Motor imagery (MI) is defined as a dynamic state during which the representation of a specific motor action is internally reactivated within working memory without any actual motor output (Jeannerod, 1995). This phenomenological experience is governed by the principles of central motor control and consists in a first-person perspective of performing a movement.

The mentally and actually executed movements share common characteristics (Grèzes and Decety, 2001; Jeannerod, 2001; de Lange et al., 2005, 2008). Studies of fMRI have demonstrated that both motor execution and MI activate brain regions involved in motor planning and execution, including the supplementary motor area (SMA), premotor area (PMA), primary sensorimotor area (M1/S1), posterior parietal lobe (PPL), striatum, cerebellum and thalamus, although at different volume and intensity of activation (Lotze et al., 1999; Gerardin et al., 2000; Lacourse et al., 2005; Munzert et al., 2009; Zhang et al., 2012; Saiote et al., 2016).

From a behavioral standpoint, the required time to perform a specific movement is tightly correlated with the required time to mentally simulate the same movement. It has been shown that healthy adults show temporal equivalence, i.e., isochrony, between the mental simulation of an action and its actual performance (Gentili et al., 2004; Guillot and Collet, 2005; Munzert et al., 2015). Conversely, the temporal uncoupling between actual and mental movements, i.e., anisochrony, may be the consequence of either abnormal developmental processes (Choudhury et al., 2007; Caeyenberghs et al., 2009) or structural and functional network changes in neural networks occurring with aging (Skoura et al., 2008; Personnier et al., 2010) and with neurological disorders (Sirigu et al., 1996; Daprati et al., 2010; Abbruzzese et al., 2015).

However, there is increasing evidence that the effect of mental practice is partially preserved in neurological diseases such as stroke and Parkinson disease (Sharma et al., 2006; Tamir et al., 2007; Zimmermann-Schlatter et al., 2008) supporting the use of MI in rehabilitative protocols aimed at re-activating sensorimotor networks and promoting motor recovery (Jackson et al., 2001; Johnson et al., 2002; Sharma et al., 2006; de Vries and Mulder, 2007).

Unlike in patients with stroke and Parkinson disease, only few studies have investigated differences of MI ability in multiple sclerosis (MS) and showed that patients had a significantly lower accuracy score than controls in correctly executing a hand rotation task (Heremans et al., 2012; Tabrizi et al., 2013). A recent work from our group confirmed the presence of anisochrony between actual and mental movement in MS patients during a motor task with the upper limbs. We showed that actual movement duration was significantly longer than mental movement duration for the non-dominant arm and suggested that these differences could be related to increased cognitive effort required for performing movements with the non-dominant limb (Tacchino et al., 2013). None of these studies, however, investigated the neural correlates of MI impairment in MS. Both focal and diffuse MS damage of white and gray matter could contribute to the development of motor and cognitive deficits, thus affecting alteration of temporal processing of imagined actions. In addition, although it is known that motor-cognitive impairments accumulate with disability progression (Johansson et al., 2007) affecting movement planning and online control it is still not clear whether anisochrony between MI and motor execution in MS patients is associated with the disease severity. The investigation of the neural correlates of MI as function of disease severity could improve our understanding of changes in behavioral performance (i.e., anisochrony) and could contribute to the development of more effective motor-cognitive strategies to be employed in rehabilitative interventions.

Therefore, the aims of our study were: (a) to investigate the neural correlates of MI in patients with MS using fMRI during a task executed with the right and left hands; (b) to explore the association among disease severity, anisochrony and brain activation patterns during MI. In particular, we focused our attention on the first stages of the disease, by comparing two populations of MS patients, clinically isolated syndrome (CIS) and relapsing-remitting (RR-MS), to healthy controls (HC) matched for age and educational level.

CIS is the initial presentation in 80% of MS cases. CIS encompasses an acute clinical attack affecting one or more central nervous system sites and converts to RR-MS with a variable rate, from 20% to 80% depending on the presence and the number of clinically silent white matter lesions on MRI. These early stages of MS are characterized by demyelination processes, thought to have a common inflammatory pathological substrate (Doshi and Chataway, 2016). Although, there is usually good recovery from an attack, a decreased whole-brain network efficiency was observed for both CIS and RR-MS patients. Disrupted network organization already emerges in CIS, even if with a lesser degree relative to MS, suggesting that differences could be present between populations in mentally predicting motor actions (Liu et al., 2017).

To the best of our knowledge, this is the first study that characterizes the neural correlates of MI in MS patients. Previous studies focused on the behavioral aspects of MI and did not explore the impact of disease severity on imagery ability (Heremans et al., 2012; Tabrizi et al., 2013; Tacchino et al., 2013).

