Title

Sex as a determinant of relapse incidence and progressive course of multiple sclerosis

Running title

Multiple sclerosis relapses: sex and age

Authors and affiliations

Tomas Kalincik; Department of Medicine, University of Melbourne, Melbourne, Australia; Department of Neurology, Royal Melbourne Hospital, Melbourne, Australia

Vino Vivek; Department of Neurology, Royal Melbourne Hospital, Melbourne, AustraliaVilija Jokubaitis; Department of Medicine, University of Melbourne, Melbourne, AustraliaJeannette Lechner-Scott; John Hunter Hospital, Newcastle, AustraliaMaria Trojano; Department of Basic Medical Sciences, Neuroscience and Sense Organs,

University of Bari, Bari, ItalyGuillermo Izquierdo; Hospital Universitario Virgen Macarena, Sevilla, SpainAlessandra Lugaresi; MS Center, Department of Neuroscience and Imaging, University ‘G.

d’Annunzio’, Chieti, ItalyFrancois Grand’Maison; Neuro Rive-Sud, Hôpital Charles LeMoyne, Quebec, CanadaRaymond Hupperts; Orbis Medicle Center, Sittard, The NetherlandsCelia Oreja-Guevara; University Hospital San Carlos, IdISSC, Madrid, SpainRoberto Bergamaschi; Neurological Institute IRCCS Mondino, Pavia, ItalyGerardo Iuliano; Ospedali Riuniti di Salerno, Salerno, ItalyRaed Alroughani; Amiri Hospital, Kuwait City, KuwaitVincent Van Pesch; Cliniques Universitaires Saint-Luc, Brussels, BelgiumMaria Pia Amato; Department NEUROFARBA, Section of Neurosciences, University of

Florence, Florence, ItalyMark Slee; Flinders University and Medical Centre, Adelaide, AustraliaFreek Verheul; Groen Hart Ziekenhuis, Gouda, The NetherlandsRicardo Fernandez-Bolanos; Hospital Universitario Virgen de Valme, Seville, SpainMarcela Fiol; FLENI, Buenos Aires, ArgentinaDaniele La Spitaleri; AORN San Giuseppe Moscati, Avellino, ItalyEdgardo Cristiano; Hospital Italiano, Buenos Aires, ArgentinaOrla Gray; Craigavon Area Hospital, Portadown, UKJose Antonio Cabrera-Gomez; Centro Internacional de Restauracion Neurologica, Havana,

CubaVahid Shaygannejad; Al-Zahra Hospital, Isfahan University of Medical Sciences, Isfahan, IranJoseph Herbert; New York University Hospital for Joint Diseases, New York, USASteve Vucic; Westmead Hospital, Sydney, AustraliaMerilee Needham; Royal North Shore Hospital, Sydney, Australia

Tatjana Petkovska-Boskova; Clinic of Neurology Clinical Center, Skopje, MacedoniaCarmen-Adella Sirbu; Central Clinical Emergency Military Hospital, Bucharest, RomaniaPierre Duquette; Hôpital Notre Dame, Montreal, CanadaMarc Girard; Hôpital Notre Dame, Montreal, CanadaPierre Grammond; Hotel-Dieu de Levis, Quebec, CanadaCavit Boz; Karadeniz Technical University, Trabzon, TurkeyGiorgio Giuliani; Ospedale di Macerata, Macerata, ItalyMaria Edite Rio; Hospital CUF, Porto, PortugalMichael Barnett; Brain and Mind Research Institute, Sydney, AustraliaShlomo Flechter; Assaf Harofeh Medical Center, Beer-Yaakov, IsraelFraser Moore; Jewish General Hospital, Montreal, CanadaBhim Singhal; Bombay Hospital Institute of Medical Sciences, Mumbai, IndiaElizabeth Alejandra Bacile; Instituto de Neurociencias Cordoba, Cordoba, ArgentinaMaria Laura Saladino; INEBA, Buenos Aires, ArgentinaCameron Shaw; Geelong Hospital, Geelong, AustraliaEli Skromne; Hospital Angeles Mexico City, Lomas, MexicoDieter Poehlau; Multiple Sclerosis Centre Kamillus-Klinik, Asbach, GermanyNorbert Vella; Mater Dei Hospital, MaltaTimothy Spelman; Department of Neurology, Royal Melbourne Hospital, Melbourne,

AustraliaDanny Liew; Melbourne EpiCentre, University of Melbourne and Melbourne Health,

Melbourne, AustraliaTrevor J Kilpatrick*; Centre for Neuroscience Research and the Melbourne Neuroscience

Institute, University of Melbourne, Melbourne, Australia; Department of Neurology, Royal Melbourne Hospital, Melbourne, Australia;

Helmut Butzkueven*; Department of Medicine, University of Melbourne, Melbourne, Australia; Department of Neurology, Royal Melbourne Hospital, Melbourne, Australia; and Department of Neurology, Box Hill Hospital, Monash University, Box Hill, Australia

on behalf of the MSBase Study Group†

*These authors contributed equally to the manuscript.†Contributing members of the MSBase Study Group are listed in the Acknowledgements.

