The patient first noticed abnormal bleeding on March 7, 2026, and hysteroscopy at a local hospital quickly confirmed endometrioid adenocarcinoma. Imaging at Qilu Hospital showed an irregular endometrial mass of roughly 4.6 by 3.7 by 2.5 centimeters with internal branching vessels, and pelvic magnetic resonance imaging revealed a slightly larger lesion with the restricted diffusion typical of malignancy. Her only laboratory abnormality was a mildly elevated CA125 of 49.7 U/mL. She had no family history of gynecologic or hereditary cancers, and after undergoing a total hysterectomy with bilateral salpingo-oophorectomy and sentinel lymph node biopsy, pathology showed a FIGO grade 2 endometrioid carcinoma invading about 20 percent of the myometrium, with focal lymphovascular invasion but no lymph node involvement and a final stage of pT1aN0, corresponding to FIGO 2009 stage IA.

What transformed this from an ordinary early-stage cancer case into a scientific puzzle was next-generation sequencing. Using a 572-gene hybrid-capture panel on tumor tissue containing roughly 45 percent tumor cells, the team identified a rare variant in the exonuclease domain of POLE, the gene encoding DNA polymerase epsilon: c.833C>A, resulting in a threonine-to-lysine substitution at position 278, or p.T278K. This variant sat at a variant allele frequency of 40.42 percent, and manual review of the raw sequencing data confirmed 440 reads at the locus with 187 supporting the alternate T allele. Alongside it, sequencing uncovered two separate frameshift variants in MSH6, a key mismatch repair gene, both hitting the same nucleotide position c.3261 within a low-complexity sequence but predicting differently truncated proteins.

The genomic consequences were dramatic. The tumor’s mutational burden reached 370.0 mutations per megabase, firmly in the ultramutated range reserved for POLE-driven cancers, and microsatellite analysis showed profound instability with an MSI score of 1279.88, far above the positivity threshold of 200. The substitution profile reinforced the POLE story: 26.35 percent of coding variants were C>A transversions and 8.87 percent were T>G substitutions, the classic nucleotide fingerprint left by a failed DNA proofreading enzyme. Additional driver mutations in PTEN, PIK3CA, ARID1A, PPP2R1A, and FGFR2, plus alterations in DNA damage response genes including BRCA2, ATM, ATR, and RAD54L, rounded out a genomic landscape typical of endometrial carcinoma, and a TP53 missense variant was detected by sequencing even though p53 immunostaining looked wild-type.

Immunohistochemistry added a second layer of complexity. The tumor showed subclonal loss of the mismatch repair proteins MLH1, PMS2, and MSH6, with retained MSH2 expression, and methylation-specific PCR detected MLH1 promoter methylation. Taken together, these findings meant the tumor carried features of not one but two of the four molecular classifiers established by The Cancer Genome Atlas: the POLE-ultramutated group, known for excellent prognosis, and the microsatellite instability-hypermutated group driven by mismatch repair deficiency. Such multiple-classifier endometrial carcinomas are uncommon, and interpreting a rare POLE variant within one of them is where the diagnostic challenge truly begins.

Most pathogenic POLE mutations cluster at well-known hotspot residues such as p.P286R, p.V411L, p.S297F, p.A456P, and p.S459F. The p.T278K variant is different. It lies within the Exo I motif of the exonuclease domain, immediately adjacent to the catalytic residue E277 and inside the DNA-binding cleft, a location that raises suspicion but does not by itself prove pathogenicity. To address exactly this problem, León-Castillo and colleagues developed an evidence-based scoring framework that integrates tumor mutational burden, the proportion of C>A and T>G substitutions, insertion-deletion frequency, and recurrence data to separate genuine proofreading-deficient variants from variants of uncertain significance.

Applying that framework to this tumor, the researchers assigned one point each for the ultrahigh tumor mutational burden, the C>A proportion, the T>G proportion, and the C>G proportion, but no points for the indel proportion of 9.17 percent or for recurrence, since p.T278K was absent from the TCGA endometrial carcinoma cohort and represented by only a single record in COSMIC. The resulting exploratory POLE score of 4, combined with the characteristic substitution spectrum, aligns the tumor squarely with the POLE-ultramutated profile, though the authors caution that the framework was built on whole-exome data and its use with targeted panels should be regarded as supportive rather than definitive. Prior evidence strengthens the case: before this report, p.T278K had been documented in only two patients with endometrial carcinoma, one of whom showed the same ultramutated phenotype and a POLE score of 4, and a yeast-based reversion assay has demonstrated a mutator phenotype for this variant, providing functional evidence of impaired proofreading.

