Neuromyelitis optica spectrum disorders with and without connective tissue disorders

Neuromyelitis optica spectrum disorders (NMOSD) is a severe central nervous system (CNS) demyelinating syndrome, characterized by optic neuritis (ON) and acute myelitis [1]. Immunoglobulin G autoantibodies against aquaporin-4 (AQP4-IgG) play a key role in the pathogenesis and diagnosis of NMOSD. Standardized diagnostic criteria for NMOSD were published in 2015, further stratifying NMOSD based on serologic testing (NMOSD with or without AQP4-IgG). According to these criteria, NMOSD has six core clinical characteristics: ON, acute myelitis, area postrema syndrome, acute brainstem syndrome, acute diencephalic clinical syndrome and symptomatic cerebral syndrome [2].

In addition to AQP4-IgG, antinuclear autoantibodies (ANAs) are also often detectable in patients with NMOSD who do not have clinical evidence of a systemic autoimmune disease [3]. However, an increasing number of reports have revealed that NMOSD often coexists with connective tissue disorders (CTD), particularly systemic lupus erythematosus (SLE) and Sjögren syndrome (SS) [3‐5]. Whether there are clinically significant differences in NMOSD with and without CTD remains unclear. In this study, we investigated and compared the demographic, clinical, laboratory, and magnetic resonance imaging (MRI) characteristics of NMOSD with and without CTD.

From 2014 to 2017, a total of 57 patients satisfied the diagnostic criteria for inclusion in this study: 39 NMOSD patients without CTD and 18 with CTD (including 7 with SS, 3 with SLE, 1 with RA, and 7 UCTD). The demographic and clinical features of NMOSD with CTD are summarized in Table 1. Among the 18 patients with CTD, 9 developed CTD before NMOSD (ranging from 2 months to 30 years), and in the other 9 patients, the diagnosis of CTD followed the diagnosis of NMOSD (ranging from 1 month to 5 years). The demographic and clinical features of the patients are summarized in Table 2. Regular therapy was defined as a full dose of immunosuppressants. NMOSD patients with CTD had higher EDSS scores and more severe sensory disability at nadir than NMOSD patients without CTD (P < 0.05). No other significant demographic or clinical features difference was found in between NMOSD patients with and without CTD (P >  0.05).

The laboratory features of the patients are summarized in Table 3. CSF white blood cell (WBC) counts, protein, and chloride (Cl) showed no significant difference between the two groups. The level of serum IgG was significantly higher in NMOSD patients with CTD than those without CTD (P = 0.013), and a similar result was found for the percentage of γ-globulin (P = 0.023). Furthermore, ANA, anti-SSA/Ro antibodies (anti-SSA), anti-SSB/La antibodies (anti-SSB), anti-Ro52, and RF were significantly higher in NMOSD patients with CTD than in those without CTD (P < 0.05). However, there was no difference between the two groups in the positivity rate of AQP4-IgG, OCB, CRP (C-reactive protein), IgE, anti-dsDNA (anti-double stranded DNA antibodies), anti-nRNP (antinuclear ribonucleoprotein), anti-Sm (anti-Sm antibodies), anti-Scl70 (anti-topoisomerase I antibodies), anti-Jo1 (anti-Jo-1 antibodies), ACA (anti-neutrophil cytoplasmic antibodies), AnuA (anti-nucleosome antibody), AHA (anti-histone antibody), ARPA (anti-ribonucleoprotein antibodies), or GPI (Glucose-6 phosphate isomerase). No significant difference was found in the level of IgA, IgM, C3, or C4 between the two groups of patients.

The spinal and brain MRI features of the patients are summarized in Tables 4 and 5, respectively. Representative MRI abnormalities (arrows) in NMOSD patients with CTD are shown in Fig. 1. The length of spinal cord lesions was longer in NMOSD patients with CTD than in NMOSD patients without CTD (P = 0.018). There was no significant difference in the frequency of short TM at onset between the two groups. However, the frequency of short TM at the initial manifestation of myelitis was significantly higher in NMOSD patients without CTD than in those with CTD (P = 0.010). The frequency of T1 hypointensity and T2 BSLs in acute myelitis was higher in NMOSD patients with CTD than in those without CTD (P = 0.001 and 0.011, respectively). No significant difference between the two groups was observed in any of the other MRI features.

