Эпидемиология, патология и патогенез MS: терапевтические последствия.

Epidemiology of MS

Multiple sclerosis (MS) is a chronic, recurrent, inflammatory disease of the central nervous system ( CNS ), which generally becomes clinically evident in young adulthood (mean age of symptom-onset: ∼30 years) – see
Table 1 & Fig. 1 . Nevertheless, it can begin as early as first or second year of life or as late as the seventh or eighth decade [ [1] , [2] , [3] , [4] , [5] , [6] ]. It can be one of the most disabling diseases of younger individuals. Currently, women are affected two to three times as often as men [ [5] , [6] , [7] ] – see also Table 1 – although this may not have been the case during earlier Time-Periods [ 7 , 8 ] – see also Table 2 . Episodes of neurologic dysfunction (interchangeably called episodes, attacks, flares, exacerbations, or bouts) occur unpredictably and often result in multifocal tissue injury, especially to the oligodendrocytes and the myelin sheaths, but also to the nerve cells and axons within the CNS [ 5 , 6 ]. Because MS is defined as a recurrent disease, individuals who experience a single episode (without other ancillary evidence of recurrence over time) are said to have a “ clinically isolated ” syndrome ( CIS ). However, many patients who are diagnosed initially with CIS will develop clinically definite MS subsequently and, in any event, CIS is thought to reflect the same underlying disease process. Consequently, the revised MS-diagnostic criteria, which are currently under development, will likely remove the requirement for MS attacks (either symptomatic or radiographic) to be recurrent. MS pathophysiology, unquestionably, involves immune mechanisms, either as a primary cause or as a secondary phenomenon. As discussed below , direct evidence of inflammation is found within the CNS tissue ( see section on Pathology of MS ). Evidence is also found in the cerebrospinal fluid ( CSF ), including elevations in the protein concentration, in the number of mononuclear cells, and in the levels of CNS -derived gamma-globulin (IgG), as reflected by the IgG index, the IgG synthesis rate, the presence of unique oligoclonal IgG bands on gel electrophoresis, and increased levels of kappa free light chains [ 5 , 6 , 9 ]. Table 1 Natural history of MS. Table 1 Italy 1 Canada 2,3 Canada 4-6 United States 7 France 8,9 Sweden 10 Number of patients 1463 2837 1099 201 1844 255   % Female (all) 67.2 ​ % 70.4 ​ % 65.7 ​ % 69.7 ​ % 64.4 ​ % 60.0 ​ %   % Female (PPMS) – 53.3 ​ % 57.1 ​ % – 53.8 ​ % – Onset age (all) (Standard deviation) 28.3 ​yrs (9.0) 30.6 ​yrs (10.0) 30.5 ​yrs (9.9) 31.2 ​yrs – 31.0 ​yrs (9.7) – –  < ​15 ​yrs 3.8 ​ % – – – – –  < ​20 ​yrs 18.7 ​ % 9.6 ​ % 12.0 ​ % – 11.7 ​ % –  > ​40 ​yrs 11.6 ​ % 13.8 ​ % 20.1 ​ % – 20.6 ​ % –  > ​50 ​yrs 1.9 ​ % 4.7 ​ % – – 5.9 ​ % – Relapsing onset ( %) 86.1 ​ % 87.6 ​ % 81.2 ​ % 94.5 ​ % 84.7 ​ % 85.9 ​ % Earliest onset – – – 12 ​yrs 5 ​yrs – Latest onset – – – 58 ​yrs 62 ​yrs – Mean age, relapsing onset – – 29.5 ​yrs Mean age, progression onset  SPMS – 49.0 ​yrs 40.1 ​yrs – 39.5 ​yrs –  PPMS – 41.0 ​yrs 38.6 ​yrs – 39.3 ​yrs – Time to EDSS ​= ​4, median – – – – 8 ​yrs –  Relapsing onset – – – – 11 ​yrs –  Progressive onset – – – – 0 ​yrs – Time to EDSS ​= ​6, median – 28 ​yrs 15 ​yrs 26 ​yrs 14 ​yrs 18 ​yrs  Relapsing onset – – – 28 ​yrs 23 ​yrs 23 ​yrs  Progressive onset – – – 7 ​yrs 7 ​yrs 6 ​yrs 1. Liguori M, Marrosu MG, Pugliatti M, Giuliani F, De Robertis F, Cocco E, Zimatore GB, Livrea P, Trojano M. Age at onset in multiple sclerosis. Neurol Sci. 2000; 21: S825–S829. 2. Tremlett H, Paty D, Devonshire V. Disability progression in multiple sclerosis is slower than previously reported. Neurology 2006; 66,172–7. 3. Tremlett H, Yinshan Zhao, Devonshire V. Natural history of secondary-progressive multiple sclerosis. Mult Scler . 2008; 14:314–24. 4. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: I. clinical course and disability. Brain . 1989; 112:133–46. 5. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: II. predictive value of the early clinical course. Brain 1989; 112:1419–28. 6. Kremenchutzky M, Rice GP, Baskerville J, Wingerchuk DM, Ebers GC. The natural history of multiple sclerosis: a geographically based study 9: observations on the progressive phase of the disease. Brain , 2006; 129: 584–94. 7. Pittock SJ, Mayr WT, McClelland RL, Jorgensen NW, Weigand SD, Noseworthy JH, Rodriguez M. Disability profile of MS did not change over 10 years in a population-based prevalence cohort. Neurology . 2004; 62:601–6. 8. Confavreux C, Vukusic S, Adeleine P. Early clinical predictors and progression of irreversible disability in multiple sclerosis: an amnesic process. Brain . 2003; 126:770–82. 9. Confavreux C, Vukusic S. Natural history of multiple sclerosis: a unifying concept. Brain . 2006; 129:606–16. 10. Runmarker B, Andersen O. Prognostic factors in a multiple sclerosis incidence cohort with twenty-five years of follow-up. Brain 1993; 116:117–34. Fig. 1 The distribution of the age at symptom-onset in a cohort of 1 , 463 patients with multiple sclerosis (MS). The age of symptom-onset in Figure is given as (mean ​± ​1 standard deviation) (Data from: Liguori M, Marrosu MG, Pugliatti M, Giuliani F, De Robertis F, Cocco E, Zimatore GB, Livrea P, Trojano M. Age at onset in multiple sclerosis. Neurol Sci. 2000; 21:S825-9). Modified with permission from: Goodin DS [ 6 ]. Fig. 1 Table 2 The sex distribution for cases of multiple sclerosis prior to 1922 a . Table 2 Europe (Historical) United States (Historical) United States (Wechsler Series) Total Male 658 (58 ​ %) 99 (60 ​ %) 117 (59 ​ %) 874 (58 ​ %) Female 484 (42 ​ %) 67 (40 ​ %) 80 (41 ​ %) 631 (42 ​ %) Total 1142 (100 ​ %) 166 (100 ​ %) 197 (100 ​ %) 1505 (100 ​ %) a Data from: Wechsler IS [ 8 ]. Mostly, the historical cases represent those reported in the medical literature prior to 1903. Wechsler's series was drawn from the experience of the Mount Sinai Hospital, the Montefiore Hospital, and the Vanderbilt Clinic (1912–1921).

Table 1 Natural history of MS. Table 1 Italy 1 Canada 2,3 Canada 4-6 United States 7 France 8,9 Sweden 10 Number of patients 1463 2837 1099 201 1844 255   % Female (all) 67.2 ​ % 70.4 ​ % 65.7 ​ % 69.7 ​ % 64.4 ​ % 60.0 ​ %   % Female (PPMS) – 53.3 ​ % 57.1 ​ % – 53.8 ​ % – Onset age (all) (Standard deviation) 28.3 ​yrs (9.0) 30.6 ​yrs (10.0) 30.5 ​yrs (9.9) 31.2 ​yrs – 31.0 ​yrs (9.7) – –  < ​15 ​yrs 3.8 ​ % – – – – –  < ​20 ​yrs 18.7 ​ % 9.6 ​ % 12.0 ​ % – 11.7 ​ % –  > ​40 ​yrs 11.6 ​ % 13.8 ​ % 20.1 ​ % – 20.6 ​ % –  > ​50 ​yrs 1.9 ​ % 4.7 ​ % – – 5.9 ​ % – Relapsing onset ( %) 86.1 ​ % 87.6 ​ % 81.2 ​ % 94.5 ​ % 84.7 ​ % 85.9 ​ % Earliest onset – – – 12 ​yrs 5 ​yrs – Latest onset – – – 58 ​yrs 62 ​yrs – Mean age, relapsing onset – – 29.5 ​yrs Mean age, progression onset  SPMS – 49.0 ​yrs 40.1 ​yrs – 39.5 ​yrs –  PPMS – 41.0 ​yrs 38.6 ​yrs – 39.3 ​yrs – Time to EDSS ​= ​4, median – – – – 8 ​yrs –  Relapsing onset – – – – 11 ​yrs –  Progressive onset – – – – 0 ​yrs – Time to EDSS ​= ​6, median – 28 ​yrs 15 ​yrs 26 ​yrs 14 ​yrs 18 ​yrs  Relapsing onset – – – 28 ​yrs 23 ​yrs 23 ​yrs  Progressive onset – – – 7 ​yrs 7 ​yrs 6 ​yrs 1. Liguori M, Marrosu MG, Pugliatti M, Giuliani F, De Robertis F, Cocco E, Zimatore GB, Livrea P, Trojano M. Age at onset in multiple sclerosis. Neurol Sci. 2000; 21: S825–S829. 2. Tremlett H, Paty D, Devonshire V. Disability progression in multiple sclerosis is slower than previously reported. Neurology 2006; 66,172–7. 3. Tremlett H, Yinshan Zhao, Devonshire V. Natural history of secondary-progressive multiple sclerosis. Mult Scler . 2008; 14:314–24. 4. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: I. clinical course and disability. Brain . 1989; 112:133–46. 5. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: II. predictive value of the early clinical course. Brain 1989; 112:1419–28. 6. Kremenchutzky M, Rice GP, Baskerville J, Wingerchuk DM, Ebers GC. The natural history of multiple sclerosis: a geographically based study 9: observations on the progressive phase of the disease. Brain , 2006; 129: 584–94. 7. Pittock SJ, Mayr WT, McClelland RL, Jorgensen NW, Weigand SD, Noseworthy JH, Rodriguez M. Disability profile of MS did not change over 10 years in a population-based prevalence cohort. Neurology . 2004; 62:601–6. 8. Confavreux C, Vukusic S, Adeleine P. Early clinical predictors and progression of irreversible disability in multiple sclerosis: an amnesic process. Brain . 2003; 126:770–82. 9. Confavreux C, Vukusic S. Natural history of multiple sclerosis: a unifying concept. Brain . 2006; 129:606–16. 10. Runmarker B, Andersen O. Prognostic factors in a multiple sclerosis incidence cohort with twenty-five years of follow-up. Brain 1993; 116:117–34.

Natural history of MS.

1. Liguori M, Marrosu MG, Pugliatti M, Giuliani F, De Robertis F, Cocco E, Zimatore GB, Livrea P, Trojano M. Age at onset in multiple sclerosis. Neurol Sci. 2000; 21: S825–S829.

2. Tremlett H, Paty D, Devonshire V. Disability progression in multiple sclerosis is slower than previously reported. Neurology 2006; 66,172–7.

3. Tremlett H, Yinshan Zhao, Devonshire V. Natural history of secondary-progressive multiple sclerosis. Mult Scler . 2008; 14:314–24.

4. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: I. clinical course and disability. Brain . 1989; 112:133–46.

5. Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study: II. predictive value of the early clinical course. Brain 1989; 112:1419–28.

6. Kremenchutzky M, Rice GP, Baskerville J, Wingerchuk DM, Ebers GC. The natural history of multiple sclerosis: a geographically based study 9: observations on the progressive phase of the disease. Brain , 2006; 129: 584–94.

7. Pittock SJ, Mayr WT, McClelland RL, Jorgensen NW, Weigand SD, Noseworthy JH, Rodriguez M. Disability profile of MS did not change over 10 years in a population-based prevalence cohort. Neurology . 2004; 62:601–6.

8. Confavreux C, Vukusic S, Adeleine P. Early clinical predictors and progression of irreversible disability in multiple sclerosis: an amnesic process. Brain . 2003; 126:770–82.

9. Confavreux C, Vukusic S. Natural history of multiple sclerosis: a unifying concept. Brain . 2006; 129:606–16.

10. Runmarker B, Andersen O. Prognostic factors in a multiple sclerosis incidence cohort with twenty-five years of follow-up. Brain 1993; 116:117–34.

The distribution of the age at symptom-onset in a cohort of 1 , 463 patients with multiple sclerosis (MS). The age of symptom-onset in Figure is given as (mean ​± ​1 standard deviation) (Data from: Liguori M, Marrosu MG, Pugliatti M, Giuliani F, De Robertis F, Cocco E, Zimatore GB, Livrea P, Trojano M. Age at onset in multiple sclerosis. Neurol Sci. 2000; 21:S825-9). Modified with permission from: Goodin DS [ 6 ].

Table 2 The sex distribution for cases of multiple sclerosis prior to 1922 a . Table 2 Europe (Historical) United States (Historical) United States (Wechsler Series) Total Male 658 (58 ​ %) 99 (60 ​ %) 117 (59 ​ %) 874 (58 ​ %) Female 484 (42 ​ %) 67 (40 ​ %) 80 (41 ​ %) 631 (42 ​ %) Total 1142 (100 ​ %) 166 (100 ​ %) 197 (100 ​ %) 1505 (100 ​ %) a Data from: Wechsler IS [ 8 ]. Mostly, the historical cases represent those reported in the medical literature prior to 1903. Wechsler's series was drawn from the experience of the Mount Sinai Hospital, the Montefiore Hospital, and the Vanderbilt Clinic (1912–1921).

The sex distribution for cases of multiple sclerosis prior to 1922 a .

Data from: Wechsler IS [ 8 ]. Mostly, the historical cases represent those reported in the medical literature prior to 1903. Wechsler's series was drawn from the experience of the Mount Sinai Hospital, the Montefiore Hospital, and the Vanderbilt Clinic (1912–1921).

The symptoms of MS reflect its multifocal CNS pathology ( see section on Pathology of MS; below ) and commonly include visual impairment, weakness, muscle spasms, sensory disturbances (including pain and paresthesias), incoordination, gait abnormalities, bladder dysfunction, and/or double vision [ 5 ]. Three basic disease courses of MS are recognized [ 5 , 6 ] – relapsing/remitting MS ( RRMS ), secondary progressive MS ( SPMS ) and primary progressive MS ( PPMS ). However, whether these different disease types reflect a variable clinical expression of a single disease-process or whether they reflect different diseases with different pathophysiological underpinnings is less clear. Nevertheless, despite the fact that the sex ratio in PPMS is more even and the age at symptom-onset is older [ 10 ] – see also
Table 1 – both the genetic and environmental disease determinants for all three clinical presentations seem quite similar [ [10] , [11] , [12] , [13] , [14] ]. Therefore, a variable clinical expression for a single disease-process seems most likely. RRMS – approximately 85–90 % of MS cases at onset – is characterized by recurrent self-limited episodes of neurological dysfunction, which develop acutely, evolve over hours to days, last days to months, and are followed by a recovery of neurological function (sometimes incomplete). Between episodes, the individual is stable neurologically. SPMS begins as typical RRMS but, at some point, the disease-course changes. The acute episodes become less frequent, and the patient begins to experience an insidious and progressive decline in their neurological function, independent of acute episodes. PPMS – approximately 10–15 % of MS cases at onset – is characterized by an insidious onset of symptoms. Moreover, from the outset, the patient has a steadily progressive decline in neurological function without acute episodes (although acute attacks and remissions can occur subsequently).

Some individuals who have pathologically-verified MS may not be discovered until autopsy confirms the diagnosis. Indeed, based on four such autopsy series, notably undertaken prior to the wide-spread use of magnetic resonance imaging ( MRI ), the prevalence of such “asymptomatic” MS was reported to be approximately 100 cases per 100,000 individuals living in northern latitudes [ [15] , [16] , [17] , [18] ]. If so, and assuming an MS-prevalence ( i.e., diagnosed cases during a comparable Time Period ) of approximately 150 cases per 100,000 population in these geographical regions ( see below ), this suggests that, over their entire lifetime, ∼ 25 % of pathologically-verified MS cases do not experience clinical symptoms sufficient to bring them to medical attention – see the legend of
Table 3 for the basis of this estimate. With advent of using MRI as a critical aid for diagnosis ( Fig. 2 ), however, the finding of radiographic disease in persons who have not yet experienced neurological symptoms is now common-place [ [19] , [20] , [21] ]. Nevertheless, about a third to a half of these patients with the so-called “ radiographically-isolated ” syndrome ( RIS ) will have experienced clinical symptoms indicative of MS after 5–10 years [ [22] , [23] , [24] ]. Moreover, of the lesions visualized on T2-weighted, fluid attenuated inversion recovery (FLAIR), brain MRI ( Fig. 2 ), only about 5–10% of them produce neurological symptoms [ 25 ] and even large spinal cord lesions can be asymptomatic [ 26 ]. Consequently, the revised MS-diagnostic criteria (under development) will likely make an MS-diagnosis possible, even in the absence of any symptoms whatsoever. Table 3 MS prevalence, MZ -twin concordance { P ( M S | M Z M S ) }, and “ genetic susceptibility ” { P ( G ) } – ( see Text ) – in different geographical locations. Table 3 Geographical location MS prevalence a P ( M S | M Z M S ) b Latitude P ( G ) c North America Canada [ 43 ] 68–248 0.253 N45-60° 0.01–0.04 Northern USA [ 46 ] 100–192 0.314 N41-45° 0.01–0.02 Southern USA [ 46 ] 22–112 0.174 N30-41° 0.005–0.02 Europe Finland [ 48 ] 52–219 0.462 N60-70° 0.004–0.02 Denmark [ 44 , 45 ] 110–189 0.240 N55-58° 0.02–0.03 British Isles [ 42 ] 74–231 0.400 N50-59° 0.007–0.02 France [ 41 ] 32–151 0.111 N44-50° 0.01–0.05 Sardinia [ 47 ] 144–342 0.222 N39-41° 0.02–0.06 Italy [ 47 ] 38–203 0.145 N38-46° 0.01–0.05 a Estimates for the MS-prevalence (in cases per 100,000 population) for each region are taken from Refs. [ [31] , [32] , [33] , [34] , [35] , [36] , [37] , [38] ]. A range is given because there are more than one estimate available for each region. b MZ- twin concordances – i.e., P ( M S | M Z M S ) – are presented in the Table as proband-wise rates. These rates represent the maximum possible estimate for MZ -twin concordance from each study. Because the actual rates are often reported as pair-wise rates [ 57 ], however, the reported rates frequently require conversion [ 57 ]. Notably, however, this conversion assumes random sampling of twin-pairs [ 57 ] – an assumption which is almost certainly violated by many studies [e.g., 46]. Therefore, the rates listed in the Table provide only an upper-bound estimate for the MZ- twin concordance. c For the purposes of determining the probability of “ genetic-susceptibility ”, P ( G ) – see Text – in each region, we used the relationship that: P ( M S ) ≈ 2 ∗ ( t h e M S p r e v a l e n c e ) ​− ​ see Refs. [ 58 , 64 ]. To compensate for the impact of twins sharing environments, the MZ -twin concordance rates have been adjusted [ 64 ], using the Canadian data [ 43 ], for the difference in concordance rate between dizygotic ( DZ )-twins ( 5.4 % ) and non-twin siblings (2.9 %), such that: P ( M S | I G M S ) = ( 2.9 5.4 ) ∗ P ( M S | M Z M S ) {NB: In these calculations, we are assuming that: P ( M S | G ) = P ( M S | I G M S ) . This assumption is probably incorrect and could be off by a factor of ( 2 ) or more. Therefore, the estimates for P ( G ) could be more than double the values presented in the Table. Taking this and other uncertainties into account, yields the supported range of : { 0.001 < P ( G ) ≤ 0.52 }, from the Canadian MS-data [ 64 ]. Finally , P ( G ) is determined from the ratio of the MS-penetrance for the population ( Z ) to that for the subset ( G ) , such that : { P ( G ) = P ( M S ) / P ( M S | G ) } – see subsection on the Genetic and Environmental Susceptibility to MS. If ( Z ) and ( G ) are identical then, during every specific Time-Period ( E T ) , the MS-penetrance of these two sets will be identical, so that this ratio will always yield: { P ( G ) = 1 } . By contrast, whenever the MS-penetrance of ( G ) is greater than that of ( Z ) , then this ratio will yield: P ( G ) < 1 and, in this case, some individuals in the population cannot get MS regardless of any environmental conditions they either have experienced of could have experienced [ 64 ] } . Fig. 2 Characteristic pattern and distribution of MRI lesions in a patient with MS. Panel A is an axial T2-weighted FLAIR image demonstrating the typical periventricular, subcortical, and central white matter lesions of MS. The white arrow points to a juxtacortical lesion. Panel B is an axial T1-weighted, gadolinium (Gd)-enhanced image demonstrating enhancement of the lesion adjacent to the frontal pole of the L lateral ventricle (white arrow). This enhancement represents the acute focal breakdown of the Blood-Brain-Barrier in an active MS lesion. Gd-enhancement generally lasts (<3 months). The other, non-enhancing, lesions are, therefore, presumably more chronic than this. Panel C is a sagittal T2-weighted FLAIR image demonstrating typical lesions of the corpus callosum in patients with MS (white arrow). These were initially described by the Scottish pathologist James Dawson in 1916 and are often referred to as “Dawson's Fingers”. They represent inflammation that surrounds the draining medullary veins. These veins are oriented at right angles to the ventricular surface and, thus, when inflamed, appear, both on MRI and on pathological examination, as vertical spikes (fingers) off the ventricular surface. Panel D is a sagittal T2-weighted image of the cervical spinal cord demonstrating a demyelinating lesion in the posterior cervical spine at the level of C3 (white arrow). Fig. 2

Table 3 MS prevalence, MZ -twin concordance { P ( M S | M Z M S ) }, and “ genetic susceptibility ” { P ( G ) } – ( see Text ) – in different geographical locations. Table 3 Geographical location MS prevalence a P ( M S | M Z M S ) b Latitude P ( G ) c North America Canada [ 43 ] 68–248 0.253 N45-60° 0.01–0.04 Northern USA [ 46 ] 100–192 0.314 N41-45° 0.01–0.02 Southern USA [ 46 ] 22–112 0.174 N30-41° 0.005–0.02 Europe Finland [ 48 ] 52–219 0.462 N60-70° 0.004–0.02 Denmark [ 44 , 45 ] 110–189 0.240 N55-58° 0.02–0.03 British Isles [ 42 ] 74–231 0.400 N50-59° 0.007–0.02 France [ 41 ] 32–151 0.111 N44-50° 0.01–0.05 Sardinia [ 47 ] 144–342 0.222 N39-41° 0.02–0.06 Italy [ 47 ] 38–203 0.145 N38-46° 0.01–0.05 a Estimates for the MS-prevalence (in cases per 100,000 population) for each region are taken from Refs. [ [31] , [32] , [33] , [34] , [35] , [36] , [37] , [38] ]. A range is given because there are more than one estimate available for each region. b MZ- twin concordances – i.e., P ( M S | M Z M S ) – are presented in the Table as proband-wise rates. These rates represent the maximum possible estimate for MZ -twin concordance from each study. Because the actual rates are often reported as pair-wise rates [ 57 ], however, the reported rates frequently require conversion [ 57 ]. Notably, however, this conversion assumes random sampling of twin-pairs [ 57 ] – an assumption which is almost certainly violated by many studies [e.g., 46]. Therefore, the rates listed in the Table provide only an upper-bound estimate for the MZ- twin concordance. c For the purposes of determining the probability of “ genetic-susceptibility ”, P ( G ) – see Text – in each region, we used the relationship that: P ( M S ) ≈ 2 ∗ ( t h e M S p r e v a l e n c e ) ​− ​ see Refs. [ 58 , 64 ]. To compensate for the impact of twins sharing environments, the MZ -twin concordance rates have been adjusted [ 64 ], using the Canadian data [ 43 ], for the difference in concordance rate between dizygotic ( DZ )-twins ( 5.4 % ) and non-twin siblings (2.9 %), such that: P ( M S | I G M S ) = ( 2.9 5.4 ) ∗ P ( M S | M Z M S ) {NB: In these calculations, we are assuming that: P ( M S | G ) = P ( M S | I G M S ) . This assumption is probably incorrect and could be off by a factor of ( 2 ) or more. Therefore, the estimates for P ( G ) could be more than double the values presented in the Table. Taking this and other uncertainties into account, yields the supported range of : { 0.001 < P ( G ) ≤ 0.52 }, from the Canadian MS-data [ 64 ]. Finally , P ( G ) is determined from the ratio of the MS-penetrance for the population ( Z ) to that for the subset ( G ) , such that : { P ( G ) = P ( M S ) / P ( M S | G ) } – see subsection on the Genetic and Environmental Susceptibility to MS. If ( Z ) and ( G ) are identical then, during every specific Time-Period ( E T ) , the MS-penetrance of these two sets will be identical, so that this ratio will always yield: { P ( G ) = 1 } . By contrast, whenever the MS-penetrance of ( G ) is greater than that of ( Z ) , then this ratio will yield: P ( G ) < 1 and, in this case, some individuals in the population cannot get MS regardless of any environmental conditions they either have experienced of could have experienced [ 64 ] } .

MS prevalence, MZ -twin concordance { P ( M S | M Z M S ) }, and “ genetic susceptibility ” { P ( G ) } – ( see Text ) – in different geographical locations.

Estimates for the MS-prevalence (in cases per 100,000 population) for each region are taken from Refs. [ [31] , [32] , [33] , [34] , [35] , [36] , [37] , [38] ]. A range is given because there are more than one estimate available for each region.

MZ- twin concordances – i.e., P ( M S | M Z M S ) – are presented in the Table as proband-wise rates. These rates represent the maximum possible estimate for MZ -twin concordance from each study. Because the actual rates are often reported as pair-wise rates [ 57 ], however, the reported rates frequently require conversion [ 57 ]. Notably, however, this conversion assumes random sampling of twin-pairs [ 57 ] – an assumption which is almost certainly violated by many studies [e.g., 46]. Therefore, the rates listed in the Table provide only an upper-bound estimate for the MZ- twin concordance.

For the purposes of determining the probability of “ genetic-susceptibility ”, P ( G ) – see Text – in each region, we used the relationship that: P ( M S ) ≈ 2 ∗ ( t h e M S p r e v a l e n c e ) ​− ​ see Refs. [ 58 , 64 ]. To compensate for the impact of twins sharing environments, the MZ -twin concordance rates have been adjusted [ 64 ], using the Canadian data [ 43 ], for the difference in concordance rate between dizygotic ( DZ )-twins ( 5.4 % ) and non-twin siblings (2.9 %), such that: P ( M S | I G M S ) = ( 2.9 5.4 ) ∗ P ( M S | M Z M S ) {NB: In these calculations, we are assuming that: P ( M S | G ) = P ( M S | I G M S ) . This assumption is probably incorrect and could be off by a factor of ( 2 ) or more. Therefore, the estimates for P ( G ) could be more than double the values presented in the Table. Taking this and other uncertainties into account, yields the supported range of : { 0.001 < P ( G ) ≤ 0.52 }, from the Canadian MS-data [ 64 ]. Finally , P ( G ) is determined from the ratio of the MS-penetrance for the population ( Z ) to that for the subset ( G ) , such that : { P ( G ) = P ( M S ) / P ( M S | G ) } – see subsection on the Genetic and Environmental Susceptibility to MS. If ( Z ) and ( G ) are identical then, during every specific Time-Period ( E T ) , the MS-penetrance of these two sets will be identical, so that this ratio will always yield: { P ( G ) = 1 } . By contrast, whenever the MS-penetrance of ( G ) is greater than that of ( Z ) , then this ratio will yield: P ( G ) < 1 and, in this case, some individuals in the population cannot get MS regardless of any environmental conditions they either have experienced of could have experienced [ 64 ] } .

Characteristic pattern and distribution of MRI lesions in a patient with MS. Panel A is an axial T2-weighted FLAIR image demonstrating the typical periventricular, subcortical, and central white matter lesions of MS. The white arrow points to a juxtacortical lesion. Panel B is an axial T1-weighted, gadolinium (Gd)-enhanced image demonstrating enhancement of the lesion adjacent to the frontal pole of the L lateral ventricle (white arrow). This enhancement represents the acute focal breakdown of the Blood-Brain-Barrier in an active MS lesion. Gd-enhancement generally lasts (<3 months). The other, non-enhancing, lesions are, therefore, presumably more chronic than this. Panel C is a sagittal T2-weighted FLAIR image demonstrating typical lesions of the corpus callosum in patients with MS (white arrow). These were initially described by the Scottish pathologist James Dawson in 1916 and are often referred to as “Dawson's Fingers”. They represent inflammation that surrounds the draining medullary veins. These veins are oriented at right angles to the ventricular surface and, thus, when inflamed, appear, both on MRI and on pathological examination, as vertical spikes (fingers) off the ventricular surface. Panel D is a sagittal T2-weighted image of the cervical spinal cord demonstrating a demyelinating lesion in the posterior cervical spine at the level of C3 (white arrow).

In general, MS-prevalence increases with increasing geographical distance from the equator [ [27] , [28] , [29] , [30] , [31] , [32] , [33] , [34] , [35] , [36] , [37] , [38] , [39] , [40] ] – see
Fig. 3 . This gradient is also evident within a single country when this has been reported. For example, the relative prevalence estimates are: North Scotland ​> ​South Scotland [ 30 ]; North USA ​> ​South USA [ 31 , 35 ]; South New Zealand ​> ​North New Zealand [ 29 , 40 ]; Tasmania ​> ​North Australia [ 28 ]. Nevertheless, in some regions (e.g., the island of Sardinia – see
Table 3 ) the prevalence estimates seem unexpectedly high compared to geographic expectations [ 32 , 37 ]. Moreover, regardless of latitude, the fact that the monozygotic ( MZ )-twin concordance rates seem to mirror the prevalence rates, even in Sardinia ( Table 3 ), strongly favors environmental events as the principal basis for the observed latitude gradient [ [41] , [42] , [43] , [44] , [45] , [46] , [47] , [48] ]. Fig. 3 The Upper Panel represents the distribution of MS-prevalence around the world. Dotted black line indicates the Equator. Note that the MS-prevalence increases with increased latitude both north and south of the equator – data taken from reviews of worldwide epidemiology of MS [ 5 , 32 ]. The Lower Panel represents the global distribution ultraviolet B ( UVB ) radiation reaching the earth's surface divided into three Exposure Zones . Zone 1 indicates regions where there is a sufficient amount UVB radiation reaching the earth's surface for adequate vitamin D synthesis during ( 12 / 12 ) months of the year. Zone 3 indicates regions where there is an insufficient amount UVB radiation reaching the earth's surface for adequate vitamin D synthesis during 12 / 12 months of the year. In Zone 2 (where there is a variable insufficiency), an increasing number of insufficient months indicated by the coloration changing from reds to oranges to yellows to greens. The red indicated that there is a sufficient amount of UVB radiation reaching the earth's surface for vitamin D synthesis during (11/12) months during the year. The darkest green indicates that there is a sufficient amount of UVB radiation reaching the earth's surface for vitamin D synthesis during only 1 / 12 months of the year. The amount of UVB radiation also varies in Zones 1 and 3 but is either always adequate ( Zone 1 ) or always inadequate ( Zone 3 ). Modified with permission from: Goodin DS [ 6 ] and from: Jablonski NG and Chaplin G [ 163 ]. Fig. 3

The Upper Panel represents the distribution of MS-prevalence around the world. Dotted black line indicates the Equator. Note that the MS-prevalence increases with increased latitude both north and south of the equator – data taken from reviews of worldwide epidemiology of MS [ 5 , 32 ]. The Lower Panel represents the global distribution ultraviolet B ( UVB ) radiation reaching the earth's surface divided into three Exposure Zones . Zone 1 indicates regions where there is a sufficient amount UVB radiation reaching the earth's surface for adequate vitamin D synthesis during ( 12 / 12 ) months of the year. Zone 3 indicates regions where there is an insufficient amount UVB radiation reaching the earth's surface for adequate vitamin D synthesis during 12 / 12 months of the year. In Zone 2 (where there is a variable insufficiency), an increasing number of insufficient months indicated by the coloration changing from reds to oranges to yellows to greens. The red indicated that there is a sufficient amount of UVB radiation reaching the earth's surface for vitamin D synthesis during (11/12) months during the year. The darkest green indicates that there is a sufficient amount of UVB radiation reaching the earth's surface for vitamin D synthesis during only 1 / 12 months of the year. The amount of UVB radiation also varies in Zones 1 and 3 but is either always adequate ( Zone 1 ) or always inadequate ( Zone 3 ). Modified with permission from: Goodin DS [ 6 ] and from: Jablonski NG and Chaplin G [ 163 ].

Measuring the population prevalence in many world-wide locations can be a somewhat imprecise (especially in remote or low-prevalence areas) although, regardless of any uncertainties that one might have regarding the estimated prevalence from any specific region, if one takes a “world-wide” view, the gradient seems readily apparent ( Fig. 3 ). Nevertheless, three estimates are noteworthy as they seem to be particularly accurate, at least for the populations they represent. The first is the nationwide Canadian Collaborative Project on Genetic Susceptibility to Multiple Sclerosis ( CCPGSMS ) study [ 7 , 43 , [49] , [50] , [51] , [52] , [53] , [54] , [55] , [56] ], which was begun in the mid to late 1990s [ [49] , [50] , [51] , [52] ] and initially identified 29,478 patients with MS in Canada who were born between the years of 1891 and 1993 [ 7 ]. Using statistical methods, which depend only upon doubly-ascertained twin-pairs [ 57 ], the authors estimate that their sample includes 65.3 ​ % of the MS patients in Canada at the time ( Table 4 ). Thus, based on a total population of 30.1 million persons in Canada in 2001 [ 43 ], the MS-prevalence in Canada can be estimated as 150.1 cases per 100,000 population ( Table 4 ). Table 4 Multiple sclerosis: epidemiological data from Canada circa ( 2000 –2015 ). Table 4 Canadian data for MZ twin-pairs a Females Males Totals Concordant for MS 22 2 24 Discordant for MS 66 43 109 Totals 88 45 133 Proband-wise concordance b 0.340 0.065 0.253 Concordance ratio ( F / M ) 5.231 − − Canadian data for MZ twin-pairs a D R B 1 ∗ 15 + D R B 1 ∗ 15 − Totals Concordant for MS 9 11 20 Discordant for MS 31 42 73 Totals 40 53 93 Proband-wise concordance b 0.309 0.288 0.297 Concordance ratio ( 15 + / 15 − ) 1.073 − − Population data for Canada in 2001-2010: T o t a l P o p u l a t i o n = 30 , 081 , 000 i n d i v i d u a l s – in 2001 ; from Reference [ 43 ] P ( F ) = 0.505 – from the 2001 Canadian Census c Case ascertainment in the CCPGSMS: Estimated from the doubly ascertained concordant twin pairs – i.e., ( 16 ) o u t o f ( 33 ) T o t a l ( 16 ) D o u b l y A s c e r t a i n e d / ( 33 − 8.5 ) = 65.3 % – from Refs . [ 43 , 57 ]{NB: This “double ascertainment” estimate is independent of the twin-rate and the MS-prevalence} Summary Data for MS-Concordance among DZ-Twins and non-twin siblings in Canada: P ( M S | D Z M S ) = 0.054 from Reference [ 43 ] P ( M S | S M S ) = 0.029 from Reference [ 43 ] Preponderance of Women among MS patients in Canada During Time-Period #1 ( 1941–1945 ): P ( F | M S ) 1 = 0.685 – from Reference [ 7 ] During Time-Period #2 ( 1976–1980 ): P ( F | M S ) 2 = 0.762 – from Reference [ 7 ] a Data drawn from the MS-patients in the CCPGSMS database as of ( ∼ 2001 ) – Reference [ 43 ]. b Proband-wise concordance according to Refs. [ 43 , 57 ] -- adjusted for doubly ascertained concordant MZ- twin pairs ( 13 24 = 54 % ) -- Proband-wise Concordance = P ( M S | M Z M S ) c Canadian Census. 2010. Available from: https://www150.statcan.gc.ca/n1/en/pub/89-503-x/2010001/article/11475-eng.pdf?st=WVL9_Ggm -- last accessed 19 February 2025.

Table 4 Multiple sclerosis: epidemiological data from Canada circa ( 2000 –2015 ). Table 4 Canadian data for MZ twin-pairs a Females Males Totals Concordant for MS 22 2 24 Discordant for MS 66 43 109 Totals 88 45 133 Proband-wise concordance b 0.340 0.065 0.253 Concordance ratio ( F / M ) 5.231 − − Canadian data for MZ twin-pairs a D R B 1 ∗ 15 + D R B 1 ∗ 15 − Totals Concordant for MS 9 11 20 Discordant for MS 31 42 73 Totals 40 53 93 Proband-wise concordance b 0.309 0.288 0.297 Concordance ratio ( 15 + / 15 − ) 1.073 − − Population data for Canada in 2001-2010: T o t a l P o p u l a t i o n = 30 , 081 , 000 i n d i v i d u a l s – in 2001 ; from Reference [ 43 ] P ( F ) = 0.505 – from the 2001 Canadian Census c Case ascertainment in the CCPGSMS: Estimated from the doubly ascertained concordant twin pairs – i.e., ( 16 ) o u t o f ( 33 ) T o t a l ( 16 ) D o u b l y A s c e r t a i n e d / ( 33 − 8.5 ) = 65.3 % – from Refs . [ 43 , 57 ]{NB: This “double ascertainment” estimate is independent of the twin-rate and the MS-prevalence} Summary Data for MS-Concordance among DZ-Twins and non-twin siblings in Canada: P ( M S | D Z M S ) = 0.054 from Reference [ 43 ] P ( M S | S M S ) = 0.029 from Reference [ 43 ] Preponderance of Women among MS patients in Canada During Time-Period #1 ( 1941–1945 ): P ( F | M S ) 1 = 0.685 – from Reference [ 7 ] During Time-Period #2 ( 1976–1980 ): P ( F | M S ) 2 = 0.762 – from Reference [ 7 ] a Data drawn from the MS-patients in the CCPGSMS database as of ( ∼ 2001 ) – Reference [ 43 ]. b Proband-wise concordance according to Refs. [ 43 , 57 ] -- adjusted for doubly ascertained concordant MZ- twin pairs ( 13 24 = 54 % ) -- Proband-wise Concordance = P ( M S | M Z M S ) c Canadian Census. 2010. Available from: https://www150.statcan.gc.ca/n1/en/pub/89-503-x/2010001/article/11475-eng.pdf?st=WVL9_Ggm -- last accessed 19 February 2025.

Multiple sclerosis: epidemiological data from Canada circa ( 2000 –2015 ).

Data drawn from the MS-patients in the CCPGSMS database as of ( ∼ 2001 ) – Reference [ 43 ].

Proband-wise concordance according to Refs. [ 43 , 57 ] -- adjusted for doubly ascertained concordant MZ- twin pairs ( 13 24 = 54 % ) -- Proband-wise Concordance = P ( M S | M Z M S )

Canadian Census. 2010. Available from: https://www150.statcan.gc.ca/n1/en/pub/89-503-x/2010001/article/11475-eng.pdf?st=WVL9_Ggm -- last accessed 19 February 2025.

Second, in 2020, Italian investigators published a systematic study that included all intra-regional local health authorities and all MS clinical units on the island of Sardinia, using both a clinical database (including all Sardinian neurologists in these clinical units to provide the clinical data) and an island-wide health administrative database [ 37 ]. Out of a total population of 1,663,251 inhabitants residing in Sardinia on “prevalence day” (3/31/16), the authors found 5680 unique individuals with definite MS (of any age) and another 412 persons who had a CIS . Excluding those individuals with a CIS , this yields an estimated crude prevalence of 341.5 cases per 100,000 population on Sardinia.

And third, in a recent study from Denmark [ 38 ], the authors used several nationwide registries with an expected coverage of 97 ​ % of the Danish population and identified 28,145 unique patients with definite MS. Notably, Denmark has a universal healthcare system, which provides equal access to diagnostics and treatment for all inhabitants and does not require informed consent or ethical approval to conduct observational register-based studies – thereby substantially increasing the likelihood of complete case ascertainment for the entire country. This study documents a steady increase both in the prevalence of MS in Denmark over the last half century and in the proportion of women among Danish MS patients – from 89.1 cases per 100,000 population and 58.7 % women in 1975 to 295.0 cases per 100,000 population and 68.5 % women in 2023 [ 38 ].

Multiple reports around the world, have similarly documented that the MS-prevalence is increasing, especially among women [ 7 , 38 , 43 , [59] , [60] , [61] , [62] , [63] , [64] ]. Indeed, based upon the point-estimates from the large CCPGSMS study from Canada cited previously [ 7 , 43 , [49] , [50] , [51] , [52] , [53] , [54] , [55] , [56] ], the MS-prevalence in Canada must have increased by more than 32 % (and likely much more) between the two 5-year birth epochs of: ( 1941 – 1945 ) and ( 1976 – 1980 ) and the proportion of women among MS patients has increased from ( 68.5 % ) to ( 76.2 % ) over this same time-interval [ 7 , 64 ] – see also
Table 4 . For example, a similar minimum estimate of a ( > 32 % ) increase in prevalence is also calculated from the change in the ( F:M ) sex-ratio for the Danish population discussed above [ 38 ]. Nevertheless, in this case, the actual increase in prevalence is known to be ( 295.0 / 89.2 = 3.1 ) or a greater than 3-fold increase − i.e., a number well above the calculated 32 % minimum [ 38 ].

Finally, when considering the environmental and genetic bases of MS, it is important to recognize that MS-pathogenesis involves two distinct components. The first is related to those environmental events, genetic factors, and interactions, which lead to the onset of the disease. The second, is related to those events, factors, and interactions, which lead to the ongoing clinical expression of the disease (i.e., to causing the MS-attacks). These two sets of events, factors, and interactions may not be the same for each component of disease pathogenesis and this topic is considered elsewhere ( see sub-section on Sex-based Therapies; below ).

Pathology of MS

Pathologically, MS is characterized by multi-focal areas of inflammation, demyelination and gliosis (also referred to as plaques or lesions) within the CNS white matter ( Fig. 4 ). These lesions appear in many different locations throughout the CNS , although certain white matter regions (e.g., optic nerves, corpus callosum, peri-ventricular, juxtacortical, and spinal cord) seem to be particularly vulnerable areas [ [65] , [66] , [67] , [68] , [69] ] – see
Fig. 2 , Fig. 4 . In many cases these plaques surround a central vein ( Fig. 4 ), which likely represents the onset-location of the inflammatory process [ 5 ]. Indeed, within these plaques, the white-matter tissue injury is often of differing ages with more chronic inflammation being present centrally (i.e., immediately adjacent to the central vein) and acute inflammation being more conspicuous peripherally [ 5 ]. Injury commonly occurs both to the myelin sheaths and to the oligodendrocytes, although axonal injury – especially within active lesions – and grey matter demyelination – especially late in the disease course – also occur [ [69] , [70] , [71] , [72] , [73] ]. Guided by pro-inflammatory cytokines and utilizing vascular cell adhesion molecule-1 ( VCAM-1 ), immune cells cross the blood–brain barrier ( BBB ) and enter the CNS to produce the inflammatory lesions and tissue injury. The VCAM-1 protein (expressed on both large and small blood vessels after the endothelial cells are stimulated by cytokines) is the principle endothelial ligand for the very late antigen-4 ( VLA-4 ), which is an integrin and is expressed on the surface of eosinophils, monocytes, macrophages, natural killer cells, and lymphocytes (both B and T cells), but not neutrophils. VLA-4 is a transmembrane heterodimer ( α 4 β 1 ) , which binds to activated VCAM-1 and mediates leukocyte attachment and transmigration through the cytokine-activated vascular epithelium. These invading cells include cluster of differentiation (CD)4+ T cells, CD20+ B cells, CD8+ cytotoxic lymphocytes, and CD68+ macrophages [ [64] , [65] , [66] , [67] ]. Lesions that, on MRI, have only T2-hyperintensity, are more likely to lack of evidence of demyelination on histopathological examination ( 55 − 80 % ) compared to only ( 17 − 20 % ) of lesions that have T2-hyperintensity, a reduced magnetization transfer ratio (MTR), and a persistent T1-hypointensity [ 74 , 75 ]. Nevertheless, regardless of whether the myelin is preserved, most active MS lesions contain activated microglia and evidence of BBB breakdown and, thus, on MRI, generally show gadolinium-enhancement ( G d + ) – e.g., see
Fig. 2 . Even within normal appearing white matter, 30 % of regions are still found to contain activated microglia on histopathological examination [ 74 , 75 ], although it is unclear what role these play in the disease process. In sum, it is clear that MS lesions have considerable histopathological diversity. The pathological changes range from chronic, gliotic, and demyelinated scars, to acute inflammatory lesions with obvious injury to the myelin sheaths, oligodendrocytes, and axons, and, finally, to less inflammatory lesions, in which myelin sheaths, oligodendrocytes and axons seem to be completely spared [ 5 , [69] , [70] , [71] , [72] , [73] , [74] , [75] ]. Fig. 4 Gross pathological and histopathological appearance of MS lesions. Panel A is a section of the ‘naked eye’ appearance of a coronal cut section from an unfixed brain of a patient with MS. Black arrows indicate the chronic lesions (plaques) of MS in the subcortical, periventricular and central white matter. Panel B represents a coronal section from the brain of an MS patient (luxol fast blue stain) and demonstrates the areas of chronic demyelination in the central and periventricular white matter. The solid black arrows indicate areas of complete demyelination. The rounded periventricular lesions are chronic; the star-like lesion is active. The dashed black arrows indicate the appearance of so-called “shadow” plaques. These shadow plaques may represent early active lesion formation, aborted lesions, or areas where partial, but incomplete, re-myelination has taken place. Panels C and D (hematoxylin and eosin stain) demonstrates the perivenular infiltrate of lymphocytes and plasma cells in an active MS lesion. Panel C is cut along the venule's course whereas Panel D is cross sectional. Modified with permission from: Adams CWM. (1989) Color Atlas of Multiple Sclerosis and Other Myelin Disorders. Wolfe Medical Publications Ltd. Printed by Cowell WS, Ipswich, Suffolk, United Kingdom. Fig. 4

Gross pathological and histopathological appearance of MS lesions. Panel A is a section of the ‘naked eye’ appearance of a coronal cut section from an unfixed brain of a patient with MS. Black arrows indicate the chronic lesions (plaques) of MS in the subcortical, periventricular and central white matter. Panel B represents a coronal section from the brain of an MS patient (luxol fast blue stain) and demonstrates the areas of chronic demyelination in the central and periventricular white matter. The solid black arrows indicate areas of complete demyelination. The rounded periventricular lesions are chronic; the star-like lesion is active. The dashed black arrows indicate the appearance of so-called “shadow” plaques. These shadow plaques may represent early active lesion formation, aborted lesions, or areas where partial, but incomplete, re-myelination has taken place. Panels C and D (hematoxylin and eosin stain) demonstrates the perivenular infiltrate of lymphocytes and plasma cells in an active MS lesion. Panel C is cut along the venule's course whereas Panel D is cross sectional. Modified with permission from: Adams CWM. (1989) Color Atlas of Multiple Sclerosis and Other Myelin Disorders. Wolfe Medical Publications Ltd. Printed by Cowell WS, Ipswich, Suffolk, United Kingdom.

The CNS endothelial cells are not fenestrated and have extraordinarily tight junctions, which help to form the BBB . The actual barrier, however, is produced by a complex interaction between the endothelial cells, the oligodendrocytes, the axons, and the vascular system. Collectively, this system is referred to as the neurovascular unit, which provides mutual trophic support to its elements and makes the entry of hydrophilic molecules into the CNS (by active transport, diffusion, or transcytosis) extremely selective [ [76] , [77] , [78] ]. A breakdown of the BBB can be caused through a variety of mechanisms including trauma, ischemia, neoplasia, inflammation, and toxic exposure, is generally thought both to be a crucial step in the development of a new MS lesion and, also, is thought to be the pathogenic basis of an acute MS attack. Nevertheless, utilizing the technique of MTR imaging, changes in the relative concentrations of free and bound water are found in focal areas of the CNS , which otherwise appear normal on MRI [ 79 , 80 ]. Notably, however, months later, these same CNS areas are destined to become ( G d + ) lesions on MRI [ 79 , 80 ]. It seems likely that these MTR changes reflect biochemical signatures related to the initial events in lesion formation, which are undetectable by Gd-enhanced MRI. If so, the breakdown of the BBB , as reflected by Gd-enhancement, would be seen as a secondary event.

Genetic and Environmental Suseptability to MS

Chronic diseases such as MS typically have complex etiological bases and both the individual's genetic make-up and their environmental experiences are critical components of whether an individual will develop the disease [ 81 ]. When assessing the role that different environmental events and genetic factors play in MS-pathogenesis, it will sometimes prove useful to define, specifically, both the general population of interest (e.g., the population of the Canada, Sardinia, Denmark, or some other specific region) and, in addition, some of its subsets [ 64 ]. In this context, therefore, a population ( Z ) will be considered. This population consists of N individuals ( k = 1 , 2 , … , N ) , each having a unique genotype ( G k ) . This uniqueness even applies to MZ- twins who, despite having “ identical ” genotypes ( I G ) , still have subtle genetic differences from one another [ 58 ]. The “ genetically-susceptible” subset of this population, ( G ) , includes everyone who has any non-zero life-time chance of developing MS under some environmental conditions. Even individuals who can only develop MS under improbable (or extreme) conditions (e.g., being intentionally inoculated with demyelinating or other substances), are still members of the ( G ) subset – i.e., they can develop MS under some environmental conditions. Each of the ( m ≤ N ) individuals in the ( G ) subset ( i = 1 , 2 , … , m ) has a unique genotype ( G i ) . The event ( G ) is defined as the event that a proband , randomly-selected from ( Z ) , is a member of the ( G ) subset and the probability of this event is: P ( G ) = m / N [ 64 ]. The event ( F ) represents the event that a proband , randomly selected from the population ( Z ) , is a female and the probability of this event is P ( F ) . The event ( M ) represents the event that this proband is a male and its probability is: P ( M ) = 1 − P ( F ) . The ( M S ) subset includes everyone in ( Z ) who either has, or will subsequently develop, MS. The event ( M S ) is defined as the event that a proband , randomly-selected from ( Z ) , is a member of the ( M S ) subset. The probability of this event is: P ( M S ) , which is also called the MS-penetrance for the population ( Z ) . The comparable MS-penetrance for the subset ( G ) is termed: P ( M S | G ) . Both penetrance values depend upon the environmental conditions of the time [ 64 ]. During any Time-Period
( E T ) , the ratio of these two penetrance values estimates the proportion of the population who are members of the genetically-susceptible subset [ 64 ] such that: P ( G ) = P ( M S | E T ) / P ( M S | G , E T )

For each member of the ( G ) subset, a family of “ sufficient ” exposures is defined to include every set of environmental exposures, each of which is sufficient , by itself, to cause MS in that person [ 64 ]. Moreover, there can be any number of “ sufficient ” exposure sets within each susceptible person's family. The event ( E ) is defined as the event that a proband , randomly-selected from the ( G ) subset, experiences any (one or more) of their sets of “ sufficient ” environmental conditions [ 64 ]. The probability of this event, during some specific Time-Period
( E T ) – see Ref. [ 64 ] − is represented as: P ( E | G , E T ) . Moreover, each of these sufficient-exposure sets is completely undefined with respect to: 1) when, during life, and in what order, these exposures need to occur; 2) the intensity and duration of the required exposures; 3) how many exposures are involved; 4) what these exposures are; 5) whether certain exposures need to be present or absent.; and 6) whether any of these exposures needs to interact with any genetic-factors. The only requirement is that every member of a susceptible-individual's exposure-set family, taken together, is sufficient , by itself, to cause MS to develop in that specific person [ 64 ].

The event ( M Z ) represents the event that a proband , randomly selected from the population ( Z ) , is an MZ- twin – i.e., is a member of the ( M Z ) subset. The twin of this proband is referred to as their “ co-twin ”. The probability that the proband is a member of the ( M S , M Z ) subset, given that their co-twin is a member of ( M Z ) , is the same as the probability that the co-twin is a member of ( M S , M Z ) , given that the proband is a member of ( M Z ) . Consequently, to distinguish between these two circumstances, the term ( M S , M Z ) represents this event (or subset) for the proband , whereas the term ( M Z M S ) represents the same event (or subset) for the co-twin , given that both twins are members of the ( M Z ) subset. Therefore: P ( M Z M S ) = P ( M S , M Z | M Z ) = P ( M S | M Z )

The comparable events (or subsets) for DZ “ co-twins ” ( D Z M S ) and for non-twin “ co-siblings ” ( S M S ) are defined in a similar manner [ 64 ]. Consequently, P ( M S | M Z M S ) indicates the life-time probability that a proband , randomly-selected from ( Z ) , is a member of ( M S , M Z ) , given that their co-twin is a member of ( M Z M S ) – a probability estimated by the proband-wise (or case-wise ) MZ- twin concordance rate [ 57 ]. This MZ -twin concordance rate – that is, P ( M S | M Z M S ) – may require some adjustment because MZ- twins, in addition to sharing their “ identical ” genotypes ( I G ) , also share some of their environments (e.g., intrauterine, early post-natal and, possibly, others). This adjusted rate – termed P ( M S | I G M S ) – isolates the genetic contribution to the observed MZ- twin concordance from these environmental similarities [ 58 , 64 ]. Notably, the subset ( I G ) is identical to the subset ( M Z ) and, thus: P ( I G ) = P ( M Z ) .

Using these definitions and based upon observations, from multiple world regions, that both the prevalence of MS and the proportion of women among MS patients are increasing [ 7 , 38 , 43 , [59] , [60] , [61] , [62] , [63] , [64] ], the pathogenic implications of the Canadian CCPGSMS data (see Table 1 ) were analyzed in detail [ 64 ]. From this analysis, two principal conclusions emerged: (1) that the MS-penetrance for the genetically-susceptible subset, ( G ) , exceeds that for the population ( Z ) ; and: (2) that, as the probability of the event ( E ) increases to 100 %, the maximum (i.e., limiting) probability of developing MS in susceptible men
( c ) is less than that for susceptible women
( d ) . These two conclusions can be stated both explicitly and quantitatively as: (1) 0.001 < P ( G ) ≤ 0.52 (2) c < d ≤ 1

The first conclusion seems inescapable (both from the Canadian CCPGSMS data [ 7 , 43 , [49] , [50] , [51] , [52] , [53] , [54] , [55] , [56] ] and other MZ -twin data [ 64 ] – see also
Table 3 , Table 4 ) and, thus, not everyone in the population is genetically-susceptible ( as defined in the
section on Genetic and Environmental Susceptibility to MS; see above ). This indicates that some members of the population ( Z ) cannot develop MS regardless of any environmental experiences that they either have had or could have had (see Table 3 ). The second conclusion indicates that, even when a susceptible man experiences an environment “ sufficient ” to cause MS in him, the likelihood of him actually developing disease is less than 100 % . Such a circumstance implies that there must be a “ truly ” random factor, which plays a critical role in MS pathogenesis and, thus, that “ truly ” random events must be an integral feature of our universe [ 64 ]. An example of a typical solution to the epidemiological equations [ 64 ] is presented in Fig. 5 . Moreover, under the stated conditions of Fig. 5 , this random factor is entirely responsible for the currently-observed difference in MS disease-expression between susceptible women and susceptible men . Indeed, the fact that this random factor is primarily responsible for the currently-observed difference in MS disease-expression between susceptible women and susceptible men , is also true for all plausible solutions to these equations [ 64 ]. Fig. 5 Using the Canadian MS data ( Table 4 ), response-curves are depicted for developing MS to an increasing odds that a proband , randomly selected from the susceptible subset ( G ) , experiences the event ( E ) – i.e., they experience a set of environmental conditions that are “ sufficient ”, by themselves, to cause MS in them [ 64 ]. {NB: exposure here is being measured as the “odds” that the proband will experience a “sufficient” exposure [ 64 ]. This measure is related to the probability of a “ sufficient ” exposure , not to any specific exposure conditions} . Response-curves representing susceptible women (black lines) and men (red lines) are depicted separately. The constant { c = P ( M S | M , G , E ) } represents the limiting probability of developing MS in susceptible men whereas the constant { d = P ( M S | F , G , E ) } represents this limiting probability in susceptible women . The probability that a proband , randomly selected from the susceptible subset ( G ) is a female is: { P ( F | G ) = p } , whereas the probability that this proband is a male is: { P ( M | G ) = 1 − p } The hazard proportionality constant, ( R ) , if the hazards are proportional, is defined elsewhere [ 64 ]. The blue lines represent the change in the ( F:M ) sex-ratio with increasing exposure, with the scale provided in the upper left. The thin grey vertical lines represent the portion of the response-curves that corresponds to the observed change in the ( F:M ) sex-ratio – ( 2.2 → 3.2 ) – which took place in Canada between Time-Periods #1 & #2 (see Table 4 ). In constructing these depicted curves, three assumptions have been made. These are: (1) at every exposure level, the randomly selected proband (who experiences a “ sufficient ” exposure”) has some non-zero probability of being either a man or a women . Nevertheless, this probability may not be the same (or even similar) for the two sexes – i.e., we are assuming only that the threshold is the same for males and females or: ( λ = 0 ) [ 64 ]; (2) with any increase in exposure – i.e., any increase in P ( E | G ) – the probability of developing MS is increased for both susceptible men and susceptible women although, again, possibly with different degrees of increase for each sex; and: (3) “male-MS” and “female-MS” represent the same underlying disease process [ 64 ]. From these three assumptions and from the observations, both in Canada and around the world, that the MS-prevalence is increasing, especially among women [ 7 , [59] , [60] , [61] , [62] , [63] , [64] ] – see also Table 4 – four conclusions directly, and necessarily, follow [ 64 ]. First, the hazards are proportional. Second, at every exposure level, susceptible men are more likely than susceptible women to experience a “ sufficient ” exposure [ 64 ]. Third, at every exposure level a susceptible man who experiences a “sufficient exposure” less likely to actually develop MS than a susceptible woman in the same circumstances. This conclusion reflects a “ truly ” random mechanism in MS-pathogenesis [ 64 ]. And fourth, the difference in disease-expression between the sexes under the stated conditions is explained entirely by this “ truly ” random mechanism [ 64 ]. Thus, this “ truly ” random mechanisms is similar to a biased coin, which favors disease-development in suscveptible women over susceptible men . Notably, however, as noted previously [ 64 ], the fact that a coin is biased only indicates that, when the coin is flipped, the two possible outcomes are not equally likely. It does not imply that the coin-flip is a non-random process. The consequences of dropping or modifying any or all of these assumptions are considered elsewhere [ 64 ]. Fig. 5

Using the Canadian MS data ( Table 4 ), response-curves are depicted for developing MS to an increasing odds that a proband , randomly selected from the susceptible subset ( G ) , experiences the event ( E ) – i.e., they experience a set of environmental conditions that are “ sufficient ”, by themselves, to cause MS in them [ 64 ]. {NB: exposure here is being measured as the “odds” that the proband will experience a “sufficient” exposure [ 64 ]. This measure is related to the probability of a “ sufficient ” exposure , not to any specific exposure conditions} . Response-curves representing susceptible women (black lines) and men (red lines) are depicted separately. The constant { c = P ( M S | M , G , E ) } represents the limiting probability of developing MS in susceptible men whereas the constant { d = P ( M S | F , G , E ) } represents this limiting probability in susceptible women . The probability that a proband , randomly selected from the susceptible subset ( G ) is a female is: { P ( F | G ) = p } , whereas the probability that this proband is a male is: { P ( M | G ) = 1 − p } The hazard proportionality constant, ( R ) , if the hazards are proportional, is defined elsewhere [ 64 ]. The blue lines represent the change in the ( F:M ) sex-ratio with increasing exposure, with the scale provided in the upper left. The thin grey vertical lines represent the portion of the response-curves that corresponds to the observed change in the ( F:M ) sex-ratio – ( 2.2 → 3.2 ) – which took place in Canada between Time-Periods #1 & #2 (see Table 4 ). In constructing these depicted curves, three assumptions have been made. These are: (1) at every exposure level, the randomly selected proband (who experiences a “ sufficient ” exposure”) has some non-zero probability of being either a man or a women . Nevertheless, this probability may not be the same (or even similar) for the two sexes – i.e., we are assuming only that the threshold is the same for males and females or: ( λ = 0 ) [ 64 ]; (2) with any increase in exposure – i.e., any increase in P ( E | G ) – the probability of developing MS is increased for both susceptible men and susceptible women although, again, possibly with different degrees of increase for each sex; and: (3) “male-MS” and “female-MS” represent the same underlying disease process [ 64 ]. From these three assumptions and from the observations, both in Canada and around the world, that the MS-prevalence is increasing, especially among women [ 7 , [59] , [60] , [61] , [62] , [63] , [64] ] – see also Table 4 – four conclusions directly, and necessarily, follow [ 64 ]. First, the hazards are proportional. Second, at every exposure level, susceptible men are more likely than susceptible women to experience a “ sufficient ” exposure [ 64 ]. Third, at every exposure level a susceptible man who experiences a “sufficient exposure” less likely to actually develop MS than a susceptible woman in the same circumstances. This conclusion reflects a “ truly ” random mechanism in MS-pathogenesis [ 64 ]. And fourth, the difference in disease-expression between the sexes under the stated conditions is explained entirely by this “ truly ” random mechanism [ 64 ]. Thus, this “ truly ” random mechanisms is similar to a biased coin, which favors disease-development in suscveptible women over susceptible men . Notably, however, as noted previously [ 64 ], the fact that a coin is biased only indicates that, when the coin is flipped, the two possible outcomes are not equally likely. It does not imply that the coin-flip is a non-random process. The consequences of dropping or modifying any or all of these assumptions are considered elsewhere [ 64 ].

Genetic Factors Related to MS-pathogenesis

Considering genetics, the aggregation of MS cases within families is well-documented. Thus, the MS-risk for a proband (who has a relative with MS) increases in rough proportion to the genetic similarity between themselves and their affected relative [ 5 , 6 , [41] , [42] , [43] , [44] , [45] , [46] , [47] , [48] , [49] , [50] , [51] , [52] , [53] , 82 , 83 ]. For example, compared to the risk in general population, MS-risk is increased ∼10-fold in nieces and nephews, ∼30-fold in non-twin siblings, and ∼250-fold in MZ - co-twins of an MS proband [ 5 , 6 , 43 , 83 ]. Moreover, there are 233 genetic traits, which have now been identified as being MS-associated. Nevertheless, as is also the case for many other chronic diseases, the genetics of MS are complex [ 84 ]. The strongest association is with the HLA Class-II haplotype, DRB1∗15:01∼DQB1∗06:02 located at ( 6p21.3 ) within the major histocompatibility complex ( MHC ) on the short arm of chromosome 6 [ 5 , 6 , [84] , [85] , [86] , [87] , [88] , [89] ]. The odds ratio ( OR ) for disease-association with this haplotype is (∼3) in heterozygotes and of (∼6) in homozygotes [ 5 , 6 , [84] , [85] , [86] , [87] , [88] , [89] ] – see also
Table 5 , Table 6 . By comparison, the other MS-associations are substantially weaker [ 84 ] – with a median OR of (1.158) and an interquartile range of (1.080–1.414). Furthermore, the DRB1∗15:01∼DQB1∗06:02 is highly “ selected ” in the sense that it accounts for 12–13 ​ % of all DRB1∼DQB1 haplotypes among European decedents [ 58 , 84 , 89 ] – see also
Table 5 , Table 6 . Indeed, this is the most frequently occurring DRB1∼DQB1 haplotype among individuals within this ethnic group [ 58 , 84 , 89 ]. Notably, in both “ European-American ” and “ African-American ” populations and in men and women , considered separately, the MS- associations with the different haplotypes in the HLA region are very similar among the two groups and the two sexes ( Table 5 ). In the African-American population, each of these haplotypes seems to be of “ European ” origin – i.e., haplotypes that seem to have been admixed ( en bloc ) into the collective genome of the African-American population [ 90 ]. Nevertheless, it is striking that both groups and both sexes have such similar HLA disease-associations ( Table 5 ). Importantly, unlike Class I molecules (which are expressed on the surface of all nucleated cells in the body and platelets but not on red blood cells), Class II molecules are expressed only on professional antigen presenting cells (e.g., dendritic cells, macrophages, thymic epithelial cells, and B cells). Moreover, also unlike Class I M H C alleles (which present processed endogenous cytosolic peptides to cytotoxic T cells), Class II molecules present processed peptides derived from extracellular proteins to helper T cells and are critical for the initiation of immune responses to invading pathogens [ 86 ]. Therefore, the strong association of MS with Class II M H C alleles underscores the closeness of the relationship between the genetic factors and the inflammatory processes involved in MS pathogenesis ( Table 5 , Table 6 ). Table 5 MS associations for HLA- haplotypes among men and women of European and African-American heritage. Table 5 CEH Name a CEHs b A∼C ​∼ ​B ​∼ ​DRB1∼DQB1∼SNP OR c (Women) p OR c (Men) p c1 (E) 01:01∼07:01∼08:01∼03:01∼02:01∼a6 3.5 < E-7 1.8 0.02 c2 (E) d 03:01∼07:02∼07:02∼15:01∼06:02∼a1 2.7 < E-78 2.5 < E-36 c3 (E) d 02:01∼07:02∼07:02∼15:01∼06:02∼a1 1.9 < E-17 1.9 < E-9 c5 (E) 02:01∼05:01∼44:02∼04:01∼03:01∼a3 0.5 < E-9 0.5 < E-6 c1 (AA) e 01:01∼07:01∼08:01∼03:01∼02:01∼a6 1.9 0.02 1.9 ns c2 (AA) e 03:01∼07:02∼07:02∼15:01∼06:02∼a1 3.0 0.002 4.5 0.004 Origen HLA haplotypes A∼C ​∼ ​B or: DRB1∼DQB1 OR c (Women) p OR c (Men) p E DRB1∗15:01∼DQB1∗06:02 – 1 copy d 3.1 < E-182 2.7 < E-82 E DRB1∗15:01∼DQB1∗06:02 – 2 copies d 6.5 < E-110 6.2 < E-70 E DRB1∗03:01∼ DQB1∗02:01 – 1 copy d 1.1 <0.05 1.2 <0.01 E DRB1∗03:01∼ DQB1∗02:01 – 2 copies d 2.9 < E-21 2.2 < E-7 E A∗02:01∼C∗05:01∼B∗44:02 – 1 copy 0.6 < E-7 0.6 < E-3 AA DRB1∗15:01∼DQB1∗06:02 – 1 copy e 1.8 < E-3 3.1 < E-3 AA DRB1∗03:01∼ DQB1∗02:01 – 1 copy e 1.5 0.005 1.4 ns AA A∗02:01∼C∗05:01∼B∗44:02 – 1 copy e 0.5 ns 0.4 ns a Arbitrary names for the most common MS-associated conserved extended HLA haplotypes ( CEHs ) in the “ European ” Welcome Trust Case Control Consortium ( WTCCC ) database [ 58 , [88] , [89] , [90] , 94 ]. Origen: E= European; AA= African American [ 89 ]. b CEHs include the HLA alleles ( A, C, B , DRB1 & DQB1 ) together haplotypes of 11 single nucleotide polymorphisms ( SNPs ) – a1, a3, & a6 – which span the HLA Class II chromosomal region [ 88 ]. The HLA Haplotypes include HLA Class I alleles ( A, C & B ) and HLA Class II ( DRB1 & DQB1 ) alleles considered separately [ 58 , [88] , [89] , [90] , 93 ]. c Odds ratio ( OR ) for MS in men and women having 2 copies of the ( c1 ) CEH (only homozygotes were MS- associated) or any number of copies of the ( c2 ), ( c3 ), or ( c5 ) CEHs (both homozygotes and heterozygotes were MS- associated). The p-values (if less than 0.001) are expressed in scientific notation as powers of 10 (E). d The difference in OR between the ( c2 ) and ( c3 ) CEHs was significant for both women ( p=0.0001 ) and men ( p=0.03 ). The difference in OR between possessing 1 or 2 copies of the DRB1∗15:01∼DQB1∗06:02 01 haplotype was significant for both women (p ​< ​ 10 −13 ) and men (p ​< ​ 10 −10 ) . Similarly, the difference in OR between possessing 1 or 2 copies of the DRB1∗03:01∼DQB1∗02:01 haplotype was significant for both women (p ​< ​ 10 −14 ) and men (p ​< ​ 0.001) – see Ref. [ 90 ]. e The same CEHs in the UCSF African American ( AA ) population [ 58 ]. Because of the much smaller numbers of cases and controls (especially among men), only single copies of the CEHs ( c1 ) ad ( c2 ) are compared between the AA women and men . Each of these CEHs represent the Caucasian haplotypes admixed into the African genome [ 58 ]. HLA Haplotypes , reported only for having 1 copy, had too few observations to report the findings for having 2 copies. Table 6 Association of MS with the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype in different regions. Table 6 Controls a DRB1∗15:01∼DQB1∗06:02 n ( %) Cases a DRB1∗15:01∼DQB1∗06:02 n ( %) AOR b Australia c [ 11 ] 390 (27.6 ​ %) 806 (52.4 ​ %) 3.0 New Zealand c [ 94 ] 185 (27.6 ​ %) 192 (58.0 ​ %) 3.6 European (WTCCC) [ 58 , 84 , 89 , 90 , 93 ] 18,682 (23.5 ​ %) 9646 (49.8 ​ %) 3.2 European (Gragert and colleagues) [ 91 ] 81,806 (23.9 ​ %) – – United States (WTCCC) [ 58 , 84 , 89 , 90 , 93 ] 18,682 (23.5 ​ %) 1614 (51.2 ​ %) 3.4 United States d (EPIC) [ 87 ] 434 (17.7 ​ %) 482 (42.9 ​ %) 3.5 Canada c [ 95 ] 104 (23.0 ​ %) 53 (46.9 ​ %) 3.0 Saudi Arabia f [ 96 ] 20 (6.3 ​ %) 45 (16.8 ​ %) 2.9 Middle East ​+ ​North Africa f [ 97 ] 1256 (20.7 ​ %) 1040 (35.7 ​ %) 2.1 Japan f [ 98 ] 394 (16.3 ​ %) 528 (33.6 ​ %) 2.5 Native Africans e [ 91 ] 1488 (1.8 ​ %) – – a n = number of subjects included for each calculation. Percentages are adjusted based upon what data is actually reported in the papers. For example, based on both the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ] and from the dataset of Gragert and colleagues [ 91 ], 92 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. Similarly, 99 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15:01 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. The “ United States ” portion and the “ European ” portion of the WTCCC cohort are listed separately. All controls in the WTCCC cohort came from Europe. b “ AOR ” ​= ​adjusted odds ratio ( OR ) to account for the relationships ( described above ) between the HLA haplotypes of DRB1∗15 and DRB1∗15:01 and the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. These adjustments are referred to as “European” . c These studies only reported the status of the HLA-DRB1∗15 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above) .{NB: These populations are basically “transplanted Europeans” and the use of “European” adjustments, therefore, are likely to be appropriate} [ 58 , 84 , 89 , 90 , 93 ]. d These patients represent the “United States” portion of a cohort [ 87 ], referred to as EPIC . e In Africa, there are too few MS cases [ 33 , 92 ] and the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype is too rare [ 91 ] so that the MS-association with this haplotype has never been reported in an African cohort. f These studies only reported the status of the HLA-DRB1∗15:01 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above ). {NB: These studies were done in populations that are not merely “transplanted Europeans” and, therefore, these “European” adjustments may be either too much or too little} [ [96] , [97] , [98] ].

Table 5 MS associations for HLA- haplotypes among men and women of European and African-American heritage. Table 5 CEH Name a CEHs b A∼C ​∼ ​B ​∼ ​DRB1∼DQB1∼SNP OR c (Women) p OR c (Men) p c1 (E) 01:01∼07:01∼08:01∼03:01∼02:01∼a6 3.5 < E-7 1.8 0.02 c2 (E) d 03:01∼07:02∼07:02∼15:01∼06:02∼a1 2.7 < E-78 2.5 < E-36 c3 (E) d 02:01∼07:02∼07:02∼15:01∼06:02∼a1 1.9 < E-17 1.9 < E-9 c5 (E) 02:01∼05:01∼44:02∼04:01∼03:01∼a3 0.5 < E-9 0.5 < E-6 c1 (AA) e 01:01∼07:01∼08:01∼03:01∼02:01∼a6 1.9 0.02 1.9 ns c2 (AA) e 03:01∼07:02∼07:02∼15:01∼06:02∼a1 3.0 0.002 4.5 0.004 Origen HLA haplotypes A∼C ​∼ ​B or: DRB1∼DQB1 OR c (Women) p OR c (Men) p E DRB1∗15:01∼DQB1∗06:02 – 1 copy d 3.1 < E-182 2.7 < E-82 E DRB1∗15:01∼DQB1∗06:02 – 2 copies d 6.5 < E-110 6.2 < E-70 E DRB1∗03:01∼ DQB1∗02:01 – 1 copy d 1.1 <0.05 1.2 <0.01 E DRB1∗03:01∼ DQB1∗02:01 – 2 copies d 2.9 < E-21 2.2 < E-7 E A∗02:01∼C∗05:01∼B∗44:02 – 1 copy 0.6 < E-7 0.6 < E-3 AA DRB1∗15:01∼DQB1∗06:02 – 1 copy e 1.8 < E-3 3.1 < E-3 AA DRB1∗03:01∼ DQB1∗02:01 – 1 copy e 1.5 0.005 1.4 ns AA A∗02:01∼C∗05:01∼B∗44:02 – 1 copy e 0.5 ns 0.4 ns a Arbitrary names for the most common MS-associated conserved extended HLA haplotypes ( CEHs ) in the “ European ” Welcome Trust Case Control Consortium ( WTCCC ) database [ 58 , [88] , [89] , [90] , 94 ]. Origen: E= European; AA= African American [ 89 ]. b CEHs include the HLA alleles ( A, C, B , DRB1 & DQB1 ) together haplotypes of 11 single nucleotide polymorphisms ( SNPs ) – a1, a3, & a6 – which span the HLA Class II chromosomal region [ 88 ]. The HLA Haplotypes include HLA Class I alleles ( A, C & B ) and HLA Class II ( DRB1 & DQB1 ) alleles considered separately [ 58 , [88] , [89] , [90] , 93 ]. c Odds ratio ( OR ) for MS in men and women having 2 copies of the ( c1 ) CEH (only homozygotes were MS- associated) or any number of copies of the ( c2 ), ( c3 ), or ( c5 ) CEHs (both homozygotes and heterozygotes were MS- associated). The p-values (if less than 0.001) are expressed in scientific notation as powers of 10 (E). d The difference in OR between the ( c2 ) and ( c3 ) CEHs was significant for both women ( p=0.0001 ) and men ( p=0.03 ). The difference in OR between possessing 1 or 2 copies of the DRB1∗15:01∼DQB1∗06:02 01 haplotype was significant for both women (p ​< ​ 10 −13 ) and men (p ​< ​ 10 −10 ) . Similarly, the difference in OR between possessing 1 or 2 copies of the DRB1∗03:01∼DQB1∗02:01 haplotype was significant for both women (p ​< ​ 10 −14 ) and men (p ​< ​ 0.001) – see Ref. [ 90 ]. e The same CEHs in the UCSF African American ( AA ) population [ 58 ]. Because of the much smaller numbers of cases and controls (especially among men), only single copies of the CEHs ( c1 ) ad ( c2 ) are compared between the AA women and men . Each of these CEHs represent the Caucasian haplotypes admixed into the African genome [ 58 ]. HLA Haplotypes , reported only for having 1 copy, had too few observations to report the findings for having 2 copies.

MS associations for HLA- haplotypes among men and women of European and African-American heritage.

Arbitrary names for the most common MS-associated conserved extended HLA haplotypes ( CEHs ) in the “ European ” Welcome Trust Case Control Consortium ( WTCCC ) database [ 58 , [88] , [89] , [90] , 94 ]. Origen: E= European; AA= African American [ 89 ].

CEHs include the HLA alleles ( A, C, B , DRB1 & DQB1 ) together haplotypes of 11 single nucleotide polymorphisms ( SNPs ) – a1, a3, & a6 – which span the HLA Class II chromosomal region [ 88 ]. The HLA Haplotypes include HLA Class I alleles ( A, C & B ) and HLA Class II ( DRB1 & DQB1 ) alleles considered separately [ 58 , [88] , [89] , [90] , 93 ].

Odds ratio ( OR ) for MS in men and women having 2 copies of the ( c1 ) CEH (only homozygotes were MS- associated) or any number of copies of the ( c2 ), ( c3 ), or ( c5 ) CEHs (both homozygotes and heterozygotes were MS- associated). The p-values (if less than 0.001) are expressed in scientific notation as powers of 10 (E).

The difference in OR between the ( c2 ) and ( c3 ) CEHs was significant for both women ( p=0.0001 ) and men ( p=0.03 ). The difference in OR between possessing 1 or 2 copies of the DRB1∗15:01∼DQB1∗06:02 01 haplotype was significant for both women (p ​< ​ 10 −13 ) and men (p ​< ​ 10 −10 ) . Similarly, the difference in OR between possessing 1 or 2 copies of the DRB1∗03:01∼DQB1∗02:01 haplotype was significant for both women (p ​< ​ 10 −14 ) and men (p ​< ​ 0.001) – see Ref. [ 90 ].

The same CEHs in the UCSF African American ( AA ) population [ 58 ]. Because of the much smaller numbers of cases and controls (especially among men), only single copies of the CEHs ( c1 ) ad ( c2 ) are compared between the AA women and men . Each of these CEHs represent the Caucasian haplotypes admixed into the African genome [ 58 ]. HLA Haplotypes , reported only for having 1 copy, had too few observations to report the findings for having 2 copies.

Table 6 Association of MS with the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype in different regions. Table 6 Controls a DRB1∗15:01∼DQB1∗06:02 n ( %) Cases a DRB1∗15:01∼DQB1∗06:02 n ( %) AOR b Australia c [ 11 ] 390 (27.6 ​ %) 806 (52.4 ​ %) 3.0 New Zealand c [ 94 ] 185 (27.6 ​ %) 192 (58.0 ​ %) 3.6 European (WTCCC) [ 58 , 84 , 89 , 90 , 93 ] 18,682 (23.5 ​ %) 9646 (49.8 ​ %) 3.2 European (Gragert and colleagues) [ 91 ] 81,806 (23.9 ​ %) – – United States (WTCCC) [ 58 , 84 , 89 , 90 , 93 ] 18,682 (23.5 ​ %) 1614 (51.2 ​ %) 3.4 United States d (EPIC) [ 87 ] 434 (17.7 ​ %) 482 (42.9 ​ %) 3.5 Canada c [ 95 ] 104 (23.0 ​ %) 53 (46.9 ​ %) 3.0 Saudi Arabia f [ 96 ] 20 (6.3 ​ %) 45 (16.8 ​ %) 2.9 Middle East ​+ ​North Africa f [ 97 ] 1256 (20.7 ​ %) 1040 (35.7 ​ %) 2.1 Japan f [ 98 ] 394 (16.3 ​ %) 528 (33.6 ​ %) 2.5 Native Africans e [ 91 ] 1488 (1.8 ​ %) – – a n = number of subjects included for each calculation. Percentages are adjusted based upon what data is actually reported in the papers. For example, based on both the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ] and from the dataset of Gragert and colleagues [ 91 ], 92 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. Similarly, 99 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15:01 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. The “ United States ” portion and the “ European ” portion of the WTCCC cohort are listed separately. All controls in the WTCCC cohort came from Europe. b “ AOR ” ​= ​adjusted odds ratio ( OR ) to account for the relationships ( described above ) between the HLA haplotypes of DRB1∗15 and DRB1∗15:01 and the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. These adjustments are referred to as “European” . c These studies only reported the status of the HLA-DRB1∗15 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above) .{NB: These populations are basically “transplanted Europeans” and the use of “European” adjustments, therefore, are likely to be appropriate} [ 58 , 84 , 89 , 90 , 93 ]. d These patients represent the “United States” portion of a cohort [ 87 ], referred to as EPIC . e In Africa, there are too few MS cases [ 33 , 92 ] and the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype is too rare [ 91 ] so that the MS-association with this haplotype has never been reported in an African cohort. f These studies only reported the status of the HLA-DRB1∗15:01 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above ). {NB: These studies were done in populations that are not merely “transplanted Europeans” and, therefore, these “European” adjustments may be either too much or too little} [ [96] , [97] , [98] ].

Association of MS with the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype in different regions.

n = number of subjects included for each calculation. Percentages are adjusted based upon what data is actually reported in the papers. For example, based on both the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ] and from the dataset of Gragert and colleagues [ 91 ], 92 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. Similarly, 99 ​ % of those individuals from “ Europe ” who carry the HLA-DRB1∗15:01 haplotype, also carry the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. The “ United States ” portion and the “ European ” portion of the WTCCC cohort are listed separately. All controls in the WTCCC cohort came from Europe.

“ AOR ” ​= ​adjusted odds ratio ( OR ) to account for the relationships ( described above ) between the HLA haplotypes of DRB1∗15 and DRB1∗15:01 and the full HLA-DRB1∗15:01∼DQB1∗06:02 haplotype. These adjustments are referred to as “European” .

These studies only reported the status of the HLA-DRB1∗15 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above) .{NB: These populations are basically “transplanted Europeans” and the use of “European” adjustments, therefore, are likely to be appropriate} [ 58 , 84 , 89 , 90 , 93 ].

These patients represent the “United States” portion of a cohort [ 87 ], referred to as EPIC .

In Africa, there are too few MS cases [ 33 , 92 ] and the HLA-DRB1∗15:01∼DQB1∗06:02 haplotype is too rare [ 91 ] so that the MS-association with this haplotype has never been reported in an African cohort.

These studies only reported the status of the HLA-DRB1∗15:01 haplotype. The percentages – but not the numbers ( n ) – have been adjusted to reflect this fact ( see above ). {NB: These studies were done in populations that are not merely “transplanted Europeans” and, therefore, these “European” adjustments may be either too much or too little} [ [96] , [97] , [98] ].

In addition, this genetic complexity is further underscored by the fact that, with the exception of MZ- twins, every individual in the population possesses a unique combination of these 233 MS-associated genetic traits [ 58 ]. Finally, considering the MZ- twin studies reported from around the world, and using the maximum possible estimates for the MZ- twin concordance rates from these studies ( Table 3 ), the range for likelihoods that an MZ -twin will develop MS, given that their co-twin is known to be a member of the ( M S ) subset, is only (0.11–0.46) – see
Table 3 . Consequently, the degree to which genetics contributes to determining the ultimate MS disease-expression must be relatively minor compared to other determinants.

Conserved extended haplotypes in the MHC

Notably, DRB1∼DQB1 haplotypes are only a component of much larger haplotypes within the MHC , as suggested by the well-documented long-range linkage disequilibrium between the many HLA alleles [ 84 , [87] , [88] , [89] ]. Indeed, the M H C region consists of a relatively small collection of highly conserved extended haplotypes ( C E H s ) , which stretch (at least) across the HLA class I and class II genes ( A , C , B , D R B 1 , a n d D Q B 1 ) – a distance, which spans nearly 3 ​mb of DNA [ 88 ] – see also
Table 5 , Table 7 . Moreover, it is the disease-associations of these CEHs that combine to yield the observed associations with the DRB1∼DQB1 portions of these haplotypes. Notably, however, the disease-association for some of these CEHs (which include the same DRB1∼DQB1 haplotype), are different [ 89 ] – e.g., CEHs c2 & c3; see
Table 7 . Also, in the large database compiled by Gragert and colleagues from populations in multiple regions world-wide [ 91 ], the haplotypic structure for these 5 allele haplotypes (in each of these populations) consists of a very small number of very high frequency CEHs and a very limited total number of unique C E H s , compared to conditions where these loci were sorting independently – see
Fig. 6 & Table 7 . Even though the actual C E H composition varies widely between the different populations [ 58 ] – presumably based, in part, upon different environmental pressures in different global regions – the overall CEH structure is very similar across all of the different populations ( see
Fig. 6 & Table 7 ). Table 7 CEH frequencies in different world populations. Table 7 Population Top 2 CEHs A∼C ​∼ ​B ​∼ ​DRB1∼DQB1 Frequency Top 2 Frequency Top 10 African a 30:01∼17:01∼42:01∼03:02∼04:02 2.2 ​ % 6.6 ​ % 36:01∼04:01∼53:01∼11:01∼06:02 Mexican a 01:01∼07:01∼08:01∼03:01∼02:01 3.3 ​ % 10.2 ​ % 29:02∼16:01∼44:03∼07:01∼02:01 South Asian Indian a 33:03∼07:01∼44:03∼07:01∼02:01 5.0 ​ % 13.1 ​ % 01:01∼06:02∼57:01∼07:01∼03:03 Native north American a 01:01∼07:01∼08:01∼03:01∼02:01 7.0 ​ % 15.0 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 European a 01:01∼07:01∼08:01∼03:01∼02:01 9.0 ​ % 17.8 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 European b 01:01∼07:01∼08:01∼03:01∼02:01 9.6 ​ % 20.7 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 Aleut a 01:01∼07:01∼08:01∼03:01∼02:01 6.8 ​ % 18.4 ​ % 02:06∼03:04∼40:02∼14:02∼03:01 Chinese a 33:03∼03:02∼58:01∼03:01∼02:01 8.6 ​ % 19.4 ​ % 02:07∼01:02∼46:01∼09:01∼03:03 Japanese a 24:02∼12:02∼52:01∼15:02∼06:01 11.8 ​ % 24.4 ​ % 33:03∼14:03∼44:03∼13:02∼06:04 Korean a 33:03∼14:03∼44:03∼13:02∼06:04 6.2 ​ % 20.6 ​ % 33:03∼03:02∼58:01∼13:02∼06:09 Southeast Asian a 33:03∼07:01∼44:03∼07:01∼02:01 4.2 ​ % 11.6 ​ % 33:03∼03:02∼58:01∼03:01∼02:01 Vietnamese a 11:01∼08:01∼15:02∼12:02∼03:01 10.4 ​ % 27.8 ​ % 29:01∼15:05∼07:05∼10:01∼05:01 Hawaiian a 24:07∼04:01∼35:05∼12:02∼03:01 7.2 ​ % 18.1 ​ % 02:01∼03:04∼40:01∼09:01∼03:03 Filipino a 24:07∼04:01∼35:05∼12:02∼03:01 7.7 ​ % 22.4 ​ % 24:02∼07:02∼38:02∼15:02∼05:02 Middle eastern a 01:01∼07:01∼08:01∼03:01∼02:01 3.3 ​ % 9.7 ​ % 33:01∼08:02∼14:02∼01:02∼05:01 Native south American a 01:01∼07:01∼08:01∼03:01∼02:01 6.8 ​ % 14.7 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 a Data from: Gragert et al. [ 91 ]. b Data from controls of the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ]. Fig. 6 The frequecy-distrubutions of 5-allele conserved extended haplotypes ( CEHs ) A ∼ C ∼ B ∼ D R B 1 ∼ D Q B 1 – from different regions around the world. The data is from the study of Gragert and colleagues [ 91 ]. For simplicity of its presentation, the populations displayed in the Figure have been restricted to African, Mexican, South Asian Indian, Native North American, European, Chinese, Aleut, and Japanese. Nevertheless, these displayed C E H frequency distributions are fully representative of every world region reported by Gragert and colleagues [ 87 ], including those not presented in the Figure – i.e., Korean, Southeast Asian, Vietnamese, Hawaiian, Filipino, Middle Eastern, and Native South American – see Table 7 . This graph plots the cumulative number of unique C E H s in each population (beginning with the highest haplotype-frequency C E H ) against the percentage of the total number of C E H s for each of these populations. In each population, the majority of the total haplotypes are accounted for by only a very small number of very high-frequency unique C E H s . Nevertheless, the composition of these very high-frequency C E H s varies markedly between populations (see Table 7 ). Moreover, C E H -diversity also varies between populations. The most diversity is found among Africans, where the top 10 C E H s account of 6.5 ​ % of the total number of C E H s present compared to Japan where the top 10 C E H s account of 24.4 ​ % of the total number of C E H s present (see Table 7 ). Notably, these haplotypes span ∼ 3 m b of DNA and involve both Class I and Class II MHC alleles. Nevertheless, there are 451 loci within the extended M H C , of which 252 are expressed [ 86 ]. Almost, certainly, some (or many) of these expressed genes will also be linked to these C E H s so that the “nominal” identity of any 5-allele CEH cannot be assumed to be an “actual” identity of the extended haplotype. Fig. 6

Table 7 CEH frequencies in different world populations. Table 7 Population Top 2 CEHs A∼C ​∼ ​B ​∼ ​DRB1∼DQB1 Frequency Top 2 Frequency Top 10 African a 30:01∼17:01∼42:01∼03:02∼04:02 2.2 ​ % 6.6 ​ % 36:01∼04:01∼53:01∼11:01∼06:02 Mexican a 01:01∼07:01∼08:01∼03:01∼02:01 3.3 ​ % 10.2 ​ % 29:02∼16:01∼44:03∼07:01∼02:01 South Asian Indian a 33:03∼07:01∼44:03∼07:01∼02:01 5.0 ​ % 13.1 ​ % 01:01∼06:02∼57:01∼07:01∼03:03 Native north American a 01:01∼07:01∼08:01∼03:01∼02:01 7.0 ​ % 15.0 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 European a 01:01∼07:01∼08:01∼03:01∼02:01 9.0 ​ % 17.8 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 European b 01:01∼07:01∼08:01∼03:01∼02:01 9.6 ​ % 20.7 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 Aleut a 01:01∼07:01∼08:01∼03:01∼02:01 6.8 ​ % 18.4 ​ % 02:06∼03:04∼40:02∼14:02∼03:01 Chinese a 33:03∼03:02∼58:01∼03:01∼02:01 8.6 ​ % 19.4 ​ % 02:07∼01:02∼46:01∼09:01∼03:03 Japanese a 24:02∼12:02∼52:01∼15:02∼06:01 11.8 ​ % 24.4 ​ % 33:03∼14:03∼44:03∼13:02∼06:04 Korean a 33:03∼14:03∼44:03∼13:02∼06:04 6.2 ​ % 20.6 ​ % 33:03∼03:02∼58:01∼13:02∼06:09 Southeast Asian a 33:03∼07:01∼44:03∼07:01∼02:01 4.2 ​ % 11.6 ​ % 33:03∼03:02∼58:01∼03:01∼02:01 Vietnamese a 11:01∼08:01∼15:02∼12:02∼03:01 10.4 ​ % 27.8 ​ % 29:01∼15:05∼07:05∼10:01∼05:01 Hawaiian a 24:07∼04:01∼35:05∼12:02∼03:01 7.2 ​ % 18.1 ​ % 02:01∼03:04∼40:01∼09:01∼03:03 Filipino a 24:07∼04:01∼35:05∼12:02∼03:01 7.7 ​ % 22.4 ​ % 24:02∼07:02∼38:02∼15:02∼05:02 Middle eastern a 01:01∼07:01∼08:01∼03:01∼02:01 3.3 ​ % 9.7 ​ % 33:01∼08:02∼14:02∼01:02∼05:01 Native south American a 01:01∼07:01∼08:01∼03:01∼02:01 6.8 ​ % 14.7 ​ % 03:01∼07:02∼07:02∼15:01∼06:02 a Data from: Gragert et al. [ 91 ]. b Data from controls of the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ].

CEH frequencies in different world populations.

Data from: Gragert et al. [ 91 ].

Data from controls of the Welcome Trust Case Control Consortium ( WTCCC ) dataset [ 58 , 84 , 89 , 90 , 93 ].

The frequecy-distrubutions of 5-allele conserved extended haplotypes ( CEHs ) A ∼ C ∼ B ∼ D R B 1 ∼ D Q B 1 – from different regions around the world. The data is from the study of Gragert and colleagues [ 91 ]. For simplicity of its presentation, the populations displayed in the Figure have been restricted to African, Mexican, South Asian Indian, Native North American, European, Chinese, Aleut, and Japanese. Nevertheless, these displayed C E H frequency distributions are fully representative of every world region reported by Gragert and colleagues [ 87 ], including those not presented in the Figure – i.e., Korean, Southeast Asian, Vietnamese, Hawaiian, Filipino, Middle Eastern, and Native South American – see Table 7 . This graph plots the cumulative number of unique C E H s in each population (beginning with the highest haplotype-frequency C E H ) against the percentage of the total number of C E H s for each of these populations. In each population, the majority of the total haplotypes are accounted for by only a very small number of very high-frequency unique C E H s . Nevertheless, the composition of these very high-frequency C E H s varies markedly between populations (see Table 7 ). Moreover, C E H -diversity also varies between populations. The most diversity is found among Africans, where the top 10 C E H s account of 6.5 ​ % of the total number of C E H s present compared to Japan where the top 10 C E H s account of 24.4 ​ % of the total number of C E H s present (see Table 7 ). Notably, these haplotypes span ∼ 3 m b of DNA and involve both Class I and Class II MHC alleles. Nevertheless, there are 451 loci within the extended M H C , of which 252 are expressed [ 86 ]. Almost, certainly, some (or many) of these expressed genes will also be linked to these C E H s so that the “nominal” identity of any 5-allele CEH cannot be assumed to be an “actual” identity of the extended haplotype.

As an example of this CEH variability between populations, the shorter DRB1∗15:01∼DQB1∗06:02 haplotype is present in native African populations, as are the longer CEHs , which are both disease-associated and which include specific combinations of all 5 of these MHC alleles [ 58 ] – see also
Table 5 , Table 7 . Nevertheless, presumably, other MHC alleles (that lie between and outside of these Class I and Class II alleles) are also linked to these “shared” 5 allele CEHs . Therefore, the African-derived, 5-allele, haplotypes may not be the same as their European counterparts, despite their nominal “5-allele identity”. Moreover, these African-derived haplotypes occur only very infrequently in the native African population compared to “ European ” populations [ 58 , 90 ] and, also, MS is extremely rare in among native Africans [ 92 ]. Consequently, the MS-association of these African-derived haplotypes (even for “nominally” identical 5-allele CEHs ) cannot be assumed and has never been tested (e.g., Table 5 , Table 6 ).

HLA associations in different populations

The MS-association of the DRB1∗15:01∼DQB1∗06:02 haplotype for different populations around the world are presented in Table 6 . In the published literature, many studies report only the HLA-DRB1∗15 or the DRB1∗15:01 status rather than the complete DRB1∗15:01∼DQB1∗06:02 haplotype. Therefore, to compare the associations in different regions, some adjustment needs to be made to reflect the full haplotype. For example, in the “ European ” populations of the both the study of Gragert and coworkers [ 91 ] and that of the Welcome Trust Case Control Consortium ( WTCCC ) [ 58 , 84 , 89 , 90 , 93 ] – see
Table 6 – 92 % of DRB1∗15 haplotypes and 99 % of DRB1∗15:01 haplotypes carry the full DRB1∗15:01∼DQB1∗06:02 haplotype. In Table 6 , this adjustment has been made for each population using the findings from these two studies. Nevertheless, it is possible (or even likely) that this adjustment is inappropriate for “ non-European ” populations ( see legend of Table 6 ). Nevertheless, the MS-association of the DRB1∗15:01∼DQB1∗06:02 haplotype is remarkably consistent across all populations reported ( Table 6 ). This is particularly evident among populations of “ European ” decent, where the DRB1∗15:01∼DQB1∗06:02 carrier-status has been essentially unchanged over the hundreds of years that these populations have been separated from each other (although, admittedly, a very short time in evolutionary terms). Thus, both the proportion of DRB1∗15:01∼DQB1∗06:02 carriers
( 17.7 − 27.6 % ) and the MS-association of this haplotype ( O R = 3.0 − 3.6 ) are about the same among the “ European ” populations of Europe, Canada, Australia, New Zealand, and the United States [ 11 , [90] , [91] , [92] , [93] , [94] , [95] , [96] , [97] , [98] ] – see also Table 6 ). This suggests that (as expected for any evolutionary change) the genetic underpinnings of MS have been relatively stable over the last several centuries and, thus, that genetics cannot account the observed changes in MS-epidemiology, which have occurred within the past 50 years ( see section on Epidemiology of MS; above; see also Table 4 ). Rather, these epidemiological changes are almost certainly due to changes in the environmental conditions of human beings around the world.

