The toxicity swimmer plot: A novel approach for evaluating patient-level toxicity following immune checkpoint inhibitor (ICI) therapy on the Checkmate 214 trial.

Abstract

11014 Background: Treatment with ICIs creates the possibility of prolonged disease control or remission but also creates the possibility of prolonged toxicity even after treatment is discontinued. Treatment-free survival (TFS) with and without toxicity has previously been described as an integral piece of a partitioned survival analysis in advanced renal cell carcinoma (aRCC). Prior analyses have reported further partition of TFS as mean duration of TFS with and without treatment-related adverse events (TRAEs), but this metric fails to provide more granular information regarding individual patients experiencing toxicity during TFS. We present a novel extended swimmer plot that characterizes patient-level toxicity data within TFS using the example of the Checkmate 214 trial to complement analyses of the population average experience. Methods: Data were analyzed from the Checkmate 214 trial which randomized 1096 patients with aRCC to receive either first-line (1L) nivolumab plus ipilimumab (n = 550) or sunitinib (n = 546). The minimum follow-up from protocol therapy initiation was 42 months. Patients’ overall survival was partitioned into 3 survival states: time on 1L protocol therapy, TFS, and survival after subsequent therapy initiation. With a re-defined time origin relative to cessation of 1L protocol therapy, we generated swimmer plots highlighting the TFS period and documenting each patient’s experience with and without grade 2+ and grade 3+ TRAEs during TFS. The swimmer plot was sorted by duration of TFS and duration of TFS with grade 2+ TRAEs in order to illustrate patterns of individuals’ experiences. Results: The 42-month mean TFS was 7.8 months (with grade 2+ TRAEs: 3.2 months) for patients treated with nivolumab plus ipilimumab and 3.3 months (1.6 months) for patients treated with sunitinib. In the respective treatment groups, TFS with grade 2+ TRAEs was experienced by 32% and 25% of patients and comprised of 13% and 9% of patients for whom it was the entire TFS period. These percentages were consistent regardless of TFS duration. TFS with grade 3+ TRAEs was experienced by 15% and 3% of patients, respectively. For most patients having TFS with TRAEs, the TRAEs persisted from the protocol therapy period, but 7% and 1% of patients in the respective treatment groups had new-onset grade 3+ TRAE during TFS. A swimmer plot will be presented as a visual presentation of this study’s results. Conclusions: We describe a novel extension of swimmer plots for presenting patient-level information and illustrating patterns in toxicity duration relative to TFS by incorporating individual toxicity data into a swimmer plot of partitioned survival. This approach allows for a more comprehensive understanding of toxicity trends and risk of prolonged toxicity after treatment is discontinued.