Evaluation of the T25FW in minimally disabled people with multiple sclerosis

Background: Walking impairment is one of the most prevalent symptoms in people with multiple sclerosis (pwMS). In this study, we aimed to explore the usefulness of a simple walking test, the Timed 25 Foot Walk (T25FW), in detecting subtle differences in “ fully ambulatory ” pwMS compared to HC. Methods: We therefore investigated retrospective data from a clinical real-life cohort of 650 pwMS. We first analyzed the amount of patients showing clinically relevant impairment in the T25FW (T25FW > 6 s) within different levels of disability according to the Expanded Disability Status Scale (EDSS). For detailed analysis in “ fully ambulatory ” pwMS, we formed four groups according to the respective levels of disability (EDSS 0, EDSS 1, EDSS 1.5 – 2, EDSS 2.5 – 3), and compared their walking speed to age-and sex-matched healthy controls (HC). Results: In our cohort, the number of patients showing clinically relevant slowing in the T25FW ranged from 15% in “ fully ambulatory ” patients (EDSS 0 – 3) to 69% in patients with moderate (EDSS 3.5 – 5.5) and 100% in patients with severe impairment (EDSS ≥ 6). Further analyses in “ fully ambulatory ” patients revealed that all EDSS-subgroups showed significant slowing compared to HC. The mean difference to walking speed of HC became gradually more pronounced from 0.15 m/s in asymptomatic patients (EDSS 0) to 0.5 m/s in patients with EDSS 2.5 – 3. Conclusion: These findings underline the ability of the T25FW to detect slowing even in patients with minimal disability. While the difference to HC was slightly below clinical relevance in asymptomatic patients (EDSS 0), slowing gradually worsened from EDSS 1 onwards and exceeded published thresholds for clinical meaningfulness.


Introduction
Walking impairment represents one of the most prevalent symptoms in people with multiple sclerosis (pwMS).Impairments strongly affect activities of daily life and, as a result, walking function is one of the main contributors to patients' perception of health [1,2].As evidenced by a recent multicentre study, pwMS already show slowing in walking speed early on (< 30 years) when compared to healthy controls (HC) and demonstrate an accelerated deterioration of walking speed with advancing age [3].
However, the EDSS has limitations in detecting mild walking impairment [5].As became evident by recent concepts such as PIRA (Progression Independent of Relapse-Activity; [7]), however, progressive changes can occur even in early stages of the disease.The earlier such progression is detected, the sooner appropriate treatment can be initiated [8].
For improved detection of changes in walking speed, objective tests such as the Timed 25-Foot Walk (T25FW) have been introduced [9].The T25FW can be administered with little temporal and spatial effort [6], is considered highly reliable and valid, and is consequently widely used as an outcome in clinical trials [10,6].About a decade ago, a wellestablished benchmark for clinically relevant slowing was created for the T25FW (time to complete the T25FW ≥6 s), that comes along with occupational changes, need for some assistance in everyday life, and walking with a cane [11].
When it comes to subtle walking impairment, however, the sensitivity of the T25FW is debated as well, due to the presence of floor effects [12,13,10].Previous studies, in which pwMS were compared to HC, usually found a significant difference between the two groups, however, those studies either had small sample sizes or did not explicitly compare patients with minimal impairment to healthy controls (for metaanalysis, see Sikes et al. [8]).Of note, it should be considered that the mean T25FW-performance in the studies included in the meta-analysis was worse than the threshold for clinically significant slowing (M = 6.7 s; [11,8]).Two recent outstanding studies focusing on patients with early MS and minimal impairment advocated the use of more complex balance tests instead of the T25FW, since included patients walked relatively fast in the T25FW [14,10].However, in those studies, patients were not evaluated against age-and sex-matched HC regarding the T25FW [14,10].On the other hand, recent findings highlight a significant slowing in walking speed in patients with low self-reported disability (assessed by the patient-determined disease step (PDDS) scale), when compared to previously published data of age-matched healthy controls [15].Hence, whether the T25FW can be a useful tool for detecting and monitoring minimal walking impairment should still be further explored [16].
In this single-centre study, we thus aimed to investigate the T25FW in a large real-life cohort of pwMS from the MS Outpatient Department of the Medical University of Graz, with a special focus on its ability to detect slowing in walking speed even in minimally disabled patients.For more detailed analyses in patients with minimal impairment, we directly compared different groups of fully ambulatory patients (including asymptomatic patients (EDSS 0)) to prospectively assessed age-and sexmatched healthy controls.