Corresponding author: Tomas Kalincik; L4 Centre, Melbourne Brain Centre at Royal Melbourne Hospital, Grattan St, Parkville VIC 3050, Australia; Tel: +61 3 9342 4404, Fax: +61 3 9349 5997

Keywords:multiple sclerosis, sex, risk factors, MSBase, prediction

Word count – manuscript: 3941

SUMMARY

Objectives

To evaluate sex differences in the incidence of multiple sclerosis relapses. To assess the

relationship between sex and primary progressive disease course. To compare effects of age

and disease duration on relapse incidence.

Methods

Annualised relapse rates were calculated using the MSBase registry. Patients with

incomplete data or less than one year of follow-up were excluded. Patients with primary

progressive multiple sclerosis were only included in the sex ratio analysis. Relapse

incidences over 40 years of multiple sclerosis or 70 years of age were compared between

females and males with Andersen-Gill and Tweedie models. Female-to-male ratios stratified

by annual relapse count were evaluated across disease duration and patient age and

compared between relapse-onset and primary progressive multiple sclerosis.

Results

The study cohort consisted of 11,570 eligible patients with relapse-onset and 881 patients

with primary progressive multiple sclerosis. Among the relapse-onset patients (82,552

patient-years), 48,362 relapses were recorded. Relapse frequency was 17.7% higher in

females compared to males. Within the initial five years, the female-to-male ratio increased

from 2.3:1 to 3.3:1 in patients with 0 versus ≥4 relapses per year, respectively. The

magnitude of this sex effect increased at longer disease duration and older age (p < 10-12).

However, the female-to-male ratio in patients with relapse-onset multiple sclerosis and zero

relapses in any given year was double that of the patients with primary progressive multiple

sclerosis. Patient age was a more important determinant of decline in relapse incidence

than disease duration (p < 10-12).

Conclusions

Females are predisposed to higher relapse activity than males. However, this difference

does not explain the markedly lower female-to-male sex ratio in primary progressive

multiple sclerosis. Decline in relapse activity over time is more closely related to patient age

than disease duration.

INTRODUCTION

Episodic exacerbations are defining phenomena for relapsing-remitting multiple sclerosis

(MS) and can be directly associated with residual disability (Hirst et al., 2008). Even though

the evidence remains inconclusive (Hutchinson, 2011), an association between high relapse

rate in the first two years after disease onset and later accumulation of irreversible

neurological impairment has been suggested (Kalincik et al., 2012; Scalfari et al., 2010).

There have been only a limited number of studies delineating predictors of relapse activity,

with only one study examining a large retrospective cohort (Held et al., 2005; Mowry et al.,

2009; Tremlett et al., 2009). These studies suggested several predisposing factors, such as

sex, age, ethnicity and time from MS onset. However, their outcomes, in particular those

related to effect of sex on relapse incidence, have often been discordant.

There is a marked predominance of women among MS patients with relapse-onset course.

The sex ratios are 3:1 in relapsing-remitting and 3:2 in secondary progressive MS, but in

primary progressive MS the both sexes are represented equally (Alonso and Hernan, 2008;

Confavreux and Vukusic, 2006b; Koch-Henriksen and Sorensen, 2010; Runmarker and

Andersen, 1993).

We have used MSBase, a longitudinal, international, observational registry of patients with

MS, to evaluate the effect of sex on relapse activity up to 40 years of MS duration or 70

years of age. We have directly compared the impacts of age and disease duration on relapse

frequency in a predominantly contemporary MS cohort. We have also tested the hypothesis

that the female-to-male ratio increases gradually with relapse MS activity and that the

primary progressive disease represents a non-relapsing extreme along this continuum.

PATIENTS AND METHODS

Ethics statement

The MSBase registry (Butzkueven et al., 2006) was approved by the Melbourne Health

Human Research Ethics Committee, and by the local ethics committees in all participating

centres (or exemptions granted, according to applicable local laws and regulations). If

required, written informed consent was obtained from enrolled patients, in accordance with

the Declaration of Helsinki.

Patients and follow-up

Longitudinal clinical data from 18,885 patients from 55 MS centres in 25 countries were

extracted from the MSBase registry in February 2012. Data recorded between years 1951

and 2012 were used. The majority of the patients were enrolled in the MSBase registry in

year 2000 or later (75%), with 20% enrolled in years 1990-2000, 3% enrolled in years 1980-

1990 and only 0.9% of patients enrolled before 1980. The participating centres contributed

between 2 and 1239 cases fulfilling the study inclusion criteria, with recorded ARR in

relapse-onset MS ranging from 0.1 to 2 and female-to-male ratio ranging from 0.9:1 to

4.4:1. The country-specific ranges were 2-3282 for the number of eligible patients, 0.2-2 for

ARR in relapse-onset MS and 0.9-4.4:1 for female-to-male ratio. Patients with incomplete

data (i.e. not fulfilling the minimal dataset requirement, see below) or less than one year of

recorded clinical follow-up were excluded. The minimal dataset consisted of patient date of

birth, sex, MS centre, dates of MS onset and clinical follow-up, disease course and disability

(at inclusion and censoring), start and end dates of disease modifying treatment exposure

and list of clinical relapses (including date of onset and relapse treatment status). Patients

with relapse-onset MS (i.e. clinically isolated syndrome, relapsing-remitting MS, secondary

progressive MS or relapsing-progressive MS) were included in all analyses. Patients with

primary progressive MS were only included in the analyses comparing female-to-male ratios

between relapse-onset and primary progressive MS (see below). Finally, patients with

relapse-onset MS and at least 10 years of recorded follow up, and all patients with primary

progressive MS were included in a subgroup analysis of female-to-male ratios stratified by

relapse activity.

