Poor accuracy of endometrial sampling in patients with uterine carcinosarcomas: a nationwide analysis

To determine the accuracy of aspiration biopsy (AB), hysteroscopic biopsy (HB), and dilatation & curettage (D&C) in detecting uterine carcinosarcoma (UCS).

Pathology reports were retrieved from the Dutch Nationwide Pathology Databank PALGA for patients with a certain or suggested diagnosis of UCS in pre- and/or postoperative histology between 2001 and 2021. Patients without available pre- or postoperative pathology reports were excluded. The accuracy measures sensitivity, positive predictive value (PPV), accuracy, and concordance using Cohen's kappa were calculated for AB, D&C, and HB, using postoperative histology as the reference. This was analyzed for 2 scenarios: Analysis A compared samples with a certain or suggested diagnosis of UCS vs. no mention of UCS. Analysis B compared samples with a certain diagnosis of UCS vs those without UCS.

The study included 1,481 patients, totaling 1,685 samples. Sensitivity was similar for AB and HB (52.4% and 50.5%, respectively, for analysis A; 45.1% and 42.2% for analysis B). D&C showed the highest sensitivity (70.8% and 64.9% for analysis A and B, respectively). AB had the highest PPV (85.3% and 90.9% for analysis A and B, respectively), HB had the lowest PPV (79.7% and 80.9%, respectively). Accuracy was highest for D&C (44.4%) compared to AB (32.8%) and HB (29.5%). All Cohen's kappa values were below 0.20, indicating poor correlation between preoperative and postoperative diagnoses.

The study reveals low accuracy measures across all conventional endometrial sampling techniques, highlighting the need for research to identify markers or tools to diagnose UCS.