Aging with multiple sclerosis- a longitudinal study of physical function, mental health, and memory in two cohorts of US women

Abstract

Background:

It is unknown how individuals with multiple sclerosis (MS) age compared to unaffected peers.

Objectives:

Describe the impact of MS on health and functioning in aging women.

Methods:

We used PF10 physical-functioning-scores (from the SF-36) and other indicators of general, physical, mental health, and memory collected repeatedly over 25 years with self-administered questionnaires among participants in the Nurses’ Health Study (n=121,700 recruited at ages 30–55) and Nurses’ Health Study II (n=116,429 recruited at ages 25–42) to compare women with MS (n=733) to unaffected peers in their health and disability, and describe/quantify the burden of aging with MS.

Results:

Women with MS had a consistently lower PF10 by 0.9–1.7 standard deviations with greater overall variability than unaffected women. PF10-scores gradually decreased with increasing age in both groups, but MS cases declined 3–4-times faster in midlife, while decline was similar in old age. The physical-function-score of 45-year-old women with MS was comparable to that of 75-year-old unaffected women; 70-year-old women with MS scored similarly to 85-year-old unaffected women. MS cases also reported worse health/ more disability throughout adulthood on the other indicators.

Conclusions:

The age-related decline in physical health is accelerated by 15–30 years in MS patients compared to unaffected peers.

Introduction

Survival in persons with multiple sclerosis (MS) has improved over the past decades. Most have a life expectancy similar or moderately reduced compared to unaffected peers and live to an old age.^{1, 2} Still, little is known about the overall effect of MS on aging. Previous studies of senior MS patients were small and cross-sectional,^{3, 4} or described challenges these patients face compared to younger patients.^5–7 Other studies included a subgroup of older patients and drew comparisons to normative population data.^8–10 However, it is unknown how the health of individuals with MS changes throughout the adult lifespan compared to unaffected peers. Addressing this question is important to identify needs an aging MS population has in relation to a normal aging population.¹¹

We therefore compared the general, physical, and mental health, as well as memory through midlife and old age of women aging with MS to that of unaffected women, all drawn from the same study population of two large US cohorts.

Methods

Study design and population

The Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII) are prospective cohort studies that started in 1976 (NHS, 121,700 participants) and 1989 (NHSII, 116,430 participants) of female registered nurses who lived in 11 (NHS) or 14 (NHSII) U.S. states and were 30–55 (NHS) and 25–42 (NHSII) years old at inclusion. They are followed over time with biennial self-administered questionnaires collecting information on lifestyle exposures and health-related outcomes. The overall response rates were at least 85% in both cohorts.¹²

Standard protocol approvals, registrations, and patient consents

The studies were approved by the Institutional Review Boards of the Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health. Return of the baseline questionnaires implied informed consent by the participants.

MS case ascertainment

Women reporting their MS diagnosis on the biennial questionnaires were asked for permission to contact their treating neurologist to review their medical record. Since 2003, our study neurologist (TC) has been reviewing all medical records for confirmation. There were 733 cases of probable or definite MS among women in the cohorts (NHS: 248 cases; NHSII: 485 cases).

Assessment of physical functioning

Women from both cohorts repeatedly reported their physical function during follow-up (Table 1). It was assessed by the physical-functioning component (PF10) of the Short-Form-36 (SF-36), a commonly used valid and reliable tool for health status assessments. The PF10 quantifies on a 3-level Likert scale how limited a person’s physical functioning is as “a lot”, “a little”, or “not limited” across ten activities or activity levels. The PF10 mean summary score ranges from 0–100 for lowest to highest functioning. The SF-36 has been validated in MS patients,^{13, 14} and the PF10 correlates strongly (r=−0.80) with the Expanded Disability Status Scale (EDSS),^{15, 16} the most commonly used clinical measure of MS physical disability. MS patients with an EDSS≤2.0, corresponding to “minimal disability on one functional system”, had significantly lower PF10 scores than the general population, supporting that PF10 is sensitive to small EDSS deteriorations.¹⁴

Table 1.

