We conducted this observational cross-sectional monocentric study at the University Hospital “Klinikum rechts der Isar” of the Technical University, Munich, Germany, a tertiary center for neurological, rheumatologic, and otorhinolaryngology (ENT) diseases. We identified all patients with pSS seen between January 2015 and September 2021 in the departments of rheumatology, ENT, or neurology. Electronic files were screened for the corresponding code of Sjögren's syndrome (M35.0-) according to the 10th revision of the International Classification of Diseases (ICD-10). Part of the data obtained from this cohort was also integrated into the Big Data Sjögren Consortium (16).

Primary Sjögren's syndrome was diagnosed according to the classification criteria proposed by the International Sjögren's syndrome Criteria Working Group and approved by the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) (7). We excluded patients with a competing classifiable rheumatic disease in terms of an overlap syndrome.

We assessed systemic disease activity with the EULAR Sjögren's syndrome disease activity index (ESSDAI) (5, 17). This index comprises 12 domains and is regarded as the gold standard for scientific studies on pSS. However, due to the retrospective nature of our study, data sets on ESSDAI domains in our patients were often incomplete. Therefore, we narrowed the clinical domains to dichotomous variables (absent vs. present). CNS manifestations were classified based on further clinical and paraclinical data [evoked potentials/nerve conduction studies, magnetic resonance imaging (MRI), and CSF (cerebrospinal fluid) analysis].

We analyzed anti-SSA/Ro and anti-SSB/La antibodies with an enzyme-linked immunosorbent assay (ELISA, Orgentec Diagnostika, Mainz, Germany). Anti-aquaporin-4 IgG antibodies (AQP4-Ab) were determined by a commercially available cell-based assay (Euroimmun Perkin Elmer, Luebeck, Germany) and anti-myelin oligodendrocyte glycoprotein (MOG) antibody (MOG-Ab) with an in-house cell-based assay, as described previously (18).

The T2-weighted whole brain and spinal MRI scans of individual patients were evaluated in a blinded manner. The distribution of T2 hyperintensities was analyzed manually in a pattern-based approach. The spinal cord was analyzed in total and divided into cervical and thoracolumbar regions according to the anatomical nomenclature.

We used Fisher's test for nominal variables, Student's t-test for age/time comparisons, and Mann–Whitney U-test for ESSDAI parameters. We did not adjust for multiple testing due to the exploratory retrospective study design. We performed all statistical tests by using IBM SPSS Statistics for Windows (version 23.0, Armonk, NY, IBM Corp).

The involvement of the CNS was defined by the presence of focal neurological deficits indicative of CNS disorder and/or supported by morphological alterations of the brain and/or the spinal cord on MRI. Twenty-one patients had CNS involvement (21/194, 10.8%) and of these, 19 patients presented with a CNS lesion pattern compatible with demyelination based on radiological findings and/or CSF parameters (referred to as the “CNS group”). A total of 173 pSS patients without CNS manifestations were available for comparison (the “control group”; Table 1).

In the CNS group, 14 patients were female (14/19, 73.7%). The median age [interquartile range (IQR)] at symptom onset and diagnosis was 45 years (31–56) and 51 years (46.5–60), respectively. The median time of diagnosis was 2 years longer in the CNS group than in the control group (p = 0.009). In the CNS group, neurological deficits were the first manifestation of pSS in 16 patients (16/19; 84.2%), and in 14 patients (14/19, 73.7%), they were the primary reasons to consult a physician.

Only 2 of 19 CNS patients (2/19, 10.5%) reported sicca symptoms [xerostomia and/or keratoconjunctivitis sicca (KCS)] at first encounter, with one of them already having a history of known pSS. Eight patients (8/17, 47.0%) developed sicca symptoms in the course of the disease. Sicca symptoms occurred more frequently in the control group [control vs. CNS group; xerostomia 155/173 (89.6%) vs. 8/19 (42.1%) with p < 0.00001; KCS 147/173 (84.9%) vs. 8/19 (42.1%) with p = 0.0001)]. Regarding other signs of systemic pSS, only hematological alterations mainly comprising anemia were more present in the CNS group (p = 0.01; Table 1).

In the CNS group, seroprevalence for anti-SSA/Ro was higher leading to a definitive diagnosis of pSS [85/173 (49.1%) vs. 15/19 (78.9%), p = 0.016], while the control group had more seronegative patients requiring a salivary gland biopsy (SGB) for their pSS diagnosis [73/159 (45.9%) vs. 3/19 (15.8%), p = 0.014].

Central nervous system (CNS) symptoms consisted mainly of sensorimotor syndromes due to myelitis or visual impairment due to optic neuritis—both accompanied or not by active/non-active brain lesions in the MRI.

We analyzed the clinical phenotypes, serological profiles, and lesion patterns on whole brain and spinal MRI. Myelitis was the index event in 13 patients (13/19; 68.4%; Table 2A, Figure 2), optic neuritis occurred in five patients (5/19; 26.3%; Table 2A, Figure 3), and four patients had both in the further course of the disease (4/19; 21.1%; Table 2A). MRI scans at the index event revealed corresponding T2 hyperintensities in all patients with myelitis (13/13, 100%) and three patients with optic neuritis (3/5, 60%). Of patients with a history of optic neuritis, MRI revealed T2 hyperintensities in six patients (6/8, 75%), three extended throughout the optic nerve (3/8, 37.5%), two were located in the anterior part of the optic nerve (2/8, 25%), and two affecting both optic nerves simultaneously (2/8, 25%). Notably, 13 patients had signs of T2 hyperintensities on index brain MRI consistent with demyelination (13/19, 68.4%). In total, 14 out of 19 patients with CNS manifestation reported sensory deficits (14/19, 73.7%), five patients each had visual disturbances or bladder dysfunction (5/19, 26.3% each), and seven patients had motor deficits (7/19, 36.8%). Most patients accrued some degree of neurological disability. Eleven patients were presented to us initially during a clinical attack. Of these, eight patients were treated with steroids with four being responsive (4/8, 50%). Only one patient had recovered entirely from her symptoms (1/19, 5.3%); in all others, residual deficits remained. In three patients, the disease course was progressive (3/19, 15.8%).

