Objectives Endometrial carcinoma (EC) is traditionally diagnosed by histopathological assessment of endometrial biopsies, leaving up to 22% of patients undiagnosed. This study explores the feasibility of the clinical implementation of detecting EC using mutational analysis of cervical cytology.

Methods This prospective multicentre study included patients that underwent a hysterectomy for histopathologically proven EC or a benign gynecological condition (control group). A Pap brush sample, cervicovaginal self sample, pipelle and hysterectomy specimen were obtained from each patient. A targeted next-generation sequencing panel was used to screen these samples for mutations in eight genes. Diagnostic accuracy was calculated, including sensitivity, specificity and predictive values.

Results Fifty-nine EC patients and 31 control patients were included. In these patients traditional histopathological diagnosis had a sensitivity of 78.9% and a specificity of 100%. For EC patients, 96.6% of surgical specimens contained at least 1 mutation. Blinded mutational analysis of Pap brush samples, self-samples, and pipelle endometrial biopsies yielded a sensitivity of 78.0%, 67.3% and 96.5% with a specificity of 96.8%, 96.8% and 93.5%, respectively. Combining these three methods with histopathological pipelle endometrial biopsies evaluations yielded a sensitivity of 95.8%, 93.0% and 96.5 respectively.

Conclusions This study has shown the potential of mutational analysis of cervical cytology and pipelle endometrial biopsies to improve diagnosis of EC, whether or not in additional to traditional histopathological assessment. Prospective evaluation is required for validation and clinical implementation.