Morphologic features in a series of 352 Spitz melanocytic proliferations help predict their oncogenic drivers

International audience; Spitz nevi are indolent melanocytic tumors arising preferentially during and after childhood. Over the last decades, recurrent oncogenic drivers, sparsely detected in melanoma, were identified in Spitz melanocytic proliferations. Therefore, the detection of such drivers appears as a relevant diagnostic tool to distinguish both entities. Interestingly, morphologic features might correlate with the oncogenic drivers. Thus, the goal of this study was to assess the performances of previously identified morphological criteria to predict the presence of specific drivers. In total, 352 Spitz melanocytic proliferations either with a genetically identified oncogenic driver or investigated for ALK, ROS1, and NTRK1 overexpression by immunohistochemistry were enrolled in the present study. The microscopic features of the cases were assessed blindly with regards to the molecular status and, performances of previously described morphological criteria to predict the molecular status were assessed applying the likelihood-ratio test (LHR). Overall, an oncogenic driver was identified in 76% of the cases (n = 268/352). No microscopic features allowed the reliable prediction of ROS1- and NTRK1-overexpressing cases. By contrast, a plexiform pattern can contribute to the recognition of ALK-overexpressing cases (LHR(+) = 6.14). Importantly, the pseudo-schwannoma variant was highly suggestive of NTRK3-rearranged cases (LHR(+) = 43). Moreover, atypical/malignant tumor (LHR(+) = 5.18), severe cellular atypia (LHR(+) = 5.07), and p16 loss (LHR(+) = 14) contribute to the recognition of MAP3K8-rearranged cases, while the presence of a sheet-like architecture (LHR(+) = 5.39) and a marked fibrosis of the stroma (LHR(+)=5.06) were predictive of BRAF-fused tumors. To conclude, our study confirms ALK-overexpressing, NTRK3-, MAP3K8-, and BRAF-rearranged cases harbored distinct morphologic features allowing their microscopic recognition.