Biological Agents in the Treatment of Crohn's Disease: A Propensity Score–Matched Analysis From the Prospective Persistence Australian National IBD Cohort (PANIC3) Study

INTRODUCTION: Comparative effectiveness research provides data on the relative benefits and risks between treatments. In Crohn's disease (CD), however, there are few head-to-head studies comparing advanced therapies and none with long-term follow-up. Real-world effectiveness, defined by treatment persistence, obtained from prospective population-based patient cohorts, may help determine the best sequencing and positioning of biological agents. METHODS: We analyzed the prospectively collected population-based Australian national Pharmaceutical Benefits Scheme dispensing data registry (2005–2019) for CD. There is no mandated biological agent prescribing order, and all citizens and permanent residents are eligible for treatment irrespective of insurance status. Propensity score matching was performed to reduce selection bias. RESULTS: There were 2,029 lines of therapy in 1,446 patients (median age 43 years, interquartile range 34–58, 44% male patients) over the 15-year period with 5,618 patient-years of follow-up. Per line of therapy, 915/2,029 (45.1%) patients used adalimumab, 722/2,029 (35.6%) used infliximab, 155/2,029 (7.6%) used vedolizumab, and 237/2,029 (11.7%) used ustekinumab. When used in biological agent–naive patients, there was no difference in persistence between any agent ( P > 0.05). Used after first line in biological agent–experienced CD, ustekinumab had significantly better persistence than non-ustekinumab biological agents ( P = 0.0018), vs anti–tumor necrosis factor (TNF) alpha therapy ( P = 0.006) or vedolizumab ( P < 0.001). Ustekinumab persistence was unaffected by prior biological agent exposure ( P = 0.51). After anti-TNF use, ustekinumab had superior persistence to an alternative anti-TNF agent ( P = 0.033) and to vedolizumab ( P = 0.026). Using a propensity score–matched analysis adjusted for age, immunomodulator use, and bio-exposed status, ustekinumab had superior persistence to anti-TNF ( P = 0.01). Multivariate predictors of worse persistence were the use of a non-ustekinumab biological agent (adjusted hazard ratio 2.10, P < 0.001), and bio-experienced status (adjusted hazard ratio 1.23, P < 0.001). DISCUSSION: This large national prospective database with nonhierarchical prescribing of biological agents did not identify superior persistence of any agent in bio-naive CD. However, for patients with bio-experienced CD, persistence was greater with ustekinumab.