A new study published in Ecotoxicology and Environmental Safety sheds light on the molecular mechanisms by which per- and polyfluoroalkyl substances — commonly known as PFAS or “forever chemicals” — may drive the development and progression of colorectal cancer, offering potential targets for future therapies.
Background: A Growing Cancer Burden and an Invisible Threat
Colorectal cancer (CRC) is one of the most common and deadly malignancies worldwide. According to current projections, the global burden of CRC is expected to surpass 3 million new cases and approximately 1.6 million deaths annually by 2040. While improved screening has helped reduce CRC rates among adults over 50, the incidence in younger adults under 50 has risen sharply since the 1990s — a trend that remains poorly understood but is widely suspected to involve dietary, lifestyle, and environmental factors.
Among the environmental factors drawing increasing scientific scrutiny are PFAS — a large family of more than 12,000 synthetic chemicals that have been used for decades in products ranging from non-stick cookware and food packaging to firefighting foam and waterproof clothing. These substances are extraordinarily resistant to environmental and biological degradation, earning them the nickname “forever chemicals.” They accumulate in soil, water, and living organisms, and have been detected in human blood, breast milk, and a wide range of tissues globally.
PFAS exposure has been linked to a range of adverse health outcomes, including hormonal disruption, metabolic disorders, immune dysfunction, and elevated cancer risk. Animal studies have associated PFAS compounds — particularly perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) — with the development of liver, pancreatic, and testicular tumors. More recently, epidemiological and laboratory evidence has begun to implicate PFAS in the disruption of intestinal homeostasis and the promotion of gastrointestinal cancers, including CRC. However, the precise biological mechanisms have remained largely unclear — until now.
The Study: Combining Bioinformatics with Laboratory Experiments
Researchers from China conducted an integrated investigation combining large-scale bioinformatics analysis with cell biology experiments to map the molecular pathways through which PFAS exposure might fuel CRC development. Their work, led by Wende Hao, Xiangyu Chen, Huixian Xin, and colleagues, represents the first attempt to construct a PFAS-based prognostic model for CRC.
The team began by identifying differentially expressed genes (DEGs) — genes whose activity is significantly elevated or suppressed in CRC compared to normal colon tissue — using two publicly available genomic datasets: GSE33113 and GSE32323. They then cross-referenced these cancer-associated gene expression changes with data from the Comparative Toxicogenomics Database (CTD), which catalogs known interactions between chemical exposures and gene expression changes.
This intersection yielded 192 genes that were both altered in CRC and known to be responsive to PFAS exposure, suggesting these genes may be key players in a pathway linking chemical exposure to cancer development. Pathway analysis of these 192 genes pointed toward disruption of the cell cycle and several oncogenic signaling cascades as the likely mechanisms of PFAS-induced cancer promotion.
Building and Validating a Prognostic Risk Model
From the 192 PFAS-related DEGs, the researchers applied univariate Cox regression analysis to the large TCGA-COAD dataset (comprising 370 CRC patients) to identify genes with genuine prognostic significance. This initial screen yielded 20 candidate genes. A further round of statistical refinement using LASSO regression — a method that prevents overfitting by penalizing model complexity — narrowed this down to a final set of 15 genes that together formed a robust prognostic signature.
The 15 genes included in the risk model are: ABCC5, PRAME, AASS, ASNS, FOXM1, PCK1, CCDC34, SLC36A4, ORC1, SCD, CCNB1, HPGD, CDKN2A, ADH1C, and KRT7. Using the expression levels of these genes weighted by their statistical coefficients, a risk score was calculated for each patient. CRC patients were then divided into high-risk and low-risk groups based on the median score.
The results were striking. Patients in the high-risk group had significantly worse overall survival than those in the low-risk group. The model’s ability to predict survival was validated by ROC (Receiver Operating Characteristic) curves, which yielded area-under-the-curve (AUC) values of 0.81, 0.80, and 0.78 at one, two, and three years, respectively — indicating strong predictive performance. The model’s accuracy was further confirmed using an independent external dataset (GSE17536), reinforcing its generalizability.
Multivariate Cox regression analysis confirmed that the risk score functioned as an independent prognostic factor for CRC overall survival, alongside tumor stage. To make the model clinically useful, the team incorporated it into a nomogram — a graphical tool combining multiple variables to predict individual patient outcomes — which demonstrated good calibration and practical clinical benefit.
Genomic Instability and Therapeutic Implications
Patients in the high-risk group displayed several other clinically relevant features. Their tumors carried a greater burden of genetic mutations and showed higher scores on the MATH (Mutant-Allele Tumor Heterogeneity) index — a measure of genetic diversity within a tumor that is associated with worse outcomes and drug resistance. Copy number variation (CNV) analysis revealed widespread chromosomal gains in several model genes, including PRAME, FOXM1, ABCC5, and ASNS.
