when several other risk factors were taken into consideration in logistic regression analysis (Chen et al., 1986).

In an ecological analysis in which cancer mortality was examined in relation to arsenic concentrations in drinking water in the villages of the blackfoot disease-endemic areas of southwestern Taiwan, Chen et al. (1985) found an association between high-arsenic artesian well water (ranging from 0.35 to 1.14 mg/L; median level 0.78 mg/L) and cancers of the bladder, kidney, lung, skin, liver, and colon. Both the SMR and cumulative mortality rate were significantly higher for cancers of the bladder, kidney, lung, skin, liver, and colon compared with the general population of southwestern Taiwan. The SMRs for cancers of the bladder, kidney, skin, lung, liver, and colon were 1100, 772, 534, 320, 170, and 160, respectively, for males and 2009, 1119, 652, 413, 229, and 168, respectively, for females. A dose-response relationship was observed between the SMRs of the cancers and blackfoot disease prevalence rate of the villages and townships in the endemic areas. An additional ecological analysis of the same southwestern Taiwanese population by Wu et al. (1989) also found significant dose-response relationships for age-adjusted rates of cancers of the bladder, kidney, skin, and lung in both sexes and cancers of the prostate and liver in men (the total numbers of cancers at each site were 181 cancers of the bladder in both sexes, 59 cancers of the kidney in both sexes, 36 cancers of the skin in both sexes, 9 cancers of the prostate in men, 123 cancers of the liver in men, and 268 lung cancers in both sexes) (Wu et al., 1989). A study examining the ecological correlations between arsenic levels in well water and mortality from various malignant neoplasms in southwestern Taiwan demonstrated a significant association between the arsenic level in well water and cancers of the liver, nasal cavity, lung, skin, bladder, and kidney in both sexes and prostate cancer in men (Chen and Wang, 1990). A later reanalysis by Chen et al. (1992) on the same southwestern Taiwanese study population calculated cancer potency indices for liver, lung, bladder, and kidney. The study population was stratified into four groups according to the median arsenic level of well water in each village. There were 13 villages with median arsenic levels below 0.10 mg/L, eight villages with levels ranging from 0.10 to 0.29 mg/L, 15 villages with levels from 0.30 to 0.59 mg/L, and six villages with levels greater than or equal to 0.60 mg/L. The total numbers of cancer-related deaths observed during the study period were as follows: 140 male and 62 female liver cancer deaths, 169 male and 135 female lung cancer deaths, 97 male and 105 female bladder cancer deaths, and 30 male and 34 female kidney cancer deaths. Mortality rates were found to increase significantly with age for all cancers in both males and females. Significant dose-response relationships were observed between the ingested arsenic level and mortality from cancer of the liver, lung, bladder, and kidney in most age groups of both males and females.

Further support for the increased incidence of lung and bladder cancers from arsenic exposure is provided by Ferreccio et al. (2000) and Chiou et al. (2001). Both of these studies differed from the southwestern Taiwan ones (Chen et al., 1985; Wu et al., 1989), in that they examined the risk factors for newly diagnosed cases of bladder cancer (Chiou et al., 2001) and lung cancer (Ferreccio et al., 2000) rather than deaths. The study by Chiou et al. (2001) established a significant dose-response relationship between risk of urinary cancers and arsenic exposure after adjustment for age, sex, and cigarette smoking. This work was, however, limited in terms of its size. Ferreccio et al. (2000) revealed a clear association between the odds ratios for lung cancer and concentrations of arsenic in drinking water. While this work further supports the association of cancer with arsenic in drinking water, it has been deemed limited because of control selection methods used (U.S. NRC, 2001).

In a case-control study of 270 children with congenital heart disease and 665 healthy children, maternal consumption of drinking water containing detectable arsenic concentrations during pregnancy was associated with a threefold increase in the occurrence of coarction of the aorta. The prevalence odds ratio adjusted for all measured contaminants and source of drinking water was 3.4, with a 95% confidence interval of 1.3-8.9 (Zierler et al., 1988). However, there was no adjustment for maternal age, socioeconomic status, or previous reproductive history. Exposure was determined by matching the results of available water analyses for the water supplies serving the mothers to their dates of conception. However, for 101 of the mothers residing in communities served by multiple water supplies, it was necessary to average contaminant concentrations from more than one source in the community; the mean interval from the date of analysis to date of conception for the entire study population was 227 days.

In a case-control study in Massachusetts of 286 women with spontaneous abortions and 1391 women with live births, elevated odds ratios for miscarriages were associated with exposure to arsenic in drinking water (Aschengrau et al., 1989). The odds ratios for spontaneous abortion adjusted for maternal age, educational level, and history of prior spontaneous abortion for women exposed to arsenic in their drinking water at undetectable concentrations, 0.0008-0.0013 mg/L, and 0.0014-0.0019 mg/L were 1.0, 1.1, and 1.5, respectively. Exposure was determined by matching each woman to the results of a tap water sample taken in her city or town during pregnancy. However, the median interval from the date of matched metal analysis sample to the date of conception was 2.1 years, and it was reported that the variability of concentrations of metals in 20 Massachusetts towns and cities over the 7-year period between 1978 and 1985 was 10- to 100-fold. It would be desirable, however, to follow up these preliminary results in studies designed to more accurately assess exposure.

