Cell Painting-Based Tool for the Risk Assessment of Mammary Carcinogens and Endocrine Disruptors

Breast cancer is the most prevalent cancer among women globally, and chemicals that interfere with estrogen or progesterone signaling are considered potential risk factors. However, the effects of chemicals on mammary gland (MG) development and function are still not thoroughly investigated, emphasizing the need for further research in this field.

To fill this knowledge gap, researchers employed high-content imaging assays, specifically Cell Painting technology, to evaluate cell morphology alterations due to chemical exposure and to identify morphological characteristics that might indicate mammary carcinogens (MC) risk factors.

By analyzing a dataset of MC and non-mammary carcinogens (Non-MC) with existing Cell Painting profiles from the JUMP-CP dataset, the researchers identified 51 compounds, consisting of 28 MC and 23 non-genotoxic Non-MC. These compounds were then analyzed through non-linear dimensionality reduction (UMAP) and hierarchical clustering techniques to characterize the morphological data.

The researchers then created a Guilt-By-Association (GBA) framework, comparing different configurations of similarity metrics, risk-score aggregation methods, and feature representations. Interestingly, the morphological profiles of MCs clustered based on their mechanism of action rather than their carcinogenicity label. Genotoxic MCs caused significant disruptions in the endoplasmic reticulum, mitochondria, nucleus, and RNA compartments.

Conversely, hormonally active compounds were virtually indistinguishable from control cells, indicating a lack of functional steroid hormone receptors in the utilized cell line.

The most effective GBA configuration yielded an AUC-ROC of 0.630 and an AUC-PR of 0.696. When tested prospectively on endocrine disruptors, this method highlighted clofentezine, 3-methylpyrazole, resorcinol, 2-tert-butyl-4-methoxyphenol, and thiabendazole as promising candidates for further confirmation testing. This study introduces a transparent and interpretable tool for prioritizing chemicals in mammary carcinogenicity assessments, while also acknowledging the limitations and underscoring the necessity of complementing Cell Painting datasets with hormone-sensitive models for a more comprehensive evaluation.