Article Text

Research paper
Long-term outcome of paediatric-onset multiple sclerosis: a population-based study
  1. Katharine E Harding1,2,
  2. Kate Liang2,
  3. Mark D Cossburn2,
  4. Gillian Ingram2,
  5. Claire L Hirst2,
  6. Trevor P Pickersgill2,
  7. Johann Te Water Naude3,
  8. Mark Wardle2,
  9. Yoav Ben-Shlomo4,
  10. Neil P Robertson1,2
  1. 1Institute of Psychological Medicine and Clinical Neuroscience, Cardiff University, University Hospital of Wales, Cardiff, UK
  2. 2Department of Neurology, University Hospital of Wales, Cardiff, UK
  3. 3Department of Paediatrics, University Hospital of Wales, Cardiff, UK
  4. 4School of Social and Community Medicine, University of Bristol, Bristol, UK
  1. Correspondence to Professor Neil P Robertson, Institute of Psychological Medicine and Clinical Neuroscience, Cardiff University, University Hospital of Wales, Heath Park, Cardiff CF14 4XN, UK; robertsonnp{at}


Background Age of onset of multiple sclerosis (MS) peaks in the 3rd and 4th decades and is rarely less than 18. Robust longitudinal studies in paediatric-onset MS (POMS) are limited, and a clearer understanding of outcome could optimise management strategies.

Methods Patients with disease onset <18 years were identified from a prospective population-based register. Clinical features including presenting symptoms, time to Expanded Disability Status Scale (EDSS) 4.0, 6.0 and 8.0 and onset of secondary progression were compared with patients with adult-onset MS (AOMS).

Results 111 POMS patients were identified from a cohort of 2068. No significant differences in sex ratio, familial recurrence, relapse rate, ethnicity or clinical symptoms at presentation were identified between POMS and AOMS. However, interval to second relapse was longer (5 vs 2.6 years, p=0.04) and primary progressive disease was less common (0.9% vs 8.5%, p=0.003) in POMS than in AOMS. POMS patients also took longer to develop secondary progressive disease (32 vs 18 years, p=0.0001) and to reach disability milestones (EDSS 4.0, 23.8 vs 15.5 years, p<0.0001; EDSS 6.0, 30.8 vs 20.4 years, p<0.0001; EDSS 8.0, 44.7 vs 39 years, p=0.02), but did so between 7.0 and 12 years younger than in AOMS.

Conclusions 5.4% of patients with MS have POMS (2.7% <16 years; 0.3% <10 years) and have distinct phenotypic characteristics in early disease. Furthermore, while patients with POMS take longer to reach disability milestones, they do so at a younger age than their adult counterparts and could be considered to have a poorer prognosis. Management strategies for these patients should take account of these data.

Statistics from


Multiple sclerosis (MS) is a chronic neuroinflammatory disease that can present at almost any age but rarely under 18 or over 50 years of age.1 Studies of selected cohorts at extremes of age allow insights into disease pathogenesis, highlight avenues of novel research, and inform clinical practice. In addition, early diagnosis and recognition of particular clinical problems associated with paediatric-onset MS (POMS) is of importance for long-term management and patient well-being. However, specific difficulties for studies of POMS include case definition2 ,3 and low disease frequency.4 ,5 As a result, direct comparison of studies has been problematic and tended to concentrate on short-term outcomes in highly selected clinic-based populations5–9 rather than long-term outcome in representative cohorts. Available data on prognosis therefore remains limited, and there is little agreement on specific characteristics, including clinical features,5 ,10–12 prognosis,5 ,6 atypical presentations,6 ,12 ethnicity11 ,13 and early relapse frequency.5 ,6 ,8 A recent consensus statement has defined age at onset in POMS as <18 years,14 facilitating more direct future comparisons.

This study explores clinical features and accumulation of disability in POMS in a representative, population-based cohort with long-term follow-up, making comparisons with a cohort of patients with adult-onset (AOMS) disease.


