Skip to main content

Evaluating H295R steroidogenesis assay data for robust interpretation

Description

Key highlights of this paper:

  • An extensive H295R database from various stakeholders has been analyzed.
  • Quality control acceptance criteria are not consistently achieved.
  • Pairwise significance testing contributes to the assay’s high positive hit rate.
  • Trend testing and/or 1.5-fold change threshold reduce the rate of positive outcomes.
  • Improving H295R data interpretation could help refine any subsequent testing needs.

The H295R steroidogenesis assay, performed according to OECD Test Guideline (TG) 456, is an important component of the Level 2 (IN VITRO) battery of assays in the OECD Conceptual Framework for Testing and Assessment of ED for evaluating endocrine mechanisms. The assay can inform on the potential of a chemical to interfere with the synthesis and metabolism of steroid hormones, in particular testosterone and estradiol.

An extensive analysis of a large compendium of H295R data has been conducted to addressthe observation by several stakeholders of an unusually high occurrence of positive outcomes, especially those of low efficacy, but nonetheless statistically significant, outcomes in the assay. The database consisted of OECD TG 456 guideline studies provided by 11 agrochemical companies and comprised of ~16,800 raw data entries, of which ~9,900 represented data for 50 test articles, and ~7,000 represented quality control (reference compound) data. The data were analysed using 4 different approaches: 1: analysis of pairwise concentration-specific significance compared to plate-matched vehicle controls (as per the guideline of 2011); 2: inclusion of a non-parametric Jonckheere-Terpstra trend statistic to confirm significant concentration-response trend across testing concentrations; 3:consideration of a 1.5-fold threshold for hormone changes prior to using the guideline statistics (as per the recently refined guideline, 2022); 4: integration of all three of the evaluated parameters in combination.

The highest number of positive interpretations for individual runs and overall assay outcomes was observed when using the pairwise comparison of OECD TG 456 (2011). The integration of either trend testing or consideration of the 1.5-fold change cut-off markedly reduced the number of positive interpretations. The number of final call outcomes for which no conclusion could be obtained was also improved using the additional criteria. Overall, theanalyses confirmed that the addition of efficacy cut-off and, to a lesser extent, trend testing criteria to the existing statistical evaluation can help reduce equivocal outcomes and identifyrobust positive calls in the H295R assay that are appropriate for decision making within a regulatory context.