Abstract
OBJECTIVE: Baricitinib preserves β-cell function in people with recently diagnosed type 1 diabetes. We aimed to determine whether simple routine clinical measures could be used to assess β-cell preservation and predict treatment response. RESEARCH DESIGNS AND METHOD: Measures of β-cell function derived from clinical and biochemical measures were calculated using data from the BAricitinib in Newly DIagnosed Type 1 diabetes (BANDIT) randomized trial of baricitinib in recent-onset type 1 diabetes. Measures that reported and predicted treatment efficacy were determined using linear regression and receiver operator characteristic analysis, respectively. Therapeutic predictors were validated using data from trials of rituximab, abatacept, and antithymocyte globulin. RESULTS: Quantitative response score (QRS), fasting C-peptide, and model-estimated C-peptide (CPest) most reliably differentiated placebo-treated from baricitinib-treated participants at 24 and 48 weeks. The Beta2 score, derived from fasting glucose, C-peptide, HbA1c, and insulin dose at 12 weeks, was optimal for predicting QRS >0 following 1 year of treatment with baricitinib and the other immunotherapies (areas under receiver operator curve 0.864 and 0.765, respectively). A 6.2% decrease in the Beta2 score at week 12 predicted significant improvement in HbA1c (-0.6% or -6 mmol/mol) and insulin use (-0.26 units/kg/day) in combined data from the rituximab, abatacept, and antithymocyte globulin trials. CONCLUSIONS: QRS, fasting C-peptide, and CPest could be used as more efficient, less burdensome primary outcome measures for future immunotherapy trials. The ability of the Beta2 score to predict treatment responses could facilitate adaptive trial designs and help guide treatment decisions in the clinic.