Novel approaches to assessing upper motor neuron dysfunction in motor neuron disease/amyotrophic lateral sclerosis: IFCN handbook chapter

(cid:1) Transcranial magnetic stimulation is an objective biomarker of upper motor neuron dysfunction in motor neuron disease (MND). (cid:1) Cortical hyperexcitability is mediated by cortical disinhibition and increased facilitation. (cid:1) Neuroimaging biomarkers of upper motor neuron dysfunction in MND exhibit utility.


Introduction
In motor neuron disease (MND), the highly evolved human nervous system undergoes progressive degeneration of its primary motor pathways, characterised by dysfunction of the upper [UMN] and lower motor neurons [LMN] (Geevasinga et al., 2016c, Kiernan et al., 2011, Vucic et al., 2023b).While a focal disease onset is typical of MND, involving the limbs, bulbar or respiratory regions, the progressive course affects other body regions to result in global muscle wasting and weakness, with respiratory dysfunction representing the terminal phase of MND (Dharmadasa et al., 2017, Vucic et al., 2014).Disease origin has been a long-standing debate, with three main hypotheses proposed: (i) a cortical onset, whereby descending corticomotoneuronal tracts mediate LMN degeneration through an anterograde glutamatergic excitotoxic mechanism (Eisen et al., 2017, Eisen et al., 1992),(ii) LMN onset and secondary dysfunction of UMN pathways (Boillee et al., 2006, Fischer et al., 2004, Pun et al., 2006, Williamson and Cleveland, 1999), or (iii) independent degeneration and dysfunction of the upper and lower motor neurons in a contiguous and random pattern (Ravits et al., 2007, Ravits andLa Spada, 2009).Ultimately, the relationship between dysfunction of the upper and lower motor neurons is fundamental to the diagnosis and understanding of MND pathophysiology.
Development of objective biomarkers of upper and lower motor neuron dysfunction in MND has gone a long way towards clarifying pathogenic mechanisms, as well as heralding diagnostic and prognostic biomarkers.In turn, such concepts would potentially enable identification of novel therapeutic targets, lead to development of treatment strategies, and enhance patient stratification and monitoring in clinical MND trials.The present review will discuss the utility of TMS and neuroimaging biomarkers in MND, focusing on recently developed TMS techniques and advanced neuroimaging modalities that interrogate structural and functional integrity of the corticomotoneuronal system, with an emphasis on their pathogenic, diagnostic, and prognostic utility.

Physiological biomarkers using TMS
Transcranial magnetic stimulation (TMS) uses electromagnetic induction to non-invasively and relatively painlessly stimulate the human brain (Fig. 1A).The technique of TMS still uses fundamentally similar principles to when it was first described in the 1980s (Barker et al., 1985) to interrogate the integrity and excitability of the descending corticomotoneuronal system, with refinement of methodology and stimulus paradigms now enabling these studies to become more robust, more insightful, and of potentially greater clinical value.
The TMS stimulator passes a brief but large current through a coil that is positioned over the scalp, generating a strong timevarying electromagnetic field (with a peak field up to $2 Tesla) perpendicular to the coil (Siebner et al., 2022).This magnetic field induces a secondary electric field in the underlying brain tissue which preferentially activates the axons of excitatory intracortical interneurons, stimulating pyramidal neurons trans-synaptically (Fig. 1B) (Vucic et al., 2023b).At higher stimulation intensities, corticospinal axons are also activated.At a peripheral level, a motor evoked potential (MEP) recorded from a target muscle (e.g., intrinsic hand or leg muscle) reflects the discharged activity of the descending corticospinal volleys.The activation of distinct neuronal elements depends on multiple factors, such as coil shape and field depth (focal [figure of eight] versus non-focal [circular]; superficial [figure of eight and circular] versus deep [cone]), direction of induced cortical currents, pulse waveform (monophasic vs biphasic), number of pulsed stimuli (e.g.single versus multiple [paired-pulse]), and the strength of stimulation (subthreshold vs threshold) (Corp et al., 2021, Di Lazzaro et al., 2002b, Di Lazzaro and Rothwell, 2014, Pavey et al., 2023, Rossini et al., 2015, Rossini et al., 2019, Siebner et al., 2022, Sommer et al., 2018).