Pearson correlation results showed that EDSS scores were positively correlated with group classification (NMOSD with or without CTD) (r = 0.286, P = 0.031), the length of spinal cord lesions (r = 0.488, P < 0.001) and T1 hypointensity (r = 0.362, P = 0.006). EDSS scores showed no correlation with T2 BSLs (r = 0.172, P = 0.202) or AQP4-IgG positivity status (r = − 0.117, P = 0.388). However, partial correlation results showed that EDSS scores had no correlation with group classification after controlling for lesion length and T1 hypointensity (r = 0.003, P = 0.985).

In the present study, we found that patients with NMOSD and CTD were similar to those without CTD in all tested demographic and clinical features except EDSS scores, especially sensory disability at nadir. Furthermore, most clinical, laboratory, and MRI features also did not show significant differences between the two groups. However, a number of autoantibodies, CSF indexes, and MRI features differed significantly.

NMOSD patients with CTD had increased amounts of T1 hypointensity and T2 BSLs on spinal MRI in acute myelitis. T1 hypointensity and T2 BSLs probably indicated intense damage of the spinal cord [11, 12]. However, the characteristics of spinal MRI did not show any significant difference between the two groups in the chronic phase. These findings may partially explain the differences in sensory disability and EDSS scores at nadir. Since CTDs can cause peripheral neuropathy leading to sensory deficits, electromyography should be performed to exclude that diagnosis. None of the patients in the present study showed clinical symptoms or signs caused by peripheral neuropathy until the last follow-up, according to their records. Pain is common in NMOSD and can lead to a reduced quality of life [13, 14]. In the present study, neuropathic pain (NP) was defined carefully according to previous reports [13, 15]. Almost half of the patients in both groups complained of NP in the present study, the characteristics and intensity of this pain should be evaluated in the future using previously reported methods [13].

The serum autoantibody AQP4-IgG is highly sensitive and specific for NMOSD [16]. In 2004, Lennon and colleagues reported this autoantibody as a frequent feature of NMOSD; the presence of the antibody was 73% sensitive and 91% specific for clinically defined neuromyelitis optica (NMO) [17]. Since then, many study groups have tested AQP4-IgG in patients with NMO or NMOSD, with the reported frequency varying according to the assay, e.g., immunofluorescence, immunoprecipitation, radioimmunoprecipitation or CBA [18‐22]. The new diagnostic criteria for NMOSD recommended CBA for the detection of AQP4-IgG because of its good sensitivity and specificity. Hence, we adopted CBA to detect AQP4-IgG in the present study. The seroprevalence of AQP4-IgG was 74.4% and 66.7% in patients with and without CTD, respectively, with no significant difference. This result was similar to those of Jarius et al. [23], who found that AQP4-IgG seropositivity in patients with CTD and co-existing neurological disorders was restricted to those with NMOSD [23]. This finding strongly suggests that AQP4-IgG was linked to the pathogenesis of NMOSD in those patients.

Although the frequency of ANAs was significantly higher in NMOSD patients with CTD than in those without CTD in our study, the frequency of ANAs in the latter group was more than 50%, which was similar to the frequencies reported in other reports [5, 11, 23‐26]. Until now, the exact relationship between NMOSD and CTD has been unknown. A prevailing notion of NMOSD pathogenesis is that it is a complication of CTD manifested predominantly in the CNS. If a complication of CTD were the cause of NMOSD, one would expect that CTD would consistently be diagnosed before the onset of NMOSD. However, of the 18 NMOSD patients with CTD in our study, only 9 developed CTD before NMOSD, and in the other 9, the diagnosis of CTD followed the diagnosis of NMOSD, which was similar to previously reported data [5]. This chronology may support the current hypothesis that NMOSD and CTD can be co-existing conditions that are clinically expressed in patients susceptible to autoimmunity [27]. In the present study, the level of serum IgG and the percentage of γ-globulin were significantly higher in patients with CTD than in those without CTD, which was in agreement with a previous study [28]. Zhang and co-workers also found higher levels of CRP in NMOSD patients with autoimmune diseases than in those without autoimmune diseases. These findings may suggest that NMOSD patients with CTD have an intensified autoimmune response.