Sample-characteristics
The patient cohort was obtained by extracting data from the research database of the Medical University of Graz, consisting of all clinical assessments at the local outpatient clinic in which the T25FW was performed, between the years 2012 and 2019.Data from each patient's first visit within that time interval were included.Inclusion criteria were diagnosis of clinically isolated syndrome or MS and at least 18 years of age, exclusion criteria were clinical relapse or change of diseasemodifying treatment (DMT) within 6 weeks, temporally overlapping injuries or missing data for EDSS or T25FW (for a detailed study flow, see Fig. 1).
For more detailed analyses in patients with minimal disability, T25FW-data of 61 HC assessed between 2021 and 2022 were analyzed (approved by the local ethics-committee; 31-432ex18/191264-2019 & 33-056ex20/21).The HC cohort was collected in order to match a multiple sclerosis cohort of mild impairment in terms of age and sex.HC had no history of neurologic or psychiatric disorders.Additionally, magnetic resonance imaging (MRI)-scans of HC were checked for abnormalities by an experienced neurologist (CE).One HC presented an outlier (>3 interquartile ranges above the third quartile) in terms of T25FW speed (i.e., showing extremely high walking speed) and was therefore excluded from all analyses (final n = 60 HC).Written informed consent was obtained from all participants and the study was conducted in accordance with the Declaration of Helsinki.

Outcomes and statistical analysis
Data was analyzed using the Statistical Package of Social Science (IBM SPSS Statistics 27).The level of significance was set to 0.05.
Primary endpoints of the study were the T25FW [9] and the EDSS [4].For the T25FW, participants were instructed to walk along a clearly marked course of 25 ft (7.62 m) as quickly as possible, but safely [17].
The task was administered twice and participants were allowed to use an assistive device.We extracted the average time (in s) to complete the task as well as the walking speed in m/s (T25FW speed; 7.62 m divided by the time in seconds, [6]).We additionally created a binary variable based on the well-established benchmark indicating clinically meaningful slowing by Goldman et al. [11], defined as an average time to complete the T25FW of ≥6 s.
For descriptive purposes and to estimate the comparability of our real-life cohort with the data of previous large clinical trials [13,16], we first assessed the T25FW and EDSS within the entire cohort.Patients were classified into three groups according to their EDSS-score (mild impairment, "fully ambulatory" according to the EDSS: EDSS 0-3; moderate impairment: EDSS 3.5-5.5;severe impairment: EDSS ≥6) [6,18].For each EDSS-group, the number of patients showing clinically meaningful slowing in the T25FW (T25FW time ≥ 6 s) was assessed.Additionally, we computed a correlation between T25FW speed and neurological impairment (EDSS) within the entire patient cohort, adjusted for age and sex.
For analysis in patients with minimal disability, we further divided the group of fully ambulatory patients (EDSS 0-3) into four subgroups based on the extent of their neurological impairment (EDSS 0, EDSS 1, EDSS 1.5-2, and EDSS 2.5-3).Note that, due to the small sample sizes in the groups of EDSS 1.5 (n = 37) and 2.5 (n = 31), we grouped them together with the respective next higher category (EDSS 2 and EDSS 3, respectively).We compared HC and the four groups of fully ambulatory pwMS in terms of gender distribution (using a Chi 2 -test) and age (oneway ANOVA with five independent groups).Lastly, we conducted an analysis of covariance (ANCOVA), adjusted for age [3], to identify differences in T25FW speed between the five groups (HC, EDSS 0, EDSS 1, EDSS 1.5-2, EDSS 2.5-3).Post-hoc tests were corrected for multiple comparisons using the Bonferroni procedure.
In order to get an approximate estimation of the implication of the speed differences found in the T25FW for longer walking distances in everyday life, we computed a prediction for the 6-min walk test (6MWT) based on a formula published on behalf of the European Rehabilitation in MS (RIMS) network [19].The 6MWT is a test where participants are asked to walk as far as possible for 6 min and, therefore, is considered a measure of walking endurance [20,21,19].Ramari et al. [19] provide means and 95% confidence intervals (CI) for the predicted 6MWT distance based on the formula detailed in section E.1 in the Appendix.

Findings in the entire study cohort
Descriptive statistics of the entire patient cohort can be found in Table 1.The majority of pwMS in this cohort (80.2%, n = 521) had EDSS-scores between 0 and 3 (overall EDSS-range: 0-7).A total of 87% (n = 567) of patients were diagnosed with clinically-isolated-syndrome (CIS) or relapsing-remitting-MS (RRMS).In patients labelled "fully ambulatory" according to the EDSS (EDSS 0-3), 77 patients (14.8%) showed clinically meaningful slowing in the T25FW (i.e., T25FW ≥ 6 s).In patients with moderate impairment according to the EDSS (EDSS 3.5-5.5),72 patients (out of 104 patients; 69%) showed clinically meaningful walking impairment in the T25FW.All patients with severe impairment according to the EDSS (EDSS ≥6) showed clinically meaningful walking impairment in the T25FW, with no patient walking faster than 8 s (n = 25; 100%).As expected, higher EDSS-scores correlated with slower walking speed within the entire patient cohort (r = − 0.63; p < 0.001, adjusted for sex and age).The distribution of walking speed across the entire study cohort is illustrated in Fig. A1 in the Appendix.

