CALUX measurements: Statistical inferences for the dose-response curve

M. Elskens, D. S. Baston, C. Stumpf, J. Haedrich, I. Keupers, K. Croes, M. S. Denison, W. Baeyens, L. Goeyens

Research output: Contribution to journalArticlepeer-review

24 Scopus citations


Chemical Activated LUciferase gene eXpression [CALUX] is a reporter gene mammalian cell bioassay used for detection and semi-quantitative analyses of dioxin-like compounds. CALUX dose-response curves for 2,3,7,8- tetrachlorodibenzo-p-dioxin [TCDD] are typically smooth and sigmoidal when the dose is portrayed on a logarithmic scale. Non-linear regression models are used to calibrate the CALUX response versus TCDD standards and to convert the sample response into Bioanalytical EQuivalents (BEQs). Several complications may arise in terms of statistical inference, specifically and most important is the uncertainty assessment of the predicted BEQ. This paper presents the use of linear calibration functions based on Box-Cox transformations to overcome the issue of uncertainty assessment. Main issues being addressed are (i) confidence and prediction intervals for the CALUX response, (ii) confidence and prediction intervals for the predicted BEQ-value, and (iii) detection/estimation capabilities for the sigmoid and linearized models. Statistical comparisons between different calculation methods involving inverse prediction, effective concentration ratios (ECR 20-50-80) and slope ratio were achieved with example datasets in order to provide guidance for optimizing BEQ determinations and expand assay performance with the recombinant mouse hepatoma CALUX cell line H1L6.1c3.

Original languageEnglish (US)
Pages (from-to)1966-1973
Number of pages8
Issue number4
StatePublished - Sep 30 2011


  • Bioassay CALUX
  • Box-Cox transformation
  • Dose-response curve
  • Statistical inference

ASJC Scopus subject areas

  • Chemistry(all)


Dive into the research topics of 'CALUX measurements: Statistical inferences for the dose-response curve'. Together they form a unique fingerprint.

Cite this