D Lee1, J Porter2, N Hertel3, A J Hatswell2, A Briggs4. 1. BresMed, North Church House, 84 Queen Street, Sheffield, S1 2DW, UK. dlee@bresmed.co.uk. 2. BresMed, North Church House, 84 Queen Street, Sheffield, S1 2DW, UK. 3. Bristol-Myers Squibb Pharmaceuticals Ltd, Uxbridge, Middlesex, UK. 4. Health Economics and Health Technology Assessment, University of Glasgow, Glasgow, Scotland, UK.
Abstract
BACKGROUND: In the absence of head-to-head data, a common method for modelling comparative survival for cost-effectiveness analysis is estimating hazard ratios from trial publications. This assumes that the hazards of mortality are proportional between treatments and that outcomes are not polluted by subsequent therapy use. Newer techniques that compare treatments where the proportional hazards assumption is violated and adjust for use of subsequent therapies often require patient-level data, which are rarely available for all treatments. OBJECTIVE: The objective of this study was to provide a comparison of overall survival data for ipilimumab, vemurafenib and dacarbazine using data from three trials lacking a common comparator arm and confounded by the use of subsequent treatment. METHODS: We compared three estimated overall survival curves for vemurafenib and the difference compared to ipilimumab and dacarbazine. We performed a naïve comparison and adjusted it for heterogeneity between the ipilimumab and vemurafenib trials, including differences in prognostic characteristics and subsequent therapy using a published hazard function for the impact of prognostic characteristics in melanoma and trial data on the impact of second-line use of ipilimumab. RESULTS: The mean incremental life-years gained for patients receiving ipilimumab compared with vemurafenib were 0.34 (95 % confidence interval [CI] -0.24 to 0.84) using the naïve comparison and 0.51 (95 % CI -0.08 to 0.99) using the covariate-adjusted survival curve. CONCLUSIONS: The analyses estimated the comparative efficacy of ipilimumab and vemurafenib in the absence of head-to-head patient-level data for all trials and proportional hazards in overall survival.
BACKGROUND: In the absence of head-to-head data, a common method for modelling comparative survival for cost-effectiveness analysis is estimating hazard ratios from trial publications. This assumes that the hazards of mortality are proportional between treatments and that outcomes are not polluted by subsequent therapy use. Newer techniques that compare treatments where the proportional hazards assumption is violated and adjust for use of subsequent therapies often require patient-level data, which are rarely available for all treatments. OBJECTIVE: The objective of this study was to provide a comparison of overall survival data for ipilimumab, vemurafenib and dacarbazine using data from three trials lacking a common comparator arm and confounded by the use of subsequent treatment. METHODS: We compared three estimated overall survival curves for vemurafenib and the difference compared to ipilimumab and dacarbazine. We performed a naïve comparison and adjusted it for heterogeneity between the ipilimumab and vemurafenib trials, including differences in prognostic characteristics and subsequent therapy using a published hazard function for the impact of prognostic characteristics in melanoma and trial data on the impact of second-line use of ipilimumab. RESULTS: The mean incremental life-years gained for patients receiving ipilimumab compared with vemurafenib were 0.34 (95 % confidence interval [CI] -0.24 to 0.84) using the naïve comparison and 0.51 (95 % CI -0.08 to 0.99) using the covariate-adjusted survival curve. CONCLUSIONS: The analyses estimated the comparative efficacy of ipilimumab and vemurafenib in the absence of head-to-head patient-level data for all trials and proportional hazards in overall survival.
Authors: José M Estrela; Rosario Salvador; Patricia Marchio; Soraya L Valles; Rafael López-Blanch; Pilar Rivera; María Benlloch; Javier Alcácer; Carlos L Pérez; José A Pellicer; Elena Obrador Journal: Am J Cancer Res Date: 2019-12-01 Impact factor: 6.166
Authors: Marta Polkowska; Paweł Ekk-Cierniakowski; Edyta Czepielewska; Wojciech Wysoczański; Wojciech Matusewicz; Małgorzata Kozłowska-Wojciechowska Journal: J Cancer Res Clin Oncol Date: 2017-06-12 Impact factor: 4.553