Highlights
Pathogenic or likely pathogenic alterations in homologous recombination repair genes were identified in only 6.0% of endometrial cancers in this single-centre retrospective cohort.
HRD-positive tumours clustered with adverse clinicopathological features, including non-endometrioid histology, high-grade disease, and poor-prognosis molecular subgroups.
Despite this association with aggressive biology, HRD status was not associated with overall survival, disease-free survival, progression-free survival, or platinum sensitivity.
The study supports the primacy of integrated molecular classification over isolated HRR-gene status for routine risk stratification in endometrial cancer.
Background
Endometrial carcinoma is the most common gynecologic malignancy in many high-income settings, but it is not a single disease. Its clinical behavior varies widely according to histotype, grade, stage, and, increasingly, molecular subtype. Over the last decade, the field has moved from morphology-based risk assessment alone toward a biologically grounded framework informed by The Cancer Genome Atlas and adopted into clinical practice by ESGO-ESTRO-ESP guidelines. This molecular taxonomy, centered on POLE-ultramutated, mismatch repair-deficient, p53-abnormal, and no specific molecular profile groups, has improved prognostic discrimination and is shaping adjuvant treatment recommendations.
At the same time, interest in homologous recombination deficiency has expanded across gynecologic oncology because of the success of PARP inhibition in ovarian cancer and its emerging role in other solid tumors. Homologous recombination is a high-fidelity DNA repair pathway that resolves double-strand DNA breaks through proteins including BRCA1, BRCA2, ATM, PALB2, RAD51 paralogs, CHEK2, and others. Defects in this pathway can create genomic instability and may confer synthetic lethality with PARP inhibition. However, the frequency, biological distribution, and clinical significance of HRR alterations in endometrial cancer remain incompletely defined.
That uncertainty is clinically relevant. If HRD is rare, biologically heterogeneous, or not independently associated with outcomes, then broad application of HRD testing in endometrial cancer may have limited value outside research or biomarker-enriched therapeutic trials. Carvalho and colleagues address this question in a contemporary cohort evaluated in the era of molecular classification.
Study Design and Methods
Design and setting
This was a retrospective observational cohort study including 298 consecutive patients with endometrial carcinoma treated at a tertiary cancer centre. The investigators performed molecular profiling and assigned patients to risk groups according to the 2025 ESGO-ESTRO-ESP framework.
Definition of HRD
For this analysis, homologous recombination deficiency was defined pragmatically by the presence of pathogenic or likely pathogenic variants in homologous recombination repair genes. This is an important methodological point. The study did not define HRD by genomic scar assays, large-scale state transitions, loss of heterozygosity scores, or mutational signatures; rather, it used gene-level pathogenic alterations as the biomarker of interest. That choice improves specificity for clearly deleterious alterations but may miss functionally HR-deficient tumors without a detectable pathogenic variant in the selected genes.
Outcomes
The investigators assessed associations between HRD status and clinicopathological characteristics using exact tests. Survival analyses included overall survival, disease-free survival, and progression-free survival using Kaplan-Meier methods. Platinum sensitivity was also evaluated, a clinically relevant endpoint given the therapeutic rationale linking HRD to DNA-damaging therapy response.
Key Findings
HRD prevalence was low
Among 298 tumours, 18 were HRD-positive, corresponding to a prevalence of 6.0%. The most frequently altered genes were BRCA1/2 and ATM. This low prevalence is itself one of the paper’s most important findings. It suggests that, at least when defined by pathogenic or likely pathogenic HRR gene variants, HRD is an uncommon event in unselected endometrial cancer populations.
That estimate is clinically meaningful because enthusiasm for PARP-directed strategies can easily outpace biomarker prevalence. A 6% rate does not preclude therapeutic importance, especially in heavily pretreated or biomarker-selected settings, but it does argue against assuming that HRD is a dominant biological driver in endometrial carcinoma as it is in a substantial subset of high-grade serous ovarian cancers.
HRD clustered with aggressive pathological and molecular features
HRD-positive tumours were more common in non-endometrioid histologies, high-grade cancers, and poor-prognosis molecular subgroups. This pattern is biologically plausible. Non-endometrioid and high-grade endometrial cancers often exhibit greater genomic instability and a more aggressive clinical phenotype, so enrichment of deleterious DNA repair alterations in these subsets is not surprising.
From a translational standpoint, the finding suggests that HRR alterations may track with aggressive tumour biology without necessarily acting as independent determinants of outcome. In other words, HRD may be more a passenger or co-traveler within already high-risk molecular contexts than a stand-alone prognostic classifier.
