OBJECTIVES: To determine the clinical evidence base, analyses, and criticisms of advanced oncology HTA submissions.

METHODS: The NICE database was searched for submissions of advanced or metastatic oncology receiving second-line treatments in the previous 12 months (to 18^th April 2023). We extracted the characteristics of the clinical evidence and types of statistical analyses. A thematic analysis was performed on the critiques. Data were extracted by one reviewer and checked by a second reviewer. The identified NICE submissions were matched to those from Canada (CADTH) and Germany (G-BA/IQWiG), to assess similarities and differences.

RESULTS: 21 submissions were identified; 15 were recommended, five were terminated and one was not recommended.

Of the 15 successful submissions and their associated systematic reviews (SRs), 10 SRs included one trial (population size = 108 to 700), four included two trials, and one included three trials. Three SRs found Phase 1/2 trials, and the remainder were Phase 2 or 3. Five SRs identified single arm studies, and 10 identified comparative data.

Two SRs included real-world evidence (RWE) in addition to trial evidence. Six submissions used RWE not identified by the SR. RWE was used to provide comparator evidence or validate clinical evidence.

Seven submissions included expert opinion to validate the clinical evidence, treatment pathways/choices, comparator choice, or prognostic variables.

Eight matched-adjusted indirect comparisons (MAICs) were identified, three Bucher indirect treatment comparisons (ITCs), three propensity-score matching ITCs, and two network meta-analyses. Additional statistical analyses were requested by the committee in six submissions.

Preliminary analysis identified the following themes reduced data confidence: low sample size, heterogeneous populations, and lack of evidence for UK populations. Themes arising from CADTH and G-BA/IQWiG will be presented.

CONCLUSIONS: Evidence for advanced oncology indications is limited and is frequently supplemented by RWE. RWE should be relevant and produced by transparent methods.