Serum peptides as candidate biomarkers for relapsing polychondritis

Abstract

AbstractBackground:For relapsing polychondritis (RP), no useful biomarkers have yet been identified. We analyzed serum peptide profiles to identify candidate biomarkers.Methods:Patients with RP or rheumatoid arthritis (RA) and healthy control (HC) subjects were divided into training set (RP, n=19; RA, n=21; HC, n=17) and testing set (RP, n=18; RA, n=21; HC, n=18). Seven patients demonstrating granulomatosis with polyangiitis (GPA) were used for validation. The ion intensity of serum peptides was comprehensively measured by matrix-assisted laser desorption/ionization time-of-flight/time-of-flight mass spectrometry.Results:One hundred sixty serum peptides were detected. In the RP group of the training set, 24, 8, and 7 peptides showed a ³1.2-fold difference in ion intensity in comparison to the HC, RA, and HC+RA (non-RP) groups, respectively (p<0.05). Based on a supervised multivariate analysis of the ion intensity of 160 peptides, we generated 3 models that completely discriminated the RP group from the HC, RA, and non-RP groups (RP/HC-160P model, RP/RA-160P model, and RP/nonRP-160P model; AUROC, 1.000). By selecting 11, 9, and 14 peptides, the RP group was also completely discriminated from the 3 groups (RP/HC-11P model, RP/RA-9P model, and RP/nonRP-14P model; AUROC, 1.000). We attempted to identify the peptides with a ³1.2-fold difference in ion intensity between the RP group and one of the 3 groups and the peptides comprising the RP/HC-11P, RP/RA-9P or RP/nonRP-14P models. Nineteen peptides were identified. Most were fragments of proteins associated with coagulation. To obtain biomarker models for RP which consists of a few peptides, we further generated 330 models, using all combinations of 3 or 4 peptides out of the 10 identified peptides of the RP/nonRP-14P model. Among them, 69 models provided ³65.0% sensitivity and specificity in the training set (AUROC, 0.789-0.823). In the testing set and the testing set with GPA group, 4 models consisting of 4 peptides (RP/nonRP-4P-2, -10, -11, and -38 models) provided ³70.0% sensitivity and specificity (AUROC, 0.779-0.815). Notably,the RP/nonRP-4P-2 model provided 83.3% sensitivity and 71.7% specificity in the testing set with GPA group (AUROC, 0.802).Conclusion:Serum peptide profiles provided useful candidate biomarkers for RP and may be implicated in the pathophysiology of RP.Trial registration:University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR), UMIN 000037212. Registered on 30 June 2019.