Advertisement

Prognosis of walking function in multiple sclerosis supported by gait pattern analysis

Open AccessPublished:April 09, 2022DOI:https://doi.org/10.1016/j.msard.2022.103802

      Highlights

      • Deterioration of walking function is poorly characterized in people with MS (PwMS).
      • Walking function was monitored over 4 years in 22 PwMS using clinical and kinematic outcomes.
      • Walking decline was pronounced in PwMS with spastic-paretic gait impairments.
      • Gait pattern analysis provides additional, valuable information on the underyling pathomechanisms of walking impairment.
      • Instrumented gait analysis represents a complementary prognostic tool for walking function in PwMS.

      Abstract

      Background

      Walking impairment is a common and highly disabling symptom in people with MS (PwMS). Ambulatory deterioration is poorly characterized in PwMS and reliable prognosis that may guide clinical decisions is elusive. This study aimed to objectively track the progression of clinical walking performance and kinematic gait patterns in PwMS over 4 years, thereby revealing potential prognostic markers for deterioration of ambulatory function.

      Methods

      Twenty-two PwMS (48.8 ± 9.9 years, 14 females; expanded disability status scale [EDSS]: 4.5 ± 0.9 points) with gait impairments were recruited at the University Hospital Zurich, Switzerland. Gait function was monitored over a period of 4 years using a set of standardized clinical walking tests (timed 25-foot walk [T25FW], 6 min walk test [6MWT], 12-item MS walking scale [MSWS-12]) and comprehensive 3D kinematic gait analysis. Walking decline was assessed in the full patient cohort and in patient sub-groups that were built according to MS type (relapsing-remitting [RRMS], progressive [PMS]) and subjects' pathological gait signature (cluster groups 1–3).

      Results

      In the total cohort (n = 22), we found a significant worsening in the 6MWT (BL vs. 4y: -41.1 m; P = 0.0053), while the performance in the T25FW, MSWS-12 and the EDSS remained unchanged over 4 years. Subjects with PMS (n = 12) showed a significant worsening in the EDSS (BL vs. 4y: +0.6 points; P = 0.0053), which was not observed in participants with RRMS (n = 10). Whereas deterioration of clinical walking function was not different between subjects with RRMS and PMS, we identified differences in clinical walking deterioration between PwMS with varying gait pattern pathologies: Subjects with spastic-paretic gait impairments (cluster 1; n = 9) demonstrated a marked worsening in the T25FW (BL vs. 4y: +2 s; P = 0.0020) and 6MWT (BL vs. 4y: -92.9 m; P < 0.0001) which was not seen in PwMS with an ataxia-like (cluster 2; n = 8) or unstable walking pattern (cluster 3; n = 5). Deterioration of clinical walking performance in cluster 1 was accompanied by a specific worsening of gait deficits that were characteristic of this cluster at baseline, a phenomenon not found in the other sub-groups. Accordingly, aggravation of cluster 1-specific gait impairments over 4 years predicted deterioration of the 6MWT in the total cohort (n = 22) with an accuracy of 90.9% (sensitivity: 90.9%; specificity: 90.9%; Nagelkerkes coefficient of determination R2: 0.721), unveiling key determinants of MS-related walking decline.

      Conclusions

      Our findings highlight the potential of quantitative, functional outcomes for objective tracking of disease progression in PwMS. Gait pattern analysis can provide valuable information on the underlying pathomechanisms of gait deterioration and may represent a complementary prognostic tool for walking function in PwMS.

      Clinical trial

      clinicaltrials.gov, NCT01576354

      Keywords

      1. Introduction

      Multiple sclerosis (MS) is a progressive, chronic inflammatory disease of the central nervous system and the most common cause of non-traumatic, disabling neurological disease in young adults (
      • Alonso A.
      • Hernan M.A.
      Temporal trends in the incidence of multiple sclerosis: a systematic review.
      ;
      • Hirtz D.
      • et al.
      How common are the "common" neurologic disorders?.
      ). Among the multifaceted MS-related deficits, impairment of walking function and mobility are perceived as most devastating by people with MS (PwMS) (
      • Heesen C.
      • et al.
      Patient perception of bodily functions in multiple sclerosis: gait and visual function are the most valuable.
      ;
      • Sutliff M.H.
      Contribution of impaired mobility to patient burden in multiple sclerosis.
      ). Three of four PwMS report gait dysfunction and restricted mobility (
      • Hobart J.C.
      • Riazi A.
      • Lamping D.L.
      • Fitzpatrick R.
      • Thompson A.J.
      Measuring the impact of MS on walking ability: the 12-Item MS Walking Scale (MSWS-12).
      ;
      • Kister I.
      • et al.
      Disability in multiple sclerosis: a reference for patients and clinicians.
      ), major factors limiting quality of life (
      • Creange A.
      • et al.
      Walking capacities in multiple sclerosis measured by global positioning system odometer.
      ). The ability to predict if, when and how walking function of PwMS might deteriorate may allow for targeted interventions to preserve ambulatory function, quality of life and independence.
      Progressive impairments in MS are related to neurodegenerative processes occurring at advanced stages of the disease (
      • Trapp B.D.
      • Nave K.A.
      Multiple sclerosis: an immune or neurodegenerative disorder?.
      ). Several studies have investigated the long-term progression of MS-related disability using outcomes including questionnaires, scales (e.g. expanded disability status scale (EDSS)) or relapse rates (
      • Hauser S.L.
      • et al.
      Ocrelizumab versus Interferon Beta-1a in relapsing multiple sclerosis.
      ;
      • Palace J.
      • et al.
      Assessing the long-term effectiveness of interferon-beta and glatiramer acetate in multiple sclerosis: final 10-year results from the UK multiple sclerosis risk-sharing scheme.
      ;
      • Jokubaitis V.G.
      • et al.
      Predictors of long-term disability accrual in relapse-onset multiple sclerosis.
      ). Despite the well-established relevance of walking function to PwMS, only a few studies have specifically monitored gait deterioration using quantitative ambulatory tests.
      • Paltamaa J.
      • Sarasoja T.
      • Leskinen E.
      • Wikstrom J.
      • Malkia E.
      Measuring deterioration in international classification of functioning domains of people with multiple sclerosis who are ambulatory.
      investigated ambulatory impairments in a cohort of 109 mildly disabled PwMS (median EDSS score: 2.0). They reported a small decline in walking endurance (based on the 6 min walk test (6 MWT)) over 2 years. In contrast,
      • Spain R.I.
      • Mancini M.
      • Horak F.B.
      • Bourdette D.
      Body-worn sensors capture variability, but not decline, of gait and balance measures in multiple sclerosis over 18 months.
      did not detect changes in balance and gait function in mildly impaired PwMS over a period of 18 months using body-worn sensors equipped with accelerometers and goniometers. A recent study analyzed functional deterioration of 57 PwMS with moderate disability over a period of 32 months. The authors reported a moderate decline in maximal walking speed (−6.9% in timed 25-foot walk (T25FW)), which was not correlated with the EDSS (
      • Fritz N.E.
      • et al.
      Longitudinal relationships among posturography and gait measures in multiple sclerosis.
      ). This study provided evidence that walking deterioration can be predicted by instrumented biomechanical parameters (i.e. posturography), highlighting the importance of objective, reliable and sensitive outcome measures to quantify functional changes over time. In a previous study, we showed a modest, but significant walking decline in 28 PwMS over 1 year. Whereas functional decline was not reflected in changes in EDSS, there was a significant reduction in walking endurance (6MWT) that correlated with longitudinal changes in specific kinematic gait parameters (e.g. reduction in knee range of motion) (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ). Interestingly, the progression of walking decline differed in patient sub-groups that were clustered according to their walking patterns, which suggests an additional role of gait pattern analysis in stratifying patients and aiding prognosis of walking function in PwMS.
      In this study, we aimed to track and characterize the evolution of gait function in PwMS over a period of 4 years using both clinical gait measures and treadmill-based kinematic gait analysis. Accurate and objective tracking of walking deterioration is rare in PwMS, likely because it is time demanding and usually conflicts with the limited resources in everyday clinical practice. However, comprehensive monitoring of walking function using quantitative, responsive and reliable outcome measures seems crucial to enhance the basic knowledge on the progression of MS-related gait dysfunction, to improve clinical decision making (i.e. treatment management) and to advance prognostic assessment in MS.

