Cerebrospinal fluid markers in incident pediatric-onset multiple sclerosis: a nationwide study

To investigate whether cerebrospinal fluid (CSF) markers differ between pediatric-onset multiple sclerosis (PoMS, onset < 18 years) and adult-onset (AoMS), and whether these markers are associated with clinical outcomes among PoMS. Prospective nationwide registry study of incident MS, including persons with a CSF sample < 3 years post-MS onset. We compared CSF oligoclonal band (OCB) status, immunoglobulin G (IgG) index levels, and mononuclear cell count between PoMS and AoMS. Within the PoMS cohort we analyzed the association between CSF markers, relapse rate and Expanded Disability Status Scale (EDSS) score, using negative binomial regression and generalized estimating equations, respectively. The cohort consisted of 130 PoMS and 3228 AoMS cases. The PoMS group had higher odds of OCB-positivity (odds ratio: 2.70; 95% CI 1.21–7.67). None of the CSF markers were associated with relapse rate in the PoMS cohort; however, OCB-positivity was associated with higher EDSS scores. This study suggested that PoMS more commonly display CSF evidence for intrathecal IgG production than AoMS. Further, we found evidence of a relationship between OCB-positivity and subsequent disability, suggesting that they could play a role in the prognostication of MS in children.

www.nature.com/scientificreports/ age of 18 years or older. Due to the rarity of primary progressive MS in the pediatric population, and to ensure comparability between groups, we selected only relapsing-onset patients in both cohorts.
Study exposures and outcomes. We analyzed CSF oligoclonal band (OCB) status, immunoglobulin G (IgG) index levels, and mononuclear cell count in the full cohort. These outcomes were selected because they are routinely collected in the neurology clinic, and therefore of clinical relevance. Detailed clinical information was collected in the SMSreg prospectively, including information on relapses and neurological disability. Relapses are recorded by the neurologist as new or recurrent neurologic symptoms not associated with fever or infection that lasted for at least 24 h and were accompanied by new neurologic signs. The expanded disability status scale (EDSS) 15 is performed in the clinic on an approximately annual basis by an MS-specialist neurologist. The scale ranges from 0 (no disability) to 10 (death due to MS) 15 .
Statistical analysis. Hypotheses were formulated before the registry data extraction. Continuous variables were compared using the student's t-test, or Wilcoxon-rank sum test. Categorical variables were compared by the chi-squared test or the Fisher's exact test. OCB status is a binary outcome (positive vs negative), but we also categorized the continuous CSF outcomes as "high" vs "normal" using established normal ranges in Sweden (IgG index normal range ≤ 0.70; mononuclear cell count normal range < 5 × 10 6 /μl). We used logistic regression analysis, with results presented as odds ratios (OR) and 95% confidence intervals (95% CI), adjusted for sex and region of Sweden (North, Central, South). Within the PoMS cohort, we explored the relationship between CSF markers and relapse rate in the 5 years following the baseline date. We used negative binomial regression, as the relapse rate was an over-dispersed count variable. Covariates that were considered included age at baseline, sex, region of Sweden, relapse status, and disease-modifying therapy (DMT) exposure, with time from baseline included as an offset. DMTs were classified as moderate-efficacy (interferon-beta, glatiramer acetate, dimethyl fumarate, teriflunomide) or highefficacy (rituximab, ocrelizumab, natalizumab, fingolimod, alemtuzumab, mitoxantrone, and cladribine) 16,17 , and analyzed as time-varying covariates in the 5 years post-baseline. Findings were reported as incident rate ratios (IRRs) with 95% confidence intervals.
EDSS was similarly analyzed in the 5 years following baseline. We used generalized estimating equations, with an exchangeable correlation structure in order to use all EDSS scores recorded during this period for each individual. The EDSS score was modeled as a linear variable, adjusting for age at onset, time from baseline, sex, region of Sweden, and time on first-and second-line DMTs, with results presented as beta-coefficients and standard deviations (SD). Given that adjusting for multiple reduction reduces the type I error for null associations and increases the type II error for those associations that are not null, we did not control for multiple comparisons in this study. Statistical analyses were performed using R: a Language and Environment for Statistical Computing v.4.0.2 (R Foundation for Statistical Computing, Vienna, Austria; 2020).
The study was approved by the Regional Ethical Review Board of Stockholm, and informed consent was provided from patients for the collection of their clinical information. All methods were performed in accordance with the relevant guidelines and regulations, and informed consent was obtained from all participants and/ or their legal guardian.

