OBJECTIVES: Real-world end points from diverse healthcare data organizations with claims and/or electronic health records data (EHR) have been compared and showed consistency in patients with aNSCLC treated with immunotherapy. We seek to assess how the real-world outcomes identified in the GuardantINFORM™ clinical-genomics database relate to those findings.

METHODS: Patients were identified from GuardantINFORM, which links cell-free circulating tumor DNA (cfDNA) results to de-identified claims data. Adult aNSCLC patients in the US, who received PD-L1 treatments post-diagnosis from January 2014 to October 2017 were included, with data up to April 30 2018. Patients were excluded if they had less than two claims in the six months prior to first lung cancer diagnosis in claims and start of first line treatment was more than 90 days after first diagnosis. To account for left truncation of clinical-genomic databases, only patients who received a cfDNA test prior to PD-L1 treatment initiation were included. Median real-world time to treatment discontinuation (rwTTD), time to next treatment (rwTTNT) and real-world overall survival (rwOS) were calculated using Kaplan Meier method, starting from PD-L1 treatment initiation.

RESULTS: 789 aNSCLC patients were identified. Mean age at treatment was 65 years, 48% were male and 81% were white. 17% and 66% of patients were on first line and second line PD-L1 therapy respectively. Median rwTTD was 3.17 months (95% CI 2.80-3.60), median rwTTNT was 7.97 months (95% CI 6.00-9.07), and median rwOS was 12.20 months (95% CI 10.67-13.80). These estimates, except rwTTNT, were comparable to previously reported values in other databases (rwTTD: 3.2 months to 4.7 months, rwTTNT: 11.6 months to 14.0 months, rwOS: 8.6 months to 13.5 months).

CONCLUSIONS: We demonstrated that the real-world end points from GuardantINFORM database is generally consistent with other real-world databases in this aNSCLC patient cohort.