Memory Dysfunction in Primary Sjögren's Syndrome Is Associated With Anti-NR2 Antibodies

Authors

Errata

This article is corrected by:

  1. Errata: Incorrect Decapeptide Sequence in the Article by Lauvsnes et al (Arthritis Rheum, December 2013) Volume 66, Issue 4, 989, Article first published online: 28 March 2014

Abstract

Objective

Our understanding of the etiology and pathogenesis of neuropsychiatric involvement in primary Sjögren's syndrome (SS) is incomplete. In systemic lupus erythematosus, it has been reported that antibodies directed against N-methyl-D-aspartate receptor subtype NR2 (anti-NR2) interfere with memory and learning function, as well as mood. This has not been investigated in primary SS; however, the present study was undertaken to advance our understanding of neuropsychiatric involvement in this disease.

Methods

Sixty-six patients with primary SS and 66 age- and sex-matched healthy control subjects underwent clinical examination and neuropsychological evaluation. Anti-NR2 antibodies were measured in serum and cerebrospinal fluid. Hippocampus volume was estimated using software extensions to SPM5.

Results

Patients with primary SS had smaller hippocampi than healthy subjects (mean ± SD 8.15 ± 0.98 cm3 versus 8.49 ± 0.88 cm3; P = 0.01). In patients with primary SS, anti-NR2 antibodies in cerebrospinal fluid were associated with a worse performance in 8 of 10 memory and learning tests, and anti-NR2 antibodies in serum were associated with a worse performance in 6 of those same tests. In addition, a higher proportion of patients with depression than patients without depression had serum anti-NR2 antibody levels above the cutoff value.

Conclusion

Results of this study indicate that anti-NR2 antibodies may represent one of the pathogenetic mechanisms for cognitive disturbances and mood disorders in patients with primary SS.

Primary Sjögren's syndrome (SS) is a systemic autoimmune disease characterized by chronic inflammation of exocrine glands, often leading to dryness of the mouth and eyes. Extraglandular manifestations are commonly reported, including central nervous system involvement such as headaches, mood disorders, and cognitive dysfunction ([1, 2]). However, the use of different criteria over time to classify patients as having primary SS, the use of data from investigations performed in different patient populations, such as tertiary hospital center cohorts as compared to more population-based cohorts, and the lack of criteria for neurologic involvement make comparison difficult and result in large differences in reported prevalence ([3]). In a recent population-based study ([1]) in which American–European Consensus Group (AECG) criteria ([4]) were used for classification of primary SS, we found that 50% of patients displayed cognitive dysfunction.

There is growing evidence that anti-NR2 antibodies may interfere with distinct cerebral functions in systemic lupus erythematosus (SLE) ([5-10]). These antibodies are directed against the NR2 subtype of N-methyl-D-aspartate (NMDA) receptors, which are considered important for memory, learning, and mood ([11-13]). NR2 receptors are distributed throughout the brain, but they occur in particularly high density in the hippocampus, a brain structure that is extremely important for learning and memory formation ([13]). Studies in animal models of SLE have demonstrated that anti-NR2 antibodies bind NR2 receptors in a dose-dependent manner and cause prolonged neuronal excitation, leading to neuronal dysfunction and eventually neuronal death due to glutamate excitotoxicity ([14, 15]). Subsequently, these mice display cognitive impairment and hippocampus atrophy, as well as altered emotional behavior and degeneration of amygdalar neurons ([16-18]).

Memory impairment is one of the cognitive dysfunctions that has been consistently reported in patients with SLE ([19-21]). Hippocampal atrophy in SLE patients has recently been found to be associated with lower scores on general memory, verbal memory, and delayed recall tests ([21]). In a previous study, a proportion of SLE patients with anti-NR2 antibodies exhibited relevant cognitive disturbances such as memory dysfunction and depression ([5]). The influence of anti-NR2 antibodies on memory function in patients with primary SS has not been investigated, and volumetric cerebral magnetic resonance imaging (MRI) studies are rare. For further exploration of these matters, a composite approach that includes immunologic, structural, and functional measures is preferred. As part of a larger study of primary SS, we focused on data related to memory tests, hippocampus size, and anti-NR2 antibodies in blood and cerebrospinal fluid (CSF).

