6‐Mercaptopurine modifies cerebrospinal fluid T cell abnormalities in paediatric opsoclonus–myoclonus as steroid sparer

Summary

The purpose of this study was to evaluate the capacity of 6‐mercaptopurine (6‐MP), a known immunosuppressant, to normalize cerebrospinal fluid (CSF) lymphocyte frequencies in opsoclonus–myoclonus syndrome (OMS) and function as a steroid sparer. CSF and blood lymphocytes were immunophenotyped in 11 children with OMS (without CSF B cell expansion) using a comprehensive panel of cell surface adhesion, activation and maturation markers by flow cytometry, and referenced to 18 paediatric controls. Drug metabolites, lymphocyte counts and liver function tests were used clinically to monitoring therapeutic range and toxicity. In CSF, adjunctive oral 6‐MP was associated with a 21% increase in the low percentage of CD4⁺ T cells in OMS, restoring the CD4/CD8 ratio. The percentage of CD4⁺ T cells that were interferon (IFN)‐γ⁺ was reduced by 66%, shifting the cytokine balance away from T helper type 1 (Th1) (proinflammatory) predominance. The percentage of natural killer (NK) cells decreased significantly in CSF (–32%) and blood (–67 to −82%). Low blood absolute lymphocyte count was more predictive of improvement in CSF lymphocyte proportions (correlated with % CD4⁺ T cells) than the 6‐thioguanine level (no correlation). 6‐MP was difficult to titrate: 50% achieved the target absolute lymphocyte count (< 1·5 K/mm); 20%, the ‘therapeutic’ 6‐thioguanine level; and 40% the non‐toxic 6‐methylmercaptopurine level. Side effects and transaminase elevation were mild and reversible. Clinical steroid‐sparing properties and lowered relapse frequency were demonstrated. 6‐MP displayed unique pharmacodynamic properties that may be useful in OMS and other autoimmune disorders. Its steroid sparer capacity is limited to children in whom the therapeutic window can be reached without limiting pharmacokinetic factors or side effects.

Introduction

Lymphocyte recruitment into the central nervous system is characteristic of neuroimmunological disorders, such as multiple sclerosis, systemic lupus erythematous and the autoimmune paraneoplastic syndrome opsoclonus–myoclonus (OMS) 1. In children with OMS, peripheral induction by neuroblastoma‐associated onconeural antigens is presumed to result in the characterized immunophenotypical abnormalities of cerebrospinal fluid (CSF) 2. In contrast to the CSF immunophenotype of paediatric controls, in which the vast majority of cells are CD4⁺ T cells and B cells are scant 2, expansion of certain αβ T cells [CD8⁺, human leucocyte antigen D‐related (HLA‐DR⁺)] with reduction in others (CD4⁺ T cells and CD4/CD8 ratio) was found in OMS. CSF CD19⁺ B cells were also expanded. Several of the abnormalities correlated with clinical severity and disease duration, suggesting that they could serve as biomarkers of disease activity 2.

The question remains as to whether the CSF immunophenotype is restorable and whether normalization (brain immune quiescence) results in clinical benefit. If so, treatment for OMS could be evidence‐based and individualized. While anti‐B cell monoclonal antibody therapies eradicate CSF B cells in OMS 3, therapeutic manipulation of T cell abnormalities is largely unstudied. Standard conventional immunotherapies for OMS, such as corticosteroids, adrenocorticotrophic hormone (ACTH) and intravenous immunoglobulins (IVIG), do not appear to alter the CSF T cell phenotype 2, perhaps accounting for the high relapse rate on steroid tapering 4. Little is known about the effect of other immunotherapy or chemotherapy on the CSF lymphoid subpopulations.

The purine anti‐metabolite 6‐mercaptopurine (6‐MP), like its prodrug azathioprine, is used clinically as a ‘steroid‐sparer’ and maintenance drug, rather than for induction. It has been administered in treating autoimmune disorders in children, such as inflammatory bowel disease 5 and certain leukaemias 6, as well as systemic lupus erythematosis, rheumatoid arthritis and inflammatory myopathies in adults 7. The therapeutically active metabolite 6‐thioguanine (6‐TG) can be measured commercially. Despite the fact that 6‐MP is one of the oldest immunosuppressants in current use 8, little is known about its effects on neuroinflammation or the central nervous system. The capacity of 6‐MP to modify CSF lymphocytic abnormalities and allow steroid doses to be weaned without relapse in neuroinflammatory diseases has not been reported previously.

