P-325 Can stored ultrasound images be used to identify adenomyosis? assessing the interobserver variability in sonographic diagnosis of adenomyosis

Abstract Study question Can different observers successfully identify adenomyosis, based on the revised MUSA (Morphological Uterus Sonographic Assessment) criteria, using archived ultrasound pictures? Summary answer Stored ultrasound pictures are insufficient to identify adenomyosis because judgment of the requisite diagnostic adenomyosis characteristics is greatly dependent on the viewer. What is known already The incidence of adenomyosis varies between 20-30% and yet an exact number is difficult to assess due to the lack of standardized diagnostic criteria. In 2022, the revised MUSA criteria were released to better diagnose and define the adenomyotic lesions. Given the often-complex ultrasound structure of a uterus throughout the fertile lifespan of women and the fact that the diagnosis of adenomyosis is based on sonographic assessment of direct and indirect signs, intra- as well as inter-observational discrepancies are highly likely and previous studies have described a variable reproducibility of the sonographic diagnostic criteria. Study design, size, duration A review of stored ultrasound images of 2210 patients attending the clinic for infertility treatment was conducted by 2 reproductive endocrinology and infertility specialists in a tertiary IVF referral center. The images reviewed were taken during women’s initial visit and the treatment course and extracted from the clinical documentation system between 2017 and 2024. Participants/materials, setting, methods Patients with clear sonographic images were included, while patients with fibroids were excluded. The 2 physicians were blinded to the original diagnosis of adenomyosis documented in the charts. The physicians reviewed the images and based on the revised MUSA criteria decided whether the patient had adenomyosis or not with a description of the lesions (direct and indirect signs) when needed. Statistical analysis using the Cohen Kappa test was then performed on the concordance rate. Main results and the role of chance The analysis revealed a fair agreement of kappa value of 0.261, 95% CI (0.203 - 0.331) in the diagnosis of adenomyosis. Subanalysis of the direct signs of adenomyosis revealed “hyperechogenic islands” as the direct sign with the highest concordance rate albeit being fair with a kappa value of 0.271, 95% CI (0.201 - 0.344), while “subendometrial buds” and “myometrial cysts” had a poor agreement with kappa values 0.156, 95% CI (0 - 0.455) and 0.13, 95% CI (0.07-0.209) respectively. Indirect signs of adenomyosis were also analysed. The “asymmetrical thickening of the myometrium” had the highest agreement rate despite being only fair with a kappa value of 0.20, 95% CI (0.126 - 0.293). The other indirect signs of “globular uterus”, “fan shaped shadowing” and “irregular junctional zone” revealed a poor agreement score with values of 0.04, 95% CI (0- 0.128), 0.057, 95% CI (0.007 - 0.18) and 0, 95% (0 - 0) respectively. Limitations, reasons for caution A limitation of the study is that the use of stored static images does not allow a dynamic assessment, the use of colour doppler as well as 3-dimensional ultrasounds, thus hindering the proper detection of adenomyosis as well as the description of the signs. Wider implications of the findings The findings of our study highlight the importance of using uniform diagnostic criteria for adenomyosis while performing advanced targeted ultrasounds to facilitate the diagnosis as well as the management of infertility in patients with adenomyosis. Stored ultrasound images are not an accurate tool for the diagnosis of adenomyosis. Trial registration number No