A Decision Logic for the Reliability Assessment and Interpretation of Vitellogenin Measurements

The egg-yolk precursor protein vitellogenin (VTG) is a biomarker for the determination of in vivo endocrine activity of chemicals in animals. Measurements of VTG in fish and amphibians are included in Organisation for Economic Co-operation and Development (OECD) test guidelines to provide support for identifying potential endocrine-active substances acting on the estrogen, androgen, and steroidogenesis (EAS) pathways. Induction of VTG in male fish is often associated with estrogenic activity, whereas inhibition in female fish may be related to substances that inhibit estrogen synthesis. The VTG protein or mRNA is measured in the plasma, liver, or whole body of fish, depending on the species and developmental stage, and on the specific test guideline requirements. Concerns have been raised regarding the variability of VTG measurements in fish, which could challenge the reliability and acceptability of VTG results for use in regulatory assessment of chemicals (Brown et al., 2023). Hence, it is important to correctly measure and interpret VTG results, because ambiguous effects could trigger additional, potentially unnecessary, higher tier testing (i.e., animal-intensive life-cycle studies) to confirm or refute the VTG result. A literature review of VTG data from standard fish species exposed to 106 substances showed high intra- and interlaboratory variability in VTG concentrations, as well as discrepancies in the interpretation of results based on large differences between fish held in dilution water versus solvent controls, or due to the presence of outlier measurements (Brown et al., 2023). For instance, the coefficient of variation of VTG concentrations in control adult fathead minnows (Pimephales promelas) was 543.2% for males and 206.1% for females. The same review also found evidence for false-positive/negative responses and situations in which the VTG results were difficult to interpret.

Findings from a laboratory survey to help understand the sources of variability in VTG protein and mRNA measurements identified three areas for improvement and harmonization: (1) sampling and storage, (2) quantification, and (3) data handling and statistical analysis (Burden et al., 2023).

The survey also highlighted a need for the development of a decision logic to assist in the acceptability, determination, and interpretation of VTG measurements. This would support the development of new OECD guidance detailing best practice for VTG methodology, applicable across relevant test guideline studies but also applicable to studies published in the open literature. In the proposed decision logic (Figure 1), the reliability of the VTG results is assessed separately from the overall study reliability, because a reliable study (as evaluated against, for example, the Klimisch criteria [Klimisch et al., 1997] or the Criteria for Reporting and Evaluating Ecotoxicity Data [Moermond et al., 2016]), may still include unreliable VTG analyses. The reliability of VTG results should therefore be based on specific quality criteria (e.g., as outlined in OECD [2012] test guideline 229: Fish short term reproduction assay, and summarized in Burden et al., 2023). A lack of reporting key information such as the calibration curve statistics, or limits of detection (LoD) or limits of quantification (LoQ) can make it difficult to assess the reliability of the VTG result.

Data handling and statistical analysis require special consideration as part of OECD guidance on VTG methodology and interpretation (Burden et al., 2023)—from the survey it was evident that laboratories have different approaches. This includes the value used to represent male VTG concentrations when they are below the LoQ (e.g., the LoQ value itself, half the LoQ value, the LoD value or zero), or the identification and handling of outlier VTG values. Outliers can be identified as values that exceed the median plus three times the interquartile range (i.e., the difference between the 75th and 25th percentiles) or via formal outlier tests (e.g., Grubb's test). Conducting statistical analysis with and without outliers is the most transparent way of presenting the data. In addition to general recommendations for the statistical analysis of VTG data based on OECD test guideline 54 (OECD, 2006), methods addressing specific aspects of VTG data handling should be more explicit in the proposed new guidance (i.e., related to hypothesis testing when the sex-related response is one sided or the value to use in case male VTG is below the LoQ).

According to the European Chemicals Agency (ECHA) and European Food Safety Authority (EFSA) guidance for the identification of endocrine disruptors (ECHA/EFSA, 2018), a decrease in VTG may also be caused by overt or systemic toxicity, nonendocrine mechanisms (e.g., hepatotoxicity), or confounding factors such as diet or infection (Dang, 2016), and should not necessarily be interpreted as being due to an endocrine mechanism. When one is considering whether inhibition of VTG is endocrine mediated, the effect should be interpreted in combination with other observations including systemic toxicity (typically based on an arbitrary threshold of 10% mortality related to the definition of the Maximum Tolerated Concentration) and other overt signs of nonlethal toxicity, such as behavioral changes (e.g., loss of equilibrium, lethargy), gill respiration rate, reduced growth, discoloration, and reduced feeding (Hutchinson et al., 2009; Wheeler et al., 2013).

The existence and shape of the concentration–response and the magnitude of VTG effects in exposed organisms compared with controls are also important considerations (e.g., responses should be considered in the context of concentration–response relationships to ensure biological relevance). This is especially the case for screening studies, which often use a wide spacing (up to 10-fold) between test concentrations. Based on the literature review, small changes in VTG (e.g., <100% induction in males and <30% induction or inhibition in females for P. promelas) are unlikely to be biologically relevant as indicators for endocrine activity, considering experimental and species-specific variability (Brown et al., 2023). The consistency of VTG changes should therefore be considered alongside other endocrine-mediated endpoints such as specific gonadal histopathology findings (Ankley & Jensen, 2014).

Finally, VTG changes should be checked for consistency with other available data (i.e., in silico, in vitro, in vivo studies in mammals and knowledge of robust adverse outcome pathways). In utilizing all available data in a Weight of Evidence (WoE) analysis, it may be possible to confirm or refute whether there is a need for additional higher tier fish testing. The WoE can also be applied where there are conflicting VTG results, for example, by giving more weight to studies that have well-reported VTG analysis and meet the data quality criteria. A WoE could also be used in determining the next steps following an equivocal VTG result. For example, an equivocal effect on male VTG, in the absence of other evidence for endocrine activity, could potentially be resolved by conducting relevant in vitro assays covering the EAS pathways (e.g., with the estrogen receptor transactivation assay [OECD, 2021a], the aromatase inhibition assay [US Environmental Protection Agency, 2009], or eleutheroembryonic (nonprotected life stage) assays, such as the EASZY assay [OECD, 2021b] or the RADAR assay [OECD, 2022]). Such a mechanistic approach may avoid conducting animal studies or justify a move directly to higher tier animal testing.

In conclusion, the proposed decision logic provides a basis for more consistent and transparent VTG measurement and reporting and will assist end users, including regulatory scientists, in interpreting the data. The overall aim is to support the development of OECD guidance on best practice for VTG assessment in aquatic vertebrates.

Rebecca Jayne Brown: Conceptualization; Writing—original draft; Writing—review & editing. Grace Panter: Conceptualization; Writing—original draft; Writing—review & editing. Natalie Burden: Conceptualization; Writing—original draft; Writing—review & editing. Lennart Weltje: Conceptualization; Writing—original draft; Writing—review & editing. James Wheeler: Conceptualization; Writing—original draft; Writing—review & editing. Edward Salinas: Conceptualization; Writing—original draft; Writing—review & editing. Yvonne Wolf: Conceptualization; Writing—original draft; Writing—review & editing. Laurent Lagadic: Conceptualization; Writing—original draft; Writing—review & editing.