OBJECTIVES

: Several disease modifying treatments (DMTs) are licensed and reimbursed for treating relapsing-remitting multiple sclerosis (RRMS). Cost-effectiveness analyses supporting HTA submissions generally assume that patients discontinuing treatment due to tolerability or disease progression receive best supportive care, assumed to have no effect on RRMS natural history. Contrary, in clinical practice patients receive DMTs in sequence. Results of said cost-effectiveness analyses do not, therefore, sufficiently inform policy regarding optimal use of new drugs. This analysis evaluates the impact on cost-effectiveness of modelling subsequent treatment in RRMS from a UK NHS perspective.

METHODS

: A cost-effectiveness Markov model was developed. Natalizumab and fingolimod were considered as licensed treatments for patients with RRMS. Analyses were run varying the cost and treatment effect of a hypothetical subsequent treatment compared with a scenario excluding subsequent treatment. Treatment efficacy and discontinuation rates for primary treatments were sourced from a network meta-analysis conducted by the Institute for Clinical and Economic Review. The hazard ratio on disability progression versus placebo was 0.56 for natalizumab and 0.68 for fingolimod, and annual discontinuation rates were 4.9% and 8.4%, respectively. Costs and utility data were sourced from published literature. A discount rate of 3.5% pa was applied.

RESULTS

: Without modelling subsequent treatment cost or effect, the ICER for natalizumab versus fingolimod was £29,500. Including subsequent treatment, the ICER ranged from £855 (most costly and least effective subsequent treatment) to £33,058 (least costly and most effective subsequent treatment).

CONCLUSIONS

: Subsequent treatment impacts on cost-effectiveness of DMTs for RRMS. More cost-effective subsequent treatment favours primary treatments with higher discontinuation rates. The resulting ICER may therefore lead to different reimbursement decisions depending on whether subsequent treatment is modelled and based on the assumptions made about its efficacy and cost. Alternative modelling approaches need to be explored and more clinical data are required to define optimal treatment sequences.