TMS is now well established as a tool to measure UMN dysfunction and as a biomarker in MND.This has led to a greater understanding of MND pathophysiology and the development of novel diagnostic approaches (Vucic et al., 2023b), with threshold tracking Fig. 1.Transcranial magnetic stimulation and principles of paired-pulse threshold tracking technique: (A) Transcranial magnetic stimulation using a figure-of-eight coil is applied over the primary motor cortex (M1) and elicits a motor evoked potential (MEP) from a target muscle.(B) Descending corticomotoneuronal pathways from the precentral gyrus (M1) contribute to the MEP response.Direct neuronal activation occurs in the lip/rim regions of the M1 hand knob region.Spread of TMS current to the rostral and caudal parts of the M1, via cortico-cortical synaptic transmission, further contributes to generation of indirect waves (I-waves).Fast-conducting monosynaptic corticomotoneuronal neurons are predominantly located in the caudal M1 (BA4p) region compared to the rostral M1 (BA4a).The exact transition between rostral M1 and caudal dorsal premotor cortex (PMd) in the lip/rim region of the gyrus is gradual and varies across subjects.Additional corticospinal pathways may be activated by TMS via excitation of postcentral primary somatosensory cortex (S1) and its cortico-cortical projections to rostral/caudal M1. (C) The paired pulse TMS paradigm is illustrated.Channel 1 records an unconditioned test stimulus, defined as TMS intensity required to generate and maintain the tracking target, signifying the resting motor threshold (RMT).Channel 2 monitors the subthreshold conditioning stimulus and does not generate an MEP response, while channel 3 records the conditioned-test stimulus at interstimulus intervals of 1-30 ms.The interstimulus intervals can be advanced serially or in a parallel (pseudo-random) manner.(D) In the paired pulse threshold tracking TMS technique, a target MEP of 0.2 mV (±20%) tracked as it lies in the steepest portion of the stimulus response curve.(E) When the MEP response is larger than the tracking target (potential-1) the subsequent stimulus intensity is reduced, while if the MEP response is smaller than the tracking target (potential-2), the subsequent stimulus intensity is reduced.As such, much larger variations in MEP amplitude intensity translate to smaller variations in thresholds (TMS intensity).(F) Short interval intracortical inhibition (SICI) is represented as increased conditioned-test stimulus intensity required to generate and maintain the tracking target (0.2 mV), developing between interstimulus intervals of 1-to-7 ms.Intracortical facilitation (ICF) is represented as reduced conditioned-test stimulus intensity required to generate and maintain the tracking target.In MND patients, SICI is reduced while ICF is increased signifying cortical hyperexcitability.(Figure adapted from Vucic and colleagues.Clinical diagnostic utility of transcranial magnetic stimulation in neurological disorders.Updated report of an IFCN committee. Clin Neurophysiol 2023;150:131-17).
paradigms emerging as a potential diagnostic technique (Menon et al., 2015a).Single and paired-pulse TMS techniques are now also being integrated into clinical trials as biomarkers of therapeutic efficacy (Wainger et al., 2021).The following section on TMS will discuss the utility of different TMS techniques as biomarkers in MND, providing a description of specific single, paired, and triple pulse techniques and their clinical applicability as pathogenic, diagnostic, and prognostic biomarkers.

TMS techniques
2.1.1.Single-pulse TMS biomarkers 2.1.1.1.Motor threshold.Motor threshold (MT) measures the integrated excitability across the cortical motor system and gives a functional evaluation of the pyramidal system by reflecting the ease with which corticomotoneurons are excited (Vucic et al., 2023b).Traditionally, relative frequency methods define the resting motor threshold (RMT) as the lowest TMS intensity required for eliciting an MEP amplitude of !50 lV in 50% of 10 trials within a target muscle at rest (Rothwell et al., 1999), or if using threshold tracking methods (see later), as the stimulus intensity required to maintain an MEP amplitude of 0.2 mV [±20%] (Fisher et al., 2002, Vucic et al., 2006).RMTs differ across regions of the healthy motor cortex and are smallest for intrinsic hand muscles, particularly influenced by the superficial depth of the target motor region with respect to the scalp and the higher density of corticospinal projections onto spinal motor neurons (Chen et al., 1998).The active MT (AMT) is measured during slight isometric tonic muscle contraction and has been defined as the lowest intensity required to elicit an MEP amplitude !200 lV.MT can also be measured using adaptive threshold hunting methods, which models the probabilistic nature of MT and the relationship between TMS intensity and MEP amplitude using an S-shaped metric function (Awiszus, 2003a, 2003b, Rossini et al., 2015).The maximal individual difference between the adaptive and relative frequency methodologies is reported to be up to 5% of maximal stimulator output, with higher thresholds generated using adaptive methods (Silbert et al., 2013).Importantly, motor thresholds are comparable between genders and dominant and non-dominant hemispheres (Livingston et al., 2010), but are not static over time, dynamically change with central nervous system (CNS) maturation and aging.Typically, they are highest in infants, falling to adult values by the second decade of life and increasing again in older adults (Bashir et al., 2014).
In MND, the findings relating to changes in MT have been heterogenous, but there is now emerging consistency of low thresholds in the early clinical disease course that precedes muscle wasting (Vucic et al., 2023b) and is associated with profuse fasciculations and hyper-reflexia (Eisen andWeber, 2001, Mills andNithi, 1997), indicative of cortical and/or spinal motor neuron hyperexcitability (Vucic et al., 2023b).When this is evident, the reduction in MTs is more prominent over the dominant motor cortex, contralateral to site of disease onset, and may contribute to disease spread (Menon et al., 2017).Others have documented normal MTs (Kohara et al., 1996, Menon et al., 2017, Mills and Nithi, 1997, Vucic and Kiernan, 2006, Vucic and Kiernan, 2007, Vucic et al., 2008, Zanette et al., 2002b), increased MTs or inexcitable motor cortices (no motor response even at maximal TMS intensity) (Attarian et al., 2005a, Berardelli et al., 1991, de Carvalho et al., 2003, Eisen et al., 1990, Miscio et al., 1999, Triggs et al., 1992, Triggs et al., 1999, Urban et al., 2001).An increase in MT has also been reported in atypical MND phenotypes such as primary lateral sclerosis, differentiating this from mimicking diseases such as hereditary spastic paraplegia (Agarwal et al., 2018, Geevasinga et al., 2015a).In certain racial backgrounds such as Asian populations, healthy subjects have inherently higher resting thresholds than in European populations, translating to higher thresholds in disease (Suzuki et al., 2022).The finding of an inexcitable motor cortex appears to be present only in a minority ($10%) of typical patients at their first clinical presentation, which, if across all four-limbs within the first year of symptom onset, is a biomarker of an adverse prognosis and a faster rate of disease progression (Dharmadasa et al., 2021).Overall, the discordant findings may relate to assessing patients at differing clinical disease stages, underpinned by clinical and pathological heterogeneity and the variable rates of disease progression.