PATIENTS AND METHODS

Patients with primary SS and healthy subjects

All patients with primary SS were recruited from Stavanger University Hospital inpatient and outpatient clinics. The only hospital in the southern part of Rogaland county in Norway, Stavanger University Hospital serves 330,000 inhabitants. Catchment and inclusion procedures have been previously described ([22]). Of the 99 patients who fulfilled the AECG criteria for primary SS ([4]), 72 (73%) provided informed consent. A further 6 patients (6%) were excluded: 2 were ineligible for MRI due to claustrophobia or cochlear implant, and 4 were excluded after visual inspection of the MRI scans revealed poor image quality. The analyzed population thus consisted of 66 patients: 57 women (86.4%) and 9 men (13.6%). The mean ± SD age was 56.8 ± 12.5 years.

Age within 2 years– and sex-matched healthy subjects were recruited through unrelated friends and neighbors of the patients and hospital staff. For the 66 control subjects, the mean ± SD age was 56.2 ± 12.5 years. All patients and healthy subjects were white.

For research purposes only, patients with primary SS and healthy subjects were admitted to the hospital and received a standardized examination by specialists in neuropsychology (SSM), internal medicine (EH), and neurology (ABT) over a 2-day period. MRI was performed within a median of 12 days after the examination (range 0–33 days) for the patients and 22 days (range 0–81 days) for the healthy subjects. Lumbar puncture was performed in all patients with primary SS except for 15 who refused the procedure. None of the healthy control subjects underwent lumbar puncture.

To establish reference values, CSF was also obtained from 24 patients at Stavanger University Hospital who underwent lumbar puncture as part of a neurologic examination and later did not display any inflammatory, autoimmune, or malignant disease. This study was approved by the regional research ethics committee and was carried out in compliance with the Declaration of Helsinki.

Cerebral MRI

Image acquisition

A 1.5T Gyroscan NT Intera Release 10 scanner (Philips Medical Systems) was used for image acquisition. The MRI protocol has been previously described ([22]).

Volumetric analysis

We use axial T1 3-dimensional turbo field-echo with a 1.6-mm slice thickness. The images were fully automatically preprocessed using the VBM5 extension (Gaser; online at http://dbm.neuro.uni-jena.de/vbm/download/) to SPM5 software (online at http://www.fil.ion.ucl.ac.uk/spm/software/spm5/). First, we created template images based on our healthy control group. All images were then spatially normalized to the same stereotactic space. Normalization and segmentation were performed using default settings and tested by viewing one normalized unsegmented slice for all patients and controls and evaluating a covariance matrix of the covariance among all volumes. Grey matter and white matter were smoothed with an 11-mm full-width half-maximum Gaussian Kernel. We used the MarsBar volume-of-interest analysis toolbox (online at www.marsbar.sourceforge.net) for SPM5 and selected right and left hippocampus region of interest (online at www.gin.cnrs.fr/spip.php?article217).

Neuropsychological testing

All 66 patients completed neuropsychological testing that began at a fixed time in the morning; the tests were administered by a trained psychometric test technician and were completed within 3–4 hours. All tests were administered in a standardized manner, and the results were analyzed by a clinical neuropsychologist. Data from memory and learning tests were selected for this specific study. Measures were treated as continuous variables, and raw scores were used in subsequent analyses.

Scores from the Wechsler Memory Scale, Revised (WMS-R) ([23]) and the Tactual Performance Test (TPT) ([24]) were obtained. We used the sum of the weighed scores of 4 indices from the WMS-R (verbal memory, visual memory, general memory, and delayed memory). WMS-R verbal memory and WMS-R visual memory tests provide composite measures of the immediate retrieval of verbal and visual material, respectively. The WMS-R general memory test provides a composite measure of the immediate retrieval of both verbal and visual material, and the WMS-R delayed memory test provides a composite measure of delayed retrieval of the same material after a 30-minute interval. The TPT includes several cognitive domains ([25]), and performance is reported as the time to complete an assigned task using the dominant hand, the nondominant hand, both hands together, and the total time to complete all 3 tests. In relation to the baseline performance of the dominant hand, the TPT nondominant hand test, TPT test for both hands together, and TPT total time test can be considered as tests that measure learning. The TPT location test and TPT memory test measure incidental tactile and spatial learning and retrieval, respectively.