The Timing of Environmental Events Related to MS-pathogenesis

Based upon the population-based data from the CCPGSMS study ( Table 4 ), the likelihood that an MZ -twin, in Canada, will develop MS, given that their co-twin is known to be a member of the ( M S ) subset is 25.3 % . Consequently, factors other than genetics must be critically involved in disease pathogenesis. These “other factors” include “ truly random ” mechanisms ( see
section on Genetic and Environmental Susceptibility to MS; above ) and, indeed, these random events seem likely to be principally responsible for the observed difference in disease-expression between susceptible women and susceptible men [ 64 ]. Nevertheless, in addition to these genetic and random mechanisms, it is clear that environmental events (including possible interactions with genetic factors) must also contribute critically to disease-pathogenesis in MS.

When considering the question of which environmental events are involved, it is noteworthy that the micro-environmental conditions, to which an individual is exposed – i.e., the environments shared among families or small groups – seem to make little contribution to the overall MS-risk. For example, studies of MS-risk in brothers and sisters of different birth order, in conjugal couples, in siblings and half-siblings raised together or apart, and in adopted individuals, have generally found that these micro-environmental influences don't alter the overall risk of developing MS [ 50 , 52 , 53 , 56 , 99 , 100 ]. Moreover, with the exception of very rare (and potentially unreliable) reports of “ MS epidemics ” (e.g., in the Faroe Islands), excessive numbers of MS cases do not occur in small, localized regions [ 5 ]. In these circumstances it seems likely that the relevant sets of environmental experiences act at a population level. Moreover, if one or more sets of these environmental events (in addition to genetic and random mechanisms) is necessary for MS to develop in an individual, then it is only natural to enquire as to how many such events there are and whether these events need to occur at any particular time or in any particular order. Several published epidemiological findings bear on these issues.

Environmental events surrounding birth

A so-called “maternal effect” in MS has been reported in MS [ 100 ]. Evidence for such a “maternal effect” in MS derives from three independent observations. First, half-siblings (i.e., siblings who share one, but not both, biological parents), who are concordant for MS, are twice as likely to share the mother as they are to share the father [ 50 , 100 ]. This suggests that MS susceptibility is passed from the mother to the child by something other than nuclear genes. An environmental exposure, which occurs either during the intrauterine period or soon thereafter, is one possibility. After birth, however, once the child has become independent of their mother, an environmental event would not be expected to produce such a maternal effect. This observation could, potentially, also be explained by mitochondrial inheritance, by genetic-imprinting causing selective expression of certain maternal genes, or by other epigenetic mechanisms [ 101 ]. Nevertheless, when the familial genetic risk is provided to the child by either the mother or the father (but, regardless, where the child's intrauterine and early post-natal environments are provided by the same biological mother), studies have failed to demonstrate evidence for the excessive maternal transmission expected if mitochondrial genes, genetic imprinting, or epigenetic factors were the basis of the “maternal effect” in MS [ 100 ].

Second, the MS concordance-rate for full-siblings is generally [ 43 , 46 , 82 , 83 ], although not invariably [ 44 , 45 ], found to be less than that for dizygotic ( DZ )-twins – see
Table 4 . Such a disparity cannot be attributed to mitochondrial inheritance, genetic imprinting, or other epigenetic factors because these factors should be the same for both siblings and fraternal twins sharing the same biological parents. Rather, if correct, this discrepancy is likely due to environmental circumstances during, or in close proximity to, the shared intrauterine period.

Third, a month-of-birth effect for MS-risk has been reported in studies from Canada, northern Europe, and Australia [ 55 , 102 , 103 ]. For example, a study combining cohorts of patients from the northern hemisphere (Canada, Denmark, and Sweden), reported that the MS-risk varies throughout the year in a circa annum pattern based on the month-of-birth of the individual [ 55 ]. Thus, the observed MS-risk was greatest for individuals born in May and least for individuals born in November [ 55 ]. This observation seemed to be confirmed by a second study of RRMS patients in Canada [ 102 ] and, finally, for 67 Canadian MS patients, born in the southern hemisphere, this month-of-birth effect seemed to be reversed [ 55 ]. Importantly, also, this hypothesis was tested in a population-based study from Australia, which reported that MS-risk has a similar circa annum periodicity although, in this region, the risk was greatest for individuals born in November/December and was least for individuals born in May/June [ 103 ].

There has been some controversy regarding the possibility that this observed month-of-birth effect might be artifactual based on seasonal and latitudinal birth patterns, which can deviate significantly from the observations in control groups, which are often based on the unweighted averaging of nationally-acquired data and which, potentially, could represent a different region of the country [ [104] , [105] , [106] , [107] , [108] ]. The counter-argument to this concern is that the original study confirmed their results in sibling pairs, born at the same latitude [ 55 ], and that a similar month-of-birth effect was found in another study, even after adjusting for both the year and the place of birth [ 107 ]. Indeed, a recent meta-analysis of 22 studies concluded that this effect was not artifactual [ 108 ]. As a result, these month-of-birth observations probably can't be explained away as a simple artifact [ 107 , 108 ] and, if so, this effect provides unequivocal evidence for an early environmental event, which is time-locked to the birth, and which is involved in MS pathogenesis. As the time-interval increases between the birth of the individual and the responsible environmental event, the coupling between birth and the event will necessarily become less precise and, as a consequence, any month-of-birth signal will be increasingly ill-defined. The fact that this signal is so clear indicates that the environmental event must occur very near to the birth itself and its circa annum periodicity indicates that it is coupled to the solar cycle [ 55 , 102 , 103 , 107 , 108 ]. Notably, mothers of May babies in the northern hemisphere (or November babies in the southern hemisphere) spend much of their pregnancy during the winter months (with less sun exposure) compared to mothers who are pregnant over the summer months. Thus, one possibility is that variations in intrauterine vitamin D levels are responsible for the effect [ 55 ]. Nevertheless, there are other possibilities. For example, seasonal infections (which also have circa annum periodicity) might produce a similar effect although, because infections of the fetus both are uncommon during the intra-uterine period and are often accompanied by obvious congenital defects, any such association would likely be a secondary (non-infectious) event following an infection in the mother.

Environmental events during childhood or adolescence

A second environmental event in MS pathogenesis is suggested by observations in migrants who move from one geographical region to another; each region having a different MS-risk [ 5 , 6 , [109] , [110] , [111] , [112] , [113] , [114] , [115] ]. For example, when an individual relocates (prior to adolescence) from a region of high prevalence to a region of low prevalence area (or vice versa ), their MS-risk becomes similar to that of the region to which they moved. By contrast, when the same relocation is undertaken following adolescence, the MS-risk is similar to that of the region from which they moved. Also, children of immigrants from low prevalence regions who are born in high prevalence areas have an MS-risk similar to their birth country rather than their country of ethnic origin [ 5 , 6 , [109] , [110] , [111] , [112] , [113] , [114] , [115] ]. If correct, these observations imply that there is an environmental event (or events) that occur between the birth and adolescence, which is (or are) related to MS pathogenesis.

Environmental events during adulthood

Third, as noted previously, in general, the initial clinical symptoms in MS are substantially delayed, often by decades, following both the birth and adolescence ( Table 1 ). Consequently, it seems likely that the timing of symptom-onset in MS must be determined subsequent environmental events.

The Nature of the Environmental Events Related to MS Pathogenesis

Over the years, many potential environmental events have been postulated as being linked to MS pathogenesis. These events include trauma, stress, tobacco, vaccinations, obesity, small pox, typhoid, Epstein-Barr virus ( EBV ), human herpes virus ( HHV ) 6, chicken pox, Chlamydia, a variety of other infections, mercury poisoning from dental amalgam tooth-fillings or other sources, low-sunlight, vitamin deficiencies, occupational hazards, cosmic-rays, dietary habits, living with domesticated animals, ciprofloxacin treatment, and other toxic exposures [ 5 , 6 ]. Of these possibilities, however, EBV infection, vitamin D deficiency, and tobacco have the most evidence in their favor and have attracted the greatest current interest as playing some role in MS pathogenesis. Therefore, the focus here will be on these possibilities. Nevertheless, some of the other potential factors have strong proponents and, although it is true that no single factor has been proven or excluded conclusively, it is also true that many (or most) of these other proposed associations lack credible scientific evidence, biological plausibility, or both [ 5 , 6 ].

Epstein-barr virus ( EBV )

EBV is a linear double-stranded DNA virus of the Herpesviridae family. It is an extremely common human infection with more than 90 % of the population ultimately becoming infected with EBV by the time they reach adulthood [ [116] , [117] , [118] , [119] , [120] , [121] , [122] , [123] , [124] , [125] , [126] , [127] , [128] , [129] ] – see also
Table 8 . After birth, as maternal EBV antibody protection wanes, the infant becomes vulnerable to infection. In many places around the world, where EBV infection occurs during early childhood, the infection is either asymptomatic or is indistinguishable symptomatically from other common childhood illnesses. However, when the infection is delayed until adolescence or young adulthood (e.g., by possibly better hygiene or other factors in the developed countries of Europe, North America, and elsewhere), between 35 and 50 % of infected individuals will develop symptoms of infectious mononucleosis (also called glandular fever). Notably, regardless of when it occurs, an EBV infection seems to target specifically the B cells and the epithelial cells of the oropharynx. Table 8 Prevalence of EBV antibodies in the sera of MS-patients and controls. Table 8 Study EBV+ MS Cases ( %) EBV+ Controls ( %) p value a Sumaya, 1980 c [ 116 ] 155/157 (98.7 ​ %) 76/81 (93.8 ​ %) 0.05 Bray, 1983 c [ 117 ] 309/313 (98.7 ​ %) 363/406 (89.4 ​ %) E-4 Larson, 1984 c [ 118 ] 93/93 (100 ​ %) 78/93 (83.9 ​ %) E-4 Sumaya, 1985 b [ 119 ] 104/104 (100 ​ %) 23/26 (88.5 ​ %) 7 x E-3 Shirodaria, 1987 f [ 120 ] 26/26 (100 ​ %) 24/26 (92.3 ​ %) ns Munch, 1998 d [ 121 ] 137/138 (99.3 ​ %) 124/138 (89.9 ​ %) 4 x E-4 Myhr, 1998 b [ 122 ] 144/144 (100 ​ %) 162/170 (95.3 ​ %) 8 x E-3 Wagner, 2000 d [ 123 ] 107/107 (100 ​ %) 153/163 (93.9 ​ %) E-2 Ascherio, 2001 e [ 124 ] 143/144 (99.3 ​ %) 269/287 (93.7 ​ %) 8 x E-3 Sundström, 2004 [ 125 ] 234/234 (100 ​ %) 693/702 (98.7 ​ %) ns Haahr, 2004 d [ 126 ] 153/153 (100 ​ %) 50/53 (94.3 ​ %) 0.05 Ponsonby, 2005 f [ 127 ] 136/136 (100 ​ %) 252/261 (96.6 ​ %) 0.05 Kreft, 2017 g [ 128 ] 665/668 (99.6 ​ %) 131/147 (89.1 ​ %) E-7 Bjornevik, 2023 h [ 129 ] 800/801 (99.9 ​ %) 1459/1566 (93.2 ​ %) E-13 Totals 3206/3218 (99.6 ​ %) 3857/4119 (93.6 ​ %) p ​< ​ 10 − 16 a (E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant. b Study measured antibodies against the Epstein-Barr nuclear antigens (EBNA), the viral capsid antigen (VCA), and the early antigens (EA). c Study measured antibodies against VCA. d Study measured antibodies against EBNA and EA. e Study measured antibodies against EBNA and VCA. f Study measured antibodies against EBNA and VCA. g Study measured antibodies against EBNA-1, EA-D, and VCA. h Study measured antibodies against EBNA-1,2,3,4, & 6 and VCA.

Table 8 Prevalence of EBV antibodies in the sera of MS-patients and controls. Table 8 Study EBV+ MS Cases ( %) EBV+ Controls ( %) p value a Sumaya, 1980 c [ 116 ] 155/157 (98.7 ​ %) 76/81 (93.8 ​ %) 0.05 Bray, 1983 c [ 117 ] 309/313 (98.7 ​ %) 363/406 (89.4 ​ %) E-4 Larson, 1984 c [ 118 ] 93/93 (100 ​ %) 78/93 (83.9 ​ %) E-4 Sumaya, 1985 b [ 119 ] 104/104 (100 ​ %) 23/26 (88.5 ​ %) 7 x E-3 Shirodaria, 1987 f [ 120 ] 26/26 (100 ​ %) 24/26 (92.3 ​ %) ns Munch, 1998 d [ 121 ] 137/138 (99.3 ​ %) 124/138 (89.9 ​ %) 4 x E-4 Myhr, 1998 b [ 122 ] 144/144 (100 ​ %) 162/170 (95.3 ​ %) 8 x E-3 Wagner, 2000 d [ 123 ] 107/107 (100 ​ %) 153/163 (93.9 ​ %) E-2 Ascherio, 2001 e [ 124 ] 143/144 (99.3 ​ %) 269/287 (93.7 ​ %) 8 x E-3 Sundström, 2004 [ 125 ] 234/234 (100 ​ %) 693/702 (98.7 ​ %) ns Haahr, 2004 d [ 126 ] 153/153 (100 ​ %) 50/53 (94.3 ​ %) 0.05 Ponsonby, 2005 f [ 127 ] 136/136 (100 ​ %) 252/261 (96.6 ​ %) 0.05 Kreft, 2017 g [ 128 ] 665/668 (99.6 ​ %) 131/147 (89.1 ​ %) E-7 Bjornevik, 2023 h [ 129 ] 800/801 (99.9 ​ %) 1459/1566 (93.2 ​ %) E-13 Totals 3206/3218 (99.6 ​ %) 3857/4119 (93.6 ​ %) p ​< ​ 10 − 16 a (E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant. b Study measured antibodies against the Epstein-Barr nuclear antigens (EBNA), the viral capsid antigen (VCA), and the early antigens (EA). c Study measured antibodies against VCA. d Study measured antibodies against EBNA and EA. e Study measured antibodies against EBNA and VCA. f Study measured antibodies against EBNA and VCA. g Study measured antibodies against EBNA-1, EA-D, and VCA. h Study measured antibodies against EBNA-1,2,3,4, & 6 and VCA.

Prevalence of EBV antibodies in the sera of MS-patients and controls.

(E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant.

Study measured antibodies against the Epstein-Barr nuclear antigens (EBNA), the viral capsid antigen (VCA), and the early antigens (EA).

Study measured antibodies against VCA.

Study measured antibodies against EBNA and EA.

Study measured antibodies against EBNA and VCA.

Study measured antibodies against EBNA and VCA.

Study measured antibodies against EBNA-1, EA-D, and VCA.

Study measured antibodies against EBNA-1,2,3,4, & 6 and VCA.

Once a cell becomes infected, EBV enters a lytic cycle where fresh viral particles are released, and new cells are infected. However, the virus can also enter a latent phase where the viral genome circularizes and persists in the cell as an episome. Ultimately, likely due to immune responses by the host, latent infections predominate over lytic infections. Latently infected B cells then proliferate and are directed to specific sites, such as the bone marrow, where the virus persists indefinitely. Episodically, the virus reactivates, thereby causing further cell lysis and producing new viral particles.

During the period of late incubation and during the early acute illness, antibodies ( Abs ) are found in the serum, which are directed against antigens associated with viral replication – e.g., the diffuse and restricted early antigens ( EA ) and the viral capsid antigen ( VCA ) [ 130 ]. The EA - Abs are of the IgG class. They are present early in the acute infection but generally drop to undetectable levels within 3 − 6 months. Consequently, the presence of EA - Abs in the serum is usually taken to be a sign of early active infection although these EA - Abs titers may persist for years in 20 − 30 % of patients. Moreover, EA - Abs can also be found in patients with chronic, but reactivated, infections or who have secondary complications of EBV (e.g., nasopharyngeal carcinoma or Burkitt's lymphoma). Initially, the VCA - Abs are of the IgM class although this acute response only lasts about 1 − 2 months. After this, the VCA - Abs switch to the IgG class, which persist for the remainder of the person's life. Latently infected B cells express the EBV nuclear antigens ( EBNA -1 to EBNA -6). Antibodies to these antigens ( EBNA - Abs ) usually appear 3 − 6 weeks after the initial infection and, like VCA - Abs , they persist for the remainder of the person's life. Consequently, the presence of EBNA - Abs in the serum is usually taken to be a sign of chronic (latent) infection. EBV -involvement has also been suggested in the pathogenesis of some malignancies (e.g., Burkitt's lymphoma, nasopharyngeal carcinoma, EBV -positive Hodgkin lymphoma) and of certain autoimmune diseases such as Sjogren's syndrome, rheumatoid arthritis, and systemic lupus erythematosus [ 131 ].

Notably, for many years there has been a consistent linkage found between an EBV infection and the development of MS (both childhood-onset and adult-onset), especially when the infection produces the syndrome of infectious mononucleosis [ 124 , 125 , 127 , [132] , [133] , [134] , [135] , [136] , [137] , [138] ]. Despite this, MS develops in less than ( 1 / 900 )
EBV- infected individuals and in only a small fraction of mononucleosis cases. Nevertheless, the evidence that EBV plays a significant role in MS pathogenesis seems overwhelming [ [116] , [117] , [118] , [119] , [120] , [121] , [122] , [123] , [124] , [125] , [126] , [127] , [128] , [129] , [130] , [131] , [132] , [133] , [134] , [135] , [136] , [137] , [138] ] – see also
Table 8 . Indeed, despite the fact that an EBV infection occurs in over 90 % of the non-MS population, a considerably greater proportion of adult MS patients ( > 99 % ) have evidence of a prior EBV infection ( Table 8 ). This difference between MS patients and non-MS controls is highly significant ( p < 10 − 16 ) . Moreover, it is very possible that these rare EBV- negative patients represent false-negative tests, which would mean that 100 % of MS-patients have evidence of a prior EBV exposure. Such a circumstance (i.e., false-negative tests) could be the result either from only measuring the antibody-response to some of the EBV antigens or from inherent errors when any fixed antibody-titer “cut-off” level is used to distinguish between the EBV- positive and EBV- negative status of an individual. Importantly, if these rare EBV- negative patients do, in fact, reflect false negative test results, then one would be forced to conclude that an EBV infection is a necessary (but not a sufficient) condition in every causal pathway currently leading to MS.

When the data has been available, the EBV infection has been documented to occur years before onset of the clinical symptoms of MS [ 124 , 127 , 129 , 130 , 132 , [134] , [135] , [136] , [137] ]. Also, this timing of events is supported by the fact that the Abs to the various EBV antigens are of the IgG class and not the IgM class and by the fact that over 98 % of patients have EBNA Abs [ 128 ]. Notably, in this context, the term “prior” means prior to the symptomatic-onset of MS. As discussed in the section on the Epidemiology of MS ( see above ), however, in most, if not all, MS cases, the actual disease-onset occurs several years prior to symptom-onset so that the order of events cannot be established reliably from the available data. Nevertheless, such an extremely strong association between EBV infection and MS is difficult to ignore.

Moreover, the near 100 % prevalence of EBV-Abs in MS ( Table 8 ) can't be explained by a high false-positive test rate due to a general hyper-immune state in MS patients because their antibody responses to other common pathogens (e.g., mumps, measles, chicken pox, cytomegalovirus, herpes simplex, etc.) are not similarly increased [ 129 , 139 , 140 ] – see also
Table 9 . Finally, because symptom-onset in MS generally occurs long after initial EBV exposure or infectious mononucleosis [ [116] , [117] , [118] , [119] , [120] , [121] , [122] , [123] , [124] , [125] , [126] , [127] , [128] , [129] , [130] , [131] , [132] , [133] , [134] , [135] , [136] , [137] , [138] ] – see also
Table 1 – it is quite plausible for this particular pathogen to be in the causal pathway leading to MS. Table 9 Prevalence of antibodies to different viruses in the sera of MS-patients and controls. a Table 9 Virus Virus+ MS Cases ( %) Virus+ Controls ( %) p-value b OR EBV c 3206/3218 (99.6 ​ %) 3857/4119 (93.6 ​ %) E-16 17.0 HSV2 d 40/138 (29.0 ​ %) 29/205 (14.1 ​ %) E-2 2.5 Measles d 661/749 (88.3 ​ %) 1309/1609 (81.4 ​ %) ns 1.72 Chlamydia e 85/129 (65.9 ​ %) 160/258 (62.0 ​ %) ns 1.18 HSV1 d 109/138 (79.0 ​ %) 159/205 (77.6 ​ %) ns 1.09 Mumps d 224/386 (58.0 ​ %) 305/524 (58.2 ​ %) ns 0.99 HSV d 507/666 (76.1 ​ %) 997/1287 (77.4 ​ %) ns 0.93 VZV d 516/637 (81.0 ​ %) 1004/1179 (85.2 ​ %) ns 0.74 CMV d 312/693 (45.0 ​ %) 447/854 (52.3 ​ %) ns 0.74 Rubella d 226/235 (96.2 ​ %) 299/307 (97.4 ​ %) ns 0.67 a Epstein-Barr Virus (EBV); Varicella-Zoster Virus (VZV); Cytomegalovirus (CMV); Herpes Simplex Virus Type 1 or oral herpes (HSV1); Herpes Simplex Virus Type 2 or genital herpes (HSV2); Herpes Simplex Virus not otherwise specified (HSV) OR = Odds Ratio. b (E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant. c Data from Table 8 . d Data from Aschiero, 2007 [ 139 ]. e Chlamydia pneumoniae; Data from Munger, 2004 [ 140 ].

Table 9 Prevalence of antibodies to different viruses in the sera of MS-patients and controls. a Table 9 Virus Virus+ MS Cases ( %) Virus+ Controls ( %) p-value b OR EBV c 3206/3218 (99.6 ​ %) 3857/4119 (93.6 ​ %) E-16 17.0 HSV2 d 40/138 (29.0 ​ %) 29/205 (14.1 ​ %) E-2 2.5 Measles d 661/749 (88.3 ​ %) 1309/1609 (81.4 ​ %) ns 1.72 Chlamydia e 85/129 (65.9 ​ %) 160/258 (62.0 ​ %) ns 1.18 HSV1 d 109/138 (79.0 ​ %) 159/205 (77.6 ​ %) ns 1.09 Mumps d 224/386 (58.0 ​ %) 305/524 (58.2 ​ %) ns 0.99 HSV d 507/666 (76.1 ​ %) 997/1287 (77.4 ​ %) ns 0.93 VZV d 516/637 (81.0 ​ %) 1004/1179 (85.2 ​ %) ns 0.74 CMV d 312/693 (45.0 ​ %) 447/854 (52.3 ​ %) ns 0.74 Rubella d 226/235 (96.2 ​ %) 299/307 (97.4 ​ %) ns 0.67 a Epstein-Barr Virus (EBV); Varicella-Zoster Virus (VZV); Cytomegalovirus (CMV); Herpes Simplex Virus Type 1 or oral herpes (HSV1); Herpes Simplex Virus Type 2 or genital herpes (HSV2); Herpes Simplex Virus not otherwise specified (HSV) OR = Odds Ratio. b (E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant. c Data from Table 8 . d Data from Aschiero, 2007 [ 139 ]. e Chlamydia pneumoniae; Data from Munger, 2004 [ 140 ].

