Is the human carcinogen arsenic carcinogenic to laboratory animals?

Abstract

Arsenic has long been known to cause cancer in humans (Hutchinson, 1987, 1988), and has been correlated convincingly with cancers of the skin, lung, liver, kidney, and urinary bladder (IARC, 1987; NTP, 2000). Paradoxically, we now know that arsenic has been shown to be “anticarcinogenic” as well, and of potential benefit in the treatment of acute promyelocytic leukemia (Zhu et al., 1999). Whereas this is a major cancer chemotherapeutic advance, we believe use of arsenicals in human medicine must be tempered by toxicological realities (Huang et al., 1998; Huff et al., 1999). Nonetheless, most if not all cancer chemotherapeutic agents are carcinogenic to animals, and cause eventual second primaries in humans. In summarizing the informative and valued proceedings from a recent symposium on arsenic, Goering et al. (1999) stated that “animal carcinogenicity data for arsenic is considered either negative or equivocal.” These authors make a more discerning statement two paragraphs later: “Animal bioassays are considered either flawed or incomplete for establishing carcinogenicity of arsenic in rodents.” They also contend that “the International Agency for Research on Cancer (IARC) has determined that arsenic is the only agent to be a definitive human carcinogen in the face of ‘inadequate’ evidence of carcinogenic potential in animals.” In fact, none of these statements is fully correct; the second quoted sentence above comes close. In 1987 IARC evaluated arsenic and arsenic compounds and concluded “there is limited evidence of carcinogenicity in experimental animals,” meaning there are tumor responses in some studies on arsenic carcinogenicity in animals but due to certain deficiencies (too few animals, low doses, short duration, no controls, arsenic mixtures) the findings were not considered sufficient evidence of carcinogenicity in animals. In these assorted and seemingly numerous studies, various forms of inorganic arsenic were associated with tumors of the lung, respiratory tract, and stomach (IARC, 1980, 1987). Moreover, as we have stated previously (Huff et al., 1998a,b), arsenic trioxide and other inorganic (and until now organic) arsenicals have in reality never been tested adequately for carcinogenesis, and never by the inhalation route. Thus, to state that arsenic has not been shown to cause cancer in laboratory animals is patently premature (Chan and Huff, 1997). Unfortunately, a common and perpetuated incorrect statement is often made that arsenic is carcinogenic to humans and not to experimental animals. Bioassays should be done, in particular with arsenic trioxide, to satisfy this frequently but wrongly opined discrepancy. As history often repeats, this reminds some of us of when benzene was also heralded as being carcinogenic to humans but not to animals, and the “lack” of carcinogenicity in animals for both arsenic and benzene was used to discount the relevancy of bioassay testing results for predicting human cancer risks. It is now clear that, after adequate testing of benzene in animals, the results have been overwhelmingly positive for carcinogenicity (Huff et al., 1989; Maltoni et al., 1989). We believe the same will occur if inorganic arsenic is tested properly in laboratory animals. In addition to arsenic, and contrary to the quoted comment by Goering et al. (1999), other IARC Group-1 human carcinogens have less than sufficient evidence of carcinogenicity in animals, largely because only one study has been reported, and in some cases there is no evidence at all because no bioassays have been done. These substances are listed in Table 1. For some of these there may be more recent experimental data that we are either unaware or have been unable to locate. Despite this collective observation, correlations between bioassays and epidemiology findings are excellent (Huff, 1994, 1998, 1999a; Tomatis et al., 1989; Wilbourn et al., 1986). And of course, in addition to these 10, plus arsenic, one must recognize that there are six mixtures, 13 industries or occupations, as well as 10 biological agents or viruses (Table 2) that are carcinogenic to humans, and that have not, and typically cannot, be studied in animals. Granted, for many of these, bioassays either cannot be done or can only be accomplished with great difficulty, although particular agents within an in1 To whom correspondence should be addressed. Fax: (919) 558-7055. E-mail: huff1@niehs.nih.gov. TOXICOLOGICAL SCIENCES 55, 17–23 (2000) Copyright © 2000 by the Society of Toxicology