B- and T-cell markers in opsoclonus–myoclonus syndrome

Subjects.

Thirty-six children with OMS were recruited through the National Pediatric Myoclonus Center, and their parents signed consent for this institutional review board–approved study. The clinical characteristics are shown in the table. A thorough search was made for occult neuroblastoma using neuroimaging and blood and urine tumor markers. Those with a previously identified tumor, which was either stage I or II in all but one case, had undergone tumor resection.

Scoring of neurologic status.

Each child was videotaped (see the supplementary material on the Neurology Web site; go to www.neurology.org). A trained observer blinded to treatment status rated motor impairment using the OMS Evaluation Scale, which we devised and validated.23 Items of the 12-part scale were rated from 0 to 3 as an index of increasing neurologic severity or impairment. Total score, which was calculated as the sum of subscores, tallied 36 in the most extreme cases.

Controls.

Control subjects included 18 children with myoclonus, ataxia, idiopathic intracranial hypertension, chronic daily headache, or developmental delay, who were undergoing a diagnostic evaluation. Routine CSF studies such as cell count, protein, and glucose were normal. None of the children ever received immunotherapy. The mean age was 8.5 ± 1.2 years (range 1.3 to 16 years). Seven were boys.

Lumbar puncture.

To prevent contamination of CSF with blood due to trauma, minimize sedation risks during lumbar puncture, standardize the degree of stress on immune function, and provide compassionate care, an anesthesiologist administered IV propofol after brief sevoflurane induction to insert the IV line. The lumbar puncture was performed in the lateral decubitus position, and the first 4 mL was sent for routine studies. An additional 8 to 10 mL, not to exceed 15% of the calculated total CSF volume, was collected for flow cytometry. Blood for parallel studies was also drawn.

CSF mononuclear cell count was 0 to 3 cells/mm³ in all children except three with OMS, who had pleocytosis with counts of 5 to 20 cells/mm³. The mean CSF white blood cell count of 1.3 ± 0.2/mm³ in control subjects was not statistically different from 1.5 ± 0.2/mm³ in OMS. The CSF red blood cell count was always ≤1, and CSF protein and glucose levels were normal.

Flow cytometry.

Cells were recovered from CSF through low-speed centrifugation (200 g, 7 minutes) and concentrated by resuspension in phosphate-buffered saline (PBS; pH 7.2) containing 0.5% bovine serum albumin (Sigma Chemicals, St. Louis, MO). CSF samples (100 μL/tube) were stained with 15 to 20 μL of monoclonal antibodies (mAb) directly conjugated with fluorescein isothiocyanate, phycoerythrin, allophycocyanin, or phycoerythrin-cyanine 5 (see table E-1 on the Neurology Web site), which were arranged as panels in several assay tubes (see table E-2 on the Neurology Web site).24 Samples were incubated for 20 minutes at room temperature. CSF leukocytes were washed and resuspended in 200 μL of PBS after final incubations.

Whole peripheral blood samples (100 μL/tube) were stained with the same panel of directly conjugated mAb used for CSF cells. They were further incubated for 10 minutes with 100 μL/tube Optilyse B (Immunotech, Marseille, France; Beckman-Coulter, Miami, FL). One milliliter of H₂O was subsequently added to each sample and incubated for a final 10 minutes at room temperature.

All samples were acquired and analyzed by flow cytometry on a FACSCalibur cytometer equipped with a 488-nm argon/633-nm HeNe laser (Becton-Dickinson, San Jose, CA). For CSF mononuclear cell count of >3 cells/mm³, 1,000 events were counted; otherwise, 500 were counted. Data acquisition and analysis were performed with CellQuest (Becton-Dickinson). Dependent on the surface marker fluorochrome, data were plotted as log vs log of fluorescence. The relative size of each lymphocyte subset (percentage of positive cells) was expressed as the percentage within the total lymphocyte population or “lympho/monocyte gate,” as determined manually by forward and side scatter characteristics.

Quality control was maintained in several ways. The lymphocyte gate was checked by CD14/CD45 double staining and deemed pure if it included ≥95% of all lymphocytes and <5% contamination with granulocytes, monocytes, or cell debris. Appropriately labeled isotypes (IgG₁) were used as internal controls in each analysis. Instrument calibration was monitored daily with Calibrite beads (Becton-Dickinson). A “lymphosum” was calculated as the sum of the percentage for T-cells, B-cells, and NK cells and found to equal 92 to 98%, as it should. For CD3, coefficients of variation were 2.0% (intrabatch) and 0.5% (interbatch), which fall within acceptable performance parameters.

Statistical analysis.