As the disease progresses, the corticomotoneuronal pool is slowly depleted and becomes more difficult to stimulate, leading to increasing MTs and finally, inexcitability of the motor cortex (Mills and Nithi, 1997), although this may not be an invariable finding (Shibuya et al., 2017b).Racial background may impact on MTs, with relatively higher values evident in Japanese MND patients (Suzuki et al., 2022).Additionally, reduced MT was reported to be an independent predictor of cognitive dysfunction in MND (Agarwal et al., 2021).
2.1.1.2.The stimulus-response curve.The summation of descending corticospinal volleys onto the spinal motor neuron after a single TMS pulse to M1 is recorded as a MEP.In healthy subjects, MEP amplitude increases gradually with an increase in stimulus intensity, and this stimulus-response (SR) relationship can be formally modelled by a cumulative Gaussian sigmoid curve.In comparison to MTs, the SR curve and MEP amplitude probably assesses the function of motor cortical neurons that are less excitable or spatially distant from the center of the target TMS field (Chen andGarg, 2000, Hallett et al., 1999).The curve is different for each target muscle and is subject to dynamic changes according to the physiological state of the motor system, resulting in a right-or leftwards shift with changes in corticospinal excitability (Möller et al., 2009, Pitcher et al., 2003).The gradient of the curve is determined by the activation of corticospinal neurons and strength of their projections onto target muscles, and positively correlates with glutamate levels in the motor cortex, increasing with enhanced adrenergic neurotransmission and decreasing with any agents that enhance GABAergic effects (Ziemann et al., 2015).Consequently, steeper SR curve gradients suggest a higher level of excitation or lower level of inhibition of the descending corticospinal tracts and are evident in muscles with lower MTs, such as intrinsic hand muscles (Chen et al., 1998).Importantly, the MEP has different neurobiological properties to the peripherally evoked motor response, being more polyphasic, more prolonged and significantly smaller than the maximal compound muscle action potential amplitude, a finding related to greater desynchronized excitation of motor units in the target muscle from descending corticospinal volleys, with consequent phase cancellation (Rosler et al., 2002).The MEP amplitude is therefore typically expressed as a percentage of the maximal compound muscle action potential (CMAP) peripheral response (i.e., MEP/CMAP), to provide insight into the UMN contribution to the MEP.
A potential limitation of MEP amplitude as a biomarker in MND relates to the highly variable trial-to-trial amplitude of this measure (Kiers et al., 1993).The reasons for this are multifactorial, related to target muscle contraction (Darling et al., 2006), intrinsic fluctuations of neuronal excitability at cortical and spinal cord levels (Rossini et al., 2015), and the timing of TMS stimulus application in relation to cortical oscillatory states (Metsomaa et al., 2021).Regarding the latter, delivering the TMS pulses at the optimal phase of individualized b oscillation has been reported to reduce the extent of this MEP variability (Torrecillos et al., 2020)..
In MND, increased MEP amplitude has been reported as a consistent early feature (Geevasinga et al., 2017, Geevasinga et al., 2015b, Menon et al., 2015a, Menon et al., 2014, Vucic and Kiernan, 2006, Vucic and Kiernan, 2007).The stimulus-response curve typically shows an early left-ward shift, supporting an increase in glutamatergic activity and the presence of corticomotor hyperexcitability (Menon et al., 2017, Vucic and Kiernan, 2006, Vucic and Kiernan, 2010).While it could be argued that this increased MEP amplitude represents a plasticity phenomenon in MND, this seems less likely given that MEP amplitudes are not increased in MND mimicking disorders, despite a comparable degree of LMN dysfunction (Menon et al., 2015a).
2.1.1.3.Triple stimulation technique.The triple stimulation technique (TST) is a collision method with increased sensitivity for detecting corticomotoneuronal function, (Magistris et al., 1999).The TST circumvents the limitation of MEP desynchronization, thereby reducing MEP variability and improving amplitude (Magistris et al., 1999, Magistris et al., 1998, Rosler et al., 2002, Z'Graggen et al., 2005).Briefly, the TST consists of three successive stimuli delivered at precisely defined ISIs, measuring response to the intrinsic hand or foot.The TMS pulse (first stimulus) is delivered at high intensity over the motor cortex, followed by the first electrical stimulus (2nd TST stimulus) delivered to the distal segment of the nerve innervating the target muscle.A third electrical stimulus (3rd TST stimulus) is then delivered proximally (at either Erb's point or the gluteal fold after a period of delay.The TMS induced descending volley collides with and 2nd antidromic stimulus leading to cancellation.Subsequently, the 3rd stimulus elicits a synchronous discharge resulting in a supramaximal compound muscle action potential (CMAP).Several parameters of the resultant CMAP response, such as amplitude and area, are compared to the response induced by the conditioned TST paradigm, resulting in amplitude and area ratios that estimate the proportion of surviving corticospinal motor neurons.When there is dysfunction of corticospinal tracts, the TMS induced descending discharges fail to collide with the 2nd antidromic peripheral stimulus, resulting in smaller CMAP responses.
The TST response is compared to a control curve obtained by triple stimulation performed solely on the peripheral nerve [Erb's point (1st stimulus)-to-wrist (2nd stimulus)-to Erb's point (3rd stimulus)] paradigm.The proportion of spinal motor neurons activated by the TMS is quantified by calculating the amplitude ratio between TST test (TMS 1st stimulus) and TST control (peripheral stimulation only) curves.A diagnostic cut-off using TST amplitude and area ratios > 93% have shown good test-to-test reliability (Buhler et al., 2001, Humm et al., 2004, Magistris and Rösler, 2003, Magistris et al., 1999), with commercial software now available to facilitate the translation of TST into clinical practice.