General, physical, mental health, and memory indicators collected during follow-up in the Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII)

PF10: Physical functioning-10; Mo: month; w: week;

^aNumber of MS cases with at least one post-onset assessment for multilevel indicators: PF10: 673, Hospitalized: 472, Walking pace: 672, Flights of stairs: 664, Balance: 194, Falls: 193, Urine leak: 679, Stool leak liquid: 576, Stool leak solid: 574, Health interferes: 521, Tiredness: 526, Pain: 530, Depressed: 620, Suicidal thoughts: 528, Memory: 433; Data from the year 2000 were included (apart from for PF10).

^bPF10 assesses physical function and is a component of the Short-Form-36 Health Survey: “Does your health limit you in any of the following ten activities and if so, how much (No, not at all; Yes, a little; Yes, a lot)? 1. Vigorous activity (running, heavy lifting etc.), 2. Moderate activity (pushing a vacuum, moving a table etc.), 3. Lifting/carrying groceries, 4. Climbing several flights of stairs or 5. One flight of stairs, 6. Bending/kneeling/stooping, 7. Walking more than a mile, 8. Walking several blocks, 9. Walking one block, 10. Bathing/dressing yourself.”

^cIn the NHSII coded as “yes or no”. In the NHS: “How many times were you hospitalized for 2+ nights during the past year? None, 1 time, 2–3 times, 4+ times.”

^dHealth interfering with social life and activities all, most, some, little, or none of the time.

^eTiredness and suicidal thoughts assessed as: all, most, good bit, some, little, or none of the time.

^f“Do you experience more trouble than usual in the ability to remember things? More trouble 1. remembering things, 2. remembering recent events, 3. remembering short lists, 4. remembering one second to the next, 5. with spoken instructions, 6. following conversations or plots, and 7. findings way on familiar streets?”

The PF10 was assessed six times between 1992 and 2012 in the NHS, and four times between 1993 and 2017 in the NHSII (Table 1). The median number of completed PF10 was four in the NHS and three in the NHSII. Among women without MS, 204,081 (86%) had at least one PF10 scale. Among women with MS, 673 (92%) had at least one and 520 (71%) had at least two PF10 scales completed after the first symptom (post-onset), while 282 had at least one completed prior to the first symptom (pre-onset).

Other indicators of health and disability

Information on other indicators covering domains of general, physical, mental health and memory was repeatedly collected during follow-up. Details about the indicators, categories, and years of reporting, for which we used the data, are listed in Table 1.

As to memory, women reported whether they had more troubles than usual on 7 specific cognitive tasks (Table 1) and were classified as having good, moderate, or poor memory (no, 1–2, or ≥3 complaints). Although this test cannot substitute for an in-depth cognitive assessment, it is a cost-effective screening method for cognitive decline in large population studies, with a substantial amount of evidence supporting its validity. The test is associated with objective cognitive status,¹⁷ predicts cognitive decline over time,¹⁸ and correlates with pathological indicators of Alzheimer’s disease, such as amyloid burden, hippocampal volume, and glucose metabolism.^19–22

Covariates

For women with MS, we had information on the date of first symptom, diagnosis, and disease duration. We imputed date of onset for 40 cases using the cohort-specific median time interval between date of first symptom and diagnosis among the cases with both dates available (6 years in NHS, 2 years in NHSII). The intervals correspond to previously reported delays between MS onset and diagnosis over the past decades.²³ For 661 (90%) MS cases, we had information on the initial disease course, which was reported by the treating neurologist at the time of review of the case’s medical record as relapsing-remitting MS (RRMS, n=508) or progressive MS (P-MS, n=153). Moreover, a subgroup of women with MS from both cohorts (n=382) reported in 2006 (NHSII)/2008 (NHS) whether they had been treated with one of the disease-modifying MS drugs (DMT) available at the time, i.e. interferon beta-1b, −1a, glatiramer acetate, and mitoxantrone (“Have you ever used Avonex, Betaseron, Copaxone, Rebif, or Novantrone?”, allowing a separate answer for each drug). In a small validation study of 37 MS cases, the concordance between the self-reported treatment and pharmacotherapy specified in the medical record was 86.5%. Finally, information on smoking habits (never, past, current smoker) was collected biennially in all women.