Prevalence of antibodies to different viruses in the sera of MS-patients and controls. a

Epstein-Barr Virus (EBV); Varicella-Zoster Virus (VZV); Cytomegalovirus (CMV); Herpes Simplex Virus Type 1 or oral herpes (HSV1); Herpes Simplex Virus Type 2 or genital herpes (HSV2); Herpes Simplex Virus not otherwise specified (HSV) OR = Odds Ratio.

(E) represents the exponent – i.e., E-4 ​= ​10 −4 ; ns ​= ​not significant.

Data from Table 8 .

Data from Aschiero, 2007 [ 139 ].

Chlamydia pneumoniae; Data from Munger, 2004 [ 140 ].

Because EBV infection doesn't occur during either the in utero or the early postnatal periods, EBV seems to be precluded, at least directly, from being the environmental event responsible for the “maternal effect” in MS. Rather, an EBV infection, like the second environmental event, seems to act during adolescence or early adulthood – i.e., when, typically, both late EBV infection and mononucleosis occur. Nevertheless, although the timing of the relevant EBV infection and the second environmental event seem to coincide, the migration data is difficult to rationalize in this manner. Thus, individuals from low-prevalence regions for MS, presumably, would have acquired EBV already so that moving after childhood from a low-prevalence area to a high prevalence area shouldn't matter. Similarly, why children moving from a high-prevalence region to a low-prevalence region (and who, presumably, have not yet acquired EBV ) should adopt a low prevalence when they move is also unclear. Regardless of these difficulties, however, it is known that there are at least two different strains EBV [ [141] , [142] , [143] ] with different geographic distributions so that, perhaps, a better understanding of EBV biology and its complexities might help to rationalize the migration data.

In sum, on the basis of strong, high-quality, and very consistent evidence ( Table 7 , Table 8 ), the conclusion seems inescapable that EBV infection plays a critical role in the pathogenesis of most (perhaps all) MS cases – likely as a component of a very important causal pathway leading to MS.

Vitamin D deficiency

In mammals, the in vivo production of active vitamin D requires a two-step conversion of 7-dehydro-cholesterol into vitamin D 3 [ [144] , [145] , [146] , [147] ]. In the first step, pre-vitamin D 3 is sy nthesized from 7-dehydro-cholesterol by a pathway involving its exposure (in the skin) to ultraviolet-B ( UVB ) radiation – i.e., radiation with a wavelength of ( 280 − 320 ) nanometers. Second, a thermally-induced rearrangement of the internal double bond structure of the pre-vitamin D 3 molecule takes place, which forms vitamin D 3 . Notably, however, vitamin D 3 , is not the active form of the vitamin. Vitamin D 3 needs to be hydroxylated, first to 25(OH) D 3 by 25-hydroxylase (primarily in the liver) and, then, further hydroxylated by 1-α hydroxylase in the tissues (in the kidneys and many other tissues, including lymph nodes, skin, neurons and glia) to form 1,25(OH) 2 D 3 – or active vitamin D [ 152 , 153 ].

The dietary intake of vitamin D 3 can bypass the UVB -dependent portion of this pathway and can permit the maintenance of normal 25(OH) D 3 serum levels in the absence of adequate UVB radiation. Nevertheless, only a few natural dietary sources (e.g., oily fishes and reindeer) have sufficient vitamin D 3 [ 148 ]. These animals derive it or its precursors from their diet (phytoplanctonic algae in the sea in the case of fish and lichen on the tundra in the case of reindeer). Interestingly, the Inuit or Eskimos (whose diet consists of large amounts of oily fish) and the Sami or Lapps (who consume reindeer meat on a regular basis) both have a notably low MS-risk [ [148] , [149] , [150] , [151] , [152] ] and, in both cases, the population's principal source of vitamin D 3 comes from their respective diets [ 148 ]. By contrast, other human populations require sufficient skin-exposure to UVB radiation or supplementation with pharmaceutical vitamin D 3 in order to maintain adequate serum levels throughout the year. In this manner, either the natural dietary consumption or the supplementation of vitamin D 3 merely provides the body with the building blocks necessary to make active vitamin D whenever and wherever it is needed.

Moreover, the grouping of vitamin D together with the other vitamins is likely inappropriate. Thus, in general, vitamins are organic compounds, that are involved in specific chemical reactions (usually as co-factors), and that humans (over the course of their evolutionary history) have lost the ability to synthesize. Presumably, we have lost this ability because these compounds are widely available in the foods we eat and, consequently, most diets provide an adequate supply. By contrast, humans have not lost the ability to synthesize truly essential organic compounds (e.g., cholesterol, which is vitally important to the function of every cell), even though they may be dietarily abundant. Clearly, vitamin D falls into this latter category. We can synthesize it, it is not generally available in our diet, and it does not participate in specific biochemical reactions as a co-factor. Rather, vitamin D acts by binding to its receptor ( VDR ), which is located on the surface of cells throughout the body, including activated T cells, B cells, and macrophages [ 146 , 154 , 155 ]. This receptor-ligand complex is then internalized to form a heterodimer, together with the retinoid X receptor – a complex, which is then translocated to the cell nucleus where it binds to a short sequence of DNA bases (the vitamin D response element or VDRE ) located in the promoter region of many genes throughout the genome [ 146 , [148] , [149] , [150] , [151] , [152] , [153] , [154] , [155] , [156] , [157] ]. Interestingly, the VDRE has been identified in the promoter region adjacent to the HLA-DRB1∗15:01 allele and has been shown to regulate the expression of this gene [ 158 ]. This observation is notable because, as discussed earlier, this HLA region (and particularly this DRB1 allele) has been linked to MS disease-expression for decades [ 5 , [83] , [84] , [85] , [86] , [87] , [88] , [89] ] – see also
Table 5 , Table 6 .

Vitamin D is a pleiotropic hormone, well-known to play a critical role in calcium homeostasis (e.g., participating in bone growth, bone cell differentiation, and mineralization of the collagen matrix). Nevertheless, the physiological effects of vitamin D are considerably far more wide-ranging. Vitamin D plays an important role in other cellular processes, including a variety of immune system functions such as modulating cell proliferation, cell differentiation, the maturation of dendritic cells, and in the modulation of antigen-specific immune responses in vivo [ 146 , [154] , [155] , [156] , [157] ]. Active vitamin D is synthesized by human decidual cells at the maternal-fetal interface during gestation (especially in early pregnancy), suggesting a possible role for vitamin D in regulating the fetal development of both acquired and innate immune responses [ 159 ]. Vitamin D also has important anti-inflammatory, anti-fibrotic, anti-neoplastic, and anti-oxidant actions [ 145 , [153] , [154] , [155] , [156] , [157] ]. In this context, therefore, it is not surprising that vitamin D deficiency has been associated with other autoimmune diseases such as such as rheumatoid arthritis, insulin dependent diabetes mellitus, and inflammatory bowel disease [ 154 , 160 ]. Moreover, because, in general, women compared to men, are more likely to develop autoimmune diseases, it is possible that there are sex-specific differences in the physiological responses to active vitamin D exposure [ 6 , 161 , 162 ].

Notably, as latitude increases (both north and south of the equator), the angle at which the sunlight strikes the Earth's atmosphere becomes increasingly oblique. This causes the light to travel a longer path through the atmosphere before reaching the earth's surface, thereby reducing the amount of UVB radiation available for vitamin D synthesis in temperate and more extreme latitudes. In equatorial Africa (where Homo sapiens evolved) adequate UVB radiation for vitamin D 3 synthesis is available throughout the year [ [163] , [164] , [165] ] – see also
Fig. 3 . By contrast, once human populations left Africa and began to inhabit the temperate (and even more extreme) regions of the earth, UVB exposure became inadequate for vitamin D 3 synthesis during some (or many) months of the year ( Fig. 3 ). For example, it has been estimated that the level of UVB radiation at the US-Canadian border during most months of the year is insufficient to produce an adequate amount of vitamin D 3 [ [163] , [164] , [165] ] – see
Fig. 3 . Indeed, it has been suggested that the lighter skin tones in people of many ethnic origins, who live in temperate (or more extreme) latitudes, reflect a convergent evolutionary adaptation to the need for more efficient utilization of the limited amount of UVB radiation available to these people compared to the much greater amount of UVB radiation available to people living equatorial regions [ [163] , [164] , [165] ]. However, despite the fact that African immigrants to higher latitude countries have lower serum-levels of vitamin D compared to those with lighter skin-tones, there is also evidence that these lower serum levels do not adversely impact bone health [ 165 ]. Nevertheless, it is unknown whether these lower serum levels might impact any of the other physiologic effects of Vitamin D. Regardless, as noted by many authors and consistent with a potential role of vitamin D deficiency in MS pathogenesis, the world-wide distribution of reduced UVB availability is strikingly similar to the world-wide distribution of MS-prevalence ( Fig. 3 ).

Most of the evidence presented above for a potential role of vitamin D deficiency in MS pathogenesis is circumstantial although there have been a few studies, which have more directly explored the potential relationship between vitamin D deficiency and MS. For example, in 2003, Van der Mei and co-workers reported the results of interviewing 136 Tasmanian MS cases retrospectively together with 272 sex-matched and birth-year-matched controls who came from the same population [ 166 ]. Participants filled out validated questionnaires and were interviewed about the amount of time they spent in the sun for weekends and holidays during each summer of their life, about measures they took to protect against sun-exposure, and about their use of vitamin D supplementation. This study found that the greater the sun-exposure that an individual had during childhood and early adolescence (6–15 years), the lower the risk of developing multiple sclerosis ( p < 0.01 ) . This study did not attempt to measure sun-exposure during the age-range of 0–5 years.

In 2004, Munger and colleagues reported the results from the two large studies in cohorts of nurses, who were followed prospectively [ 167 ]. The first began in 1976 and consisted of 121,700 nurses (aged 30–55 years) and the second was begun in 1989 and consisted of 116,671 nurses (aged 25–42 years). On several occasions during the course of each study, the nurses in both cohorts completed a validated semi-quantitative questionnaire regarding food-frequency. The authors estimated the total vitamin D consumption (both supplemental and dietary) and each person was ranked by their quintile of vitamin D intake. For each quintile of these cohorts, the number of MS-cases per person-year were calculated. The authors of these studies reported that the total vitamin D consumption at baseline was inversely associated with MS risk. Thus, comparing the highest with the lowest quintile of vitamin D consumption, the pooled, age-adjusted, relative risk ( RR ) of MS was: R R = 0.67 ( p = 0.03 ) . Analyzing the intake of dietary and supplemental vitamin D separately, the authors reported that the effect was principally related to the intake of supplemental vitamin D.

Investigators of another study, analyzed, retrospectively, 315 people with MS who had previously served in the armed services [ 168 ]. Cases were identified using the Physical Evaluation Boards’ database, which includes records from all active-duty personnel. Participants were required to have, available for analysis, one or more stored serum samples from the time when they entered the military. Controls (two for each MS case) were selected randomly from the same population and were matched on the bases of race/ethnicity, sex, age, dates of sample collection, and their service-branch of the military. Among whites (but not among blacks), a regression analysis demonstrated that for every 20 n g / m l increase in 25-(OH)-vitamin D serum level, there was a 41 % decrease in MS risk ( p = 0.04 ) and, comparing the top quintile of 25-(OH)-vitamin D level to the bottom quintile, demonstrated a reduced MS risk such that: R R = 0.38 ( p = 0.006 ) .

Because 25-(OH)-vitamin D serum levels have a circa annum periodicity in temperate regions [ 48 ], similar to the “maternal” factor in MS pathogenesis ( see section on the Timing of Environmental Events; above ), vitamin D deficiency would seem to be a natural candidate for this factor. Moreover, not only is the 25-(OH)-vitamin D level coupled to the solar cycle, the world-wide distribution of MS mirrors that of reduced UVB radiation ( Fig. 3 ), vitamin D is involved in maturation of the immune system [ 146 , [154] , [155] , [156] , [157] ], other autoimmune disorders are associated with its deficiency [ 154 , 160 ], and populations with high dietary consumption of vitamin D 3 (e.g., the Inuit and Sami people) have a very low MS-prevalence [ [148] , [149] , [150] ].

However, despite the attractiveness of equating vitamin D deficiency to the maternal factor in MS, the reality is that, if it acts at all, vitamin D deficiency could act at any time. In fact, the direct data ( cited above ), which suggests a role for vitamin D deficiency in MS pathogenesis, actually supports the view that this effect occurs during childhood, adolescence, or early adulthood [ [166] , [167] , [168] ]. The potential design of any clinical trial to test the efficacy of vitamin D 3 supplementation, will be greatly impacted by such uncertainties. For example, if a clinical trial is designed to be enriched for individuals at high-risk of developing MS – e.g., by studying first-degree relatives of MS probands – then the study will fail if the critical time of vitamin D deficiency occurs prior to symptom-onset in the proband , when most of their brothers and sisters, likely, also, have already passed their therapeutic window of opportunity for supplementation. A similar concern is applicable to any clinical trial that targets an environmental event, which is thought to act early.

To this point, a 2018 Cochrane review considered 12 published randomized-trials of vitamin D supplementation in patients with MS (a position paper re-affirmed in 2023) and the authors of this report concluded that: “to date, very low-quality evidence suggests no benefit of vitamin D for patient-important outcomes among people with MS. Vitamin D appears to have no effect on recurrence of relapse, worsening of disability measured by the Extended Disability Status Scale (EDSS), and MRI lesions. Effects on health-related quality of life and fatigue are unclear” [ 169 ]. More recently, two other studies have explored the efficacy if vitamin D supplementation in preventing the conversion of CIS to MS [ 170 , 171 ]. One study showed no benefit to supplementation after 48 weeks [ 164 ], whereas the other reported a significant 34 ​ % reduction in such conversion ( p = 0.004 ) after 2 years [ 171 ]. All of these randomized trials, however, do not address the important question of whether vitamin D deficiency was in the causal pathway leading to MS. Rather, these studies are concerned solely with the use of vitamin D supplementation in the context of treating existing MS and MS attacks.

Because a randomized clinical trial (RCT) is impractical in this circumstance ( see section on Therapies Targeting Environmental Events; below ), a preferable design to test the hypothesis that vitamin D deficiency is in the causal pathway leading to MS would be to conduct a non-randomized prospective observational trial. Newly pregnant women, together with their physician (and based on their own beliefs and preferences), could decide on what degree of supplementation, if any, was desirable. Vitamin D levels could then be monitored – in the mother during pregnancy, and in the children after birth – and, subsequently, these levels could then be correlated with the children's health outcomes decades later. Such a study, potentially, could provide a definitive answer and it would not be that expensive (levels are already measured routinely in many practice settings) and the monitoring would be clinically useful to physicians when making individual pre-natal and post-natal treatment recommendations. Moreover, in addition to permitting the vitamin D hypothesis to be tested, such a prospective study (especially if it included other, more wide-ranging, patient health-data) could provide important public health information to the society. There are, of course, difficulties with any such project. First, the data acquisition and data entry would require an ongoing and sustained effort. Second, it will take a very long-term follow-up and the answers to many questions could take decades. Third, the study would need to be population-based (preferably national) and it would also require a national health-care system that permitted researchers to track easily both children and adults over time and that allowed researchers to have consent-free access to an anonymized medical records database. Unfortunately, in the United States, any such public health trial is not currently feasible, given both the patch-work nature of our insurance coverage and the general anxiety that our society has regarding privacy issues. Perhaps other countries (e.g., Denmark) could be more successful [ 38 ].

Tobacco

A case/control study, published approximately 60 years ago, suggested a possible relationship between MS and tobacco smoking [ 6 ]. Subsequently, numerous case/control studies have been reported, which have generally supported this association [ [172] , [173] , [174] , [175] ]. Indeed, two recent large meta-analyses (including 26 and 31 carefully screened studies) concluded that smokers are statistically more likely to develop MS cases compared to non-smokers ( p o o l e d O R ≈ 1.5 ) and a similar association was found even when considering only smoking behavior prior to symptom-onset and the exposure to passive smoke [ 172 , 173 ]. In aggregate, therefore, the association between smoking behavior and MS seems reasonably well-established. The pathophysiological basis for this relationship, however, is less clear. Nevertheless, smoking has also been linked to the occurrence of other autoimmune diseases such as Crohn's disease, rheumatoid arthritis, systemic lupus erythematosus, and thyroid disease [ 176 ]. Moreover, nicotine or other chemicals in tobacco smoke (or perhaps smoke and pollutants, more generally) could exert an effect through a molecular cascade that begins with damage to the lung epithelium, leading to decreasing mucus clearance, and immune suppression, which, in turn, could lead to an autoimmune cascade impacting the vascular system and resulting more severe atherosclerosis, myocardial infarctions, strokes and, possibly, other conditions [ 175 , 176 ]. Clearly, such a pathophysiological explanation for this relationship is highly speculative, lacks specificity, and lacks any direct evidence other than fact that there are well-documented adverse health outcomes from smoking in other contexts. In aggregate, however, regardless of its basis, the association between tobacco smoke and MS, at least, seems genuine. However, there are four reasons that, currently, the causality of this relationship should be considered only tentative. First, it is possible that smoking represents one possible “trigger” for an MS attack (including the 1st attack), in which case it would be associated with symptom-onset but not with disease-onset ( see section on Epidemiology of MS; above; and sub-section on Sex-based Therapies; below ). Second, case/control studies only provide evidence of an association and, generally, these studies can't establish causation ( Fig. 7 ) – see the above section on EBV regarding a possible exception to this rule . In fact, of all Bradford Hill's criteria for establishing a causative relationship [ 81 ], only the criterion of temporality seems sacrosanct (i.e., the effect must follow the cause). And, yet, even this criterion only applies to the pathway being considered ( Fig. 7 ). If alternative pathways exist, even this temporal criteria can be violated [ 81 ]. Indeed, all of Bradford Hill's other criteria can be either violated or confirmed, both with and without causation being implied [ 81 ]. Thus, it is not clear from the available evidences that smoking is even in the (or a) causal pathway leading to MS ( Fig. 7 ). Third, in Canada, it is well-established that the prevalence of MS has increased by more than 32 % over the past half-century [ 64 ], whereas the prevalence of smoking has declined by a similar amount over the same time-frame [ 177 ]. Naturally, this observation could be due to competing effects – i.e., some environmental exposures may be decreasing while others, perhaps more important to MS-pathogenesis than smoking, are increasing. Nevertheless, these population-based epidemiological observations in Canada seem counter-intuitive. And fourth, the majority of MS patients don't smoke, and the vast majority of smokers don't develop MS [ 175 , 177 ]. Consequently, even if smoking were somehow causative, there must be other, more important, causal pathways leading to MS that don't include this particular environmental exposure ( Fig. 7 ). Fig. 7 An example illustrating the difficulties that arise when trying to make inferences regarding causation from association studies. Illustrated are different causal pathways that might explain a possible association between epilepsy and MS. Approximately 0.5 − 1 % of the general population suffers from epilepsy whereas, among MS patients, several studies suggest that this percentage may be as high as ( 2 t o 4 − f o l d ) greater [ 5 ]. However, regardless of whether this association is genuine, the schemes and connections presented in this Figure are hypothetical and speculative. In Pathway ( 1 ) , it is envisioned that there is some antecedent event ( Event A ), which is common to both condition – e.g., a shred genetic pre-disposition, a traumatic event, a toxic exposure, etc. This creates the association. In Pathway ( 2 ) , both arms produce the same “side effect ( Event I ), which, potentially, can create an association. Otherwise, these causal pathways involve completely unrelated and non-overlapping events. Prevention of MS will not prevent epilepsy and vice versa . These relationships are considered “ spurious associations ”. Also, even if these pathways do create the association, there may be other, more important, pathways leading to MS and epilepsy, which don't include any common events (e.g., the dashed part of Pathway 5), and that weaken the observed association. In Pathway ( 5 ) , the antecedent event ( Event M ) only predisposes to epilepsy; MS follows automatically (if other necessary intervening Events also occur). However, because epilepsy generally occurs late in the MS disease course (i.e., long after the symptom-onset), the solid arrow pathway is excluded because it violates Bradford Hill's criteria of temporality that effect must follow cause [ 81 ]. Notably, however, most individuals with epilepsy don't have MS, so that following an Event N (e.g. a developmental anomaly, trauma, genetic factors, etc), an alternative dashed arrow path to the epilepsy must exit, which leads to epilepsy without MS occurring. As a result, if epilepsy occurs, depending upon circumstances, it can precede or follow MS. In Pathways ( 3 & 4 ) , the antecedent event ( Event K ) only predisposes to MS; epilepsy follows automatically (again, presuming that certain necessary intervening Events occur). However, these two pathways represent alterative explanations for the connection between MS and epilepsy. In MS, lesions are sometimes located in the cortex or in the white matter adjacent to the cortex [ 5 , [68] , [69] , [70] , [71] , [72] ]. In Pathway ( 3 ) some of these lesions are presumed to occasionally be epileptogenic and cause epilepsy. This is the pathway invoked by most authorities to explain the possible association [ 5 ]. However, it is well-established that MS can cause incoordination, imbalance, and falls [ 5 ]. In Pathway ( 4 ) , it is presumed that these symptoms lead to more frequent head strikes and brain injuries and, thus, more epilepsy. However, potentially, one can enter this second causal pathway to epilepsy at the brain injury stage following numerous antecedent traumatic events (e.g., motor vehicle and bike accidents, blunt trauma, serious falls, work-related injuries, blast injuries, etc), designated here as ( Event L ). In this manner, the antecedent event of MS can be bypassed entirely. Indeed, in any large case series of post-traumatic epilepsy, the likelihood of finding any MS-patients among the cohort is vanishingly small so that, undoubtedly, the most common paths to epilepsy for Pathways ( 3 ) , ( 4 ) and ( 5 ) are the alternative paths where MS and epilepsy occur separately from each other. Thus, the prevention of MS will have no impact on the occurrence of post-traumatic epilepsy and vice versa , despite MS and epilepsy both being components of one possible causal pathway. Fig. 7

An example illustrating the difficulties that arise when trying to make inferences regarding causation from association studies. Illustrated are different causal pathways that might explain a possible association between epilepsy and MS. Approximately 0.5 − 1 % of the general population suffers from epilepsy whereas, among MS patients, several studies suggest that this percentage may be as high as ( 2 t o 4 − f o l d ) greater [ 5 ]. However, regardless of whether this association is genuine, the schemes and connections presented in this Figure are hypothetical and speculative. In Pathway ( 1 ) , it is envisioned that there is some antecedent event ( Event A ), which is common to both condition – e.g., a shred genetic pre-disposition, a traumatic event, a toxic exposure, etc. This creates the association. In Pathway ( 2 ) , both arms produce the same “side effect ( Event I ), which, potentially, can create an association. Otherwise, these causal pathways involve completely unrelated and non-overlapping events. Prevention of MS will not prevent epilepsy and vice versa . These relationships are considered “ spurious associations ”. Also, even if these pathways do create the association, there may be other, more important, pathways leading to MS and epilepsy, which don't include any common events (e.g., the dashed part of Pathway 5), and that weaken the observed association. In Pathway ( 5 ) , the antecedent event ( Event M ) only predisposes to epilepsy; MS follows automatically (if other necessary intervening Events also occur). However, because epilepsy generally occurs late in the MS disease course (i.e., long after the symptom-onset), the solid arrow pathway is excluded because it violates Bradford Hill's criteria of temporality that effect must follow cause [ 81 ]. Notably, however, most individuals with epilepsy don't have MS, so that following an Event N (e.g. a developmental anomaly, trauma, genetic factors, etc), an alternative dashed arrow path to the epilepsy must exit, which leads to epilepsy without MS occurring. As a result, if epilepsy occurs, depending upon circumstances, it can precede or follow MS. In Pathways ( 3 & 4 ) , the antecedent event ( Event K ) only predisposes to MS; epilepsy follows automatically (again, presuming that certain necessary intervening Events occur). However, these two pathways represent alterative explanations for the connection between MS and epilepsy. In MS, lesions are sometimes located in the cortex or in the white matter adjacent to the cortex [ 5 , [68] , [69] , [70] , [71] , [72] ]. In Pathway ( 3 ) some of these lesions are presumed to occasionally be epileptogenic and cause epilepsy. This is the pathway invoked by most authorities to explain the possible association [ 5 ]. However, it is well-established that MS can cause incoordination, imbalance, and falls [ 5 ]. In Pathway ( 4 ) , it is presumed that these symptoms lead to more frequent head strikes and brain injuries and, thus, more epilepsy. However, potentially, one can enter this second causal pathway to epilepsy at the brain injury stage following numerous antecedent traumatic events (e.g., motor vehicle and bike accidents, blunt trauma, serious falls, work-related injuries, blast injuries, etc), designated here as ( Event L ). In this manner, the antecedent event of MS can be bypassed entirely. Indeed, in any large case series of post-traumatic epilepsy, the likelihood of finding any MS-patients among the cohort is vanishingly small so that, undoubtedly, the most common paths to epilepsy for Pathways ( 3 ) , ( 4 ) and ( 5 ) are the alternative paths where MS and epilepsy occur separately from each other. Thus, the prevention of MS will have no impact on the occurrence of post-traumatic epilepsy and vice versa , despite MS and epilepsy both being components of one possible causal pathway.

