Introduction

Multiple sclerosis (MS) is a chronic, inflammatory and neurodegenerative disease of the CNS characterized by inflammation, demyelination and neurodegeneration [1]. Motor dysfunction, deficits in gait and balance, and cognitive impairment are clinical hallmarks of MS that impact daily life activities [2]. Cognitive impairment affects 43–70% of MS patients. The most frequently involved cognitive domains are information processing speed, executive functions, attention, working memory and episodic memory [3].

Cognitive and motor symptoms often occur simultaneously and their relationship is well known in numerous neurological disorders [4,5,6,7,8]. In Alzheimer’s disease, impaired executive functions can lead to a significant increase of stride time and stride time variability during walking [4]. Gait deficits in patients with Parkinson's disease are exacerbated while performing simultaneous cognitive or motor tasks, with reduction in walking speed, left and right steps coordination and stride length [7]. In elderly individuals, the coexistence of cognitive deficits and mobility impairment (i.e., slow gait speed) is associated with elevated risk of progression to dementia [9, 10]. There is a growing recognition that mobility and cognitive impairments should be considered interrelated to develop new risk assessment procedures in clinical and research settings [11]. Since the likelihood of cognitive impairment increases in subjects with slower walking speed, gait speed assessment was proposed to detect prodromal dementia [11].

The simultaneous performance of cognitive tasks while walking, termed dual tasking, has been suggested as an useful tool to detect the impact of motor–cognitive impairment in patients suffering from MS [12, 13]. Several existing studies demonstrated that dual task generates cognitive–motor interference causing loss of efficacy in one or both tasks [14]. In MS patients, cognitive-motor interference intensifies functional impairments by reducing gait speed [15] and increasing fall risk [16]. Alterations in motor performance under dual-task condition are often explained within the context of the “limited-capacity model of attention”, according to which both the motor and cognitive tasks compete for finite attentional resources [16]. More recently, Plummer and Eskes [17] proposed the “task prioritization model” suggesting that the magnitude and direction of dual-task interference is also influenced by the way in which individuals spontaneously prioritize their attention.

In MS, deficits in executive functions and information processing speed have been reported to be correlated with slower gait speed measured with Timed 25-Foot Walk test (T25FW) [6, 18,19,20,21], poorer walking endurance assessed with the 6-min walk test [18, 20, 21] and lower functional mobility [18]. Moreover, cognitive dysfunction is associated with higher step length variability and increased fall frequency [22, 23]. One possible explanation is that slower information processing speed prevents a prompt response to postural threats, resulting in increased gait variability and risk of falling [22, 23]

In physiological normal conditions, walking is mainly automatic and rhythmic; however, moving across a complex environment requires higher-order cognitive functions to plan movements and avoid obstacles [5]. Gait control involves both the direct and indirect motor pathways, the former including the primary motor cortex, cerebellum, and spinal cord, and the latter involving the basal ganglia, premotor area and prefrontal cortex [5]. Existing studies of neurological patients and older adults [4, 7, 8] have mainly focused on associations between cognitive dysfunction and gait impairment, while the link between cognitive performance and hand motor function has received little attention. A few investigations demonstrated an association between hand dexterity and cognitive tests measuring information processing speed and executive functions in older adults [24] and MS patients [6]. Yet, the pathological mechanisms underpinning motor and cognitive dysfunctions remain unclear.

In one study [6], 211 MS patients underwent extensive functional and neuropsychological evaluation to assess the influence of cognitive impairment on motor dysfunction. Information processing speed and executive function tests predicted variability in gait speed and hand dexterity. However, the possible role of structural brain damage has not been explored. Gray matter atrophy and white matter abnormalities are known to correlate with both motor and cognitive dysfunction [25, 26], which can therefore develop in parallel, and not necessarily directly influencing each other. The detected association between motor and cognitive performance may be due to the tendency of MS patients with preserved mobility to perform better on neuropsychological tests than those with motor impairment.

Aim of this work is to investigate the association between cognitive function, motor performance and brain structural damage (i.e., WM lesions and atrophy) in patients with MS. To gain additional insights into the mechanisms underlying the different aspects of motor deficits in these patients and their interplay with cognitive dysfunction, we also assessed the effect of demographic, clinical, cognitive, and brain structural measures on gait speed measured with the T25FW and hand dexterity evaluated with the nine-hole peg test (9-HPT).

Methods

Ethics committee approval. The local ethical standards committee on human experimentation approved this study. Written informed consent was obtained from all participants.