Gene set enrichment analysis comparing high- and low-risk tumors revealed that high-risk CRC was associated with activation of multiple oncogenic pathways: mitotic spindle assembly, Wnt/β-catenin signaling, Hedgehog signaling, Notch signaling, ERBB signaling, and mTOR signaling. These pathways are well-established drivers of tumor growth and invasiveness, suggesting that PFAS-associated gene expression changes funnel into core cancer-promoting networks.
On the therapeutic side, the risk model was used to predict sensitivity to various drugs. Several compounds — including Lenalidomide, Shikonin, DMOG, and ABT-263, among others — showed greater predicted efficacy in high-risk patients. While most of these agents are not yet standard CRC treatments, the findings may point the way toward future repurposing opportunities. Notably, analysis using the TIDE (Tumor Immune Dysfunction and Exclusion) algorithm suggested that high-risk patients may paradoxically be better candidates for immunotherapy, as they showed lower immune dysfunction scores and a higher predicted likelihood of responding to checkpoint inhibitor treatment.
PRAME: The Key Gene at the Interface of PFAS Exposure and Cancer
Of the 15 prognostic genes, PRAME — Preferentially Expressed Antigen in Melanoma — emerged as the most significant finding. Bubble plot analysis confirmed PRAME as the most significantly upregulated gene in CRC across both gene expression datasets. Its expression rose progressively from normal colon tissue to primary tumors to metastatic lesions. In laboratory experiments, PRAME was found to be strongly overexpressed in multiple CRC cell lines compared to normal colonic epithelial cells. Kaplan-Meier analysis linked high PRAME expression to significantly shorter overall survival and poorer recurrence-free survival in CRC patients.
PRAME is a cancer-testis antigen first identified in melanoma as a target for immune cells. It has since been found to be overexpressed in a broad range of cancers, where it appears to suppress differentiation and promote malignant behavior — but its role in CRC had not previously been characterized. The new study provides the first detailed look at how PRAME operates in colorectal cancer, finding it to be associated with activation of MYC target genes (which drive cell proliferation) and spliceosome pathways (which regulate gene expression processing and have been linked to CRC invasiveness).
Most importantly, the researchers demonstrated experimentally that PFOS exposure — the most widely studied PFAS compound — directly induces PRAME upregulation in CRC cells. When HCT-116 and SW480 CRC cell lines were treated with 5 µmol/L PFOS, proliferation and invasive capacity increased significantly. These effects were accompanied by upregulation of PRAME protein levels.
The team then used RNA interference (siRNA) to silence PRAME in the same cell lines. Knockdown of PRAME substantially reduced both proliferation and invasion, confirming its functional importance. Western blot analysis revealed that PFOS treatment activated key proteins in the PI3K/Akt/mTOR signaling cascade — specifically the phosphorylated (active) forms of PI3K, Akt, and mTOR, as well as downstream effectors cyclin D1 and c-Myc — while PRAME knockdown suppressed this same cascade. Similar effects were observed when normal colonic epithelial cells were exposed to PFOS, suggesting that the pathway activation begins before malignant transformation.
Taken together, these findings outline a coherent mechanistic chain: PFOS exposure upregulates PRAME expression in colon cells; elevated PRAME activates the PI3K/Akt/mTOR pathway; this in turn promotes cell proliferation, cell cycle progression, and invasiveness — the core hallmarks of cancer progression.
Significance and Limitations
This study makes several important contributions. It is the first to build a validated multi-gene prognostic model for CRC grounded in PFAS biology, offering a new tool for risk stratification that could guide clinical decision-making. It also provides the first mechanistic account of PRAME‘s role in colorectal cancer and establishes PFOS-driven PI3K/Akt/mTOR activation via PRAME as a plausible oncogenic mechanism — one that could be targeted therapeutically.
The authors acknowledge several limitations. The study is retrospective and relies heavily on publicly available datasets, meaning an independent prospective cohort will be needed to validate the risk model’s real-world clinical utility. The experimental work focused exclusively on PFOS, one member of a structurally diverse PFAS family; whether other PFAS compounds drive the same pathway remains to be determined. And while PRAME has been identified as a key mediator, the mechanisms connecting PFOS exposure to PRAME transcriptional upregulation at the molecular level are not yet fully understood.
Conclusion
The research adds important new evidence to the growing body of science linking environmental pollutant exposure to cancer at the molecular level. As PFAS contamination of drinking water and food supplies continues to be documented worldwide, understanding the biological pathways through which these chemicals may promote cancer is both a scientific imperative and a public health priority. The identification of PFOS-PRAME-PI3K/Akt/mTOR as a potentially actionable oncogenic axis offers a new direction for CRC research — and highlights PRAME as a candidate biomarker and therapeutic target worthy of further investigation.
Source:https://www.sciencedirect.com/science/article/pii/S0147651326001995