Data collection

The neuroinflammatory service at the University Hospital of Wales operates a regional service for South-East Wales, providing dedicated, consultant-led, multidisciplinary clinics and rapid-access clinics supplemented by a network of community-based nurse clinics. The service covers a geographical region incorporating the major urban centres of Cardiff and Newport, the former industrial centres of the South Wales Valleys, and surrounding semi-rural areas, with an estimated population of 1.3 million,15 with contemporary disease prevalence recently estimated at 146 per 100 000 population.16 Patients with MS in this region have been well characterised, as a prevalence study in 198517 and data collection via networked clinics and periodic regional re-surveillance ensures the clinic sample remains representative of the resident population. Data are collected prospectively in a standardised manner and stored in a secure database, including demographic details, a full medical history including comprehensive history of disease onset and course, examination findings including dated Expanded Disability Status Scale (EDSS)18 assessments, details of investigations and family history. To date, this database includes information on more than 3000 individuals. Information including symptoms at first presentation and extent of recovery from first event is recorded using an adapted European Database for Multiple Sclerosis (EDMUS) system,19 using specific symptom codes. Onset syndromes are defined as being monoregional if symptoms were attributable to a single neuroanatomical location and polyregional if not. This study has been approved by the South East Wales Ethics Committee (reference No 05/WSE03/111).


Patients with a diagnosis of MS were identified from the South-East Wales regional database. To qualify for inclusion, patients must have been assessed and diagnosed by a neurologist with a specialist interest in MS using criteria appropriate to the period of diagnosis.20–22 Patients with alternative neuroinflammatory diagnoses, clinically isolated syndromes without adequate para-clinical evidence for a diagnosis of MS, or incomplete datasets including details of disease onset were excluded. Age at onset was calculated for each patient, and patients were classified as (a) POMS disease (<18 years), (b) AOMS disease (18–49 years) or (c) late-onset MS disease (>50 years). Clinical information on sex, ethnic group, familial disease recurrence, presenting symptoms, disease course and time to disease milestones, including time to EDSS 4, 6, 8 and secondary progression, were identified for each patient with either POMS or AOMS and used for further analysis and comparison.

Statistical analysis

Statistical significance was calculated using Student's t test or Wilcoxon's signed-rank test for continuous variables, and χ2 test or Fisher's exact test for categorical variables. Kaplan–Meier survival analysis was used to explore time elapsed to defined disease milestones, and differences between AOMS and POMS cohorts were tested using the log-rank test. Cox regression was used to analyse contribution of demographic factors and features of early disease to rate of accumulation of disability within the POMS cohort. Statistical analysis was performed using R v.


Patient identification

Records from 3048 patients registered with the regional neuroinflammatory services since 1985 were examined. A diagnosis of MS was established for 2654 patients, 2068 of which fulfilled recognised diagnostic criteria for MS,20–22 had been reviewed by a study clinician, followed prospectively, and had complete datasets from disease onset. Of these, 23 (1.1%) had been registered with the service before the first cross-sectional survey of MS in the region in 1985, and 142 (6.9%) had been registered with the service between 1985 and 1999, during which time additional periodic cross-sectional studies were performed.24 ,25 After 1999, data were collected in a systematic longitudinal fashion. Age of disease onset was <18 years for 111 patients (5.4%; 95% CI 4.4 to 6.2; female/male ratio 2.9:1), >18–50 years for 1825 patients (88.2%) and >50 years for 132 patients (6.7%). The paediatric cohort had been followed for a mean of 25.3 years (SD 15.8), and the adult cohort for 17.2 years (SD 12).