In MND, the TST technique can be utilized to detect corticomotoneuronal degeneration and UMN dysfunction, even at a subclinical stages of the disease (Kleine et al., 2010, Komissarow et al., 2004, Rösler et al., 2000, Wang et al., 2019).In a recent multicenter study, TST amplitude reduction was reported in 62% of MND patients and was more sensitive than reduction in MEP amplitude or prolonged central motor conduction time (Grapperon et al., 2021), correlating with clinical measures of functional disability.
2.1.1.4.The cortical silent period.The interruption of voluntary electromyographic (EMG) activity in the target muscle and the subsequent period of electric silence following a single suprathreshold TMS pulse to M1 is described as the cortical silent period (CSP) [Fig.2A] (Cantello et al., 1992).The duration of this period gradually increases with TMS intensity, but is not impacted by strength of target muscle contraction (Inghilleri et al., 1993, Kimiskidis et al., 2005) nor amplitude of the facilitated MEP response (Triggs et al., 1992).Muscle fatigue can inadvertently prolong the duration of CSP, so keeping muscle contraction at a low level is recommended to avoid the effects of fatigue (Hunter et al., 1985).
In addition to upper and lower limbs (Ziemann et al., 1993), the CSP may be recorded from a variety of other muscle targets including from the face (Werhahn et al., 1995), diaphragm (Lefaucheur and Lofaso, 2002) and sphincter (Lefaucheur, 2005).CSP duration appears to be influenced by density of corticomotoneuronal projections to the spinal motor neuron, with duration longest for distal upper limb muscles (50-300 ms in hand muscles).A preceding MEP response may not be always evident prior to CSP (Trompetto et al., 2001), suggesting that mechanisms underlying CSP are different to those mediating MEP.Although the physiolog- ical mechanisms remain to be fully elucidated, spinal inhibitory mechanisms appear to contribute to the earliest phase (first 50 ms), including Renshaw cell and IA inhibitory afferent activation as well as refractoriness of spinal motor neurons (Fuhr et al., 1991, Pierrot-Deseilligny and Burke, 2012, Rossini et al., 2015), while the predominant input to CSP is from the later segments, which is of cortical origin (Vucic et al., 2023b), mediated by cortical inhibitory circuits in M1 and GABAergic neurotransmission acting via GABA B receptors (Classen andBenecke, 1995, Stetkarova andKofler, 2013).As such, CSP is largely a measure of motor cortical inhibition and therefore usually only altered by cortical mechanisms, which makes it a useful measure to detect intracortical excitability changes in disease.
The ipsilateral silent period (iSP) is induced by ipsilateral motor cortex stimulation (i.e., on the same side of the contracting muscle), which interrupts voluntary EMG activity in ipsilateral muscles (Chen et al., 2008).The iSP commences 30-40 ms after the stimulus, typically lasting for 20-25 ms (Meyer et al., 1995).This phenomenon reflects transcallosal inhibition (Meyer et al., 1995) with some contribution from non-callosal pathways that are caudal to the corpus callosum (Compta et al., 2006).For an optimal iSP recording, TMS intensity should be set to at least 60% of maximum stimulator output and a low-level of muscle contraction is recommended to avoid fatigue (Hupfeld et al., 2020, Meyer et al., 1995).The iSP onset latency and duration, along with rectified area, are typically measured (Kuo et al., 2017).
In MND patients, a significant reduction in CSP duration (Fig. 2B), and in some cases an absence of CSP, has been reported in the early stages of the disease process (Desiato and Caramia, 1997, Geevasinga et al., 2017, Geevasinga et al., 2015a, 2015b, Grieve et al., 2015, Mills, 2003, Prout and Eisen, 1994, Siciliano et al., 1999, Vucic and Kiernan, 2006, Vucic and Kiernan, 2008, Vucic et al., 2010, Wittstock et al., 2007, Zanette et al., 2002b), regardless of family history or specific MND clinical phenotype (Menon et al., 2016, Vucic andKiernan, 2007).This appears to be a specific TMS biomarker of MND when compared to neuromuscular mimicking diseases (including those with pyramidal features), with modest diagnostic utility (Menon et al., 2015a, Vucic et al., 2011b, Vucic and Kiernan, 2008, Vucic et al., 2010).A possible explanation for this excitability phenomenon in MND is the degeneration and dysfunction of longer latency cortical GABAergic inhibitory circuits acting via GABAB B receptors, although a contribution of spinal inhibitory circuits cannot be discounted.Interestingly, abnormalities in ipsilateral CSP duration (i.e.absent or prolonged) have also been identified early in the course of the clinical disease, potentially reflecting degeneration of the transcallosal glutamatergic fibres that project onto inhibitory interneurons in the non-stimulated motor cortex (Wittstock et al., 2007), a notion that is supported by neuroimaging studies (Bede andPradat, 2019, Menke et al., 2017).
2.1.1.5.Central motor conduction time.Central motor conduction time (CMCT) measures the time for impulse propagation through the central motor system, starting from the initial activation of motor cortical neurons (via trans-synaptic excitation by cortical interneurons) in the primary motor cortex through to the excitation of spinal motor neurons via the corticospinal tract.CMCT is calculated by subtracting the final MEP latency from the peripheral motor conduction time (PMCT) (i.e., conduction time between the spinal nerves and the muscle) to distinguish between a delay along the proximal nerve root from a delay along the corticospinal tract.Two methods are commonly used to estimate PMCT: (i) spinal nerve stimulation and (ii) the F-wave technique (Rossini et al., 1985;Merton and Morton, 1980;Claus, 1990).The first method involves stimulating spinal nerves (i.e.proximal motor nerve roots) at the spinal foramen and subtracting the spinal MEP latency from the cor-tical MEP latency (Matsumoto et al., 2013, Mills andMurray, 1986).Using this technique, CMCT is measurable from almost any muscle but is prone to overestimation due to the inclusion of conduction along the motor axon, which is particularly relevant for lower limb muscle targets because of longer conduction pathways through the cauda equina (Matsumoto et al., 2010).Other challenges include inevitable contamination of the target muscle EMG recording due to costimulation of neighbouring axons, particularly at plexus stimulation sites (Rossini et al., 2015).The second approach uses F-wave principles of conduction (i.e., antidromic conduction to the spinal motoneuronal pool, 1 ms delay at the motoneuronal pool, followed by orthodromic conduction back to the muscle) to estimate PMCT using the formula (F + MÀ1 ms)/2 (Rossini et al., 1987), where F represents the shortest F-wave latency and M the distal motor latency, with 1 ms subtracted for the delay time at the spinal motoneuronal pool (Mills, 1999).Compared to the spinal stimulation method, conduction time is 1-1.5 ms longer using this methodology due to the differences in the site of peripheral stimulation, and overall, the Fwave technique is more accurate for measuring CMCT.It should be noted, however, that F-waves are routinely only recorded from distal muscles, limiting the use of this method for more proximal targets.Motoneuron recruitment thresholds and low F-wave persistence may be further barriers to sampling the fastest motor axons as required for this technique.Choice of optimal method is ultimately based on considerations of these specific limitations against the clinical context.
CMCT is measured during voluntary contraction of the target muscle (or contraction of the homologous contralateral muscle if precluded by weakness) and at least 5 recordings should be superimposed in order to obtain the shortest MEP latency (Mariorenzi et al., 1991).CMCT evolves with the maturation of the corticospinal tract, with longer latencies in the neonatal period (Duron and Khater-Boidin, 1998) shortening to adult values by approximately 3 years of age (Caramia et al., 1993).In the healthy adult population, there is no clear correlation between CMCT and age (Matsumoto et al., 2012, Claus, 1990), sex (Claus, 1990;Mills and Nithi, 1997), or handedness (Eisen and Shtybel, 1990;Mills and Nithi, 1997).CMCT strongly correlates with height in the lower limbs, but not in the upper limbs (Ugawa et al., 1989a, Claus, 1990, Matsumoto et al., 2010a, Furby et al., 1992).
In MND, CMCT may be abnormally prolonged due to corticospinal tract dysfunction (Eisen and Shtybel, 1990;Mills, 2003;Tokimura et al., 2020) and can reveal subclinical UMN involvement in a third of patients, particularly when recording from proximal muscles (Tokimura et al., 2020, Di Lazarro et al., 1999), suggesting potential diagnostic utility.There is no apparent correlation between degree of UMN impairment and prolongation of CMCT (Tokimura et al., 2020).

Paired-pulse TMS biomarkers
Paired pulse TMS techniques involve the delivery of two stimuli at varying interstimulus intervals (ISI), whereby a conditioning stimulus modulates the effects of a second test stimulus.This method has proven invaluable for integrating the role of cortical interneuronal circuits in MND pathophysiology and has resulted in development of novel biomarkers.Specifically, implementation of the pairedpulse technique has given rise to several emerging biomarkers that have shown utility in the understanding of ALS pathogenesis, diagnosis and clinical research, including short interval intracortical inhibition (SICI), intracortical facilitation (ICF), short interval intracortical facilitation (SICF) and long interval intracortical inhibition (LICI) (Rossini et al., 2015, Vucic et al., 2013b).
2.1.2.1.Assessment of cortical circuits.2.1.2.1.1.Short interval intracortical inhibition.First described in 1993, short interval intracortical inhibition (SICI) is a sensitive method to evaluate motor cortex excitability within M1 and is the most frequently used paired-pulse TMS paradigm (Kujirai et al., 1993), with the reliability and reproducibility of mean SICI in healthy populations recently confirmed (Matamala et al., 2018).Specifically, SICI is elicited when a preceding subthreshold conditioning stimulus (CS) suppresses a later suprathreshold test stimulus (TS) when separated by inter-stimulus intervals (ISI) of 1-to-7 ms.Using the conventional ''constant stimulus" method, where CS and TS are kept at a constant intensity, SICI is reflected bya reduction in MEP amplitude when compared to the MEP amplitude produced by an unconditioned TS (Kujirai et al., 1993).A threshold tracking paired-pulse TMS technique (TT-TMS) was subsequently developed, whereby the MEP amplitude is fixed (0.2 mV ± 20%) and changes in the TS intensity to maintain this target response are automatically tracked at ISIs increasing in a sequential ascending order [Fig.1C-E] (Fisher et al., 2002, Vucic et al., 2006) or in a pseudorandom parallel fashion (Tankisi et al., 2021a), which has been shown to be comparable with the ''constant stimulus" method in terms of reproducibility (Nielsen et al., 2021).Using the TT-TMS paradigm, SICI is reflected by a higher conditionedtest stimulus intensity required to generate and maintain the target MEP response between 1-to-7 ms when compared to unconditioned TS (Fisher et al., 2002, Vucic et al., 2006) At a physiological level, it has been suggested that this inhibitory phenomenon may serve to inhibit the undesired coactivation of other muscles during targeted movement (Stinear andByblow, 2003, Zoghi et al., 2003).While SICI is most prominent when recording from distal hand muscles, this inhibitory effect is also evident from other muscle targets including in the lower limb (Chen et al., 1998, Menon et al., 2018), face (Paradiso et al., 2005), diaphragm (Demoule et al., 2003) and sphincter (Lefaucheur, 2005).Two peak phases of SICI have been reported; a smaller phase at ISI 1 ms and a larger phase at ISIs 2.5-to-3 ms (Fisher et al., 2002, Hanajima et al., 2003, Roshan et al., 2003, Vucic et al., 2006).