Findings in pwMS with minimal disability
Details on demographics and walking speed for HC and the four "fully ambulatory" groups of pwMS (EDSS 0, EDSS 1, EDSS 1.5-2, EDSS 2.5-3) are provided in Table 2. Further descriptive statistics (sex, age and clinical characteristics) on the four groups of pwMS are shown in Table A1 in the Appendix.
The ANCOVA comparing HC and the four fully ambulatory groups of pwMS (EDSS 0, EDSS 1, EDSS 1.5-2, EDSS 2.5-3) revealed a significant group difference in walking speed (F(4,575) = 29.4,p < 0.001), indicating a large effect (partial eta 2 = 0.17).Fig. 2 illustrates the distribution of walking speed within the five groups as well as the results from post-hoc comparisons.Post-hoc comparisons corrected for Bonferroni revealed a statistically significant decrease in walking speed in asymptomatic patients (EDSS 0) compared to HC (p = 0.013).Additionally, all other contrasts between any two fully ambulatory patient groups or between any patient group and HC revealed significant differences (all p < 0.005) except for the comparison between patients with EDSS 0 and EDSS 1 (p = 1.00).Table 2 shows the mean difference in walking speed between each fully ambulatory patient group and HC.The mean difference to HC was 0.15 m/s for asymptomatic patients (EDSS 0) and the magnitude of the difference was greater in groups with higher EDSSscores, up to a slowing of 0.5 m/s in patients with EDSS-scores between 2.5 and 3.As can be seen in Table 2, predicted mean 6MWT performance was 569 m in HC.The difference to HC in terms of predicted mean 6MWT performance was 44 m in patients with EDSS 0, 63 m in patients with EDSS 1, and over 100 m in patients groups with EDSS >1.

