IgG4 surge and relapse: determining the predictive threshold for IgG4 rising kinetics in IgG4-related disease.

Abstract

To investigate longitudinal IgG4 dynamics in IgG4-related disease (IgG4-RD), identify the predictive value of IgG4 rising kinetics for relapse and evaluate outcomes of intensified treatment following IgG4 surge. This cohort study was conducted at the Chinese PLA General Hospital from February 2016 to May 2023, including patients with >6 months of follow-up and ≥3 IgG4 measurements. Single-timepoint IgG4 level and IgG4-Ratio (sequential visit-to-visit fold-change) were dichotomized as binary indicators using variable thresholds to predict relapse assessed at a third consecutive visit in time-dependent Cox regression models. We used restricted cubic splines (RCS) to identify the optimal IgG4-Ratio threshold with adjustment for confounders. Subsequently, relapse-free survival after intensified treatment was analysed. Among 155 patients (median follow-up: 33 months; inter-visit interval: 3 months), 149 had baseline IgG4 positivity. IgG4 normalized in 63.8% (95/149) patients, of whom 61.1% (58/95) developed IgG4 re-positivity. Single-timepoint IgG4 level, whether as continuous or dichotomized variables, showed no significant predictive value for relapse. IgG4-Ratio thresholds >1.1-1.3 and ​>1.6-1.8 significantly predicted relapse (P < 0.05), with RCS confirming a nonlinear dose-response relationship (P = 0.008). IgG4-Ratio >1.6 was identified as the optimal threshold to predict relapses. Among 45 patients with IgG4-Ratio >1.6, intensified treatment reduced relapse risk by 68.3% (HR 0.317, 95% CI 0.102-0.988; P = 0.048). A paradigm shift from relying on static IgG4 levels to monitoring IgG4 dynamics is needed. The IgG4 surge defined as an IgG4-Ratio >1.6 (indicating a >60% increase in IgG4 levels between consecutive measurements taken at median 3-month intervals) predicts relapse in the majority of seropositive IgG4-RD patients. Intensified treatment in patients with IgG4 surge can reduce recurrence risk.