A neurophysiological approach to mirror movements in amyotrophic lateral sclerosis

(cid:1) Quantifying mirror activity in ALS patients can be readily achieved using a novel algorithm. (cid:1) Increase of mirror activity seems to precede changes in other measures of motor function and transcallosal inhibition. (cid:1) Further studies are warranted to evaluate the potential utility of this approach as a biomarker for ALS.


Introduction
Performing unilateral motor tasks requires a reduction in the output activity of the ipsilateral primary motor cortex (M1) (Carson, 2005).When the motor programs and inhibitory neural circuits associated with the unilateral voluntary movements are dysfunctional, motor overflow across the midline can occur (Cincotta and Ziemann, 2008).This phenomenon may lead to unintended activation of the contralateral muscles analogous to the activated ones, which has been named in different ways across the literature -motor overflow, mirror activity or mirror movements.There are two suggested pathophysiological mechanisms for this phenomenon (Cincotta andZiemann, 2008, Sehm et al., 2010).The first explanation involves activation of the uncrossed corticospinal projection to the lower motor neurons on the same side of the body (Farmer et al., 1990, Mayston et al., 1997).The second possibility suggests that reduced interhemispheric inhibition leads to the activation of the motor cortex on the opposite side of the body through transcallosal pathways (Cernacek, 1961, Hubers et al., 2008).These are not mutually exclusive explanations.
Mirror movements (MM) may be present in normal children (Cohen et al., 1967), probably due to an immature motor system.In adults, subtle mirroring can be present in healthy subjects (Armatas et al., 1994, Castro et al., 2023) and overt MM are characteristic of a wide range of congenital or neurological conditions (Cincotta and Ziemann, 2008).MM have been described in amyotrophic lateral sclerosis (ALS) (Hubers et al., 2021b, Krampfl et al., 2004, Krampfl et al., 2003, Wittstock et al., 2011, Wittstock et al., 2020, Wittstock et al., 2007) but with variable frequency.It has been proposed that MM in ALS patients occur due to degeneration of callosal pathways (Wittstock et al., 2011, Wittstock et al., 2007).More recently, studies using diffusion tensor imaging and diffusion weighted spectroscopy have shown that functional changes in transcallosal fibers, particularly a decrease in interhemispheric inhibition measured by paired-pulse transcranial magnetic stimulation, occur even without the presence of microstructural alterations (Hubers et al., 2021a, Hubers et al., 2021b, Wittstock et al., 2020).An objective quantification of mirror activity could provide a simple measure of early transcallosal dysfunction in ALS.
Moreover, it has been suggested that a link between MM and executive dysfunction exists in brain injured patients (Tisseyre et al., 2018).Given that executive functions are commonly affected in ALS patients (Kasper et al., 2015), we considered as important to explore this possible correlation in these patients.
Our group has recently developed a simple mathematical algorithm allowing a quick, objective quantification of MM, by recording EMG activity in both the active and mirror muscles during a brief isometric maximal contraction of a given muscle (Castro et al., 2023).
The main purpose of the current study was to apply a new algorithm of MM quantification in a group of ALS patients to evaluate dysfunction of the interhemispheric inhibition.Additionally, we assessed possible correlations of MM with measures of upper motor neuron, transcallosal and cognitive function.

Subjects
This study recruited individuals with probable or definite ALS, as per the Awaji criteria (de Carvalho et al., 2008), from our ALS clinic in Lisbon.These diagnostic criteria require evidence of progressive neurogenic weakness through clinical and/or EMG analysis, and the exclusion of other potential diagnoses.Eligible participants were required to have a minimum of grade 4 strength on the Medical Research Council (MRC) scale for the abductor digiti minimi (ADM) and tibialis anterior (TA) muscle in at least both upper or both lower limbs.In addition, participants had to have normal nerve conduction studies in ulnar and peroneal nerves (motor conduction studies and sensory nerve action potentials).
Individuals who were unable to give informed consent, could not tolerate the recumbent position or electrical stimulation, were not able to cooperate with the required muscle contraction, or had other neurological diseases such as peripheral neuropathy and epilepsy, had metallic brain implants or pacemakers, or signs of ulnar or peroneal nerve lesion, or were taking drugs that could affect cortical excitability, were excluded from this study.
All patients had been stabilized on riluzole for more than one month at the time of assessment.

Ethics approval and consent to participate
This work was approved by the Ethics Committee of the Lisbon Academic Medical Center (protocol 385/20).The study was performed according to the Declaration of Helsinki, and written informed consent was obtained from all participants.

Clinical assessment
Clinical and EMG assessments (MdeC) were carried out prior to the neurophysiological investigation, which was always scheduled within one month of the initial clinical evaluation.A clinical upper motor neuron (UMN) score, as described by Geraldo et al. (Geraldo et al., 2018), was derived: the total UMN score for each patient ranged from 0 to 31 (maximal abnormality).Functional disability was assessed on the same day using the ALSFRS-R scale (maximum healthy score 48) (Cedarbaum et al., 1999).

Cognitive assessment
As part of our clinical routine, we evaluate cognition in ALS patients.We use the validated Portuguese version of the Edinburgh Cognitive and Behavioral ALS Screen (ECAS) (Tomsic, 2015), which is a multi-domain neuropsychological battery specifically developed for assessing cognitive status in ALS patients (Abrahams et al., 2014).This battery assesses domains that are described as impaired in ALS (executive functions, social cognition, verbal fluency, and language), but also other domains (memory and visuospatial performance).

Mirror movements
Patients were positioned supine in a silent room, with their arms extended along their body and hands resting comfortably on the bed.They were instructed to perform brief isometric full force contractions in only one hand (fifth finger abduction, abductor digit minimi, ADM) or one foot (foot dorsiflexion, tibialis anterior, TA), which was considered the active muscle.The subjects were instructed to maintain the contralateral part of the body fully relaxed.Muscle relaxation was monitored by the device loudspeaker to ensure that there was no muscle contraction outside the experimental paradigm.Each subject performed three trials (2-3 seconds maximal contraction and 15 seconds resting) in each muscle of one side, followed by the same protocol on the opposite side.The order of muscle (ADM or TA) and side (right or left) contraction was randomly assigned between subjects.The recordings for the 4 muscles took less than 10 minutes per subject.
Surface electrodes (reference 9013L0203, Natus Inc) were used for recording EMG activity.For the upper limb, recordings were made with the active electrode over the belly of the ADM, while the reference electrode was placed on the palmar side of the proximal inter-phalangeal joint of the 5th finger.The ground electrode was placed on the wrist.For the lower limb, recordings were performed with the active electrode over the belly of the TA muscle and the reference electrode 5-7 cm distally, over the tibial bone.The ground electrode was placed on the ankle.Standard amplifier filter settings of 30-Hz and 10-kHz were used.Signals were digitized at a sampling frequency of 3 kHz.Recordings were made on a 10 second window and stored for offline analysis.
Mirror activity was quantified using a custom built MATLAB algorithm (MATLAB R2018a, The Mathworks, Inc., Natick, Massachusetts) developed by our group, and published elsewhere (Castro et al., 2023).Briefly, our algorithm detects the starting and finishing points of the active muscle contraction and estimates the amount of EMG signal between those two points by calculating the area under the curve for both the active and the mirror mus-cles.The amount of mirror activity is then expressed as a percentage of the EMG signal of the active muscle.The algorithm is accessible at https://ars.els-cdn.com/content/image/1-s2.0-S030439402300143X-mmc1.pdf.

Transcranial magnetic stimulation (TMS)
A Magpro x100 (MagVenture, Inc, Alpharetta, Georgia USA) was used for TMS.Stimulation was performed over the contralateral hand and lower limb muscle cortical areas, defined in preliminary recordings by the lowest resting motor threshold (RMT), using a round coil (inner diameter 35 mm; outer diameter 121 mm).Motor evoked potentials (MEP) were obtained using the recording settings described above.RMT was calculated for each muscle, defined as the minimum stimulus intensity needed to elicit at least 50 % of responses with a minimum amplitude of 0.1 mV (Rothwell et al., 1999).Stimulation was set at 120 % of RMT.Ten consecutive traces were recorded in each muscle, separated by more than 30 seconds interval, with the target muscle at rest, as controlled by audio monitoring.Recordings with artefact or noise from adjacent muscle contraction were discarded.The motor evoked potential (MEP) with highest peak-to-peak amplitude was selected to define motor latency and motor evoked amplitude.Amplitude was recorded as a ratio between MEP and compound muscle action potential (CMAP) peak-to-peak amplitudes.Central motor conduction (CMCT) was calculated using the F-wave method (Robinson et al., 1988), by stimulating the ulnar nerve at wrist and the peroneal nerve at fibula (20 supramaximal stimuli,1 Hz).Taking into account normative data from our laboratory (de Carvalho et al., 2003), TMS was considered abnormal for the studied limb if one of the following were met: no reproducible MEP; MEP amplitude below 5 % of CMAP amplitude; CMCT longer than 8 ms for the ADM or 16 ms for the TA.

Ipsilateral silent period (iSP)
We assessed iSPs in both ADM muscles, using the same TMS setting.Stimulation was applied, during maximum voluntary contraction of the ipsilateral ADM, and recording 6 to 8 responses, allowing for a 20 second rest between trials.Traces were recorded for offline analysis, and then rectified and averaged.The latencies of the iSP were derived using the mathematical method proposed by Garvey et al (2001).The onset latency was calculated from the cortical stimulus to a consistent fall in the amplitude of the EMG trace to less than the mean amplitude of the 100 ms of EMG preceding the stimulus -[MCD*2.66].The duration of the iSP was calculated from its onset latency to the return of the EMG signal to this mean pre-stimulus EMG amplitude -[MCD*2.66].iSPs were classified as abnormal when either onset latency or duration were outside the normal limits for our laboratory (derived from investigating 20 gender and age-matched controls, unpublished): latency longer than 48.8 ms; duration shorter than 12.4 ms or longer than 61.8 ms).

Statistical analysis
Descriptive data is shown with mean values and standard deviations, or median and first and third interquartile (1st and 3rd IQR), as appropriate.Data distribution was assessed with the Shapiro-Wilk test and, since most neurophysiological results did not follow a normal distribution, we subsequently used non-parametric tests.We considered p values lower than 0.05 as significant.Correlations between variables were tested using the Spearman correlation coefficient.Comparisons between 2 groups were performed using the Mann-Whitney U test.The Kruskal-Wallis H test was used for comparisons of more than 2 groups.Post-hoc pairwise compar-isons were performed using Dunn's procedure.False discovery rates for multiple comparisons were controlled by the Benjamini-Hochberg procedure (Benjamini and Hochberg, 2018), using a false discovery rate of 5 %.

Results
Forty-one ALS patients, 13 women (32 %), with a median age of 59 years (IQR 52.0-71.0),and a median disease duration of 14 months (IQR 9.0-24-0), were studied.The median ALSFRS-R at the time of the assessment was 44.0 (IQR 40.0-45.0).Thirty-nine patients were right-handed, as assessed by the Edinburgh Handedness Inventory (Oldfield, 1971), 1 patient was left-handed and 1 was ambidextrous.These patients were early affected patients consecutively recruited in our Unit.Detailed characteristics of the patients included are displayed in Supplementary Table S1.

Mirror activity
Measurements of mirror activity were obtained from 72 ADM and 56 TA muscles (Table 1), after excluding results from cervical or lumbo-sacral segments in which at least one muscle was weak (MRC < 4, see methods).
There were no significant differences in the amount of mirror activity between genders.There was also no significant correlation between mirror activity and age, disease duration, UMN score and ALSFRS-R, as assessed by the Spearman rank-order correlation (p > 0.05 for all correlations) -see Supplementary Table S2.

Cognitive data
In a subgroup of 19 ALS patients (13 male; median age 61; IQR 52-71 years) neuropsychological status was measured with the ECAS battery (median disease duration at evaluation 16; IQR 9-34 months).The subjects' ECAS total score ranged between 21 and 128 with a median of 97 (IQR 71-117.5),while the median ALS-Specific score (executive functions and social cognition; fluency; language) was 71 (IQR 49.5-85.5).In this group of patients, 52.6 % were found to have cognitive impairment on the ECAS total score and 57.9 % on ALS-specific score.
When comparing the amount of mirror activity between subjects with or without cognitive impairment, we found no significant differences in both muscles, both on the ECAS total score (p = 0.545 for the ADM; p = 0.781 for the TA) and on the ECAS ALS-specific score (p = 0.736 for the ADM; p = 0.899 for the TA).

Transcranial magnetic stimulation
Conventional TMS was performed in all subjects.Values for RMT, amplitude and CMCT are displayed in Table 2.A reproducible motor evoked response was absent in 7 ADM and 15 TA of the assessed muscles.Abnormal TMS responses were found in 68 % of ADM and 52 % of TA muscles.
We compared the amount of mirror activity from limbs with normal and abnormal TMS results, and with our normative data from healthy subjects investigated with the same method (Castro et al., 2023) (Fig. 2).The Kruskal-Wallis H test showed significant differences between the 3 groups, both in ADM (v2(2) = 43.606,p < 0.001) and in TA (v2(2) = 39.162,p < 0.001).Subsequent pairwise comparisons (Table 3) disclosed significant differences between all groups, both in ADM and TA muscles.Patients with   3.
normal TMS results had more mirror activity than healthy subjects, while patients with abnormal TMS results had more mirror activity than patients with normal TMS results.No differences in age, disease duration, strength or CMAP amplitude were seen between patients with normal and abnormal TMS results.