Discussion
In this study, we aimed to investigate the ability of the T25FW to detect walking impairment in a real-life cohort of pwMS, with a special focus on patients with minimal disability.In our cohort, already 15% of fully ambulatory pwMS (EDSS 0-3) showed a clinically relevant slowing in the T25FW (T25FW ≥ 6 s), and this rate amounted to 69% in patients with moderate (EDSS 3.5-5.5)and 100% in patients with severe impairment (EDSS ≥6).For further analyses in mildly impaired pwMS, we divided the fully ambulatory pwMS into four subgroups based on disability level (EDSS 0, EDSS 1, EDSS 1.5-2, and EDSS 2.5-3) and compared their walking speed to age-and sex-matched healthy controls.Analyses revealed that all investigated subgroups showed statistically significant slowing in the T25FW compared to HC.The mean difference to walking speed of HC ranged from 0.15 m/s in asymptomatic patients (EDSS 0) to 0.5 m/s in patients with EDSS 2.5-3.
The cohort presented in this study is comparable to the largest cohorts of T25FW of previous literature [13,16,8] in terms of sex, age and disease duration, but displays a lower median EDSS impairment and faster T25FW times (e.g., median EDSS = 1.25 vs 3 & mean T25FW = 5.7 vs 9.2 s in Kalinowski et al. [16]).Other than that, our cohort shows the typical negative association between T25FW speed and EDSS, as reported in various other studies [6,22,8].The result that already 15% of "fully ambulatory" patients (EDSS 0-3) show clinically meaningful slowing in the T25FW further underlines the weakness of the EDSS in detecting changes in walking function in mildly impaired pwMS.Likewise, the steep slope in the amount of patients showing clinically relevant slowing from mild (15%) to moderate (69%) and severe impairment (100%) is in line with previous literature on construct validity in the T25FW [6,22].Regarding mild impairment, our analyses revealed significant slowing in walking speed in all ambulatory patient groups, which gradually worsened with higher levels of neurological disability.These results are in concordance with a recent meta-analysis including 31 studies in which walking speed in pwMS was compared to that of HC [8].Since only a few of those studies explicitly compared patients with mild impairment according to the EDSS with HC (with varying cutoff-values used for the definition of mild impairment), and if so, the studies had small sample sizes (patients with mild impairment: n max = 64) [8], our results add additional weight to these findings.Likewise, a recent paper found that patients with mild to no perceived gait issues performed worse in the T25FW compared to normative reference data from HC (n = 95; [15]).These results underline the notion that the T25FW provides important additional information not only to commonly used clinical scores, but also to subjective gait indices measured with patientreported outcomes, revealing even subtle impairment.
On the other hand, our results differ from another interesting, recently published study on asymptomatic patients (EDSS 0; [10]).In the cohort assessed by Krieger et al. [10], walking speed in patients with EDSS 0 was similar to that of our HC, with no patients taking >5 s to complete the test.One possible explanation for this might be the difference in inclusion criteria: in Krieger et al. [10], only patients between 20 and 50 years of age, who had a shorter disease duration than 5 years were included [14], while no such criteria were used for our cohort.This   led to a mean age of 33 years (vs M = 37 years in our EDSS 0 cohort) and a median disease duration of 2 years, which was shorter than the median disease duration in our asymptomatic patients by >4 years.Another important point of discussion is whether the group differences found in our fully ambulatory groups of pwMS reflected clinically meaningful differences to HC.In absence of measures indicating a subjective change, we decided upon a two-step approach to investigate the size of the differences.Firstly, we compared the mean difference with published thresholds of minimally important change, which are ~0.20 m/s or ~ 12% in people with mild impairment, [23,24,25].Note that these thresholds are lower than the frequently used threshold of 20%change, since it was repeatedly shown that the variability in the T25FW is lower in less disabled patients [23,24,25].In our cohort, the mean difference to HC is below the thresholds for clinical relevance in patients with EDSS 0 (0.15 m/s vs threshold: 0.20 m/s), while patients with EDSS 1 already exceed the thresholds of minimally important change (0.21 m/ s; ~12% change to HC).Of course, the difference was even stronger in patients with EDSS>1 (>0.36 m/s).To get a rough estimation of what a those differences mean for typical walking distances of everyday life, we further computed an estimate of the mean walking distance in the 6MWT [19].Here, the same pattern was observed: asymptomatic patients (EDSS 0) showed a difference to healthy controls that was slightly below the threshold for minimal important change (actual mean difference = 44 m vs threshold = 54 m; [24]).Accordingly, the threshold was exceeded in patients with EDSS 1 (mean difference to HC: 64 m) and higher (mean difference > 100 m when EDSS>1).
There are several methodological considerations worth mentioning.The main strength of the study is the large, real-life cohort of (fully ambulatory) people with multiple sclerosis, which enabled us to examine people with minimal impairment more thoroughly.Another strength is that we assessed age-and sex-matched healthy controls at the same center and setting, rather than using reference data from other studies, which enabled a direct comparison to pwMS.On the other hand, collecting our own cohort of HC raises the question whether the observed differences in walking speed stem from an unusually high walking speed in our HC.However, the mean T25FW walking speed in our HC (M speed = 1.95 m/s; Median Time = 3.9 s) was comparable or even a little slower than that of published healthy cohorts of similar age (M speed = 2.05 m/s, n = 188 [26,27,3]; Median Time = 3.7 s, n = 104 [22]).Major weaknesses of our study are the retrospective nature of the data and the cross-sectional setting.With this design, we were not able to assess whether the group differences we found came along with other subjective or objective changes in daily activities and, hence, were not able to thoroughly investigate whether the differences found in this cohort are clinically meaningful.Therefore, our results need to be confirmed in prospective (and longitudinal) trials.Another point of note is that the T25FW merely focuses on walking speed and does not measure other important aspects of walking function, such as subtle balance disturbances, endurance, and fatigability.In an ideal scenario, therefore, a set of different ambulation tests would be administered [15], however, a higher number of tests come with a higher burden in clinical routine.

Conclusion
In conclusion, our results underline the ability of the T25FW to detect slowing in walking speed even in cohorts that primarily consist of pwMS walking faster than the currently known threshold for clinically relevant slowing (i.e., T25FW <6 s).These findings support earlier suggestions, such as creating additional benchmarks in the lowest T25FW spectrum (<6 s) [11], or normative data (z-scores) [9] based on age and sex.
= people with multiple sclerosis; HC = healthy controls; EDSS = Expanded Disability Status Scale; IQR = interquartile range; T25FW = Timed 25-Foot Walk; diff.= difference; 6MWT = 6 minute walk test; CI = confidence interval.a numbers represent total amounts (percentages) b numbers represent mean values, round brackets indicate standard deviations and square brackets denote 95% confidence intervals c numers represent median values (interquartile range)1 The walking distance of the 6MWT was predicted based on the T25FW, using a previously published formula by Ramari et al.[19].

Fig. 2 .
Fig. 2. Distribution of walking speed in the T25FW among healthy controls (HC) and fully ambulatory people with multiple sclerosis.Brackets and asterisks indicate significant group differences shown in post-hoc t-tests (Bonferroni-corrected) of an analysis of covariance (ANCOVA, adjusted for age): All group comparisons revealed significant differences except for the comparison between patients with EDSS 0 and patients with EDSS 1. EDSS = Expanded disability status scale; T25FW = Timed 25 Foot Walk.

Table 1
Characteristics of the entire patient cohort, including walking test scores.

Table 2
Distribution of the T25FW among healthy controls and fully ambulatory patients (EDSS 0-3).