      2. Methods

      2.1 Participants

      Twenty-two PwMS (with relapsing-remitting, primary- or secondary-progressive MS) and clinical walking impairment were assessed at the University Hospital Zurich in the framework of the FAMPKIN study (clinicaltrials.gov, NCT01576354) (
      • Zorner B.
      • et al.
      Prolonged-release fampridine in multiple sclerosis: improved ambulation effected by changes in walking pattern.
      ). For inclusion in this 4-year sub-study, patients needed to be able to achieve at least 50 m in the 6MWT and to walk unassisted, without handrail support, on an instrumented treadmill at each measurement timepoint. PwMS were allowed to use their usual walking aids (i.e. walking sticks, crutches, walking frames) for the clinical walking tests. Normative data of gait patterns from age-matched, healthy control participants originate from a separate two-center study (University Hospital Zurich and Balgrist University Hospital) that was performed independently with the aim to provide comparative gait pattern data for future clinical studies (
      • Killeen T.
      • et al.
      Minimum toe clearance: probing the neural control of locomotion.
      ,
      • Killeen T.
      • et al.
      Increasing cognitive load attenuates right arm swing in healthy human walking.
      ). These participants were thoroughly screened for neurological and orthopedic impairments before inclusion. Both studies were approved by the local ethics committee (Zurich cantonal ethics committee; KEK-2011–0445, KEK-2014–0004) and were conducted according to the guidelines of the Declaration of Helsinki and Good Clinical Practice. Written, informed consent was obtained from all participants.

      2.2 Study design and experimental procedures

      Study assessments were performed yearly over 5 years for all PwMS included in the extension study of the FAPMKIN trial. At baseline, all participants underwent a neurological examination including the EDSS (
      • Kurtzke J.F.
      Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).
      ). Additional EDSS assessments were performed 2 and 5 years after baseline. Assessments of clinical walking function (i.e.T25FW (
      • Motl R.W.
      • et al.
      Validity of the timed 25-foot walk as an ambulatory performance outcome measure for multiple sclerosis.
      ), 6MWT (
      • Goldman M.D.
      • Marrie R.A.
      • Cohen J.A.
      Evaluation of the six-minute walk in multiple sclerosis subjects and healthy controls.
      ), 12-item MS walking scale (MSWS-12)) and instrumented gait analysis were conducted yearly up to 4 years after baseline. All patients in this study were treated with prolonged-release (PR-) fampridine within the framework of the FAMPKIN study (
      • Zorner B.
      • et al.
      Prolonged-release fampridine in multiple sclerosis: improved ambulation effected by changes in walking pattern.
      ;
      • Filli L.
      • et al.
      Monitoring long-term efficacy of fampridine in gait-impaired patients with multiple sclerosis.
      ). However, all assessments and results presented here were performed during placebo treatment (year 1 to 5) or drug holidays (baseline), where subjects were ≥14 days without fampridine treatment.
      Instrumented gait analysis was performed in the gait laboratories of the University Hospital Zurich and Balgrist University Hospital equipped with infrared cameras using Nexus 2.7 (Vicon, Oxford, UK) motion capture software at a sampling rate of 200 Hz. Thirty-one reflective markers (14 mm diameter) were placed on the skin overlying anatomical landmarks of the abdomen, upper and lower extremities, allowing for 3D reconstruction of movements using a standard, full-body gait model (Plug-in-Gait, Vicon, UK). Kinematic gait analysis was performed while participants walked barefoot for at least 30 s per trial on an instrumented treadmill (120 Hz, FDM-T, Zebris Medical GmbH, Germany) without holding the handrails. The individual maximal walking speed was determined based on the T25FW for each patient and half of this speed (vmax50%) was used for instrumented gait assessments on the treadmill (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ;
      • Killeen T.
      • et al.
      Increasing cognitive load attenuates right arm swing in healthy human walking.
      ). Consequently, PwMS with varying degrees of gait dysfunction walked at a speed that was proportional to their walking ability and which was generally perceived as a comfortable pace by subjects. In parallel, we analyzed kinematic gait patterns of 20 healthy participants that walked at a series of different speeds ranging from 0.4 to 3.1 km/h in intervals of 0.3 km/h. Hence, each subject with MS was compared to 20 healthy subjects walking at the same speed (with a resolution of 0.1 km/h). This procedure enabled us to perform precise assessments of patients' pathological gait patterns by eliminating confounding effects resulting from different walking speeds in patients vs. healthy subjects. Hence, we were able to compute patients' gait parameters as z-scores relative to normative, speed-controlled gait data (see section "Statistical analysis"). Healthy controls and PwMS underwent familiarization with treadmill walking prior to recordings (
      • Meyer C.
      • et al.
      Familiarization with treadmill walking: how much is enough?.
      ).