The analysed data were recorded as part of routine clinical practice. The usual data entry

practice at most centres was real time or near-real time data entry in relation to clinical

visits. The MSBase protocol stipulates minimum annual updates of the minimum dataset,

but patients with less frequent visits were not excluded from the analysis. Data entry portal

was either the iMed patient record system or the MSBase online data entry system.

A relapse was defined as occurrence of new symptoms or exacerbation of existing

symptoms persisting for at least 24 hours, in the absence of concurrent illness or fever, and

occurring at least 30 days after a previous relapse (Schumacher et al., 1965). Date of onset

was recorded for each clinical relapse. Formal scoring of relapse-associated disability was

not required upon relapse entry. Disability was scored by accredited scorers (online

Neurostatus certification was required at each centre) using the Expanded Disability Status

Scale (EDSS). Duration of MS was calculated as the time from the patient-reported first

clinical manifestation of the disease and relapsing / progressive onset of disease was

assessed by participating neurologists. Primary progressive MS was defined as the disease

with at least one year of disease progression from its first clinical manifestation and with

zero recorded relapses. In each patient, unadjusted annualised relapse rates (ARRs)

stratified by disease duration or patient age were calculated by including the initial relapse

and the number of relapses for years for which clinical data were available. Censoring was

defined to occur on the day of the last recorded clinical visit. Information about patient

death and its relation to MS was not collected in the database.

To assure quality of the analysed data, only information from centres contributing at least

10 active records (i.e. cases with regular annual updates of clinical information) to the

MSBase registry was used, as stipulated in the study protocol. A date of onset was required

for all recorded events, including relapses, visit (with or without EDSS score), changes in

disease course and changes in treatment. Prior to the analysis the recorded data were

verified using a series of automated procedures to identify any invalid or inconsistent

entries.

Statistical analysis

Statistical analyses were carried out using Statistica 10 (Statsoft, Tulsa, OK, USA) and R

(http://www.R-project.org). All hypotheses were tested at the two-tailed 0.05 level of

statistical significance, after applying the Benjamini-Hochberg correction for multiple

hypothesis testing. Two proportional hazards models with robust variance estimation

(Andersen-Gill) and Efron approximation method, adjusted either for patient age or disease

duration, were used to estimate cumulative hazard* of relapses for each sex. The outcome

variable was time to relapse with multiple entries per patient allowed. The models were

adjusted for pregnancy (proportion of time pregnant from the study enrolment or the

previous recorded event to the next event/censoring, treated as time-varying variable) and

immunomodulatory therapy, when any change in therapy with or without relapse was

treated as a relapse (response) or non-relapse (censoring) event. The immunomodulatory

agents considered were interferon β-1a 22 μg or 44 μg subcutaneous injection thrice

weekly, interferon β-1a 30 μg intramuscular injection once weekly, interferon β-1b 250 μg

subcutaneous injection every other day, glatiramer acetate 20 mg subcutaneous injection

daily, natalizumab 300 mg intravenous infusion once monthly, fingolimod 0.5 mg orally once

daily, dimethyl fumarate 240 mg orally twice or thrice daily, cladribine 10 mg orally once

monthly for two months, and teriflunomide 7 mg or 14 mg orally once daily.

Two Tweedie models with index parameters 1.5 (i.e. compound Poisson-Gamma models)

adjusted for sex, pregnancy and treatment (the latter two variables expressed as the

proportion of time pregnant or treated with disease modifying agent, respectively,

throughout the recorded follow-up) were built to evaluate separately the effects of age and

disease duration on ARR. Each patient contributed a single entry of overall ARR calculated as

the number of relapses divided by the duration of the follow-up. A similar final Tweedie

model including both age and disease duration was used to compare the effects of age and

disease duration.

We evaluated the effect of relapse count on female-to-male ratio at any completed

observational year using two logistic mixed effect models. The outcome in these models was

the probability of being female (i.e. the female gender recorded in individual patients), with

multiple entries per patient allowed (one entry per completed year of follow-up /

chronological age) and the random effect being the patient unique identifier. The predictors

of interest were the number of relapses recorded in any completed follow-up year, patient

age or MS duration (either age or MS duration was used in either model), and their

interaction term. The models were adjusted for age at MS onset, disease course, pregnancy

(proportion of time in any completed year) and immunomodulatory therapy (proportion of

time in any completed year).

RESULTS

Of the 18,885 patients in the MSBase registry (17,021 with relapse-onset MS, 1018 with

primary progressive MS and 846 with the information about disease course missing), data

from 11,570 relapse-onset and 881 (7%) primary progressive patients fulfilled the inclusion

criteria (Figure 1). The female-to-male ratio among the excluded patients (n = 6434) was

2.4:1 (4516 females, 1918 males, in no instance the information about patient sex was

missing). Tables 1 and 2 show demographic and clinical characteristics of the included

patients with relapse-onset and primary progressive MS, respectively. Patient characteristics

were comparable between females and males, including visit frequency, with only

marginally shorter disease duration and larger end-of-follow-up disability among males.