Therapeutic Implications – General Considerations

As noted earlier, the pathogenic basis for MS seems to include three distinct elements – i.e., genetic factors, environmental events, and random mechanisms [ 64 ] – perhaps with important interactions between the first two [ 64 ]. All three elements of this pathogenesis have been targeted either in actual or in proposed clinical trials. Nevertheless, although the design, execution, and evaluation of clinical trials has improved markedly in the last several decades, largely due to both the more rigorous methodological and statistical training of clinical researchers and, also, the diligent efforts of regulatory agencies, there is still a need to match the outcomes measured during a trial to its intended purpose.

For example, in general, disease-modifying therapies ( DMTs ) are intended to postpone or prevent a negative patient outcome (e.g., MS-related disability or MS-related death) decades in the future. Short-term outcomes (e.g., the attack rate, short-term changes in disability measures, or short-term changes on MRI imaging) are useful only to the extent that they serve as valid surrogates for these longer-term negative patient outcomes [ 178 ]. Yet, no patient would agree to (and no clinician would recommend) participation in any multi-decades long RCT, so that a full validation in this manner is probably not feasible. Moreover, while it is true that a significant correlation with a “true endpoint” is insufficient to establish surrogacy [ 178 ]; the opposite is also true – i.e., the absence of such a correlation indicates that the surrogate is not valid. Also, the outcomes themselves need to be valid in the sense that they actually measure what they purport to measure [ 179 ]. For example, the extended disability status scale (EDSS) is well-known to be problematic [ 180 ]. At the low end of the scale ( E D S S = 1 − 4 ) , the high inter- and intra-rater variability in scoring makes a 1-point change unreliable, whereas in the mid-range ( i . e . , E D S S = 6 ) a 1-point deterioration represents the difference between walking or not. For the purposes of determining “fixed” disability over time, negative patient outcomes are probably best objectified by defining so called “hard” endpoints such as unremitting ( E D S S ≥ 6 ) , unremitting ( E D S S ≥ 7 ) , conversion to SPMS, or death [ 181 ]. However, this approach converts the E D S S scale into a measure of only ambulatory status or, at the high-end of the scale ( i . e . , E D S S ≥ 8 ) , into a measure of very advanced disease (including loss of ambulation). Even so, such endpoints, determined at the end of an RCT (or whenever the follow-up period ends), are much preferred to short-term measures evaluated either during the RCT or thereafter. For example, “confirmed disability progression” during an RCT can't even reliably predict the patient's status at the end of the RCT [ 182 ] – see also
Table 10 . This conclusion was confirmed in the 16-year and 21-year follow-up studies of the interferon beta-1b (IFNβ-1b) clinical trial [ 181 ] where the measure of “confirmed disability progression” either was not correlated, or was the least well-correlated, with the final outcome among the disability measures considered ( Table 11 ). Thus, “confirmed disability progression” should not be considered a valid measure of long-term, fixed, disability outcome. The same conclusion applies to the measures of “no evidence of disease activity” ( NEDA ) and on-RCT MRI activity ( Table 11 ). Table 10 Relationship of “confirmed” disability progression to RCT outcome a . Table 10 EDSS Progression Duration EDSS Change ( %) PPV Confirmed Sustained 1 point 3 months 32.2 15.3 0.48 1 point 6 months 21.4 14.1 0.67 2 points 3 months 12.1 6.4 0.53 2 points 6 months 9.3 5.1 0.55 a RCT = randomized controlled tiral; EDSS = extended disability status scale; EDSS Progression = legnth of time that EDSS progression is required to be sustained in order to qualify as "confirmed progression"; Confirmed = % of population who reached the "confirmed progression" outcome during the RCT; Sustained = % of population who remained as "confirmed progression" at the enmd of the RCT; PPV = positive predictive value of the "confirmed progression" outcome. Data from Liu C and Blumhardt LD [ 182 ]. Table 11 Relationship of different “On-RCT” outcomes to a negative disability outcome 16- and 21-years after RCT-onset a . Table 11 Baseline variables 16-year outcomes b 21-year outcomes b OR p-value HR p-value  EDSS 1.87 < E–4 1.14 ns  MRI T2 lesion burden 1.03 < E–3 1.01 0.005  3rd ventricular width 1.19 3 x E–3 1.09 ns  Relapse rate prior to RCT 1.00 ns 0.94 ns On RCT variables OR p-value HR p-value  Annual relapse rate 1.70 < E–4 1.15 0.04  Actual EDSS change over RCT 1.37 E–2 1.23 0.003  ≥1 point EDSS change at year 2 1.85 ns 1.97 0.03  Confirmed 1 point EDSS change 1.00 ns 1.71 0.03  New T2 lesions on MRI 0.98 ns 1.10 0.02  Change, MRI T2 lesion burden 1.00 ns 1.01 ns  NEDA 0.51 ns 0.30 ns a RCT = randomized controlled trial; EDSS = extended disability status scale; Disability = a patient had reached a negative disability outcome (NDO). At 16-years an NDO was defined if a patient had died, or had converted to SPMS, or had reached and unremitting EDSS >= 6.0 (i.e., reached these endpoints during follow-up and sustained them until the end of the follow-up period). At 21-years an NDO was defined if the patient had died by the end of the follow-up period. OR = odds ratio; HR = hazard ratio; NEDA = no evidence of disease activity -- defined as no clinical attacks, no "confirmed" disability progression, and no new or enlarging lesions on T2-weighted MRI. b Data from: Goodin DS, et al. [ 181 ].

Table 10 Relationship of “confirmed” disability progression to RCT outcome a . Table 10 EDSS Progression Duration EDSS Change ( %) PPV Confirmed Sustained 1 point 3 months 32.2 15.3 0.48 1 point 6 months 21.4 14.1 0.67 2 points 3 months 12.1 6.4 0.53 2 points 6 months 9.3 5.1 0.55 a RCT = randomized controlled tiral; EDSS = extended disability status scale; EDSS Progression = legnth of time that EDSS progression is required to be sustained in order to qualify as "confirmed progression"; Confirmed = % of population who reached the "confirmed progression" outcome during the RCT; Sustained = % of population who remained as "confirmed progression" at the enmd of the RCT; PPV = positive predictive value of the "confirmed progression" outcome. Data from Liu C and Blumhardt LD [ 182 ].

Relationship of “confirmed” disability progression to RCT outcome a .

RCT = randomized controlled tiral; EDSS = extended disability status scale; EDSS Progression = legnth of time that EDSS progression is required to be sustained in order to qualify as "confirmed progression"; Confirmed = % of population who reached the "confirmed progression" outcome during the RCT; Sustained = % of population who remained as "confirmed progression" at the enmd of the RCT; PPV = positive predictive value of the "confirmed progression" outcome. Data from Liu C and Blumhardt LD [ 182 ].

Table 11 Relationship of different “On-RCT” outcomes to a negative disability outcome 16- and 21-years after RCT-onset a . Table 11 Baseline variables 16-year outcomes b 21-year outcomes b OR p-value HR p-value  EDSS 1.87 < E–4 1.14 ns  MRI T2 lesion burden 1.03 < E–3 1.01 0.005  3rd ventricular width 1.19 3 x E–3 1.09 ns  Relapse rate prior to RCT 1.00 ns 0.94 ns On RCT variables OR p-value HR p-value  Annual relapse rate 1.70 < E–4 1.15 0.04  Actual EDSS change over RCT 1.37 E–2 1.23 0.003  ≥1 point EDSS change at year 2 1.85 ns 1.97 0.03  Confirmed 1 point EDSS change 1.00 ns 1.71 0.03  New T2 lesions on MRI 0.98 ns 1.10 0.02  Change, MRI T2 lesion burden 1.00 ns 1.01 ns  NEDA 0.51 ns 0.30 ns a RCT = randomized controlled trial; EDSS = extended disability status scale; Disability = a patient had reached a negative disability outcome (NDO). At 16-years an NDO was defined if a patient had died, or had converted to SPMS, or had reached and unremitting EDSS >= 6.0 (i.e., reached these endpoints during follow-up and sustained them until the end of the follow-up period). At 21-years an NDO was defined if the patient had died by the end of the follow-up period. OR = odds ratio; HR = hazard ratio; NEDA = no evidence of disease activity -- defined as no clinical attacks, no "confirmed" disability progression, and no new or enlarging lesions on T2-weighted MRI. b Data from: Goodin DS, et al. [ 181 ].

Relationship of different “On-RCT” outcomes to a negative disability outcome 16- and 21-years after RCT-onset a .

RCT = randomized controlled trial; EDSS = extended disability status scale; Disability = a patient had reached a negative disability outcome (NDO). At 16-years an NDO was defined if a patient had died, or had converted to SPMS, or had reached and unremitting EDSS >= 6.0 (i.e., reached these endpoints during follow-up and sustained them until the end of the follow-up period). At 21-years an NDO was defined if the patient had died by the end of the follow-up period. OR = odds ratio; HR = hazard ratio; NEDA = no evidence of disease activity -- defined as no clinical attacks, no "confirmed" disability progression, and no new or enlarging lesions on T2-weighted MRI.

Data from: Goodin DS, et al. [ 181 ].

By contrast, the goal in using agents to treat acute MS episodes is to promote both faster more complete recovery from an MS attack and short-term outcomes are, by definition, necessary. Also, if the goal of therapy is to interrupt the causal pathway leading to MS (i.e., a primary prevention trial), then the outcome should be the prevention of disease. For this purpose, there need to be clear criteria for making a definite diagnosis and very long-term follow-up. As discussed above ( and also discussed in the subsection on Vitamin D Deficiency ), this makes an RCT infeasible and, in a country such as the United States, even a large-scale observational study is likely impractical.

Therapies Targeting Inflammation

Corticosteroids

Following the introduction of cortisone treatment for rheumatoid arthritis in 1949, corticosteroids were successfully used to treat a number of autoimmune and other inflammatory conditions [ 183 , 184 ]. Corticosteroids represent a class of natural biological chemicals also referred to as glucocorticoids. In humans, hydrocortisone (cortisol) – a cholesterol-derived steroid hormone – is the principal endogenous member of this class. It is produced in the zona fasciculata of the adrenal cortex and its release is controlled by the secretion of adrenocorticotropic hormone (ACTH) into the blood stream by the anterior pituitary gland and also by autonomic neurons, which innervate the adrenal cortex. In turn, ACTH secretion is controlled by the hypothalamus through the secretion of corticotropin-releasing hormone (CRH). CRH is produced by neuroendocrine cells within the paraventricular nucleus of the hypothalamus and, from these neurons, it is released into the capillary plexus of the hypophyseal-portal system, which supplies blood to the anterior lobe of the pituitary gland. The secretion of CRH (and, thus, the release of ACTH) is increased in response both to stress and to low circulating glucocorticoid blood-levels. Cortisol mediates the suppression of numerous immune functions, for example, through its impact on leukocyte differentiation, proliferation, function, and survival; its induction of T cell apoptosis; its promotion of myeloid-derived suppressor cell expansion; its suppression of leukocyte effector functions; its suppression of proinflammatory cytokine synthesis – e.g., interleukin (IL)-2, IL-2R, IL-4, IL-8, interferon gamma (IFNγ) and tumor necrosis factor alpha (TNFα); and its inhibition of immune cell trafficking across the BBB [ [184] , [185] , [186] ]. It also produces other notable biologic effects, including elevating blood glucose concentrations (through gluconeogenesis); increasing blood pressure; participating in the metabolism of proteins, fats, and carbohydrates; and in reducing bone formation [ [184] , [185] , [186] ]. Glucocorticoid production is also increased during late pregnancy where it promotes the formation of lung surfactant and, thereby, stimulates lung maturation [ 187 ]. Importantly, however, it is the impact of glucocorticoids on non-hematopoietic cells, which are primarily responsible for the metabolic, neuromuscular, cardiovascular, and cognitive side-effects that patients often experience while on therapy [ [184] , [185] , [186] ].

The plethora of physiologic and cellular effects that glucocorticoids have are principally mediated through the glucocorticoid receptor (GCR) – a member of the nuclear receptor superfamily, which is ubiquitously expressed in cells throughout the body [ [184] , [185] , [186] ]. Glucocorticoids can either bind to the GCR on the cell surface (membrane-bound receptor) or they can enter the cell and bind to GCRs located within the cytosol (thereby triggering release of heat-shock and other cytosolic signaling proteins). Regardless, however, these receptor–ligand complexes are translocated into the cell nucleus where they bind to specific short sequences of DNA bases (the glucocorticoid response elements), which are located in the promotor regions of many nuclear genes where they activate or inhibit the transcription of glucocorticoid-responsive genes [ 186 ].

Moreover, based on the pathological evidence of inflammation within acute MS lesions and the genetic evidence implicating inflammatory mechanisms ( see sections on the Pathology of MS and
Genetic Factors Relateded to
MS- Pathogenesis ; above ), beginning in the 1960's and early 1970s, the use of glucocorticoid treatments for MS, using both natural ACTH and synthetic glucocorticoids (e.g., prednisone, prednisolone, dexamethasone, and methylprednisolone), began to be investigated in several small, poorly designed, clinical trials with equivocal results [ 188 ]. However, in 1991 the results of the large Optic Neuritis Treatment Trial (ONTT) were published regarding the use of high-dose glucocorticoids to improve visual recovery following acute attacks of optic neuritis [ 189 ]. Although deeply flawed for most, if not all, of their secondary outcomes and conclusions [ 190 ], the ONTT did convincingly demonstrate that high-dose glucocorticoid treatment sped the recovery of visual function compared to placebo. In the ONTT, visual recovery was also more complete in the treated group compared to the control group although, after 6 months, this difference was not statistically significant. Following this publication, treatment regimens, modeled on that used by the ONTT, became a standard therapy for the treatment of acute MS attacks.

Immunosuppressant medications

Nevertheless, because of the marked toxicity of long-term glucocorticoid therapy, several other broad-spectrum immunosuppressive therapies were investigated in the 1970s and 1980s as a means of controlling inflammation over the long-term and, in general, these trials met with little success [ 188 , 191 ]. More recently, likely due to marked improvements in the design of clinical trials and the use of more robust statistical analysis methods, some broad-spectrum immunosuppressants have demonstrated convincing efficacy in the treatment of MS.

For example, teriflunomide, which inhibits the mitochondrial enzyme dihydroorotate dehydrogenase, and thereby blocks the de novo synthesis of pyrimidines, interferes with the replication of rapidly dividing cells, including activated T cells and B cells. However, teriflunomide spares the “salvage” pathway of pyrimidine synthesis, possibly, thereby mitigating some of the potential toxicity of general immune suppression. In two double-blind, placebo-controlled, RCTs in MS, teriflunomide significantly reduced the clinical relapse rate by (31–36 %) compared to placebo [ 192 , 193 ].

Also, cladribine, a purine analogue of deoxyadenosine, is incorporated into the DNA of lymphocytes during cell division, resulting in DNA stand breaks and the disruption of both DNA synthesis and DNA repair – ultimately leading to cell death by apoptosis. The selective action of cladribine on lymphocytes seems to be due to lymphocytes possessing high-levels of the enzyme deoxycytidine kinase, which phosphorylates cladribine, thereby facilitating its activation. Also, in the cladribine molecule, unlike adenosine, a chlorine replaces a hydrogen at position 2 on the purine ring. This renders the molecule partially resistant to deamination by the enzyme adenosine deaminase, which is involved in purine-degradation. Consequently, the residence-time of activated cladribine inside the lymphocytes is prolonged. In a double-blind, placebo-controlled, RCT in MS, cladribine significantly reduced the clinical relapse rate by (54–58 %) compared to placebo [ 194 ]. Moreover, in the extension (follow-up) trial after completion of the initial RCT [ 195 ], cladribine treatment for 2 years followed by 2 years of placebo-treatment seemed to produce the same clinical benefits as that provided by 4 years of cladribine treatment – suggesting that, similar to alemtuzumab ( see below ), cladribine treatment might lead to persistent alterations in immune function.

Because of their relatively selective depletion of T and B cells from the peripheral blood compartment, perhaps, both cladribine and teriflunomide might be better classified with the therapies targeting lymphocyte trafficking ( see section on Targeting Lymphocyte Trafficking; below ) rather with the broad-spectrum immunosuppressants. Regardless, however, the use of broad-spectrum immune-suppression for the treatment of MS (including these two agents) has been limited due to their potential, or actual, toxicity.

Therapies (Possibly) Targeting Down-Stream Consequences of MS Genetics

Interferons

In the late 1970s and 1980s, therapeutic interest shifted to the use of interferons (IFNs) as a means of managing MS in the long-term, beginning with a small study of IFNα, which did not demonstrate significant efficacy but was thought to suggest a trend and another small dose-finding study of IFNγ that was thought to cause immune activation and to be harmful. First described in the 1950s, the IFNs consist of a large family of secreted proteins, which participate in regulating the growth and proliferation of cells, in the defense against viral infections, and in the modulation of immune responses [ [196] , [197] , [198] , [199] , [200] ]. There are two basic types of IFN – the type I IFNs (α and β) and the type II IFNs (γ). IFNβ is a glycoprotein with 166 amino acid residues. Both IFNα and IFNβ bind to the same two-subunit cell-surface receptor and both activate the same Janus kinase/signal transducer and activator of transcription ( Jak/STAT ) signaling pathway. The activation of this signaling pathway ultimately leads to the binding of interferon-stimulated gene factor 3 ( ISGF3 ) to the interferon stimulated response element ( ISRE ). The ISRE is a short sequence of DNA bases, which are located in the promoter region of many nuclear genes [ 196 , 200 ]. Following the binding of ISGF3 to the ISRE , the activated transcription of ISRE -containing genes occurs. In the absence of such activation, however, the expression-level of these ISRE -containing genes is low. IFNβ induces the expression of many gene products including the Mx family of GTPases (e.g., MxA and MxB ), 2′-5′ oligoadenylate synthase, β2 microglobulin, tumor necrosis factor-related apoptosis-inducing ligand ( TRAIL ), and neopterin [ [196] , [197] , [198] , [199] , [200] ]. Notably, despite binding the same receptor, the downstream expression-pattern following exposure of these ISRE -containing genes to IFNα and IFNβ, is not identical [ 196 , 198 ].

By contrast, IFNγ binds to a different two-subunit receptor and activates a different (but related) Jak/STAT signaling pathway. Activation of this pathway ultimately leads to the binding of gamma activated factor ( GAF ) to a short DNA sequence (the gamma activation sequence or GAS ), which is located within the promoter region of many nuclear genes. As is the case for the Type I IFNs, this binding activates transcription of many GAS -containing genes including neopterin, proteasomal subunit, and transfer molecules [ [196] , [197] , [198] , [199] , [200] ].

There are two forms of IFNβ that have been used in the clinical management of MS patients. The first is IFNβ-1a, which is a glycosylated protein and has the full 166 amino acid sequence of native human IFNβ. Nevertheless, it is genetically engineered using a Chinese hamster ovary (CHO) cell line, so that, potentially, the glycosylation pattern is different from that of a human being. The second is IFNβ-1b, which is produced in a bacterial cell line ( E. coli ) and, thus, is non-glycosylated (because bacterial cells don't glycosylate their proteins). Consequently, the cysteine at position 17 has been substituted by a serine (replacing a sulfur atom with an oxygen atom), thereby limiting the formation of incorrect disulfide bonds and, thus, the number of improperly folded molecules. Moreover, the N-terminal methionine has also been deleted so that the final protein is only 165 amino acids residues in length. Theoretically, at least, it is possible that these chemical differences between IFNβ-1a and IFNβ-1b could have important clinical consequences. Nevertheless, in practice, any observed efficacy-difference between the IFNβ products seems to be due to the dose and dosing schedule of the products rather than to any of these subtle chemical differences between molecules [ 191 ].

The successful trial of IFNβ-1b in the treatment of RRMS led to the first ever approval of a therapy for MS by the Federal Drug Administration (FDA) in 1993. This three-year trial, begun in 1988, showed a 31 % reduction in the clinical attack rate over 3 years with the use of IFNβ – a finding that the FDA considered bolstered by the non-validated surrogate of a marked reduction in the amount of on-RCT MRI activity. Because this was the only approved therapy from 1993 to 1996, effectively, this was the only option for therapy in randomized patients for 8 years after RCT start. A follow-up study (spear-headed, encouraged, funded, and otherwise supported by the pharmaceutical manufacturer of IFNβ-1b), ascertained the death-status for 98.4 % of the randomized cases after 21-years. This long-term follow-up study demonstrated a significant (and similar) reduction in MS mortality by ∼50 % in both arms that were randomized to receive active treatment during the RCT, compared to those randomized to receive placebo [ 184 ]. This data represents the best and most reliable data on long-term DMT outcome available for any product on the market. Regrettably, however, despite the widely-held view that current competing products are more effective than IFNβ, similar studies supporting this belief in the long-term will likely never be undertaken in the future by other product manufacturers because of the immense amount of the time, effort, and money required to conduct this kind of research.

Glatiramer acetate

Glatiramer acetate (alternatively referred to as GA , Copolymer-1, or Cop-1), is a random polypeptide (average molecular mass 6.4 kD) composed of four amino acids (L-glutamic acid, L-lysine, L-alanine, and L-tyrosine) in a specific molar ratio (1.4, 3.4, 4.2, and 1.0, respectively). GA was produced by researchers who were making synthetic polypeptides of those amino acid residues, disproportionately present in myelin basic protein ( MBP ), in hope of producing synthetic antigens, capable of inducing experimental autoimmune encephalomyelitis ( EAE ) – the widely studied animal model for MS. Unexpectedly, however, these synthetic polymers were found to be protective against EAE (especially GA ) rather than being encephalitogenic [ 201 ]. These results ultimately led a phase III double-blind, placebo-controlled RCT, published in 1995, which reported that GA treatment significantly reduced the clinical attack rate by 29 % ( p = 0.007 ) compared to placebo over a two-year period although it had no significant impact on measures of disability [ 202 ].

In analyzing their data, however, the authors employed a statistical method (adjusting for already balanced covariates), which was not part of their a priori analysis plan. The FDA was aware of this and required the company to re-analyze their data according to the original plan. This adjustment made the observed reduction in attack rate statistically non-significant ( p = 0.055 ) . However, having done this, inexplicably, the FDA approved glatiramer acetate for marketing in the United States despite the fact that GA did not have a statistically significant benefit on either the clinical relapse rate or disease progression. MRIs were not included in this study.

Moreover, the mechanism of action glatiramer acetate is unknown and it seems a rather odd biological pharmaceutical. Because the synthetic polymers are random, every molecule is different from every other; every vial is unique. Nevertheless, it was suggested that its mechanism of action might relate to a number of its immunological effects such as the induction of regulatory T cells of the helper 2 (Th2)-type, thereby causing an immune deviation in CD4 + T cells from an inflammatory Th1 phenotype to an anti-inflammatory Th2 phenotype, the induction of MBP- specific suppressor T cells, inhibition of antigen presentation, displacing bound MBP , or promiscuous binding to a variety of MHC class II molecules [ [203] , [204] , [205] ]. Nevertheless, the actual mechanism of action (if any) in human MS remained opaque.