      2.3 Data analysis

      Data were processed in Nexus 2.7 and walking parameters were extracted using ProCalc (Vicon, Oxford, UK). Heel strike and toe off events were set according to the definition by
      • Zeni J.A.
      • Richards J.G.
      • Higginson J.S
      Two simple methods for determining gait events during treadmill and overground walking using kinematic data.
      . The more- (MI) and less-impaired (LI) leg of each patient was defined based on the neurological examination (
      • Zorner B.
      • et al.
      Prolonged-release fampridine in multiple sclerosis: improved ambulation effected by changes in walking pattern.
      ). Twenty-eight gait parameters were used to objectively quantify the walking patterns of PwMS and healthy participants. Gait parameters were assessed per step cycle and comprised different functional domains including limb excursion, leg joint range of motion (ROM), left-right asymmetry of leg joint excursion, measures of balance, inter-limb coordination, gait phases and gait variability (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ).
      In a previous study, a hierarchical cluster analysis was performed to group differential walking patterns in a larger population of PwMS (n = 37) (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ). Here, we applied the same cluster definitions (i.e. if a subject belonged to cluster 1 in the previous study, she/he was also allocated to cluster 1 in the present study) and monitored walking function in PwMS assigned to one of these three cluster groups over time. There is no overlap of patients' gait analysis data between the present study and
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      as kinematic assessments were performed at different time points. In addition, walking function in subjects with relapsing-remitting (RRMS) and primary or secondary progressive MS (PMS) was analyzed separately.

      2.4 Statistical analysis

      Statistical analysis was conducted using SPSS (V23, SPSS Inc., Armonk, NY, USA) and Matlab (Mathworks, Inc., Natick, MA, USA). The level of significance was set at 0.05 for all statistical tests. All tests were adjusted for multiple comparisons via post-hoc Bonferroni correction. Normality of data was checked using the Kolmogorov–Smirnov test. Comparison of demographic factors between cluster groups was performed by 1-way ANOVA (continuous data) and the Fisher's Exact Test (categorical data). Demographic differences between RRMS and PMS were evaluated by two-tailed, unpaired t-tests (continuous data) and Pearson's Χ2 test (categorical data). Changes over time were assessed by repeated measures 1-way ANOVA (for normally distributed data) and the Friedman test (non-normally distributed data). To identify changes in clinical walking test data over time in the different sub-groups, repeated measures 2-way ANOVA with the within-subject factor "time" and the between-subject factor "sub-group" was used.
      To assess changes in kinematic gait patterns over time in the specific sub-groups, two-tailed, paired t-tests were conducted. Patients' walking parameters were expressed as z-scores based on the normative values of the 20 healthy volunteers. The z-score-based analysis of gait impairments allowed for a direct comparison of different patient groups and gait parameters.
      Hierarchical cluster analysis (Ward's method with z-standardized variables and Euclidian distance as measure of dissimilarity) was conducted to identify patient sub-groups with different walking deficits (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ;
      • Watelain E.
      • Barbier F.
      • Allard P.
      • Thevenon A.
      • Angue J.C.
      Gait pattern classification of healthy elderly men based on biomechanical data.
      ). Walking data generated from the cluster analysis were examined by principal component analysis (PCA). Principal components were calculated using eigenvalue decomposition on the centered and standardized dataset. The number of relevant principal components was determined with a Kaiser-Guttman stopping criteria. Binary logistic regression (Wald forward stepwise) was used to identify kinematic correlates (28 kinematic gait parameters) of patients' deterioration in the 6MWT. P values were set at <0.05 for stepwise entry of a given variable in the model (no specific assumptions were made). The dependent variable was set as deterioration in 6MWT performance over 4 years (i.e. 4y vs. baseline) resulting in a patient group with mild progression (n = 11) and a group with pronounced deterioration (n = 11). Model fit at each step was assessed by Nagelkerke R square and −2 Log Likelihood values

      3. Results

      3.1 Participants

      Walking function was analyzed in 22 PwMS (48.8 ± 9.9 years, 14 females; EDSS: 4.5 ± 0.9 points) over a period of 4 years (Table 1). Patients with a progressive type of MS (PMS) were older than subjects with relapsing-remitting MS (RRMS; P = 0.0497; two-tailed, unpaired t-test). Subjects with RRMS were more frequently treated with disease modifying therapies than those with PMS (P = 0.0060; Pearson's Χ2 test; Table 1). There were no changes in diagnosed MS type during the observational period. In patients with RRMS, there were 3 relapses in 2 patients with all relapses occurring between baseline and the 1-year visit (mean duration between relapses and following assessment visits: 194 ± 120 days). Demographic factors and MS therapies were not different between the three patient sub-groups that were determined by their specific kinematic gait patterns (cluster groups 1–3). Cluster group 1 was the only patient sub-group showing a difference in the EDSS functional systems with pyramidal impairments higher than cerebellar (P = 0.0199; repeated-measures 1-way ANOVA with Bonferroni's post-hoc test) and sensory deficits (P = 0.0090). Three patients switched their disease modifying treatment (from no treatment to fingolimod (between baseline (BL) and 1y), from natalizumab to dimethylfumarate (between 1y and 2y), from interferon to rituximab (between 3y and 4y)).
      Table 1Demographic and clinical data for the study population at baseline. Demographic factors are shown for the overall patient cohort (n = 22), as well as for patient sub-groups that were allocated according to MS type (RRMS, PMS) and their kinematic gait patterns (clusters 1–3). Abbreviations: DMT: disease modifying therapy; EDSS: expanded disability status scale; f: female; m: male; PMS: progressive multiple sclerosis; PPMS: primary progressive multiple sclerosis; pts: points; RRMS: relapsing remitting multiple sclerosis; SD: standard deviation; SPMS: secondary progressive multiple sclerosis.
      Total cohort

      (n = 22)
      RRMS

      (n = 10)
      PMS

      (n = 12)
      Cluster 1

      (n = 9)
      Cluster 2

      (n = 8)
      Cluster 3

      (n = 5)
      Age [years ± SD]48.8 ± 9.944.2 ± 9.352.6 ± 9.448.3 ± 10.549.0 ± 8.749.2 ± 10.5
      Sex [m / f]8 / 142 / 86 / 63 / 63 / 52 / 3
      MS Type

      (RRMS / PPMS / SPMS)
      10 / 2 / 1010 / 0 / 00 / 2 / 105 / 0 / 44 / 1 / 31 / 1 / 3
      Disease duration since diagnosis [years ± SD]11.6 ± 6.89.1 ± 4.613.7 ± 7.914.8 ± 8.09.4 ± 4.79.3 ± 4.4
      EDSS [pts ± SD]

      Pyramidal FS [pts ± SD]

      Cerebellar FS [pts ± SD]

      Sensory FS [pts ± SD]
      4.5 ± 0.9

      2.3 ± 0.8

      2.3 ± 0.9

      1.8 ± 1.1
      4.1 ± 0.2

      2.4 ± 0.7

      2.1 ± 0.9

      1.8 ± 1.2
      4.8 ± 1.2

      2.2 ± 0.9

      2.5 ± 0.9

      1.8 ± 1.0
      4.3 ± 0.7

      2.7 ± 0.5

      1.7 ± 0.9

      1.6 ± 0.7
      4.2 ± 0.8

      1.8 ± 1.0

      2.8 ± 0.7

      2.1 ± 1.4
      5.2 ± 1.2

      2.4 ± 0.5

      2.8 ± 0.4

      1.6 ± 1.1
      DMTinterferon312111
      fingolimod110100
      natalizumab660321
      no treatment12210453

      3.2 Progression of walking impairment in full patient population over 4 years

      Walking function in PwMS (n = 22) was clinically evaluated with the T25FW, 6MWT and MSWS-12 (Fig. 1). For the total cohort, maximal walking speed (i.e. T25FW) and self-perceived walking capacity (i.e. MSWS-12) did not change over 4 years (Fig. 1A). In contrast, walking endurance as assessed with the 6MWT showed a significant deterioration (6MWT: P = 0.0038; repeated-measures 1-way ANOVA), reducing by 46 m (i.e. −10.9%, CI −17.7% to −4.8%) at year 3 (P = 0.0015; Dunnett's post-hoc test) and by 41.1 m (i.e. −9.7%, CI −20.1% to 3.7%) at year 4 (P = 0.0053; Dunnett's post-hoc test) compared to baseline. The EDSS was only assessed in 20 PwMS, as two patients were not examined at the last study visit, and remained unchanged over time (Fig. 1A).
      Fig 1
      Fig. 1Change of clinical walking function and neurological disability over 4 years. Progression of walking impairment and neurological disability were assessed in the full patient cohort (A), in patients with RRMS and PMS (B) and in different cluster groups that were determined according to kinematic gait signatures (C). Significant changes over time are indicated with asteriks and highlighted in the color of the respective sub-group. Statistical analysis was performed by repeated measures 2-way ANOVA (with the independent factors time and cluster groups) followed by post-hoc analysis (Dunnett's multiple comparison test). *: P < 0.05; **: P < 0.01; ***: P < 0.001. Abbreviations: BL: baseline; EDSS: expanded disability status scale; MSWS-12: 12-item multiple sclerosis walking scale; PMS: progressive multiple sclerosis; RRMS: relapsing remitting multiple sclerosis; T25FW: timed 25-foot walk; 6MW: 6 min walk test; y: year.
      For kinematic gait analysis, patients' walking patterns were compared to those of healthy participants walking at matched speeds (n = 20; age: 48.8 ± 10.1 years, 12 female). Twenty-eight key gait parameters were assessed to generate comprehensive, individual gait profiles (Fig. 2) and were monitored over 4 years. Gait impairments of the full cohort (n = 22) were mainly characterized by enhanced left-right asymmetry of knee ROM, disturbed interlimb coordination (i.e. phase coupling) and pronounced variability of step lengths (Fig. 2A). Enhanced trunk sway (i.e. C7 trajectory) was the only significant change in the gait pattern over 4 years in the overall patient cohort (P = 0.0007; two-tailed, paired t-test).
      Fig 2
      Fig. 2Kinematic gait profiles of patient sub-groups over time. Comprehensive gait patterns were assessed at baseline (black line) and at 4 years (red line) for the overall patient cohort (A), for subjects with RRMS and PMS (B) and for the three cluster sub-groups (C). Patient values were calculated as z-scores based on normative gait data from 20 healthy volunteers. Data in the spider plots represent mean values ± standard error of the mean. Z-score values between −2 to +2 (yellow circle) were defined as the physiological range. Walking outcomes lying outside the yellow circle were considered to be different from healthy controls (i.e. pathological). Gait parameters were assessed for domains of limb excursion (orange), leg joint ROM (light blue), gait asymmetry (purple), balance (green), inter-limb coordination (gray), gait phases (bright green), variability (dark blue). Gait parameters highlighted in green color indicate significant changes of the respective parameters over 4 years. Abbreviations: AP: antero-posterior; ASI: asymmetry index; CoM: center of mass; CoV: coefficient of variation; il. coord.: inter-limb coordination; LI: less-impaired leg; MI: more-impaired leg; ML: medio-lateral; PD: phase dispersion; PMS: progressive multiple sclerosis; ROM: range of motion; RRMS: relapsing remitting multiple sclerosis; traj.: trajectory.

      3.3 Evolution of gait function in patients with relapsing-remitting and progressive MS

      Based on the different forms of MS, we separately analyzed the course of walking disability in subjects with RRMS (n = 10) or PMS (n = 12; PPMS n = 2, SPMS n = 10; Fig. 1B). Two-way ANOVA (with independent factors "time" and "MS type") revealed a significant main effect of MS type on the T25FW (F (1, 20) = 6.474; p = 0.0193) and EDSS (F (1, 18) = 7.613; p = 0.0129), indicating worse walking function in subjects with PMS across all time points. For the 6MWT, there was a significant main effect of time (F (4, 80) = 4.657; p = 0.0020) indicating a gradual reduction of walking endurance across both groups over 4 years. A significant interaction effect (group by time) was only found for the EDSS (F (2, 36) = 4.571; P = 0.0170). Here, we found a significant increase in neurological disability in PMS (EDSS: BL vs. 5y: P = 0.0053; Fig. 1B). Given the absence of significant interaction effects (group by time) for the T25FW, 6MWT and MSWS-12, we did not conduct post-hoc tests to assess changes in walking function over time for these outcome measures.
      Individual gait profiles of subjects with RRMS and PMS were similar at baseline, both mainly showing reduced knee ROM, enhanced left-right asymmetry of knee movements, and increased variability of step length compared to healthy controls (Fig. 2B). There were no significant changes in the gait pattern over 4 years in either patient sub-group (Fig. 2B).

      3.4 Walking deterioration in patient sub-groups with different kinematic gait signatures

      In a previous study, we performed hierarchical cluster analysis to disentangle different walking patterns in 37 PwMS (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ). In the present study, we applied the same cluster allocations to our sub-population of PwMS (n = 22) who remained in the study over 4 years and that were all part of the original population (cluster 1: n = 9; cluster 2; n = 8; cluster 3: n = 5). To validate the cluster characteristics of the patient sub-groups, the three cluster groups were plotted in the principal component (PC) space (Supplementary Fig. 1A). Group shapes and separation, as well as group-specific mean PC scores were highly similar to the original cluster groups (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ). As in the former study, cluster group 1 was mainly characterized by a spastic-paretic gait pattern, group 2 by an ataxia-like walking pattern and subjects in cluster group 3 primarily presented with an unstable gait (Supplementary Fig. 1B).
      Monitoring of outcome in clinical walking tests over 4 years demonstrated differences in the extent of worsening between the three cluster groups (Fig. 1C). Two-way ANOVA (with independent factors "time" and "cluster") revealed a significant main effect of time on the 6MWT (F (4, 76) = 3.199; P = 0.0175) suggesting an overall decline in walking endurance over time across groups. Whereas no main effects were found for clusters, we found significant interaction effects (cluster by time) for the T25FW (F (8, 76) = 2.113; P = 0.0446) and 6MWT (F (8, 76) = 2.773; P = 0.0096). Hence, post-hoc analysis assessing changes in walking function over time was conducted in the T25FW and 6MWT only. There was a significant deterioration in the T25FW for cluster group 1 (BL vs. 4y: P = 0.0020; Dunnett's multiple comparison test), but not for cluster groups 2 and 3 (Fig. 1C). Similarly, post-hoc analyses demonstrated a specific worsening in the 6MWT in cluster group 1 (BL vs. 2y: P = 0.0437; BL vs. 3y: P = 0.0007; BL vs. 4y: P < 0.0001), but not for patients assigned to the other two cluster groups (Fig. 1C). Although self-reported walking function (MSWS-12) and the EDSS showed a descriptive worsening in cluster group 1 that was greater than in the other clusters, there were no statistically significant differences in the amount of worsening observed across the clusters (Fig. 1C).
      Whereas cluster group 1 primarily presented with an impaired ROM of the knee and ankle joint and an enhanced left-right asymmetry of knee ROM at baseline (Fig. 2C, Supplementary Fig. 1B), the principal gait impairments of cluster group 2 were characterized by enhanced toe height of the LI leg, as well as an enhanced variability of trunk movements and step length (MI leg). The pathological hallmarks of cluster group 3 included enhanced trunk sway (i.e. C7 trajectory) and medio-lateral excursions of the center of mass (CoM ML) as well as increased step width (Fig. 2C, Supplementary Fig. 1B) (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ).
      Longitudinal analysis revealed no significant changes for all kinematic gait parameters in either of the three cluster groups. However, group-specific changes in gait parameters were found when the analysis was restricted to the above-mentioned three cardinal gait pathologies of each cluster group. Cluster group 1 demonstrated a specific worsening of its principal pathological hallmarks over time (Fig. 3). In contrast, cluster group 2 showed a slight improvement of its group-specific gait impairments over 4 years. There were no changes regarding the group-specific gait abnormalities in cluster group 3. Cumulated changes in group-specific gait impairments were significantly different between groups (P = 0.0475; one-way ANOVA). Post-hoc analysis demonstrated that deterioration of group-specific gait impairments was more pronounced in cluster group 1 than in cluster group 2 (P = 0.0133; Fig. 3).
      Fig 3
      Fig. 3Changes of cluster group-specific gait impairments over time. Cardinal gait pathologies in cluster group 1–3 at baseline (A). Changes in these cluster group-specific gait hallmarks were monitored over 4 years (B). Values represent mean summated z-scores over the three group-specific gait hallmarks for each cluster group. Positive values represent improvements, negative values deterioration over time. Cluster group 1 shows focal worsening of group-specific gait pathology. Abbreviations: ASI: asymmetry index; CoM: center of mass; CoV: coefficient of variation; LI: less-impaired leg; MI: more-impaired leg; ML: medio-lateral; ROM: range of motion.