Paediatric-onset patients

Demographic features

Age of disease onset within the paediatric cohort ranged from 4 to 17 years (mean 15; median 15.6) (figure 1). Seven patients (6.3%; six female, one male) had disease onset at <10 years of age, 51 (45.9%; 39 female, 12 male) between 10 and <16, and 53 (47.7%; 37 female, 16 male) from 16 to 18. Median age at onset was lower in female (14.7) than male (15.7) patients, but this difference was consistent with chance (p=0.12). Most patients (107; 96.4%) were of Caucasian origin. Only four patients were from other ethnic groups: two (1.8%) Afro-Caribbean, and two (1.8%) Asian-Indian. Fourteen patients (12.6%) had an additional family history of demyelinating disease, comprising two (1.8%) with an affected father, two (1.8%) with an affected brother, two (1.8%) with an affected daughter, four (3.6%) with an affected aunt, one with an affected grandmother, and the remainder with more distant affected relatives. Thirty-eight patients (34.2%) had received disease-modifying treatment at some point during their disease. The mean age at which disease-modifying therapy was started was 25.9 years, and the mean interval from onset of disease to the start of disease-modifying therapy was 11.7 years. However, it is worth noting that disease-modifying therapies were only widely available in Wales from 2002; of POMS patients diagnosed since that time, 60% have received treatment.

Figure 1

Bar chart showing age at onset of paediatric cases, for all cases and by gender.

Index event and initial disease course

In 110 patients, disease onset was relapsing, and only one patient had a progressive disease course from onset. Of the 110 patients with relapsing onset disease, 93 (83.8%) recovered completely from the index event, but 17 (15.3%) had persistent neurological sequelae. Motor symptoms were the most common initial manifestation of disease (52.8%) followed by optic neuritis (26.4%). There was no significant variation in symptoms at presentation by gender (p=0.86) or age group (p=0.11). Most patients (76.4%) presented with monoregional symptoms.

Differential diagnosis

In four patients, the initial disease course was considered atypical for MS, and an alternative diagnosis was suggested. In two patients, acute demyelinating encephalomyelitis (ADEM) was considered, and MS was only diagnosed after a period of more than a year during which symptoms and radiological findings evolved. Two patients (including the only patient with primary progressive disease) were investigated for neuromyelitis optica following episodes of optic neuritis with poor recovery, but aquaporin 4 antibodies were negative and subsequent investigations and clinical course confirmed a diagnosis of MS.

Paraclinical findings

Ninety per cent of patients with POMS had cranial MR changes typical of MS on initial investigation. The remainder did not fulfil radiological criteria for MS, comprising two with a scan suggestive of inflammatory disease but not fulfilling radiological criteria, two with a scan initially suggestive of ADEM, and two with a normal scan. A diagnostic lumbar puncture was performed in 46% of POMS cases, of which 93% demonstrated positive oligoclonal bands. Median time from disease onset to diagnostic lumbar puncture was 5.2 years (IQR 2.1–12.6). Thirty per cent of patients had visual evoked potentials, with unilateral delay identified in 61% and bilateral delay in 16%.

Disease milestones

One-third of patients entered a secondary progressive phase of disease during the course of the study (SPMS). Median time to reach SPMS was 32 years, and median age at onset of secondary progression was 44.7 years. Median time to EDSS 4, 6 and 8 from disease onset was 23.2, 30.8 and 44.7 years, respectively, and median age at which the POMS patients reached these milestones was 40.1, 45.4 and 54.8 years (figure 2).

Figure 2

Kaplan–Meier survival curves for time to, and age at, disability end points for adult-onset cohort (solid line) and paediatric-onset cohort (dotted line), with median times marked. EDSS, Expanded Disability Status Scale; SPMS, secondary progressive multiple sclerosis; y, years.

Prognostic features

Incomplete recovery from initial event was significantly associated with a shorter time to EDSS 4 (HR 2.44, 95% CI 1.20 to 4.95, p=0.01) (figure 3). However, sex, age at onset, neurological symptoms at onset and polyregional onset were not associated with time to EDSS 4. The associations between clinical variables and time to secondary progression were consistent with random variation (table 1).

Table 1

Results of Cox regression model for the paediatric cohort for time to EDSS 4 and secondary progressive disease

Figure 3

Time to Expanded Disability Status Scale (EDSS) 4 for paediatric-onset cohort, grouped by degree of recovery from initial event.

Comparison with adult-onset disease

No significant differences between the AOMS and POMS cohorts were identified in proportion of patients from ethnic minorities (3.1% in AOMS vs 3.6% in POMS; p=0.99), sex ratios (female/male 2.4:1 in AOMS, 2.8:1 in POMS; χ2=0.66, p=0.42), family history of demyelination (12.1% in AOMS vs 13.3% in POMS; χ2=0.013, p=0.91), or anatomical site of index event (χ2=2.74, p=0.60). In addition, similar proportions of patients had monoregional and polyregional disease onset (χ2=0.0042, p=0.95). Median time from onset to diagnosis was longer in POMS (POMS 5.8 years, IQR 2.0–14.1; AOMS 2.0 years, IQR 0.6–6; p<0.0001). Fewer adult-onset patients recovered from the initial event: 61.9% in AOMS vs 83.8% in POMS, although this was of borderline significance (χ2=3.65, p=0.06). Disease-modifying therapy had been received by 441 (24.1%) AOMS patients; this was not significantly different from the proportion of POMS patients who were treated (p=0.13). Thirty-seven per cent of AOMS patients who were diagnosed after disease-modifying therapies were available in Wales in 2002 have received disease-modifying therapy; this is significantly less than the proportion of the POMS cohort receiving disease-modifying therapy diagnosed after 2002 (p=0.006). Mean time from onset of disease to starting disease-modifying therapy was shorter in the AOMS group than in the POMS group (7.3 years, p=0.02), and patients were older when they started disease-modifying treatment (37.7 years, p<0.0001). Demographic characteristics are summarised in table 2.

Table 2

Comparison of demographic characteristics of POMS and AOMS cohorts

There was no difference in annualised relapse rates (ARRs) between AOMS and POMS patients (0.44 in AOMS, 0.44 in POMS, p=0.92). In the first 5 years of disease, POMS patients had fewer relapses (median ARR 0.2 in POMS and 0.4 in AOMS; p=0.05), but there was no difference in ARR over the first decade of disease (median ARR 0.2 in POMS and 0.2 in AOMS; p=0.26). In addition, there was a significantly longer time from first to second event in the POMS cohort (2.55 years in AOMS, 5.02 years in POMS, p=0.04).

Time to, and age at, disability milestones differed significantly between AOMS and POMS. Median time to SPMS was 14 years longer (p=0.001) in POMS, but median age at onset of SPMS was 7 years younger (p=0.002). In addition, although median time to all EDSS milestones was significantly longer for POMS (EDSS 4, 23.8 vs 15.5 years (p=<0.0001); EDSS 6, 30.8 vs 20.4 years (p=<0.0001); EDSS 8, 44.7 vs 39.0 years (p=0.02)), patients with POMS reached these milestones at a significantly younger age (EDSS 4, 40.1 vs 49.6 (p=<0.0001); EDSS 6, 45.4 vs 53.9 (p=<0.0001); EDSS 8, 54.8 vs 66.8 (p=0.0004)) (figure 2).


By using an established and robust population-based epidemiological framework which has identified and documented incident cases of MS in south Wales since 1985, we have been able to identify 111 patients with POMS from a cohort of 2068 patients with MS, representing 5.4% (95% CI 4.4 to 6.4%) of patients within this region of the UK. In addition, 2.7% (95% CI 1.8 to 3.2%) had disease onset <16 years of age and 0.3% (95% CI 0.1 to 0.7%) under the age of 10. Although there was no overall difference in the sex ratios between POMS and AOMS, only one male patient was identified with onset under the age of 10 years, and female patients had a lower median age at onset. All but one patient had relapsing onset disease, and 84% of patients recovered completely from their first clinical demyelinating episode. No differences in the proportion of patients with a family history of MS or from ethnic minorities were observed.