Mental depression

The Beck Depression Inventory (BDI) was used to evaluate mood. A score of ≥13 was defined as reflecting clinical depression ([26, 27]).

Laboratory tests

Routine hematologic and biochemical tests were analyzed in the hospital's laboratory. Screening for anti-SSA and anti-SSB antibodies was performed with a Quanta Lite ENA 6 assay, and positive results were confirmed by Quanta Lite SSA and SSB enzyme-linked immunosorbent assay (ELISA) (both from Inova Diagnostics). Screening to detect IgM and IgG anticardiolipin antibodies was performed with the Quanta Lite ACA IgM and IgG ELISA (Inova Diagnostics).

Anti-NR2 antibodies in blood

Serum anti-NR2 antibodies were analyzed using a previously described ELISA-based technique ([5]), with minor modifications. Briefly, 96-well microtiter plates (Greiner Bio-One) were each coated with 100 μl of synthetic decapeptide H2N DWEYSVWLS (1 μg/ml). The plates were left overnight at 4°C and blocked with 10% fetal calf serum in phosphate buffered saline (PBS) for 1 hour. The serum samples were diluted 1:50 with 10% fetal calf serum in PBS. The diluted samples were added in triplicate and incubated for 2 hours before bound antibodies were detected using Polyvalent Anti-Human Ig (G, A, M) Peroxidase Conjugate (Sigma-Aldrich). O-phenylenediamine dihydrochloride (DakoCytomation) was used as the detection substrate, and after 30 minutes of incubation, the reaction was stopped by adding 100 μl of 1M H2SO4. Optical density (OD) was read at 492 nm with a Multiskan Ascent microplate photometer (Thermo Scientific). Except for coating, all steps were conducted at room temperature, and the plates were washed with PBS 4 times after each incubation.

Anti-NR2 antibody levels were determined in 95 healthy blood donors, and the cutoff for positivity was initially set at 3 SD above the mean value in the group, i.e., 0.82. In one patient, the OD was close to this value (0.86), and thus this patient's serum was analyzed on all plates and this measurement was used as the cutoff value. The ratio against this cutoff was calculated for all samples; a ratio of ≥1.0 was considered positive and a ratio of <1.0 was considered negative for the presence of anti-NR2 antibodies.

Anti-NR2 antibodies in CSF

Anti-NR2 antibodies in CSF were analyzed by electrochemoluminescence on a Sector Imager 2400 platform (Meso Scale Discovery). A high bind plate (L15XB-3; Meso Scale Discovery) was coated with 25 μl of synthetic decapeptide (H2N DWEYSVWLS) at a concentration of 2 μg/ml and incubated overnight at 4°C. The plate was blocked with 150 μl of 3% bovine serum albumin (MSD Blocker A; Meso Scale Discovery) for 1 hour. Then, 25 μl of each sample was added in duplicate to the wells and incubated for 2 hours. Twenty-five microliters (1 μg/ml) of anti-human antibody (goat) with Sulfo-Tag (Meso Scale Discovery) was added and incubated for 1 hour. Read Buffer T 2× (150 μl; Meso Scale Discovery) was added and results were read on a Sector Imager 2400 (Meso Scale Discovery). With the exception of coating, all incubation steps were conducted at room temperature on a plate shaker (300–400 revolutions per minute), and the plate was thereafter washed 3 times with PBS (pH 7.2–7.3) plus 0.05% Tween 20.

The CSF sample (from the 24 Stavanger University Hospital patients) with the highest signal (a signal slightly higher than 3 SD above the mean of all samples) was chosen as the cutoff value/internal calibrator and measured together with the samples on each plate. For each sample, the ratio against the internal calibrator was calculated. Samples with a ratio ≥1.0 were considered positive, and samples with a ratio <1.0 were considered negative for the presence of anti-NR2 antibody.

Statistical analysis

For normally distributed variables, data were reported as the mean ± SD. Hippocampus volumes in patients and matched healthy subjects were compared using paired-sample t-tests; otherwise, normally distributed variables were compared using independent-sample t-tests. Non-normally distributed data were reported as the median and range and were analyzed using the Mann-Whitney U test. Categorical variables were analyzed using the chi-square test.