Participants. We recruited 106 relapsing–remitting MS patients (RRMS) and 81 sex- and age-matched healthy controls (HC). To mitigate a potential bias due to the exhaustion of functional compensatory mechanisms and the severity of structural damage, which could mask any effects of cognition on motor performance [27], we excluded patients with the progressive phenotypes of the disease. Inclusion criteria were: age ≥ 18 years, right handedness (Edinburgh Handedness Inventory pre-MS condition ≥ 50) [28], Italian native language, no previous history of neurological disorders (other than MS for patients), and no psychiatric disorders. RRMS patients were steroid-free for at least one month, and treatment-stable for at least three months.

Clinical and functional assessment. Within 48 h from the MRI acquisition, all participants underwent a neurological and functional assessment. The neurological evaluation included the Expanded Disability Status Scale (EDSS) [29], right- and left 9-HPT, and T25FW. All functional tests were repeated twice and results were averaged.

Neuropsychological assessment. Patients also underwent cognitive assessment. Experienced neuropsychologists conducted the neuropsychological assessment on the day of MRI, using the Brief Repeatable Battery of Neuropsychological Tests (BRB-N), which assesses verbal memory (Selective Reminding Test [SRT]), visuospatial memory (10/36 Spatial Recall Test [SPART] and delayed-recall), attention and information processing speed (Symbol Digit Modalities Test [SDMT], Paced Auditory Serial Addition Test [PASAT] 3″ and 2″) and verbal fluency (Word List Generation [WLG]) [30]. Corrected scores for age, sex, and education according to normative values [30] were derived based on normative data, obtaining z-scores for each cognitive test.

MRI acquisition. Using a Philips Intera 3.0 T scanner, the following sequences of the brain were acquired from all subjects: (1) dual-echo turbo spin echo (DE-TSE) (TR/TE = 2599/16–80 ms; flip angle 90°; FOV 240 mm2; matrix 256 × 256; ETL 6; 44 contiguous, 3-mm-thick axial slices) and (2) 3D T1-weighted fast field echo (TR/TE = 25/4.6 ms, flip angle 30°, FOV 230 mm2, matrix 256 × 256, slice thickness 1 mm, 220 contiguous axial slices, in-plane resolution 0.89 × 0.89 mm2).

MRI Analysis. T2-hyperintense and T1-hypointense lesion volumes (LV) were measured on DE-TSE and 3D T1-weighted images, using a local thresholding segmentation technique (Jim 7.0, Xinapse Systems Ltd, Colchester, UK). After T1-hypointense lesion refilling [31], normalized volumes of the whole brain (NBV), GM (NGMV) and WM (NWMV) were measured using the SIENAx software [32]. Normalized deep GM volume (NDGMV), corresponding to the sum of the normalized volumes of bilateral thalamus, caudate, putamen, pallidum, amygdala and accumbens, measured with the FIRST tool [33], was also calculated.

Statistical analysis. The statistical analysis was performed using R-4.0.3 and SPSS software, version 22.0. Between-group comparisons of demographic and brain MRI measures were assessed with the chi-square test, Mann–Whitney U test, or two-sample t test (according to normality assumption).

The inverse of 9-HPT and T25FW scores were converted to z-scores using HC’s mean and standard deviation. The significance of z-scores deviation from the HC population was tested with a one-sample t test.

T2-LV and T1-LV underwent a logarithmic transformation before statistical analysis. Right and left 9-HPT z-scores were averaged to reduce the number of variables.

Univariate analyses (Pearson’s correlations) were used to assess the association between MRI measures, cognitive scores and motor performance (z-scores of T25FW and 9-HPT). We investigated the differences of correlation between the two considered functional scores by Steiger’s z-test [34]. Hierarchical regression analysis was performed to select the independent predictors of motor performance. We used a within-block stepwise approach: block 1 included demographic and clinical variables (age, sex, disease duration and education); block 2 comprised all MRI measures (NBV, NGMV, NWMV, NDGMV, T2-LV and T1-LV) and block 3 contained BRB-N tests z-scores (SRT, SPART, SDMT, PASAT 3″, PASAT 2″ and WLG). Since the EDSS score is not an independent variable explaining disability, but a measure of disability itself, it was not included among the demographic and clinical variables of the models. For all analyses, statistical significance was set at p < 0.05.

Results

Demographic, clinical, functional, neuropsychological and MRI features. Table 1 summarizes the main demographic, clinical, functional, neuropsychological and MRI findings in MS patients and HC. Compared to HC, RRMS patients showed significant motor impairment measured with 9-HPT (mean z-score =  − 1.0 ± 0.8; p < 0.001) and T25FW (mean z-score =  − 0.5 ± 0.8; p = 0.001).

