Lung tumors in mice induced by “whole-life” inorganic arsenic exposure at human-relevant doses

Abstract

In mice, inorganic arsenic in the drinking water in the parts per million (ppm) range via the dam during in utero life or with whole life exposure is a multi-site carcinogen in the offspring. However, human arsenic exposure is typically in the parts per billion (ppb) range. Thus, we studied “whole life” inorganic arsenic carcinogenesis in mice at levels more relevant to humans. Breeder male and female CD1 mice were exposed to 0, 50, 500 or 5000 ppb arsenic (as sodium arsenite) in the drinking water for three weeks prior to breeding, during pregnancy and lactation, and after weaning (at week 3) groups of male and female offspring (initial n = 40) were exposed for up to 2 years. Tumors were assessed in these offspring. Arsenic exposure had no effect on pregnant dam weights or water consumption, litter size, offspring birth weight, or weight at weaning compared to control. In male offspring mice, arsenic exposure increased (p < 0.05) bronchiolo-alveolar tumor (adenoma or carcinoma) incidence at 50 ppb (51%) and 500 ppb (54%), but not at 5000 ppb (28%) compared to control (22%). These arsenic-induced bronchiolo-alveolar tumors included increased (p < 0.05) carcinoma at 50 ppb (27%) compared to controls (8%). An increase (p < 0.05) in lung adenoma (25%) in the 50 ppb group compared to control (11%) occurred in female offspring. Thus, in CD1 mice whole life arsenic exposure induced lung tumors at human relevant doses (i.e. 50 and 500 ppb).

Response to Druwe and Burgoon

“…We have determined that the conclusion, reported by Waalkes et al (2014), that lung tumors are induced in CD-1 mice by exposure to inorganic arsenic at human-relevant doses is not supported by the data presented. The key reasons for our position are: (1) lack of consistency and reproducibility (Garry et al 2015); (2) not taking into account the high variability in incidences of lung tumors in control CD-1 mice, including the variability in two studies reported by the Waalkes' group (Tokar et al 2011;Waalkes et al 2014); (3) lack of information regarding historical controls; (4) the statistical criterion of significance used by Waalkes et al does not follow the criterion of Haseman et al (1986); and (5) the biologic implausibility of the dose response being suggested.…”

“…Druwe and Burgoon (2016) employed Bayesian statistics that included modeling control tumor incidences as Bernoulli distributions and an approach that sets a region of practical equivalence (ROPE) to demarcate a region around zero difference that is functionally equivalent to no difference, to test the null hypothesis that the control tumor incidences from the Tokar et al (2011) and Waalkes et al (2014) studies are the same. The null hypothesis was accepted, and the authors concluded that the tumor incidences in both studies are likely from the same distribution.…”

“…Druwe and Burgoon (2016) next tested the null hypothesis that arsenic exposure does not increase the lung tumor response in male mice. A Bayesian approach was used which involved combining control incidences of lung tumors in CD-1 mice from three studies (Tokar et al 2011(Tokar et al , 2012Waalkes et al 2014). Druwe and Burgoon (2016) indicate that their analysis confirms that the Waalkes et al (2014) study demonstrates low-dose (50 and 500 ppb) inorganic arsenic increases lung tumor incidences in male CD-1 mice.…”

“…The key reasons for our position are: (1) lack of consistency and reproducibility (Garry et al 2015); (2) not taking into account the high variability in incidences of lung tumors in control CD-1 mice, including the variability in two studies reported by the Waalkes' group (Tokar et al 2011;Waalkes et al 2014); (3) lack of information regarding historical controls; (4) the statistical criterion of significance used by Waalkes et al does not follow the criterion of Haseman et al (1986); and (5) the biologic implausibility of the dose response being suggested. When evaluating the incidence of common tumors (background incidence ≥1 %, which the background incidence of lung tumors in CD-1 mice clearly meets), Haseman et al (1986) from the National Toxicology Program (NTP) recommended a p value <0.01 be utilized for statistical significance, not p < 0.05 as used by Waalkes et al (2014). Using this criterion, the mouse lung tumor data reported by Waalkes et al (2014) are not significant, even using concurrent controls for comparison.…”

“…When evaluating the incidence of common tumors (background incidence ≥1 %, which the background incidence of lung tumors in CD-1 mice clearly meets), Haseman et al (1986) from the National Toxicology Program (NTP) recommended a p value <0.01 be utilized for statistical significance, not p < 0.05 as used by Waalkes et al (2014). Using this criterion, the mouse lung tumor data reported by Waalkes et al (2014) are not significant, even using concurrent controls for comparison.…”