Background Primary progressive multiple sclerosis (PPMS) is the least common MS disease course and carries the worst prognosis. In relapsing–remitting multiple sclerosis (RRMS) disability accumulation occurs in two distinct phases, but it is unclear whether this is also true for PPMS. Here we investigate factors associated with early and late disability accumulation in PPMS.
Methods We used Kaplan–Meier survival analyses and Cox regression to investigate the influence of sex, age at disease onset and onset symptoms on time to, and age at, Expanded Disability Status Scale (EDSS) 4 and 6, as well as the time from EDSS 4 to 6 in patients with PPMS.
Results We identified 500 patients with PPMS. The analyses on time to EDSS 4 included 358 patients, and those on time to EDSS 6 included 392 patients. The median times to EDSS 4 and EDSS 6 were 5 and 9 years. The analyses on age at EDSS 4 included 360 patients, and those on age at EDSS 6 included 402 patients. The median ages at EDSS 4 and EDSS 6 were 51 and 55 years. Older age at onset and bilateral motor onset symptoms were independently associated with a shorter time to both EDSS 4 and EDSS 6. Sex and other onset symptoms were not associated with time to, or age at, landmark disability. Only age at onset was significantly associated with the time from EDSS 4 to EDSS 6.
Conclusions Age at disease onset is the most important predictor of disability accumulation in PPMS. Bilateral motor onset symptoms were associated with quicker disease progression. In contrast to RRMS, we found no evidence for distinct phases of disability accumulation in PPMS. Disability accumulation in PPMS appears to be affected by the same factors throughout its course.
- MULTIPLE SCLEROSIS
Statistics from Altmetric.com
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.
Primary progressive multiple sclerosis (PPMS) differs from the more common relapsing–remitting multiple sclerosis (RRMS) in several key aspects. In comparison to RRMS, it begins later in life, (with a median onset age of around 40 years in comparison to an onset age around 30 years in RRMS) and proportionally affects more men (with a gender distribution of close to 1:1 compared to RRMS, where women outnumber men at a ratio of at least 2:1).1 ,2 The pathophysiology of PPMS is driven less by inflammatory demyelinating lesions, while diffuse axonal degeneration and cortical atrophy are more prominent.2 The natural history of PPMS is difficult to investigate because this disease course affects only about 10–20% of patients with MS, so that there are few natural history cohorts with sufficiently large numbers of patients. Maybe because of the difficulty in establishing representative cohorts, estimates of times to landmark disability differ significantly between studies, with estimates of the median time from disease onset to the landmark disability score of Expanded Disability Status Scale (EDSS)3 6 (the time where patients require a cane for walking) ranging from 7.14 to 14 years.5 Other factors that can contribute to the variability in estimates are the difficulty in establishing disease onset in PPMS, and inter-rater variability in assigning disability scores.
Large natural history studies from two French cohorts have shown that disability accumulation in RRMS is a two-stage process, with an ‘early’ phase (from disease onset to EDSS 3 or 4), and a ‘late’ phase (from EDSS 3 or 4 to EDSS 6); the influence of risk factors differs between these two phases.6 ,7 It is currently unknown if such disparity between the early and late disease phase exists in PPMS as well.
In this study, we investigate the natural history of PPMS, how potential risk factors influence the time to and age at landmark disability and whether the influence of these risk factors differs between the early and late disease phase in a large population-based patient cohort.
This study was a retrospective analysis of prospectively collected data. Patients were identified from the Calgary MS clinic database, which collects demographic (date of birth and sex) and disease-specific (date of disease onset, diagnosis, disease course and disability) data on cases of MS seen at the Calgary MS clinic since its inception in 1976. The Calgary MS clinic is the main MS referral centre in Southern Alberta with a wide catchment area including the cities of Calgary, Lethbridge and Medicine Hat and southeastern British Columbia. Historical data are obtained at the first visit and follow-up data are prospectively recorded at each clinic visit by the treating neurologist.
