Pragmatic preoperative molecular classification of endometrial cancers; Replacing POLE sequencing with hormone receptor staining

Molecular classification including POLE sequencing is encouraged for all endometrial cancer (EC) patients, although the relevance of POLE sequencing has been questioned. We aimed to determine whether POLE sequencing can be omitted when introducing hormone receptor (HR) immunohistochemical staining. Preoperative EC biopsies were molecularly classified according to two different algorithms: a pragmatic approach including staining of HRs, p53 and mismatch repair (MMR) proteins but omitting POLE sequencing (n = 534), and the conventional algorithm including POLE sequencing and with patients completely overlapping with the pragmatic cohort (n = 505). The two algorithms were tested by Cox proportional hazard analysis calculating the probability of disease-specific survival (DSS) (hazard ratios) with 95 % confidence intervals. Both classifiers identified four patient groups with significantly different prognosis (p < 0.001). However, with the pragmatic approach, the group with lowest risk included 44 % of the patients, as compared to 7 % using the conventional algorithm. In the preoperative setting, Cox proportional hazard analysis demonstrated that the pragmatic algorithm was stronger to predict DSS than the conventional algorithm. Only 1.4 % (7/498) of patients failed to be identified for potential POLE-driven de-escalation of treatment using the pragmatic algorithm. Preoperative evaluation of ER, PR, MMR and p53 identifies four molecular and prognostic groups. This pragmatic classification allows POLE sequencing to be omitted. Our proposed algorithm saves time and resources and is a valuable alternative to full molecular classification.