Dependent variables (CSF and blood lymphocytes and the CSF/blood ratio) were analyzed as means by two-tailed t-tests or one-way analysis of variance (ANOVA) on the Statistical Analysis System (SAS, Cary, NC), a computer software program for statistical analysis,25 using a significance level of p < 0.05. The Duncan Multiple Range Test was utilized for post-hoc subgroup comparisons of all statistically significant results from ANOVA to allow comparisons between OMS categories as well as with controls. The principal independent variables were diagnosis (OMS vs controls), etiology (tumor vs no tumor), neurologic severity (mild, moderate, severe), syndrome duration (acute, subacute, chronic), and treatment (current treatment vs no treatment, prior chemotherapy vs no chemotherapy). Pearson correlation coefficients were used for correlational analysis. Demographic data were analyzed by χ² and χ² goodness-of-fit tests. To evaluate unequal variances, we performed nonparametric tests, but as the results were substantially the same as the unequal-variance t-tests, only t-tests are reported here.

Because lymphocyte populations, in blood at least, may vary with age and there are no CSF data in healthy children, we looked for correlations between age and each type of lymphocyte in our control subjects. No significant correlations were found, so the data were pooled. This had the effect of increasing the mean age of our control subjects somewhat, but we felt it was more important, given the small sample size, to boost the statistical power. The same argument applies to blood, as the relative size of various lymphocyte populations in healthy children ages 2 to 5 years and 5 to 10 years differs by only 1 to 6%.26 Published normative CSF data for healthy young adults24 as well as children with various neurologic disorders19 were used to assess the comparability of our neurologic controls.

Distribution of lymphocyte population in CSF.

In control subjects, the prominent characteristic of normal CSF lymphocytes was the dominance of αβ T-cells (CD3⁺) and almost total lack of B-cells. The rank order of CSF lymphocytes was helper/inducer T-cells (CD3⁺CD4⁺) > cytotoxic/suppressor T-cells (CD3⁺CD8⁺) > γδ T-cells (T-cell receptor [TCR]-γδ⁺), NK cells (CD3⁻CD16/56⁺), NK-like T-cells (CD3⁺CD16/56⁺) >> B-cells (CD45⁺CD3⁻CD19⁺).

In OMS, abnormalities in the percentage of various cell types were found in all cases, despite the absence of CSF leukocytosis in most (figure 1). Percentages were lower for CD3⁺CD4⁺ cells (−19%; p = <0.0001) and minimally higher for CD3⁺CD8⁺ cells (+5%; p = 0.047). As a result, the mean ratio of CD4⁺ to CD8⁺ T-cells was less (−53%; p = 0.017). Percentages were higher for γδ T-cells (2.7-fold; p = <0.0001) and CD19⁺ B-cells (6.5-fold; p = 0.0004). There was a trend toward lower CD3⁺ cells (p = 0.055) but no intergroup difference in CD3⁻ or CD3⁺ NK cells. The immunophenotype was not changed by presence of tumor or treatment.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Distribution of CSF lymphocyte population. (A) In opsoclonus– myoclonus syndrome (OMS), the percentage of CD3⁺ cells and CD4⁺ T-cells was reduced compared with controls, but increases were found for CD19⁺ B-cells, γδ T-cells, and CD8⁺ T-cells. There were no statistically significant differences between tumor and no-tumor groups (B) or untreated and treated groups (C). Data are means ± SEM. The B-cell percentage for controls was 0.7 ± 0.2%. Asterisks signify statistical significance by t-tests: *0.01 ≤ p < 0.05, ***0.0001 ≤ p < 0.001, **** p < 0.0001.

Certain cell types were found in neither controls nor OMS. Based on CD45 staining and light scatter characteristics, no plasma cells or macrophages were detected in CSF. Monocytes regularly constituted <5% within the lympho/monocyte gate (data not shown).

Expression of T-cell activation markers in CSF.

We evaluated both human leukocyte antigen (HLA) DR and CD25 (interleukin-2 receptor) T-cell activation markers. HLA-DR⁺ T-cells accounted for approximately 12% of CD3⁺ cells in the CSF of control subjects, whereas few T-cells were CD25⁺ (figure 2). In OMS, the percentage of HLA-DR⁺ T-cells was 51% greater (p = 0.0006), as can be seen in representative flow cytometric displays (see figure E-1 on the Neurology Web site). The percentage of CD3⁺CD25⁺ cells was not significantly different in OMS. OMS etiology and treatment had no significant effect.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. CSF CD3⁺ T-cell activation markers in control subjects vs opsoclonus–myoclonus syndrome (OMS) (A), tumor vs no tumor (B), and untreated vs treated OMS (C). Data are means ± SEM. Asterisks signify statistical significance by t-tests: ***0.0001 ≤ p < 0.001. In OMS, 21 of 36 children had values above the upper confidence limit of the control group for human leukocyte antigen DR⁺ T-cells (up to 34%) but only 11 children for CD3⁺CD25⁺ cells (up to 25%).

CSF T-cell markers of maturation.

In a comparison of CD45R isoforms on CD3⁺ cells from control subjects, the percentage of “memory” (CD45RO⁺) T-cells was about twofold higher than that of “naive” (CD45RA⁺) T-cells (p = 0.0004). However, there were no significant differences between controls and OMS for CD45RO⁺ (47.3 ± 4.4 and 43.5 ± 4.0, respectively) or CD45RA⁺ (21.7 ± 4.8 and 18.5 ± 1.9) T-cells. OMS etiology and treatment also had no significant effects with the exception of slightly lower CD45RA⁺ cells in the no-tumor group (−9%; p = 0.019) (data not shown).

