Optimizing Ulcerative Colitis maintenance treatment: Indirect comparison on number needed to treat among interleukin-23p19 inhibitors highlights mirikizumabs efficacy

Abstract

Background: We indirectly compared the interleukin-23p19 inhibitors (IL-23p19i) mirikizumab (MIRI), risankizumab (RIS), and guselkumab (GUS) in adults with ulcerative colitis (UC) by estimating the number of patients needed to treat (NNT),1 versus placebo (PBO), for 1 additional patient to benefit after approximately (approx.) 1 year of maintenance treatment. Method(s): In each phase 3 trial (MIRI: NCT03518086/NCT035240922; RIS: NCT03398148/NCT033981353; GUS: NCT040334454), adults with UC received IL-23p19i or PBO intravenously for 12 weeks. Those who achieved a clinical response (CRes) on IL-23p19i after the 12-week induction were rerandomized to receive IL-23p19i (MIRI 200mg every 4 weeks Q4W]; RIS 180 or 360mg every 8 weeks Q8W]; GUS 100mg Q8W or 200mg Q4W]) or PBO subcutaneously. 2-4 Efficacy endpoints included CRes, clinical remission (CRem), endoscopic improvement (EI), and histo-endoscopic mucosal improvement (HEMI). We estimated NNTs1 for all 3 drugs compared with PBO after approx. 1 year of maintenance, based on published (unadjusted) response rates from those phase 3 trials. 3,4 We also estimated NNTs1 for MIRI based on adjusted response rates from anchored matching-Adjusted indirect comparison (MAIC) with RIS and GUS at study and arm level. Patients were matched for age, sex, UC duration, UC location, modified Mayo Score, concomitant corticosteroid and/or immunosuppressant use, and inadequate response/intolerance to prior advanced UC treatment. Result(s): NNTs for MIRI 200mg, based on adjusted rates from MAIC per treatment arm (2.8 for CRes, 3.2 for CRem, 2.9 for EI, and 3.6 for HEMI), were lower than NNTs for RIS 180mg (6.1 for CRes, 6.6 for CRem, and 5.2 for EI and HEMI). NNTs for MIRI 200mg, based on adjusted rates from MAIC per treatment arm, were also lower than NNTs for RIS 360mg (Fig 1). NNTs for MIRI 200mg, based on adjusted rates from MAIC per treatment arm (3.1 for CRes, 4.0 for CRem, 3.3 for EI, and 3.8 for HEMI), were comparable to NNTs for GUS 100mg (2.9 for CRes, 3.8 for CRem, 3.3 for EI, and 3.7 for HEMI). NNTs were also comparable for MIRI 200mg and GUS 200mg (Fig 2). MIRI NNTs were similar for each efficacy endpoint whether calculated from unadjusted or matching-Adjusted response rates at the study or arm level. Conclusion(s): MIRI achieved consistently lower NNTs than RIS across all disease-control endpoints, suggesting clinicians may need to treat approx. half as many patients with MIRI as with RIS for 1 year for 1 additional patient to benefit over PBO. NNTs for MIRI and GUS were similar, suggesting comparable efficacy. Though further studies are needed to confirm these findings, they provide valuable evidence to support the efficacy of MIRI across clinical endpoints for UC and to inform IL-23p19i selection in practice.