tions of arsenic with several internal cancers at exposure concentrations of several hundred micrograms per liter of drinking water. However, they provide few data about the degree of association at exposure concentrations below a few hundred micrograms per liter.
An extensive literature describes cases of skin and internal cancers following medicinal treatment with potassium arsenite (Fowler's solution) for a variety of conditions (Sommers and McManus 1953; Rosset 1958; Robson and Jelliffe 1963; Jackson and Grainge 1975; Popper et al. 1978; Prystowsky et al. 1978; Reymann et al. 1978; Nagy et al. 1980; Falk et al. 1981; Roat et al. 1982; Robertson and Low-Beer 1983), exposure to arsenical pesticides (Sommers and McManus 1953; Kjeldsberg and Ward 1972; Popper et al. 1978), or consumption of industrially contaminated drinking water or pesticide-contaminated wine (Roth 1957). The case reports and case series do not provide the needed data for quantitative risk assessment. However, the occurrence of these tumors in high numbers after long-term ingestion of arsenic in relatively young patients, or at anatomic sites where cancer is an extremely rare occurrence (e.g., liver angiosarcoma), increases the likelihood that many of the documented cancers were induced by arsenic. The observations also assist in identifying major cancer end points. The most common types of malignancy described in the reports are skin cancer, lung cancer, angiosarcoma of the liver (probably noted because of its rarity), prostate cancer, and bladder cancer. Reports of other cancers also appear: leukemia; other hematopoietic cancers; and cancers of the breast, colon, stomach, parotid gland, nasopharynx, larynx, buccal cavity, kidney, and others. Additional case reports describe internal cancers after the appearance of Bowen's disease, a type of superficial intraepidermal carcinoma that has been linked with arsenic exposure (Graham and Helwig 1959; Epstein 1960; Peterka et al. 1961; Hugo and Conway 1967).
The second group of studies comprises epidemiological investigations. Most of them did not provide the informational quality necessary for interpretation of dose-response relationships. However, many of the studies included data that are valuable in establishing the level of risk of particular internal cancers associated with a range of likely arsenic exposures (see Table 4-1). The form of arsenic was not specified in the epidemiological studies cited except for Cuzick et al. (1992), who observed mortality in a cohort of patients medicinally treated with potassium arsenite. Ecological studies are considered first, followed by cohort studies. Studies are summarized in Tables 4-1 through 4-6. When evaluating the epidemiological evidence to help judge whether arsenic in drinking water is a likely cause of internal cancers or other diseases, the subcommittee used the evaluation criteria that have been discussed by Hill (1965) and others (Cox 1972; Susser 1973; Rothman 1986).