Evaluation of change in RECIST tumor size and survival in patients with rare cancers treated with checkpoint inhibitor therapy (SWOG S1609)

2504 Background: RECIST has been used for over twenty years to evaluate changes in tumor burden. At each scan, the sum of maximum diameters of “target” lesions are measured. Changes in the sum are categorized into an “objective status” at each scan, including complete and partial response and stable and progressive disease. Prior research among patients treated with chemotherapy did not find a threshold effect corresponding to the RECIST categories; instead a linear association was observed. Herein we use a unique clinical trial resource, a basket trial of patients with rare cancer treated with combination anti-CTLA-4 and anti-PD-1 therapy, to evaluate the association between quantitative change in RECIST tumor burden and survival among patients treated with checkpoint inhibitor therapy. Methods: The study population was 796 patients enrolled on the S1609 DART (Dual Anti-CTLA-4 & Anti-PD-1 blockade in Rare Tumors) trial, a basket immunotherapy trial evaluating ipilimumab plus nivolumab across solid rare tumor subtypes. The analysis population was 720 patients; patients not eligible (n=50; mostly due to non-rare diagnosis), THOSE who did not receive protocol therapy (n=22), OR WHOSE DISEASE WAS NOT evaluated by RECIST (n=4; gestational trophoblastic disease evaluated by -hCG) were excluded. All patients had RECIST-measurable disease at study entry. Associations between overall and progression-free survival (OS and PFS) and change in tumor burden between baseline and the first scan were evaluated using log-rank tests and Cox regression models stratified by basket. Landmark analyses (at day 65) controlled for lead-time bias. Linear regression models and R 2 were estimated for median and 2-year estimates across tumor change intervals. Results: The median age of the cohort was 61, 52% werethe distribution of percent change in tumor burden at the first scan. The linear correlation between median and 6-month PFS and change in tumor burden had R 2 = 0.92 and 0.74.; median and 1-year OS had R 2 = 0.79 and 0.79. Martingale residuals for both PFS and OS showed linear trends up to tumor increase of +50% with no thresholds at definitions of partial response or stable disease. Conclusions: This is the first evaluation of the association between numeric changes in RECIST tumor burden among patients receiving checkpoint therapy. Our results suggest that similar to results among patients who received chemotherapy there is a linear correlation with PFS and OS. As a consequence, the thresholds for RECIST categories do not correspond to thresholds in patient outcomes. [Table: see text]