Table 1 Main demographic, clinical, functional, neuropsychological and structural MRI characteristics of the subjects included in the study
Full size table

RRMS patients performed below the normative values in all neuropsychological tests (mean z-score ≤ -0.3) except for SPART (mean z-score = 0.0 ± 0.9) and SPART recall (mean z-score = 0.0 ± 0.9).

Compared to HC, RRMS patients had lower NBV (p = 0.02), NGMV (p = 0.01) and NDGMV (p < 0.001).

Correlations between cognitive, MRI and motor performance. Table 2 shows the univariate associations between functional measures (z-scores) and neuropsychological tests (z-scores) as well as between functional measures (z-scores) and MRI measures in RRMS patients. Linear relationship between motor and cognitive performance is presented in Figs. 1 and 2.

Table 2 Univariate association (Pearson’s correlations) between neuropsychological tests z-scores, MRI measures and functional measures in RRMS patients
Full size table
Fig. 1
figure 1

Scatterplots for significant positive correlations between 9-HPT and cognitive performance tests. Linear relationship between 9-HPT and SPART (r = 0.359; p < 0.001), SPART recall (r = 0.297; p = 0.002), SDMT (r = 0.388; p < 0.001) and PASAT 2″ (r = 0.294; p = 0.002). 9-HPT nine-hole peg test, PASAT paced auditory serial addition test, SDMT symbol digit modalities test, SPART 10/36 spatial recall test

Full size image
Fig. 2
figure 2

Scatterplots for significant positive correlations between T25FW and cognitive performance tests. Linear relationship between T25FW and SPART (r = 0.236; p = 0.015) and SDMT (r = 0.205; p = 0.035). SDMT symbol digit modalities test, SPART 10/36 spatial recall test, T25FW 25-foot walk test

Full size image

Visuospatial memory assessed with SPART was associated with 9-HPT (r = 0.359; p < 0.001) and T25FW (r = 0.236; p = 0.015). Reduced information processing speed measured with SDMT, PASAT 3″ and PASAT 2″ was related to poorer performance on 9-HPT (r ≥ 0.254; p ≤ 0.009). SDMT z-scores resulted also significantly associated with T25FW (r = 0.205; p = 0.035). Verbal fluency significantly correlated with hand dexterity assessed with 9-HPT (r = 0.226; p = 0.020).

All functional measures were significantly associated with NBV (r ≥ 0.346; p < 0.001), NGMV (r ≥ 0.340; p < 0.001), NWMV (r ≥ 0.234; p ≤ 0.016) and NDGMV (r ≥ 0.317; p ≤ 0.001). Higher T2-LV and T1-LV were correlated with worse performance at 9-HPT (r ≤ -0.430; p < 0.001) and slower walking speed (r ≤ -0.254; p ≤ 0.009).

Applying Steiger’s z-test to assess differences between correlations, 9-HPT showed stronger association with PASAT 3″ (p = 0.018), PASAT 2″ (p = 0.021) and NGMV (p = 0.035) than T25FW.

Predictors of motor performance. Table 3 summarizes the results of the hierarchical regressions analysis.

Table 3 Hierarchical regression models assessing the predictors of motor performance for RRMS patients
Full size table

Older age (p = 0.001; ΔR2 = 0.103), male sex (p = 0.042; ΔR2 = 0.035), lower NGMV (p < 0.001; ΔR2 = 0.187) and lower scores at PASAT 2″ (p = 0.017; ΔR2 = 0.037) predicted worse performance at 9-HPT (adjusted R2 = 0.337). Older age (p = 0.013; ΔR2 = 0.058) and lower NDGMV (p = 0.005; ΔR2 = 0.068) predicted slower gait speed measured with the T25FW (adjusted R2 = 0.109).

Discussion

In this study, we assessed the correlation between functional measures of upper and lower limbs, cognitive performance, and structural brain damage in RRMS patients. We subsequently identified the predictors of hand motor dexterity and walking speed among a set of demographic, clinical, MRI and cognitive variables using hierarchical regression analysis.

Better performance at tests measuring visuospatial memory and information processing speed correlated with higher degree of hand dexterity and faster walking speed. Motor performance resulted also associated with all MRI measures analysed (T2-LV, T1-LV, NBV, NGMV, NWMV and NDGMV).

As a second step, we searched for the predictors of motor impairment. Younger age, female sex, higher NGMV and higher PASAT 2″ scores predicted better hand motor performance assessed with 9-HPT, while younger age and higher NDGMV were selected as significant predictors of faster walking speed.