Raw scores from the neuropsychological tests were compared to the normative scores, yielding so-called standard scores for the WMS-R tests and T scores for the TPT ([23, 24]). For scale-free comparison, these scores were further transformed into Z scores, which we calculated using a ComputerPsych Online Normal Distribution Curve Calculator ([28]).

When establishing models for memory function and for hippocampus size, potential explanatory variables were tested by univariable linear regression to generate unadjusted effect estimates. Based on results from univariable regression analyses, multivariable regression analyses were performed for memory tests and for hippocampus volumes. Variables with P values of ≤0.25 in univariable analyses and variables thought to have clinical importance independent of their P values were added in the multivariable models. Variables that lacked significant effect as determined by multivariable regression analysis were excluded from the final models, with the exceptions of age, sex, disease duration, and anti-NR2 antibodies, which were regarded as important variables regardless of their significance levels. In addition, education level and hippocampus volume were always included in the final models of memory function. P values less than or equal to 0.05 were considered significant. Analyses were performed with IBM SPSS Statistics 20.

RESULTS

Selected patient characteristics are summarized in Table 1. Twenty-five patients (38%) were receiving antimalarials, 4 patients (6%) were receiving azathioprine, 4 patients (6%) were receiving antidepressants, and 2 patients (3%) were receiving beta-blockers, while 6 patients (9%) were taking sleeping pills when needed.

Table 1. Clinical characteristics of the patients with primary Sjögren's syndrome (n = 66)
  1. * Except where indicated otherwise, values are the number (%). Total number of patients examined is shown for variables for which data were not available for all 66 patients. MSG = minor salivary gland; ANA = antinuclear antibody; CSF = cerebrospinal fluid; aPL = antiphospholipid antibody; BDI = Beck Depression Inventory; AECG = American–European Consensus Group.
MSG focus score ≥152/65 (80)
ANA56 (90)
Anti-SSA antibody51/57 (86)
Anti-SSB antibody30/57 (53)
Anti-NR2 antibody in serum13 (20)
Anti-NR2 antibody in CSF6/51 (12)
aPL8 (12)
Present use of prednisolone15 (23)
BDI score ≥1320 (30)
Disease duration, mean ± SD years7.2 ± 5.0
No. of AECG criteria fulfilled, median (range)5 (3–6)

Imaging

MRIs of 1 healthy subject were missing; the MRI data from the corresponding patient with primary SS were therefore excluded from comparisons between patients and healthy subjects, but were included in the regression analyses of the associations between hippocampus volume and clinical characteristics of the patients with primary SS. One of the patients and 3 of the healthy subjects (P = 0.32) exhibited lacunar infarcts on MRI, while cortical infarctions were not seen on MRIs of either the patients or the healthy controls. Hippocampus volumes that were corrected for total intracerebral volumes were lower in the patient group than in the control group (n = 65 for both) (mean ± SD 8.15 ± 0.98 cm3 versus 8.49 ± 0.88 cm3; P = 0.01).

Memory and learning tests and depression

Figure 1 shows the 95% confidence intervals of the Z scores for the memory and learning tests taken by the patients with primary SS. For each test, the mean Z score of the patients with primary SS as a group was within 1 SD of the expected scores, but for each individual test there were some patients with impairment. There was a trend toward higher-than-expected scores in the WMS-R tests and lower scores in the TPT (Figure 1). Patients had a median BDI score of 9 (range 0–38). In healthy subjects, this value was 2 (range 0–16) (P < 0.001).

Figure 1.

Mean Z scores and 95% confidence intervals for the memory and learning tests in the 66 patients with primary Sjögren's syndrome. WMS-R = Wechsler Memory Scale, Revised; TPT = Tactual Performance Test.

