In this study, a total of 258 patients with NMOSD in the Department of Neurology, Huashan Hospital, Fudan University, from September 2015 to September 2021 were enrolled, from which a total of 143 AQP4-IgG-positive NMOSD patients were included ( Figure 1 ). All patients met the 2015 International Panel for NMO diagnostic criteria (7). The clinical data of the NMOSD patients were collected, including the gender, age, clinical symptoms and signs, location and length of the affected spinal cord segments, status of OCBs, Expanded Disability Status Scale (EDSS) at the first attack and at the 36-month follow-up, relapse times within 36 months, concomitant connective tissue disease (CTD), and other autoimmune antibodies (oAIA).

The exclusion criteria were as follows: 1) patients with other serious medical diseases, chronic infection (e.g., syphilis, HIV, etc.), epilepsy, and tumors; 2) patients positive for anti-myelin oligodendrocyte glycoprotein immunoglobulin G (MOG-IgG), anti-glial fibrillary acidic protein immunoglobulin G (GFAP-IgG), or onconeuronal antibodies; 3) less than 36 months of follow-up; and 4) incomplete clinical data.

Fresh paired serum and CSF samples were obtained within 30 days from the onset of the recent episode. Samples were stored at 2–8°C and analyzed within 15 days. This study was approved by the Ethics Committee of Huashan Hospital, Fudan University.

All patients were enrolled at the time of lumbar puncture and had received corticosteroid therapy. During 36 months of follow-up, 60 patients received azathioprine, 62 received mycophenolate mofetil, 14 received rituximab, and 7 received other medications (e.g., tacrolimus, cyclophosphamide, methotrexate, and cyclosporine) ( Table 1 ).

The serum AQP4-IgG, MOG-IgG, and GFAP-IgG concentrations of all enrolled patients were determined with a cell-based assay (CBA) using the kit from Simcere Medical Diagnosis Co., Ltd. (Nanjing, China). Antibody titers ≥1:10 were considered as positive. Other autoimmune antibodies [i.e., antinuclear antibody (ANA), anti-Sjogren’s syndrome type A (SSA) and type B (SSB) antibodies, anti-double-stranded DNA (ds-DNA) antibody, etc.] were detected using the diagnostic kit from EUROIMMUN Medical Diagnostics (Beijing, China). To avoid inter-assay variations, the same positive control, negative control, and blank control were used as the reference samples for each detection.

The IgG and albumin concentrations in the serum and CSF were determined using the turbidimetric scattering method. The blood–CSF barrier (BCB) function was assessed using the CSF/serum albumin quotient: (Q _Alb) = CSF − Alb (mg/L)/serum − Alb (g/L), with BCB dysfunction defined as Q _Alb > 4 + (age/15) (5, 14).

The IgG index, calculated as (CSF IgG/serum IgG)/(CSF albumin/serum albumin), was a measure of the intrathecal IgG synthesis adjusted for the degree of damage of the BCB. A value below 0.7 was considered normal (14).

The detection of OCBs was performed using a Sebia HYDRASYS 2 Isofocusing system (PN1211, Lisses, France) according to the manufacturer’s instructions. Briefly, the CSF and serum samples were run in parallel on an agarose gel. After electrophoresis, the gel was incubated with peroxidase-labeled anti-IgG antibodies, and the bands were displayed using TTF1/2 chromogenic agents (15). The two inspectors interpreted the electrophoresis results independently according to the following interpretations: type I, no bands in both the serum and CSF; type II, more than two bands in the CSF, but no band in the serum; type III, additional bands in the CSF despite bands in the serum; type IV, symmetrically separated OCB bands in the serum and CSF; and type V, twin bands with regular and periodic spacing in both the serum and CSF (16, 17). Types II and III were considered positive for CSF OCBs (OCBs+), while the remaining three types were considered negative (OCBs−).

SPSS 19.0 software was used for statistical analysis. The normality of continuous data was assessed using the Shapiro–Wilk test. Data with normal distribution are expressed as χ ¯ ± SD, while those with non-normal distribution are expressed as median (interquartile range, IQR). Comparison of the classification variables between groups was performed using Pearson’s χ ² test or Fisher’s exact probability method or the Cochran–Mantel–Haenszel test. Pairwise comparisons were performed using an independent two-sample t-test or the Mann–Whitney U-test, as applicable. A p < 0.05 was considered statistically significant.

A total of 143 patients with NMOSD, including 14 men and 129 women (men/women = 1:9.2), were enrolled in this study ( Figure 1 ). The median age at enrollment was 44 (18) years. In the OCBs+ group, 15 patients were included, two men and 13 women (men/women = 1:6.5), with a median age of 43 (22) years. The OCBs− group comprised 128 patients, 12 men and 116 women (men/women = 1:9.7), with a median age of 46 (18) years. There were no significant differences in the gender and age distribution between the two groups (p = 0.643 for gender and p = 0.685 for age).

Of the 143 patients, there were 16 (11.2%) patients with CTD and 46 (32.2%) patients with oAIA.

The median relapse time within 36 months was 2.0 (2.0), and the EDSS at baseline and at 36 months were 4.0 (2.0) and 3.0 (2.0), respectively. With regard to the lesion distribution revealed by MRI, of the 143 cases, 56 (39.2%) had intracranial lesions, 78 (54.5%) had optic nerve lesions, and 113 (79.0%) had spinal cord (SC) lesions with the median length of 4.0 (4.0) vertebral segments.