Low disease frequency and variable applications of age limits make determining the frequency of POMS problematic.4 Few conventional epidemiological studies of disease prevalence have sufficient paediatric-onset cases to provide narrow, informative CIs. Most studies of paediatric-onset disease therefore rely on clinic-based registers as a source population. One of the largest studies of this kind originated from a French and Belgian consortium and identified 394 cases with disease onset under the age of 16 in a registered multicentred clinic-based population of 17 934 patients (2.2%).6 A similar study from the University of British Colombia MS clinic found 129 cases with onset <16 years during a 19-year period, representing 3.6% of the total MS case load,5 and disease registers from four Italian centres identified 149 (3.8%) cases in a register population of 3375.12 More recently, an MS centre in the north-eastern USA reported a frequency of 3.06% (135/4399) using age at onset of <18 years as the inclusion criterion,11 and a recent study in Iraq of patients with onset <18 years found a frequency of 6.84%.10 One of the highest prevalence figures observed was derived from an Italian hospital-based cohort, which found 83 of 710 patients with onset <16 years (11.7%).8 Where clinic-based cohorts are representative of the local population, a figure of approximately 2.5% for patients below the age of 16 and 5% for those below 18 years seems to be a consensus and is concordant with data from our population-based study. Higher figures identified in some studies are likely to reflect case mix in specialist services. Case ascertainment may be more difficult, as paediatric patients might also be preferentially referred to general paediatric services rather than specialist MS services. In our population, the paediatric and adult MS services are integrated, so there is less chance of ascertainment bias. In addition, it should be recognised that classification inaccuracies may occur through the retrospective nature of studies that assign paediatric-onset status from historical data. Case series derived from cross-sectional data may be particularly liable to this rather than prospectively collected data.

Natural history studies of POMS have also been hampered by small patient numbers and limited follow-up in a disease that evolves over decades. In the majority of studies examining outcome in POMS, the EDSS has been used, with identification of key disability milestones to facilitate the inclusion of some retrospective data. Some studies have reported on cross-sectional data with proportion of patients achieving standardised disability levels which may be prone to selection bias,12 ,10 ,26 but only a few have analysed systematically collected longitudinal disability data, and, of these, only the most recent have used appropriate survival analyses to account for variability of follow-up.5 ,6 ,8 No systematic studies examining prognosis have been performed since the introduction of the consensus statement in 2007,14 and studies before that have concentrated on the younger age groups.5 ,6 ,8 All have used an AOMS cohort as a comparator. Analysis of disability data has suggested a benign disease course so that, in a Canadian cohort of 125 patients, 51 patients had disability data at 10 years, of which 60% had an EDSS of <3.0, 24% between 3.0 and 6.5, and 16% of 7.0 or more.27 However, many studies of small cohorts of POMS also identified that rapid evolution of disability occurred in a proportion of patients.12 ,10 ,26 Our data suggest that, while prognosis may be considered more benign in early disease, as reflected by an enhanced ability to recover from acute inflammatory events and a greater interval to second relapse, later stages of the disease trajectory are more similar to those in AOMS and lead to an earlier age-related accumulation of disability.

Three previous studies (table 3) have used similar survival analysis to our study but have identified cohorts of POMS with age at onset of <16 years of age.5 ,6 ,8 Despite the application of appropriate statistical methodology, all remain limited by the relatively small proportion of patients reaching the higher disability end points. All studies identified a longer time to disability end points when compared with AOMS. However, even within these studies, there is considerable variability in outcome, with time to EDSS 6 ranging from 19.4 to 30.8 years and time to onset of secondary progression from 16.1 to 32.0. What is clear from all these studies is that age at which these disability end points were achieved is consistently younger, so that, while previous observers may have suggested that POMS has a relatively benign disease course, its age-related outcome is poorer.