We observed isolated myelitis or optic neuritis without further morphological correlation of demyelination on whole brain and whole spine MRI in five patients (5/19, 26.3%). Of these, one patient had already been diagnosed with pSS, and in the remaining four patients, initial diagnostic workup led to the detection of anti-SSA/Ro and/or anti-SSB/La and consequently to the diagnosis of pSS. Thus, isolated myelitis or optic neuritis was classified as a demyelinating event associated with pSS. This association turned out to be less straightforward in all other patients whose MRI scans revealed additional asymptomatic lesions that were disseminated in space and thus suggestive of chronic inflammatory CNS diseases such as MS or NMOSD.

A total of eight patients had been diagnosed with MS and were treated as such, and three more patients fulfilled the McDonald's criteria of 2017 (10). Notably, only 8 of these 11 patients (8/11, 72.7%) were positive for CSF-specific oligoclonal bands (OCB) with 13 of the total CNS group being OCB positive (13/18; 72.2%). Two patients were positive for anti-AQP4 antibodies (2/13, 15.4%) and diagnosed with NMOSD according to the IPND criteria of 2015 (11). Of 14 tested patients, none were positive for anti-MOG antibodies (0/14). Another three patients fulfilled the diagnostic criteria for AQP4 antibody-negative NMOSD. Both patients with anti-AQP4-antibody-positive NMO presented with an initial demyelinating event that left debilitating residual deficits. After initiating disease-modifying therapy based on CD20 depletion, further relapses could be prevented (Figure 4). The three other patients had a history of both myelitis and optic neuritis, which was rather mild compared with prognostic outcomes from observational NMOSD studies.

In all these cases, the diagnosis of pSS was finally made in the course of the disease either when the disease remained active under therapy (e.g., recurring optic neuritis) or when the diagnosis was facilitated by signs of systemic disease or elevated ANA titers. In 10 patients (10/19; 52.6%), symptoms suggestive of sicca syndrome did not occur until a median of 5.5 years (2.25–17) after the onset of neurological symptoms. Two patients reported Raynaud's phenomenon (2/19, 10.5%), and six patients had arthralgia (6/19, 31.6%). Five patients from the CNS group without subjective signs of sicca syndrome had a positive Schirmer's test (5/9, 55.6%), and in four of these five patients, salivary gland biopsy (SGB) disclosed pathognomonic lymphocytic infiltrates with a median Chisholm–Mason focus score of 4 (3, 4). Similar histological results after SGB could be obtained in the remaining CNS group (14/16, 87.5%).

A total of eight patients were started on disease-modifying antirheumatic drugs (DMARD) after pSS diagnosis, and ongoing immunomodulatory treatment was adjusted in three patients with prior MS diagnosis. Of note, five patients receiving first-line agents approved for MS had ongoing disease activity (Figure 4). Most successful was B-cell depletion with rituximab with a positive response (stable disease or improvement) in eight out of nine patients (8/9, 88.9%). Three patients (3/19, 15.8%) refused long-term immunosuppression, and two patients (2/19, 10.5%) did not require immunomodulatory therapy due to stable disease. Four patients required more than three drugs (4/19, 21%), and six patients deteriorated at least once in the course of the disease (6/19, 31.6%), requiring apheresis therapy as an acute rescue intervention. The median EDSS (extended disability status scale) was 2.0 (IQR 2.0–3.5) with a disease duration of 9 years (6–21.5) and a yearly attack rate of 0.33 (0.25–1.0; Table 2) indicating an overall disabling disease course with permanent deficits in our pSS patients with CNS manifestations.

While spinal cord lesions in MS are supposed to occur in the periphery of the spinal cord and usually extend over less than three vertebral segments, lesions in NMOSD and other systemic autoimmune may preferentially affect the center of the spinal cord and may exceed more than three vertebral segments (19). We took a pattern-based approach identifying the localization of T2-hyperintense lesions within the spinal cord of all patients from the CNS group and identified a similar pattern distribution as proposed by Goh et al. (19).

A total of three patients were devoid of T2 hyperintensities along with the spinal cord (3/19, 15.8%). Of the remaining patients with spinal cord lesions, eight had at least two or even more T2 hyperintensities (8/16, 50%), and we could identify five patients with T2 hyperintensities that exceeded more than three vertebral segments (5/16, 31.25%) fulfilling the criteria for longitudinal extensive transverse myelitis (LETM) (20). On sagittal images, lesions were distributed over the entire spinal cord without a distinct site of predilection. On axial images, T2 hyperintensities were distributed mainly in three patterns. We observed that lesions were localized at the periphery affecting the ascending and descending tracts, central lesions covering the gray matter and its surroundings, and lesions at the dorsal region involving the lemniscal tracts. Most lesions were localized in the central region (12 out of 27 lesions in total, 44.4%), followed by lesions in the periphery (10/27, 37%) and the dorsal region (5/27, 18.5%). The relative distribution of lesions in the cervical and thoracolumbar regions of the spinal cord is presented in Figure 5.