Results
Between 2001 and 2019, 429 PoMS and 7842 AoMS had a lumbar puncture (LP) recorded in the SMSreg. Among whom, 130 incident cases of PoMS and 3228 AoMS had the LP performed within 3 years of their disease onset. There were no differences between the groups in terms of sex ratio, disease course, DMT or steroid exposure at baseline, nor region of Sweden (Table 1). The PoMS baseline ages ranged from 6 to 17, but the cohort was comprised predominantly of adolescents (≥ 13 years; 93%). Less than 5% of the full cohort (143/3358) were exposed to a DMT at the time of the CSF analysis.
When exploring crude CSF levels, we found that the pediatric cohort had a higher proportion of OCB-positive individuals, IgG index, and mononuclear cell count compared to the adult-onset cohort, though only OCB status and IgG index were statistically different between groups (Table 2).
When adjusted for sex and region of Sweden (relapse status was non-significant and not included in the models), the PoMS group were found to have 2.70 (95% CI 1.21-7.67) times higher odds of detectable OCB bands in their CSF (Table 3). When the continuous outcomes (IgG index and mononuclear cell count) were dichotomized into "normal" or "high" values we found no differences between groups.
In the 5 years following baseline (median follow-up time: 5.0 years [interquartile range: 5.0-5.0 years), 48% of PoMS patients experienced no relapses, while 22% had one, and 30% had more than one recorded. None of the CSF markers were associated with subsequent relapse rate (Table 4).
Of the full cohort, 118 (90.8%) of the PoMS patients had at least one EDSS score recorded within 5 years of baseline and were included in the analysis. The median number of EDSS scores recorded was 5 (IQR: [3][4][5][6][7][8]. At a population level, the cohort had mild disability (median EDSS: 1.0; IQR: 0.0-2.0) during this period. OCBpositivity within the first 3 years of disease was associated with a higher EDSS score, by 0.72 points (SD: 0.27), on average. Neither IgG index level nor mononuclear cell count were associated with subsequent disability levels ( Table 5).

Discussion
In this nationwide cohort study of cerebrospinal fluid markers in pediatric-onset MS, we found notable differences between PoMS and AoMS, and evidence of a link between OCB-positivity and MS disability early in disease among PoMS. The PoMS cohort had a higher proportion of OCB-positive individuals, and IgG index than the www.nature.com/scientificreports/ AoMS cohort. While we found no evidence of a relationship between CSF and relapse rate in PoMS, we did find that having OCBs corresponded with higher EDSS levels within 5 years, suggesting that this marker may offer some prognostic utility among persons with pediatric-onset MS.  Table 3. Odds ratios and 95% confidence intervals for the odds of having "positive" or "high" CSF values in the pediatric-onset patients compared to the adult-onset patients (reference group). a Adjusted for sex and region of Sweden.

