Sunday, July 17, 2005

Mega mouse study - First look, no threshold

BrooklynDodger returns to yesteryear, before quantitative risk assessment became dominant. In those days, the issue was threshold for carcinogenesis by chemicals. Threshold means there is region where there is NO exposure-response relationship for a chemical - increasing doses below the threshold cause no increased response. If no threshold, then any reduction in exposure will bring a reduced risk, and the only zero risk is zero exposure.

[Quantitative risk assessment, fought vigorously by public health advocates, later came to drive protective OSHA and EPA standards. This force opponents of public health advocates to adopt a new methodology, the Houdini risk assessment, which adds untestable mechanistic hypotheses to reduce or eliminate predicted risks...a commentary for another day.]

A buzz surrounded the "mega mouse study," portrayed as a central mission of the National Center for Toxicological Research (NCTR); NCTR was the Pine Bluff, Arkansas, biological warfare center converted to peaceful use.

Mega-mouse recognized that treatment groups must be large to detect lower rates of response. The 50 animal bioassay can observe statistically significant results at a 10% incidence rate against a zero background, the ED10. The mega-mouse study aimed at finding an ED01, which now the Dodger can't figure is 1% or 0.1%. Probably 1%, given the background rate of these tumors in control animals. Even at 1/1000, the effect is 3 orders of magnetude above the 1/1,000,000 then talked about as the "virtually safe dose." Instead of the million mice implied, "only" 25,000 were used.

Unfortunately, the study compound, a model carcinogen, was a material of no commerical importance, therefore no application of this data other than theoretical was possible.

The answer was, initial models of the data, found no threshold for one site [liver], and maybe or maybe not a threshold for the other site, bladder. That's what NCTR published.

The results had no impact on the debate over threshold.

Subsequently, other statisticians [or the same biostatisticians moved to other institutions] remodeled this data to assert there was a threshold. Check ED01 on pub med.

PS: an early OSHA transcipter [coke oven] misheard and published the term "biased statistician."


J Environ Pathol Toxicol. 1980;3(3 Spec No):17-34.

Effects of dose and time in a long-term, low-dose carcinogenic study.

Littlefield NA, Farmer JH, Gaylor DW, Sheldon WG.

...Mice were exposed for up to 33 months to low doses of 2-acetylaminofluorene (2-AAF)... urinary bladder neoplasms and liver neoplasms, resulted in 2 different types of dose response relationships. Although bladder neoplasms exhibited a minimum effect level (or a nonlinear response) for specific conditions, the total results were consistent with a "no threshold concept." The late appearing liver neoplasms displayed a nearly linear type response that extrapolated directly to zero dose. ... a positive response was noted at the next lower dose as time was extended. Discontinuing dosing and sacrificing at 18 and 24 months also demonstrated the effects of exposure to the carcinogen 2-AAF. Induction of bladder neoplasms was shown to occur early in the study, but was dependent upon the continuous presence of 2-AAF. The liver neoplasms appeared very late in the study but were shown to be induced at a very early point in the exposures and did not require the continuous presence of the carcinogen in order to develop. A standard 18 month bioassay study, if conducted under the same conditions, would have classified this chemical as a weak acting carcinogen.

No comments: