Epidemiology of Progressive Multiple Sclerosis

Study

Time to EDSS endpoint

Cottrell [5]

216

Tremlett et al. [18]

352

13.3

Andersson et al. [1]

10.2

Koch et al. [25]

552

Debouverie et al. [11]

359

3.0

9.9

17.0

Confavreux et al. [6]

282

7.1

13.4

Runmarker and Andersen [7]

Simone et al. [16]

114

9.3

Leray et al. [2]

445

It is, however, clear that the rate of disability accumulation as reported is inconsistent and the reasons for this have generated some debate. A study by Tremlett et al. revealed a wide variation in rates of disease progression for PPMS, with median time to DSS 6 of 13.5 years, of which 25% of the population had reached EDSS 6 by 7.3 years after disease onset and yet 25% of the population had not reached this endpoint even after 25 years of follow-up [18]. In addition, 15% of this same cohort had reached EDSS 8 20 years after disease onset, 25% by 25 years, and 30% by 30 years [18]. This rate of accumulation of disability in this study is slower than reported in most others, including our own, although later stages of disease are more difficult to compare as time to EDSS 8 has been assessed in only one other large study reporting a median of 18 years [5], and also underlines the fact that even when a patient enters a progressive phase of disease, the outcome remains unpredictable. The reason for slower accumulation of disability observed in the British Columbia study may be due to the fact that patients who had already reached the specified disability outcome before the first clinic visit were excluded and this may have led to patients with more rapid disease progression being excluded, resulting in an overestimation of time to reach disability milestones. However, a further illustration of this unpredictability of outcome is that despite a progressive disease from onset, 9% of patients in one study were observed to have benign disease (EDSS ≤ 3 after 10 years) [26] even after prolonged follow-up. In contrast, it has also been demonstrated that patients recruited from and followed within the setting of a specialist MS center reach disability milestones earlier than those followed outside this construct [27], presumably due to the fact that specialist centers tend to accumulate patients with more severe disease. These issues underline that although there are undoubtedly emerging temporal changes in the disability profile of disease, it remains important to examine differences in the construct, setting, and representative nature of the patient cohort to the wider population in the interpretation and comparison of these studies.

As described above, there have been a number of studies examining disability outcome in PPMS employing scales and measurements for which there are established and significant inter- and intra-observer variations. However, there have been few that have utilized the most robust of endpoints, namely, that of mortality. Those studies that have examined mortality have been of relatively small numbers. In one study of 216 patients with primary progressive disease, nearly 50% had died, of which 70% had died of MS-related causes. The mean age at death was 58.8 years compared to 66.8 years for those who died of unrelated causes [5]. Of those with death due to MS, the mean time from onset to death was 22 years. In a further study, median survival from onset was longer (43 years) for BOMS than PPMS (26 years), and those with PPMS had a higher relative mortality [28].

3.2.4 Prognostic Factors in Primary Progressive Disease

Overall, there appears to be little disagreement that a progressive disease course, once established, carries a poor prognosis with a number of studies showing shorter survival times [29–35] and shorter time intervals to moderate disability [7, 36–42] when compared to BOMS. As a marker of prognosis, progressive disease at onset was noted in one study to be the strongest independent risk factor for death, having an adjusted risk ratio of 2.6 compared to relapsing disease [35]. Tremlett et al. also showed a striking difference in outcome between groups with median time to EDSS 6 from disease onset of 13.3 years for PPMS and 30.3 years for BOMS [8], while Ebers has reported 8 years and 15 years, respectively, in the London, Ontario study for similar milestones, and Confavreux et al. reported 7.1 years and 23.1 years, respectively, for the MS population in Lyon, France [41, 43]. Identifying additional factors which reliably impact on outcome in PPMS has proven more difficult. In general, there seems to be little evidence of an influence of gender on outcome [1 4 12, 18, 37], although a limited number of studies have demonstrated some more subtle effects such as shorter time to death in males [5] and increased odds ratios of deterioration on the EDSS for males over a 10 years period [44]. Similarly, evidence for the effect of ethnicity is limited and studies scarce, but data from one small study of Brazilians of African descent compared to white Brazilians with PPMS has suggested that ethnicity may influence progression with those patients of African descent taking, on average, 1 year less to reach EDSS 3 and 2 years less to reach EDSS 6 [45]. The impact of age is more controversial; some studies have shown no effect of age on time to disability endpoints in patients with primary progressive disease [1, 4]. One study showed a trend for those with the highest age at onset to have a shorter time to EDSS 10, though did not reach statistical significance [5]. In contrast, a study from British Columbia reported that those with a younger age at onset took a significantly longer time to reach EDSS 6. However, when the authors also studied the age at which this milestone was reached, they found that despite taking longer to reach EDSS 6, the patients were significantly younger at reaching EDSS 6 [25] which suggested that a younger age does not necessarily indicate a better prognosis and is reflected in our series in which age-related outcomes are marginally worse for BOMS than PPMS (see Fig. 3.1)