Methods

Patients

Children with OMS were recruited through the National Pediatric Myoclonus Center, an international centre for paediatric‐onset OMS, in Springfield, IL, and evaluated by the first two authors to confirm the clinical diagnosis of OMS. Written parental consent was obtained for an observational Institutional Review Board‐approved study to gather clinical and immunological data, including CSF and blood samples, for immunological studies of OMS (Southern Illinois University School of Medicine). The initial lumbar puncture was to diagnose neuroinflammation through lymphocyte subset analysis 2. Given the high morbidity and propensity for permanent neurological, neuropsychiatric and cognitive sequelae in OMS 4, the parents of children with neuroinflammation were reconsented for a second lumbar puncture after treatment to demonstrate no evidence of disease activity (NEDA) 9 and for clinical management issues, such as relapse, insufficient response or other clinical concerns 1. Videotapes of the neurological examination were made with written parental consent.

The 11 children treated with 6‐MP as a steroid sparer were those who displayed CSF T cell abnormalities without B cell expansion while being on conventional immunotherapy (usually ACTH and IVIG). Patients with CSF B cell expansion were not treated with 6‐MP because anti‐CD20 therapy has been shown to be so effective in that OMS setting 3, whereas 6‐MP is not known to alter CSF B cell frequency. The choice of treatment modality was empirical and clinically based, and drugs were administered by the treating physicians, not as part of a drug trial. Western Institutional Review Board (Puyallup, WA, USA) designated exempt review for retrospective analysis of responses to immunotherapy and other longitudinal data obtained in the context of clinical care.

Clinical characteristics are shown in Table 1. Neurological symptomatology ranged from mild to severe, mostly mild‐to‐moderate. Some had been treated previously with chemotherapy or related agents, but not within at least 6 months of testing. None had evidence of recurrent neuroblastoma by neuroimaging studies or blood and urine tumour markers. Controls were 18 children with non‐autoimmune neurological and non‐neurological disorders, whose data were published previously 2, 10, and compared favourably with those from healthy young adults 11.

Table 1.

Clinical characteristics

n	11
Age (years) *	3·7 ± 0·9
Gender (n)
Boys	6 (55%)
Girls	5 (45%)
OMS onset (years) *	1·4 ± 0·2
OMS duration (years) *	2·3 ± 0·8
OMS severity category (total score)†	11·7 ± 2·9
OMS aetiology
No tumour	6 (55%)
Tumour	5 (45%)
Ongoing treatments
ACTH	9
IVIG	9
ACTH and IVIG	7
Prednisone	1
Prior chemotherapy or related treatment
Cyclophosphamide	4
Rituximab	4
Methotrexate	1
Mycophenolate	2
β‐interferon	2

*Means ± standard error of the mean. †Based on total score: mild = 0–12, moderate = 13–24, severe = 25–36. OMS = opsoclonus–myoclonus syndrome; ACTH = adrenocorticotrophic hormone; IVIG = intravenous immunoglobulin.

Flow cytometry

Lumbar punctures were performed under sedation 12, and blood for lymphocyte phenotyping was drawn at the same time. The expression of lymphocyte surface antigens was investigated in CSF and whole blood, as described previously 2. Briefly, cells were recovered from CSF through low‐speed centrifugation and resuspended in 200 μl phosphate‐buffered saline (PBS) containing 0·5% bovine serum albumin. They were incubated in the presence of various combinations of monoclonal antibodies to adhesion and activation proteins in combination with anti‐CD3 and anti‐CD45 antibodies. The panel was used to identify T cells [CD4⁺, CD8⁺, CD3⁺natural killer (NK), T cell receptor (TCR)‐γδ⁺)], T cell activation (HLA‐DR⁺, CD25⁺), T cell maturation (CD45RA⁺, CD45RO⁺), NK cells (CD16/56⁺CD3^–) and B cells (CD19⁺CD45⁺CD3^–) through four‐colour fluorescence‐activated cell sorting (FACS) 2. After final incubations and washing, the cells were resuspended in 200 μl of PBS. Blood samples were handled in the same manner, except that erythrocytes were lysed 10 and NK T cells and T cell activation and maturations markers were not measured.

