A risk assessment of bisphenol A in luminal A breast cancer via fault tree analysis

Yang, Jaemin; Blake, Catherine; Flaws, Jodi

Permalink

https://hdl.handle.net/2142/133279 Copy

Description

Title

A risk assessment of bisphenol A in luminal A breast cancer via fault tree analysis

Author(s)

• Yang, Jaemin
• Blake, Catherine
• Flaws, Jodi

Issue Date

2025-03-16

Keyword(s)

• Bisphenol A
• Breast Cancer
• Fault Tree Analysis
• Risk Assessment
• Luminal A
• Endocrine Disruptors

Date of Ingest

2026-05-15T08:59:32-05:00

Abstract

Background and Purpose: Bisphenol A (BPA), an industrial chemical used in a variety of settings including the manufacture plastics. There is little debate that BPA disrupts the endocrine system, but rather on establishing the maximum tolerable daily intake (MTDI). Experts from 41 institutions wrote in support of the European Food Safety Authority’s (EFSA) proposal to lower the tolerable daily intake of BPA by 20,000-fold. At the heart of the argument was that risk assessment conclusions are highly influenced by the authors choice of which data reported in experimental studies should be included, and which to data to discard. In this paper we introduce fault tree analysis (FTA) to the toxicology community. FTA enables decision makers to systematically structure to capture and quantify risk and has conventionally been deployed in high-risk engineering fields such as nuclear power plants and aerospace. In this paper we focus on the diverse biological pathways that are shared between BPA and breast cancer as a first step, but FTAs enable decision makers to integrate evidence at different scales, from molecular mechanisms to epidemiological data and thus are aligned with the transparency needed by modeling methods used for public health policy. In contrast to the macro-BPA discussion around the MTDI, we take a micro view that considers only the intersection between BPA and breast cancer. Breast cancer remains one of the most common malignancies on a global basis, and of the breast cancers. Luminal A is the most prevalent subtype of breast cancer, which is characterized with a positive estrogen receptor and progesterone receptor and low expression of HER2. Our goal is to demonstrate that fault tree analysis (FTA) can accurately capture the key biological mechanisms that underpin luminal A breast cancer and quantify the impact of BPA with respect to those biological mechanisms. Method: The biological pathways implicated in breast cancer were identified using the Kyoto Encyclopedia of Genes and Genomes database (KEGG, https://www.genome.jp/kegg/) and Rat Genome Database (RGD, https://rgd.mcw.edu/). The KEGG pathway includes estrogen receptor activation, MAPK/ERK signaling, modulation of the Wnt pathway, and changes in gene expression that have been previously associated with the development of breast cancer. We then manually developed a fault tree where the development of breast cancer is the root node, and specific events that lead to this outcome are captured by sub-trees and branches that capture logical relationships between gene activation, and breast cancer risk. The GSE65194 dataset from the GEO database was used to obtain gene expression profiles for breast cancer patients and normal controls, providing baseline intervals for comparison in the model. We then identified intersections between the mechanisms known to be involved with breast cancer and mechanisms reported with BPA exposure by searching PubMed to identify systematic reviews, and relevant databases. We compiled a dataset of concentration-dependent gene expression changes by extracting from each paper for each reported level of BPA. Where papers presented data as a figure only, we converted the figures to quantities. This data was converted so that each event in the fault tree had a probability drawn from approximately 20 peer-reviewed studies that are BPA-associated in the RGD database. Result: The quantitative FTA focused on the key events of ER activation, MAPK/ERK signaling, and Wnt pathway modulation. The FTA output illustrates how these pathways interact to drive cancer development. For example, the development of breast cancer requires (1) Activation of ER AND (2) downstream signaling pathway that (2a) mimics the activity of estrogen OR (2b) suppresses ERα co-repressors AND (2c) involves the MAPK/ERK Pathway OR Wnt Pathway. This text output highlights one of the key pathways, but the full model integrates additional biological interactions and gene expression changes relevant to BPA exposure. Using the probabilities from the literature the FTA showed a non-monotonic, inverted U-shaped dose-response curve typical of endocrine disruptors, with following the equation: 𝑦 = −0.0027𝑥^2 + 