No significant association with survival outcomes
Despite this enrichment in biologically adverse disease subsets, HRD status was not associated with statistically significant differences in survival.
Median overall survival was 115.9 months in HRD-positive tumours versus not reached in HRD-negative tumours, with p = 0.355.
Median disease-free survival was 41.6 months versus 31.3 months, respectively, with p = 0.689.
Median progression-free survival was 8.0 months versus 13.8 months, respectively, with p = 0.252.
These findings are notable for two reasons. First, they argue against using HRR-gene alteration status alone as a prognostic biomarker in routine practice. Second, they highlight a recurring theme in endometrial cancer biomarker development: markers associated with aggressive pathology are not always independently prognostic once broader molecular architecture is considered.
The apparent lack of outcome separation should be interpreted carefully. Only 18 HRD-positive cases were identified, which substantially limits statistical power and increases the risk of type II error. Nonetheless, in real-world clinical decision-making, a biomarker that is both uncommon and underpowered for clear prognostic discrimination will have limited immediate utility.
No observed association with platinum sensitivity
HRD status was also not associated with platinum sensitivity, with p = 0.234. This result may temper assumptions derived from ovarian cancer, where homologous recombination deficiency often correlates with benefit from platinum-based chemotherapy and PARP inhibition.
Several explanations are possible. Endometrial cancer is biologically distinct from ovarian high-grade serous carcinoma, and the mechanistic consequences of HRR-gene alterations may differ by tissue context. In addition, a gene-variant definition of HRD may not capture the full functional phenotype relevant to platinum responsiveness. Finally, sample size likely constrained the ability to detect moderate associations.
Clinical Interpretation
Why molecular classification remains central
The study’s main clinical message is that integrated molecular classification remains the most informative framework for risk stratification in endometrial cancer. Current practice already recognizes that POLE-mutated tumours generally have an excellent prognosis, p53-abnormal tumours carry higher risk, mismatch repair-deficient cancers have both prognostic and therapeutic implications, and NSMP cancers require further refinement using additional biomarkers and clinicopathological factors.
Within that landscape, HRR-gene alterations do not appear ready to function as an isolated decision-making tool. Their presence may enrich for biologically aggressive disease, but this signal does not translate into independent prognostic value in the present dataset.
Implications for PARP inhibitor development
The negative prognostic result should not be misread as evidence against PARP inhibitor investigation in endometrial cancer. Prognostic biomarkers and predictive biomarkers are not interchangeable. A marker can fail to predict natural history yet still identify vulnerability to a targeted therapy. The current study did not test PARP inhibitor efficacy, so it cannot answer the key predictive question.
However, it does indicate that future PARP studies in endometrial cancer should avoid simplistic biomarker assumptions. Trials may need more refined HRD phenotyping, incorporating genomic scars, biallelic inactivation, functional assays, or composite biomarkers integrated with molecular subtype. It is also plausible that PARP inhibition will prove most useful in rational combinations, such as with immune checkpoint blockade, antiangiogenic therapy, ATR inhibition, or DNA-damaging agents, rather than as a biomarker-monotherapy paradigm borrowed directly from ovarian cancer.
Strengths and Limitations
Strengths
This study has several strengths. It evaluates a consecutive real-world cohort from a tertiary cancer centre, reducing some selection bias associated with highly filtered datasets. It also situates HRR-gene alterations within the modern molecular classification framework, which is essential for clinically relevant interpretation. Finally, it examines both pathological associations and outcome endpoints, providing a more complete view of clinical relevance than prevalence-only studies.
Limitations
The limitations are equally important. The retrospective design introduces inherent risks of missing data, treatment heterogeneity, and residual confounding. The number of HRD-positive cases was small, limiting power for subgroup analysis and survival comparisons. HRD was defined exclusively by pathogenic or likely pathogenic variants in homologous recombination genes, which likely underestimates functional HR deficiency and may misclassify some tumours. The abstract does not provide hazard ratios or confidence intervals, which would have helped quantify effect size precision. Single-centre data may also limit generalizability across populations with different histologic distributions, sequencing platforms, and treatment pathways.
Another conceptual limitation is that not all HRR-gene alterations are biologically equivalent. BRCA1/2 loss, ATM alteration, CHEK2 mutation, and RAD51-pathway changes may have different consequences for homologous recombination competency and drug sensitivity. Grouping them together under a single HRD umbrella is practical for exploratory study but may blur clinically meaningful distinctions.
