Onset of progressive phase is an age-dependent clinical milestone in multiple sclerosis

Study populations

All patients were seen and followed at the Mayo Clinic Center for MS and CNS Demyelinating Diseases, a tertiary care provider for complicated referral patients with MS and a primary care provider for regional patients with MS.

We studied two cohorts fulfilling McDonald diagnostic criteria for MS: ^11,12 a population-based MS cohort from Olmsted County, MN, USA, established in 1992 ¹³ and re-ascertained in 2002 (n=221, age ≥18 years), ^14,15 and a clinic-based progressive MS cohort (n=859, age ≥18 years) established from patients seen from 2003 through 2007, with observed progression of at least 1 year (Figure 1).

OPEN IN VIEWER

Figure 1.

Study design, patient populations and final disease course classification.

Definition of progression

We defined progression as insidious and irreversible worsening brain, brainstem–cerebellar and spinal cord syndromes most commonly characterized by progressive weakness, ataxia or bladder dysfunction of ≥1 year. We did not include patients with exclusive sensory progression due to subjectivity and variability in recording change in intensity of sensory symptoms. Typical of a standard clinical practice setting, the most detailed and rigorous historical documentation and clinical examination closest in time to progression onset was used to establish the date of onset of progression.

Disease course classification

PPMS is defined as progression without previous relapses. SAPMS is defined as progression following a CIS. ⁵ SPMS is defined as progression following RRMS. Although separated for understanding potential impact of a single critical relapse before progression onset, SAPMS and SPMS are also grouped together as BOPMS when appropriate for the analyses.

Rather than as a separate relapsing–progressive MS category, ongoing relapses after onset of progression were considered as part of a continuum of the symptomatic (BOPMS) or subclinical (PPMS) relapsing phase as it evolved into progressive phase of MS.

The population-based cohort captured RRMS patients with or without conversion to SPMS despite a ≥20-year follow-up. As a tertiary care referral practice, the clinic-based cohort captured a population enriched for less common forms of progressive MS (PPMS and SAPMS) than SPMS.

Study variables and outcomes

We extracted the following variables: sex; age at MS onset; age at progression onset; year of progression onset; age at sustained disability of the clinically well-recognized and well-recorded milestone of EDSS6 ^5,10 or needing a cane to walk; age at last relapse; occurrence of pre-progression and post-progression relapses. We studied the following outcomes: time from onset of MS to onset of progression, age at progression onset, and time from progression onset to EDSS6 (time to EDSS6).

Data extraction and validation

In the population-based cohort with prospective follow-ups in our center, two neurologists with subspecialty training in MS (JT, MT) independently reviewed all neurologic and non-neurologic records and agreed on date of progression and date of EDSS6 in each case.

Two additional neurologists with MS subspecialty training (NA, NK) reviewed all neurologic and non-neurologic records for the clinic-based cohort. Year of progression onset was identical according to the two reviewers in 26 of the first 30 reviewed patients. The year of attaining EDSS6 matched in 26 of 28 patients (two of 30 had not reached EDSS6 in this group). In the cases where records at different time points indicated different dates for progression onset or EDSS6, decision was made to use the earliest record closest to the event date. Any final discrepancies were arbitrated by the principal investigator (OK). Despite consensus that the course was progressive, we could not clarify presence or absence of pre-progression relapses in 19 (2.5%) patients from the clinic-based cohort, who were excluded from the final analyses (Figure 1).

Data analyses

To assess for potential impact of referral and recall biases on age at progression onset in the largely retrospective tertiary care clinic population, we compared sex ratio, age at MS onset, age at progression onset and time to EDSS6 in the two cohorts.

Demographic and clinical features in different models were compared by Chi² or t-tests as appropriate. Kaplan–Meier analysis was used to account for censoring, age- or time-scaled outcomes (i.e. where RRMS patients that are yet-to-progress are included in the analyses). Mean values with standard deviations from the mean were provided when appropriate (i.e. where all patients included in the analyses have reached the outcome).

Do sex ratio, age at progression onset and time to EDSS6 differ between clinic-based and population-based MS patients?

Table 1 shows the demographic aspects of patients with progressive MS from the population-based cohort compared with the clinic-based cohort. Analyses were done both by including and excluding PPMS patients. Sex ratio did not differ between populations (Table 1).

