SAT352 Reliability Of Two- And Three-dimensional Ultrasonographic Assessments Of Ovarian Morphology In Adolescents

Abstract

Abstract Disclosure: E. Rahman: None. B. Smith: None. H. Vanden Brink: None. N. Zaman: None. A. Malik: None. T. Burgert: None. R. Barral: None. M.E. Lujan: None. Background: Distinct features of ovarian morphology using transabdominal ultrasonography (TAUS) have been observed in adolescents with polycystic ovary syndrome (PCOS). Standardized approaches to assess ovarian morphology using transvaginal ultrasonography result in excellent reliability and close agreement among multiple raters, which can lead to more precise estimates of follicle counts and ovarian size. The degree to which these methods translate to TAUS is unknown. Establishing standardized methods to analyze ovarian features via TAUS are needed to clarify the utility of ovarian morphology in the clinical evaluation of PCOS in adolescence. Objectives: To determine the reliability of follicle counts and ovarian size assessments obtained using 2D and 3D TAUS in early adolescence. Methods: Three raters evaluated images of 30 ovaries garnered from adolescents via TAUS (n=19, 6 - 20 months post-menarche) for the following 2D parameters: follicle number per ovary (2D FNPO), follicle number per cross-section (FNPS), ovarian volume (2D OV), ovarian area (OA) and stromal area (SA). Three dimensional assessments of follicle counts (3D FNPO) and ovarian volume (3D OV) were also performed using SonoAVCantralTM and VOCALTM software (GE Healthcare), respectively. Intraclass correlation coefficients (ICCs) were used to determine interrater reliability for each variable. An ICC <0.5 was defined as poor, 0.50-0.75 as moderate, 0.75-0.90 as good, and >0.900 as excellent reliability. Results: Across 2D measures of ovarian morphology, interrater reliability was highest for 2D FNPO (ICC = 0.698), followed by OA (0.655), SA (0.619), and FNPS (0.516). Ovarian volume measurements by 2D methods demonstrated poor reliability (0.297). Three-dimensional assessments of follicle counts were not associated with improved interrater agreement (0.609). By contrast, better reliability was detected for 3D OV versus 2D OV - albeit 3D measures still met criteria for poor reliability (0.397 vs 0.297). Conclusions: Assessments of ovarian morphology using TAUS are associated with moderate-to-poor reliability using standardized approaches. Follicle counts per ovary were moderately reliable with 2D and 3D TAUS, in contrast to OV which demonstrated poor reliability across raters irrespectively of modality. Poor reliability for measures of OV is concerning given current practice guidelines for the ultrasonographic assessment of PCOS advocate for measures of OV over FNPO when using TAUS. Factors impacting reliability, such as gynecological age, body habitus, and image quality will be evaluated in subsequent investigations to refine training, image acquisition protocols, and method development. Presentation: Saturday, June 17, 2023