TREATMENT OF POST-MENOPAUSAL ADVANCED BREAST CANCER- NETWORK META-ANALYSIS USING SECOND-ORDER FRACTIONAL POLYNOMIALS MODELS
Author(s)
Harvey R
DRG Abacus, Manchester, UK
OBJECTIVES: A multi-dimensional approach to evidence synthesis has previously been explored as an alternative to analysis of hazard ratios (HR) frequently used for time-to-event data. The objective of this study was to explore and compare results using a more complex approach to network meta-analysis (NMA) with those using conventional methods. The methods were applied to an evidence base previously identified to evaluate comparative efficacy for the treatment of advanced breast cancer in postmenopausal women following failure of prior endocrine therapy. METHODS: Data were identified from a published systematic literature review previously conducted in this population; these data were synthesised within a Bayesian NMA utilising fractional polynomial models to estimate relative efficacy between different interventions for overall survival. Kaplan–Meier curves reported by all studies were digitised to recreate virtual individual patient data. Extrapolated long-term survival profiles, expected (mean) survival and ranking probabilities were also estimated for each intervention. RESULTS: Seven trials were identified, which assesed seven comparators of interest. Fulvestrant 500 mg was found to be the most effective intervention; mean survival was over 3 years and there was around a 40% chance of fulvestrant 500 mg being ranked as the best treatment over a 10-year time horizon. Findings and trends were consistent with previously published NMAs utilising conventional methodology, synthesising HRs. CONCLUSIONS: A comprehensive approach to evidence synthesis may yield more informative, robust and reliable estimates of comparative efficacy than traditional NMA methods analysing trial-level HRs. These efficacy estimates may subsequently be incorporated into cost-effectiveness models without the assumption of proportional hazards between treatment arms that underpins the use of HRs, which have been shown to be a limited measure of treatment effect. An alternative approach to synthesis, utilising fractional polynomials models, may lead to more informed decision-making, which is critical when there are budget constraints associated with reimbursement of new interventions.
Conference/Value in Health Info
2018-11, ISPOR Europe 2018, Barcelona, Spain
Value in Health, Vol. 21, S3 (October 2018)
Code
PCN18
Topic
Clinical Outcomes
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
Oncology