Clinicopathological analysis of ROS1 gene alterations and immunohistochemistry screening for ROS1 gene rearrangement in non-small cell lung cancer

Abstract

Introduction: ROS1 rearrangement, found in a subset of lung cancers, have therapeutic significance, as ROS1-rearranged tumors are sensitive to ALK kinase inhibitors. This study sought to evaluate the clinicopathological implications of ROS1 gene alterations and the histomorphological characteristics of ROS1-rearranged tumors, especially micropapillary and aerogenous spread growth, and to investigate the usefulness of ROS1 immunohistochemistry as a diagnostic test for ROS1 rearrangement. Methods: Characterization of ROS1 gene alterations using fluorescent in situ hybridization, and characterization of ROS1 and E-cadherin protein expression using immunohistochemistry were performed in 754 non-small cell lung cancer (NSCLC) surgical samples. Results: ROS1 rearrangement was identified in ten patients (1.3%, NSCLC

1.9%, adenocarcinoma). Histologically, all ten ROS1-rearranged tumors harbored an adenocarcinoma component. Importantly, ROS1 rearrangements were highly associated with micropapillary components (p < 0.001), aerogenous spread status (p = 0.002), and loss of E-cadherin expression (p = 0.049). Kaplan-Meier survival curves with log-rank test showed that ROS1 rearrangement was significantly associated with a higher risk of tumor recurrence (p = 0.024). Nine of the ten ROS1-rearranged tumors showed moderate-to-strong ROS1 immunoreactivity, with a 100?300 H-score range (median, 240), whereas most ROS1 wild-type cancers (73.3%) lacked detectable immunoreactivity (H-score range, 0?240

median, 0). The criterion that best differentiated between ROS1-rearranged and ROS1 wild-type tumors was an H-score of ≥100, with a sensitivity and specificity of 90% and 93.5%, respectively. On the other hand, ROS1 gene copy number gain (CNG) was found in 4.8% (18/375) of the tumors. ROS1 gene CNG was significantly associated with shorter disease-free survival (DFS, 12 vs. 58 months

p = 0.003) and shorter overall survival (OS, 40 vs. 67 months

p < 0.001) compared to that observed in the group without CNG. Multivariate analysis confirmed that ROS1 gene CNG was significantly associated with poorer DFS (hazard ratio [HR] = 2.16, 95% confidence interval [CI] = 1.22?3.81, p = 0.008), and poorer OS ([HR] = 2.53, 95% [CI] = 1.31?4.89, p = 0.006). Conclusion: This study demonstrated that ROS1 gene rearrangement was detected in 1.9% of surgically resected adenocarcinoma. The patients harboring ROS1-rearrangement showed shorter DFS. ROS1-rearranged lung adenocarcinoma exhibited distinct morphological and clinicopathological features, including high prevalence of cribriform and/or micropapillary pattern with aerogenous spread status and decreased membranous E-cadherin expression. Cutoff value of H-score 100 best predicts ROS1 rearrangement. On the other hand, ROS1 gene CNG is an independent indicator of poor prognosis in surgically resected NSCLC.