Here, MI ability and corresponding neural correlates of RR-MS patients with mild disability, CIS patients and a group of healthy subjects were compared. Based on the IP, our results suggest that MI ability depends on disease severity as shown by the significantly higher anisochrony in RR-MS patients than CIS patients and controls. Our task consisted of simple squeeze ball movements to be repeatedly imagined in blocks of long duration (30 s). This design require prolonged attention that may have led to a slowdown in the mental motor simulation and a consequent overestimation of movement duration during mental execution (Sabaté et al., 2004; Stinear et al., 2007; Lebon et al., 2012). This might have been further stressed in our task because we asked participants to count how many movements they were performing during MI, which might increase the mental effort, forcing two different mental tasks taking place simultaneously. However, the strict temporal coupling between the two tasks (i.e., an increment in counting concomitant to each imagined movement) and the minimal memory and calculus resources demand (i.e., adding 1) let suppose that these effects should equally influence all the participants. Consequently, the significantly higher anisochrony in patients could be mainly explained by the impact of MS pathology on the forward internal models used in mental prediction of motor actions (Miall and Wolpert, 1996; Wolpert and Flanagan, 2001).

Task-dependency may also explain why here we found imagined movements slower than those actually executed, differently from our previous study (Tacchino et al., 2013). There, RR-MS patients had to perform overt and covert pointing tasks as accurately and as fast as possible between targets of several sizes. The results showed that they imagined faster than when the task was actually performed, probably as consequence of two coexistent causes: the difficulty of the task led to slowness in actual execution and patients probably adopted a simulation strategy that did not fully integrate task constraints (i.e., target size and movement speed) during the mental process, allowing to speed up the mental representation (Guillot and Collet, 2005).

No cognitive and fatigue differences were found, but worse fine hand movements were observed in RR-MS patients with respects to the other groups. The upper limbs physical effects due to MS progression, albeit significant, were probably not consciously perceived, remaining subtle (Solaro et al., 2007), and consequently did not affect a process of self-evaluation such as that required while they rated own imagery vividness through KVIQ (Heremans et al., 2012; Tabrizi et al., 2013; Saiote et al., 2016).

As expected, during actually executed tasks all subjects recruited a network of cortical and subcortical brain regions deputed to motor and sensory control that became more extensive in patients likely as a compensatory mechanism to maintain a normal performance despite the presence of focal and diffuse brain injury (Guillot and Collet, 2005; Pantano et al., 2005; Rocca et al., 2005; Lenzi et al., 2007). A similar pattern of activation was elicited during MI tasks, though M1 and S1 activations were less extensive due to a less essential role of the sensory-motor network during motor simulation and due to the minimal activation of the bulk of the thalamo-cortical projections from sensory inputs (Sharma et al., 2008; Munzert et al., 2009; Di Rienzo et al., 2014). Furthermore, during MI with the dominant hand, CIS and RR-MS patients showed an additional recruitment of frontal, parietal and occipital areas, which may represent a compensatory strategy to achieve a better functional performance (Mizuguchi et al., 2016). Indeed, the activation of the left SFG, generally involved in tasks requiring attention and working memory (Owen, 2000; du Boisgueheneuc et al., 2006; Li et al., 2013) may indicate an increased need of attentional and memory resources to produce accurate mental images (Pokryszko-Dragan et al., 2016). Moreover, the contribute of left SFG to motor tasks (Nachev et al., 2008; Chouinard and Paus, 2010; Martino et al., 2011) together with the higher recruitment of pre-SMA in CIS and PMd and right cerebellum in RR-MS patients suggests a loss of automatism in action simulation with the consequent request of major engagement of brain areas involved in motor control during mental tasks. Indeed, the frontal regions (i.e., pre-SMA) recruited by CIS patients are involved in “higher” level functions (i.e., movement ideation, visuomotor associations, time perception and discrimination, action intention representation and internally guided actions; Nachev et al., 2007), whereas frontal areas (i.e., PMd) preferentially recruited by the RR-MS patients are deputed to movement preparation and selection (Kantak et al., 2012). In addition, the right cerebellum is known to play an important role in coordinating fine motor movements such as those used in this study (Stoodley and Schmahmann, 2009).

Besides frontal areas, during MI with the right hand, both groups of patients showed increased recruitment of parieto-occipital regions such as the SPL and the IPS. The former receives a great deal of sensory inputs, participates in tactile localization and spatial orientation and plays a crucial role in visuomotor coordination especially in producing purposeful and skilled movements. Similarly, the anterior IPS is associated to visuomotor processes such as visual attention and visuomotor coordination, object manipulation and cross-modal transfer of object information between the visual and sensorimotor systems (Grefkes and Fink, 2005). Their increased activation, together with that of LOC, involved in shape object recognition, representation and processing, highlight their critical role in imaging somatosensory and visual aspects, in coding spatial properties and in predicting the temporal feature of the movement to be imagined. Also, bilateral and not only contralateral activation of SPL and IPS in CIS group seems to stress the crucial role of sensory information for these patients in producing accurate MI. Moreover, right IPS is found to work as a bridge between visual cortex and PMA (Alahmadi et al., 2015), furtherly validating the notion that the manipulation of mental images is governed by control processes located in fronto-parietal networks according to current theories on MI (Corbetta and Shulman, 2002; Ganis and Schendan, 2011).

Taken together, the results related to MI with the dominant hand suggest that unlike HC, CIS patients, to produce accurate images, needed not only more attention, memory and the aid of brain areas deputed to motor control, as well as for RR-MS patients, but also the engagement of cortical regions devoted to somatosensory elaboration to abstract features of movement control such as ideation, representation and timing. It is like that, at the first stages of the disease, patients were still able to adopt all these resources in order to counteract the changes due to the isolated MS event. Conversely, patients with definite MS encountered more difficulties in engaging visual, sensory and high level cortical regions and, consequently, in preserving aspects, such as timing, relevant to the correct imagination of movements. Additional activations of motor cerebral areas could represent an ineffective attempt to compensate this deficit.

The loss of ability to compensate the effect of MS severity on MI is more evident when the results related to the task with the non-dominant hand are considered. RR-MS patients did not activate extra cerebral regions with respect to the other two groups; only CIS patients showed more recruitment of parietal and frontal areas when compared with HC subjects, especially sensory areas (S1 and right SPL) and motor areas deputed to movement preparation, selection (PMd bilaterally) and execution (M1), defining a pattern of activation similar to that shown by RR-MS patients during MI with dominant hand. Moreover, CIS with respect to RR-MS patients showed more activation of putamen, interconnected with many other structures to control many motor skills such as, for examples, motor learning, motor preparation and movement sequences. Results suggest that more challenging is the task, more difficult the recruitment of compensatory brain areas, with a consequent deficit in behavioral performance. Thus, the neuroimaging results may explain the differences found in anisochrony among groups and between dominant and non-dominant hand.

These considerations are further supported by results from voxelwise correlation analysis and comparison between MM and AM. The former showed that, during MI with right hand, increased activity in sensory areas correlated with the increment of IP in the whole patient group. Moreover, worse performance according to the IP correlated with activations in cerebellum and several frontal areas usually involved in motor control in RR-MS group. In addition, voxelwise correlation analysis confirmed the shift of activation depending on the disease stage and complexity of the task. Indeed, for non-dominant hand increased cerebellar activations was again associated with higher IP in RR-MS, whereas a relationship between IP and controlateral frontal regions (right dorsolateral prefrontal cortex) was observed in CIS group when considered alone; negative correlations with other regions involved in motor control were found in ipsilateral frontal areas (left medial prefrontal cortex and left IFG) and basal ganglia (left putamen and pallidum). The associations between IP and MI activation in the patient groups are more distributed and distinct from those previously observed in healthy participants, where worse performance was associated with higher activation in the left superior parietal lobe and better performance was related to increased right prefrontal cortex recruitment (Saiote et al., 2016).

Similarly, the comparison of MM and AM showed that CIS patients activated a more extensive MI network than HC for both dominant and non-dominant side, involving frontal, parietal and subcortical regions. Moreover, they showed a higher activation difference between MM and AM with respect to RR-MS group in regions sub-serving sensory processing and higher level functions. RR-MS differed from HC group in MI with dominant hand principally activating brain structures mainly involved in motor control, whereas, as expected from above considerations, no differences were found between these two groups for the non-dominant hand.

In conclusion, we found that MS patients with different disease courses showed different pattern of neural engagement during MI and that the networks of brain activations depended on the hand used to execute/imagine the movement. The association between brain activations and anisochrony is reflected in the IP suggesting that it could be considered as a personalized measure of actual and mental execution performance, correlated with disease severity. Therefore, due to the MS impact on motor and cognitive domains already after the first attacks, an indicator such as the IP, taking into account not only motor performance but also cognitive and mental aspects, could be useful as a surrogate behavioral marker of MS severity, especially in the early stages of the disease.

AT and CS: substantial contribution to the conception, design, acquisition, analysis and interpretation of the work; drafted and then revised the work after the revision of the coauthors; finally, approved the version to be published; agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. GBr, CC, MAB and GLM: substantial contribution to the interpretation of the work; revised the work; finally, approved the version to be published; agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. GBo and LR: substantial contribution to the acquisition and analysis of the work; revised the work; finally, approved the version to be published; agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. MI: substantial contribution to the conception, design, acquisition, analysis and interpretation of the work; revised the work; finally, approved the version to be published; agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.