MS onset was 3.5 years older for patients with BOPMS in the clinic-based cohort compared with the population-based cohort (mean ± SD: 33.5 ± 9.6 years vs. 29.9 ± 8.8 years; p=0.001) (Table 1). Age at progression onset (p=0.37) did not differ between clinic- and population-based cohorts (Table 1). Time to EDSS6 did not differ between clinic- and population-based cohorts, regardless of inclusion (p=0.16) or exclusion (p=0.86) of PPMS (Figure 2).

OPEN IN VIEWER

Figure 2.

Kaplan–Meier estimates (with life-tables) of time to moderate disability after progression onset are shown for the population-based and clinic-based MS cohorts. All progressive MS patients (a) and bout-onset progressive MS patients (b) are shown separately. ‘Progression’ refers to insidious and irreversible worsening brain, brainstem-cerebellar and spinal cord syndromes most commonly characterized by progressive weakness, ataxia or bladder dysfunction of ≥1 year. In the population-based cohort from the Olmsted County, MN (n=210), patients with RRMS (n=109) and patients where age at onset of progression was not clear (n=1 SAPMS) were excluded. In the clinic-based referral patients (n=773), patients where pre-progression disease course was not clear (n=19) were excluded. This figure illustrates that for patients with progressive MS, there is no referral bias due to more severe disease after progression onset in the clinic-based cohort compared with the population-based cohort.

Despite a long follow-up (mean ± SD: 22.8 ± 11.8 years), only two of 109 (1.8%) patients in the RRMS group without secondary progression reached EDSS6. Similarly, two of 85 (2.4%) patients from the population-based cohort and seven of 447 (1.6%) patients from the clinic-based cohort with established BOPMS reached EDSS6 before progression (Table 1).

Do all patients with RRMS ultimately progress?

Four models were applied to the population-based cohort using Kaplan–Meier estimates of time to progression and age at progression onset (Figure 3). Model 1 includes the patients with established SPMS only. Model 2 includes patients with established SPMS and the 109 patients still with RRMS as of their last follow-up; nine patients who died and 100 patients who had their last follow-up before the year 2010 were excluded. To consider extreme estimates for the excluded patients, Model 3 (upper limit) assumed that all RRMS patients develop SPMS within a year from their last follow-up or death, whereas Model 4 (lower limit) assumed that all RRMS patients remain RRMS as of 2010.

OPEN IN VIEWER

Figure 3.

Kaplan–Meier estimates (with life-tables) of time to progression (a) and age at progression (b) are shown in the longitudinally followed population-based cohort of patients with MS older than age 18 in Olmsted County, MN (n=210). Patients whose date of onset could not be precisely defined [SPMS (n=4), SAPMS (n=6) and PPMS (n=15)] are excluded from the analyses. Model 1 assumes that some patients with MS will never develop progression and therefore includes only patients with already established SPMS (n=80). Model 2 assumes that patients with MS may at one point develop progression (censored data) and therefore includes patients still with RRMS (n=105) as of their last follow-up together with patients with already established SPMS. Model 3 assumes that all RRMS patients with censored data due to death or last follow-up before 2010 would have converted to SPMS within 1 year of their last follow-up establishing the upper extreme limit for Model 2. Model 4 assumes that all of the RRMS patients who are deceased or lost to follow-up remain RRMS as of 2010 establishing the lower-extreme limit for Model 2. These models predict that that not all patients with MS develop progression despite sufficient follow-up. ‘Progression’ refers to insidious and irreversible worsening brain, brainstem-cerebellar and spinal cord syndromes most commonly characterized by progressive weakness, ataxia or bladder dysfunction of ≥1 year.

Within 35 years of MS onset, 94% of patients with established SPMS (Model 1) had already developed progression. When patients with RRMS as of their last follow-up were included (Model 2), 45% of patients had not yet developed progression by 35 years post-MS onset (Models 3 and 4 suggested a range of 13–52%).

At age 75, 99% of patients with established SPMS (Model 1) had already developed progression. When patients with RRMS as of their last follow-up were included (Model 2), 38% of patients had not yet developed progression at age 75 (Models 3 and 4 suggested a range of 3–42%).

Does the presence (BOPMS) or absence (PPMS) of a pre-progression symptomatic relapsing–remitting phase impact age at progression onset?

Our hypothesis required comparison of SPMS and SAPMS with PPMS for age at progression onset. Studying the pre-progression disease course in PPMS is akin to ascertaining all radiologically isolated syndrome (RIS) patients in the population ^{16 -19} and is not practical. Since about one-third of RRMS patients may also never develop progressive disease course it was more appropriate to exclude yet-to-progress RRMS patients for a head-to-head comparison of age at progression onset between BOPMS and PPMS.

Given the similar sex ratio, age at progression onset, and ratio of attainment of EDSS6 before progression between the two cohorts, patients with progressive MS from each cohort were combined. Distribution of age at first clinical symptoms of MS and age at progression onset stratified by disease course (PPMS, SAPMS, SPMS) revealed that age at MS onset was earlier for patients with SPMS (31.8 ± 9.3) versus SAPMS (37.4 ± 9.4) versus PPMS (45.7 ± 10.8) (p<0.001) (Figure 4). However, age at progression onset did not differ by disease course (SPMS: 44.9 ± 9.6, SAPMS: 45.5 ± 9.6, PPMS: 45.7 ± 10.8, p=0.23). We also found that while the proportion of patients who experienced relapses after progression onset differed by disease course (SPMS 27.3%, SAPMS 12.3%, PPMS 0.3%; p<0.0001), age at last relapse after progression did not differ by disease course (SPMS 48.1±9.5, SAPMS 49.6±8.8, PPMS 47.8±10.2; p>0.05).

OPEN IN VIEWER

Figure 4.

Age at developing first clinical symptoms MS (a) and age at reaching progression (b) curves, along with life-table analyses and mean (± standard deviation) values are shown for SPMS versus SAPMS versus PPMS. All patients with progressive MS from both cohorts are included in the analyses. In the population-based cohort from the Olmsted County, MN (n=210), we excluded patients with RRMS (n=109) and patients whose age at onset of progression was not clear (n=1 SAPMS). In the clinic-based referral patients (n=773), we excluded patients where presence or absence of pre-progression relapses were not well documented (n=19). Age at progression does not differ according to whether there were preceding relapses, but age at onset of MS, defined either by the first relapse (SPMS and SAPMS) or progression onset (PPMS), is significantly earlier for SPMS compared with SAPMS and compared with PPMS. ‘Progression’ refers to insidious and irreversible worsening brain, brainstem-cerebellar and spinal cord syndromes most commonly characterized by progressive weakness, ataxia or bladder dysfunction of ≥1 year.

Switching from relapses to progression in MS is an age-dependent process independent of pre-progression disease course or disease duration

We novelly show that almost all patients with MS develop a progressive disease course before age 75 (or by 35 years post-MS onset), while 38% of patients with RRMS are predicted not to develop progression by age 75. Identifying the intriguing underlying biology and predictors of non- progression can help design novel therapeutics to prevent progression, improve patient counseling and help better stratify patients for clinical trials.

Age at progression onset was independent of pre-progression clinical relapse-based classification of PPMS, SAPMS or SPMS in the current study, as previously reported. ^2,5 We confirm that onset of progression in MS is age- rather than pre-progression disease duration-dependent. ^7,8 PPMS occurs only infrequently in children, ^20,21 further supporting that progression is an age-dependent process.

In the current study, ongoing relapses after progression onset were more common in SPMS than in SAPMS or PPMS, mirroring the frequency of pre-progression clinical relapses. However, a novel finding is that the age at last relapse was similar regardless of disease course classification of PPMS, SAPMS or SPMS. The end of the relapsing phase, like the onset of progressive phase in MS, also seems to be an age- rather than disease duration-dependent process.

Recent imaging studies suggest that myelination of the compact white matter ends by the fourth decade, followed by a slow degeneration in the white matter tracts in the following decades. ^22,23 Continued white matter tract repair (myelin and axons) may compensate for degeneration in previously damaged areas of the central nervous system until the end of fourth decade. An advancing degenerative phenotype afterwards is possibly reflected by the clinical progression surfacing in the fifth decade, as in our study.

The relapsing phase of MS can manifest as symptomatic relapses (i.e. RRMS) or subclinical lesions (i.e. RIS). ^{16 –19} The current diagnostic criteria for MS utilize subclinical lesions by magnetic resonance imaging (MRI) to establish dissemination in time and space during the first encounter with the patient who has already had a clinical relapse or has a slowly progressive disease course. ²⁴ Patients with PPMS may have an ‘established’ lesion load on MRI at onset of progression indistinguishable from that in patients with SPMS. SPMS follows RRMS clinically, while PPMS follows RIS in a similar but subclinical continuum. ^25,26 In patients matched for the total number of brain and spinal cord lesions, SPMS and PPMS differ by the presence of previous and ongoing inflammatory demyelination and recovery potential of individual lesions, but share a similar spectrum of slowly expanding demyelination and axonal injury. ²⁷ Our study further contributes to the unifying clinical scheme that SPMS, SAPMS and PPMS likely share a common neurodegenerative mechanism of progression but differ by the extent to which inflammatory–demyelinating activity manifests as preceding clinical relapse(s). ^2,5

