The research, led by Chunhong Li of The Second Affiliated Hospital of Guilin Medical University, together with Xin Zeng and Yuhua Mao, addresses a stubborn gap in environmental health science. While epidemiological studies have suggested links between endocrine-disrupting chemicals and breast cancer, the precise molecular mechanisms through which MEP exposure might drive oncogenesis and tumor progression have remained poorly understood. Laboratory work in model systems has added to the concern: recent studies have shown that methylparaben can induce metabolic disorder and liver damage at human-relevant exposure levels, disrupt endocrine function and impair reproduction in adult zebrafish, and correlate with altered semen quality in reproductive-aged men. Meanwhile, Mendelian randomization analyses combined with network toxicology have begun to suggest causal relationships between the compound and cancers including glioblastoma and breast cancer. What has been missing is a systematic, integrated picture of how MEP’s molecular targets intersect with the genomic and immunological landscape of actual breast tumors.

To build that picture, the team assembled breast cancer-related targets from three authoritative disease databases: the Comparative Toxicogenomics Database, GeneCards, and OMIM. On the chemical side, they interrogated MEP-related targets from ChEMBL, PharmMapper, and the Similarity Ensemble Approach, applying stringent filters to ensure that only high-confidence chemical-protein associations survived. The intersection of these two target sets formed the seed of the analysis, informing the construction of protein-protein interaction networks designed to reveal which human proteins sit at the crossroads of MEP’s toxicological activity and breast cancer biology. Molecular docking studies then tested, at the structural level, whether MEP could plausibly bind the key proteins identified by the network analysis, providing a physical rationale for the computational associations.

The results converged on five core putative toxicological targets that appear to play critical regulatory roles in MEP-associated molecular alterations: HSP90AA1, CTNNB1, TP53, MYC, and EGFR. For anyone familiar with cancer biology, this list reads like a hall of fame of oncogenic machinery. HSP90AA1 encodes the molecular chaperone HSP90, which stabilizes hundreds of client proteins and has been independently implicated in breast cancer progression and doxorubicin resistance through PI3K/AKT signaling; elevated plasma HSP90AA1 has even been proposed as a predictor of breast cancer onset and distant metastasis. CTNNB1 encodes beta-catenin, the transcriptional co-activator at the heart of the Wnt signaling pathway, which has recently become the target of novel covalent degrader drugs. TP53, the guardian of the genome, shapes long-term responses to CDK4/6 inhibitors in breast cancer through its role in cellular senescence. MYC, the archetypal oncogenic transcription factor, has been shown to suppress STING-dependent innate immunity in triple-negative breast cancer. And EGFR acts as what researchers have called a master switch between immunosuppressive and immunoactive tumor microenvironments in inflammatory breast cancer.

That a single cosmetic preservative would dock onto and potentially modulate this particular quintet of proteins is the study’s most striking finding, because it suggests a plausible mechanistic route by which chronic, low-level MEP exposure could touch some of the most fundamental circuits of tumor biology: chaperone stabilization of oncogenic clients, Wnt-driven proliferation, loss of genomic surveillance, immune evasion, and growth factor signaling. The authors are careful to frame these as putative targets identified through computational inference rather than proven causal agents, but the convergence of network toxicology, docking evidence, and prior experimental literature on each of these proteins gives the hypothesis a weight that few single-method studies could muster.

But the team did not stop at target identification. In a second major analytical phase, they derived consensus molecular subtypes of breast cancer by applying ten different clustering algorithms to multi-omics data from patient cohorts, drawing on datasets from The Cancer Genome Atlas, UCSC XENA, and the GEO repositories. This consensus clustering approach, implemented through purpose-built R packages for multi-omics integration, guards against the well-known problem that any single clustering method can impose artificial structure on high-dimensional data. By requiring agreement across ten algorithms, the researchers aimed to identify molecular subgroups of breast cancer that reflect genuine biological differences rather than statistical artifacts. These consensus subtypes then served as the foundation for the study’s most clinically oriented deliverable.

