Performance Characteristics of Endometrial Sampling in Diagnosis of Endometrial Carcinoma

Outpatient endometrial biopsy can give false-negative results, with a 0.9% reported posttest probability for endometrial carcinoma (EC) after a negative result. Our objective was to determine if there has been any improvement in the performance characteristics of endometrial biopsy over the last 15 yr. All hysterectomy specimens with a diagnosis of EC or atypical hyperplasia (AH), reported between May 2011 and May 2015, were identified and cross-referenced for any negative endometrial sampling results during the 5 yr before hysterectomy. Negative endometrial samples were defined as either benign findings or insufficient/nondiagnostic, excluding those diagnosed as AH or EC and those for which follow-up sampling was recommended because of atypia. Of 1677 hysterectomy specimens showing AH or EC there were previous negative biopsies in 172: 116 benign and 56 insufficient/nondiagnostic. Over the same period 22,875 negative endometrial biopsy specimens were reported in our region. The posttest probability of having EC or AH in the hysterectomy specimen, given a negative endometrial biopsy result, was 0.74%. In a subset of 90 cases in which a negative biopsy was followed by a diagnosis of AH or EC in a hysterectomy specimen, the slides were independently reviewed. There were no cases where a diagnosis of carcinoma was missed. In 12 samples atypia or possible atypia was identified, and the level of agreement with the original diagnosis was excellent κ=0.83±0.05. In a prospective comparison of examination of 3 levels from each block versus a single slide in 319 cases, the routine preparation of additional slides did not yield clinically significant information. Although there has been evolution in the diagnostic criteria for AH and for recognition of morphologically subtle forms of AH or EC, our results demonstrate a significant lack of sensitivity of outpatient endometrial sampling in the diagnosis of endometrial malignancy/premalignancy. The sensitivity problems are mainly attributable to failure to sample abnormal endometrium. Independent review of slides or examination of additional levels did little to increase the diagnostic yield.