Fig. 3.1

Onset and age-dependent EDSS outcomes in a population-based cohort of 2009 patients from south Wales (a) Time to EDSS 4 (b) Time to EDSS 6 (c) Time to EDSS 8 (d) Age at EDSS 4 (e) Age at EDSS 6 (f) Age at EDSS 8 (Courtesy of Dr. K Harding, with permission)

3.2.5 Clinical Features at Disease Onset

Identifying clinical features that convey additional information on disease outcome is becoming increasingly important even in patients with PPMS. Unfortunately, no symptoms at disease onset are universally accepted to influence prognosis. Whereas some observers have suggested that those patients with visual or brainstem or cerebellar onset have a worse prognosis [12, 37], these data have been difficult to replicate. Further studies have suggested that a higher number of functional systems 3, 4, and 5 years after disease onset as well as the presence of bladder and bowel involvement within 5 years both correlate with a progression of EDSS scores [1] as well as shorter time to death [5]. In contrast, sensory symptoms at onset have been suggested as a good prognostic factor and associated with longer time and older age to EDSS 6 [25], although others have shown no prognostic value of any initial symptoms [4, 18].

Several studies have also suggested that those who have a slower initial progression of disability have a better prognosis with longer interval to EDSS 6 associated with a longer interval to EDSS 8 [18]. There is also a suggestion that those who reach even lower disability milestones early may also progress more rapidly [5] and a more rapid deterioration in EDSS over 2 years associating with a worse prognosis [44]. A further study has suggested that the best predictor of deterioration in disability was a slower timed walk [44], and in addition, reduction in brain volume recorded by serial cranial MR scanning may also be an effective predictor of outcome being associated with greater long-term disability [44].

3.3 Secondary Progressive Multiple Sclerosis

3.3.1 Disease Frequency

Approximately 80–90% of patients with MS will have an initial disease course which is relapsing. After a variable interval, a proportion of these patients will also develop a progressive disease course. The rate of conversion to secondary progression (SPMS) has been suggested to be approximately 2–3% per year [46], and the longer the period of follow-up, the larger the overall proportion of convertion. Weinshenker et al. demonstrated that after 5 years, 12.1% of a north American cohort with BOMS had converted to progressive disease, whereas between 6–10, 11–15, 16–25 years, and more than 26 years, 41.3%, 57.6%, 65.6%, and 88.9% had developed progressive disease, respectively [4]. The proportional increase of SPMS over time in studies only limited by the length of follow-up has led some observers to suggest that all patients would eventually develop SPMS if observed up for long enough [9]. Table 3.2 shows the proportion of patients with relapsing–remitting disease reaching secondary progression over time in published studies.

Table 3.2

Proportion of patients with relapsing–remitting disease reaching secondary progression

	N	Proportion reaching SPMS	Length of follow-up (mean, years)
British Columbia [19]	1,445	58.2	20
Goteborg, Sweden [7]	162	80.2	25
London, Ontario [17]	551	66.9	25
Belgrade, Yugoslavia [31]	119	76.8	21.7
Groningen, Netherlands [11]	108	50.5	18
Lyon, France [47]	496	31.8	12
Florence, Italy [40]	190	27.4	9.78
Rennes, France [2]	1,609	38.4	16.0
Bergen, Norway [42]	179	35.2	14.4
Burgundy, France [48]	288	18	6
Bari, Italy [16]	596	24	6.34

3.3.2 Subsequent Disease Course

The median time to the onset of secondary progressive disease varies between 1.83 and 21.4 years (Fig. 3.2). Once again, differences are likely to be methodological, so that in early data from London, Ontario mean time to secondary progression in the population based subgroup was 7.14 years [13]. In contrast, in the “seen from onset” subgroup, the time interval was only 1.83 years. However, it is likely that this latter figure is an underestimate since it is calculated from a group of patients seen from disease onset in a specialist clinic [13], the problems of which have been previously described. In addition, analysis of data did not distinguish PPMS patients from the wider patient cohort. When this population was reassessed after an interval of 17 years, median time to progression had increased to a more realistic 10.3 years for BOMS and to 7.6 years for patients who had experienced only a single attack before the onset of a progressive phase [17]. This latter analysis did not include PPMS patients and a much greater proportion of at least two thirds of patients had reached secondary progression, thus making calculation considerably more reliable. A further analysis was undertaken a further 4 years later by which time the median time to progression had once again increased to 15 years [49]. These data effectively demonstrate the impact that increasing length of follow-up and greater proportions of patients have on calculating median times to conversion, and it is likely that the real figure will exceed even these values.

