A continuously updating database of clinical and genomic details on patients with non–small cell lung cancer accurately represented correlations between genomics and outcomes, based on an analysis of more than 4,000 patients.
“Most efforts to identify clinicogenomic associations currently rely on clinical trials, single-institution series, or national registries,” wrote Gaurav Singal, MD, of Foundation Medicine in Cambridge, Mass., and colleagues in JAMA.
To explore the feasibility of a clinicogenomic database, the researchers combined clinical data from electronic health records with comprehensive genetic profiling data from 28,889 patients; 4,064 adults with non–small cell lung cancer were included in the analysis of associations among tumor genomics, patient characteristics, and clinical outcomes. The data were collected between Jan. 1, 2011, and Jan. 1, 2018, from 275 U.S. oncology practices.
The researchers examined implications of clinical and genomic features for 3,522 patients with advanced disease. Among these, the median overall survival was 10.3 months and the 5-year survival rate was 3.8%. Factors influencing a longer overall survival included never smoking and having nonsquamous pathology; the presence of mutations in genes TP53 and RB1 were associated with shorter survival.
For each patient, researchers calculated the tumor mutational burden (TMB), defined as “a measure of the number of somatic mutations identified per megabase of DNA sequenced.” TMB was significantly higher among smokers, compared with nonsmokers, and “alterations in EGFR, ALK, ROS1, and RET were associated with significantly lower TMB than wild-type cases,” the researchers wrote.
Overall, the results “replicated previously described associations between clinical and genomic characteristics, driver mutations and response to targeted therapy, and TMB and response to immunotherapy,” the researchers wrote.
The findings were limited by several factors, notably the quality and completeness of mortality data, as well as potential biases from the inclusion of comprehensive genetic profiling results and analysis of therapeutic exposures in an unrandomized trial, as well as a study population limited to patients with advanced stage disease, the researchers noted.
However, the results support data from similar studies and further show that clinicogenomic databases can be used in research to augment drug development and improve the design of clinical trials, they wrote.
The study was supported by Flatiron Health and Foundation Medicine, which are both owned by the Roche Group. Dr. Singal and several coauthors are employees of Foundation Medicine.
SOURCE: Singal G et al. JAMA. 2019;321:1391-9.