Real-World Response Endpoints in Patients With mNSCLC Treated With Chemotherapy Across Real-World Datasets

Authors Brittany Avin McKelvey, Elizabeth Garrett-Mayer, Andrew J. Belli, Thomas D. Brown, Jessica Dow, Janet L. Espirito, Paul Kluetz, Xinran Ma, Andrea McCracken, Pallavi Shruti Mishra-Kalyani, Yanina Natanzon, Danielle Potter, Donna Rivera, Hillary Stires, Mark Stewart, Jeff Allen

Background:Response Evaluation Criteria in Solid Tumors (RECIST) based response rate (RR) is used for efficacy evaluation in clinical trials and relies on imaging data collected at specified timepoints for uniform assessment. In routine clinical practice, the method and timing of response assessment can vary, and imaging data from electronic health records (EHR) and other real world (rw) sources may not be available, making RECIST-based assessment of rw-response rate (rwRR) using rw data (RWD) challenging. Friends of Cancer Research formed a multi-stakeholder partnership to assess available data attributes to measure response across RWD sources to inform development of a consistent method for measurement.

Methods:The study included seven EHR data partners who identified and analyzed a cohort of 1,380 patients (pts) with metastatic non-small cell lung cancer (mNSCLC) treated with first-line platinum doublet chemotherapy, following a common protocol and statistical analysis plan. The availability and frequency of data components to assess response including raw images, radiology imaging reports, and clinician response assessments from provider notes were assessed. Response endpoints measured included rwRR, rw-duration of response (rwDOR), and the association of rwR with rw-overall survival (rwOS), rw-time to treatment discontinuation (rwTTD), and rw-time to next treatment (rwTTNT).

Results:The availability of data components varied across RWD sources (Table). Images were not widely accessible, thus response was analyzed using clinician response assessments (median proportion of pts evaluable, 77.5%). Of these assessments, the majority relied on imaging interpretation. The median rwRR was 46% with a median rwDOR of 119 days. The table provides median rwTTD, rwTTNT, and rwOS across data sources.

Conclusions:The rwRR among pts with mNSCLC calculated using the clinician assessment was relatively consistent across all RWD sources, with consistent trends in time to event endpoints. While variability in the availability of data components to assess response was observed, the demonstrated feasibility of response endpoints based on clinician assessment suggests further exploration may inform drug effectiveness evaluation with RWD.

Group* Pts Evaluable for rwR (Pts Ev) by Images Pts Ev by Radiology Reports Pts Ev by Clinician Response Assessment rwRR Median rwDOR, days
(95% CI)
Median rwTTD, days: Responders /Non-Responders (R/NR) Median rwTTNT, days: R/NR Median rwOS, days: R/NR
A 3.5% 73% 79.5% 42% 115 (86, 199) 142/69 200/100 375/245
B 0.5% 55% 80.5% 53% 133 (108, 182) 128/84 209/98 464/314
C 40.5% 77% 77.5% 46% 146 (102, 210) 147/63 234/93 832/213
D 0% 0% 74% 40% 100 (74,-) 105/70 140/115 614/414
E 79.5% 79.5% 76% 38% 119 (98, 231) 132/48 235/93 474/184
F 0% 66.5% 69% 52% 182 (147, 287) 99/43 219/109 436/353
G 0% 85.5% 88.3% 49% 105 (7, 672) 112/21 198/61 392/86

*n=200 pts except G: n=180.