      3.5 Kinematic predictors of decline in clinical walking function

      Changes of single gait parameters associated with reduced performance in clinical walking tests (i.e. T25FW and 6MWT) over 4 years were investigated using binary logistic regression analysis. The dependent variable was decline in 6MWT performance over 4 years (i.e. BL vs. 4y) resulting in a patient group (n = 11) with mildly progressing walking deficits and a group of patients (n = 11) with a more pronounced deterioration of walking function. In the latter group, worsening in the 6MWT was significantly higher than in the group with mild deterioration in walking endurance (P < 0.0001; two-tailed, unpaired T-test). Regression analysis revealed that subjects with a more pronounced deterioration of walking function showed greater reductions in ankle and knee joint ROM over 4 years. Decline in ankle ROM of the LI leg was identified as strongest single predictor for decline in walking function with an accuracy of 77.3% (sensitivity: 81.8%; specificity: 72.7; Nagelkerkes coefficient of determination R2: 0.466; −2 Log Likelihood: 21.028). Combining ankle ROM (LI) and knee ROM (MI) predicted poorer walking outcome with an accuracy of 90.9% (sensitivity: 90.9%; specificity: 90.9; Nagelkerkes coefficient of determination R2: 0.721; −2 Log Likelihood: 13.366).

      4. Discussion

      Walking function was monitored in 22 PwMS over 4 years using a comprehensive set of quantitative ambulatory outcomes. Gait deterioration was pronounced in certain patient sub-groups defined by MS type (i.e. RRMS vs. PMS) or specific kinematic gait patterns. Objective clustering of patients based on kinematic gait profiles revealed a patient sub-group (cluster group 1) with marked deterioration in walking function, pointing towards the potential of advanced functional outcomes as instrument for improved patient stratification and prognosis. Gait pattern analysis showed that impaired knee and ankle joint movements are significant determinants of walking deterioration in PwMS, revealing potential key functional targets for rehabilitative interventions in MS.
      The overall patient cohort demonstrated reduced walking endurance (i.e. 6MWT: −9.7%), but unchanged maximal walking speed (T25FW), self-perceived walking function (MSWS-12) and neurological disability (EDSS) over 4 years. Moderate progression of walking deficits over time conforms to earlier studies (
      • Paltamaa J.
      • Sarasoja T.
      • Leskinen E.
      • Wikstrom J.
      • Malkia E.
      Measuring deterioration in international classification of functioning domains of people with multiple sclerosis who are ambulatory.
      ;
      • Spain R.I.
      • Mancini M.
      • Horak F.B.
      • Bourdette D.
      Body-worn sensors capture variability, but not decline, of gait and balance measures in multiple sclerosis over 18 months.
      ;
      • Fritz N.E.
      • et al.
      Longitudinal relationships among posturography and gait measures in multiple sclerosis.
      ) and might be partially attributed to the effective treatment options now available to attenuate disease progression. Progressive walking impairments were better detected by long-distance (i.e. 6MWT) than short-distance walking test (i.e. T25FW) or the EDSS, confirming that the 6MWT is more responsive in detecting ambulatory changes in mildly and moderately gait-impaired PwMS (
      • Filli L.
      • et al.
      Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
      ,
      • Filli L.
      • et al.
      Predicting responsiveness to fampridine in gait-impaired patients with multiple sclerosis.
      ;
      • Gijbels D.
      • et al.
      Which walking capacity tests to use in multiple sclerosis? A multicentre study providing the basis for a core set.
      ;
      • Goldman M.D.
      • Motl R.W.
      • Rudick R.A.
      Possible clinical outcome measures for clinical trials in patients with multiple sclerosis.
      ;
      • Bigland-Ritchie B.
      • Cafarelli E.
      • Vollestad N.K.
      Fatigue of submaximal static contractions.
      ).
      Clinical walking impairment was more pronounced in subjects with PMS than RRMS. Although deterioration of walking decline is typically reported to be stronger in subjects with PMS (
      • Confavreux C.
      • Vukusic S.
      • Moreau T.
      • Adeleine P.
      Relapses and progression of disability in multiple sclerosis.
      ; ), we did not find significant group differences with regards to longitudinal changes in the T25FW, 6MWT or MSWS-12 over time. However, we observed a significant worsening of the EDSS in subjects with PMS over time, which was not present in people with RRMS.
      Progression of walking impairments markedly differed between sub-groups with different pathological gait signatures: subjects with a spastic-paretic gait and pyramidal deficits (cluster group 1) showed the highest degree of gait deterioration over 4 years as demonstrated by objective outcome measures (T25FW, 6MWT). Subjects with ataxia-like (cluster group 2) and unstable gait patterns (cluster group 3) revealed no appreciable decline in walking function over the same period. These results suggest that gait pattern analysis provides additional, unique information for improved prognosis of walking impairment. The gait of subjects with a spastic-paretic pattern was characterized by diminished excursions in knee and ankle joints as well as increased left-right asymmetry of knee joint movements. Ambulatory deficits were previously reported to be more severe in patients with impaired knee and ankle joint control (
      • Chung L.H.
      • Remelius J.G.
      • Van Emmerik R.E.
      • Kent-Braun J.A.
      Leg power asymmetry and postural control in women with multiple sclerosis.
      ;
      • Kalron A.
      • Givon U.
      Gait characteristics according to pyramidal, sensory and cerebellar EDSS subcategories in people with multiple sclerosis.
      ). In keeping with these findings, we demonstrated that progressive deficits in ankle and knee joint control were strongly associated with a more pronounced decline in walking performance (i.e. 6MWT and T25FW) using logistic regression analysis. Given the small number of participants in the individual cluster groups, the results of gait classification need to be interpreted carefully.
      For all sub-groups (i.e. RMMS, PMS, cluster 1–3), there were no statistically significant changes in gait pattern over 4 years when considering the full set of kinematic gait parameters. However, we found significant inter-cluster differences when only the principal kinematic hallmarks of each cluster group were considered. Patients in cluster 1, characterized by a spastic-paretic gait pattern, demonstrated further worsening of their key baseline locomotor impairments, while cardinal pathologic gait parameters of the other two clusters did not show further decline over time. Worsening of the main pathologic kinematic hallmarks of a given gait pattern maybe the driver of functional deterioration in clinical walking tests.
      An additional finding of our gait pattern analysis was that the gait profiles of cluster group 2 (ataxia-like gait) and 3 (unstable gait pattern) seemed to become more similar over time. This convergence of the gait patterns might be explained by new MS lesions that occur in previously unaffected areas of the nervous system, resulting in additional, overlapping deficits that are less characteristic of the particular cluster phenotype.
      A key constraint of this study is the limited number of participants; a direct result of the comprehensive study design required to accurately track patients' walking function over 4 years, potentially limiting generalizability of these findings. Moreover, study assessments were performed under double-blinded placebo treatment. Whereas direct effects by PR-fampridine are unlikely (all subjects stopped their PR-fampridine treatment ≥14 days before the assessments), the double-blinded design might have resulted in some placebo effects regarding the walking outcomes.
      In summary, our results indicate that PwMS with a spastic-paretic gait pattern have an unfavorable prognosis with regard to clinical walking function. Pronounced deterioration of walking function in this sub-group might be due to the fact that either the restoration (e.g. physical therapy) or compensation (e.g. walking aids) of walking dysfunction is particularly challenging in these patients. Regarding the latter, it might by hypothesized that walking aids are effective in improving gait function in subjects with an ataxia-like (cluster 2) or unstable gait pattern (cluster 3) through assistance of balance, while they may be of less benefit for patients with impaired distal leg joint control (cluster 1). Functional deficits that were characteristic for subjects with a spastic-paretic gait pattern (cluster group 1) may represent key therapeutic targets for rehabilitative interventions aiming for improved clinical gait function in PwMS. Our findings also highlight the importance of monitoring patients' walking function by quantitative and objective outcome measures. Analysis of the gait pattern in addition to clinical, electrophysiological and imaging measures may improve prognosis, especially for the progression of gait impairments (
      • Fritz N.E.
      • et al.
      Longitudinal relationships among posturography and gait measures in multiple sclerosis.
      ).