Our results confirm the positive correlation between motor and cognitive performance in MS [6, 19, 22, 35]. However, our work differs from available literature as we assessed the influence of potential confounding variables such as brain atrophy and focal WM lesions on the association between motor and cognitive dysfunction. It could therefore be hypothesized that information processing speed and executive function deficit contribute to explain hand motor dysfunction. Complex movements require cognitive planning and monitoring, especially in subjects with motor impairment [5, 7, 19] and a high order cognitive control dysfunction can affect the performance at complex motor tasks independently from physical disabilty and structural brain damage.

Significantly higher correlation coefficients were found between 9-HPT and PASAT 3″ and PASAT 2″ than T25FW. Our results are in line with those of previous studies which observed stronger correlations between SDMT [6, 35], PASAT [6, 35], the Brief Visuospatial Memory Test-Revised [6] and hand dexterity compared to walking speed. We hypothesize that eye-hand coordination needed to execute timely and skilful movements requires more cognitive resources than movement coordination and integration of sensory feedback necessary for walking forward [5, 36]. The greater level of cognitive complexity assumed for hand dexterity tasks compared to gait speed measures can thus explain why correlations with cognitive performance were more often statistically significant and higher for 9-HPT than T25FW.

Hand motor functioning and gait speed resulted significantly correlated with all measures of structural brain damage, as consistently demonstrated by previous studies [37]. In particular, 9-HPT scores resulted more closely associated with NGMV than walking ability. Such aspect might be due to the higher functional demand required for upper limb function, which could make it more influenced by brain damage accumulation. On the other hand, it is well-known that involvement of CNS areas, such as the spinal cord, may be more clinically relevant to explain locomotor dysfunction [38].

Univariate analysis results confirmed significant relationships between neuropsychological and motor outcomes [6, 18,19,20,21]. However, in our study, lower correlation coefficients were found compared with previous works [18, 21]. This can be explained by the fact that our sample was younger and less heterogeneous in terms of clinical phenotypes than the samples included in earlier studies [18, 21].

Applying hierarchical regression analysis, we identified the predictors of hand motor function and gait speed. Age was selected as significant predictor of motor dysfunction, confirming that slower ambulation and poorer hand dexterity are associated with advancing age in MS [39, 40]. The regression models also indicated that NDGMV is associated with T25FW. Our results align with previous works demonstrating the involvement of thalamic and basal ganglia structures in motor control [41, 42].

For the 9-HPT, predictors of performance were age, sex, NGMV and PASAT performance. The effect of sex on hand motor function is in agreement with previous studies showing that men need more time to complete the task than women [43]. A previous study in a large cohort of RRMS patients showed that NGMV was one of the main predictors of upper limb motor performance [44]. As reported by the literature, atrophy of cerebellum, basal ganglia and fronto-parietal regions may play a critical role in eye-hand coordination [25, 36].

Nine-HPT was confirmed significantly associated with PASAT 2″ even after controlling for demographic variables, clinical features and structural brain damage. PASAT has been developed to measure the rate of information processing [45]; however, numerous studies pointed out that PASAT is also related to other cognitive functions such as working memory and sustained attention [45]. Sherman and colleagues [46] tested PASAT construct validity and observed that, despite mathematical ability and attention accounted for a substantial amount of variance, verbal ability and even complex motor skills were also moderately correlated with its performance. The amount of variance shared between PASAT and complex motor skills such as manual dexterity is thought to be due to attentional/executive functioning deficits.

Evidence supports high-order cognitive functions involvement in complex movements. Our study confirmed the 9-HPT as a complex task requiring the interaction between motor- and cognitive abilities beyond the effect of disease progression and structural brain damage.

This study has a few limitations. First, our protocol did not include a spinal cord MRI acquisition that could have improved the ability of the final models to predict motor impairment. Second, we did not consider fatigue, depression and anxiety symptoms, which may influence cognitive performance in MS patients [3]. Third, it is cross-sectional, thus not allowing evaluating whether the observed predictors are related to motor worsening. Finally, the selected demographical, clinical, neuropsychological and MRI variables were able to explain only 34% and 11% of variance in the 9-HPT and T25FW, respectively. The inclusion of other MRI measures, such as those of structural and functional connectivity, could have increased the proportion of variance accounted.

Overall, this study provides evidence for an additional contribution of high order cognitive control dysfunction to motor hand dexterity impairment in MS, beyond the effect of structural brain damage.

Future studies should investigate the longitudinal evolution of cognitive and motor dysfunction in MS and clarify eventual common pathological mechanisms.

Our results support the idea that a multidimensional approach should be implemented in the clinical setting to improve the assessment of functional and cognitive deficits and enhance rehabilitation programs [11]. Also, this study provides further evidence for using composite scores for MS evaluation, including both motor and cognitive outcomes [35].