Statistical analyses

The statistical analyses were conducted in SAS 9.4 (SAS Institute Inc, Cary, NC) and R 3.6 (R Core Team 2019). For the first set of analyses of physical functioning we used post-onset PF10 scores among the cases. We assessed whether the median PF10 scores differed (Wilcoxon rank-sum test) among women with and without MS within 10-year age groups (≤40, >40–50, >50–60, >60–70, >70 years). For further comparison, we subdivided women with MS according to their initial disease course (Kruskal-Wallis). Next, we examined whether disease status (MS, no MS) adjusted for age (continuously) predicted the PF10 score in a linear mixed model including a random intercept to account for differences between the women. We also assessed whether the change in PF10 over time differed by group including an interaction term of disease status and age into the model. In a separate model restricted to the cases, we examined whether the initial disease course predicted later PF10 scores, adjusting for age and disease duration (continuously). We adjusted all analyses also for smoking status (categorically), as smoking could confound the association between MS and an accelerated decline in physical functioning. Moreover, to illustrate the PF10 trajectory throughout adult life comparing women with and without MS, we fitted a generalized additive model using a penalized cubic spline smooth with knots placed evenly across the age range and reported PF10 and 95% confidence intervals (CI). These analyses were repeated among the cases by initial disease course, smoking status, and separately also in the subgroup with treatment information, restricted to those with a diagnosis in 1993, the year of approval of the first DMT, to 2006–2008, the year of treatment reporting (in total 250 cases of whom 32 reported no treatment).

Subsequently, using both pre-onset and post-onset PF10 scores, we fitted another generalized additive model with smooth term to examine changes in PF10 among women with MS according to the time from the first symptom. As our comparison group we randomly selected among women without MS 5 controls for each case matched on age and year of PF10 assessment.

Finally, to estimate the proportions and 95% CIs of women with difficulties on the other indicators of health and disability through midlife and old age, we fitted a smooth curve using local polynomial regression (LOESS). We estimated how many women with and without MS were out of work due to disability, had been hospitalized in the previous year, were unable to walk, could only walk 2 flights of stairs daily, had balance difficulties, reported one or more falls, suffered a hip fracture, reported an at least weekly urine leak, and an at least monthly stool leak. Moreover, we reported the proportion of women who felt their health interfered all the time with social life, were tired all the time, had severe body pain, were depressed for 2 or more weeks, ever thought about suicide, and had at least 3 specific cognitive complaints of worsening memory.

Results

Women with MS were of similar age as women without MS when first recruited into the NHSII, while they were younger when recruited into the NHS (Table 2). Women with MS were more likely to have ever smoked. Compared to women without MS, women with MS had a shorter follow-up time due to the use of PF10 scales collected post-onset (for the main analysis). As the participants in the NHSII were younger, women with MS had a shorter follow-up from MS onset in this cohort compared to women with MS in the NHS.

Table 2.

Characteristics of women in the study and their mean follow-up time

	Women in the NHS		Women in the NHSII
	with MS	without MS	with MS	without MS
Age at recruitment, mean, yearsa	38.4	42.5	34.6	34.4
Smokingb, %
Never smoker	33.2	43.6	55.5	64.9
Past smoker	52.2	41.8	31.4	24.0
Current smoker	13.8	14.4	13.2	11.0
Age at MS onset, mean, years	40.9	n.a.	39.6	n.a.
Follow-up from first PF10, mean (median), yearsc	16 (18)	17 (20)	16 (20)	19 (23)
Follow-up from MS onset, mean (median), yearsd	30 (29)	n.a.	22 (22)	n.a.

