Two‐ and three‐dimensional saline contrast sonohysterography: interobserver agreement, agreement with hysteroscopy and diagnosis of endometrial malignancy

Abstract

The aims of our study were to compare the interobserver reproducibility of two-dimensional (2D) and three-dimensional (3D) saline contrast sonohysterography (SCSH) and agreement of these techniques with hysteroscopy, and to determine which SCSH findings best discriminate between benign and malignant endometrium. Consecutive women with postmenopausal bleeding and endometrial thickness > or = 4.5 mm underwent 2D and 3D SCSH; the results were videotaped and stored electronically, respectively, for analysis by two independent experienced examiners who were blinded to each other's results. A histological diagnosis was obtained by dilatation and curettage, hysteroscopic resection or hysterectomy. The hysteroscopist was blinded to the ultrasound results and used the same standardized research protocol to describe the uterine cavity as the ultrasound examiners. Of 170 consecutive women with postmenopausal bleeding and endometrial thickness > or = 4.5 mm, 84 (14 with endometrial malignancy) fulfilled our inclusion criteria. Hysteroscopy findings in 54 women (one with endometrial malignancy) were used to determine agreement with SCSH. Interobserver agreement of 2D and 3D SCSH was 95% (80/84) vs. 89% (75/84) with regard to presence of focal lesions, 89% (75/84) vs. 88% (74/84) for presence of focal lesions with irregular surface, 67% (54/81) vs. 63% (51/81) for number of focal lesions, and 77% (46/60) vs. 70% (42/60) for location of focal lesions. The agreement between 2D and 3D SCSH and hysteroscopy was 94% (51/54) vs. 93% (50/54) with regard to presence of focal lesions, 74% (40/54) vs. 76% (41/54) for presence of focal lesions with irregular surface, 63% (34/54) vs. 54% (29/54) for number of focal lesions, and 66% (29/44) vs. 64% (28/44) for location of focal lesions. The SCSH finding that best discriminated between benign and malignant endometrium was the presence of focal lesion(s) with irregular surface (for 2D SCSH: sensitivity 71%, specificity 97%, positive likelihood ratio 25, negative likelihood ratio 0.3; for 3D SCSH: sensitivity 43%, specificity 97%, positive likelihood ratio 15, negative likelihood ratio 0.6). 3D SCSH does not seem to be superior to 2D SCSH when performed by experienced ultrasound examiners either with regard to reproducibility, agreement with hysteroscopy findings or diagnosis of endometrial malignancy. The presence of focal lesion(s) with irregular surface is the best SCSH variable for discrimination between benign and malignant endometrium.