Epidural recordings have suggested a cortical origin to SICI due to the observation that a subthreshold CS suppresses recruitment of TS provoked indirect (I)-waves (especially later I 3 waves) (Di Lazzaro and Rothwell, 2014).More specifically, SICI is thought to be mediated by inhibitory interneuronal circuits acting via GABA A receptors, which particularly influence the second peak phase (Ziemann et al., 2015).The first peak phase may be additionally affected by increased axonal refractoriness as well as synaptic inhibition (Chen, 2004, Fisher et al., 2002, Hanajima et al., 2003, Paulus and Rothwell, 2016, Vucic et al., 2011a, Vucic et al., 2009).SICI is influenced by coil type and muscle activity (Dharmadasa et al., 2019, Menon et al., 2018, Van den Bos et al., 2018b, Vucic et al., 2011a), and is critically dependent on stimulation intensities (Daskalakis et al., 2004, Garry and Thomson, 2009, Roshan et al., 2003, Sanger et al., 2001).The latter relationship is neither linear nor easily predicted, typically represented as a U-shaped curve with eventual recruitment of facilitatory circuits when the CS intensity increases (Chen et al., 1998, Peurala et al., 2008b, Vucic et al., 2009).As such, although SICI is generally characterized as an inhibitory effect that is largely mediated by interneuronal inhibition, it reflects the balance between inhibition and facilitation.
The influence of handedness on SICI is still unclear (Cahn et al., 2003, Dharmadasa et al., 2019, Menon et al., 2019, Shibuya et al., 2017a), although a reduction of SICI in the dominant hemisphere has been reported by some (Hammond et al., 2004, Ilic et al., 2004).It should be noted that SICI is not influenced by racial background (Suzuki et al., 2021) or sex (Cahn et al., 2003, Hermsen et al., 2016), suggesting utility as a biomarker across diverse patient population groups.
Abnormalities of cortical inhibition have also been reported in atypical MND phenotypes as well as familial MND cohorts.Specifically, reduced SICI has been established in the flail arm and leg variants of MND as well as primary lateral sclerosis (Geevasinga et al., 2015b, Menon et al., 2016, Vucic and Kiernan, 2007).Importantly, cortical hyperexcitability correlated with neurodegeneration, underscoring its importance in the pathogenesis of atypical MND phenotypes.Separately, recent studies utilizing the parallel threshold tracking paradigm disclosed a more prominent reduction of SICI in patients with absent or less prominent UMN signs (Tankisi et al., 2021b, Tankisi et al., 2023), suggesting greater utility of the parallel paradigm in atypical MND phenotypes.In familial MND cohorts, reduction in SICI has been reported in phenotypes linked to mutations in the superoxide dismutase-1 (Vucic et al., 2008), fused in sarcoma (Williams et al., 2013) and c9orf72 genes (Geevasinga et al., 2015a).Importantly, asymptomatic mutation carriers exhibit normal cortical function (Geevasinga et al., 2015a), with SICI reduction preceding the clinical development of familial MND by months (Vucic et al., 2008).
It has been argued that reduction in SICI represents a compensatory mechanism in response to peripheral neurodegeneration (Zanette et al., 2002b).The findings of normal cortical excitability in MND mimicking disorders (Menon et al., 2015a, Vucic and Kiernan, 2008, Vucic et al., 2010), along with partial and transient normalization of SICI with the anti-glutaminergic agent riluzole (Geevasinga et al., 2016b, Vucic et al., 2013a), argues against a compensatory mechanism.Rather, dysfunction or degeneration of GABAergic interneuronal circuits, acting via GABA A receptors, seems the most plausible mechanism for mediating SICI reduction in MND, a notion that is supported by pathological studies in patients and mouse models (Clark et al., 2021, Nihei et al., 1993, Zhang et al., 2016) as well as magnetic resonance spectroscopy (MRS) studies that reveal a reduction of GABA levels in the MND motor cortex (Foerster et al., 2012).In addition to reduced GABAergic function, SICI reduction in ALS may also be mediated by increased glutamatergic neurotransmission (Geevasinga et al., 2016b, Vucic et al., 2013a).These findings support the hypothesis that MND is a multistep process (Al-Chalabi et al., 2014, Vucic et al., 2020, Vucic et al., 2019), with the development of cortical hyperexcitability emerging as an important pathogenic step.