Longitudinally extensive transverse myelitis (LETM) lesions on spinal cord MRI are regarded as one of the six core clinical characteristics of NMOSD. The spinal cord lesions of NMOSD patients with CTD were longer than those of NMOSD patients without CTD. An increased frequency of both ‘T1 dark’ and T2 BSLs was found in NMOSD patients with CTD compared to those without CTD, in contrast to the findings of another study [28]. As previously reported, ‘T1 dark’ signals on spinal MRI was relatively specific to NMOSD, and were rarely found in MS [11, 12]. This feature may indicates early necrosis and cavity formation caused by intense cord damage. In previous reports, BSLs on axial T2WI were the most distinctive finding of NMO on spinal MRI, with frequencies of 54% and 86.1%, respectively [11, 12]. The pathophysiology of BSLs remains unclear. Compared to BSL-negative patients, BSL-positive patients had a higher frequency of contrast-enhanced lesions. Hence BSLs probably reflect severe damage to the spinal cord, which may destroy the blood-brain barrier (BBB) [11]. These significant differences between the two groups may also indicate that NMOSD patients with CTD are prone to intense autoimmune responses, which is in accordance with the elevated IgG, increased EDSS scores at nadir and intensified sensory disability in these patients compared to those without CTD. CTD-induced tissue damage probably promotes AQP4-IgG-induced pathology. ANAs or other inflammatory mechanisms in CTD may contribute to vascular damage and disruption of the BBB induced by vasculitis, which makes the AQP4-IgG accessible to the CNS and triggers the AQP4-IgG-mediated inflammatory cascade [23, 27]. EDSS scores were positively correlated with group classification (NMOSD and NMOSD with CTD), the length of spinal cord lesions and T1 hypointensity. However, partial correlation results showed that EDSS had no correlation with different groups after adjusting for the length of spinal cord lesions and T1 hypointensity. Therefore, the elevated EDSS scores in NMOSD patients with CTD may be due to the increased length of spinal cord lesions and frequency of T1 hypointensity. The results may also be influenced by other factors, which we will explore and investigate in the future.

Short TM (< 3 vertebral segments) is considered non-characteristic of NMOSD. In our study, we found that short TM is not uncommon in NMOSD. A short TM episode as an onset symptom was found in 17.9% of NMOSD patients without CTD and 5.6% of those with CTD, which were higher than those published in a previous report [29]. This difference was probably because Flanagan’s study included only AQP4-IgG-positive NMOSD cases. The frequency of short TM at the initial manifestation of myelitis was lower in NMOSD patients with CTD than in those without CTD, which may also indicate that NMOSD patients with CTD have a more intense autoimmune response. This finding is also consistent with the elevated levels of IgG and EDSS at nadir in NMOSD patients with CTD. Short TM can be the first presentation of NMOSD, which may delay diagnosis and treatment. Therefore, short TM, especially complicated with ‘T1 dark’ and/or T2 BSLs, should not exclude consideration of AQP4-IgG and ANAs testing, which may be helpful for avoiding delayed diagnosis and treatment of NMOSD.

Brain MRI in NMOSD is historically thought to be normal or non-specific, especially at the onset of the disease. However, the presence of certain lesions described as specific to NMOSD may be helpful in its diagnosis. Classic NMOSD lesions are those located at sites of high AQP4 expression, such as the periependymal areas, which include the hypothalamus, periaqueductal grey and area postrema [30]. Brain lesions are present in approximately half of patients at presentation [30, 31], and increase in number with disease progression [30, 32]. We investigated brain MRI findings at onset in all patients in the study. The frequency of abnormal brain MRI findings was 35.9% in NMOSD patients without CTD and 50% in those with CTD; these rates were not significantly different. Nearly all of the supratentorial lesions were non-specific and asymptomatic. These lesions were dot-like or patchy, < 3 cm in diameter, and located in the cerebral deep white matter, as previously described [33]. Infratentorial lesions were more common and specific than supratentorial lesions. In our study, patients had a higher frequency of medulla oblongata lesions than pons or midbrain lesions. This result was consistent with previous reports [28, 34‐37]. Although the medulla was the most common brainstem lesion location in NMO, only a few patients showed MRI lesions in the area postrema at onset in our study, which was consistent with the findings of our previous study [10], and all but one of the patients were free of apparent hypothalamic lesions at onset. ‘Classic-’ or ‘typical NMO lesions’ located at the hypothalamus and area postrema are highly specific to the diagnosis of NMOSD; however, these lesions are seen in only a minority of patients, especially at onset [30, 38, 39].

This study had certain limitations. First, as this study is retrospective, bias is inevitable. Second, all the patients came from a single centre, and were insufficient in number. In the future, we will recruit more NMOSD patients with and without CTD and test the titer of AQP4-IgG, which may lead to a deeper understanding of the significance of CTD in NMOSD.

In conclusion, NMOSD patients with and without CTD were similar in most of the demographic, clinical and laboratory features that we examined. NMOSD patients with CTD have more frequent of T1 hypointensity and T2 BSLs, longer spinal cord lesions on MRI and a lower frequency of short TM than those without CTD.