Diagnostic value of serum IgG4 for IgG4-related disease

Abstract

Many studies about serum IgG4 for the diagnosis of IgG4-related disease (IgG4-RD) have been reported. However, these studies had relatively small sample sizes and the diagnostic accuracy values varied much between them. The aim of this study was to perform a meta-analysis to evaluate the diagnostic value of serum IgG4 for IgG4-RD. We conducted a search of relevant articles using MEDLINE, EMBASE, Web of Science, SCOPUS, and Cochrane Library databases published before December 2015. Studies those assessed the diagnostic accuracy of serum IgG4 for IgG4-RD and those provided the cut-off value for serum IgG4 were included. Data were synthesized using the random-effect model. Statistical analysis was performed using STATA with the MIDAS module and Meta-DiSc 1.4 software. A total of 9 case-control studies were analyzed, which included 1235 patients with IgG4-RD and 5696 overall controls. The pooled estimate, for a cut-off value ranged from 135 to 144mg/dL, produced a sensitivity of 87.2% (95% CI, 85.2-89.0%) and a specificity of 82.6% (95% CI, 81.6-83.6%). The positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were 6.48 (95% CI, 3.98-10.57), 0.14 (95% CI, 0.09-0.21), and 45.15 (95% CI, 23.41-87.06), respectively. The area under the curve (AUC) of the summary receiver operating characteristic curve (SROC) was 0.94 (0.92-0.96). When a cut-off value of 2-fold the upper limit of normal was used (ranged from 270 to 280mg/dL), the pooled sensitivity was 63% (95% CI, 60.0-66.0%), and the specificity was 94.8% (95% CI, 94.1-95.4%). The PLR, NLR, and DOR were 13.3 (95% CI, 7.39-24.0), 0.41 (95% CI, 0.29-0.58) and 33.42 (95% CI, 13.88-80.43), respectively. The AUC of the SROC was 0.92 (0.90-0.94). Only a relatively small number of studies were included, and significant heterogeneity was observed in this meta-analysis. Serum IgG4 is a modestly effective marker to diagnose IgG4-RD. Doubling the cut-off value for IgG4 could not improve the overall test characteristics. A high specificity inevitably accompanies with a significant sacrifice in sensitivity.