Considering the possible impact of drugs on the laboratory indicators, the differences in drug use among patients were compared with the different OCB status. In the OCBs+ group, six received azathioprine, six received mycophenolate mofetil, two received rituximab, and one received other medications (tacrolimus). In the OCBs− group, 54 received azathioprine, 56 received mycophenolate mofetil, 12 received rituximab, and 6 received other medications (e.g., tacrolimus, cyclophosphamide, methotrexate, and cyclosporine). There were no differences in the therapies between patients with positive and negative OCBs (p = 0.944) ( Table 1 ).

In general, the median CSF cell count and protein level were 5 (15)/mm³ and 487 (284) mg/L, respectively. Among the 143 cases, 49 (34.3%) had BCB destruction and 20 (14.0%) had an elevated IgG index. Moreover, there were 128 (89.5%) patients with negative OCBs and 15 (10.5%) patients with positive OCBs, including 11 type II and 4 type III.

For CSF biochemistry, OCBs+ patients had higher white blood cell counts than OCBs− patients [15.0 (27.0) vs. 5.0 (12.0)/mm³, p = 0.008]. Detailed white cell differentiation analysis of the 143 patients revealed that 130 (90.9%) exhibited a predominance of mononuclear cells in their CSF [11/15 (73.3%) in the OCBs+ group and 119/128 (93.0%) in the OCBs− group], while 13 (9.1%) showed a predominance of polynuclear cells [4/15 (26.7%) in the OCBs+ group and 9/128 (7.0%) in the OCBs− group]. Notably, the proportion of patients with a predominance of polynuclear cells in the CSF was significantly higher in the OCBs+ group compared with the OCBs− group (4/15 vs. 9/128, p = 0.032). There were no statistically significant differences for the CSF protein levels [571.0 (381.0) mg/L vs. 467.0 (286.0) mg/L, p = 0.070] between these two groups.

In the OCBs+ group, 8 out of 15 (53.3%) patients had elevated Q _Alb with a median IgG index of 0.68 (0.23), whereas in the OCBs− group, 41 out of 128 (32.0%) cases had elevated Q _Alb with a median IgG index of 0.52 (0.15). OCBs+ patients had a higher IgG index than OCBs− patients (p < 0.001) ( Table 1 ).

In contrast to OCBs− patients, more OCBs+ patients had concomitant CTD [5/15 (33.3%) vs. 11/128 (8.6%), p = 0.014] and oAIA [9/15 (60.0%) vs. 37/128 (28.9%), p = 0.020] ( Table 1 ).

During the baseline MRI evaluation at enrollment, brain/brainstem involvement was detected in 8 out of 15 (53.3%) OCB+ patients compared with 48 out of 128 (37.5%) in the OCB− group. Optic nerve lesions were identified in 9 out of 15 (60.0%) and in 69 out of 128 (53.9%) patients. SC abnormalities comprised the most prevalent imaging feature, which were present in 14 out of 15 (93.3%) OCB+ cases and in 99 out of 128 (77.3%) OCB− cases. No statistically significant intergroup differences were observed in the lesion distribution patterns (all p > 0.05) ( Table 1 ).

At the time of enrollment, myelitis was present in 14 out of 15 (93.3%) OCB+ patients and in 99 out of 128 (77.3%) OCB− patients, with median SC lesion lengths of 6.0 (7.0) and 4.0 (4.0) vertebral segments, respectively. No statistically significant intergroup differences were identified in either myelitis prevalence or lesion length (both p > 0.05) ( Table 1 ).

For OCBs+ patients, the median EDSS at enrollment was 4.0 (1.0), while it was 3.0 (2.0) at the 36-month follow-up; the median relapse times within 36 months was 2.0 (3.0). For OCBs− patients, the median EDSS was 4.0 (2.0) at lumbar puncture and was 3.0 (2.5) at the 36-month follow-up, and the median relapse time at the 36-month follow-up was 1.0 (2.0). OCBs+ cases had more relapses than OCBs− patients (p = 0.039). However, there were no significant statistical differences in the EDSS between the two groups, either at enrollment or at the 36-month follow-up (both p > 0.05) ( Table 1 ).

Among the 15 OCBs+ patients, the onset episode manifested as myelitis in 10 (66.7%) cases, ON in 8 (53.3%), and area postrema syndrome in 2 (13.3%). Among the 128 OCBs− patients, the onset episode included myelitis in 83 (64.8%) cases, ON in 54 (42.2%), and area postrema syndrome in 16 (12.5%). No statistically significant differences were observed between these two groups (all p > 0.05) ( Table 2 ).

During the 36-month follow-up period, disease relapses occurred in 11 OCBs+ patients, manifesting as myelitis in 10 (90.9%) cases, ON in 3 (27.3%), and area postrema syndrome in 0 (0%). Among the OCBs− patients, 66 experienced relapses, presenting as myelitis in 51 (77.3%), ON in 30 (45.5%), and area postrema syndrome in 1 (1.5%). Again, no statistically significant differences were observed (all p > 0.05) ( Table 2 ).

At the 36-month follow-up, the cumulative clinical involvement among all patients was as follows: in the 15 OCBs+ patients, myelitis was observed in 14 (93.3%) cases, ON in 9 (60%), and area postrema syndrome in 2 (13.3%). In the 128 OCBs− patients, myelitis was observed in 107 (83.6%) cases, ON in 68 (53.1%), and area postrema syndrome in 17 (13.3%). The observed variations demonstrated no statistically significant differences between the two groups (all p > 0.05) ( Table 2 ).