Determining hormone metabolite concentrations when enzyme immunoassay accuracy varies over time. Issue 5 (12th February 2015)

• Record Type: Journal Article
• Authors: Davidian, Eve; Benhaiem, Sarah; Courtiol, Alexandre; Hofer, Heribert; Höner, Oliver P.; Dehnhard, Martin; Fisher, Diana
• Is Part Of: Methods in ecology and evolution. Volume 6:Issue 5(2015:May)
• Page Start: 576
• Page End: 583
• Publication Date: 2015-02-12
• Subjects: Ecology -- Periodicals; Evolution -- Periodicals; 577
• DOI: 10.1111/2041-210X.12338
• Languages: English
• ISSNs: 2041-210X

Abstract

<abstract abstract-type="main" id="mee312338-abs-0001"> <title>Summary</title> <p> <list id="mee312338-list-0001" list-type="order"> <list-item> <p>Enzyme immunoassays (EIAs) are widely used to quantify concentrations of hormone metabolites. Modifications in laboratory conditions may affect the accuracy of metabolite concentration measurements and lead to misinterpretations when results of different accuracy are combined for a statistical analysis. This issue is of great relevance to studies in behavioural and evolutionary ecology because these usually aim at understanding how hormone concentrations vary between individuals, environments or experimental conditions.</p> </list-item> <list-item> <p>We present a method based on re‐assaying a subset of samples to standardize hormone metabolite concentrations when changes in EIA accuracy occur. We used glucocorticoid metabolite concentrations (fGMCs) measured in faeces of spotted hyaenas (<italic>Crocuta crocuta</italic>) between 2011 and 2013 with a previously validated EIA. Changes in accuracy were assessed by monitoring the metabolite concentration of faecal control 'pools' that were systematically assayed with faecal samples. A cluster analysis on these pools identified two distinct sample sets with different EIA accuracy; 'Cluster 1' and 'Cluster 2'. We then re‐assayed all samples of Cluster 1 (<italic>n</italic> = 138) with an EIA accuracy similar to that of Cluster 2 and fitted a linear regression to the remeasured fGMCs against the initial fGMCs to predict fGMCs in Cluster 2. To determine the minimum number of samples to re‐assay that allows reliable predictions, we assessed the variation in the quality of model predictions by fitting linear regressions on decreasing numbers of re‐assayed samples. This revealed that re‐assaying 27 samples would be sufficient to generate reliable predictions considering our data set.</p> </list-item> <list-item> <p>To test the robustness of our method, we fitted a new linear regression to 27 randomly chosen samples and used its equation to standardize all fGMCs of Cluster 1. The standardized fGMCs were similar to the remeasured fGMCs, and the regression on 27 samples was as effective at standardizing fGMCs as the regression fitted on the complete data set.</p> </list-item> <list-item> <p>Our standardization method permits the combination of results of different accuracy. It is a simple and reliable alternative to the costly, time‐consuming and often impractical re‐assaying of complete sample sets that can be applied to a wide variety of species and sample types.</p> </list-item> </list> </p> </abstract>