Ki-67 Index of 55% Distinguishes Two Groups of Bronchopulmonary Pure and Composite Large Cell Neuroendocrine Carcinomas with Distinct Prognosis

Background: Little information is available concerning prognostic factors for bronchopulmonary large cell neuroendocrine carcinomas (BP-LCNECs) and even less is known about combined LCNECs (Co-LCNECs). We investigated whether an integrated morphological, immunohistochemical, and molecular approach could be used for their prognostic evaluation. Methods: Morphological (including combined features), proliferative (mitotic count/Ki-67 index), immunohistochemical (napsin A, p40, TTF-1, CD44, OTP, SSTR2A, SSTR5, mASH1, p53, RB1, and MDM2), and genomic (TP53, RB1, ATM, JAK2, KRAS, and STK11) findings were analyzed in BP-LCNECs from 5 Italian centers, and correlated with overall survival (OS). The Ki-67 index was expressed as the percentage of positive cells in hot spots as indicated in the WHO 2019 Digestive System Tumors and, for Co-LCNECs, the Ki-67 index was evaluated only in the LCNEC component. Results: A total of 111 LCNECs were distinguished into 70 pure LCNECs, 35 Co-LCNECs (27 with adenocarcinoma [ADC] and 8 with squamous cell carcinoma [SqCC]), and 6 LCNECs with only napsin A immunoreactivity. The Ki-67 index cutoff at 55% evaluated in the neuroendocrine component was the most powerful predictor of OS (log-rank p = 0.0001) in all LCNECs; 34 cases had a Ki-67 index p = 0.0001) were also observed between pure and Co-LCNECs. A significant difference in OS was found between pure LCNECs-A and Co-LCNECs-A (p < 0.05) but not between pure LCNECs-B and Co-LCNECs-B. Co-LCNEC-ADC and LCNEC napsin A+ cases had longer OS than pure LCNEC and Co-LCNEC-SqCC cases (log-rank p = 0.0001). On multivariable analysis, tumor location, pure versus combined features, and napsin A, but no single gene mutation, were significantly associated with OS after adjustment for Ki-67 index and study center (p < 0.05). Conclusions: The Ki-67 proliferation index and the morphological characterization of combined features in LCNECs seem to be important tools for predicting clinical outcome in BP-LCNECs.