The dual MSH6 frameshift variants tell their own fascinating story. Because they arise as reciprocal insertion and deletion events at the same nucleotide within a repetitive, low-complexity sequence, and because their variant allele frequencies differ unequally at 17.64 and 34.94 percent, they are most consistent with secondary somatic events arising through replication slippage in an already hypermutable genome, a phenomenon previously described in tumors that lose both polymerase proofreading and mismatch repair. Yet both variants are classified in ClinVar as pathogenic in the germline context, and because sequencing was performed on tumor tissue without matched normal DNA, a germline origin could not be excluded. The team therefore recommended peripheral-blood germline testing, a step with major implications for the patient’s relatives. Similarly, MLH1 promoter methylation plausibly explains the subclonal MLH1 and PMS2 loss, making this apparently only the third reported endometrial carcinoma with co-occurring POLE alteration and MLH1 promoter methylation, although the bulk-tumor assay used cannot establish spatial correspondence between the methylation and the patchy protein loss.

Why does all this molecular detective work matter clinically? Because molecular classification now directly determines treatment intensity. Under the 2023 FIGO molecular staging system, a pathogenic POLE mutation in a tumor like this one places it in stage IAmPOLEmut, a category with such an excellent prognosis that adjuvant therapy can often be safely omitted. Even setting the POLE question aside, the tumor’s grade 2 endometrioid histology, limited myometrial invasion, focal lymphovascular invasion, negative sentinel nodes, and mismatch repair deficiency would satisfy FIGO 2023 IA2 criteria and place the patient in the low-risk category. Accordingly, the patient received no adjuvant therapy, a decision consistent with the 2025 ESGO-ESTRO-ESP guidelines, and at four months of follow-up her CA125 had normalized, pelvic ultrasound showed no disease, and there was no evidence of recurrence. Her mismatch repair-deficient status nonetheless remains clinically valuable: should the cancer ever return or progress, MSI-H tumors are prime candidates for immune checkpoint inhibitors.

The case does carry limitations. Tumor-only sequencing cannot definitively distinguish somatic from germline variants, targeted-panel analysis rather than whole-exome sequencing constrains mutational signature analysis, the MLH1 methylation test was qualitative on bulk tissue, and follow-up remains short. The discordance between the TP53 sequencing result and the wild-type p53 staining pattern also remains unexplained, possibly reflecting intratumoral heterogeneity or differences in protein regulation, though it did not alter classification or management. Still, by converging from three independent directions, genomic phenotype, prior clinical observations, and functional assay data, the report substantially strengthens the argument that p.T278K is a genuine proofreading-deficient variant rather than a passenger, and it offers a practical template for pathologists confronting rare POLE variants in tumors that carry multiple molecular fingerprints. For the growing number of patients whose cancers are now classified at the DNA level, that template may help ensure the right patients receive less treatment, and the right ones receive more.

Subject of Research: Molecular classification of endometrial carcinoma harboring a rare POLE p.T278K variant and MSH6 frameshift mutations

Article Title: Case Report: Endometrial carcinoma harboring a rare POLE p.T278K variant and dual MSH6 frameshift variants

Article References: Yang, Y., Luo, H., Yan, H., Yu, H., Ma, X., Teng, Q., Zhang, L., Ye, J., Yuan, Z., & Zhang, Q. (2026). Case Report: Endometrial carcinoma harboring a rare POLE p.T278K variant and dual MSH6 frameshift variants. Frontiers in Oncology, 16, Article 1964637. https://doi.org/10.3389/fonc.2026.1964637

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1964637

Keywords: endometrial carcinoma, POLE, p.T278K, MSH6, mismatch repair deficiency, microsatellite instability, tumor mutational burden, MLH1 promoter methylation, molecular classification, León-Castillo scoring, next-generation sequencing, gynecologic oncology

The research team, led by scientists at the Mahavir Cancer Sansthan and Research Centre in Patna along with collaborators across several Indian institutions, enrolled nine women from the arsenic-endemic districts of Saran, Buxar and Bhojpur. These participants were selected on the basis of the highest blood arsenic concentrations recorded in earlier surveys conducted in the region. Each woman provided samples of blood, urine and breast milk, which were acid-digested in the laboratory and analyzed for total arsenic using graphite furnace atomic absorption spectrometry, a technique sensitive enough to detect the element at concentrations below one-tenth of a microgram per liter.