Using three machine learning algorithms applied to the subtype-classified multi-omics data, the researchers developed what they call a consensus MEP-related signature, abbreviated CMEPRS, a prognostic model for breast cancer patients. The signature functions as a molecular classifier that stratifies patients according to the activity of MEP-toxicity-related genes in their tumors. In essence, the model asks not whether a patient was exposed to methylparaben, but whether the molecular programs that MEP is predicted to perturb are active in that patient’s tumor, and whether that activity pattern carries prognostic information. The resulting classifiers and the CMEPRS prognostic model, the authors report, may facilitate patient stratification and support personalized clinical management, offering oncologists a new computational lens through which to view tumor biology that is rooted in environmental toxicology rather than conventional pathological categories.

The immunological findings that emerged from applying the signature are particularly provocative. Patients with high CMEPRS scores displayed prominent infiltration of macrophages, myeloid-derived suppressor cells, and cancer-associated fibroblasts, three cell types that collectively form the cellular architecture of an immunosuppressive tumor microenvironment. This triad is familiar to immunotherapy researchers for an unwelcome reason: tumors dominated by these populations tend to exclude cytotoxic T cells and resist checkpoint blockade. The finding resonates with established literature showing that EGFR can act as a switch governing whether the inflammatory breast cancer microenvironment is immunosuppressive or immunoactive, and that MYC can suppress innate immune sensing in triple-negative disease. In other words, the MEP-associated molecular signature appears to mark tumors whose microenvironmental composition would be expected to blunt immunotherapy response, a hypothesis that could be tested directly in future clinical studies.

Drug sensitivity predictions added a further layer of translational interest. Using computational tools that predict in vivo drug response from cell line screening data, the team found that, apart from the HER2-targeted drug lapatinib, high-CMEPRS patients showed higher predicted sensitivity to most conventional chemotherapeutic drugs. If validated, this pattern would carry practical implications: the MEP-associated molecular state might not only flag a more immunosuppressive and prognostically distinct tumor, but also one that could be managed effectively with existing cytotoxic regimens, while raising questions about the relative benefit of specific targeted agents. The study’s authors emphasize that their work is computational and therefore preliminary, providing insights into molecular alterations linked to MEP exposure rather than definitive proof of harm or clinical utility.

Nevertheless, the broader significance of the study lies as much in its methodology as in its specific findings. By fusing network toxicology with multi-omics consensus clustering and machine learning, the researchers have demonstrated a feasible analytical framework for connecting environmental chemical exposure to cancer patient stratification and therapeutic-target exploration, a template that could be applied to the dozens of other endocrine-disrupting chemicals that populate modern life. As global breast cancer incidence continues to climb across 185 countries, and as recent commentary in leading cancer journals urges a rethinking of the origins of early-onset estrogen receptor-positive disease in light of environmental endocrine disruptors, studies of this kind sharpen the questions that laboratory and epidemiological work must now answer. Whether methylparaben truly helps set the stage for breast cancer will require experimental validation, but this research has mapped, with unusual precision, exactly where to look.

Subject of Research: Computational analysis of molecular mechanisms linking the endocrine-disrupting preservative methyl 4-hydroxybenzoate to breast cancer

Article Title: Integrated multi-omics, machine learning, network toxicology, and molecular docking reveal potential mechanisms underlying methyl 4-hydroxybenzoate-associated breast cancer

Article References: Integrated multi-omics, machine learning, network toxicology, and molecular docking reveal potential mechanisms underlying methyl 4-hydroxybenzoate-associated breast cancer. (n.d.). https://doi.org/10.1007/s11030-026-11712-1

Image Credits: AI Generated

DOI: 10.1007/s11030-026-11712-1

Keywords: breast cancer, methylparaben, endocrine-disrupting chemicals, network toxicology, molecular docking, multi-omics, machine learning, prognostic signature, tumor microenvironment, HSP90AA1, TP53, EGFR