2.1.2.1.2.Long interval intracortical inhibition.Intracortical inhibition can also be demonstrated at longer interstimulus intervals, termed long-interval intracortical inhibition (LICI).This phenomenon describes the inhibition of a test MEP and is elicited by delivering a suprathreshold CS at 50-200 ms prior to a suprathreshold TS using the constant stimulus technique (Valls-Sole et al., 1992, Vucic et al., 2006, Wassermann et al., 1996), or 50-300 ms prior using the threshold-tracking technique (Vucic et al., 2006).Several points of evidence support a cortical origin of LICI, particularly due to the absence of spinal excitability changes at more than 50 ms after suprathreshold TMS (Fuhr et al., 1991), but also from the observed absence of LICI using paired transcranial electrical stimulation (Inghilleri et al., 1993) and the significant reduction of descending corticospinal volleys in epidural studies (Chen et al., 1999, Di Lazzaro et al., 2002a, Nakamura et al., 1997)..More specifically, this cortical inhibition appears to be mediated by longer latency inhibitory neuronal populations acting via GABA B post-synaptic receptors (McDonnell et al., 2006, Pierantozzi et al., 2004, Ziemann et al., 2015), although these are distinct to those mediating CSP duration and SICI (Vucic et al., 2006).Despite this, LICI and SICI are not completely independent processes, as LICI has been shown to inhibit SICI using a triple pulse stimulation paradigm probably through a pre-synaptic GABA B receptor mediated inhibitory mechanism (Ni et al., 2011).
In MND, a reduction of LICI has been reported alongside a reduction in SICI, inversely correlating with disease severity as well as upper motor neuron dysfunction (Zanette et al., 2002a(Zanette et al., , 2002b)).This has been proposed to be due to degeneration of the longlatency inhibitory circuits that mediate this phenomenon (Ziemann et al., 2015), although pathological and functional studies are required to confirm this.
2.1.2.1.3.Intracortical facilitation.Intracortical facilitation (ICF) is thought to reflect a true form of cortical facilitation, and can be elicited using a similar paired pulse paradigm as SICI but at longer ISIs (8-30 ms), with peak facilitation evident from 10 to 15 ms (Kujirai et al., 1993, Vucic et al., 2006).ICF does not appear to occur due to rebound disinhibition post SICI, as evident from a number of observations that distinguish these two phenomena, including different sensitivities to current direction and the need for a higher threshold to elicit ICF (Chen et al., 1998, Ziemann et al., 1996b).The physiological mechanisms mediating ICF remain to be fully clarified but are thought to involve recruitment of excitatory glutamatergic cortical circuits in M1 (Di Lazzaro et al., 2006), although spinal mechanisms have not been excluded (Chen et al., 2008, Di Lazzaro andRothwell, 2014).Neurobiochecmially, glutamatergic and serotonergic neurotransmission enhances ICF (Gerdelat-Mas et al., 2005, Ziemann et al., 1998).
Fig. 3. Short interval intracortical facilitation.(A) Experimental paradigm for short interval intracortical facilitation (SICF).SICF was recorded using a two channel paradigm.On channel 1, an unconditioned stimulus tracked a fixed target of 0.2 or 1.0 mV.The intensity (% maximal stimulator output) required to produce and maintain the tracking target was used to define the resting motor threshold (RMT).On channel 2, the conditioned test response was recorded.For SICF recording, the test response (T) was suprathreshold and preceded the conditioning (C) response which was set to 95% of RMT.(B) Short interval intracortical facilitation, across interstimulus intervals (ISI) of 1 to 5 ms, was significantly increased in MND patients when compared to controls.
2.1.2.1.4.Short interval intracortical facilitation.Short interval intracortical facilitation (SICF) reflects the activity of higher threshold facilitatory circuits that are elicited by a CS set to either threshold or suprathreshold levels over ISIs of 1-5 ms (Chen et al., 2008, Tokimura et al., 1996, Ziemann et al., 1998), and is a biomarker of excitatory function.The constant stimulus method identified three discrete SICF peaks at ISIs between (i) 1.1-1.5 (SICF1), (ii) 2.3-3.0 (SICF2), and (iii) 4.1-4.5 ms (SICF3) (Ziemann et al., 1998;Chen and Garg, 2000).SICF can also be elicited between ISIs of 1 and 5 ms using the threshold tracking paradigm (Fig. 3)As for the constant stimulus method, facilitation is more prominent during the same three distinct intervals when using a threshold tracking method with 0.2 ms steps between 1 and 5 ms (Cengiz and Kuruog ˘lu, 2020), while two peaks were evident (at ISI of 1.5 and 2.5-3 ms) when using 0.5 ms steps (Van den Bos et al., 2018b).
While the precise physiological mechanisms mediating SICF remain to be clarified, it has been suggested to be cortically mediated through facilitatory interactions of I-waves at the level of the corticomotoneurons (Ziemann et al., 2015, Ziemann et al., 1998) or through disinhibition of inhibitory neuronal circuits (Wagle-Shukla et al., 2009).The former hypothesis is supported by TMS modelling studies (Rusu et al., 2014) and the similar periodicity of the specific SICF peaks to the frequency of I-waves (Amassian et al., 1987).Pharmacological studies have further implicated a variety of neurotransmitter systems that also underscore a cortical contribution to SICF generation (Di Lazzaro and Ziemann, 2013, Ilic et al., 2003, Ilic et al., 2002, Korchounov and Ziemann, 2011, Ziemann et al., 2015).The recruitment of SICF partially overlaps with TMS intensities and paired-pulse intervals for eliciting SICI, which may lend some explanation as to why SICI decreases at high CS intensities (Ni et al., 2013, Peurala et al., 2008a).
