The CALUX bioassay: Comparison of non-linear and linear transformed Hill regression models for dioxin potency assessment of sludge samples - Poster @ ChemCYS 2012

Polychlorinated dibenzo-p-dioxins, -furans and biphenyls form a group of ubiquitous contaminants that can cause adverse health effects, even at very small amounts.
Analytical methods for detection of dioxins are based on GC-MS confirmatory analysis or bioassay screening. The latter method (e.g. the mammalian based CALUX receptor assay) typically employs non-linear regression based on the 4 parameter Hill plot for calibration of the assays' response vs. TCDD standard concentrations and unknown samples.
A linear regression model approach1, based on the Box-Cox transformation of data, will be presented here as both an alternative and amelioration of the sigmoid Hill regression for determination of dioxin concentrations. The linear model is indeed expected to reduce the uncertainty assessment associated with a standard or sample curve and hence also the ratio of both for the BEQ (Bioassay Equivalence) prediction. Furthermore, the linearization of data overcomes the need of full dose sample curves.