Ipsilateral silent period
iSPs were studied in 70 ADM muscles.Values of onset latency and duration are shown in Table 2.A reproducible iSP was absent in 18 (26 %) of the assessed ADM muscles.Considering our norma-tive values, abnormal iSPs were found in 36 (51.4 %) ADM muscles (absent iSP was considered an abnormal result).
We categorized muscles into 2 groups according to the presence or absence of iSP abnormalities (Fig. 3).We compared the amount of mirror activity from muscles with normal and abnormal iSP values, and with data from our healthy subjects group (Fig. 3).The Kruskal-Wallis H test showed significant differences between the 3 groups (v2(2) = 35.577,p < 0.001).Subsequent pairwise comparisons disclosed significant differences between groups, as shown in Table 4. Patients with normal iSP values had more mirror activity than healthy subjects, while patients with abnormal iSP values had more mirror activity than patients with normal iSP values.4.
No differences in age, disease duration, strength or CMAP amplitude were seen between patients with normal and abnormal iSPs.
Fig. 4 displays an illustrative example of an absent and a normal iSP, with the mirror activity of the corresponding ADM muscle.

Discussion
In this work we studied a group of 42 ALS subjects, using a novel algorithm that allows the quantification of mirror activity during a brief maximal isometric contraction.All patients were in a high functional status (median ALSFRS-R = 44), with normal CMAP amplitudes in the assessed muscles.Additionally, we only included patients with strong muscles (MRC 4), in at least one cervical or lumbo-sacral region.
The experiments were well tolerated by all patients.The algorithm used performed significantly well and was able to ignore the presence of brief bursts of EMG activity (e.g., fasciculations).
To the best of our knowledge, this is the first attempt to systematically quantify EMG signal in MM in ALS patients.Similarly to what has been reported in the literature regarding clinical MM (Hubers et al., 2021b, Krampfl et al., 2004, Krampfl et al., 2003, Wittstock et al., 2011, Wittstock et al., 2020, Wittstock et al., 2007), we found significantly increased mirror activity in ALS subjects in both upper and lower limbs.Interestingly, contrary to our  previous findings in healthy subjects (Castro et al., 2023) and to what has also been reported using clinical assessment (Armatas et al., 1994(Armatas et al., , 1996)), we did not find differences in mirror activity between sides in ALS subjects.Further analysis showed that mirror activity was significantly increased in both upper and lower limbs of ALS patients with abnormal TMS results, but also in upper limbs with abnormal iSP, as compared with patients with normal results.However, even in muscles from ALS patients with normal TMS and iSP values, mirror activity was also significantly increased when compared to controls.Mirror movements may be seen in children, but they are usually considered pathological in adulthood.However, as reported before (Armatas et al., 1994, Castro et al., 2023, Hubers et al., 2008) our results show that some healthy subjects have minor mirror activity.Previous investigation on transcallosal inhibition by TMS revealed that this phenomenon is very consistent but the amount of inhibition is variable between subjects (Ferbert et al., 1992), indeed the magnitudes of transcallosal inhibition and MM are inversely correlated in healthy subjects (Hubers et al., 2008).In healthy subjects, there is an increased amount of mirror activity in the dominant side (Castro et al., 2023), which has been attributed to an asymmetry in transcallosal connections (Armatas et al., 1996).
Our results support early inhibitory transcallosal pathway dysfunction in ALS patients (Castro et al., 2021, Filippini et al., 2010, Hubers et al., 2021a).The loss of left/right differentiation in the amount of mirror activity in ALS, both in upper and lower limbs, opens the possibility that functional changes in inhibitory pathways occur asymmetrically in ALS.This inhibitory transcallosal pathway dysfunction favors mirror phenomenon (Cernacek, 1961, Hubers et al., 2021b), and is associated with abnormal iSPs.Transcallosal inhibition through pathways crossing the posterior corpus callosum is the mechanism of the iSPs as demonstrated by several groups (Hupfeld et al., 2020, Meyer et al., 1998, Netz et al., 1995).Moreover, our findings are consistent with the concept that transcortical inhibition is associated with cortical motoneuronal loss, since TMS changes are related to stronger MM.Our earlier studies investigating the silent period after cutaneous stimulation have indicated that delayed onset-latency is associated with clinical and TMS signs of UMN lesion (Castro et al., 2023, Castro et al., 2021).In addition, in patients without clinical signs of UMN lesion, altered transcallosal inhibition causes onset-latency shortening by ipsilateral cortical activation (Castro et al., 2021).These neurophysiological findings are consistent in associating dysfunction of transcallosal inhibition with UMN dysfunction, in ALS.
Imaging and TMS studies have suggested that, in ALS, there is an early impairment of callosal pathways, in addition to the involvement of the primary motor cortex (Caiazzo et al., 2014, van den Bos et al., 2021).Recent studies observing clinical signs of MM have shown that functional deficits precede structural changes of the corpus callosum (Hubers et al., 2021b, Wittstock et al., 2020).The present work strengthens the possibility that MM may be a simple early marker of callosal dysfunction.Additionally, the increase in mirror activity in ALS patients precedes changes in corticospinal tract function, as assessed by TMS, possibly suggesting early UMN lesion.
Cognitive changes in ALS have been correlated with structural changes in the brain (Trojsi et al., 2019), including in the corpus callosum (Chenji et al., 2021).In our cohort, these differences did not reach statistical significance, probably reflecting impairment of different cortical mechanisms.
In our population, UMN score did not influence MM activity or iSP.In ALS, however, the motor system degeneration has some particularities as the relative preservation of the inhibitory corticoreticulospinal pathway, and the involvement of the spinal cord interneurons and gamma motor neurons, which influence the clinical observation of UMN signs, as extensively discussed elsewhere (Swash, 2012, Swash et al., 2020).
The present study has some limitations.We did not perform longitudinal evaluations of mirror activity, but we found no correlation between the physiological findings and disease duration.Although longitudinal studies would be useful to assess the effect of disease progression on the amount of mirror activity, the rapid progression of weakness would demand recruitment of a large population of patients.
In conclusion, this work demonstrates that the EMG signal of MM can be quantified by using a simple paradigm, allowing an objective quantification of mirror activity, which seems to correlate with early transcallosal dysfunction.This approach merits further studies, in order to assess its potential use as a measure of disease progression and as a biomarker of disease in ALS.

