Discrete Event Simulation and Treatment Sequencing Cost-Effectiveness Model in Second-Line Highly Active Relapse Remitting Multiple Sclerosis for a NICE Multiple Technology Appraisal
Author(s)
Ayman Sadek, MSc, MSc1, Catalina Lopez Manzano, MSc1, Chris Cooper, MSc, PhD1, Eve Tomlinson, MSc1, Claire M Rice, PhD, FRCP2, Emma C. Tallantyre, PhD, FRCP3, Javier Sanchez Alvarez, PhD4, Jeff W. Rodgers, PhD5, Howard Thom, MSc, PhD1.
1Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, United Kingdom, 2Department of Neurology, Southmead Hospital, North Bristol NHS Trust., Clinical Neurosciences, Translation Health Sciences, Bristol Medical School, University of Bristol., Bristol, United Kingdom, 3Division of Psychological Medicine and Clinical Neurosciences, School of Medicine, Cardiff University, Cardiff, United Kingdom, 4Evidera, A Coruna, Spain, 5Population Data Science, Swansea University Medical School, Swansea, United Kingdom.
1Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, United Kingdom, 2Department of Neurology, Southmead Hospital, North Bristol NHS Trust., Clinical Neurosciences, Translation Health Sciences, Bristol Medical School, University of Bristol., Bristol, United Kingdom, 3Division of Psychological Medicine and Clinical Neurosciences, School of Medicine, Cardiff University, Cardiff, United Kingdom, 4Evidera, A Coruna, Spain, 5Population Data Science, Swansea University Medical School, Swansea, United Kingdom.
OBJECTIVES: We undertook a Multiple Technology Assessment (MTA) for the UK National Institute for Health and Care Excellence (NICE) assessing the cost-effectiveness of natalizumab as second line treatment for Highly Active Relapsing Remitting Multiple Sclerosis (HARRMS) in comparison to cladribine, ofatumumab, ocrelizumab and ublituximab. We addressed the shortcomings of prior Excel-based cohort Markov models by incorporating more flexible natural history modelling with up-to-date data, treatment switching, and severity-dependent mortality data.
METHODS: We developed an individual patient level discrete event simulation (DES) with up to three lines of treatment in R. Events included relapse, adverse events, treatment discontinuation, disability progression, and onset of Secondary Progressive Multiple Sclerosis (SPMS). Natural history was informed by new analyses of the UK Multiple Sclerosis (MS) Registry. Relative treatment effects were informed by Bayesian Network Meta-Analysis (NMA). We modeled mortality stratified by severity using recent MS studies. Model code was independently validated, and outputs were validated against clinical opinion and long-term data. Results were summarized using incremental net monetary benefit (INMB) at £20,000/QALY with 95% Bayesian Credible Intervals (CrI). Decision uncertainty was assessed by Value of Information (VOI) analysis.
RESULTS: Model predictions were better aligned with external data and clinical opinion than previous Markov models, and the NICE committee found the model suitable for decision making. The NMB of natalizumab biosimilar was higher than natalizumab originator, INMB £7,510 (95% CrI: £-5,963; £23,291). Cladribine had highest INMB £48,302 (£-379; £115,102). VOI indicated relative treatment effects to be the greatest source of decision uncertainty.
CONCLUSIONS: This was the first NICE MTA to use a DES model built in R and it was found to be robust for decision making. In HARRMS, natalizumab biosimilar was found to have greater INMB than originator but cladribine had greatest INMB. The 95% CrI were wide and decision uncertainty, driven by treatment effects, was high.
METHODS: We developed an individual patient level discrete event simulation (DES) with up to three lines of treatment in R. Events included relapse, adverse events, treatment discontinuation, disability progression, and onset of Secondary Progressive Multiple Sclerosis (SPMS). Natural history was informed by new analyses of the UK Multiple Sclerosis (MS) Registry. Relative treatment effects were informed by Bayesian Network Meta-Analysis (NMA). We modeled mortality stratified by severity using recent MS studies. Model code was independently validated, and outputs were validated against clinical opinion and long-term data. Results were summarized using incremental net monetary benefit (INMB) at £20,000/QALY with 95% Bayesian Credible Intervals (CrI). Decision uncertainty was assessed by Value of Information (VOI) analysis.
RESULTS: Model predictions were better aligned with external data and clinical opinion than previous Markov models, and the NICE committee found the model suitable for decision making. The NMB of natalizumab biosimilar was higher than natalizumab originator, INMB £7,510 (95% CrI: £-5,963; £23,291). Cladribine had highest INMB £48,302 (£-379; £115,102). VOI indicated relative treatment effects to be the greatest source of decision uncertainty.
CONCLUSIONS: This was the first NICE MTA to use a DES model built in R and it was found to be robust for decision making. In HARRMS, natalizumab biosimilar was found to have greater INMB than originator but cladribine had greatest INMB. The 95% CrI were wide and decision uncertainty, driven by treatment effects, was high.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EE336
Topic
Economic Evaluation, Study Approaches
Topic Subcategory
Value of Information
Disease
Biologics & Biosimilars, Neurological Disorders, No Additional Disease & Conditions/Specialized Treatment Areas