OBJECTIVES: Mean survival, estimated by extrapolation of incomplete data, is commonly used by health technology assessment agencies to assess the value of new oncology treatments. A prolonged tail in the survival distribution, attributable to a subset of patients with durable response to treatment, will tend to improve estimates of mean survival. However, if the proportion of durable responders is small or moderate, the potential for increased mean survival may only become apparent with longer-term follow-up. This research seeks to identify the impact of longer follow-up on estimates of mean survival from clinical trials of immuno-oncology (IO) therapies.

METHODS: A targeted literature search was conducted to identify published long-term follow-up from clinical trials of immune checkpoint inhibitors (ICI) for the treatment of melanoma, non-small-cell lung cancer (NSCLC) and renal cell carcinoma (RCC). First line indications were excluded to reduce the potential impact of subsequent therapies. Earlier published results were identified and Kaplan-Meier curves for the short-term and long-term follow-up were digitized and converted to pseudo-IPD using an established algorithm. Six standard parametric and three flexible parametric models were used to extrapolate and estimate mean survival for each set of short-term and long-term follow-up data.

RESULTS: Ten comparisons of short-term and long-term follow-up were identified for ICI therapy in melanoma, NSCLC and RCC. Estimates of mean survival increased with longer follow-up for all 9 models in 6 comparisons, and for 8 out of 9 models in a further 2 comparisons. The variance of mean estimates was reduced for 6 comparisons. The average increase in mean survival across models for each comparison ranged from 4% to 74%.

CONCLUSIONS: Estimates of mean survival systematically increased and variability across extrapolation models reduced with longer follow-up in clinical trials of IO treatments. Developers of innovative IO therapies should plan for and publish long-term follow-up from clinical trials wherever feasible.