Peptides From the Variable Domain of Immunoglobulin G as Biomarkers in Chronic Inflammatory Demyelinating Polyradiculoneuropathy

Abstract

Background and objectives: Chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) is a clinically heterogeneous immune-mediated disease. Diagnostic biomarkers for CIDP are currently lacking. Peptides derived from the variable domain of circulating immunoglobulin G (IgG) have earlier been shown to be shared among patients with the same immunologic disease. Because humoral immune factors are hypothesized to be involved in the pathogenesis of CIDP, we evaluated IgG variable domain-derived peptides as diagnostic biomarkers in CIDP (primary objective) and whether IgG-derived peptides could cluster objective clinical entities in CIDP (secondary objective).

Methods: IgG-derived peptides were determined in prospectively collected sera of patients with CIDP and neurologic controls by means of mass spectrometry. Peptides of interest were selected through statistical analysis in a discovery cohort followed by sequence determination and confirmation. Diagnostic performance was evaluated for individual selected peptides and for a multipeptide model incorporating selected peptides, followed by performance reassessment in a validation cohort. Clustering of patients with CIDP based on IgG-derived peptides was evaluated through unsupervised sparse principal component analysis followed by k-means clustering.

Results: Sixteen peptides originating from the IgG variable domain were selected as candidate biomarkers in a discovery cohort of 44 patients with CIDP and 29 neurologic controls. For all 16 peptides, univariate logistic regressions and ROC curve analysis demonstrated increasing peptide abundances to associate with increased odds for CIDP (area under the curves [AUCs] ranging from 64.6% to 79.6%). When including age and sex in the logistic regression models, this remained the case for 13/16 peptides. A model composed of 5/16 selected peptides showed strong discriminating performance between patients with CIDP and controls (AUC 91.5%; 95% CI 84.6%-98.4%; p < 0.001). In the validation cohort containing 45 patients and 43 controls, 2/16 peptides demonstrated increasing abundances to associate with increased odds for CIDP, while the five-peptide model demonstrated an AUC of 61.2% (95% CI 49.3%-73.2%; p = 0.064). Peptide-based patient clusters did not associate with clinical features.

Discussion: IgG variable domain-derived peptides showed a valid source for diagnostic biomarkers in CIDP, albeit with challenges toward replication. Our proof-of-concept findings warrant further study of IgG-derived peptides as biomarkers in more homogeneous cohorts of patients with CIDP and controls.

Classification of evidence: This study provides Class III evidence that the pattern of serum IgG-derived peptide clusters may help differentiate between patients with CIDP and those with other peripheral neuropathies.