Diabetes Mellitus Is a Possible Risk Factor for Nodo-paranodopathy With Antiparanodal Autoantibodies

Frequencies of Antiparanodal Antibodies in the Cohort

Our cohort included 191 (84.1%) seronegative patients and 36 (15.9%) patients IgG seropositive for antiparanodal antibodies. The predominant antibody subclass was IgG4 in 18/36 patients, IgG3 in 12/36 patients, IgG2 in 3/36 patients, IgG1 in 1/36 patients, and not determinable in 2/36 patients. Table 1 displays serostatus and demographic data.

Increase in Frequency of DM in Seropositive Patients

A disorder of glucose metabolism was diagnosed in 17.2% of the entire cohort (39/227; according to the World Health Organization criteria¹¹: n = 2 DM type 1; n = 33 DM type 2; n = 4 impaired glucose tolerance). In seropositive patients, the frequency of DM was 33.3% and thus significantly higher compared with seronegative patients (14.1%), especially in anti–contactin-1-seropositive patients (58.3%; Table 2 and Figure, A). Performing a subanalysis in the CIDP and GBS cohort, we could show a significant increase in the frequency of DM in the CIDP subcohort (seropositive 33.3% vs seronegative 15.1%). In the GBS subcohort, we found a similar tendency that did not reach statistical significance (Table 2). Although patients with DM were significantly older than patients without DM in the total cohort (64.82 vs 56.57, p < 0.002), the mean age and female-to-male ratio did not differ between seropositive and seronegative patients (Table 2). In patients aged >60 years, the frequency of DM was still significantly elevated in seropositive vs seronegative patients.

• Download figure
• Open in new tab
• Download powerpoint

Figure Frequency of DM and Immunofluorescence Stainings on Pancreatic Tissue

(A) Frequency of diabetes mellitus is significantly elevated in patients seropositive for antiparanodal antibodies (33.3%) compared with seronegative patients (14.1%, p = 0.014) and with the general German population (9.9%, p < 0.001), especially in anti–contactin-1-seropositive patients (58.3% vs 14.1% in seronegative, p < 0.001 and 9.9% in the German population, p < 0.001). Significance levels are marked with asterisks: *p < 0.05, **p < 0.01, ***p < 0.001. (B) In seropositive patients not having received corticosteroid treatment within the last 28 days and who were therapy naive to rituximab, HbA1c levels (y-axis, %) were determined in 14 patients at the time point of serum withdrawal and correlated significantly with the autoantibody titer, displayed on a logarithmic scale (r = 0.58, p = 0.029). (C.a–l) Photomicrographs show human pancreatic normal tissue sections with nucleus staining (DAPI) shown in blue (C.a, C.d, C.g, and C.j) and double staining with synaptophysin as marker for the islets of Langerhans (displayed in green, C.b, C.e, C.h, and C.k) and serum or antiparanodal antibodies (displayed in magenta, C.c, C.f, C.i, and C.l). Serum of a patient with CIDP and DM type 1 with GAD antibodies binds to β cells in pancreatic islets of Langerhans (C.a–c), whereas serum of a patient with anti–contactin-1 antibodies (C.d–f) and commercial goat anti–contactin-1 (C.g–i) and commercial chicken anti–pan-neurofascin (C.j–l) do not show any binding. Photomicrographs of binding of the other patients' sera or commercial antibodies tested in the assay are not shown. Scale bar = 10 μm. CNTN = contactin-1; DAPI = 4′,6-diamidino-2-phenylindole; DM = diabetes mellitus; HbA1c = hemoglobin A1C.

Treatment with plasma exchange (PE), IVIg, and corticosteroids was assessed retrospectively in the last 28 days before serum withdrawal and rituximab or further immunosuppressive treatment until 1 year before the withdrawal. There were no significant differences in previous PE, IVIg, and corticosteroid treatment in patients with and without DM (PE 2/12 [16.6%] vs 2/24 [8.3%], p = 0.59; IVIg 2/12 [16.7%] vs 8/24 [33.3%], p = 0.44; corticosteroids 1/12 [8.3%] vs 13/24 [54.2%], p = 0.22). Nevertheless, patients having received corticosteroids (n = 8) were excluded from the titer analysis to avoid bias. They were mainly found in the nondiabetic group because corticosteroids are often avoided in patients with diabetes. Furthermore, corticosteroid treatment influences total IgG levels until 2–4 weeks after application.¹² None of the patients had received rituximab treatment or further immunosuppressive treatment before antibody testing. Titers in the remaining 28 seropositive patients ranged from 1:100 to 1:40,000 and were significantly higher in patients with DM than without DM (median of 1:2,000 vs 1:500, p = 0.035).

Hemoglobin A1c (HbA1c) was determined at the onset of neurologic symptoms in 103 (45.4%) patients. The maximum HbA1c values were significantly higher in patients with DM compared with individuals without diabetes (mean of 6.5 vs 5.5, p < 0.001), but did not differ in seropositive and seronegative patients with DM (Table 2). We performed a subanalysis of the frequency of DM with patients whose HbA1c values were measured and documented at the time point of serologic testing. Here, the frequency of DM stayed significantly higher in seropositive vs seronegative patients (Table 2). Furthermore, HbA1c levels correlated significantly with the autoantibody titer (r = 0.584, p = 0.029; Figure, B) in n = 14 patients whose titer and HbA1c were assessed simultaneously and considered in the analysis (see above).

In all seropositive patients, the diagnosis of DM preceded the acute onset of nodo-paranodopathy without any close temporal connection. In 2/12 seropositive patients, the time point of diagnosis was documented >10 years before the onset of neurologic symptoms. In the other patients, the exact time point of DM diagnosis was not documented, but all patients carried an established diagnosis of diabetes before the onset of nodo-paranodopathy, and 10/12 patients had received long-term antidiabetic treatment.

DM type 2 occurred independently of the predominant IgG subclass: in 1/1 (100%) patients with predominant IgG1, in 1/3 (33%) patients with predominant IgG2, in 3/12 (25%) patients with predominant IgG3, and in 6/18 (33.3%) patients with predominant IgG4. In 1 patient, DM type 1 was diagnosed 15 years before the onset of nodo-paranodopathy. This patient had reported normal total IgG4 levels 3 years before the onset of nodo-paranodopathy. At the onset of neurologic symptoms, IgG4 antibodies against pan-neurofascin were detected.

Relative Risks and Comparison to Previous Studies

The relative risk of DM compared with the general German population according to health insurance data¹³ was 3.4-fold higher in seropositive patients (33.3% vs 9.9%, p < 0.001; Figure, A) and 1.88-fold higher in our entire CIDP cohort (18.6% vs 9.9%, p < 0.01). The frequency of DM in our total CIDP cohort did not differ significantly from previously described European CIDP cohorts⁷ (n = 29/156, 18.6% vs n = 48/257, 18.7%, p > 0.999).

No Binding of Antiparanodal Antibodies to Pancreatic β-Cell Islets

On normal pancreatic tissue sections, commercial antibodies against synaptophysin and patient anti-GAD antibodies as positive controls bound specifically to insulin-producing β cells in the Langerhans islets (Figure, C). Neither the commercial antibodies against nodo-paranodal antigens nor the patient sera with anti-contactin-1, anti-Caspr-1, and anti-neurofascin antibodies showed any binding to β cells (Figure, C representatively illustrates binding assays with serum and commercial anti–contactin-1 and commercial anti–pan-neurofascin, other data not shown).