Table 3

Comparison of time and age to EDSS or DSS end points in studies of POMS using survival analysis

It has been suggested that prognosis in non-Caucasian patients is less good28 and may be reflected by over-representation of ethnic minorities in younger age groups, more rapid accumulation of disability, or enhanced early relapse rates, the last of these characteristics having already been reported in Afro-Caribbean children.29 In our study, no differences were found between the frequency of Caucasian and non-Caucasian patients in POMS and AOMS cohorts, and there were also no differences in disability profiles, relapse rates or age at onset, but our study may have been underpowered to detect this, as the number of ethnic minority patients was relatively small.

Recovery from first event was the sole clinical feature predicting time to accumulation of fixed disability in our study. Patients with residual disability after the first event reached EDSS 4 in a median time of 13.5 years, compared with 25.4 years for those with complete recovery, which is consistent with previous studies.8 ,30 Associations between gender, age at onset and neurological system involved at onset of disease, and time to EDSS 4 were consistent with chance variations. Furthermore, no features of early disease were associated with time to secondary progression. However, the number of patients reaching secondary progression was relatively small, and it is likely that this study is currently underpowered to detect an effect, as reflected in the wide CIs.

In addition, some studies have noted an increase in atypical features at presentation, including encephalopathy in younger patients,2 ,8 ,9 although direct comparison with ADEM suggests that atypical features are more likely in that disease.31 In our cohort, atypical features were found in only four patients.

Although numbers are small, there is a suggestion of a bimodal distribution, with a small cluster of patients having disease onset around the age of five, and then disease onset increasing from the age of nine onwards, which is similar to the pattern seen in the Canadian POMS cohort.5 This has not been described elsewhere, but may suggest alternative aetiologies for disease ignition in this small subset of POMS patients and may merit further investigation.

In the first detailed population-based study of POMS in the UK, we have used newly published consensus criteria14 together with a robust, geographically representative cohort of patients with MS to determine disease frequency and have identified distinct phenotypic characteristics of early phase disease. In contrast to some other investigators, we have failed to identify evidence for more active disease compared with AOMS, as reflected by ARR and time to second relapse. However, although POMS patients from our cohort took a significantly longer time to reach fixed levels of disability, they did so at a younger age than their adult counterparts, suggesting that, overall, age-related prognosis is worse. The main predictive factor for more rapid accumulation of fixed disability was degree of recovery from first event, but otherwise prognosis was unaffected by age, gender or neurological system involved at onset. This study provides epidemiological data to inform service provision for young patients with MS together with further evidence that age at onset acts as a modifying factor for the rate of accumulation of disability in MS. An improved understanding of the disease characteristics and trajectory of POMS may allow more timely and appropriate therapeutic intervention to reduce the long-term impact of MS for this vulnerable group of patients.


View Abstract


  • Contributors KEH wrote the paper, collected data and performed statistical analysis. KL collected data and drafted the paper. MDC collected data and drafted the paper. GI collected data and drafted the paper. CLH collected data and drafted the paper. TPP collected data and drafted the paper. JTWN collected data and drafted the paper. MW designed the study and drafted the paper. YB-S designed the study and drafted the paper. NPR designed the study, supervised the study, collected data and drafted the paper.

  • Funding Genzyme has supported a neuroimmunology research fellow within the department of neurology in Cardiff University since 2009.

  • Competing interests KEH has received a travel grant from Biogen-Idec. TPP served on scientific advisory boards of both Teva Pharmaceuticals and Biogen-Idec during 2009, has received delegate fees and travel expenses from Merck-Serono and Biogen-Idec and owned stock in Genzyme Inc from 2010 to 2011. No stock of any relevant company currently owned. YB-S has received publishing royalties from books, is a member of the Department of Health Scientific Advisory Group for the multiple sclerosis risk sharing scheme, and receives grant support from Cancer Research UK, The Tourettes Syndrome Association USA, British Heart Foundation, The Medical Research Council, The Wellcome Trust, Economic and Social Research Council, Bristol Research into Alzheimer's Care of the Elderly.

  • Ethics approval South East Wales Ethics Committee (reference No 05/WSE03/111).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.