Maxime Verhoeven

82 Chapter 5 In total, four RCTs with TCZ intravenously at least in one treatment arm as monotherapy and in one treatment arm combined with MTX (i.e., TCZ+MTX) were identified; namely, ACT-RAY (n=553), 6 FUNCTION (n=1164), 7 SURPRISE (n=105), 8 and U-Act-Early (n=317). 9 The sharp van der Heijde (SvdH) score was used to measure radiographic progression in all trials. One RCT used a tight-control treat-to-target approach, indicating that treatment could continuously (4-weekly interval) be intensified when the treatment target was not achieved. 9 Two RCTs were performed in early RA, 7,9 and two tapered TCZ in case of remission. 8,9 In 2 RCTs, a stable dose of oral glucocorticoid (GC) use (≤10 mg/day prednisone or equivalent) was permitted next to study treatment. 6,7,10 Of 3 RCTs (n=2034), IPD could be obtained. 6,7,9 All 3 RCTs are registered at ClinicalTrials.gov. (ACT-RAY: NCT00810199, FUNCTION: NCT01007435, and U-Act-Early: NCT01034137), and all patients gave written informed consent. In short, inclusion criteria were RA according to the classification criteria, age above 18, and a moderate-to-active disease. Early RA patients were DMARD-naïve; established RA patients had responded insufficiently to MTX. Radiographs were assessed by a single reader in FUNCTION and U-Act-Early, and by two independent readers in ACT-RAY; of the latter study, the average score was used in our analysis. For detailed information see Supplementary Data S2. The percentage of missing data on radiographic outcome in the individual trials is reported in their respective publications and was on average 29%. Patients with missing radiographic data were not different from patients with radiographic data present regarding joint damage, disease duration and disease activity, all at baseline, and nor were different between treatments arms within trials, hence no imputation of missing data was performed. Statistical analysis The primary endpoint was defined as: not having radiographic progression (versus having any radiographic progression, i.e., a score >0) after two years, termed as “prevention of radiographic progression”; primary analyses were performed on total scores, and secondary on erosions and joint space narrowing (JSN) scores. The different TCZ-regimens that were compared consisted of TCZ8mg/kg q4w i.v. + MTX median 15mg/weekly (TCZ+MTX) and TCZ8mg/kg q4w i.v. (TCZ). Due to heterogeneity of patient populations and study results, we decided to analyze data of trials in early, MTX-naïve RA patients and trials in established RA patients separately. All analyses were according to the intention-to-treat principle and adjusted for gender and age. Treatment effect modification by baseline joint damage, disease duration and DAS28 was explored using these factors as covariates as well as in interaction terms with treatment in the models. Logistic mixed effect models with random intercept and random effect of treatment, both at study level, were used to analyze the data. If analysis results indicated that