Immunohistochemical Assessment of β-Catenin in Endometrial Carcinoma

To the Editor The TCGA molecular classification is progressively being introduced into the guidelines for the management of endometrial carcinoma, but several critical issues should be resolved. One major criticism lies in the highly heterogeneous prognosis of the “no specific molecular profile” (NSMP) group. This group does not have molecular signatures and is defined by the exclusion of the signatures of the other groups (ie, POLE mutation, TP53 mutation/p53 abnormality, microsatellite instability/mismatch repair deficiency). A possible substratification of the NSMP group might be based on the presence of CTNNB1 exon 3 mutation, which seems to define a more prognostically homogenous group characterized by intermediate prognosis. In a previous meta-analysis conducted by 3 of us, nuclear β-catenin expression was assessed as an immunohistochemical surrogate of CTNNB1 mutation, showing 88% sensitivity and 85% specificity, with an area under the curve (AUC) of 91%.1 Remarkably, the only study that adopted next-generation sequencing (NGS) to detect CTNNB1 mutations reported a 100% specificity for β-catenin as a surrogate marker.2 We therefore hypothesized that the use of NGS techniques would lead to a better definition of the diagnostic accuracy of β-catenin.1 Herein, we report an update of the previous meta-analysis, performed by searching electronic databases from October 2018 to September 2021. We screened all studies assessing both β-catenin immunohistochemical expression and CTNNB1 gene status and selected only studies that adopted NGS as the reference standard. We included 3 new studies in addition to the 1 included in our previous meta-analysis, for a total of 322 patients.2-5 We found that nuclear β-catenin expression had a sensitivity and specificity of 85% and 98%, respectively, in predicting CTNNB1 exon 3 mutations. The positive and negative likelihood ratios were 28.41 and 0.14, respectively; the diagnostic odds ratio was 250.71, and the AUC was 96.4% (Supplemental Figure 1; all supplemental materials can be found at American Journal of Clinical Pathology online). These results confirm nuclear β-catenin accumulation as an accurate surrogate of CTNNB1 mutation, revealing a considerably higher specificity than in the previous meta-analysis (likely because of the higher sensitivity of NGS as a reference standard). In contrast, the sensitivity of β-catenin still appears suboptimal. Remarkably, the prognostic value of CTNNB1 mutations seems particularly applicable to low-grade endometrioid carcinomas. In these tumors, the nuclear localization of β-catenin is often found in areas of altered differentiation, especially in squamous/morular metaplasia (with or without keratinization) (Supplementary Figure 2) and in the rare corded and hyalinized pattern.6,7 As a result, we suggest that if areas with altered differentiation are present, sensitivity may be improved by assessing β-catenin in these areas rather than in conventional glandular areas. Further studies are encouraged to assess the causes of the discrepancy between CTNNB1 and β-catenin status, including analytical issues, different subsets of CTNNB1 mutations, or distinct biological tumor subtypes. Given the potential role of the Wnt–β-catenin pathway as a therapeutic target in endometrial carcinoma,8 this topic could acquire even further importance in the future.