In other assay tubes, CSF CD4⁺ T cells were stained intracellularly for the representative Th1 cytokine IFN‐γ and Th2 cytokine IL‐4. After completion of the 20‐min incubation, intracellular cytokine panel samples were fixed, permeabilized (IntraPrep‐Immunotech) and stained for IL‐4 and IFN‐γ (Beckman‐Coulter, Miami, FL, USA). CSF samples were washed and resuspended in PBS after final incubations.

All samples were acquired and analysed on a FACSCalibur cytometer equipped with a 488‐nm argon/633‐nm HeNe laser (Becton‐Dickinson, San Jose, CA, USA). Data acquisition and analysis were performed with CellQuest (Becton‐Dickinson). Dependent on the surface marker fluorochrome, data were plotted as log versus log of fluorescence. Quality control was maintained as described previously 2. 

Clinical treatment

Patients with OMS were evaluated before and after 7 ± 1 months of 6‐MP treatment. 6‐MP (Purinethol, 50 mg/tablet) had been given as add‐on therapy at a starting dose of 1·5–2·0 mg/kg/day orally in a single daily dose. Following clinical guidelines, the dose was titrated to achieve a therapeutic level of 6‐thioguanine (6‐TGN), while avoiding elevation of 6‐methylmercaptopurine (6‐MMPN) or the hepatic transaminases aspartate transaminase (AST) and alanine transaminase (ALT) 13, and to achieve a target absolute lymphocyte count (ALC) of 1·0–1·5 K/mm³. According to the manufacturer, the timing of blood drawing in relation to dose is not important, because 6‐MP has a plasma half‐disappearance time of 21 min in children. Erythrocyte 6‐TGN and 6‐MMPN levels were measured at intervals by Prometheus Laboratories Inc. (San Diego, CA, USA), using a proprietary and patented high‐performance liquid chromatography method. Patients were allowed to reach a steady‐state of at least 3 weeks after a dose change before metabolite levels were drawn. One child, presumed to be a non‐metabolizer 14, exhibited a toxic 6‐MMPN level without a 6‐TGN level when 6‐MP was introduced, so the drug was discontinued. For the others, the mean 6‐MP dose was 2·0 ± 0·2 mg/kg/day.

Children already receiving ACTH or IVIG were continued on those treatments. ACTH was tapered slowly over several months on an alternate‐day intramuscular injection schedule, and IVIG was infused once monthly at 1 g/kg. In almost all patients, IVIG dose and dosing remained constant throughout the study, because we do not attempt to reduce the IVIG dose or lengthen the interval between doses until ACTH or corticosteroids are discontinued 1.

Scoring of neurological status

Childhood‐onset OMS usually affects toddlers, who may lose their ability to speak, sit or walk. Each child was videotaped before and after treatment. A trained observer, blinded to treatment status, scored motor impairment using the OMS Evaluation Scale 15. Each item of the 12‐item scale was rated from 0 to 3 as an index of increasing neurological severity or impairment. Subscores were summated to a maximum total score of 36.

Statistical analysis

Data were analysed statistically using Microsoft Excel and the Statistical Analysis System (SAS Institute, Cary, NC, USA). Comparisons between the means of controls and OMS were made by two‐tailed t‐tests. Pre‐ and post‐treatment comparisons of means were made by paired t‐tests and of numbers of subjects by McNemar's test. Our published paediatric control data for CSF 2 and peripheral blood mononuclear cell (PBMC) subsets 10, obtained by the same FACS methods, were used for comparison. P < 0·05 was considered statistically significant.

Results

Effects on CSF lymphocyte phenotype

6‐MP was associated with multiple effects on the CSF immunophenotype (Fig. 1). It increased significantly the low percentage of CD4⁺ T cells (+21%) and CD45RO⁺ (memory) T cells (+73%), while reducing the frequency of NK cells (–32%). The CD4/CD8 ratio did not differ significantly pre‐ versus post‐treatment, but after treatment it was no longer significantly different to controls (data not shown). There were no significant effects on γδ T cells, NK T cells, B cells, activated T cells or CD45RA⁺ (naive) T cells.

Figure 1.

Effect of 6‐mercaptopurine (6‐MP) on the relative size of cerebrospinal fluid (CSF) lymphocyte subsets compared to our published control data 2, 7. Data are means ± standard error of the mean (s.e.m.). Statistical comparisons were made pre‐ and post‐treatment (*) and between controls and each opsoclonus–myoclonus syndrome (OMS) subgroup (†). Pre‐ and post‐treatment comparisons were made by paired t‐tests; intergroup comparisons were made by two‐tailed t‐tests. The level of statistical significance is indicated by the number of symbols as follows: *0·01 ≤ P < 0·05, **0·001 ≤ P < 0·01.

Prior to treatment with 6‐MP there was a Th1 predominance, with the percentage of IFN‐γ⁺ cells exceeding that of IL‐4⁺ cells. 6‐MP reduced the percentage IFN‐γ⁺CD4⁺ T cells (–66%). The IFN‐γ/IL‐4 ratio dropped from 14·2 ± 12·0 (range = 0·2–133) to 0·9 ± 0·1 (range = 0·3–1·9). The effect was not significant due to the huge variance.

Secondary analysis was performed to determine the effect of prior chemotherapy/immunotherapy before treatment with 6‐MP. Prior treatment with chemotherapy or related drugs (n = 6) did not alter the percentage of CSF B cells significantly (0·32 ± 0·15%). However, children who had never been treated with anything but ACTH, corticosteroids or IVIG (n = 5) exhibited an increased percentage of CSF B‐cells (2·7‐fold, P = 0·036) following 6‐MP therapy compared to their counterparts.

Effects on the CSF/blood lymphocyte ratio and PBMC immunophenotype

Treatment with 6‐MP was associated with a significant reduction in the CSF/blood lymphocyte ratio for total T cells and CD8⁺ T cells and increased it marginally but significantly for CD19⁺ B cells (Fig. 2). 6‐MP altered the relative size of the PBMC pool (Fig. 3a). It increased total T cells (+24%), CD4⁺ T cells (+33%) and CD8⁺ T cells significantly (+15%). It decreased NK cells (–67%). Compared to controls, there were fewer B cells. There were no significant effects on γδ T cells or activated T cells. 6‐MP also altered the absolute size of the PBMC pool (Fig. 3b), reducing NK cells (–82%) and HLA‐DR⁺ activated T cells significantly (–71%).

Figure 2.

Effect of 6‐mercaptopurine (6‐MP) on the cerebrospinal fluid (CSF)/blood ratio of lymphocyte percentages. Data are means ± standard error of the mean (s.e.m.). Statistical comparisons were made pre‐ and post‐treatment (*) and between controls and opsoclonus–myoclonus syndrome (OMS) (†). The level of statistical significance is indicated as follows: *0·01 ≤ P < 0·05; **0·001 ≤ P < 0·01; ***0·0001 ≤ P < 0·001; ****0·00001 ≤ P < 0·0001; *****0·000001 ≤ P < 0·00001.

Figure 3.

6‐Mercaptopurine (6‐MP) treatment effect on the (a) relative and (b) absolute size of peripheral blood mononuclear cell (PBMC) subsets. Data are means ± standard error of the mean (s.e.m.). Statistical comparisons were made pre‐ and post‐treatment (*) and between controls and opsoclonus–myoclonus syndrome (OMS) (†). The level of statistical significance is indicated as follows: *0·01 ≤ P < 0·05; **0·001 ≤ P < 0·01; ***0·0001 ≤ P < 0·001; ****P < 0·0001.

Laboratory monitoring

The mean ALC of 6‐MP‐treated children fell within our target zone of 1·0–1·5 K/mm³, but 50% of the children did not reach the target ALC. Approximately 80% did not achieve a therapeutic blood level because dose titration had to be halted due to escalation of liver function enzymes (Table 2). Mean hepatic transaminase concentrations, which were elevated in 72%, were increased 31% (AST) and 46% (ALT) above the reference range. However, none were at a clinically serious level. Sixty per cent of the children had an elevated level of 6‐MMPN, but the 6‐MMPN concentration did not correlate with either the ALT or AST levels. There was no significant correlation between 6‐TPN and 6‐MMPN concentrations. The 6‐TGN/6‐MMPN ratio was 0·036 ± 0·10. There was a significant negative correlation between the 6‐TGN concentration and the total leucocyte count (r = −0·68, P = 0·03). The mean absolute neutrophil count, haemoglobin and haematocrit fell within reference ranges.

Table 2.

Test results for monitored blood parameters

Laboratory test (units)	Mean ± s.e.m. (range) *	Reference range
6‐TGN level (pmol/8 × 10∧8 RBC)	188 ± 22 (91–312) *	230 400
6‐MMPN level (pmol/8 × 10∧8 RBC)	8503 ± 2102 (1420–22 978) *	< 5700
ALC (K/mm3)	1·4 ± 0·2 (0·36–2·7) *	2·7–8·7
ANC (K/mm3)	4·8 ± 1·3 (0·6–13·7)	1·2–8·1
ALT (U/l)	67 ± 25 (20–262) *	0–35
AST (U/l)	60 ± 1 (25–210) *	0–35
Hgb (g/dl)	12·7 ± 0·2 (11·7–13·9)	10·5–13·5
Hct (%)	37·4 ± 0·7 (33·4–42)	33–40

Relation to monitored parameter	n (%)
6‐TGN level	10
Subtherapeutic (< 230)	8 (80%)
Therapeutic (≥ 230)	2 (20%)
6‐MMPN level	10
Non‐toxic (< 5700)	4 (40%)
Toxic (≥ 5700)†	6 (60%)
ALC	10
Above target zone (≥ 1·5)	5 (50%)
Within target zone (< 1·5)	5 (50%)
Increased liver transaminases	9
ALT	6 (67%)
AST	7 (78%)

*Means were outside the clinical laboratory paediatric reference ranges. †The designation ‘toxic’ was set by the reference laboratory to identify individuals at higher risk for clinical toxicity, not based on clinical symptoms or signs. TGN = thioguanine nucleotide; RBC = red blood cells; MMPN = methylmercaptopurine; ALC = absolute lymphocyte count; ALT = alanine transaminase; AST = aspartate transaminase; s.e.m. =standard error of the mean.

Immunological correlations

There was a significant negative correlation between the ALC and the percentage of CSF CD4⁺ T cells (r = −0·68, P = 0·03), indicating that treatment helped restore the relative size of the CD4⁺ T cell subset. For CD8⁺ T cells (r = 0·57, P = 0·08) and HLA‐DR⁺ T cells (r = −0·58, P = 0·08), the correlations were not significant, but there may have been a trend. The 6‐MP dose correlated only with the percentage of CD45RO⁺ (memory) T cells (r = 0·65, P = 0·03). There were no significant correlations between 6‐TGN levels and the percentage of CSF lymphocytes.

When the data set was divided at the threshold of the predetermined target ALC (< 1·5 K/mm³) into ‘low ALC’ (n = 7) and ‘high ALC’ (n = 4), there were significant differences in 6‐MP effects on the CSF lymphocyte phenotype. In the low ALC subgroup, the percentage of CD4⁺ T cells increased by 27% with 6‐MP treatment (P = 0·009) and CD8⁺ T cells decreased by 24% (P = 0·03). As a result, the CD4/CD8 ratio increased by 67% to 2·0 ± 0·3 (P = 0·01). All these effects were in the direction of normalizing the CSF lymphocyte phenotype.

Clinical effects

6‐MP had several statistically significant clinical effects (Fig. 4). While on 6‐MP therapy, the dose of ACTH was reduced in all children. The mean dose decreased by 76% (P = 0·02, paired t‐test). Two children were weaned off ACTH for 1–2 months prior to the second lumbar puncture. In the one child taking prednisone, the prednisone dose was tapered from 10 mg on alternate days and discontinued.

Figure 4.

Clinical effects of 6‐mercaptopurine (6‐MP) on (a) adrenocorticotrophic hormone (ACTH) dose, (b) percent of children relapsing, and (c) percentage of children in various severity categories. Statistical comparisons were made between pre‐ and post‐treatment. There were significant reductions both in ACTH dose and relapse rate. (c) Due to the paucity of severe cases, severe and moderate categories were combined for statistical analysis. There was a significant difference in opsoclonus–myoclonus syndrome (OMS) severity category with 6‐MP treatment. The level of statistical significance is indicated as follows: *0·01 ≤ P < 0·05; **0·001 ≤ P < 0·01.