The cohort included all patients with a diagnosis of PPMS according to the Poser8 or McDonald9 diagnostic criteria, who attended the Calgary MS clinic before October 2012. Primary progressive disease course was defined clinically as a progressive course from onset.10 Sex, year of birth, year of disease onset, onset symptoms and year of first reaching EDSS 4 (patients are fully ambulatory despite severe disability) and EDSS 6 (patients require a cane for walking) were extracted from the database or determined from patient records. Age at onset was divided into four categories: <30, 30 to <40, 40 to <50 and ≥50 years. Symptoms in the functional systems ‘motor’, ‘sensory’, ‘visual’ and ‘cerebellar/brainstem’ were coded as either present or absent at disease onset. Motor and sensory onset symptoms were furthermore coded as occurring unilaterally or bilaterally.
The outcome measures in this study were the time from disease onset to EDSS 4 and EDSS 6; the age at EDSS 4 and EDSS 6; and the time from EDSS 4 to EDSS 6. Potential risk factors included sex, age at onset and onset symptoms (motor, sensory, visual and cerebellar/brainstem). Time to EDSS 4 was also examined as a risk factor for the time from EDSS 4 to EDSS 6 outcome, and grouped into three categories: 0–5, 6–10 and >10 years. The study was approved by the University of Calgary Research Ethics Board. All patients gave their written informed consent to the use of their data.
Median time to, and age at, landmark disability scores of 4 and 6 was calculated with Kaplan–Meier survival analyses. Survival curves were compared between groups with the log-rank-test. In order to estimate the independent relationship between the potential risk factors (sex, age at onset and onset symptoms) and the time to and age at landmark disability scores, we fit multivariable Cox regression models. Age at disease onset was not included in the models of age at landmark disability because the value of this variable is necessarily smaller than the outcome variable in every case. Patients who did not reach the outcome of interest before October 2012 were censored on the date of their last known disability score. Statistical significance was taken to be at the two-tailed 0.05 level. All statistical analyses and graphs were produced with the R statistical software package for Windows V.188.8.131.52
From a total cohort of 4862 patients, we identified 500 patients (10.3% of the total cohort) with PPMS. Informed consent could not be obtained in 54 patients, resulting in a cohort of 446 patients. The median length of follow-up from disease onset was 14 years (95% CI 13 to 15 years), the female to male ratio was 1.12 (236 women and 210 men). The time of reaching EDSS 4 could not be established in 86 patients, and the time of reaching EDSS 6 could not be established in 44 patients, resulting in sample sizes of 360 for the analysis of time to EDSS 4, of 402 for analysis of time to EDSS 6 and 310 for analysis of time from EDSS 4 to EDSS 6. The time of onset of MS could not be established in 30 patients, which further reduced the number of patients in the analysis of time to EDSS 4 to 358 and that in the analysis of time to EDSS 6 to 392. The median age at disease onset was 44 years (IQR 37–51 years) for the whole cohort (n=416). We performed additional analyses between patients included in the analyses and those excluded because of missing data. Patients who were excluded from the analyses on the timing of EDSS 4 and EDSS 6 were more often female (EDSS 4: 64% female vs 50% female Fisher's exact test p=0.03, EDSS 6: 68% female vs 51% female, p=0.02), and had a younger onset age (EDSS 4: 42 vs 44 years, Mann-Whitney U test p=0.01, EDSS 6: 34 vs 44 years and p=0.001), but no different distribution of onset symptoms. Figure 1 shows the overall survival curves for the time to, and age at, landmark disability scores. The median times from MS onset to EDSS 4 and EDSS 6 were 5 years (95% CI 4 to 6 years) and 9 years (95% CI 8 to 10 years); and the median ages at EDSS 4 and EDSS 6 were 51 years (95% CI 49 to 53 years) and 55 years (95% CI 54 to 56 years).