Numbers of patients with relapse-onset MS included in the analysis stratified by disease

duration and age are shown in Figure 2A and B. In 82,552 patient-years of observation

recorded among the relapse-onset patients, 48,362 relapses were recorded. Of these, 42%

were concurrent with clinical visits. Figure 2C and D shows unadjusted ARR with respect to

disease duration and age. Within the first year of MS, the relapse rate was relatively high

(1.1 relapse / year), which was a result of its inflation by inclusion of the first recorded

relapse. Thereafter, the ARR gradually decreased from 0.5 at 2 years to 0.1 at 40 years. Also,

a marked decline in ARR with age was seen, when ARR decreased from 0.9 at the age 17 to

0.1 at the age 70. Importantly, females tended to show higher unadjusted ARR during the

initial 18 years of MS duration and between 27 and 49 years of age. These differences were

further enhanced by adjusting the ARR for MS duration / patient age, pregnancy and

treatment (Figure 2E and F). Females showed consistently higher adjusted ARR (on average

by 17.7%, 95% confidence interval = 15.1 – 20.3%) than males throughout the disease

duration and at every chronological age (risk ratio = 1.08, 95% confidence interval = 1.06 –

1.11, p ≤10-16, Tweedie models), with a relative decrease in this difference at long MS

durations or at an older age. Accordingly, the cumulative hazard of relapses (see Figure 2G

and H) was consistently higher among females and reached 9.3 (95% confidence interval =

9.1-9.5) in women and 8.4 (95% confidence interval = 8.1-8.7) in men at 40 years after MS

onset (hazard ratio = 1.1, 95% confidence interval = 1.05 - 1.14, p = 10-5, Andersen-Gill

model). At the age of 70, the cumulative hazard reached 28.5 (95% confidence interval =

26.6-30.7) in women and 26.7 (95% confidence interval = 24.7 – 28.8) in men (hazard ratio =

1.1, 95% confidence interval = 1.07 - 1.14, p < 10-12, Andersen-Gill model).

These observations in relapse-onset MS were further confirmed by the fact that the female-

to-male ratio was higher among the patients with higher annual relapse count (see Figure 3,

β = 0.03, p < 10-12, logistic mixed models). Within the initial five years of disease onset, the

ratio increased from 2.3:1 in patients with no relapses to 3.3:1 in patients with at least four

relapses per year. Interestingly, this effect was accentuated by longer disease duration, as

the sex ratio among the frequently relapsing patients exceeded 4:1 (interaction term p < 10-

12, logistic mixed model). A similar interaction effect was seen for age (p < 10-12, logistic

mixed model). Overall, the sex ratios stratified by disease course were 2.1:1 for clinically

isolated syndrome (as defined by the McDonald 2005 criteria (Polman et al., 2005)), 2.8:1

for relapsing remitting MS, 2.2:1 for secondary progressive MS and 1.3:1 for relapsing

progressive MS. Compared to the patients with relapse-onset MS, the population diagnosed

with primary progressive disease showed a significantly lower female-to-male ratio across

all disease durations or patient age groups - approximately 1.2:1 (β = 1.13 – 1.16, p < 10-12,

logistic mixed model; see Figure 3). In fact, the female-to-male ratio in patients with relapse-

onset MS and zero relapses in any given year was 2:1 and therefore almost double that of

the patients with primary progressive MS. In a subgroup analysis including 2969 patients

with relapse-onset MS and data recorded over at least 10 years, the sex ratio increased from

2.2:1 in those with zero overall relapses† to 3.2:1 in those with 8 or more relapses captured

during their 10-year follow-up (see Figure 4). In 881 patients with primary progressive MS

(and therefore no recorded relapses) regardless of the duration of their follow-up, the

female-to-male ratio was 1.2:1.

Among the patients with relapse-onset MS, the incidence of relapses declined with both MS

duration (by 0.01 per year, 95% confidence interval = 0.0001 – 0.03) and patient age (by

0.01 per year, 95% confidence interval = 0.006 – 0.02). This was confirmed by the adjusted

Andersen-Gill models (MS duration: hazard ratio = 0.87, 95% confidence interval = 0.864 –

0.872, p < 10-12; age: hazard ratio = 0.95, 95% confidence interval = 0.949 – 0.953, p < 10-12).

The two separate Tweedie models (dispersion parameters = 0.52 and 0.51) further

confirmed the significant effects of MS duration (rate ratio = 0.99, 95% confidence interval =

0.992 – 0.996, p = 10-11) and age (rate ratio = 0.98, 95% confidence interval = 0.982 – 0.985,

p < 10-12) on ARR. Interestingly, in the final Tweedie model containing both MS duration and

patient age (dispersion parameter = 0.51), only the patient age was found to have an

independent negative association with ARR (rate ratio = 0.98, 95% confidence interval =

0.979 – 0.983, p < 10-12).