Fig. 3 .
Fig. 3. Comparison of mirror activity between upper limbs with vs without ipsilateral silent period abnormalities.ADM -Abductor digiti minimi; iSP -Ipsilateral silent period; Bars represent mean ± 2 standard error of mean; Pairwise group comparisons are shown in Table4.

Fig. 4 .
Fig. 4. Illustrative example of absent and normal ipsilateral silent period, with the corresponding mirror activity.A -Amyotrophic lateral sclerosis (ALS) subject with absent ipsilateral silent period (iSP) (top trace) on the right abductor digiti minimi (ADM), contraction of the left (active) ADM (middle trace), and clear mirror activity on the right (mirror) ADM (bottom trace).B -ALS subject with normal iSP (top trace) on the right ADM, contraction of the left (active) ADM (middle trace), and less mirror activity on the right (mirror) ADM (bottom trace).

Table 1
Amount of mirror activity in amyotrophic lateral sclerosis subjects, and percentage of muscles with abnormal mirror activity considering previously published normative values.
(Castro et al., 2023)minimi; TA -Tibialis anterior; Mirror activity is expressed as a percentage of the EMG signal of the active muscle (see Methods); *Cut-off values published previously(Castro et al., 2023); Values are median and IQR.

Table 2
Transcranial magnetic stimulation and Ipsilateral Silent Period values for amyotrophic lateral sclerosis subjects.

Table 3
Comparison of mirror activity between groups according to transcranial magnetic stimulation results (pairwise comparisons using Dunn's procedure).Abductor digiti minimi; TA -Tibialis anterior; ALS -Amyotrophic lateral sclerosis; TMS -Transcranial magnetic stimulation; ms -milliseconds; Mirror activity is expressed as a percentage of the EMG signal of the active muscle (see Methods); Values are median and IQR; p values shown are adjusted values using the Benjamini-Hochberg procedure (FDR 5%).

Table 4
Comparison of mirror activity between groups according to ipsilateral silent period results (pairwise comparisons using Dunn's procedure).Abductor digiti minimi; ALS -Amyotrophic lateral sclerosis; iSP -ipsilateral silent period; Mirror activity is expressed as a percentage of the EMG signal of the active muscle (see Methods); values are median and IQR; p values shown are adjusted values using the Benjamini-Hochberg procedure (FDR 5%).