NHS: Nurses’ Health Study; NHSII: Nurses’ Health Study II; n.a.: not applicable; PF10: physical functioning-10

^aRecruitment into the NHS in 1976 and into the NHSII in 1989.

^bSmoking status as assessed at the time of the first (available) PF10 measurement.

^cEstimated from the interval between first and last available PF10 scale in women with MS (post-onset) and women without MS.

^dEstimated from the interval between the first MS symptom and the last available PF10 scale.

Physical functioning

Women with MS had consistently worse physical functioning and showed a greater variability in PF10 scores compared to women without MS (Figure 1). Depending on the age group, MS patients scored about 0.9–1.7 standard deviations lower than their unaffected peers during follow-up. Having MS was associated with a 25.2-points lower PF10 score (95% CI: −26.5, −23.8) independent of age compared to women without MS (RRMS −19.0 points; progressive MS −47.8 points). Women with progressive MS scored 27.5 points lower than those with RRMS (95% CI: −32.3, −22.6), independent of age and disease duration.

Figure 1.

Physical functioning as assessed by the PF10 among women with and without MS by age groups.

The boxplots show the 25^th, 50^th (median), and 75^th percentiles and the range of the distribution of the PF10 physical functioning scores (a component of the SF-36 Health Survey) by 10-year age groups as assessed at repeated occasions in the Nurses’ Health Study and Nurses’ Health Study II. (A) Median PF10 scores were significantly lower among women with MS than unaffected peers within each age group (Wilcoxon, p<0.0001). (B) There was a significant difference in median PF10 scores among women without MS, women with relapsing-remitting MS (RRMS), and women with progressive MS (P-MS). (Kruskal-Wallis, p<0.0001).

The decline in PF10 over time differed in women with and without MS (p for interaction <0.0001). Throughout adult life, women with MS declined on average 2 times faster per additional life year than women without MS (−0.8 vs. −0.4 points/year). The difference was most marked in midlife (Figure 2A, 2B). In their 30s-50s, the average change over 5 years was −6.5 points in women with MS (95% CI: −7.4, −5.6) versus −1.8 points in women without MS (95% CI: −1.83, −1.77). In older age, the decline was more similar among the comparison groups (Figure 2A, 2B). In their 60s-90s, the average worsening over 5 years was −6.9 points in women with MS (95% CI: −8.1, −5.7) versus −8.1 points in women without MS (95% CI: −8.11, −8.02). The results did not change after adjusting for smoking.

Figure 2.

Physical functioning throughout the adult lifespan as assessed by the PF10 in women with and without MS according to age.

Shown are changes in PF10 physical functioning scores (a component of the SF-36 Health Survey) and 95% confidence intervals with increasing age in women with and without MS in the Nurses’ Health Study and Nurses’ Health Study II. (A) PF10 decline is greater in midlife in women with MS than unaffected peers, while decline is more similar in old age. (B) Throughout adult life, women with progressive MS score lower on the PF10 compared to women with relapsing-remitting MS. (C) In a subgroup analysis, women with MS diagnosed in 1993–2008 who reported treatment with disease-modifying drugs of lower potency also scored lower than women without MS.

The subgroup of cases with treatment information also scored lower on the PF10 than women without MS (Figure 2C). Treated women with MS performed worse on the PF10 compared to the small group of untreated women.

