Less is more? Comparison between genomic profiling and immunohistochemistry-based models in endometrial cancer molecular classification: A multicenter, retrospective, propensity-matched survival analysis.

Genomic profiling-based model (GP-M) is the gold-standard for endometrial cancer (EC) molecular classification, but several issues related to the availability of genomic sequencing in low-income settings remain and health disparities in the management are increasing. This study aims to investigate the non-inferiority of the immunohistochemistry-alone model in classifying ECs compared to the standard genomic profiling-based model in terms of oncologic outcomes. All preoperative uterine-confined ECs undergoing surgical staging were retrospectively included. Patients classified by IHC-M were stratified into: MMR-proficient (MMRp), p53 wild type (p53wt) and estrogen receptor (ER) positive, 2) MMRp, p53wt and ER-negative, 3) MMRd, and 4) p53abn. A case-control comparison was performed between the IHC-M and GP-M cohorts. Then, a propensity-matched analysis was performed: ECs classified by IHC-M were matched in a 3:1 ratio with patients classified by GP-M. 1587 patients with EC were included. The Kaplan-Meier survival curves for disease-free survival and overall survival demonstrated that the two models performed similarly in risk-stratifying the study population (p < 0.0001). Moreover, the AUC-ROC showed overlapping results: 0.77 (0.66–0.87) for IHC-M and 0.72 (0.63–0.81) for GP-M, indicating that both models were able to successfully identify patients at high-risk and low-risk of disease recurrence/progression. The IHC-M showed overlapping classification performance compared to the GP-M in terms of oncologic outcomes. This study may lay the basis to further investigate the real-life clinical impact of POLE sequencing in molecular classification and the potential stand-alone prognostic role of ER status for further allocation of EC patients into risk classes. • The ProMisE model for endometrial cancer molecular classification is the standard. • However, genome sequencing has limited availability and cost-effectiveness. • Immunohistochemistry-alone model and ProMisE model predicted prognosis similarly. • Both models were able to successfully identify patients at high-risk of recurrence. • Estrogen receptor status should be integrated in the molecular classification model. [ABSTRACT FROM AUTHOR]