Fig. 3.2

Scatter plot of proportion of patients with relapsing–remitting disease onset converting to secondary progressive disease over time from multiple studies

Interpretation of other studies should be made with these issues in mind. For example, median time to reach secondary progression of 8 years was reported in a study from Paris, France, and 7.7 years from Pavia, Italy. However, follow-up for both studies was short and only between 18% and 23% of patients had reached the study endpoint [23]. A further study in Belgrade, Yugoslavia demonstrated mean time of 6.2 years to develop SPMS but in contrast with usual epidemiological data analysis, failed to take into account patients who had not yet reached this endpoint [31].

Tremlett et al., in one of the largest studies available, in which more than half of patients had reached the designated endpoint, calculated median time to secondary progression of 18.9 years [19, 50]. In a later study, once again using data available from British Columbia, the same group reported median time to secondary progression had increased to 21.4 years from onset and a median age at progression of 53.7 years [3]. However, in this analysis, 24% of patients had at some point been prescribed immune-modifying drugs and had been censored at the first date of their use. This raises yet more complications in comparison of contemporary analyses of time to SPMS data since it is possible that this group represented those patients with more severe disease and by censoring, time to secondary progression may have been overestimated. However, for the time being, the true impact of disease-modifying treatments on overall profile of longer term epidemiological data such as these remains unknown. Additional data from Lyon, France, has shown similar results with median times to secondary progression reported to be 19 years [46] and marginally longer than the median of 16 years reported from a study in Rennes [2].

The wide variation in the time taken to reach secondary progression is striking and counterintuitive to the concept of a gradually progressive disorder of predictable disability accumulation and requires some further consideration. This variability is emphasized by one study which demonstrated that a quarter of patients had reached secondary progression in less than 11.4 years and under 45.1 years of age, but a further quarter reached progression after more than 32 years and at an age of older than 63.1 years [3]. This variability was also noted in a further study with 9.6% reaching SPMS within 5 years and yet 25% not reaching secondary progression by 30 years [50]. There is no doubt that the recognition of onset of progression, which has been determined retrospectively, is subject to inter-rater variability, and is often difficult to estimate with confidence. Although some observers have shown good inter-rater agreement for assigning onset of progression [11], this is commonly only within a calendar year and is not universal. In the British Columbia dataset, 24.6% of patients reached SPMS at or before EDSS 3, 80% at or before EDSS 6, and 99.5% at or before EDSS 8 [50]. In contrast, in the London, Ontario dataset, the time of onset of progression in those with a single attack prior to progression was felt to be at or before DSS 2 in 84.8%. In comparison, those with multiple relapses prior to progression, the onset of progression was felt to be recognized at DSS 3 or less in 48%, DSS 4 or less in 91.0%, and DSS 6 or less in 99.7% [17]. This rather striking difference in the proportion of patients entering the progressive phase at differing disability levels suggests either a fundamental difference in the course of MS between these areas or more likely a different approach in the assignation of progressive disease status by treating physicians between the two centers. This is almost certainly one of the main reasons for the large range of reported data in these and other studies. Table 3.3 summarizes the time to secondary progression in the major studies.

Table 3.3

Median time to secondary progression in some of the major studies

		Location	N	Mean length of follow-up	Proportion reaching SP	Median time to secondary progression	Mean age at secondary progression
Confavreux et al. [9]	Hospital longitudinal	Lyon, France	349	9	29.1	6.8^a
Vukusic and Confavreux [46]	Hospital based	Lyon, France	1,844	11^b	31.8	19
Weinshenker et al. [13]^c	Total population	London, Ontario	1,099	11.9^b		5.8
	Population based subgroup	London, Ontario	196			7.14
	Seen from onset subgroup	London, Ontario	197			1.83
Kremenchutzky et al. [17]	Population based	London, Ontario	824	25	66.8	10.4	39.2
Scalfari et al. [49]	Population based	London, Ontario	806	23^b	66.2	15	40.2
Levic et al. [31]	Hospital based	Belgrade, Yugoslavia	82	21.7	76.8	6.2
Bergamaschi et al. [51]	Hospital based	Pavia, Italy	186	7	18	7.7
Tremlett et al. [19 50]	Hospital based	British Columbia, Canada	2,484	20.5^b	58.2	18.9	49.0
Koch et al. [26]	Hospital based	British Columbia, Canada	5,207	ns	35	21.4	53.7
Simone et al. [16]	Hospital Based	Bari, Italy	596	5	24	6.88
Minderhoud et al. [11]	Population based	Groningen, Netherlands	450	ns	31.6	8.2^a	37.5
Eriksson et al. [52]	Population based	Gothenburg, Sweden	135	25	46.8	12.0
Stankoff et al. [23]	Hospital Based	Paris, France	884	ns	23.2	8
Leray et al. [2]	Hospital based	Rennes, France	1,609	12.8	38.4	16.0	40.4