Table 1 shows the Kaplan-Meier estimates of the median time to, and age at, landmark disability scores. Age at disease onset was strongly associated with all four outcomes, with younger age at onset associated with a longer time to EDSS 4 (p=0.005) and EDSS 6 (p<0.0001), but with younger age at each landmark disability (p<0.0001). These associations remained significant after correction for multiple comparisons with a simple Bonferroni correction. The presence of motor onset symptoms was only moderately associated with the time to EDSS 4 (p=0.04, not significant after Bonferroni correction), however, the occurrence of bilateral motor onset symptoms was strongly associated with a shorter time to EDSS 4 (p=0.007, figure 2, this remains significant after Bonferroni correction). Similarly, the overall presence of sensory onset symptoms was not associated with the time to EDSS 4 (p=0.12) and EDSS 6 (p=0.32), but the presence of unilateral sensory onset symptoms was associated with a longer time to both EDSS 4 (p=0.008, this remains significant after Bonferroni correction) and EDSS 6 (p=0.03, not significant after Bonferroni correction; figure 2). Visual onset symptoms were associated with a younger age at EDSS 4 (p=0.01) and EDSS 6 (p=0.04), both of these associations were not significant after Bonferroni correction.
The results of the multivariable Cox regression models for the time to, and age at, landmark disability scores are shown in table 2. Older age at onset was independently associated with a shorter time to EDSS 4 (HR value per 1 year increase=1.03, 95% CI 1.02 to 1.04) and EDSS 6 (HR value per 1 year increase=1.04, 95% CI 1.02 to 1.05). Having bilateral motor onset symptoms was also independently associated with a shorter time to EDSS 4 (p=0.005) and EDSS 6 (p=0.03). The occurrence of unilateral sensory onset symptoms was not associated with the time to landmark disability after adjusting for other risk factors. Visual onset symptoms were associated with a younger age at EDSS 4 (HR value=2.65, 95% CI 1.37 to 5.14) and EDSS 6 (HR value=2.32, 95% CI 1.16 to 4.61). Sex and the other onset symptoms were not associated with time to, or age at, landmark disability.
Table 3 shows the Kaplan–Meier estimates and multivariable Cox regression HRs for the time from EDSS 4 to EDSS 6. Older age at onset of MS (p=0.03), a shorter time to EDSS 4 (p=0.02) and the presence of motor onset symptoms (p=0.02) were each associated with a shorter time from EDSS 4 to EDSS 6 in the Kaplan-Meier analyses, none of these associations remains statistically significant after Bonferroni correction. In the multivariable Cox regression model, only age at onset was significantly associated with the time from EDSS 4 to EDSS 6 (HR value=1.02, 95% CI 1.00 to 1.04). There was also some evidence of an association with the time from disease onset to EDSS 4, although this relationship was not statistically significant (HR value=0.97, 95% CI 0.95 to 1.00, p=0.05).
Our study adds to the current literature in several ways. First, the median time to EDSS 6 in our cohort was 9 years, which is in keeping with a recent large cohort study from Rennes, France (10 years, n=445)7 and a classical natural history study from London, Ontario in the late 1990s (8.5 years, n=216).12 The shortest time to EDSS 6 comes from the Lyon natural history cohort, with an estimate of only 7.1 years. However, in the same study, the time from onset of MS to EDSS 4 is reported as 0 years, which does not reflect clinical practice.4 At the other end of the extreme is the natural history cohort in British Columbia in which the time to EDSS 6 was 14 years.5 In that study, 109 of 552 patients (20%) were excluded because they had reached EDSS 6 before their first clinic visit. This approach, if it excluded patients with particularly quick disease progression, may have led to selection bias and an overestimation of the time to EDSS 6. An accurate estimate of the time to landmark disability is important because these data impact the design of clinical trials. For instance, the failure of the PROMISE trial on treatment with Glatiramer acetate in PPMS has been associated with the fact that the design of this trial was on the shorter estimates of disability accumulation from the Lyon cohort. In the trial the observed disability accumulation turned out to be less than was anticipated based on these estimates.13 It is currently unclear whether the EDSS score is the best measure of disability in progressive MS,14 however, it may be advisable not to base the design of future trials on either of the extreme estimates.
Strengths of our study are the large patient cohort, and the fact that this cohort is population based. A weakness of the study is the fact that a number of patients were excluded from the analyses because of missing data. In additional analyses we found that patients who were excluded from the analyses because of missing data were more often female, and had a younger age at onset than the patients without missing data. It is not clear how this could have affected the analyses. Given the sufficiently large number of participants included in the analyses we are confident that this would not unduly affect the conclusions drawn from the data.