During a follow‐up period of 1·4 ± 0·1 years, only one of 11 children relapsed compared to nine of 11 prior to 6‐MP treatment (P = 0·0047, McNemar's test). The child who relapsed did so immediately following an IVIG infusion, so the relapse was attributed to IVIG, which was discontinued.

There was a statistically significant shift towards lesser motor severity in an analysis of the percentage of subjects in each severity category before and after 6‐MP treatment (P = 0·046, McNemar's test). However, total score remained within the mild category: 9·2 ± 1·1 post‐treatment [not significant (n.s.)]. There was no statistically significant difference in the perception of clinical improvement by the parents (seven of 10 improved) or examiners (five of 10).

6‐MP was associated with nausea, vomiting or stomach pains in three children, one of whom was treated with oral ondansetron. Another child complained of headache. None of their parents discontinued the medication.

Discussion

To our knowledge, this is the first report of 6‐MP effects on the CSF immunophenotype and its use in paediatric OMS. 6‐MP was applied in the context of steroid‐sparing for already‐treated but still‐symptomatic children, rather than as a drug to induce neurological remission in untreated cases. It allowed ACTH to be tapered without inducing relapse in almost all cases in more than a year of follow‐up. The rate of relapse after 6‐MP was lower than we have encountered previously with ACTH, corticosteroids or IVIG alone 4.

6‐MP had the capacity to correct certain CSF T cell abnormalities in OMS, whereas ‘conventional’ agents (ACTH, corticosteroids and IVIG) 2 and cyclophosphamide did not 16. There are few mechanistic data by which to discriminate whether the 6‐MP effect is exerted on haematogenous leucocytes that then enter the CNS (through the same barriers crossed by immune cells 17) or on intrathecally activated neuroinflammatory cells. In the monkey, the predominant route for 6‐MP entry into the CNS from plasma (after i.v. bolus) is the blood–CSF barrier (choroid plexus), but 12% enters through the blood–brain barrier 18. In CSF and brain tissue, the concentration of 6‐MP is very low in comparison with that in unbound plasma, indicative of restricted distribution, due in part to the efflux transport system in the blood–brain barrier 19. Because of these pharmacokinetic factors, we postulate that 6‐MP altered leucocyte trafficking into the CNS 20. Accordingly, reduction of circulating NK cells would reduce the pool of NK cells available to permeate into the CNS, and restoration of the CD4/CD8 ratio was achieved by an increase in CSF CD4⁺ T cells. However, mechanistic studies are indicated.

Our data support the hypothesis that the CSF immunophenotype is indeed responsive to treatment, and that the changes have clinical relevance. Reduction in the percentage of IFN‐γ‐producing helper/inducer T cells in CSF may be beneficial, given evidence that such Th1 cytokine production is typically proinflammatory 21. Similarly, restoration of a more normal ratio of helper/inducer to cytotoxic/suppressor T cells makes sense as a therapeutic goal, given that the ratio improves during the course of the disease in individuals who are clinically improved and not relapsing (unpublished observations). These data support the use of CSF lymphocyte phenotyping in OMS as an immunological ‘staging’ procedure upon which to base therapeutic decisions.

The capacity of 6‐MP to lower NK cells in CSF and blood is a novel property among immunotherapies that we have evaluated. Both the relative and absolute size of the NK pool were reduced in blood. We have not encountered that effect after treatment in OMS with cyclophosphamide 16 or mycophenolate mofetil 22. What is also noteworthy is that blood and CSF NK changes mirrored each other. The blood lymphocyte phenotype in OMS is not usually predictive of most abnormalities found in CSF 2. Thus far, the NK cell has not been implicated in the pathophysiology of OMS, and functional studies would be necessary to explore it further. In some autoimmune disorders there is reduction of peripheral blood NK cell activity 23. Effects of 6‐MP on NK cells may have therapeutic benefit in other autoimmune disorders, such as multiple sclerosis, in which CSF NK cell counts (not frequency) are increased 24.