Although some effects have been observed in children and pregnant women, the U.S. NRC concluded that "there was insufficient scientific information to permit separate cancer risk estimates for potential subpopulations such as pregnant women, lactating women and children and that factors that influence sensitivity to or expression of arsenic-associated cancer and non-cancer effects need to be better characterized" (U.S. EPA, 2001a).

Most studies to date on arsenic exposure through drinking water have reported links between cancer and high concentrations of arsenic. A few recent studies in the United States, however, report no clear association between lung and bladder cancer risks and arsenic levels in drinking water that are lower than those reported in Taiwan (350-1140 µg/L). A historical case-control study by Steinmaus et al. (2003) looked at arsenic ingestion and bladder cancer incidence in individuals in seven counties in the western United States exposed to arsenic at concentrations ranging from 0 to greater than 120 µg/L. Cancer incidence was recorded from 1994 to 2000, and individual data on water sources, water consumption patterns, smoking, and other factors were collected for 181 cases and 328 controls. No increased risks of bladder cancer were observed for arsenic intakes greater than 80 µg/day (equivalent to ingesting 1.5 L of water daily containing arsenic at 53 µg/L) lagged over 5 years (odds ratio = 0.94, 95% confidence interval = 0.56-1.57). For similar intakes 40 or more years prior to diagnosis of cancer (i.e., 40 years or more lag period), the odds ratio was greater than 1 (odds ratio = 1.78, 95% confidence interval = 0.89-3.56); however, since the confidence intervals were quite large and include the null hypothesis (odds ratio of 1.0), it was concluded that there was no significant association between bladder cancer and arsenic exposure above 80 µg/day. For smokers with exposures greater than 80 µg/day 40 years prior to diagnosis of cancer, an odds ratio of 3.67 (95% confidence interval = 1.43-9.42) was reported, which provides some evidence that smokers ingesting arsenic at levels above 80 µg/day may be at increased risk of bladder cancer. Results from this study suggest that the latency period for arsenic-mediated carcinogenicity may be 40 years or longer, although conclusions from this study should be made with caution, given a few weaknesses in the authors' statistical analysis. These weaknesses include arbitrarily categorized arsenic levels (which may mask a potential dose-response relationship), very small sample sizes in the categories above 10 µg/L, and the use of odds ratios instead of person-years to calculate cancer incidence rates (odds ratios give only a snapshot of cancer incidence at a given time and dose).

A study by Lamm et al. (2004) reported no arsenic-related increase in bladder cancer mortality in 2.5 million white males ( from 1950 to 1979) with exposures ranging from 3 to 60 µg/L in drinking water within 133 U.S. counties in 26 states, with 65% of the counties and 82% of the population exposed to arsenic in the 3-5 µg/L range. However, it should be noted that the analysis of cancer risks using bladder cancer mortality data is limited, since bladder cancer generally does not result in mortality (U.S. EPA and Awwa Research Foundation, 2004).

In a similar study by the U.S. EPA and Awwa Research Foundation (2004), lung and bladder cancer incidence and mortality rates were examined in 32 U.S. counties in 11 states (comprising approximately 1.5 million people) with mean drinking water arsenic levels of 10 µg/L or greater during 1950-1999. No associations were observed between arsenic in drinking water at levels greater than 10 µg/L and incidence of, or mortality from, bladder or lung cancer. The authors cautioned that it is possible for elevated risks of lung and bladder cancer mortality or incidence to be present but not apparent in the analysis, since the analysis of cancer risks from bladder cancer mortality data is limited, given that people with bladder cancer generally do not die from it; the latency period between arsenic exposure and death from cancer is relatively long, so that migration and death from other causes may mask health outcomes from arsenic exposure; and an ecological study relates exposures and outcomes in groups of individuals that may not be representative of individual responses to arsenic exposure. The authors also indicated that further research is being conducted to confirm these results.

9.2 Effects in experimental animals and in vitro

Arsenic presents unique problems for quantitative risk assessment because there is no test animal species for studying carcinogenicity. It appears that test animals do not respond to inorganic arsenic exposure in a way that makes them useful as a model for human cancer assessment. Their metabolism of inorganic arsenic is also quantitatively different from that by humans (U.S. EPA, 2001a).

The specific form or valence of arsenic that is responsible for teratogenesis in animals is not known, although there is evidence to suggest that it is arsenite (As(III)) rather than arsenate (As(V)) (Hanlon and Ferm, 1986b).