      Statements and declarations

      The trial was supported by the Betty and David Koetser Foundation, the Clinical Research Priority Program (CRPP) “NeuroRehab” of the University of Zurich, the Swiss MS Society and Biogen.

      CRediT authorship contribution statement

      Björn Zörner: Conceptualization, Investigation, Funding acquisition, Formal analysis, Visualization, Data curation, Writing – original draft. Pascal Hostettler: Data curation, Writing – review & editing, Project administration. Christian Meyer: Data curation, Formal analysis. Tim Killeen: Writing – review & editing. Pauline Gut: Formal analysis. Michael Linnebank: Conceptualization, Funding acquisition, Writing – review & editing, Supervision. Michael Weller: Supervision, Writing – review & editing. Dominik Straumann: Supervision, Writing – review & editing. Linard Filli: Supervision, Funding acquisition, Formal analysis, Visualization, Data curation, Writing – original draft.

      Declaration of Competing Interest

      PH, CM, TK, PG, DS and LF declare no conflict of interest. BZ, ML and MW report grants and honoraria from Biogen during the conduct of the study.

      Acknowledgments

      We thank the subjects who participated in this study and the institutions supporting the trial (Betty and David Koetser Foundation, the Clinical Research Priority Program (CRPP) “NeuroRehab” of the University of Zurich, the Swiss MS Society and Biogen). This work was supported by the Swiss Center for clinical Movement Analysis (SCMA), Balgrist Campus AG, Zürich, Switzerland.