The exposure data were sobering. Every one of the nine women carried blood arsenic levels above the ten micrograms per liter reference threshold, with the highest measured value reaching an extraordinary 491.2 micrograms per liter. Six of the nine exceeded the CDC’s permissible limit of fifty micrograms per liter in urine, peaking at 921.4 micrograms per liter, a pattern consistent with ongoing exposure through drinking water and locally grown food. Most alarming for public health was the breast milk analysis: eight of the nine lactating women showed arsenic concentrations above the one microgram per liter reference level set by a joint WHO/IAEA collaborative study, with a maximum of 438.6 micrograms per liter. Because breast milk represents a direct exposure pathway for nursing infants, the authors highlight this as evidence of potential maternal-to-infant transfer during a uniquely vulnerable window of development.

To probe what such sustained toxic burden might be doing at the genomic level, the team turned to whole exome sequencing, a next-generation sequencing approach that captures and reads all protein-coding regions of the genome. This strategy is well suited to the task because more than eighty-five percent of known disease-causing variants reside in exons. DNA extracted from each participant’s peripheral blood was prepared using targeted capture libraries and sequenced on an Illumina platform with paired-end chemistry, achieving mean coverage depths above eighty-five to one hundred times. Variants were called following the Genome Analysis Toolkit best-practices framework, aligned to the GRCh38 human reference genome, and classified according to the American College of Medical Genetics and Genomics guidelines using population frequency databases such as gnomAD and the 1000 Genomes Project alongside computational prediction tools including SIFT, PolyPhen-2, MutationTaster and Mutation Assessor.

Out of the nine women sequenced, two carried notable variants in genes that sit at the heart of genomic maintenance. Subject 1 harbored a novel heterozygous missense variant in the MSH6 gene, designated c.3716 T>C (p.Ile1239Thr), located in exon 8 on chromosome 2. The substitution replaces isoleucine with threonine at codon 1239 within a protein that, together with MSH2, forms the MutSα complex responsible for recognizing replication errors during DNA synthesis. Computational algorithms predicted the change to be damaging, and the affected nucleotide showed strong evolutionary conservation across one hundred vertebrate species, yet the variant is absent from all individuals in gnomAD and the 1000 Genomes Project. Under ACMG criteria it was classified as a variant of uncertain significance, a label reflecting both its suspicious features and the current limits of scientific evidence.

Subject 2 carried an even more concerning alteration: a heterozygous missense variant in the ATM gene, c.590G>A (p.Gly197Glu), in exon 6 on chromosome 11. This change, swapping glycine for glutamic acid at codon 197, was predicted damaging by both SIFT and PolyPhen-2, with the glycine residue conserved across all mammalian species examined. The variant is extremely rare, observed in only seven of 30,766 South Asian alleles in gnomAD and absent entirely from the 1000 Genomes cohort. Critically, it was classified as likely pathogenic under ACMG guidelines. The ATM gene encodes a master checkpoint kinase of the PI3/PI4-kinase family that orchestrates the cellular response to DNA double-strand breaks, phosphorylating tumor suppressors such as p53 and BRCA1, the checkpoint kinase CHK2, and the DNA repair protein NBS1. Variants in ATM have been associated with hereditary cancer susceptibility, and the remaining seven participants showed no clinically notable exomic alterations.

The biological significance of these two genes makes their disruption in arsenic-exposed individuals particularly noteworthy. MSH6 is a cornerstone of the mismatch repair pathway, working in concert with MSH2, MLH1 and PMS2 to excise mispaired bases and preserve fidelity during cell division. Germline MSH6 mutations are a recognized cause of Lynch syndrome, the most common hereditary colorectal cancer predisposition condition, and a meta-analysis of twenty-seven studies found MSH6 mutations in roughly seven percent of colorectal and nearly ten percent of endometrial cancer cases. Prior experimental work has shown that MSH6 also interacts with Ku70 to regulate non-homologous end joining of double-strand breaks, and that Msh6-deficient mice accumulate mutation frequencies up to five-fold higher than controls, particularly after ionizing radiation. Arsenic’s toxicological mechanism, meanwhile, is known to generate reactive oxygen species that inflict oxidative DNA damage, including 8-oxoG lesions that mismatch repair helps to process in cooperation with the SETD2 methyltransferase and the ATM signaling pathway.