Oligoclonal bands
Positive ( www.nature.com/scientificreports/ CSF was first analyzed within the context of inflammatory nervous system diseases in the 1950s and 1960s when both IgG and OCBs were documented in MS patients 18,19 . OCBs represent measurable IgG antibody within the CSF in the absence of corresponding IgG in the serum. Our study suggests that patients with pediatric MS are more likely to be OCB-positive than adult-onset patients. We found higher levels of IgG index, but comparable levels of mononuclear cell counts between PoMS and AoMS. Previous research has mostly explored PoMS and AoMS separately, but three studies compared the groups [20][21][22] . Most recently, a single-centre study from the Netherlands reported a high proportion of OCB-positive individuals in the pediatric group (mean age of 15; 91.3% in PoMS vs 87.8% in AoMS) 20 . Two studies from the same clinic in Germany, in which early-onset MS was defined as < 16 years of age, compared PoMS and AoMS 21,22 and reported no differences in terms of OCB status nor IgG index 21,22 . The PoMS cohort was younger than the current cohort, with a median age of 12.7 years. It's possible the different findings relate to the age structures of the PoMS cohorts. A 2010 multi-centre study of CSF in PoMS found age-related differences, such that younger children were less likely to exhibit OCBs and elevated IgG indices 23 . The age structure of our cohort precluded us from running a similar analysis, as too few children (< 13 years of age) were included.
Previous efforts to study the relationship between CSF markers and MS disability in PoMS have concentrated on MS risk following an initial demyelinating event [5][6][7] . Consistently, the evidence has shown that OCB-positive children are much more likely to develop MS than OCB-negative children [5][6][7] . In the adult MS population, there is a wider literature regarding OCBs and disability, summarized in a 2013 meta-analysis 4 . The pooled analysis of four studies with EDSS disability milestones as the outcome, found an OR of 1.96 (95% CI 1.31-2.94) comparing OCB-positive to OCB-negative individuals 4 . While our study suggests that OCBs may also be associated with worse MS disability, the number of OCB-negative patients was quite low, therefore this analysis should be replicated on larger sample sizes if possible.
Prior research has suggested a relationship between high IgG levels and MS progression, but the focus has largely been on adult cases [24][25][26] . A single study of children with MS found no evidence of a relationship between elevated IgG index and time to second event 23 , similar to our findings.
Mononuclear cells were elevated in a high proportion of the PoMS cohort, but they were not significantly different from the AoMS, nor were they related to relapse rate or disability.
A strength of our study was the inclusion of incident cases, the vast majority of whom were treatment-naïve, thereby limiting the potential confounding effect of DMT exposure on CSF levels. The SMSreg is a populationbased nationwide registry, containing a large number of pediatric patients, followed prospectively. Not all patients enrolled in the SMSreg have CSF values recorded, and fewer have values available within 3 years of their MS onset. It is possible that these patients differ from persons who did not have an LP early in their disease. This should not have affected the analysis comparing AoMS and PoMS, but our full cohort may not be representative of the wider MS population, including persons who are not examined so early in their disease. As always, the true date of MS onset is challenging to elucidate. While we only included CSF values taken within 3 years of reported MS onset, it is possible that the 'true' date of MS onset may have preceded this time. This imprecision is unlikely to have affected one group differentially, and therefore should not have biased our results. Last, we modelled the ordinal EDSS as a continuous variable, and did not have the statistical power to stratify based on relapse status. A difference on the EDSS has a clinically different meaning depending on where on the scale it Table 4. Relationship between CSF values and relapse rate among pediatric-onset patients in the 5 years following baseline, using negative binomial regression. IRR incident rate ratio. a Adjusted for age at sampling, sex, region of Sweden, and time on first-and second-line DMTs.  www.nature.com/scientificreports/ occurs. The PoMS cohort were largely on the low end of the disability scale, as the analyses were limited to the first 5 years of disease, thereby minimizing this potential imprecision. This study suggested that PoMS tend to more commonly have CSF evidence for intrathecal IgG production than AoMS, including OCBs. Further, we provide evidence for a relationship between incident OCB status and subsequent disability, suggesting that it could play a role in the prognostication of MS in children. Whether active lesions are a component of this relationship in pediatric patients, as observed among adults 27 , remains to be seen. Future studies should aim to incorporate MRI data to better understand these dynamics.

Data availability
Data related to the current article are available from Jan Hillert, Karolinska Institutet. To share data from the Swedish MS Registry, a data transfer agreement needs to be completed between Karolinska Institutet and the institution requesting data access. This is in accordance with the data protection legislation in Europe (GDPR [General Data Protection Regulation]). Persons interested in obtaining access to the data should contact Jan Hillert (jan.hillert@ki.se).