Anti-NR2 antibodies

The median ratio of anti-NR2 antibodies in serum was 0.66 in patients with primary SS (range 0.27–4.6) and 0.71 in healthy subjects (range 0.29–3.2) (P = 0.99). In 13 patients (20%) and 7 healthy subjects (11%), serum anti-NR2 antibody levels were above the cutoff value (P = 0.15). In CSF samples, the median ratio of anti-NR2 antibodies in patients was 0.41 (range 0.15–3.0). In CSF from 6 patients, anti-NR2 antibody levels were above the cutoff value; in 2 of these patients, the cutoff value was exceeded in serum as well. Five of the 6 patients with primary SS with elevated CSF anti-NR2 antibodies had an increased IgG index; in the patient who had a normal IgG index, the anti-NR2 antibody ratio in CSF was lowest of these 6 patients. Three patients with primary SS had an increased albumin CSF/serum index, indicating a disrupted blood–brain barrier. One of these patients had a serum anti-NR2 antibody level above the cutoff value, but anti-NR2 antibodies in CSF were not above the cutoff value in any of them. In patients taking antimalarials (hydroxychloroquine), the number of patients with serum anti-NR2 antibodies was lower than expected (by chi-square test, 1 observed versus 5 expected; P = 0.01).

Associations between hippocampus volume and other variables

Univariable associations between hippocampus volume and selected important clinical and laboratory variables are summarized in Table 2. Patients with CSF anti-NR2 antibodies above the cutoff value had lower total hippocampus volumes than those whose CSF anti-NR2 antibodies were below the cutoff value (mean ± SD 7.43 ± 0.71 cm3 and 8.37 ± 0.92 cm3, respectively; P = 0.02). There was no difference in volume between patients with primary SS with serum anti-NR2 antibodies and those without serum anti-NR2 antibodies (data not shown). Also, as expected, increasing age and female sex were associated with lower hippocampus volume ([29]).

Table 2. Associations between hippocampus volume and selected clinical and laboratory variables in univariable regression models in the 66 patients with primary Sjögren's syndrome*
 βR2P
  1. aPL = antiphospholipid antibody; CSF = cerebrospinal fluid.
  2. an = 51 patients.
Age−0.020.070.02
Sex1.30.22<0.01
Disease duration−0.030.030.16
Anti-SSA antibodies−0.50.030.17
aPL0.01<0.010.98
Anti-NR2 antibodies in CSFa−0.90.10.02
Present use of prednisolone−0.20.010.43

Multivariable regression revealed that anti-NR2 antibodies in CSF did not influence hippocampus volume (P = 0.23) (Table 3). The hippocampus sizes in patients with primary SS were equal whether the patient had a BDI score of ≥13 or <13 (data not shown).

Table 3. Final multivariable regression model of the association between hippocampus volume and selected clinical and laboratory variables in 51 patients with primary Sjögren's syndrome*
 βP
  1. Adjusted R2 = 0.46. Anti-NR2 antibodies were measured in cerebrospinal fluid for this analysis.
Age−0.04<0.01
Sex1.3<0.01
Disease duration0.030.15
Anti-NR2 antibodies−0.400.23

Associations between results of memory tests and other variables

Two sets of multivariable models for memory and learning function were developed; in one set, anti-NR2 antibody levels in CSF were included as a predictor, and in the other set, anti-NR2 antibody levels in serum were included instead (Tables 4 and 5, respectively).

