Presentation (CTI)

OBJECTIVES: To assess differences in patient and clinical characteristics, treatment patterns, and clinical outcomes across electronic health record-derived databases with and without genomic linkage.
METHODS: The US-based, deidentified Flatiron Health-Foundation Medicine non-small cell lung cancer (NSCLC) Clinico-Genomic Database (CGDB) and Flatiron Health Research Database (FHRD) were used to compare demographics, clinical characteristics, treatment patterns, and overall survival (OS) among patients with NSCLC, stratified by stage. Subgroup analyses were performed for patients with actionable genomic alterations (AGAs) and non-AGAs.
RESULTS: CGDB and FHRD included 2,036 and 8,108 early-stage and 12,737 and 22,248 advanced NSCLC patients, respectively. CGDB and FHRD cohorts showed similar age distributions (≥65 yrs: 76.5% vs. 78.0% for early-stage; 68.1% vs. 65.5% for advanced) and gender proportions (female: 54.6% vs. 54.0% for early-stage; 50.4% vs. 49.2% for advanced). The proportion of Non-Hispanic White patients between CGDB and FHRD differed in early-stage disease (50.8% vs. 57.1%) but was similar in advanced disease (54.7% vs. 53.5%). Among advanced NSCLC, CGDB and FHRD had similar proportion of patients receiving systemic treatment (81.0%). Carboplatin-pemetrexed-pembrolizumab was the most common first-line (1L) regimen overall and among non-AGA patients, while osimertinib was the most common 1L regimen for AGAs patients across databases. Among patients with early-stage NSCLC, the median OS from diagnosis was 53.6 months in CGDB after adjusted for left truncation and 72.0 months in EHRD. Among those with advanced NSCLC, the median OS from diagnosis, 1L initiation, 2L initiation, and 3L initiation was 14.4, 14.6, 11.2, and 8.6 months, respectively, in CGDB, and 19.9, 17.4, 12.0, and 9.4 months, respectively, in FHRD.
CONCLUSIONS: CGDB and FHRD cohorts showed generally similar demographic profiles, treatment patterns, and OS, particularly in advanced NSCLC population, suggesting that CGDB and FHRD provide complementary and comparable real-world evidence for NSCLC research. However, potential overlap in patient populations between two cohorts cannot be ruled out.