Recruitment of patients with progressive MS to studies through referral practices does not introduce any significant additional biases than expected in population-based cohorts that could impact age at progression onset determination

While population-based studies are the standard for studies like ours, we had a paucity of SAPMS and PPMS patients when compared with SPMS in our population-based cohort. Therefore we studied an additional clinic-based cohort established from our practice. The higher incidence of SAPMS (14%) and PPMS (41%) in the clinic-based cohort than in our population-based cohort (6% and 15%) and another population-based cohort ⁵ confirms that patients with less common forms of progressive MS than SPMS are over-represented among patients referred to tertiary care practices.

Inclusion of the clinic-based cohort of patients could have potentially introduced additional recall and referral biases to our analyses. To address this we compared the two populations with each other. We found that when ascertainment and follow-up are conducted in a relatively uniform clinical practice setting, age at progression onset and attaining the disability milestone of EDSS6 in the retrospective tertiary clinic-based cohort were similar to that of the prospective population-based cohort. We noted that the age at progression onset was 1 year younger in the population-based cohort, a small difference that did not reach statistical significance within the power limitations of our study.

Age at MS onset in RRMS did not differ between those with and those without secondary progression in the population-based cohort, but was older in the clinic-based cohort compared with the population-based cohort. This suggested a potential recall bias for age at MS onset in the largely retrospective clinic-based cohort. However, our primary outcome measure of age at progression onset did not differ significantly between clinic-based and population-based cohorts or between regional or non-regional referrals. Therefore, age at progression onset, a more recent and potentially more life-changing event than the first MS event, did not appear to be as prone to a potential recall bias as the age at MS onset.

Men are generally accepted to have older age at onset of MS and more severe disease than women ^1,9 but men were only slightly over-represented in our clinic-based practice; this did not reach statistical significance within the power limitations of our study. However, it is possible that the non-significant 1-year delay in age at progression onset in the clinic-based cohort is due to a slightly higher representation of men in this population. Time to EDSS6 did not differ between clinic-based and the population-based cohorts of progressive MS, either. We confidently concluded that our results regarding age at progression onset were therefore unlikely to be confounded by a major recall or referral bias specific to the clinic-based cohort.

Commonly used disease-modifying medications (DMDs) in MS do not seem to impact progressive disease course after it is established, except for some benefit on preventing additional disability burden that may ensue because of superimposed relapses. ²⁸ It is, however, unclear if DMDs can delay onset of progression in the first place in patients with BOPMS. Since our study is not planned as a prevention trial and PPMS patients naturally did not have pre-progression DMD use, there was no way to properly control for this factor in our analyses when comparing age at progression onset between BOPMS and PPMS.

EDSS6 is a disability milestone that is reached predominantly after the onset of progressive disease course in MS

EDDS6 and higher levels of the scale are biased towards the longest corticospinal tracts and naturally could reflect the impact of progression-related disability. ^5,10 Progressive disease course has been the strongest determinant of sustained disability measured by EDSS. ^{1 -4,9,29 -31} Attainment of the disability milestone of sustained EDSS4 is inevitably followed by further worsening of disability. ³² Disability progression can result from relapses that partially recover, or a progressive disease course, or both. The contribution of relapses on disability progression decreases in time. ³³ Individual relapses affecting the efferent motor pathways are less likely than relapses affecting afferent pathways to lead to chronic impairment in patients with MS. ³⁴

While almost all patients with SPMS develop the progressive phase of MS when their disability is EDSS6 or less, ⁵ it was not known what percentage of patients with RRMS without progression reach sustained EDSS6. A recent clinic-based study found that only 0.3% of individual relapses resulted in a sustained disability of EDSS6 or greater. ³⁵ In our study, 2% of patients without progression reached sustained EDSS6, representing the population-based frequency of ‘malignant relapsing–remitting’ cases of MS. ³⁶ In contrast, regardless of the cohort, 98% of patients who developed progression reached EDSS6 after progression onset. We conclude that EDSS6 is a disability milestone that is reached predominantly after the onset of progressive disease course in MS.

Interestingly, as in the case for developing progressive disease course in MS, the speed of reaching the EDSS milestones seems to be age- rather than pre-progression disease course-dependent. ^32,37,38 This intricate relationship between age of the patient, onset of progressive disease course and evolution of disability milestones needs to be accounted for in clinical trials and epidemiology studies in MS.