Table 4. Associations between raw scores on neuropsychological tests and selected clinical and laboratory variables in 51 patients with primary Sjögren's syndrome*
 WMS-R verbalWMS-R visualWMS-R generalWMS-R delayedTPT dominant hand timeTPT nondominant hand timeTPT both hands timeTPT total timeTPT memoryTPT location
  1. Multivariable regression analysis included anti-NR2 antibody levels in cerebrospinal fluid as a variable. For Tactual Performance Test (TPT) dominant hand time, nondominant hand time, both hands time, and total time, higher raw scores indicate poorer function; for the remaining TPT tests and for Wechsler Memory Scale, Revised (WMS-R) tests, higher raw scores indicate better function. aPL = antiphospholipid antibodies.
Age          
β−0.43−0.51−0.94−0.730.050.020.030.11−0.01−0.02
P0.04<0.01<0.01<0.010.570.680.350.270.690.45
Sex          
β−8.0−1.2−9.1−6.21.00.50−0.211.20.33−0.03
P0.210.740.300.380.290.730.840.730.750.97
Education          
β0.980.501.91.2−0.15−0.29−0.14−0.570.050.05
P0.110.130.020.070.310.020.130.060.600.54
Disease duration          
β0.140.350.890.81<0.010.01−0.04−0.020.050.08
P0.740.120.120.070.970.930.550.930.430.16
Hippocamus volume          
β−0.101.20.212.40.05−0.430.39−0.620.410.36
P0.970.430.960.430.940.470.370.660.330.36
Anti-SSA antibodies          
β15.7
P0.01
aPL          
β−10.1−15.81.6
P0.050.030.05
Anti-NR2 antibodies          
β−8.3−4.0−11.6−11.33.11.02.86.9−1.6−1.2
P0.040.080.040.01<0.010.24<0.01<0.010.010.04
Adjusted R20.440.530.490.480.250.160.480.360.170.14
Table 5. Associations between raw scores of neuropsychological tests and selected clinical and laboratory variables in the 66 patients with primary Sjögren's syndrome*
 WMS-R verbalWMS-R visualWMS-R generalWMS-R delayedTPT dominant hand timeTPT nondominant hand timeTPT both hands timeTPT total timeTPT memoryTPT location
  1. Multivariable regression analysis included anti-NR2 antibody levels in serum as a variable. For Tactual Performance Test (TPT) dominant hand time, nondominant hand time, both hand time, and total time, higher raw scores indicate poorer function; for the remaining TPT tests and for Wechsler Memory Scale, Revised (WMS-R) tests, higher raw scores indicate better function.
Age          
β−0.24−0.39−0.54−0.400.030.020.060.11−0.02−0.03
P0.12<0.010.010.020.300.460.040.110.470.17
Sex          
β−16.9−8.0−26.1−21.02.30.05−0.071.9−1.0−0.73
P<0.01<0.01<0.01<0.010.110.970.950.520.270.49
Education          
β0.740.501.60.87−0.06−0.23−0.10−0.35−0.02−0.06
P0.200.070.040.150.580.050.290.170.780.47
Disease duration          
β−0.230.220.110.36−0.09<−0.01−0.05−0.230.070.10
P0.520.200.820.330.220.960.390.170.170.10
Hippocamus volume          
β4.23.88.48.8−0.37−0.14−0.41−0.720.580.41
P0.07<0.01<0.01<0.010.460.770.320.510.070.28
Anti-SSA antibodies          
β14.8
P0.02
Anti-NR2 antibodies          
β−3.5−3.2−6.6−6.62.30.131.13.8−1.2−0.49
P0.180.010.070.02<0.010.790.01<0.01<0.010.22
Adjusted R20.320.590.420.440.290.040.250.270.190.02

Anti-NR2 antibodies

In the models using anti-NR2 antibodies in CSF, data from one patient's TPT location test were censored due to a large standardized residual (>3.0). A considerable and consistent explanatory effect appeared for all WMS-R tests, with an adjusted R2 that ranged from 0.44 to 0.53. The adjusted R2 for the TPT with both hands together was 0.48, and the adjusted R2 for TPT total time was 0.36; otherwise, the adjusted R2 values for the TPTs were 0.14–0.25 (Table 4). Anti-NR2 antibodies were associated with worse performance in 8 of 10 memory and learning tests, with the exceptions of WMS-R visual memory (P = 0.08) and TPT nondominant hand time (P = 0.24). The associations between the scores of the 10 memory and learning tests and the clinical and laboratory variables, including anti-NR2 antibodies in CSF, are shown in Table 4.

When developing multivariable models with anti-NR2 antibodies in serum as a predictor, one patient's data were censored due to standardized residuals of >3.0 in WMS-R visual memory, TPT dominant hand time, and TPT total time. Adjusted R2 values for the WMS-R tests were 0.32–0.59. For the TPTs, the adjusted R2 value was as low as 0.02 and 0.04 for TPT location and TPT nondominant hand time, respectively; otherwise, these values ranged between 0.19 and 0.29. The presence of anti-NR2 antibodies in serum had a negative influence on WMS-R visual memory and WMS-R delayed memory scores in multivariable models, and on all of the TPTs, except TPT nondominant hand time and TPT location (Table 5). The results of multivariable models for association between memory and learning test scores and anti-NR2 antibodies in serum are given in Table 5.

Hippocampus volume

In univariable analyses, larger hippocampus volume was associated with better function in all WMS-R tests, except for WMS-R verbal memory test, and in TPT time for both hands together (data not shown). However, in the multivariable analyses using anti-NR2 antibodies in CSF, the association of larger hippocampus volume with test results was no longer significant (Table 4). In the multivariable models using anti-NR2 antibodies in serum, increasing hippocampus volume was associated with better performance in WMS-R visual memory, WMS-R general memory, and WMS-R delayed memory tests (Table 5).

Anti-SSA antibodies

The presence of anti-SSA antibodies was associated with higher WMS-R verbal memory scores in univariable analyses as well as in both of the multivariable regression models (Tables 4 and 5). However, this effect is hard to explain and is probably coincidental, as the explanatory effect in the univariable model was small (R2 < 0.01). The presence of anti-SSA antibodies was not associated with changes in the scores of the other neuropsychological tests.

Antiphospholipid antibodies (aPL).

The presence of aPL was associated with worse performance on the WMS-R verbal memory test, WMS-R general memory test, and in TPT time for both hands together in the model using anti-NR2 antibodies in CSF (Table 4), but not in the model using anti-NR2 antibodies in serum (Table 5).

Drugs administered for primary SS

In univariable regression analyses, the use of antimalarials was associated with a significant or close-to-significant beneficial effect on performance in several of the WMS-R tests, i.e., WMS-R verbal memory (β = 0.02, R2 = 0.05, P = 0.07), WMS-R visual memory (β = 0.01, R2 = 0.06, P = 0.05), WMS-R general memory (β = 0.03, R2 = 0.07, P = 0.04), and WMS-R delayed memory (β = 0.03, R2 = 0.05, P = 0.06). No effect of antimalarials was observed in these tests when multivariable regression analyses were applied.

Mental depression

Neuropsychological test scores showed no difference between patients with BDI scores of ≥13 and patients with BDI scores of <13 (data not shown).

Mental depression and anti-NR2 antibodies

Among the 20 patients with BDI scores of ≥13, serum anti-NR2 antibody levels were above the cutoff value in 7 (35%); 6 of 46 patients with BDI scores of <13 (13%) had anti-NR2 antibody levels above the cutoff value (P = 0.04). Among patients with anti-NR2 antibodies in CSF, no difference was found between patients with depression and those without depression (P = 0.91). Drugs administered for primary SS did not influence BDI scores.

DISCUSSION

The important finding of this study is that impaired memory and learning, as well as depression, occurred more frequently in patients with primary SS with anti-NR2 antibodies than in those patients who did not have anti-NR2 antibodies. In addition, as a group, patients with primary SS had smaller hippocampus volumes than age- and sex-matched healthy subjects. This indicates that anti-NR2 antibodies may influence cognitive function in primary SS, a hypothesis that has been proposed for SLE ([5, 16, 17]). Our findings support the theory that the antibodies interfere with the NR2 subtype of the NMDA receptors on neurons mainly located in the hippocampus, as has been observed in animal models of SLE ([14, 15]). This process is accompanied by a functional decline in memory ([15-17]). Whether this decline is due to neuronal loss only, to disturbed neuronal function via anti-NR2 antibody interference with NMDA receptors, or to a combination of these mechanisms cannot be deduced from these results, but the weak association between anti-NR2 antibodies and hippocampus volume suggests that neuronal death may not dominate in this context.

Our observations extend the group of diseases in which anti-NR2 antibodies seem to play a pathogenetic role ([5-10, 30, 31]). One previous study revealed that anti-NR2 antibodies were present in patients with myasthenia gravis and autoimmune polyendocrine syndrome, although occurrence of the anti-NR2 antibodies was considerably less frequent in those patients than in patients with SLE ([32]). To the best of our knowledge, however, there are no studies of other diseases or conditions that would allow conclusions to be drawn as to whether anti-NR2 antibodies could represent a more global mechanism for cognitive impairment or cerebral dysfunctions in autoimmune diseases. Such investigations have yet to be performed.

We detected clear associations between serum anti-NR2 antibodies and impaired performance in several memory and learning tests, as well as between depression and serum anti-NR2 antibodies. An association of serum anti-NR2 antibodies with cognitive dysfunction and depression has been found in some human SLE studies ([5, 6]), although findings have not been consistent ([33-38]). It has been hypothesized, mainly based on experiments in murine models, that anti-NR2 antibodies are produced extrathecally and must cross the blood–brain barrier into the CSF to be able to influence brain function ([17]). It is possible that repeated infections allow anti-NR2 antibodies to access the brain through a leaky blood–brain barrier; alternatively, it is possible that chronic inflammation due to the autoimmune disease opens the blood–brain barrier. Recently, it has been demonstrated that anti-NR2 antibodies from SLE patients interact with endothelial cells, and thus could represent a “specific” mechanism for the passage of these antibodies ([39]).

Conversely, anti-NR2 antibodies could originate inside the brain. Activated T cells are able to pass the blood–brain barrier ([40]); over long periods of time, anti–NR2-specific T cells may activate anti–NR2-specific B cells that already reside intrathecally to produce anti-NR2 antibodies. The possibility of such a mechanism is supported by our observation that anti-NR2 antibodies were present in the CSF of 5 of 6 patients with an increased IgG index. Also, considerably increased concentrations of BAFF and APRIL have been found in the CSF of patients with SLE ([41]). These molecules are essential for B cell survival and function, and might well play a role in sustained autoantibody production in the central nervous system.

Perhaps the avidity and fine specificity of the anti-NR2 antibodies are also of importance for the biologic activity induced in neurons. This hypothesis is supported by our observation that there was not a universal association between presence of anti-NR2 antibodies and the results of all the memory tests that we performed. In this context, it is highly likely that more than 1 mechanism for neuronal disturbance contributes to cognitive dysfunction in primary SS, and it is possible that these mechanisms operate simultaneously in the individual patient. Sustained high levels of aPL have previously been associated with cognitive deficits in patients with SLE ([42, 43]). Such a mechanism could be operative in our patients as well, since only 12% of the patients with primary SS had aPL, but aPL nevertheless had a significant impact on cognitive function.

Interestingly, administration of antimalarials seemed to be associated with better performance, and anti-NR2 antibodies were detected less frequently in these patients. Whether this is a relevant finding, a bias, or a statistical coincidence remains to be seen in further studies.

Our study has several limitations. Healthy subjects did not undergo the neuropsychological tests that were administered to the patients, and thus we could not compare the performance of the patients with primary SS with the age- and sex-matched subjects. We therefore had to make use of the normative values to illustrate the cognitive performance of our patients. Also, we chose to focus on neuropsychological functions previously reported to be affected by anti-NR2 antibodies, which allowed us to minimize problems associated with multiple significance testing. Inclusion of a wider battery of tests and other psychometric variables or neurologic manifestations may reveal associations with other unexpected clinical phenomena. The cross-sectional design of this study allowed us to detect associations only; a future longitudinal approach would yield more precise information about the affected functions.

Including a larger number of patients with primary SS could have been valuable, but it is difficult to recruit patients for such extensive and invasive research purposes. Only 6 patients were found to have anti-NR2 antibodies above the cutoff value in CSF. Further, the cognitive performance of patients with primary SS was close to that expected, although some patients clearly exhibited impairment. These factors may have interfered with real associations that we were not able to detect due to low power.

However, our study was strengthened by the use of an unselected group of well-classified patients with primary SS who underwent comprehensive and standardized systematic evaluation. The use of age- and sex-matched healthy subjects and the measurement of anti-NR2 antibody levels in both serum and CSF also strengthened this study. The analytical cutoff level for anti-NR2 antibodies in sera and in CSF was based on large samples.

In conclusion, our important findings indicate that anti-NR2 antibodies may serve as a general mechanism for cerebral dysfunction in autoimmune diseases, not just in SLE.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Lauvsnes had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Lauvsnes, Maroni, Appenzeller, Greve, Harboe, Omdal.

Acquisition of data. Maroni, Beyer, Greve, Harboe, Gøransson, Tjensvoll, Omdal.

Analysis and interpretation of data. Lauvsnes, Maroni, Appenzeller, Beyer, Greve, Kvaløy, Gøransson, Omdal.

Acknowledgments

We thank Tone Berit Heng and Ingeborg Kvivik for performing the analyses of anti-NR2 antibodies in serum and cerebrospinal fluid.

Ancillary