CGE21-031: Genomic Biomarker Testing, Treatments, and Survival Outcomes Among Patients With Advanced or Metastatic NSCLC in the US: A Retrospective Cohort Study

Objectives: This study identified factors associated with genomic biomarker testing (“tested” vs “not tested”) and examined the association of the receipt of treatment that is guided by genomic biomarker testing results with survival outcomes among patients with advanced/metastatic NSCLC in the US. Methods: This retrospective observational study used Flatiron Health’s electronic health records data (01JAN2015 to 31DEC2019) and included patients with non-squamous advanced/metastatic NSCLC. Eligible patients were categorized as “tested” if they received testing for genomic biomarkers including ALK, EGFR, ROS1, and/or BRAF prior to receiving first-line treatment and as “not tested” if otherwise. These patients were further categorized into four groups: (a) “tested-positive-concordant” (patients who were tested positive for a biomarker and received NCCN-recommended first-line treatment), (b) “tested-positive-discordant” (patients who were tested positive for a biomarker but did not receive NCCN-recommended first-line treatment), (c) “tested-but no positive results” (those who did not have a positive biomarker result), and (d) “not tested”. Logistic regression and inverse weighting cox regression models were used to evaluate factors associated with testing and the association of testing-guided treatment with survival outcomes, respectively. Results: 13,770 patients (48.3% female; mean age (standard deviation) 68.0 (9.6) years) were included. Most patients (82.6%) were tested for one of the four biomarkers before initiating first-line treatment. Factors associated with testing included smoking status, gender, body mass index, performance status, stage at initial diagnosis, year of advanced diagnosis, practice site and hospital’s practice volume (all p<0.05). After adjusting for all baseline covariates, the hazard of death was 17% lower in patients in the “tested-positive-concordant” group compared those in the “tested-positive-discordant” group (adjusted hazard ratio [aHR]=0.83, 95% confidence interval [CI]=0.80, 0.86). Likewise, the adjusted hazard of death was 50% higher in patients in the “not tested” group versus patients in the “tested-positive-discordant” group (aHR=1.50, 95%CI=1.46, 1.55). Conclusions: Real world data suggest that receipt of treatment that is guided by biomarker testing results is associated with improved survival outcomes in this patient population.