DISCUSSION

In a large mostly contemporary population of patients followed in the MSBase registry,

women have a higher relapse rate than men throughout the course of MS. The sex ratio in

progressive-onset forms of MS is markedly different from that in relapse onset MS and is

discrete from the gradient determined by the relapse frequency. We have confirmed a

strong inverse relationship between relapse frequency and either disease duration or age.

Of these two collinear factors, chronological age is more closely related to the relapse

incidence.

In the only available retrospective cohort study of predictors of relapse incidence, Tremlett

and colleagues showed that women had 14.3% higher ARR than men (Tremlett et al., 2008).

We have confirmed this observation by demonstrating a similar sex effect of 17.7%.

Compared to Tremlett et al., our recorded ARR was approximately double (0.59 vs. 0.23,

range 0.1-0.9 vs. 0.08-0.29 depending on patient age, respectively), despite the presumed

higher proportion of patients receiving disease modifying treatment. This could be

attributed to the inclusion of the first clinical episodes in the relapse count in our study and

the near real-time data entry in the MSBase registry, which could have partially ameliorated

the expected under-reporting of relapses.

Other literature reporting incidence of relapses in MS only describes relatively short follow-

up periods. A series of studies identified young age (Mowry et al., 2009; West et al., 2006)

and short disease duration (Held et al., 2005) but not sex as being independently predictive

of higher relapse incidence during one-year observational periods. The discrepancies

between these studies and our present analysis could be explained by the differences in

statistical power determined by the size of the cohorts and the analytical designs. With

respect to the low effect-to-noise ratio, the other studies were most probably

underpowered to assess the effect of sex, as they only evaluated information derived from

105 – 727 patient-years.

We have seen that women are over-represented among patients with high relapse

frequency, in particular later in the disease course and at older chronological age. This

suggests that the attenuation of clinically apparent episodic inflammatory activity is delayed

in women compared to men. In addition, higher mortality from MS-unrelated causes in men

could potentially contribute to this phenomenon; however, the lack of detailed mortality

data in the MSBase registry prevented the evaluation of this potential confounder.

Importantly, the follow-up duration, which did not differ between the enrolled women and

men, did not support sex-dependent study drop-out.

The difference in relapse risk associated with sex probably only plays a small role in

explaining the reported differences in sex ratios between patients with relapse- and

progressive-onset MS (Confavreux and Vukusic, 2006b; Kampman et al., 2013). Since men

show less frequent relapse activity throughout the disease course and life span, one could

argue that their disease is more likely to be classified as progressive. However, the sex ratio

among patients with primary progressive MS is still very distinct from that among patients

with relapse-onset MS and no relapse activity observed during the study period (see Figures

3 and 4). Thus, we rejected the hypothesis that sex difference in propensity to relapses is

fully accountable for the different sex ratios between relapse- and progressive-onset MS.

From this perspective, primary progressive MS can not be viewed as a mere extreme along

the continuum of relapse activity in relapsing-remitting MS.

It is of further interest how the relapse incidence interacts with progression of disability. It

has been previously suggested that early relapses may be associated with the development

of permanent disability (Hirst et al., 2008; Kalincik et al., 2012; Leray et al., 2010; Lublin et

al., 2003; Scalfari et al., 2010; Tremlett et al., 2009). There is also evidence both supporting

(Confavreux and Vukusic, 2006a; Confavreux et al., 2003; Leray et al., 2010) and opposing

(Amato et al., 1999; Simone et al., 2002; Trojano et al., 1995) the common view of faster

disability accrual in men. Clearly, the associations between sex and disability warrant further

clarification in observational studies, with the perspective of elucidating the complex

interactions between sex, relapses and disability.

MS duration and patient age are two largely collinear variables, whose effect on disease

outcomes is difficult to differentiate. Whether the MS relapse activity is determined

predominantly by the time from MS onset, or by patient age, is not known. While some

works identified disease duration as an independent predictor of relapse incidence (Held et

al., 2005), others reported independent effect of age (Inusah et al., 2010). Tremlett et al.

described a faster decline in ARR with disease duration in patients with later MS onset. It is

possible, that this interaction effect was in fact driven by the effect of age, but a direct

statistical comparison between the effects of age and disease duration was not reported

(Tremlett et al., 2008). In our present study, we have directly compared these two collinear

variables using a multivariate Tweedie model. Our analysis indicates a dominant role of

chronological age (approximately 2% per year decrease in ARR) over that of disease duration

(approximately 1% per year decrease in ARR). This observation is complementary to

previous works, which demonstrated that age at onset of progressive phase is remarkably

similar between patients with primary progressive vs. relapsing-remitting / secondary

progressive disease course, suggesting that presence and duration of relapsing phase is, to a

great extent, defined by patient age (Confavreux and Vukusic, 2006b; Leray et al., 2010;

Scalfari et al., 2011).

