iVEMPS: QSAR Predicting Thyroid Receptor Disruption

iVEMPS: A High-Accuracy QSAR Predicting Thyroid Receptor Disruption as a First Step Toward Teratogenic Risk Assessment

Emel Ay-Albrechta, Marie Darracq-Ghitalla-Ciocka1, Franklin J. Bauera, Zlatomir Todorova, Carole Charmeau-Genevoisa, Paul C. Thomasa

1 KREATiS SAS, 23 rue du Creuzat, ZAC de St-Hubert, 38080 L’Isle d’Abeau, France

In 2018, EFSA & ECHA regulatory agencies developed a framework involving testing levels for identification of Estrogenic, Androgenic, Thyroid and Steroidogenesis (EATS) modality perturbations using in silico, in vitro and in vivo methods1. While several validated E, A and S receptor assays are already available, in vitro thyroid receptor-specific assays are still needed. The thyroid system plays a critical role in growth and neurodevelopment including teratogenic effects via thyroid-receptor disruption2. Current regulatory methods for thyroid system investigation rely on thyroid hormone concentration measurements from in vivo reproductive and developmental toxicity studies, which fail to reveal specific interactions with thyroid receptors (α and β). To address these limitations, a new high-accuracy quantitative structure-activity relationship model was developed.

A curated dataset, mainly sourced from EPA ToxCast, was refined through a semi-automated curation process guided by expert knowledge, ensuring the selection of high-quality data. The data splitting was performed using the Kennard-Stone algorithm. The machine learning algorithm, namely a support vector machine using circular molecular fingerprints, was trained and validated. A sophisticated applicability domain (AD) methodology was applied using a k-nearest neighbours’ model, based on the identification of structural analogues and the concordance of results with these analogues.

Without including AD limitations, the model already demonstrated robust performance, with high sensitivity, specificity and accuracy, i.e. 92.05, 100 and 96.03% for the training set and 87.65, 97.26, and 92.46% for the external test set. Following application of the AD model, statistics of the external test set increased respectively to 92.30, 100 and 96.15% for sensitivity, specificity and accuracy.

This new OECD QAF-compliant3 methodology enhances prediction reliability, filling critical gaps in thyroid receptor research. By linking thyroid receptor disruption with potential developmental and teratogenic outcomes, the iVEMPS model supports early-stage assessment of endocrine-mediated teratogenicity, enabling prioritisation of low concern chemicals and reducing animal testing.

Références :

1. ECHA and EFSA with support from JRC, Guidance for the identification of endocrine disruptors in the context of Regulations (EU) No 528/2012 and (EC) No 1107/2009, EFSA journal, 16, 6, e05311, 2018.

2. Norman AW and Henry HL, Chapter 5 – Thyroid hormones, Hormones - Third edition, 89-107, Elsevier, 2015.

3. OECD, (Q)SAR Assessment Framework: Guidance for the regulatory assessment of (Quantitative) Structure Activity Relationship models and predictions - Second Edition, OECD Series on Testing and Assessment, OECD Publishing, 2024.

Keywords : thyroid, in silico, endocrine disruption, teratogenicity, NAMs