The authors are careful to frame the findings as hypothesis-generating rather than conclusive. With only nine participants, no unexposed control group, and no matched tissue samples to distinguish somatic from germline variants, the study cannot establish a causal or dose-response relationship between arsenic burden and DNA repair gene mutation. Indeed, several women with equally severe arsenic exposure showed no detectable genomic alterations, hinting that individual genetic background, arsenic metabolism profiles and hormonal factors may shape who develops molecular damage. Sex-specific differences in arsenic methylation capacity, reflected in varying ratios of methylated metabolites, could help explain why women represent a potentially under-recognized high-risk group for environmentally driven cancers, including colorectal cancer, which the researchers identify as a particular concern in chronically exposed populations.

Even within those limits, the study’s integrated design, pairing multi-matrix toxicology with exome-wide sequencing, points toward a new model for environmental health surveillance. The detection of potentially pathogenic variants that conventional screening would miss underscores the value of genomic approaches in exposed communities, and the breast milk findings add an urgent transgenerational dimension, since arsenic has been shown in animal models to induce epigenetic alterations that can be inherited across generations. The research team calls for larger cohort studies incorporating matched controls, functional validation of the identified variants through mismatch repair assays, ATM kinase signaling measurements and CRISPR-based modeling, and exploration of epigenetic silencing in ATM and MSH6. If validated, these molecular markers could serve as early-warning biomarkers of arsenic-induced carcinogenesis, enabling targeted screening and precision public health interventions for the estimated ten million people in Bihar alone who live with poisoned groundwater, and millions more worldwide who share the same invisible exposure.

Beyond the immediate findings, the study sits within a broader epidemiological landscape. Chronic arsenic exposure through groundwater is estimated to affect well over one hundred million people worldwide, with South Asia bearing a disproportionate share of the burden. The metalloid is classified as a Group 1 human carcinogen, and epidemiological work across Bangladesh, India and parts of South America has linked prolonged exposure not only to cancers of the skin, bladder, lung and colorectum, but also to cardiovascular disease and reproductive toxicity. What distinguishes the present research is its attempt to move from population-level risk statistics toward the individual molecular events that may underlie them.

The choice of sequencing platform also merits attention. Unlike targeted gene panels, which interrogate a predefined set of loci, whole exome sequencing permits the discovery of rare, novel or compound variants that no a priori hypothesis would have flagged. It additionally opens the door to mutational signature analysis, in which distinctive patterns of base substitution can be matched to specific mutagenic processes, including the oxidative stress that arsenic is known to induce through reactive oxygen species generation. Such signatures, once validated, could serve as fingerprints of environmental carcinogenesis within a tumor or normal tissue genome.

The breast milk results carry particular weight for regional health policy. Arsenic is efficiently methylated in the body to mono- and dimethylated species, and interindividual variation in this metabolism is thought to influence both retention in tissues and excretion into milk. If lactational transfer proves to be a consistent pathway, interventions such as provision of arsenic-safe drinking water to nursing mothers, nutritional supplementation, and routine screening of breast milk in endemic districts could become practical priorities alongside existing water decontamination programs.

Ultimately, the work illustrates how environmental toxicology and clinical genomics can be woven together at small scale to generate testable hypotheses. Larger cohorts with matched unexposed controls, longitudinal sampling and functional assays will be needed to determine whether the DNA repair variants observed here are truly arsenic-driven, incidental, or markers of broader genomic instability in exposed communities.

Subject of Research: Whole exome sequencing of women chronically exposed to arsenic through contaminated groundwater to identify DNA repair gene variants

Article Title: Whole exome sequencing reveals DNA repair gene variants in women exposed to chronic arsenic poisoning

Article References: Agarwal, R., Kumar, A., Pandey, T., Verma, D., Kumar, K., Chayal, N. K., Ali, M., Srivastava, A., Verma, K. S., Kumar, D., Sharma, A., Singh, M., & Ghosh, A. K. (2026). Whole exome sequencing reveals DNA repair gene variants in women exposed to chronic arsenic poisoning. Discover Toxicology, 3(1), Article 15. https://doi.org/10.1007/s44339-026-00059-9

Image Credits: AI Generated

DOI: 10.1007/s44339-026-00059-9

Keywords: arsenic poisoning, whole exome sequencing, DNA repair genes, MSH6, ATM, groundwater contamination, breast milk arsenic, Bihar India, colorectal cancer risk, genotoxicity, mismatch repair, environmental health