Our analysis was carried out in a large international observational cohort - the MSBase

registry. The studied population was representative of patient populations managed at large

academic MS centres, which might limit generalisation of our observations to a prevalent

population. Inclusion in the MSBase might have posed a risk of recruitment bias; this was

subject to the criteria for patient inclusion at the participating centres. Also, preferential

data entry of patients with relapse-onset disease course at the study sites could have

resulted in a relative under-representation of patients with primary progressive disease

(which in our study was approximately 7%). Selection bias was potentially introduced by the

applied inclusion criteria. However, the fact that female-to-male ratios were comparable

between the excluded and included populations rendered this situation unlikely. Survival

bias could potentially have been introduced in case that the patients with more pronounced

disability were more often lost to follow-up. This hypothetical situation might favour the sex

with slower disability accrual. On the other hand, given the parallel patient disposition of

both sexes with respect to disease duration and patient age, the survival bias was

improbable (see Figures 2A and B). The information prior to year 2000 represented data

transferred from other clinical databases, and constituted only a small proportion of the

evaluated data. The ARR and subsequently the differences studied in our analysis could have

been underestimated by the possible relapse under-reporting. The differences in relapse

reporting at the participating centres, as implied by the variability in the centre-specific

ARRs, could have inflated the under-reporting error. On the other hand, the fact that only

42% of the recorded relapses were concurrent with a recorded clinical visit and that EDSS

validation of relapses was not required indicates that relapse over-reporting could have

occurred. For all of these potential confounders, we presume that their influence applied

consistently across the studied population, regardless of patient sex, age or disease

duration. It should be noted that our analysis contained remarkable statistical power

provided by 82,552 patient-years from a longitudinal patient follow-up with the mean

follow-up period being 6 years. Overall, we were able to analyse ARR over 40 years of

disease duration and 70 years of age.

In this study we evaluated effect of sex on the incidence of MS relapses throughout the

disease- and life span and in the relapse-onset versus primary progressive course in a

longitudinally followed large patient population. Its outcomes help elucidate patterns of MS

activity determined by sex and in the future might lead to our improved understanding of

MS etiopathogenesis.

FUNDING

The work was supported by the Multiple Sclerosis Research Australia Postdoctoral

Fellowship to TK [11-054], NHMRC Career Development Award (Clinical) to HB [ID628856],

NHMRC Project Grant [1032484], NHMRC Centre for Research Excellence [Grant ID

1001216] and the MSBase Foundation. The MSBase Foundation is a not-for-profit

organization that receives support from Merck Serono, Biogen Idec, Novartis Pharma, Bayer-

Schering, Sanofi-Aventis and BioCSL.

ACKNOWLEDGEMENTS

MSBase study group contributors: From MS-Centrum Nijmegen, The Netherlands, Dr Cees

Zwanikken; from Francicus Ziekenhuis, The Netherlands, Ms Leontien den Braber-Moerland;

from Jeroen Bosch Ziekenhuis, The Netherlands, Dr Erik van Munster; from Centre

hospitalier del’Universite de Montreal, Hopital Notre-Dame, Canada, Dr Elaine Roger and Dr

Pierre Despault; from the Royal Melbourne Hospital, Australia, Dr Mark Marriott, Dr Anneke

Van der Walt, Dr John King , Dr Katherine Buzzard, Dr Jill Byron and Ms Lisa Morgan; from

Box Hill Hospital, Monash University, Australia, Dr Olga Skibina and Ms Jodi Haartsen; from

St Vincent’s Hospital, Australia, Dr Mark Paine; from Department of Neuroscience and

Imaging, University ‘G. d’Annunzio’, Italy, Dr Giovanna De Luca, Dr Valeria Di Tommaso, Dr

Daniela Travaglini, Dr Erika Pietrolongo, Dr Maria di Ioia and Dr Deborah Farina; from

Hospital Italiano, Argentina, Dr Juan Ignacio Rojas and Dr Liliana Patrucco; from Consultorio

Privado, Buenos Aires, Argentina, Dr Aldo Savino; from Hopital Tenon, Paris, France, Dr

Etienne Roullet; from FLENI, Argentina, Dr Jorge Correale and Dr Celica Ysrraelit; from

Ospedale di Macerata, Italy, Dr Elisabetta Cartechini and Mr Eugenio Pucci; from John

Hunter Hospital, Australia, Dr David Williams and Dr Lisa Dark; from Sheba Medical Center,

Tel Hashomer, Israel, Dr Joab Chapman; from Hospital Fernandez, Argentina, Dr Norma Deri;

from Hospital Ecoville, Brazil, Dr Walter Oleschko Arruda; from HIGA Gral. San Martin La

Plata, Argentina, Dr Santiago Vetere; from Jahn Ferenc Teaching Hospital, Hungary, Dr Csilla

Rozsa and Dr Krisztian Kasa; and from Thomas Jefferson University, Philadelphia, USA, Dr

Donald McCarren.

NOTES

*Cumulative hazard function is interpreted as the most probable count of repeatable events

that would be expected for each individual by defined time, if the exposure to the risk

started at time 0.

†In these patients, the start of the 10-year follow-up period did not coincide with their

disease onset. The diagnosis of relapsing-remitting MS was based on their relapse activity

preceding the recorded follow-up.

REFERENCES

Alonso A, Hernan MA. Temporal trends in the incidence of multiple sclerosis: a systematic

review. Neurology 2008; 71(2): 129-35.

Amato MP, Ponziani G, Bartolozzi ML, Siracusa G. A prospective study on the natural history

of multiple sclerosis: clues to the conduct and interpretation of clinical trials. J

Neurol Sci 1999; 168(2): 96-106.

