In a retrospective cohort study of patients undergoing surgical treatment for deep infiltrating endometriosis, researchers established that elevated ubiquitin-specific protease 14 mRNA expression in lesions is independently associated with postoperative symptomatic recurrence. Integrating this molecular biomarker with standard clinical factors produced a combined clinico-molecular predictive model that demonstrated superior prognostic accuracy compared to a model relying solely on clinical features. This approach provides measurable prognostic improvements for identifying patients facing higher likelihoods of postoperative symptom return.
Deep infiltrating endometriosis frequently exhibits high postoperative recurrence rates, presenting a persistent clinical challenge following surgical intervention. Although current risk models are used to anticipate patient trajectories, existing frameworks lack molecular insights required for precise individualized risk stratification. To address this diagnostic gap, investigators examined whether ubiquitin-specific protease 14, or USP14, mRNA expression in surgically resected lesions is independently associated with postoperative symptomatic recurrence and whether combining this biomarker with clinical variables could enhance overall predictive accuracy.
The research team evaluated a retrospective cohort of 235 surgically treated patients diagnosed with deep infiltrating endometriosis. To isolate the most informative predictor variables from the clinical and molecular data, investigators applied the least absolute shrinkage and selection operator, widely known as LASSO. Following this variable selection procedure, Cox multivariate regression analysis was conducted to establish independent risk factors and protective determinants associated with time to postoperative symptomatic recurrence across the cohort.
Multivariate Cox regression demonstrated that high USP14 expression was an independent risk factor for symptomatic recurrence, yielding a hazard ratio of 1.656 with a 95 percent confidence interval of 1.307 to 2.099 and a p-value of less than 0.001. Anatomical and clinical characteristics also emerged as independent risk factors: pelvic adhesions yielded a hazard ratio of 1.861 (95% CI: 1.013–3.419; P = 0.045), bowel lesions showed a hazard ratio of 2.168 (95% CI: 1.278–3.679; P = 0.004), and coexisting adenomyosis conferred a hazard ratio of 2.645 (95% CI: 1.568–4.462; P < 0.001).
The multivariate Cox regression simultaneously identified two independent protective factors associated with significantly lower rates of postoperative recurrence. Patients who received gonadotropin-releasing hormone agonist therapy for at least six months demonstrated a markedly reduced recurrence risk, recording a hazard ratio of 0.126 with a 95 percent confidence interval of 0.044 to 0.363 and a p-value under 0.001. In addition, achieving postoperative pregnancy served as an independent protective factor, associated with a hazard ratio of 0.402 (95% CI: 0.200–0.807; P = 0.010).
When investigators integrated USP14 expression with these identified clinical determinants, the combined model achieved a concordance index, or C-index, of 0.896 with a 95 percent confidence interval of 0.879 to 0.912. This performance outperformed the clinical-only predictive model, which achieved a C-index of 0.870 with a 95 percent confidence interval of 0.849 to 0.892. To assess model optimism, the researchers conducted internal bootstrap validation with 2,000 resamples, yielding optimism-corrected C-indices of 0.888 for the combined model and 0.863 for the clinical-only model.
Evaluation of discrimination and reclassification supported the incremental predictive utility of incorporating USP14. The integrated discrimination improvement was 0.044 (95% CI: -0.013 to 0.11; P = 0.047); because the bootstrap confidence interval narrowly included zero, investigators interpreted this finding as borderline. However, complementary evaluation metrics affirmed the biomarker’s value, showing a continuous net reclassification improvement of 0.655 (P = 0.026) and a bootstrap-validated difference in C-index of 0.020 (95% CI: 0.002–0.048; P = 0.033).
The study concluded that high USP14 expression in lesions independently correlates with postoperative symptomatic recurrence, and its addition to clinical variables provides modest but consistent prognostic improvement. Nonetheless, because this clinico-molecular model was developed retrospectively and evaluated solely through internal validation, the investigators emphasized that it remains hypothesis-generating, concluding that rigorous external, multi-center validation is required before any potential clinical implementation.
Subject of Research: Prediction of postoperative symptomatic recurrence in deep infiltrating endometriosis using lesion USP14 gene expression and clinical factors
Article Title: USP14 Gene Expression Combined with Clinical Factors for Predicting Symptomatic Recurrence After Surgery for Deep Infiltrating Endometriosis: A Retrospective Cohort Study
Article References: Zhang, P., & Xiao, X. (2026). USP14 Gene Expression Combined with Clinical Factors for Predicting Symptomatic Recurrence After Surgery for Deep Infiltrating Endometriosis: A Retrospective Cohort Study. Reproductive Sciences. https://doi.org/10.1007/s43032-026-02259-5
Image Credits: AI Generated
DOI: 10.1007/s43032-026-02259-5
Keywords: deep infiltrating endometriosis, USP14, ubiquitin-specific protease 14, symptomatic recurrence, postoperative recurrence, retrospective cohort study, LASSO regression, Cox multivariate regression, pelvic adhesion, bowel lesions, adenomyosis, GnRH-a therapy, postoperative pregnancy, C-index, bootstrap internal validation, integrated discrimination improvement, net reclassification improvement, prognostic models
Cite Scienmag News
Ophelia Keating. (November 18, 2026). Adding USP14 Expression Improves Deep Infiltrating Endometriosis Recurrence Prediction Models. Scienmag. https://scienmag.com/adding-usp14-expression-improves-deep-infiltrating-endometriosis-recurrence-prediction-models/
Ophelia Keating. "Adding USP14 Expression Improves Deep Infiltrating Endometriosis Recurrence Prediction Models." Scienmag, 18 November 2026, https://scienmag.com/adding-usp14-expression-improves-deep-infiltrating-endometriosis-recurrence-prediction-models/. Accessed 18 November 2026.
Ophelia Keating. "Adding USP14 Expression Improves Deep Infiltrating Endometriosis Recurrence Prediction Models." Scienmag. November 18, 2026. https://scienmag.com/adding-usp14-expression-improves-deep-infiltrating-endometriosis-recurrence-prediction-models/








