The Predictive Values of Advanced Non-Small Cell Lung Cancer Patients Harboring Uncommon Mutations-The Mutation Patterns, Use of Different Generations of -TKIs, and Concurrent Genetic Alterations.

Introduction: Epidermal growth factor receptor () and mutation are known as "common mutations" in non-small cell lung cancer (NSCLC) and predict sensitivities to tyrosine kinase inhibitors (TKIs), whereas and mutations confer drug-resistance to -TKIs. The role of the remaining uncommon mutations remains elusive.

Methods: We retrospectively screened a group of NSCLC patients with uncommon mutations other than and . The mutation patterns, use of different generations of -TKIs, and concurrent genetic alterations were analyzed. Meanwhile, a cohort of patients with single or were included for comparison.

Results: A total of 180/1,300 (13.8%) patients were identified. There were 102 patients with advanced or recurrent NSCLC that received first-line therapy of gefitinib/erlotinib/icotinib and afatinib and were eligible for analysis. The therapeutic outcomes among patients with common mutations (cm, n = 97), uncommon mutation plus common mutations (um+cm, n = 52), complex uncommon mutations (complex um, n = 22), and single uncommon mutations (single um, n = 28) were significantly different (ORRs: 76.3%, 61.5%, 54.5%, and 50.0%, respectively, p = 0.023; and mPFS: 13.3, 14.7, 8.1, and 6.0 months, respectively, p = 0.004). Afatinib showed superior efficacy over gefitinib/erlotinib/icotinib in cm (ORR: 81.0% . 75.0%, p = 0.773; mPFS: 19.1 . 12.0m, p = 0.036), um+cm (ORR: 100% . 54.5%, p = 0.017; mPFS: NE . 13.6m, p = 0.032), and single um (ORR: 78.6% . 21.4%, p = 0.007; mPFS: 10.1 . 3.0m, p = 0.025) groups. Comprehensive genomic profiling by Next Generation Sequencing encompassing multiple cancer-related genes was performed on 51/102 patients; the mPFS of patients without co-mutation (n = 16) and with co-mutations of tumor-suppressor genes (n = 31) and driver oncogenes (n = 4) were 31.1, 9.2, and 12.4 months, respectively (p = 0.046). mutation was the most common co-alteration and showed significantly shorter mPFS than wild-type patients (7.0 . 31.1m, p < 0.001). Multivariate analysis revealed that concurrent and tumor-suppressor gene alterations independently predicted better and worse prognosis in patients with uncommon mutations, respectively.

Conclusions: Uncommon mutations constitute a highly heterogeneous subgroup of NSCLC that confer different sensitivities to -TKIs with regard to the mutation patterns. Afatinib may be a better choice for most uncommon mutations. Concurrent and tumor-suppressor gene alterations, especially , can be established as prognostic biomarkers.