Butzkueven H, Chapman J, Cristiano E, Grand'Maison F, Hoffmann M, Izquierdo G, et al.

MSBase: an international, online registry and platform for collaborative outcomes

research in multiple sclerosis. Mult Scler 2006; 12(6): 769-74.

Confavreux C, Vukusic S. Age at disability milestones in multiple sclerosis. Brain 2006a;

129(Pt 3): 595-605.

Confavreux C, Vukusic S. Natural history of multiple sclerosis: a unifying concept. Brain

2006b; 129(Pt 3): 606-16.

Confavreux C, Vukusic S, Adeleine P. Early clinical predictors and progression of irreversible

disability in multiple sclerosis: an amnesic process. Brain 2003; 126(Pt 4): 770-82.

Held U, Heigenhauser L, Shang C, Kappos L, Polman C. Predictors of relapse rate in MS

clinical trials. Neurology 2005; 65(11): 1769-73.

Hirst C, Ingram G, Pearson O, Pickersgill T, Scolding N, Robertson N. Contribution of relapses

to disability in multiple sclerosis. J Neurol 2008; 255(2): 280-7.

Hutchinson M. Relapses do not matter in relation to long-term disability: commentary. Mult

Scler 2011; 17(12): 1417.

Inusah S, Sormani MP, Cofield SS, Aban IB, Musani SK, Srinivasasainagendra V, et al.

Assessing changes in relapse rates in multiple sclerosis. Mult Scler 2010; 16(12):

1414-21.

Kalincik T, Vaneckova M, Tyblova M, Krasensky J, Seidl Z, Havrdova E, et al. Volumetric MRI

Markers and Predictors of Disease Activity in Early Multiple Sclerosis: A Longitudinal

Cohort Study. PLoS One 2012; 7(11): e50101.

Kampman M, Aarseth J, Grytten N, Benjaminsen E, Celius E, Dahl O, et al. Sex ratio of

multiple sclerosis in persons born from 1930 to 1979 and its relation to latitude in

Norway. Journal of neurology 2013.

Koch-Henriksen N, Sorensen PS. The changing demographic pattern of multiple sclerosis

epidemiology. Lancet Neurol 2010; 9(5): 520-32.

Leray E, Yaouanq J, Le Page E, Coustans M, Laplaud D, Oger J, et al. Evidence for a two-stage

disability progression in multiple sclerosis. Brain 2010; 133(Pt 7): 1900-13.

Lublin FD, Baier M, Cutter G. Effect of relapses on development of residual deficit in multiple

sclerosis. Neurology 2003; 61(11): 1528-32.

Mowry EM, Pesic M, Grimes B, Deen SR, Bacchetti P, Waubant E. Clinical predictors of early

second event in patients with clinically isolated syndrome. J Neurol 2009; 256(7):

1061-6.

Polman CH, Reingold SC, Edan G, Filippi M, Hartung HP, Kappos L, et al. Diagnostic criteria

for multiple sclerosis: 2005 revisions to the "McDonald Criteria". Ann Neurol 2005;

58(6): 840-6.

Runmarker B, Andersen O. Prognostic factors in a multiple sclerosis incidence cohort with

twenty-five years of follow-up. Brain 1993; 116 ( Pt 1): 117-34.

Scalfari A, Neuhaus A, Daumer M, Ebers GC, Muraro PA. Age and disability accumulation in

multiple sclerosis. Neurology 2011; 77(13): 1246-52.

Scalfari A, Neuhaus A, Degenhardt A, Rice GP, Muraro PA, Daumer M, et al. The natural

history of multiple sclerosis: a geographically based study 10: relapses and long-term

disability. Brain 2010; 133(Pt 7): 1914-29.

Schumacher GA, Beebe G, Kibler RF, Kurland LT, Kurtzke JF, McDowell F, et al. Problems of

Experimental Trials of Therapy in Multiple Sclerosis: Report by the Panel on the

Evaluation of Experimental Trials of Therapy in Multiple Sclerosis. Ann N Y Acad Sci

1965; 122: 552-68.

Simone IL, Carrara D, Tortorella C, Liguori M, Lepore V, Pellegrini F, et al. Course and

prognosis in early-onset MS: comparison with adult-onset forms. Neurology 2002;

59(12): 1922-8.

Tremlett H, Yousefi M, Devonshire V, Rieckmann P, Zhao Y. Impact of multiple sclerosis

relapses on progression diminishes with time. Neurology 2009; 73(20): 1616-23.

Tremlett H, Zhao Y, Joseph J, Devonshire V. Relapses in multiple sclerosis are age- and time-

dependent. J Neurol Neurosurg Psychiatry 2008; 79(12): 1368-74.

Trojano M, Avolio C, Manzari C, Calo A, De Robertis F, Serio G, et al. Multivariate analysis of

predictive factors of multiple sclerosis course with a validated method to assess

clinical events. J Neurol Neurosurg Psychiatry 1995; 58(3): 300-6.

West T, Wyatt M, High A, Bostrom A, Waubant E. Are initial demyelinating event recovery

and time to second event under differential control? Neurology 2006; 67(5): 809-13.

