Fitting a Model to Categorical Response Data With Application to Species Extrapolation of Toxicity

Abstract

The evaluation of toxicity data for noncarcinogens is complicated by the multiplicity of possible end points, and variations in both severity of effect and response rate. Often the response rates are not reported, so that "dose-response" analysis involves the relation between dose and severity of effect. Severity is usually reported as a description of the nature of the effects; measured values are rare. One approach is to then assign severity descriptions to ordered categories and to model the dose-category relationship. The application presented here is to interspecies scaling, i.e., the estimation of parameters in the model used to convert animal doses into equally toxic human doses.

Combination Exposure to Zidovudine plus Sulfamethoxazole-Trimethoprim Diminishes B-Lymphocyte Immune Responses toPneumocystis murinaInfection in Healthy Mice

“…Interspecies scaling in drug dosing has received much attention, as it is a continual problem when animals are used to model and investigate conditions in humans (7,48). Mice have been shown to require much higher doses of cytotoxic drugs than humans to produce similar levels of cell death (37,63). The extent to which the concentrations of SMX-HA, SMX-NO, and ZDV in mouse bone marrow correlate with those in the bone marrow of humans treated with these agents is unknown.…”

Zidovudine plus sulfamethoxazole–trimethoprim adversely affects B lymphocyte maturation in bone marrow of normal mice

“…Interspecies scaling has received much attention, as it is a continual problem when using animals to model and investigate human conditions [55,56]. Mice have been shown to require much higher doses of cytotoxic drugs than humans to produce similar levels of cell death [57,58]. The extent to which the concentrations of SMX-HA, SMX-NO, and AZT in mouse bone marrow correlate with humans treated with these agents is unknown.…”

Current intelligence bulletin 69: NIOSH practices in occupational risk assessment.

“…For modeling of continuous response data, individual animal data are strongly recommended, although information on central tendency and variability may be sufficient; typically, data on the number of subjects, mean of the response, and variability measure (e.g., standard deviation, standard error, or variance) for each group are adequate to perform the analysis. For dose-response modeling of categorical responses, the number of animals examined and the counts for each response category of each dose group are generally sufficient [EPA 2012b;Hertzberg 1989].…”