There was no difference between the median pre-onset PF10 scores of women who went on to develop MS and the median scores of unaffected peers within 10-year age groups (data not shown). However, a difference became apparent when examining time to the first symptom. Women who went on to develop MS had a lower PF10 score within about 5 years prior to the first symptom compared to the age-and-year matched controls (Figure 3). Excluding cases with imputed onset date yielded similar results (data not shown). Further, PF10 scores declined with longer disease duration among women with MS, most rapidly within the first 10 years (Figure 3). The yearly decline in PF10 (95% CI) slowed down from −3.5 (−5.0, −1.9) points within the first 5 years after symptom onset to −1.0 (−1.4, −0.7) points among cases with a disease duration >30 years.

Figure 3.

Physical functioning as assessed by the PF10 in women with MS according to the years from the first MS symptom and in matched controls

Shown are changes in PF10 physical functioning scores (a component of the SF-36 Health Survey) and 95% confidence intervals in MS cases according to the years from the first MS symptom in the Nurses’ Health Study and Nurses’ Health Study II. Control women without MS were matched to the cases by age and year of PF10 assessment in a 5:1-ratio. (A) Women with MS scored lower on the PF10 several years prior to the first symptom. The longer the disease duration, the lower the PF10 scores, but the decline was most pronounced within the 10 first years after the first MS symptom. (B) These changes were more marked in women with progressive MS (P-MS) compared to those with relapsing MS (RRMS).

Other indicators of health and disability

Women with MS were more likely to report worse health and more disability than their peers and experience more challenges in all the assessed domains (Figure 4). They were more likely out of work because of disability, they more often reported hospitalizations, reduced or inability to walk, balance difficulties, falls, hip fractures, problems with bladder and bowel control, interference of their health with social activities, tiredness, severe pain, having felt depressed for 2 weeks or longer, having thought about suicide, and a worsening on cognitive tasks indicating poor memory.

Figure 4.

Health and disability in midlife and old age among women with MS and unaffected peers.

The panels show data on the general, physical, mental health, and memory of women with and without MS participating in the NHS and NHSII (those marked with * include only data from the NHS). The total on the y-axis includes the missing categories. The age range on the x-axis can further vary due to the year the question was asked during follow-up. The indicator “Health interferes all the time” referred to health interfering with social life. Regarding memory, women reported whether they experienced more troubles than usual and poor memory was defined as 3–7 specific cognitive complaints of worsening memory.

CI: confidence interval; NHS: Nurses’ Health Study; NHSII: Nurses’ Health Study.

Discussion

In this large and rigorously controlled longitudinal study, we found that the age-related decline in physical functioning is accelerated by 15–30 years in women aging with MS compared to women aging without MS. The gap in physical functioning widened during midlife, when women with MS declined more rapidly, suggesting a primarily disease-driven decline. In old age, the gap remained more stable and the decline continued at a similar pace in both groups, indicating a primarily aging-related functional loss. We were also able to capture more comprehensively the burden of aging with MS compared to aging without MS by reporting an array of health-related problems on aspects of general, physical, mental health, and memory.

Our findings in women aging with MS are consistent with findings from previous investigations of older patients compared to younger patients or normative data.^{5–8, 11} Although we know that individuals with MS can accumulate substantial disability throughout life, the disease burden has to date not been systematically quantified and compared in its different facets to aging peers from the same study population. We found that the physical functioning of a 45-year-old woman with MS was comparable to that of a 75-year-old unaffected woman, and that of a 70-year-old woman with MS to that of an 85–90-year-old woman without MS. These differences need to be considered in patient care to assess additional needs of an aging MS patient in a timely manner.¹¹ As physical decline and disability are associated with frailty^{24, 25}, our findings provide some evidence that women aging with MS are at a higher risk of frailty earlier in life compared to women aging without MS. Preventive geriatric measures addressing problems related to polypharmacy, limited mobility, injury and infection risk, challenges with activities of daily living, social isolation, depression, and other aspects, could slow down functional decline, prevent frailty-related comorbidities, and increase resilience.^{11, 26} By giving adequate attention to physical and psychosocial components of health, healthy or successful aging can be achieved in spite of a chronic disease such as MS.²⁷ Further, quality of life and years lost due to MS morbidity are substantial, and identifying modifiable risk factors of progression and accelerated aging as well as protective factors for healthy aging should become a priority in MS research.