TABLES

Table 1

Characteristics of the studied population with relapsing MS course

females males Cohen’s d

patients, number 8295 3275

age

- at inclusion, years 37.0 ± 11.2 37.2 ± 11.2 0.02

- at censoring, years 44.1 ± 11.7 44.4 ± 11.5 0.02

ethnicity

Caucasian, number (%) 5102 (62%) 1992 (61%)

Hispanic, number (%) 123 (2%) 64 (2%)

Asian, number (%) 269 (3%) 7 (2%)

African, number (%) 44 (0.5%) 12 (0.4%)

other, number (%) 227 (3%) 117 (4%)

not specified, number (%) 2530 (31%) 1023 (31%)

follow-up duration, years* 5.8 (3, 9.5) 5.9 (3.1, 9.7) 0.01

disease duration

- at inclusion, years* 2.9 (0.4, 9.3) 2.5 (0.4, 8.5) 0.11

- at censoring, years* 11.3 ( 6.2, 18.1) 10.8 (6.1, 17.5) 0.04

relapses, number 35725 12637

disease course

- at inclusion

CIS, number (%) 2981 (36%) 1240 (38%)

RRMS, number (%) 4731 (57%) 1721 (53%)

SPMS, number (%) 412 (5%) 182 (6%)

RPMS, number (%) 171 (2%) 132 (4%)

- at censoring

CIS, number (%) 1521 (18%) 669 (20%)

RRMS, number (%) 5488 (66%) 1984 (61%)

SPMS, number (%) 1115 (13%) 490 (15%)

RPMS, number (%) 171 (2%) 132 (4%)

disability

- at inclusion, EDSS* 2 (1.5, 3.5) 2.5 (1.5, 4) 0.08

- at censoring, EDSS* 2.5 (1.5, 5) 3 (1.5, 6) 0.14

received DMDs, number (%) 3956 (48%) 1617 (49%)

*, median (interquartile range); otherwise mean ± standard deviation are shown

CIS, clinically isolated syndrome; DMDs, disease modifying drugs; EDSS, Extended Disability Status

Scale; RPMS, relapsing-progressive multiple sclerosis; RRMS, relapsing-remitting multiple sclerosis;

SPMS, secondary progressive multiple sclerosis

Table 2

Characteristics of the group with primary progressive MS course

patients, number(females, %) 881 (55%)

age

- at inclusion, years 49.4 ± 11.0

- at censoring, years 53.9 ± 11.2

follow-up duration, years* 2.6 (0.5, 7.0)

disease duration

- at inclusion, years* 5.8 (2.6, 11.4)

- at censoring, years* 10.8 (6.2, 17.5)

disability

- at inclusion, EDSS* 5 (3.5, 6.5)

- at censoring, EDSS* 6 (4, 7)

received DMDs, number (%) 190 (22%)

*, median (interquartile range); otherwise mean ± standard deviation are shown

DMDs, disease modifying drugs; EDSS, Extended Disability Status Scale

FIGURE LEGENDS

Figure 1

CONSORT flowchart of patient disposition

CIS, clinically isolated syndrome; MS, multiple sclerosis; PPMS primary progressive MS;

RPMS, relapsing-progressive multiple sclerosis; RRMS, relapsing-remitting multiple sclerosis; SPMS,

secondary progressive multiple sclerosis

Figure 2

Incidence of relapses in women and men with relapse-onset disease course, stratified by

disease duration and chronological age

Patient disposition (A-B) is only shown for patients with complete respective observational

years. Annualised relapse rate unadjusted for other potential confounders (C-D) and

adjusted for age (E) or disease duration (F), treatment and pregnancy using Tweedie models

(E-F). Each patient contributed a single ARR at censoring. P-values for effects of sex, disease

duration and age are shown (Tweedie regression). Cumulative hazard functions for relapses

in female and male sub-populations, adjusted for age (G) or disease duration (H), treatment

and pregnancy using Andersen-Gill models (G-H), with each patient contributing multiple

time-to-relapse entries.

All results are stratified by disease duration (A, C, E, G) and patient age (B, D, F, H). Error

bars and dashed lines show 95% confidence intervals.

Figure 3

Female-to-male ratios stratified by disease duration and chronological age

Unadjusted distributions of the female-to-male (F:M) ratios by disease duration (A) and

patient age (B) are shown for patients with relapse-onset MS (blue marks) stratified by

annual relapse number and patients with primary progressive MS (yellow marks). Each point

represents a F:M ratio among the patients with corresponding number of relapses in a

completed observational year. Numbers of patients per timepoint are indicated. Patients

were allowed to contribute entries at multiple years. Estimated mean is shown as distance-

weighted least squares surface (relapse-onset MS) and red least squares line (primary

progressive MS).The mean estimation does not extend beyond 2 relapses in patients with

more than 20 years of MS duration or 50 years of age, as the amount of the corresponding

data was insufficient.

Figure 4

Female-to-male ratios in relapse- vs. progressive-onset MS

Female-to-male ratios (F:M) by total number of relapses recorded over a 10-year follow-up

period. Patients with relapsing-remitting (n = 2969) and primary progressive MS course (n =

881) with 10-year follow-up data available are compared. Circle sizes indicate number of

patients within each stratum.