Diagnostic accuracy of endometrial polyps in pipelle endometrial samples: a clinicopathologic study of 195 cases

Abstract

BACKGROUND: Endometrial polyp is a common cause of abnormal uterine bleeding. Initial investigation in affected patients usually includes a pipelle endometrial (PE) sampling at an outpatient setting. When polyp is suspected histologically, subsequent confirmation would normally be done by diagnostic hysteroscopy (DH) and biopsy, or even operative hysteroscopy under general anaesthesia in some cases. Although the histological features of polyps are well-described, their presence is more variable in small and fragmented PE samples. Diagnostic accuracy in these samples has seldom been studied. DESIGN: 195 women who had undergone DH and/or polypectomy were identified in a University teaching hospital. All patients had a prior polyp diagnosis in the PE sample. The histology of these samples were compared and analyzed with subsequent DH findings and final hysteroscopic biopsies. Slides were reviewed by 2 gynecological pathologists. RESULTS: 162 were premenopausal (age 26-58, median, 46) and 33 were postmenopausal (age 44-85, median, 57). The commonest indications for a PE sampling were menorrhagia, metrorrhagia and postmenopausal bleeding. The median time intervals between initial PE sampling and DH for those with and without a final polyp diagnosis were 6 and 7.5 months, respectively. Presence of polyp was confirmed by DH in 56.3% (111/195) cases. Of these, 81.1% (90/111) were confirmed histologically. The commonest histologic feature in PE samples was thick-walled/ectatic vessels, found in 91.8% of cases (179/195). However, a polyp was only detected during the ensuring DH in 59.2% (n=106) of these 179 cases. By univariate analysis, the most reliable histologic features that can predict the presence of an underlying polyp were fibrous stroma (p= 0.01) and focal glandular clustering (p=0.03). The accuracy of detection of polyps in PE samples for premenopausal and postmenopausal women were 53.7% and 72.7%, respectively (p=0.05). CONCLUSIONS: The positive predictive value of PE samples in detecting endometrial polyp was only 56.3%. Although the presence of thick-walled/ectatic vessels was the commonest histologic finding, the only reliable features in our series that can predict the presence of an underlying polyp were fibrous stroma and focal glandular clustering. Pathologists should be cautious in making a definitive diagnosis of polyp in pipelle endometrial samples as this may lead to unnecessary operative procedures.