In contrast to some previous results,²⁸ our findings do not suggest a synergistic effect of MS and aging on the functional decline.²⁹ This is radiologically supported by a longitudinal study that reported a similar rate of brain volume change in old age in MS patients and healthy controls.³⁰ This phenomenon could be related to senescence of the immune system and chronic low-grade inflammation prevailing in old age over acute neuroinflammation,³¹ which is clinically also supported by a decreasing relapse rate with increasing age.³² However, although survival of MS patients has improved over the decades, we cannot exclude the possibility that the trend we observed is partially influenced by a premature death of patients with an aggressive disease.

In this study, we assessed the overall burden of aging with MS. Both women with and without MS suffer from chronic diseases in midlife and old age and, given that MS is associated with a higher comorbidity burden,³³ the functional decline in women aging with MS could, to some extent, be mediated through comorbidities. Exploring the role of comorbidities and other potential mediators of accelerated decline among women with MS requires additional study.

The PF10 and the other disability indicators discriminated well between women with and without MS. Patient-reported outcomes have been suggested to be more sensitive than typically used measures such as the EDSS in assessing the health of elderly individuals with MS,¹¹ and could routinely be used in a clinical setting to monitor the physical frailty of patients throughout adulthood.

It is difficult to draw definite conclusions from our findings in MS cases with treatment information. During the early years, DMTs were selectively used to treat more severely affected patients. Conversely, the small group of untreated women with MS who seemed to experience less functional decline could represent a group that was at the time considered “benign MS” and thus less likely to be treated, according to previous studies (indication bias).³⁴ The concept of “benign MS” remains controversial, as even patients with little to no initial disability show progression and symptoms such as cognitive impairment or fatigue over time.^{35, 36} The fact that physical function declined faster in the untreated women with MS than in unaffected women, supports this hypothesis.³⁴ Regardless, it is important to note that despite disease-modifying treatment being effective, being treated does not seem to rescue women with MS from a more rapid functional decline compared to unaffected peers, which could be due to continued progression/neurodegeneration or other disabling symptoms such as fatigue, for which there is to date no proven treatment. Future studies should address how functional changes unfold among patients treated with more potent disease-modifying drugs.

The strengths of our study lie in the use of prospectively collected data from large well-characterized cohorts allowing the comparison of women aging with and without MS from the same study population.¹⁰ The long follow-up and repeated measurement of health indicators, not specific for a disease or a certain age, make it possible to assess aging effects throughout the adult lifespan in both groups.

Our study has some limitations. As we are studying women, we cannot draw direct conclusions regarding men aging with MS, especially given that some studies reported sex-differences.³⁷ The PF10 is prone to floor effects and might not discriminate as well among severely disabled MS patients with a different functional decline over time.^{13, 15} We may also underestimate the difference between MS patients and unaffected women, as elderly MS patients with marked disability are more likely to be lost to follow-up if they can no longer respond or move into a nursing home, possibly at a younger age than elderly women without MS. Moreover, we had repeated measures for most, but not all, additional disability indicators that we examined, which could have reduced precision of the results for some health domains. Further, we cannot distinguish primary and secondary progressive MS and the disease course (RRMS or progressive MS) refers to how the MS case was classified at verification of the diagnosis and does not capture how it evolved over time. Another limitation is related to the subgroup of MS patients with treatment information, for whom we lack details on treatment start and duration and use of newer drugs such as natalizumab (approved in 2004 in the US). Finally, we cannot characterize more closely women with MS who reported not being treated.

In conclusion, our findings give a picture of the burden of aging with MS and provide evidence that the age-related decline in physical health of women aging with MS is accelerated by 15–30 years. This information is important in providing care to a growing population of individuals aging with MS and for defining new urgent research priorities.