      References

        • Alonso A.
        • Hernan M.A.
        Temporal trends in the incidence of multiple sclerosis: a systematic review.
        Neurology. 2008; 71: 129-135https://doi.org/10.1212/01.wnl.0000316802.35974.34
        • Bigland-Ritchie B.
        • Cafarelli E.
        • Vollestad N.K.
        Fatigue of submaximal static contractions.
        Acta Physiol. Scand. Suppl. 1986; 556: 137-148
        • Chung L.H.
        • Remelius J.G.
        • Van Emmerik R.E.
        • Kent-Braun J.A.
        Leg power asymmetry and postural control in women with multiple sclerosis.
        Med. Sci. Sports Exerc. 2008; 40: 1717-1724https://doi.org/10.1249/MSS.0b013e31817e32a3
        • Compston A.
        • Coles A.
        Multiple sclerosis.
        Lancet. 2008; 372: 1502-1517https://doi.org/10.1016/S0140-6736(08)61620-7
        • Confavreux C.
        • Vukusic S.
        • Moreau T.
        • Adeleine P.
        Relapses and progression of disability in multiple sclerosis.
        N. Engl. J. Med. 2000; 343: 1430-1438https://doi.org/10.1056/NEJM200011163432001
        • Creange A.
        • et al.
        Walking capacities in multiple sclerosis measured by global positioning system odometer.
        Mult. Scler. 2007; 13: 220-223https://doi.org/10.1177/1352458506070667
        • Filli L.
        • et al.
        Monitoring long-term efficacy of fampridine in gait-impaired patients with multiple sclerosis.
        Neurology. 2017; 88: 832-841https://doi.org/10.1212/WNL.0000000000003656
        • Filli L.
        • et al.
        Profiling walking dysfunction in multiple sclerosis: characterisation, classification and progression over time.
        Sci. Rep. 2018; 8: 4984https://doi.org/10.1038/s41598-018-22676-0
        • Filli L.
        • et al.
        Predicting responsiveness to fampridine in gait-impaired patients with multiple sclerosis.
        Eur. J. Neurol. 2019; 26: 281-289https://doi.org/10.1111/ene.13805
        • Fritz N.E.
        • et al.
        Longitudinal relationships among posturography and gait measures in multiple sclerosis.
        Neurology. 2015; 84: 2048-2056https://doi.org/10.1212/WNL.0000000000001580
        • Gijbels D.
        • et al.
        Which walking capacity tests to use in multiple sclerosis? A multicentre study providing the basis for a core set.
        Mult. Scler. 2012; 18: 364-371https://doi.org/10.1177/1352458511420598
        • Goldman M.D.
        • Marrie R.A.
        • Cohen J.A.
        Evaluation of the six-minute walk in multiple sclerosis subjects and healthy controls.
        Mult. Scler. 2008; 14: 383-390https://doi.org/10.1177/1352458507082607
        • Goldman M.D.
        • Motl R.W.
        • Rudick R.A.
        Possible clinical outcome measures for clinical trials in patients with multiple sclerosis.
        Ther. Adv. Neurol. Disord. 2010; 3: 229-239https://doi.org/10.1177/1756285610374117
        • Hauser S.L.
        • et al.
        Ocrelizumab versus Interferon Beta-1a in relapsing multiple sclerosis.
        N. Engl. J. Med. 2017; 376: 221-234https://doi.org/10.1056/NEJMoa1601277
        • Heesen C.
        • et al.
        Patient perception of bodily functions in multiple sclerosis: gait and visual function are the most valuable.
        Mult. Scler. 2008; 14: 988-991https://doi.org/10.1177/1352458508088916
        • Hirtz D.
        • et al.
        How common are the "common" neurologic disorders?.
        Neurology. 2007; 68: 326-337https://doi.org/10.1212/01.wnl.0000252807.38124.a3
        • Hobart J.C.
        • Riazi A.
        • Lamping D.L.
        • Fitzpatrick R.
        • Thompson A.J.
        Measuring the impact of MS on walking ability: the 12-Item MS Walking Scale (MSWS-12).
        Neurology. 2003; 60: 31-36
        • Jokubaitis V.G.
        • et al.
        Predictors of long-term disability accrual in relapse-onset multiple sclerosis.
        Ann. Neurol. 2016; 80: 89-100https://doi.org/10.1002/ana.24682
        • Kalron A.
        • Givon U.
        Gait characteristics according to pyramidal, sensory and cerebellar EDSS subcategories in people with multiple sclerosis.
        J. Neurol. 2016; 263: 1796-1801https://doi.org/10.1007/s00415-016-8200-6
        • Killeen T.
        • et al.
        Minimum toe clearance: probing the neural control of locomotion.
        Sci. Rep. 2017; 7https://doi.org/10.1038/s41598-017-02189-y (2017)
        • Killeen T.
        • et al.
        Increasing cognitive load attenuates right arm swing in healthy human walking.
        R. Soc. Open Sci. 2017; 4160993https://doi.org/10.1098/rsos.160993
        • Kister I.
        • et al.
        Disability in multiple sclerosis: a reference for patients and clinicians.
        Neurology. 2013; 80: 1018-1024https://doi.org/10.1212/WNL.0b013e3182872855
        • Kurtzke J.F.
        Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).
        Neurology. 1983; 33: 1444-1452
        • Meyer C.
        • et al.
        Familiarization with treadmill walking: how much is enough?.
        Sci. Rep. 2019; 9: 5232https://doi.org/10.1038/s41598-019-41721-0
        • Motl R.W.
        • et al.
        Validity of the timed 25-foot walk as an ambulatory performance outcome measure for multiple sclerosis.
        Mult. Scler. 2017; 23: 704-710https://doi.org/10.1177/1352458517690823
        • Palace J.
        • et al.
        Assessing the long-term effectiveness of interferon-beta and glatiramer acetate in multiple sclerosis: final 10-year results from the UK multiple sclerosis risk-sharing scheme.
        J. Neurol. Neurosurg. Psychiatry. 2019; 90: 251-260https://doi.org/10.1136/jnnp-2018-318360
        • Paltamaa J.
        • Sarasoja T.
        • Leskinen E.
        • Wikstrom J.
        • Malkia E.
        Measuring deterioration in international classification of functioning domains of people with multiple sclerosis who are ambulatory.
        Phys. Ther. 2008; 88: 176-190https://doi.org/10.2522/ptj.20070064
        • Spain R.I.
        • Mancini M.
        • Horak F.B.
        • Bourdette D.
        Body-worn sensors capture variability, but not decline, of gait and balance measures in multiple sclerosis over 18 months.
        Gait Posture. 2014; 39: 958-964https://doi.org/10.1016/j.gaitpost.2013.12.010
        • Sutliff M.H.
        Contribution of impaired mobility to patient burden in multiple sclerosis.
        Curr. Med. Res. Opin. 2010; 26: 109-119https://doi.org/10.1185/03007990903433528
        • Trapp B.D.
        • Nave K.A.
        Multiple sclerosis: an immune or neurodegenerative disorder?.
        Annu. Rev. Neurosci. 2008; 31: 247-269https://doi.org/10.1146/annurev.neuro.30.051606.094313
        • Watelain E.
        • Barbier F.
        • Allard P.
        • Thevenon A.
        • Angue J.C.
        Gait pattern classification of healthy elderly men based on biomechanical data.
        Arch. Phys. Med. Rehabil. 2000; 81: 579-586https://doi.org/10.1016/s0003-9993(00)90038-8
        • Zeni J.A.
        • Richards J.G.
        • Higginson J.S
        Two simple methods for determining gait events during treadmill and overground walking using kinematic data.
        Gait Posture. 2008; 27: 710-714https://doi.org/10.1016/j.gaitpost.2007.07.007
        • Zorner B.
        • et al.
        Prolonged-release fampridine in multiple sclerosis: improved ambulation effected by changes in walking pattern.
        Mult. Scler. 2016; 22: 1463-1475https://doi.org/10.1177/1352458515622695