OBJECTIVES: Recently, quality-adjusted time without symptoms or toxicity (Q-TWiST) analysis has been considered a valuable decision-making tool to assess risk-benefit trade-offs and treatment impacts. The objective of this analysis is to focus on Q-TWiST methodology and the challenges in performing the analysis in oncology trials.

METHODS: Toxicity (TOX), progression-free survival (PFS) and overall survival (OS) data were simulated by assuming an adverse event (AE) duration that follows a log-normal distribution and time-to-progression and time-to-death events that come from the family of Weibull distributions. Parameters were selected such that simulated survival times reflected those observed in immune-oncology trials. The three health states, TOX, TWiST and relapse (REL), were calculated using the area under the Kaplan-Meier curves. Two censoring rules for TOX (no censoring vs PFS censoring) were considered and the impact studied. Q-TWiST was calculated using utility values of 0.5 for TOX, 1 for TWiST and 0.5 for REL. The restricted mean survival time (RMST) was estimated for all health states for varying proportions (80%, 60%, 40%, 20%) of subjects with an AE and progression events. A total of 36 scenarios were explored and the trends in Q-TWiST were studied using 1000 simulations for each scenario with 200 patients per sample.

RESULTS: Considering the two censoring rules, no-censoring resulted in a smaller value for TOX and a larger Q-TWiST compared with PFS censoring. As the proportion of subjects with AEs increased, the variation in Q-TWiST between the two censoring rules increased. However, as PFS events increased, the variation reduced.

CONCLUSIONS: Q-TWiST analysis is a valuable tool for health authorities to support cancer treatment evaluation. The censoring rule for TOX could have an impact in the resultant Q-TWiST. Sufficient follow-up and data maturity are crucial in Q-TWiST and provide greater stability of estimates. Further research into generalizable goodness of fit is warranted.