The Genomic Landscape of Alterations and Associations with Outcomes in Patients with Lung Cancer.

Purpose: mutations are among the most common recurrent alterations in non-small cell lung cancer (NSCLC), but the relationship to other genomic abnormalities and clinical impact has not been established.

Experimental Design: To characterize alterations in NSCLC, we analyzed the genomic, protein expression, and clinical outcome data of patients with alterations treated at Memorial Sloan Kettering.

Results: In 4,813 tumors from patients with NSCLC, we identified 8% ( = 407) of patients with -mutant lung cancer. We describe two categories of mutations: class 1 mutations (truncating mutations, fusions, and homozygous deletion) and class 2 mutations (missense mutations). Protein expression loss was associated with class 1 mutation (81% vs. 0%, < 0.001). Both classes of mutation co-occurred more frequently with , and mutations compared with wild-type tumors ( < 0.001). In patients with metastatic NSCLC, alterations were associated with shorter overall survival, with class 1 alterations associated with shortest survival times ( < 0.001). Conversely, we found that treatment with immune checkpoint inhibitors (ICI) was associated with improved outcomes in patients with -mutant tumors ( = 0.01), with class 1 mutations having the best response to ICIs ( = 0.027).

Conclusions: alterations can be divided into two clinically relevant genomic classes associated with differential protein expression as well as distinct prognostic and treatment implications. Both classes co-occur with , and mutations, but individually represent independent predictors of poor prognosis. Despite association with poor outcomes, -mutant lung cancers may be more sensitive to immunotherapy.