There were significant reductions in cardiac output and stroke volume in male Wistar rats and female New Zealand rabbits ingesting drinking water containing As(III) at 50 µg/mL for 18 and 10 months, respectively. In contrast, there was no effect on cardiac function in rats following ingestion of the same concentration of As(V) for 18 months (Carmignani et al., 1985).

In a multi-organ tumour initiation-promotion study, Yamamoto et al. (1995) reported positive results in rat bladder, kidney, liver, and thyroid. DMA significantly enhanced tumour induction in the urinary bladder, kidney, liver, and thyroid gland in rats treated with DMA at 400 mg/L in the drinking water. Induction of preneoplastic lesions (glutathione S-transferase placental form-positive foci in the liver and atypical tubules in the kidney) was also significantly increased in DMA-treated rats. Ornithine decarboxylase activity in the kidneys of rats treated with 100 mg DMA/L was significantly increased compared with control values (P < 0.001). Subsequent studies have also shown positive results for promotion of carcinogenesis when examined in a single initiator-promoter protocol in the rat liver (Wanibuchi et al., 1997) and bladder (Wanibuchi et al., 1996).

Other studies have shown carcinogenic effects in mice and rats (IPCS, 2001), although many of the studies of carcinogenicity of arsenic in animals have resulted in negative findings (ATSDR, 2000). An extensive review of animal models of arsenic carcinogenicity is presented in U.S. NRC (1999), Kitchin (2001), and Wang et al. (2002).

Arsenic has been known to induce chromosome breakage, chromosomal aberrations, and sister chromatid exchange in a linear, dose-dependent fashion in a variety of cultured cell types, including human cells (Jacobson-Kram and Montalbano, 1985; U.S. EPA, 1988). Most of the chromosomal aberrations are lethal events, so that the cells do not survive more than one or two generations (U.S. EPA, 1988). Trivalent arsenic is approximately an order of magnitude more potent than As(V) in this respect. The clastogenic effect of arsenic appears to be due to interference with DNA synthesis, as arsenic induces sister chromatid exchange and chromosomal aberrations only when present during DNA replication (Crossen, 1983). Arsenic has also been shown to block dividing cells in the S and G2 phases (Petres et al., 1977). While the mechanism of arsenic genotoxicity remains unknown, mechanisms such as reactive oxygen species and the inhibition of DNA repair have been proposed (IPCS, 2001; WHO, 2003). Several possible modes of action for arsenic carcinogenesis, including chromosomal abnormalities, oxidative stress, altered DNA repair, altered DNA methylation patterns, altered growth factors, enhanced cell proliferation, promotion/progression, gene amplification, and suppression of p53, have been reviewed by Kitchin (2001).

In early studies, teratogenic effects of arsenic in chicks, golden hamsters, and mice were reported (Hood and Bishop, 1972; Zierler et al., 1988). Arsenate was found to be teratogenic in the offspring of pregnant hamsters following exposure on days 4-7 of gestation by minipump implantation (Ferm and Hanlon, 1985). The threshold blood level for teratogenesis was 4.3 µmol/kg (Hanlon and Ferm, 1986a). In studies with mice and hamsters, MMAV and DMAV have been considerably less teratogenic than As(III) or As(V). However, teratogenicity was not observed in mice or rabbits upon oral administration of arsenic acid at 48 mg/kg bw per day during gestation days 6-15 and at 0-3 mg/kg bw per day during gestation days 6-18 (Nemec et al., 1998).

9.3 Relative toxicity of arsenic compounds in humans

While earlier studies reported organic forms of arsenic (MMAV , DMAV, MMAIII, and DMAIII) to be less toxic than their inorganic counterparts (i.e., As(III) and As(V)) (U.S. NRC, 1999), recent evidence suggests that the conversion of inorganic arsenic into organic arsenic may not represent a detoxification pathway. In humans, MMAV and DMAV, as well as MMAIII and DMAIII, result from the sequential reduction and methylation of inorganic arsenic by the liver (Buchet and Lauwerys, 1985; Lovell and Farmer, 1985). Inorganic arsenic that is not immediately removed from the body undergoes these reduction and methylation steps. Recent isolation of MMAIII in urine from humans suggests that, contrary to previous belief, MMAIII is actually more toxic to hepatocytes than MMAV and arsenite (As(III)) (Aposhian et al., 2000; Petrick et al., 2000; Styblo et al., 2000; U.S. NRC, 2001). Work on human hepatocytes performed by Petrick et al. (2000) has established a relative order of toxicity: MMAIII > arsenite (+3) > arsenate (+5) > MMAV = DMAV. A study by Mass et al. (2001) provides some evidence that organic arsenic is more effective than inorganic arsenic in altering chromosomal integrity in cultured human lymphocytes and phage DNA. Both MMAIII and DMAIII were found to be more effective at inducing DNA damage than As(III). Although these studies provide some initial evidence that organic arsenic may be more toxic than inorganic arsenic, further research is required to confirm these findings.

Page details

Date modified: