Characteristics of cerebrospinal fluid cytology in anti-N-methyl-D-aspartate receptor encephalitis

Background The aim of the current study was to explore the characteristics of cerebrospinal fluid (CSF) cytology in anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis. Methods CSF was collected from patients with anti-NMDAR encephalitis at and from 01 January 2015 to 31 December 2018, and cytological characteristics and other parameters of the CSF were analyzed. Results CSF cytological data were obtained from 164 patients with anti-NMDAR encephalitis. Visible signs of inflammation were identified in cytological analyses of 112 patients’ CSF, including 46 cases of mild inflammation, 58 of moderate inflammation, and 8 of severe inflammation. With regard to inflammation type, 89 cases were classified as lymphocytic inflammation, 22 as mixed inflammation with both lymphocytes and neutrophils, and 1 case was classified as mixed inflammation with lymphocytes, neutrophils, and eosinophils. Activated lymphocytes were detected in 51 patients, and plasma cells were detected in 16. Oligoclonal bands were detected in 111/164 patients, and 51 patients were positive for specific oligoclonal bands in CSF. The positivity rate was 45.9%. In comparison with routine CSF examination with an inflammation positivity threshold of > 5 × 10 6 cells/L, the inflammation positivity rate of CSF cytology was significantly higher (68.3% versus 64.0%, p < 0.001). Conclusions The comparatively higher inflammation positivity rate and the superiority of evaluating cell morphology render CSF cytology an effective tool for identifying anti-NMDAR encephalitis and investigating its pathogenesis.

immunodeficiency virus infection, meningitis, brain abscess, prion disease, cerebral malaria, brain tumor, diagnosis of a noninfectious central nervous system disease such as acute demyelinating encephalomyelitis, or laboratory evidence of infectious encephalitis, e.g., viral, bacterial, mycobacterium tuberculosis-associated, parasitic, or fungal [3].
General patient data and the results of routine CSF examination, biochemical tests, cytological investigations, and OCB tests were collated. All CSF data were obtained from patients who had acute or recurrent anti-NMDAR encephalitis. The modified Rankin Scale

Determination of antibodies to NMDAR
Lumbar punctures of all patients were performed within 3 days after they were admitted to the hospital, and serum and CSF samples were obtained simultaneously. Anti-NMDAR antibodies were evaluated using the indirect immunofluorescence test kit autoimmune encephalitis mosaic 1 (catalog number FA 112d-1, Euroimmun Ag, Germany). Samples were considered positive if three immunohistochemical criteria were met [1].

CSF cytology
For cytological examination slides were prepared using a spontaneous sedimentation chamber (Sayk's sedimentation technique) with 0.5 mL CSF. First, 0.5 mL CSF was put into the spontaneous sedimentation chamber and stored overnight at 4℃, during which time the cells precipitated onto the slide. The slide was then air-dried and subjected to May-Grunwald-Giemsa staining. The morphology, classification, and numbers of white cells were determined under a light microscope [4,5]. On the basis of cell counts determined via microscopy, degree of inflammation was divided into four levels: normal, ≤ 200 cells/0.5 mL; mild, 201-500 cells/0.5 mL; moderate, 501-2000 cells/0.5 mL; and severe, > 2000 cells/0.5 mL. The cell collection rate via natural sedimentation is approximately 15%.
The proportions of various cell types present were calculated, and based on these proportions inflammation was classified as lymphocytic, lymphocytic and neutrophilic, or lymphocytic, neutrophilic, and eosinophilic. Levels of activated lymphocytes and plasma cells were also evaluated.

Other CSF examinations
Intracranial pressure was evaluated using a CSF pressure gauge, and pressure > 180 mmH 2 O was considered "increased". Integrated analyses included routine CSF examinations, biochemical CSF tests, and OCBs in both CSF and serum. Abnormally elevated cell counts were defined as total cell counts > 5 × 10 6 /L without erythrocytosis, and CSF protein > 0.45 g/L was considered abnormally elevated. OCBs that were present in CSF but not in serum were considered specific OCBs, which are markers of autochthone intrathecal IgG production.

Statistical analyses
SPSS 21 statistical software was used to analyze the general patient data and the characteristics of CSF. Continuous values are expressed as means ± the standard deviation (SD) or medians with 25 th and 75 th percentiles, and dichotomous values are expressed as numbers and percentages. Correlation analyses were conducted to evaluate associations between mRS score and inflammation parameters such as white blood cell count determined via routine CSF examination, degree of inflammation determined via routine CSF examination, white cell count determined via CSF cytology, and degree of inflammation determined via CSF cytology. Correlations between routine CSF examination results and CSF cytology results were also investigated, as were associations between plasma cells and specific OCBs. Spearman's rank correlational analysis was used to assess relationships between two continuous variables in the case of non-normally distributed data, the χ2 test was used to analyze associations between two categorical variables, and the kappa test was used for consistency analysis. Statistical significance was set at p < 0.05.

General data
One hundred and sixty-four patients with anti-NMDAR encephalitis were enrolled in the study. Of them, 95 were from Peking Union Medical College Hospital and 69 were from Henan Provincial People's Hospital. Ninety-four (57.3%) were female and 70 (42.7%) were male, yielding a female:male ratio of 1.34:1.00. The mean age was 25.85 ± 12.62 years (range 2-65 years), and the median mRS score was 3 (range 2-5).

CSF cytology
All 164 patients underwent CSF cytology examination (Table 1), and lymphocytes and monocytes were detected in all samples (100%). Neutrophils were detected in 23/164 (14.0%) samples, and eosinophils were detected in 1/164 (0.6%). Activated lymphocytes were detected in 51/164 (31.1%) samples and plasma cells were detected in 16/164 (9.8%). Inflammatory reactions were evident in samples from 112/164 (68.3%) patients, of which 46 were deemed mild, 58 moderate, and 8 severe. With regard to inflammation type, 89 patients exhibited lymphocytic inflammation, 22 exhibited lymphocytic and neutrophilic mixed inflammation, and 1 exhibited lymphocytic, neutrophilic, and eosinophilic mixed inflammation. The proportion of neutrophils in the total number of leukocytes ranged from 1% to 80%. In 11 patients the proportion of neutrophils was < 10%, in 8 patients it was between 10% and 50%, and in 3 patients it exceeded 50%. In the 1 patient with lymphocytic, neutrophilic, and eosinophilic mixed inflammation there was 3% neutrophils and 2% eosinophils. Representative images of CSF cytology in patients with anti-NMDAR encephalitis are shown in Figure 1. and 75 th percentiles 115.5 x 10 6 /L and 210 x 10 6 /L, range 101 x 10 6 /L-242 x 10 6 /L). All the above results are presented in Table 1.

CSF biochemical tests
As shown in Table 1

Specific OCBs
Of the 164 patients in the study 111 underwent OCB examination, and in 51 patients specific OCBs were detected in the CSF ( Table 1). The positivity rate in the 111 patients examined was 45.9%.

Correlations between mRS score and CSF parameters, and associations between plasma cells and specific OCBs
Spearman's rank correlations between mRS score and inflammation parameters showed that mRS score was not significantly correlated with white cell count determined via routine CSF examination (r = -0.055, p = 0.485), degree of inflammation determined via routine CSF examination (r = -0.028, p = 0.726), white cell counts in CSF determined via cytology (r = 0.019, p = 0.813), or degree of inflammation in CSF determined via cytology (r = -0.003, p = 0.969). The χ2 test was used to assess the correlation between plasma cells and specific OCBs (Table 2), and the two parameters were not significantly correlated (χ 2 = 0.831, r = 0.086, p = 0.362).

Comparisons between routine CSF examination and CSF cytology
White cell count determined via routine CSF examination was significantly positively correlated with white cell count determined via cytology (r = 0.599, p < 0.001), and degree of inflammation determined via routine CSF examination was significantly positively correlated with degree of inflammation determined via cytology (Table 3). At a positivity threshold of 5 × 10 6 white blood cells/L the rate of positive detection via routine CSF examination was 64.0% and the rate of positive detection via CSF cytology was 68.3%, and the difference was statistically significant (p < 0.001). The associated kappa value was 0.388 (p < 0.001), thus the consistency between the two methods was poor (    In evaluations of encephalitis in the current study the positivity rate of CSF cytology was higher than that of routine CSF examination with respect to pleocytosis. Furthermore, while cytologically determined inflammation was positively correlated with pleocytosis determined via routine CSF examination, the consistency between the two methods was poor (kappa = 0.338). This indicates that abnormal CSF cytology is not inherently associated with pleocytosis as determined via routine CSF examination. This lack of consistent parallel detection may be because the amount of CSF used in CSF cytology (500 µl) is far greater than that used in routine CSF examination (5 µL). It can be inferred from this that CSF cytology can provide more information on cell morphology and cell type proportions than routine CSF examination. In addition, in the present study more than a third of patients with anti-NMDAR encephalitis did not exhibit pleocytosis via routine CSF examination. This result is concordant with a previously reported case series in which initial non-inflammatory CSF varied widely in 15%-96% of patients at the first evaluation [9]. It is notable that initial non-inflammatory CSF evidently does not exclude a diagnosis of anti-NMDAR encephalitis, and in the past some patients may have been misdiagnosed leading to delays in the initiation of immunotherapy. The results of the current study suggest that CSF cytology is an effective candidate for the detection of an inflammatory state that has not been detected via routine CSF examination.
CSF cytological characteristics suggest that lymphocytes and plasma cells play vital roles in immune processes involved in anti-NMDAR encephalitis. Lymphocytes include B cells and T cells. Following activation by B cell-activating factor[10], undifferentiated B cells can become fully differentiated B cells known as plasma cells, which secrete antibody [11][12][13]. In patients with anti-NMDAR encephalitis memory B cells reach the brain then undergo clonal expansion triggered by antigen stimulation, affinity maturation of their antigen receptors, and differentiation into anti-NMDAR antibody-producing plasma cells [14]. The role of T cells in the immunopathogenesis of anti-NMDAR encephalitis has not been fully elucidated, but recent studies suggest that they are also involved [15,16].
It has been reported that Th17 cells appear to support B cell responses outside germinal centers and promote antibody production more efficiently than their Th1 counterparts in animal models [16].
In the current study neutrophils were detected in patients with anti-NMDAR encephalitis via CSF cytology, and in a few patients the proportion of neutrophils was higher than the proportion of lymphocytes. Similar CSF changes have only been described previously in one case report [17] . The effects of neutrophils on the immune processes involved in anti-NMDAR encephalitis remain unclear. It has been reported that neutrophil depletion after subarachnoid hemorrhage improves memory via NMDARs [18], and that in neuromyelitis optica spectrum disorders the presence of neutrophils is a very common characteristic of the associated lesions [19]. It has also been reported that complement activation by aquaporin-4-IgG results in a marked rise in the amount of circulating neutrophils, which can then enter the central nervous system and participate in early neuromyelitis optica lesion development via a neutrophil elastase-dependent mechanism [20]. Based on this, we speculate that neutrophils may also participate at least some of the immunopathology processes involved in anti-NMDAR encephalitis.
The relatively high rate of specific OCB positivity and the presence of plasma cells determined via CSF cytology in patients with anti-NMDAR encephalitis in the current study constitutes strong evidence supporting the concept that anti-NMDAR antibody was generated via intrathecal synthesis. The intrathecal source of pathogenic antibodies in anti-NMDAR encephalitis has been suggested to originate from CD138 + plasma cells identified in perivascular and interstitial spaces in biopsy and autopsy studies [21,22]. It has previously been reported that expanded intrathecal plasma cells may be the main source of intrathecal immunoglobulin, often detectable as OCBs [23,24]. Malviya et al.
[25] reported that CD19 + B cells and CD138 + CD19 + plasma cells were significantly elevated in the CSF of patients with anti-NMDAR encephalitis, but not in their serum.
These data suggest intrathecal B cell and plasma cell accumulation as a likely intrathecal source of antibody production in patients with anti-NMDAR encephalitis.
In some previous studies CSF titers of anti-NMDAR antibody were significantly correlated with outcome, and decline in CSF titer was positively correlated with relapse [1,26].
Therefore, we speculate that there may be a correlation between disease severity and the degree of inflammation evident in the CSF. Notably however, no such relationship was apparent in the current study, and the level of pleocytosis in CSF did not correlate with disease severity.

Conclusions
In the present study investigating the cytological characteristics of CSF in patients with anti-NMDAR encephalitis, inflammatory reactions were detected in two thirds of patients.
The most common type of inflammation was lymphocytic, but neutrophils were also seen in a few patients. A substantial number of eosinophils was only detected in 1 of the 164 patients in the entire study. Activated lymphocytes, plasma cells, and specific OCBs were detected in some CSF samples. Such findings may constitute initial evidence of potential anti-NMDAR encephalitis, and may also facilitate better understanding of the immune mechanisms involved in the condition. The relatively high positivity rate associated with CSF cytology, and its superiority for assessing cell morphology render it an effective tool for the evaluation of anti-NMDAR encephalitis.

Consent
Written informed consent were obtained from all the patients or their guardians for publication of this work and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.