Clinical Significance of Diabetes-Mellitus-Associated Antibodies in Rheumatoid Arthritis

Abstract

Rheumatoid arthritis (RA) is a canonical autoimmune disease that shares numerous risk factors with diabetes mellitus (DM). The production of autoantibodies is a characteristic feature in both diseases. To determine the frequency and specificity of DM-related antibodies (DMab) in RA patients and to study whether DMab associates with new DM cases in RA patients, we measured DMab defined as IgG against glutamic acid decarboxylase (GADA), tyrosine phosphatase (IA2-ab), and zinc transporter (ZnT8-ab) in a cohort of 290 RA patients (215 women and 75 men, median disease duration 11 years). Of those, 21 had a DM diagnosis at baseline. The development of new DM cases and mortality were traced in a 10-year prospective follow-up. Predictive analyses for DM and mortality were carried out by the Mantel-Cox regression. We found that 27 of the patients (9.3%) had DMab, equally often men and women. The presence of DMab was more frequent in patients with DM (p = 0.027. OR 4.01, 95%CI [1.20; 11.97]), suggesting their specificity for the disease. Men had more prevalent incidental DM at the baseline (12% vs. 5%, p = 0.030) and among the new DM cases (p = 0.012. HR 6.08, 95%CI [1.57; 25]). New DM developed equally frequently in DMab-positive and DMab-negative patients. DM, but not DMab, significantly increased the estimated mortality rate in RA patients (p = 0.021, OR 4.38 [1.2; 13.52]). Taken together, we conclude that DMab are associated with DM in RA patients, but they are not solely enough to predict disease development or mortality in those patients.

Keywords: autoantibodies; diabetes mellitus; prospective follow-up; rheumatoid arthritis.

Figure 1

Flow chart of RA patients included in the study. (A) Baseline cohort. (B) Follow-up cohort. DM, diabetes mellitus; DMab, diabetes-mellitus-specific antibodies; RA, rheumatoid arthritis; M, male; F, female.

Figure 2

Clinical associates of DM-specific antibodies in RA patients. (A) Clinical characteristics of RA patients at baseline. (B) Heatmap of DM, rheumatoid factor (RF), antibodies-to-cyclic-citrullinated peptides (ACPA), and gender distribution among the patients with the DM antibodies GADA, anti-ZnT8, and anti-IA2. Red box indicates a case with new DM. (C) A 2 × 2 table of coexistence between DM and DMab. (D) Bar graphs showing the frequencies of DM, DMab with DM, and DMab in men and women. (E) Box plots of clinical and serological variables in patients with DM without DMab (n = 14, green), in patients with DMab with and without DM (n = 27, blue), and in patients without DM and DMab (n = 249, white). (F) Scatter plot of principal component analysis (PCA) shows variables associated with DMab. Significantly associated variables are indicated with p-value (calculated with least-squares regression). Box plots present median interquartile range, the whiskers indicate min and max values except for when the max is very high (indicated with a line becoming a dotted line). p-values were calculated with mixed exact chi-square test and the Kruskal–Wallis test, followed by Dunn’s multiple comparisons test. AU, arbitrary units; BMI, body mass index; DD, disease duration; DM, diabetes mellitus; TC, total cholesterol.

Figure 3

Predictive value of DM-specific antibodies for new cases of DM and mortality in RA patients. (A) The Kaplan–Meier curves show the development of new cases of DM in DMab+ (n = 27 red line) and DMab-negative (n = 257, blue line) patients and in RAab+ (n = 234, dash red line) and RAab-negative (n = 31, dash blue line) patients during the follow-up period of 120 months. The hazard ratio (HR) between the groups was calculated by the Mantel–Cox (MC) regression analysis. (B) The Kaplan–Meier curves show the development of new cases of DM in male (n = 66, red line) and female (n = 205, blue line) patients. The hazard ratio (HR) between the groups was calculated by the Mantel–Cox regression analysis. (C) Scatter plot of principal component analysis (PCA) shows variables associated with new DM cases. Significantly associated variables are indicated with p-value (calculated with least-squares regression). (D) Forest plot of variables in the final step of the binary logistic regression analysis predicting new DM cases. p-values were calculated with Wald statistics. (E) The Kaplan–Meier curves show mortality in RA patients with DM (n = 30, red line) and without DM (n = 254, blue line) and in DMab+ (n = 27, dash red line) and DMab-negative (n = 257, dash blue line) patients during the follow-up period of 120 months. The hazard ratio (HR) between the groups was calculated by the Mantel–Cox regression analysis.