Another shortcoming of our study is the lack of subclassification of the PPMS disease course. Very recently, revised criteria for the classification of the disease course of MS have been published, which suggest that adding a description of disease activity can be meaningful for the classification of the disease course.15 Since our cohort predates the establishment of these criteria, we cannot confidently reclassify our cohort to reflect this change. However, it is not entirely clear whether the subdivision of PPMS into forms with and without relapses is meaningful in natural history studies, as a study from the London (Ontario) natural history cohort found no difference in the time to the landmark disability scores EDSS 6, 8 and 10 between these subtypes.16
The end point used in this study is the time when a person first reached the landmark disability score, without a confirmation period. This endpoint is similar to those used in the London, Ontario cohort, but different from those used in the Rennes, Lyon and British Columbia natural history cohorts. The choice of end point may have consequences for the estimated survival times. However, differences in survival times between cohorts do not clearly reflect end point definitions: the time to EDSS 6 in patients with PPMS varies from 7.16 to 107 to 14 years5 in the ‘sustained’ in Lyon, Rennes and British Columbia natural history cohorts, in comparison to 8 and 9 years in the ‘unsustained’ cohort in London, Ontario17 and our present study. Even if our choice of outcome measure would have led to shorter estimates in the overall survival times, it would likely not have affected the HRs from the multivariable regression models.
There are few prognostic factors in PPMS. In the British Columbia cohort, having sensory symptoms at onset was associated with a better prognosis,5 whereas in the London, Ontario cohort, onset symptoms were not associated with times to landmark disability.12 Our findings suggest that laterality of onset symptoms may be an important factor to consider in natural history studies. We found that patients with bilateral motor onset symptoms had significantly quicker progression to EDSS 4 and EDSS 6 than patients with unilateral motor onset symptoms. These findings reflect clinical experience, which suggests that early spinal cord involvement, presenting as bilateral motor symptoms, is associated with a worse prognosis.18 In our study visual onset symptoms were associated with a younger age at EDSS 4 and EDSS 6, however, this result is based on a very small number of patients and the CIs around the estimates are wide. This makes the interpretation of this finding more difficult, and raise the possibility of a spurious association, especially since there is no similar relationship between visual onset symptoms and any of the other outcomes, and since visual onset symptoms were not associated with time to EDSS 6 in previous natural history studies.5 ,12
Age at disease onset was the most significant prognostic factor in our patient cohort, with a younger age at onset associated with longer times to landmark disability. However, it should be kept in mind that, similar to the findings in other cohorts,5 ,7 ,19 an earlier age at onset is also associated with an earlier age at disability. Therefore, early onset age should not be considered a ‘positive’ prognostic indicator. Finally, sex does not appear to influence the course of PPMS, which has also been shown in other comparable cohorts.5 ,7 ,12 ,20
Confavreux and coworkers were the first to show in the Lyon natural history cohort that progression of disability in RRMS is a two-stage process, with an early phase (from disease onset to EDSS 4), in which sex, age at disease onset and onset symptoms are all significantly associated with progression of disability, while no such associations exist in the second phase (from EDSS 4 to EDSS 6).6 Leray and coworkers later performed similar analyses in the Rennes natural history cohort (using EDSS 3 as the cut-off point) and also found that the factors sex and age at onset had a large impact on the first phase of disability accumulation in RRMS, but no effect on disability accumulation beyond EDSS 3. They also showed that the time from disease onset to EDSS 3 had no influence on the time from EDSS 3 to EDSS 6.7 In contrast to these studies in RRMS, our study shows no evidence for a similar two-stage process in PPMS. Age at disease onset is the most important independent predictor of disability progression throughout the disease course of PPMS: an older age at disease onset is predictive of a quicker disability progression to both EDSS 4 and EDSS 6. Disability accumulation in PPMS appears to be affected by the same factors throughout its course.
Contributors MWK took part in study concept and design, analysis and interpretation of data, writing of the manuscript and critical revision of the manuscript. JG was involved in analysis and interpretation, critical revision of the manuscript for important intellectual content. OJ, SD and WW were involved in acquisition of data, interpretation and critical revision of the manuscript for important intellectual content. LMM took part in study supervision, acquisition of data, interpretation and critical revision of the manuscript for important intellectual content.
Competing interests None.
Ethics approval University of Calgary Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.