The ALC was a more useful and less expensive index of improvement in the CSF lymphocyte phenotype in OMS than the 6‐TPN concentration. The role of monitoring and optimizing thiopurine medication therapy through thioguanine metabolite determinations is controversial. However, adverse effects can be a limiting problem, leading to drug discontinuation in up to 25% of 6‐MP‐treated patients 25. The goal of monitoring is to identify medication toxicity and non‐adherence 26, 27. Repeat TGN measurements are required for an unambiguous index of active metabolite exposure 28. We found that serial monitoring of thioguanine nucleotide metabolites helped guide our decisions about when to push the dose or to retreat from safety considerations, but the tests are expensive, not always covered by insurance and, for some families, prohibitive. According to Prometheus Laboratories, the 6‐TPN and 6‐MMPN reference ranges are not specifically paediatric. The fact that 6‐MP was able to induce clinical benefit in the group as a whole suggests a sufficient pharmacodynamic was delivered regardless of a lower therapeutic blood level or ALC in several patients, lending credence to the view that target ‘therapeutic levels’ specific to the paediatric population are needed. The high intrapatient variability in TGN production over time has led some to the conclusion that TGN measurements should not be advocated for routine clinical use 28.

How 6‐MP exerts its pharmacodynamic effects on the immune system is not entirely known. The biochemical pharmacology of anti‐metabolite thiopurine drugs suggests that immunosuppression is mediated by 6‐thioguanine nucleotide metabolites through a combination of anti‐metabolic and pro‐apoptotic actions 7. In‐vitro studies found an inhibitory effect on activated T cells and tumour necrosis factor‐related apoptosis‐inducing ligand protein expression 29. In murine microglial cultures, 6‐MP attenuated TNF‐α production, and potential therapeutic use to down‐regulate inflammation mediated by microglia was proposed 30. 6‐MP also inhibits atherosclerosis in mice by reducing local CCL2 and macrophages 31; affords neuroprotection to focal cerebral occlusion in rats 32; and regulates the Bcl‐2/Bax ratio 33, transcriptional activity at the nuclear receptor NR4A level 34, 35 and Rac1‐mediated signalling 36, 37. In OMS, we did not find a reduction in CSF T cell activation, but our patients did not manifest the very high percentages we often encounter 2.

6‐MP is not for all children with OMS. Striking interindividual differences in pharmacokinetics have been reported and attributed to highly variable bioavailability of orally administered 6‐MP 38. In our study, pharmacokinetic factors limited the number of children who could achieve a reduction in the ALC, a therapeutic blood level or stay in the non‐toxic reference range without side effects. The effect of the drug on CSF B cell expansion, which was outside the purview of this study, is not established. 6‐MP may have a place as a steroid sparer in the pharmacotherapy of some patients who have persistence of OMS or relapse with a reduced CSF CD4/CD8 ratio and/or an increased percentage of CSF IFN‐γ⁺ helper/inducer T cells.

This study had strengths and limitations. Each patient was evaluated carefully both clinically and immunologically by experienced OMS experts. Documentation of the laboratory endpoints reached assessed adequacy of the drug dose. All immunological studies were performed by a single flow cytometry laboratory for consistency using a published protocol. There were sufficient patients and data to allow clinical–immunological comparisons and correlations. A limitation was the small sample size, given the rarity of the disorder, the variety of immunological agents already used to treat OMS and the retrospective nature of treatment analysis. Lymphocyte frequencies describe immune cell proportions, which is not the same as measuring absolute cell counts. Because patients with B cell expansion were treated with anti‐B cell therapy instead, no extrapolation of 6‐MP effectiveness can be made for de‐novo OMS. Immunologically active drugs can exert many effects on the immune system, including those not measured in a study, so we cannot infer that the studied parameters were the only ones relevant to the clinical outcome. However, these data should aid the design of a prospective study to compare promising steroid sparers for the treatment of OMS.

Conclusions

The use of 6‐MP as a steroid sparer in children with OMS showed some promising pharmacodynamic properties and some limiting pharmacokinetic ones. Abnormalities of the CSF immunophenotype are the known cellular biomarkers of disease activity in OMS, therefore drug effects on them are of interest. Also, some of the 6‐MP effects on CSF lymphocyte subsets, which differed from those of cyclophosphamide and mycophenolate mofetil, may have applications for other neuroinflammatory disorders. The main limitations of 6‐MP are pharmacokinetic, sometimes making it difficult or impossible to reach a therapeutic level or sufficiently lower absolute lymphocyte counts without engendering laboratory indications of toxicity or side effects. Although 6‐MP is suboptimal as a general steroid sparer in OMS, it has a role in some symptomatic children with OMS who have not responded fully to other immunotherapies and can tolerate it.

Disclosure

There were no competing interests.