Cancer Risk at Ultra-low Dose: Lessons Learned from 40,000-animal Cancer Dose-response Studies

Abstract

There continues to be a shortage of ultra-low dose/response animal data on which to conduct modeling of human cancer risk. We have conducted two large-scale cancer and biomarker dose-response studies, one with dibenzo[def,p]chrysene (DBC, formerly called dibenzo[a,l]pyrene DBP) and a more recent study with a‰atoxin B 1 (AFB 1), to address this need. These experiments used rainbow trout (Oncorhyncus mykiss), an animal model well suited to ultra low-dose carcinogenesis research, to explore dose-response down to a targeted 10 excess liver tumors per 10,000 animals (ED 001). In study one, 40,800 trout were fed 0-225 ppm DBC for four weeks, sampled for biomarker analyses, and returned to control diet for nine months prior to gross and histologic examination. Suspect tumors were conˆrmed by pathology, and resulting incidences were modeled and compared to the default EPA LED 10 linear extrapolation method. The study provided observed incidence data down to 2 above-background liver tumors per 10,000 animals at lowest dose (that is, an un-modeled ED 0002 measurement). Among nine statistical models explored, three were determined toˆt the liver data well-linear probit, quadratic logit, and Ryzin-Rai. None of theseˆtted models is compatible with the LED 10 default assumption, and all fell increasingly below the default extrapolation with decreasing DBC dose. Low-dose tumor response was also not predictable from hepatic DBC-DNA adduct biomarkers, which accumulated as a power function of dose (adducts＝100*DBC 1.31). Two-order extrapolations below the modeled liver tumor data predicted DBC doses producing one excess cancer per million individuals (ED 10-6) that were 500-1500-fold higher than that predicted by theˆve-order LED 10 extrapolation. Study two was of similar design, but using AFB 1. Analysis of the results is underway, and complicated by several diŠerences from study 1, especially presence in some quartiles and treatment groups of a fatty liver syndrome. Preliminary logistic regression analysis excludinĝ sh with this syndrome did not support the EPA linear default assumption (i.e., logistic slope 1.0), rather indicated a sublinear dose-response with slope of 1.42 (95% CI 1.23-1.61), and an extrapolated ED 10-6 that is 32-fold greater than the LED 10 default extrapolation. Inclusion of allˆsh also yielded a sublinear dose-response, with slope 1.31 (95%CI 1.13-1.50), and an extrapolated ED 10-6 17fold greater than the LED 10 default extrapolation. Thus two genotoxins with diŠering biological properties yielded ultra-low dose-response curves in the same animal model that are statistically incompatible with the linear default assumption. These results are considered speciˆc to the animal model, carcinogens, and protocols used. They provide theˆrst experimental estimations in any model of the degree of conservatism that may exist for the EPA default linear assumption for a genotoxic carcinogen.

Exposure to Ethylating Agents: Where Do the Thresholds for Mutagenic/Clastogenic Effects Arise?

“…Interestingly, however, a recent study on tumour induction in trout (36) exposed to the genotoxin dibenzo[a,l]pyrene, which induces a host of bulky adducts, has yielded strong evidence for a deˆnitely sub-linear dose response, despite a near linear induction of complex bulky DNA adducts down to very low doses. Similar conclusions have been drawn in a study with a‰atoxin as detailed in the presentation by G. Bailey (37). It is evident that linear back extrapolation from TD10 (or TD50) doses to the`virtual safe' dose (VSD, the dose inducing 1 cancer in a million individuals) yields a value that would be`too low' by orders of magnitude compared to the estimate by appropriate non-linear curveˆtting.…”

Letter to the Editor Regarding the Article by Bailey et al., 2009

“…Bailey et al report on a similar trout experiment involving aflatoxin but analyzed the data using the same approach as that used in the DBP study.…”