Congenital Uterine Malformation by Experts (CUME): T-shaped uterus.

OBJECTIVES: To assess whether is there any uterine measurement that is reliable and accurate to distinguish between T-shaped and normal/arcuate uterus considering the most voted option by 15 experts as the reference standard. METHODS: This was a prospectively designed multi-rater reliability/agreement study with elements of diagnostic accuracy study performed between Nov-2017 and Dec-2018 in a sample of 100 3D datasets of different uteri acquired in consecutive women with the presence of lateral uterine cavity indentations between 2014-2016. Fifteen blinded representative experts (5 clinicians, 5 surgeons, and 5 imaging specialists) provided their independent opinion whether that the uterus was T-shaped or not regarding anonymized images of the coronal plane of each uterus. Two other blinded experienced observers performed 15 measurements using the originally acquired 3D data-sets. The agreement between experts was assessed by kappa and percent agreement. The inter-observer reliability of measurements was assessed using the concordance correlation coefficient (CCC). The diagnostic test accuracy was assessed using the area under ROC curve (AUROC) and the best cut-off value was assessed using Youden's index, using the most voted option by the 15 experts as the reference standard. Sensitivity, specificity, negative and positive-likelihood ratio (LR- and LR+) and post-test probability were calculated. RESULTS: There were 20 T-shaped and 80 normal/arcuate uteri using CUME as reference standard (at least 8 votes). Single experts recognized from 5 to 35 (median = 19) T-shaped uteri by subjective judgments. The agreement among experts was 82% with kappa = 0.43. We identified three measurements with good diagnostic test accuracy considering CUME as reference standard: lateral indentation angle (AUROC=0.95), lateral indentation depth (AUROC=0.92), and T-angle (AUROC=0.87). From these three measurements, T-angle was the one with the best inter-observer reproducibility: CCC = 0.87 vs 0.82 vs. 0.62 (T-angle, lateral indentation depth and angle respectively). The best cut-offs values for these measurements were: lateral indentation angle ≤ 130° (sensitivity = 75%, specificity = 96%, LR- = 0.21, LR+ = 21, positive post-test probability = 83%), lateral indentation depth ≥ 7 mm (sensitivity = 85%, specificity = 78%, LR- = 0.06, LR+ = 4.2, positive post-test probability = 49%), and T-angle ≤ 40° (sensitivity = 65%, specificity = 91%, LR- = 0.23, LR+ = 6.4, positive post-test probability = 64%). We suggest considering as borderline T-shaped when only 2 of these 3 criteria (PPV = 50%) are present and definitely T-shaped uterus when meeting all the three criteria (PPV = 93%), with 75% and 93% post-test probability for the definitions, respectively. CONCLUSIONS: The diagnostic of T-shaped uterus is not easy and the agreement among top-experts is only moderate, and single expert judgment is commonly insufficient for accurate diagnosis. The study has identified three measurements with cut-offs that had good diagnostic test accuracy and fair to moderate reliability (lateral indentation depth ≥ 7 mm, lateral indentation angle ≤ 130° and T-angle ≤ 40°), and when applicated together they have provided high post-test probability of this condition. Based on the CUME criteria of T-shaped uterus, the prevalence, clinical implication and the management, as well as assessment of post-surgical morphologic outcomes of this condition may be determined with enough accuracy, reliability and with a known probability of disease after negative and positive test results. The CUME definition of T-shape uterus may help on the development of interventional randomized controlled trials, observational studies, and diagnostics of uterine morphology in every day practice, and therefore could be adopted by guidelines on uterine anomalies to enrich their classification systems. T