ADDITIONAL SUPPLEMENTARY INFORMATION
Title of article: Cost-effectiveness analysis of regorafenib for gastrointestinal stromal tumour (GIST) in Germany
Authors:David Tamoschus1, Katja Draexler1, Jane Chang2, Christopher Ngai2, Matthew Madin-Warburton3, Ashley Pitcher3
1Bayer HealthCare Germany, Leverkusen, Germany;
2Bayer HealthCare Pharmaceuticals, Whippany, NJ, USA;
3QuintilesIMS, London, UK
Corresponding author: Ashley Pitcher, 210 Pentonville Road, London N1 9JY United Kingdom. Email: . Tel: +44 (0)2030 755877.
Testing of proportional hazards assumption
The proportional hazards assumption was tested for the treatment effect of regorafenib on overall survival and progression free survival in the GRID trial.Due to a lack of available data it was not possible to conduct a similar testsfor imatinib in the RIGHT trial, however the study reported that the proportional hazards assumption was checked by the authors.
The tests of proportional hazards found that there was no evidence that the proportional hazards assumption had been violated in either case. Test results and a cumulative hazard plot can be found below.
Parameter / Rho / Chi-squared / Probability > Chi-squaredOverall survival: Regorafenib vs. BSC / 0.152 / 1.04 / 0.301
Progression-free survival: Regorafenib vs. BSC / -0.036 / 0.19 / 0.661
Abbreviations: BSC, best supportive care
Log cumulative hazard plots
Supplementary Fig.1: Log cumulative hazard plot(Overall Survival)
Supplementary Fig.1: Log cumulative hazard plot(Progression-free survival)
Summary of parametric model fits to regorafenib in GRID
Parametric model fit statistics and a graphic summary of model fits to regorafenib can be found below.
Note that the Gompertz distribution did not provide clinically plausible results for overall survival and was therefore excluded from sensitivity analyses.
Parametric model / BSC / Regorafenib / Total AICExponential / 158.3 / 329.2 / 487.6
Weibull / 167.4 / 328.4 / 495.8
Gompertz / 168.3 / 330.2 / 498.5
AICs for parametric model fits in GRID (progression-free survival)
Parametric model / BSC / Regorafenib / Total AICExponential / 111.5 / 186.8 / 298.2
Weibull / 113.4 / 186.3 / 299.8
Gompertz / 113.4 / 186.9 / 300.3
AICs for parametric model fits in GRID (overall survival)
Supplementary Fig.3:Parametric models for progression-free survival (regorafenib)
Supplementary Fig.4:Parametric models for overall survival(regorafenib)
Table of parameters included in one-way sensitivity analysis
Parameter / Base case / Low value / High value / Scenario valueDiscount rate costs / 3.5% / 0% / 6% / -
Discount rate utilities / 3.5% / 0% / 6% / -
Treatment costs following disease progression (Imatinib patients escalate to 600mg) / No treatment costs following progression / - / - / Included
Treatment costs following disease progression (Imatinib patients escalate to 800mg) / No treatment costs following progression / - / - / Included
Utility of progression-free health state / 0.77 / 0.72 / 0.82 / -
Utility of progressed health state / 0.65 / 0.57 / 0.72 / -
OS extrapolation parametric model / Exponential / - / - / Weibull
PFS extrapolation parametric model / Exponential / - / - / Weibull or Gompertz
OS extrapolation / OS extrapolation applied from end of trial data / - / - / OS extrapolation using parametric model applied from last observed event (death) or OS based on parametric model for all time points
PFS extrapolation / PFS extrapolation applied from end of trial data / - / - / PFS based on parametric model for all time points*
PFS HR (imatinib versus regorafenib) / 1.72 / 0.90 / 3.28 / -
OS HR (imatinib versus regorafenib) / 1.30 / 0.56 / 2.98 / -
Discontinuation / Discontinuation applied as observed in GRID trial / - / - / No discontinuation applied
Half cycle correction / Half cycle correction applied / - / - / No half cycle correction applied
Time horizon in years / Lifetime time horizon (40 years) applied / - / - / 5 year time horizon applied
OS survival data cut-off / 12 month data from GRID trial applied / - / - / 36 month follow-up data applied
*A scenario of PFS extrapolation from last observed event was not applied as the last point in the trial data was an event
Abbreviations: HR, hazard ratio; OS, overall survival; PFS, progression-free survival
Table of parameters included in probabilistic sensitivity analysis
Parameter / DistributionHealth-related utility of pre-progression state / Beta
Health-related utility of post-progression state / Beta
Kaplan-Meier OS (each point) / Beta
Kaplan-Meier PFS (each point) / Beta
Exponential OS extrapolation model parameter / Normal
Exponential PFS extrapolation model parameter / Normal
PFS HR (regorafenib versus imatinib) / Lognormal
OS HR (regorafenib versus imatinib) / Lognormal
Abbreviations: HR, hazard ratio; OS, overall survival; PFS, progression-free survival; QALY, quality-adjusted life year.
Supplementary Fig.5: One-way sensitivity analysis results for incremental QALYs
Abbreviations: HR, hazard ratio; OS, overall survival; PFS, progression-free survival; QALY, quality-adjusted life year.
Supplementary Fig.6: One-way sensitivity analysis results for incremental costs
Abbreviations: HR, hazard ratio; PFS, progression-free survival.