OBJECTIVES: Median overall survival (OS) data may be immature, or non-significant, when new oncology-therapies undergo health technology assessment (HTA); indeed, survival data may be limited to progression free survival (PFS). This study was designed to identify drivers of positive HTA decisions in submissions with PFS data that lack median OS.

METHODS: NICE technology appraisal guidance was searched to identify positive recommendations from submissions with PFS but no median OS data. Advice from other HTA bodies (IQWiG, G-BA, HAS, CADTH and PBAC) was then assessed for technologies achieving positive NICE recommendation, to identify discrepancies between HTAs, and to highlight drivers of positive decisions.

RESULTS: A total of 23 positive recommendations were identified from NICE January 2013 - March 2018, of which 20 were with restriction (13 patient access schemes, 3 commercial access agreements and 4 managed access agreements). Corresponding HTA submissions for each drug/indication identified differences in decisions: IQWiG (4 proven added benefit, 5 minor added benefit, 8 no added benefit) G-BA (5 proven added benefit, 6 minor added benefit, 5 no added benefit), HAS (14 recommended, 1 recommended with restriction), CADTH (1 recommended, 12 recommended with restriction, 3 not recommended) and PBAC (7 recommended, 4 recommended with restriction, 5 not recommended). PFS data, alongside advice from clinicians and patient groups on likely patient-relevant benefits were considered by NICE, HAS, CADTH and PBAC, resulting in positive decisions. IQWiG and G-BA did not consider PFS data; instead, drivers of positive decisions were symptom relief, improved functioning/QoL and reduction in adverse events compared to standard of care.

CONCLUSIONS: PFS data and clinician/patient views on patient benefits are drivers of positive HTA decisions in the absence of median OS data for non-German HTAs. However, in Germany, data demonstrating improvements in morbidity (symptoms) and QoL (functioning) can result in favourable benefit assessment without median OS.