Relation of neurologic severity of OMS to CSF cell types.

The effect of neurologic severity on the percentage of CD19⁺ B-cells as well as CD4⁺ and TCR-γδ⁺ T-cells was statistically significant (figure 3). Reciprocal relations were seen between αβ T-cells and γδ T-cells. The two children with the highest percentage of CSF γδ T-cells (26%) or B-cells (29%) were most severe (total scores 34 and 36, respectively). Severity correlated negatively with the percentage of CD3⁺ cells (r = −0.56, p = 0.0012) and CD3⁺CD4⁺ cells (r = −0.38, p = 0.024). A positive correlation was found with the percentage of CD19⁺ cells (r = 0.37, p = 0.03), TCR-γδ⁺ cells (r = 0.54, p = 0.0019), and CD3⁻ NK cells (r = 0.38, p = 0.04). There were no other significant correlations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Relation between neurologic severity (total score) in opsoclonus–myoclonus syndrome (OMS) and percentage of CSF CD4⁺ T-cells (A), γδ T-cells (B), and CD19⁺ B-cells (C). Sample size is shown at the base of each column. Data are means ± SEM. Asterisks indicate statistically significant differences between OMS severity categories on Duncan test, p < 0.05. Dagger indicates significant differences between OMS and control subjects. Analysis of variance with linear trend analysis revealed that the more severely affected children (higher total score) had a lower percentage of CSF CD4⁺ T-cells (F = 20.6, p ≤ 0.0001). In contrast, the percentage of CSF γδ T-cells increased with severity (F = 25.6, p ≤ 0.0001), being nearly double in the severe category, and the percentage of CSF CD19⁺ B-cells was also higher (F = 21.2, p ≤ 0.0001).

Relation of syndrome duration to CSF cell types.

The duration of illness correlated with the CD4/CD8 ratio (r = 0.40, p = 0.015) and the percentage of CD4⁺ T-cells (r = 0.44, p = 0.007) and CD3⁺ cells (r = 0.38, p = 0.039). There was a negative correlation with TCR-γδ⁺ cells (r = −0.37, p = 0.043) and total score (r = −0.45, p = 0.008).

CSF/blood lymphocyte ratios.

In control subjects, the percentage of all cells in CSF exceeded that in paired peripheral blood, except for B-cells and CD45RA⁺ T-cells. The magnitude of CSF/blood ratios differed significantly between cell types (figure 4). The percentage of CD3⁺HLA-DR⁺ cells was sevenfold higher in CSF than in blood, whereas the percentage of CSF CD19⁺ B-cells was only one-twentieth of that in blood. The means of other ratios were just above 1.0, with the exception of the CD3⁺ NK cell ratio and CD45RO⁺ cell ratio, which were about threefold higher.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. CSF/blood lymphocyte ratios. (A) In opsoclonus–myoclonus syndrome (OMS), the ratios for γδ T-cells, CD19⁺ B-cells, and CD3⁺CD16/56⁺ cells were significantly increased. The B-cell ratios were 0.05 ± 0.01 for controls and 0.16 ± 0.03 for OMS. (B) There were no statistically significant differences between the tumor and no-tumor groups. Data are means ± SEM. Asterisks signify statistical significance by t-tests: *0.01 ≤ p < 0.05, ***0.0001 ≤ p < 0.001.

In OMS, the CSF/blood ratio for CD19⁺ B-cells was threefold higher than in control subjects (p = 0.0005). The γδ T-cell ratio was 2.4-fold higher (p = 0.0005) and correlated with total score (r = 0.61, p = 0.0004), syndrome duration (r = −0.42, p = 0.022), and the CD4/CD8 ratio (r = −0.49, p = 0.0075). There was a 2.4-fold higher ratio for CD3⁺NK cells in OMS (p = 0.04). Presence of a tumor and treatment had no effect. There were no other significant differences in lymphocyte ratios, with the exception of a lower HLA-DR⁺ cell ratio in chemotherapy-treated children (−57%; p = 0.0019).

The relative size of the peripheral blood mononuclear cell pool did not reflect the lymphocyte subset abnormalities found in CSF. There were no significant differences between control and OMS subjects. In the tumor group, approximately 1.5-fold higher percentages of CD19⁺ B-cells (p = 0.038) and HLA-DR⁺ T-cells (p = 0.048) were found (data not shown).

Comparability of pediatric neurologic control subjects and healthy young adults.

No major discrepancies were found between our control subjects and data from healthy young adults24 in the general distribution and relative proportions of CSF lymphocyte subsets. Our controls had a slightly lower percentage of CD3⁺ cells, CD4⁺ T-cells, and NK-like T-cells and a slightly higher percentage of NK cells and HLA-DR⁺ T-cells. Although we could find no study of CSF γδ T-cells in normal individuals, the percentage of γδ T-cells in our controls was comparable with that of a group of pediatric neurologic control subjects.19