Diagnostic rate and accuracy of the ESHRE-ESGE classification for septate uterus and other common uterine malformations: why do we not see that the Emperor is naked?

Abstract

In our opinion, the results of the recent retrospective study of Graupera et al.1 on the European Society of Human Reproduction and Embryology–European Society for Gynaecological Endoscopy (ESHRE–ESGE) classification system are biased2. The main issues are described below. The ESHRE–ESGE system2 for classification of congenital anomalies of the female genital tract introduced ‘normal uterus’ as a separate class, whereas the original American Society for Reproductive Medicine (ASRM, previously the American Fertility Society (AFS)) classification does not define normal uterus as a class3. Graupera, in the original study of these patients4, included those suspected on two-dimensional ultrasonography to have a congenital uterine malformation according to the ASRM classification. Thus, the population may have been inappropriate for estimating the accuracy of three-dimensional ultrasonography (3D-US) using the ESHRE–ESGE system. Additionally, in the primary study, the patients were assigned arbitrarily to undergo magnetic resonance imaging (MRI) or hysteroscopy, which could have introduced a differential-verification bias4. In patients enrolled consecutively, the rate of diagnosis of uterine malformations is higher with the ESHRE–ESGE classification, because of the chances of overdiagnosis of septate uterus when no anomalies are detected according to the ASRM3. Additionally, the criteria for ESHRE–ESGE subclass U1c (uterine cavity shape: internal indentation < 50% myometrial thickness), overlap with those of ESHRE–ESGE normal uterus (uterine cavity shape: straight, curved interostial line or internal indentation < 50% myometrial thickness)2. If this confusing subclass is not excluded, there is an unacceptable relative risk (RR, 4.5; 95% CI, 3.4–6; P < 0.01) of overdiagnosis of all the anomalies3. Thus, in this study1, an important part of the population (patients without suspicion of uterine malformations according to ASRM, i.e. who have normal or U1c dysmorphic uterus, some with partially septate uteri according to ESHRE-ESGE) is omitted2. In the total group that Graupera et al.1 excluded from secondary analysis (i.e. patients without ultrasound suspicion of malformations and patients who underwent only hysteroscopy as the reference standard), the ESHRE–ESGE classification may be most unsatisfactory with regards to detecting the degree of distortion of the internal and external structure of the uterus3, reliability (κ < 0.9)5 and, ultimately, diagnostic accuracy. Uterine wall thickness differs in various parts of the uterus6. The method for measuring uterine wall thickness is not included in the ESHRE–ESGE classification, and it was alluded to during, but not completely clarified by, our correspondence with the authors of the ESHRE–ESGE classification6. Therefore, the method for measuring wall thickness needs to be presented in detail (with respect to coronal or sagittal view, and location of the measurement). Comparison of the primary data and their secondary analysis shows that the ESHRE–ESGE classification using MRI resulted in a significantly higher rate of diagnosis of septate uterus than did the ASRM classification (42/60 vs 17/60; RR, 2.47; 95% CI, 1.6–3.8; P < 0.01). The enrollment of women with a normal uterus could increase this risk further. These findings confirm our earlier assumptions6, 7 and recent results with 3D-US3. It would be valuable to determine the diagnostic accuracy of the ESHRE–ESGE system for the detection of septate uterus by 3D-US and compare it with the hysteroscopic findings in the primary study. Such a comparison may highlight the inability of the ESHRE–ESGE classification for accurate recognition of common distortions of the uterine cavity architecture (Figures 1 and 2), even in a dichotomous situation (between septate uterus and other morphological forms or septate + bicorporeal uterus and other morphological forms). In the current study, we think that the conclusion of Graupera et al. is overly optimistic8, and that further analysis based on our suggestions would help streamline their interpretation. A. Ludwin* and I. Ludwin Department of Gynecology and Oncology, Jagiellonian University, ul. Kopernika 23, Krakow, 31-501, Poland; Ludwin & Ludwin Gynecology, Private Medical Center, Krakow, Poland *Correspondence. (e-mail: [email protected])