Decision tree combined with a Markov model

1. bDMARD inadvisable: Biologic-naive patients who have had treatment failure with NSAIDs (due to inadequate response or intolerance) and for whom treatment with a biologic (TNF or IL-17 inhibitors) is inadvisable.

2. bDMARD-inadequate response: Patients who have had treatment failures with NSAIDs (due to inadequate response or intolerance) as well as at least 1 biologic therapy.

Etanercept

Golimumab

Infliximab

Secukinumab

Conventional therapy (corticosteroids, NSAIDs, or csDMARDs, such as sulfasalazine, methotrexate, leflunomide).

Sponsor-submitted NMA

• The comparative effect of upadacitinib on key clinical outcomes is uncertain due to a lack of direct comparative evidence to relevant comparators and the high degree of uncertainty in the sponsor’s network meta-analysis. The CADTH clinical review concluded there were no differences in efficacy between therapies used to treat AS considered in the sponsor’s NMA for both the bDMARD-inadvisable and -inadequate response subpopulations. No comparative safety or discontinuation information was available.
• The timing for the initial treatment effect for BASDAI and BASFI, as well as the baseline scores upon which the treatment effects were applied, were incorrectly implemented. This affected the total QALYs estimated for all treatments included in the analyses.
• The sponsor assumed that conventional therapy would have no effect on treatment response status or disease progression. This was inconsistent with the submitted NMA and clinical expectations.
• Although evidence was available for the 150 mg and 300 mg doses of secukinumab, only the data for the 150 mg dose was used in the sponsor’s analysis. However, ███ dose split was assumed when calculating costs, thus overestimating costs associated with the 150 mg dose and omitting the efficacy information specific to the 300 mg dose.
• Although certolizumab pegol satisfied the criterion for relevant comparators, the absence of subpopulation-specific trial data led to its exclusion from the sponsor’s base case. Further, several relevant comparators (i.e., infliximab, golimumab, etanercept, adalimumab) were omitted from the bDMARD-inadequate response subpopulation due to a lack of evidence.

• CADTH conducted a reanalysis incorporating changes to the sponsor’s economic submission to address key limitations. These changes included more appropriate assumptions regarding the calculation and timing of treatment effects, use of overall instead of response-stratified baseline BASDAI and BASFI, use of consistent dosing for secukinumab to calculate costs, assumption of a treatment effect with conventional therapy, and the inclusion of relevant comparators as permitted by the evidence available for each subpopulation.
• In the biologic-inadvisable subpopulation: Upadacitinib was dominated by secukinumab and etanercept (i.e., patients receiving secukinumab or etanercept gained more QALYs at a lower cost).
• In the biologic-inadequate response subpopulation (which did not consider all relevant TNF inhibitors): Upadacitinib dominated secukinumab. The ICER for upadacitinib relative to conventional therapy was $52,442 per QALY gained.
• The results of these analyses are dependent on estimates of treatment effect from the sponsor’s NMA, which are associated with uncertainty, and these differences in treatment effect translate into clinically meaningful improvements in disease status for patients.