1.3829𝑥 +38.2199. The curve shows a change around 256 nM, suggesting that BPA may influence biological responses at this concentration. When evaluating gene expression data from the GSE65194 dataset, our model predicted values for the luminal A subtype of breast cancer that ranged between 49.45 and 151.75, with an average of 80.72, while for normal controls, the predicted values ranged between 19.88 and 58.82, with an average of 36.00. Since the GSE dataset does not provide information on the body weights of patients or controls, we explored the relationship between body weight and BPA concentration by varying the assumed body weight from 50 to 100 kg. At 0.04 ng/kg/day, BPA concentrations ranged from 8.76 nM (50 kg) to 17.52 nM (100 kg), with predicted gene expression values from 50.13 to 61.62. At 0.2 ng/kg/day, BPA concentrations ranged from 43.80 nM (50 kg) to 87.60 nM (100 kg), with predicted values ranging from 93.61 to 138.65. Applying EFSA’s recommended tolerable daily intake of BPA (0.04 ng/kg/day), a 75 kg adult corresponds to a BPA concentration of 13.14 nM, yielding a predicted value of 55.92. This value aligns closely with the lower bound of the luminal A interval (49.45), further supporting the relevance of our model in capturing key biological responses to BPA exposure. The final values generated by our FTA model represent normalized relative gene expression levels rather than raw expression values. These unitless values ensure consistency, though they reflect comparative changes rather than absolute quantities. This normalization helps the model robustly integrate multiple biological pathways and accurately assign probabilities to their interactions. Conclusion: This study introduces FTA to decision makers in toxicology as a new way to model mechanistic pathways and explore potential links between BPA exposure and breast cancer, serving as a proof of concept for its feasibility in environmental health risk assessment. Using data from empirical studies reported in the literature, the FTA captures the known non-linear inverted U-shaped dose-response relationship between BPA and cancer and makes visible the risk derived from different biological pathways. This study specifically focuses on modeling the metabolism pathway as a critical part of BPA’s endocrine-disrupting effects, with future work aimed at integrating FTA with event tree analysis to cover additional absorption, distribution, and excretion processes. This framework helps establish a better understanding of the various levels of BPA exposure and their contribution to the total risk of developing breast cancer, especially in the luminal A subtype. It is important that these risks be quantified as regulatory agencies, including the EPA and FDA, continue to reassess safety thresholds for BPA exposure. Since the current estimation assumes that all BPA doses are fully absorbed, future research will need to consider absorption-related fault trees as part of the event tree framework to accurately reflect real-world pharmacokinetic variability. Integrating ADME, is essential for a complete and robust risk assessment. One of the strengths of FTA is its ability to integrate probabilities at each node, enabling a cumulative and quantitative assessment of risk. Furthermore, FTA captures interdependencies between multiple biological pathways, which is essential for substances like BPA that can activate several pathways simultaneously—something traditional models often treat as independent. This scalability and flexibility allow FTA to begin with a simple structure and expand into a more comprehensive framework by incorporating event tree analysis to model additional ADME processes in future research. Importantly, multi-scale evidence, from molecular mechanisms to pathway-level interactions, was integrated into this study to build a systematic environmental risk assessment. Our findings align with the exposome framework, demonstrating how both external exposures and biological responses interact over time to influence health outcomes. Coupling systems biology with computational risk modeling is necessary to develop better public health strategies and regulatory policies, bridging the gaps between experimental research and risk modeling to reduce the impact of endocrine disruptors like BPA on human health.

Publisher

Society of Toxicology

Series/Report Name or Number

The Toxicologist, a Supplement to Toxicological Sciences, Abstract #3493

Type of Resource

text

Genre of Resource

conference poster

Language

eng

DOI

https://doi.org/10.5281/zenodo.16370014

Copyright and License Information

© 2025 Society of Toxicology. Deposited under Creative Commons Attribution 4.0 International (CC BY 4.0).

Owning Collections