How These Findings Fit With Existing Literature
Prior genomic studies have shown that endometrial cancer harbors diverse DNA damage repair alterations, but the prevalence of canonical homologous recombination deficiency has generally appeared lower and less therapeutically validated than in ovarian cancer. The TCGA molecular analysis established the heterogeneity of endometrial carcinoma and laid the foundation for subtype-driven management. Subsequent ESGO-ESTRO-ESP guidelines incorporated molecular classification into risk stratification and treatment recommendations, shifting the field away from a one-size-fits-all approach.
In parallel, interest in DNA repair targeting in endometrial cancer has expanded, including studies of PARP inhibitors alone and in combination regimens. Yet robust biomarkers remain unsettled. The present paper is therefore useful because it injects discipline into the conversation: HRR-gene alterations are real, but uncommon; associated with aggressive features, but not clearly prognostic; and not sufficient, as currently defined, to displace established molecular classification.
Practice Implications
For clinicians managing endometrial cancer today, the practical message is straightforward. Routine risk assessment should continue to prioritize validated clinicopathological variables together with integrated molecular classification. If HRR-gene alterations are detected on broad next-generation sequencing panels, they may be biologically interesting and potentially trial-relevant, but they should not be overinterpreted as independent prognostic markers.
For pathologists and molecular tumor boards, the study supports cautious reporting language. A pathogenic HRR-gene variant in endometrial carcinoma should prompt contextual interpretation: histology, grade, stage, molecular subtype, and whether there is evidence for germline significance or therapeutic trial eligibility. It should not be assumed to imply platinum sensitivity or favorable candidacy for PARP inhibition outside data-supported settings.
For investigators, the study underscores the need for next-generation biomarker strategies. These should distinguish prognostic from predictive utility, account for biallelic loss and functional deficiency, and evaluate therapy response prospectively in biomarker-stratified cohorts.
Conclusion
Carvalho and colleagues provide a clinically relevant caution against overextending the HRD concept in endometrial cancer. In this 298-patient cohort, pathogenic or likely pathogenic homologous recombination repair gene alterations were present in only 6.0% of tumours and were enriched in aggressive histologic and molecular subsets. Yet they did not predict overall survival, disease-free survival, progression-free survival, or platinum sensitivity. The findings reinforce that integrated molecular classification, not isolated HRR-gene status, should remain the backbone of prognostic assessment in endometrial carcinoma. HRD may still prove useful as a predictive biomarker for selected targeted therapies, but better definitions and prospective validation are needed before it can guide routine clinical decision-making.
Funding and Trial Registration
The abstract provided does not report funding information or a ClinicalTrials.gov registration number. The study was retrospective and observational.
Citation
Carvalho SN, Soldin I, Silva RM, Domingues TD, Pinto C, Barbosa A, Teixeira MR, Bartosch C, Abreu MH. Clinical relevance of homologous recombination repair gene alterations in endometrial cancer patients in the era of molecular classification. Gynecologic Oncology. 2026-06-04;210:124-130. PMID: 42242176. URL: https://pubmed.ncbi.nlm.nih.gov/42242176/
References
Cancer Genome Atlas Research Network, Kandoth C, Schultz N, Cherniack AD, Akbani R, Liu Y, Shen H, et al. Integrated genomic characterization of endometrial carcinoma. Nature. 2013;497(7447):67-73.
Concin N, Matias-Guiu X, Vergote I, Cibula D, Mirza MR, Marnitz S, Ledermann JA, Bosse T, Chargari C, Fagotti A, et al. ESGO/ESTRO/ESP guidelines for the management of patients with endometrial carcinoma. International Journal of Gynecological Cancer. 2021;31(1):12-39.
Oaknin A, Tinker AV, Gilbert L, Samouelian V, Mathews C, Brown J, Barretina-Ginesta MP, Moreno V, Gravina A, Abdrakhmanov R, et al. Clinical activity and safety of the anti-programmed death 1 monoclonal antibody dostarlimab for patients with recurrent or advanced mismatch repair-deficient endometrial cancer: a non-randomised phase 1 clinical trial. JAMA Oncology. 2020;6(11):1766-1772.
Konstantinopoulos PA, Drew Y, O’Cearbhaill RE, et al. PARP inhibitors in gynecologic malignancies: current landscape and future directions. Clinical Cancer Research. 2020;26(18):4662-4671.