In response to the clear evidence of efficacy using IFNβ and equivocal evidence (at best) using GA to treat MS, and with GA gaining market-share, the manufacturers of two IFNβ products, independently, embarked on large head-to-head studies comparing the efficacy of GA to that of IFNβ – ironically having to purchase GA from its manufacturer on the open market to conduct the trials [ 206 , 207 ]. The results of these two trials were both striking and highly consistent [ 206 , 207 ]. There was no difference between products on measures of clinical relapse or disability although IFNβ performed better on the non-validated surrogate of on-RCT MRI activity. Together, these two trials demonstrated, convincingly, and for the first time, both that GA has clinical efficacy, and that this efficacy is approximately equivalent to that of IFNβ in the management of MS patients.

Dimethyl and monomethyl fumarate

Although not specific to MS and not related directly to MS genetics, both reactive oxygen species ( ROS ), such as H 2 O 2 , and electrophilic molecules contribute to DNA damage and to tissue injury in many circumstances [ 208 , 209 ]. Consequently, following the oxygenation of the earth's atmosphere ( ∼ 2.5 b i l l i o n y e a r s a g o ) , all organisms living on Earth at the time (i.e., single-celled organisms) had to adapt. The progenitor or progenitors of many multi-cellular organisms did this through the evolution of elaborate defense mechanisms to counter the potential damage caused by oxidative stress. A major component of this cellular defense in many animals is activation of the nuclear factor erythroid 2-related factor 2 ( Nrf2 ) antioxidant signaling pathway, which regulates the expression of the many genes involved in the detoxication and elimination of ROS and other electrophilic compounds. This regulation occurs primarily at the level of transcription and is mediated by the binding of Nrf2 to antioxidant responsive elements ( AREs ), which are located in the promoter region of many cytoprotective genes, including those that code for the two principal detoxication enzymes – glutathione S-transferase A2 ( GSTA2 ) and NADPH :quinone oxidoreductase-1 ( NQO1 ). Both dimethyl and monomethyl fumarate activate the Nrf2-ARE pathway and also inhibit the expression of adhesion molecules and cytokines, suggesting a possible role for such molecules in the treatment of disease-processes that result in tissue injury [ 208 , 209 ].

In 2012, the results of two randomized, double-blind, placebo-controlled trials using dimethyl fumarate to treat MS were published [ 210 , 211 ]. Both trials demonstrated that treatment with oral dimethyl fumarate resulted in a significant ( 42 − 52 % ) reduction in clinical attacks and marked reductions measures of on-RCT MRI activity. Also, one of these trials (but not the other) reported a benefit to treatment on the measure of 3-month confirmed EDSS progression [ 210 , 211 ].

Therapies Targeting Lymphocyte Trafficking

Monoclonal antibodies

The recent explosion in the development of monoclonal antibodies ( M A b s ) as pharmaceutical agents has allowed the precise targeting of very-specific biological functions [ 212 ]. In the setting of MS, as discussed earlier, a critical step in the initiation of an MS lesion is the trafficking of lymphocytes across the BBB from the blood and into the brain. In the circumstances of MS, a variety of mechanisms, many of which employ monoclonal antibodies, have permitted the very specific blockage of these transits.

The first such agent to receive regulatory approval for the treatment of MS was natalizumab, a M A b directed against the ( α 4 ) subunit of the VLA-4 heterodimer ( see section on the Pathology of MS; above) . This therapy proved remarkably effective, reducing the clinical attack-rate by 68 ​ % compared to placebo [ 213 ]. One year later, however, the FDA withdrew natalizumab from the market because of the unexpectedly frequent occurrence of progressive multi-focal leukoencephalopathy ( PML ) – a potentially fatal condition due to the reactivation of a latent infection by the JC virus ( JCV ). Prior to the experience with natalizumab, PML was considered an extremely rare condition ( p r e v a l e n c e ≈ 0.5 c a s e s / 10 5 p o p u l a t i o n ) and generally occurred in circumstances characterized by a marked suppression of the immune system (e.g., HIV/AIDs infection, hematological malignancies, and the use of broad-spectrum immunosuppressants). JCV is a member of the Polyomaviridae family and, like EBV , it is a circularized double-stranded DNA virus, it ultimately infects a large percentage of the population ( 70 − 90 % ) , and once infection occurs (usually asymptomatic and often in childhood) it remains latent and persists indefinitely in certain tissues (e.g., kidneys, hematopoietic progenitor cells in the bone-marrow, peripheral B cells, and, possibly, the brain). However, apparently, natalizumab is too effective at blocking lymphocyte-trafficking such that, not only does it impact the transit of activated lymphocytes, it also prevents the normal trafficking of “surveilling” lymphocytes meant to detect and eliminate any incipient infection by re-activated viruses. After a year, the FDA reapproved natalizumab for clinical use although, currently, precautions are taken to mitigate the PML risk (e.g., the measurement of the presence and titer of serum antibodies to JCV both prior to and during therapy, the use of extended-interval dosing to permit, at least, some degree of immune surveillance to occur, and the use of routine MRI-screening to identify any new lesions, which appear more suggestive of PML than of MS).

Alemtuzumab is a humanized M A b directed against the cell-surface glycoprotein CD52 , which is of unknown function. CD52 is expressed on T and B cells, macrophages, and monocytes, but not on normal hematopoietic stem/progenitor cells. Alemtuzumab causes the depletion (cell-lysis) of both B and T cells through a combination complement-dependent cytotoxicity ( CDC ) and antibody-dependent cellular cytotoxicity ( ADCC ), and it leads to long-lasting changes in adaptive immunity. It does not directly impact lymphocyte trafficking although, by markedly lowering lymphocyte counts in the peripheral blood-compartment, it greatly restricts the entry of these cells into the CNS . Two large double-blind, active-comparator controlled, randomized trials of alemtuzumab in MS demonstrated a highly significant ( 49 − 55 % ) reduction in the clinical attack rate relative to IFNβ [ 214 , 215 ]. Long-term follow-up studies, conducted 6 and 12 years after the RCT, demonstrated remarkably persistent immunological changes and clinical benefits with just two courses of treatment administered one year apart [ 216 , 217 ]. The clinical use of this therapy, however, has been limited because the high percentage of treated patients ( 20 − 40 % ) who experience secondary autoimmune diseases involving the thyroid gland, kidney, lungs, and platelets [ 218 ].

Rituximab is a M A b directed against the 33–37 ​kDa, non-glycosylated, cell surface, transmembrane protein, CD20 [ 219 ]. Rituximab was approved by the FDA for use in the treatment of B cell malignancies in 1997, although the physiological role and regulation of CD20 remain uncertain. CD20 is expressed (both as a homo-dimer and as a homo-tetramer) on the surface of most B cells beginning with late pre-B lymphocytes. However, CD20 is not expressed by pro-B lymphocytes and its expression is lost in terminally differentiated plasmablasts and plasma cells. Similar to alemtuzumab, rituximab causes cell lysis through a combination CDC and ADCC mechanisms and results in a prolonged ( > 6 m o n t h s ) depletion of the B cells that express CD20 . In 2008, a large preliminary clinical trial of rituximab in MS demonstrated marked efficacy on both clinical and radiographic outcome measures [ 220 ]. However, in 2008, rituximab was coming off patent and the pharmaceutical company decided against pursuing the use of this agent in a phase-III clinical trial. Rather, on the basis of these results, they chose to undertake the development of a novel, and more fully humanized, anti- CD20
M A b called ocrelizumab – thereby delaying any approved anti- CD20 therapy in MS by more than 5 years. In 2017, the results of two large double-blind, active-comparator controlled, randomized trials of ocrelizumab, demonstrated a highly significant ( 46 − 47 % ) reduction in the clinical attack rate relative to IFNβ [ 221 ]. Since then, several other anti- CD20
M A b s have entered the market, each with very similar efficacy.

Modulation of sphingosine-1-phosphate (S1P) receptors

The first S1P- receptor modulator to be studied in MS was fingolimod, which is derived from myriocin – a metabolite of the fungus Isaria sinclarii , and which is phosphorylated in the cell, principally by sphingosine kinase 2, to produce phospho-fingolimod. There are five sphingosine-1-phosphate receptors ( S 1 PR 1 – S 1 PR 5 ) in the body to which phospho-fingolimod binds. In the context of MS, the most important of these is thought to be S 1 PR 1 , which (when activated by ligand-binding) facilitates the egress of lymphocytes from lymphoid tissues – from the white pulp to the red pulp and then back into the peripheral circulation. Phospho-fingolimod binding causes the internalization of the S 1 PR 1 receptor, which, by preventing the egress from lymphoid tissues, sequesters the lymphocytes in secondary lymphoid organs such as the lymph nodes, spleen, mucosa-associated lymphoid tissue and Peyer's patches. This action effectively removes both B and T cells from the peripheral blood-compartment and, thereby, prevents them from trafficking into the CNS . However, unlike many of the M A b s ( discussed in the subsection on Monoclonal Antibodies; above ), the lymphocytes are not lysed but, rather, are merely sequestered.

Two large double-blind, placebo-controlled, randomized trials of the use of fingolimod in the treatment of MS, demonstrated a highly significant ( 50 − 60 % ) reduction in the clinical attack rate relative to placebo [ 222 , 223 ]. Since then, several other S1P- receptor modulators have entered the market, each with similar efficacy.

Therapies (Potentially) Targeting the Random Mechanisms

Sex-based therapies

The “ truly ” random mechanism of MS-pathogenesis of MS seems to favor disease-development in women over and men and also seems to be principally responsible for the observed differences in disease-expression between susceptible men and susceptible women [ 64 ] – see also
Fig. 5 . Therefore, in some sense, any sex-based hypotheses in MS, at least potentially, crosses paths with this random mechanism. Nevertheless, it is unclear what might account for this sex-bias although, perhaps, the general differences between males and females in their anatomy, physiology, and gene or RNA expression might set the bias differently for each sex. Regardless of its actual basis, however, because it is random, it is not possible to manipulate this mechanism directly.

Nevertheless, as noted earlier ( see section on Epidemiology of MS; above ) MS-pathogenesis and its expression has two distinct components – the first is the development of the disease and the second is the occurrence of recurrent attacks of neurological dysfunction, which seem, at least superficially, to occur at random times after disease-onset. Clearly, the environmental events that drive the first component are not necessarily the same (or even related to) those events that drive the second. However, if the difference in disease-expression between sexes is primarily driven by a “ truly ” random mechanism, then the apparently random occurrence of attack-expression might also be driven by a similar random mechanism. Moreover, if both components include this “ truly ” random factor, then it seems reasonable to guess that the bias favoring disease-development in susceptible
women might similarly favor attack-development in susceptible
women . Such a notion, obviously, is purely speculative with no evidence behind it. However, if correct, one might expect that, if certain external environmental events are identified, which increase or decrease the likelihood of an MS-attack, any such event might disproportionately affect women . Three such environmental events are known or suspected.

The first event is the observed increase in the frequency of MS attacks, which occurs in the setting of non-specific viral syndromes such as fever, malaise, cough, rhinorrhea, nausea, abdominal pain, and diarrhea [ [224] , [225] , [226] ]. Typically, these studies divide a patient's time in the study into periods when they are “ at-risk ” and periods when they are “ not at-risk ”. The “ at-risk ” period is defined somewhat variably between studies but generally extends from 1 − 2 weeks before, until to 2 − 5 weeks after, the onset of infections symptoms. Any attack, which occurs outside the defined interval are considered to have occurred during the “ not at-risk ” period. All of these studies report a significant increase in attack frequency during the at-risk periods compared to not-at-risk periods [ [224] , [225] , [226] ], with relative risk ratios ( R R ) of: ( R R = 1.5 − 2.8 ) . However, unfortunately, these studies do not break down their results by sex, so it is unknown whether this increase disproportionately affects women.

The second event is the apparent increase in disease activity in MS patients treated with TNFα inhibitors [ 227 , 228 ]. The basis for this apparent connection between MS and TNFα inhibition, however, is not well understood and, with the notable exception of the Lenercept RCT [ 227 ], the impact of TNFα inhibition on MS has mostly been documented in case reports [ 228 ]. TNFα has two receptors (TNFR1 and TNFR2), each of which, when ligand-activated, produce different down-stream biological effects [ 227 , 228 ]. Thus, TNFR1 signaling mediates pro-inflammatory pathways and can result in cellular apoptosis [ 227 , 228 ]. By contrast, TNFR2 signaling modulates immune functions by promoting oligodendrocyte differentiation, myelin repair, and tissue preservation and, thereby, may have a protective role in the context of CNS autoimmunity [ 227 , 228 ]. Thus, one possibility to explain the apparently deleterious effects of TNFα inhibition in MS might be that, for different disease states, the importance of these two signaling pathways also differ. Nevertheless, regardless of the explanation, as was the case for the role of infections in “triggering” MS attacks ( see above ), the potential for sex-specific impacts of TNFα inhibition on these biological functions has not been explored.

The third event is the decrease in the frequency of MS attacks, which occurs during pregnancy, especially during the second and third trimester ( R R = 0.29 − 0.62 ), and the increased frequency of MS attacks during the first 3 months post-partum ( R R = 1.6 = 1.7 ) – both in comparison with pre-pregnancy attack rates [ [229] , [230] , [231] ]. The basis for these variations in the MS attack-rate is unknown but, it is possible that it relates to fluctuations in hormonal levels both during and after pregnancy and that these fluctuations influence the likelihood of an MS attack. Nevertheless, there are many physiological changes (both hormonal and non-hormonal) that occur during pregnancy [ 232 ] and there is no reason to discount any of these factors as potentially being responsible for the observed pregnancy-related changes in the attack rate. Moreover, the hormonal changes during pregnancy are quite complex.

For example, human chorionic gonadotropin ( HCG ) is made once a fertilized egg attaches along the uterine wall, about six days after fertilization and its levels rises throughout most of the first trimester. Human placental lactogen ( HPL ) is made by the placenta and helps to regulate both cellular metabolism and glucose levels. It usually peaks around 34 weeks and is associated with stimulation of the mammary glands. During pregnancy, estrogen (which is normally made in the ovaries) is also produced by the placenta, which leads to a higher concentration during pregnancy than otherwise. It peaks during the first trimester and it remains elevated throughout. The estrogen, estriol, is made by the fetal/placental unit. In the nonpregnant states, its levels are minimal but, during pregnancy, its levels increase progressively. Notably, estriol, administered to nonpregnant female mice in levels sufficient to induce serological pregnancy changes, has been reported to mitigate both encephalomyelitis and rheumatoid arthritis [ 233 ].

During pregnancy, also, progesterone is made by the ovaries and placenta. It increases markedly during the first trimester and then its concentrations level out for the second and third trimesters. It is involved in thickening the uterine lining to facilitate implantation of the fertilized egg and it increases blood flow to the uterus. Follicle-stimulating hormone ( FSH ) is secreted by the pituitary gland, and it promotes egg maturation and fertilization. It is made prior to and during first trimester. After this, it drops below non-pregnancy levels and remains low throughout pregnancy. Prolactin promotes breast growth and milk production, especially during the third trimester. After delivery, prolactin participates in causing milk to be released during breastfeeding. The hormone's level is highest in the first trimester, when it participates in the prevention of uterine contractions. Relaxin helps to regulate the cardiovascular system allowing for adjustment to an increased demand for oxygen. Relaxin levels peak at the end of the first trimester and then fall to intermediate levels for the remainder of the pregnancy, As the baby comes to term, relaxin helps to loosen the muscles, ligaments, and joints of the pelvis, thereby helping to prepare the mothers body for birth.

The non-hormonal effects of pregnancy on the immunological function of the individual are equally complex [ 232 ] so that any attempt to recreate the “pregnancy state” in an individual and thereby capture the protective effect of pregnancy on disease activity in MS will be challenging [ 233 , 234 ]. The three principal estrogens are estradiol, estrone and estriol. Estrogens act by regulating gene transcription via the nuclear estrogen receptors – ERα and Erβ – which (when ligand-activated) each exhibit distinct down-stream transcriptional effects. Estriol only very weakly interacts with ERα and preferentially binds to ERβ , which may be important because the consequences of binding to one or the other ER can lead to the opposite biological effects. Moreover, estriol seems to be the safest of the three estrogens [ 233 ]. Perhaps, in part, because of the many physiological complexities of pregnancy, the results of several preliminary clinical trials of hormonal therapy in MS, mostly of estriol by itself, have not demonstrated any definite clinical benefit [ 233 , 234 ].

Therapies Targeting Environmental Events

Trials targeting the environmental factors important for MS pathogenesis are the easiest to imagine (i.e., identify the putative environmental factor and demonstrate that modification of it reduces the occurrence at disease at follow-up). Nevertheless, such trials are, simultaneously, the most difficult to execute and interpret. There are several reasons for this: (1) Association studies don't always identify those factors that are in the causal pathways, or in the most important pathways, which lead to disease – i.e., association is not the same as causation (e.g., Fig. 7 ). Therefore, at the beginning of such a study, one would have relatively low confidence that any factor, identified by association only, was actually in the principle causal pathway or even in a causal pathway at all ( Fig. 7 ). (2) Elements in any important pathway of causation must precede the onset of the disease. For example, in a rare disease like MS, the life-time probability that a randomly selected individual from the population will develop disease is only ∼ 0.3 % [ 59 ]. This means having to follow 333 asymptomatic individuals to identify one expected MS patient. In this circumstance, to prove that any intervention reduced the likelihood of disease-occurrence will, therefore, take a huge number of subjects, the vast majority of whom won't develop disease. This is particularly problematic in MS, where the critical timing of the events ( see section on Timing of the Environmental Events; above ) is such that many of the antecedent events (e.g., possibly vitamin D deficiency and EBV infection) occur decades before symptom-onset (and, therefore, before diagnosis is possible). This further requires that this huge number of subjects will need to be followed for a very long time – likely exceeding the research careers of many of the investigators who start the study. (3) To conduct such a long-term study, therefore, requires a robust and reliable means of actually tracking and following individuals over a very long period of time. As discussed previously ( see sub-section on Vitamin D deficiency; above ), this would require a very large, population-based (preferably national) study and it would also require a national health-care system, which permitted researchers to track easily both children and adults over time and allowed researchers to have consent-free access to an anonymized medical records database.

Association studies don't always identify those factors that are in the causal pathways, or in the most important pathways, which lead to disease – i.e., association is not the same as causation (e.g., Fig. 7 ). Therefore, at the beginning of such a study, one would have relatively low confidence that any factor, identified by association only, was actually in the principle causal pathway or even in a causal pathway at all ( Fig. 7 ).

Elements in any important pathway of causation must precede the onset of the disease. For example, in a rare disease like MS, the life-time probability that a randomly selected individual from the population will develop disease is only ∼ 0.3 % [ 59 ]. This means having to follow 333 asymptomatic individuals to identify one expected MS patient. In this circumstance, to prove that any intervention reduced the likelihood of disease-occurrence will, therefore, take a huge number of subjects, the vast majority of whom won't develop disease. This is particularly problematic in MS, where the critical timing of the events ( see section on Timing of the Environmental Events; above ) is such that many of the antecedent events (e.g., possibly vitamin D deficiency and EBV infection) occur decades before symptom-onset (and, therefore, before diagnosis is possible). This further requires that this huge number of subjects will need to be followed for a very long time – likely exceeding the research careers of many of the investigators who start the study.

To conduct such a long-term study, therefore, requires a robust and reliable means of actually tracking and following individuals over a very long period of time. As discussed previously ( see sub-section on Vitamin D deficiency; above ), this would require a very large, population-based (preferably national) study and it would also require a national health-care system, which permitted researchers to track easily both children and adults over time and allowed researchers to have consent-free access to an anonymized medical records database.

These concerns essentially preclude any randomized trial to test the impact of modifying any hypothesized environmental factor thought to be the causal chain leading to MS. Thus, any study would have to be purely observational and would have to rely on the very large numbers of patients followed to reach any conclusions. Despite these challenges and the uncertainty of our knowledge, however, some sensible recommendations can still be provided to patients. For example, regardless of whether smoking is in a causal path to MS, smoking is known to have many negative health outcomes (e.g., lung malignancies, cardiovascular complications and obstructive pulmonary disease) so that patients would be well-advised to stop (or not to start) smoking.

The same is basically true for the use vitamin D supplementation in MS. As noted previously, of the published RCTs [ [169] , [170] , [171] ], all but one concluded that treatment with supplemental vitamin D had no effect on the occurrence of relapses, the worsening of disability, the occurrence of MRI activity, or alterations in immune function. Nevertheless, this evidence is not dispositive. The studies reviewed lasted a year or less in all but two cases (including the one successful trial) where they lasted 2 years. If vitamin D supplementation was clinically helpful, a much longer period of follow-up would be necessary. Also, there is no reason to believe that long-term benefit of supplementation would be manifested either as a reduction in the short-term relapse rate or in measurable short-term changes in EDSS. More to the point, on average in these trials, supplementation increased the serum concentration of 25(OH) D 3 from ( 22 n g / m l ) to ( 72 n g / m l ) . Although, the Cochrane reviewers noted the occurrence of some “serious” adverse events in the supplemented groups [ 169 ], none of these events actually seem likely to reflect any real toxicity of vitamin D (e.g., single episodes of pyelonephritis, breast cancer, uterine cancer, syncope, depression, pneumonia, menorrhagia, cellulitis, uterine polyps, and hypertension). Also, another study, specifically designed to look for evidence of toxicity at very high-dosage supplementation, raised the serum concentration of 25(OH) D 3 from ( 29 n g / m l ) to ( 165 n g / m l ) without any evidence of toxicity in terms of either adverse events or changes in serum calcium concentrations [ 235 ]. Thus, it seems that supplementation is well tolerated.

Finally, in equatorial Africa, native populations, living traditional life-styles, have an average serum 25(OH) D 3 concentration of ( 46 n g / m l ) , in comparison to an average serum 25(OH) D 3 concentration of ( 18 n g / m l ) for people living in the United Kingdom [ 236 , 237 ]. If Jablonski and Chaplin are correct that the lighter skin tones found in many ethnic groups living in more temperate global regions represents a convergent evolutionary adaptation to the need for more vitamin D synthesis in temperate (or more extreme) global regions [ 157 , 163 ], then this too suggests the need for supplementation, particularly in circumstances where, regardless of its relationship to MS, vitamin D deficiency seems to have other negative health consequences [ 154 , 160 ]. Therefore, based on the above considerations and on its safety, and although not of any proven benefit, it seems reasonable to recommend vitamin D supplementation in MS patients sufficient to sustain 25(OH) D 3 serum levels in the range of ( 50 − 70 n g / m l ) . A similar conclusion has also been reached by other authors [ 238 ]. The amount of supplementation necessary to achieve this serum level varies between individuals although, in general, ( ∼ 5 , 000 ) international units (IU) of vitamin D 3 per day or ( ∼ 35 , 000 ) IU/week are sufficient. Nevertheless, the actual level of supplementation recommended should be adjusted based on the person's measured serum levels.

The last environmental event to consider is an infection by EBV . This is the environmental factor, which is both most likely to be in the causal pathway leading to MS and, at the same time, also seems to be the most difficult factor to mitigate. One obvious strategy would be to develop a safe and effective vaccine against EBV that would establish sterile immunity in vaccinated infants or children without a primary EBV infection. Nevertheless, to date, the goal of developing such a vaccine has proven quite difficult to achieve [ 239 ]. Even if this is ever accomplished, however, the vaccine would need to provide a lasting immunity – a fact that would need to be proven over decades of follow-up after vaccine development. Otherwise, as noted previously ( see sub-section on EBV; above ), if the immunity were to wane over time and thereby permit an actual EBV infection during adolescence, this might increase the likelihood of developing both mononucleosis acutely and MS in the future [ 124 , 125 , 127 , [132] , [133] , [134] , [135] , [136] , [137] , [138] ]. Even if subsequent “booster” vaccinations were able to provide such long-lasting immunity, many originally vaccinated individuals would likely not follow through with this “booster” program, resulting in the same concerns. Alternatively, one might contemplate the idea of introducing an actual EBV infection into young children. However, both because of the known potential negative health consequences of EBV infection [ 131 ], and because only ( ∼ 0.3 % ) of individuals in the population will ever develop MS, the risks of any such approach would outweigh any possible benefit and, thus, are unacceptable.

Author contributions

DSG prepared the initial manuscript draft, wrote and edited the final manuscript, wrote the computer programs (when necessary), and performed all analyses presented in this article.

Funding sources and other support

This project received no funding or support from any source.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.