In MND, increased activity of cortical facilitatory interneuronal populations has been supported by increases in ICF and SICF (Fig. 3B) measures (Vucic et al., 2023b), and was accompanied by SICI reduction (Van den Bos et al., 2018a).Increased ICF has also been reported in atypical MND phenotypes, although as this is an inconsistent finding it is of limited diagnostic utility (Menon et al., 2016, Menon et al., 2019, Vucic et al., 2013a).The imbalance in the facilitatory and inhibitory neuronal function can be quantified through a novel measure, the index of excitation, which is increased in MND, suggesting that overactivity of facilitatory circuits contributes to hyperexcitability.Clinically, this correlates with a greater degree of functional disability and the development of UMN signs, supporting pathogenic importance of facilitatory circuit overactivity in MND, and thereby providing a potential clinical biomarker.A recent study also reported an increase of SICF in ALS patients, that was most prominent at ISIs of 1-to-1.8ms (Cengiz and Kuruog ˘lu, 2020).When SICF increase was combined with SICI reduction, the diagnostic utility in discriminating ALS form healthy controls was increased (Cengiz and Kuruog ˘lu, 2020).

Imaging biomarkers in MND
Methodological advances in neuroimaging have become synonymous with advancing current understanding of neurodegenerative disorders.Routine clinical brain imaging techniques are capable of (i) localising brain abnormalities with millimetre accuracy, (ii) capturing whole-brain spatial and temporal patterns of pathological change, (iii) quantifying multi-modal markers of discrete brain tissue properties.In MND, magnetic resonance imaging (MRI) has demonstrated strong potential utility as disease biomarkers of UMN pathology (Ashhurst et al., 2022, Huynh et al., 2016).MRI is a robust non-invasive technique for inferring brain tissue properties by stimulating the spin of protons using varying radiofrequency (RF) pulses within a magnetic field, i.e., measuring the resonant release of energy following external RF energy transmission into the body.By modifying the RF pulse parameters (time-to-echo; time-to-repetition; time-to-inversion; flip angle) varying brain tissue properties can be observed, includ- ing T1 relaxation to assess degeneration of cortical motor neurons, Brownian motion of water diffusivity to quantify axonal degeneration of the corticospinal white matter, and chemical exchange saturation transfer (CEST) to assess imbalance of cortical metabolites.Neuroimaging studies in MND have highlighted the pathological presence of widespread cortical and subcortical brain changes associated with extra-motor features of disease.The clinical utility, however, remains an area of active development, providing a 'supporting' rather than a 'primary' role for MND patient diagnosis and monitoring (Shefner et al., 2020).
Early MRI reports in MND patients highlighted the presence of a hyper-intense signal along the length of the corticospinal tract (CST) and ribbon bands of focal hypo-intense signal along the precentral sulcus on T2-weighted images as an in-vivo marker of UMN abnormality (Goodin et al., 1988, Oba et al., 1993, Terao et al., 1995).The presence of T2 signal hyperintensity may serve as a surrogate biomarker of 'advanced' UMN pathology.Mean signal intensity of the CST appears higher in patients with primary lateral sclerosis (Fabes et al., 2017) and in young MND patients with a rapid disease course (Goodin et al., 1988).The nature of underlying T2 signal intensity changes in MND remains unclear and does not appear to be associated with axonal integrity (Rajagopalan et al., 2011), but potentially reflects neuroinflammatory mechanisms (Evans et al., 2014) [Fig. 4 ].Progress in advanced neuroimaging techniques, combined with multimodal approaches, has accelerated the characterisation of structural and functional changes that occur in the MND brain.This section will discuss the potential utility of imaging biomarkers in MND and future directions for their clinical translation.
High-resolution T1-weighted volumetric MRI form the basis of structural brain analysis, allowing identification of cortical atrophy and regional differences in tissue volume.A broad variety of automated and manual morphometry approaches have been used to assess CNS volume loss, with the two most commonly used analyses methods being voxel-based or surface-based (Ashburner and Friston, 2000).The former uses voxel-wise comparisons of grey and white matter volumes to infer regional brain atrophy between groups, while the latter analyses reconstructions of structural boundaries (e.g., between grey and white matter) to provide independent cortical thickness and surface area measurements (Das et al., 2009).
In MND, reduced total brain volume (or brain parenchymal fraction) has been reported with higher rates noted in those with concomitant frontotemporal dementia (MND-FTD), particularly if harboring a C9orf72 hexanucleotide repeat expansion (Agosta et al., 2017, Mahoney et al., 2012).Cross-sectional and longitudinal cohort studies in MND demonstrate relatively consistent patterns of atrophy, primarily affecting the motor cortices, thalamus, and brainstem (Bede and Hardiman, 2018, Grosskreutz et al., 2006, Menke et al., 2018, Senda et al., 2017, Tu et al., 2018), with cortical (M1) thinning proposed as a potential early biomarker of UMN dysfunction (Walhout et al., 2015).Volume loss in the corpus callosum has also been varyingly identified in sporadic cohorts (Chapman et al., 2012, Menke et al., 2014) and in PLS (Müller et al., 2020).In advanced stages of MND, motor cortex hypointensity represents an indirect marker of motor cortex atrophy (the ''motor band" or ''black ribbon" sign), which occurs due to an abnormally large central sulcus (Boll et al., 2019).
The utility of volumetric brain changes as an imaging biomarker in MND remains challenging due to low sensitivity when translated to the patient level.This is reflected by the absence of their clinical adoption despite the availability of robust fully automated analysis pipelines (Grosskreutz et al., 2006, Henschel et al., 2020), and ease of access to structural T1-weighted MRI data.Recent evidence suggests that this is perhaps due to a lack of precision from using standard Brodmann area cortical delineations.Notably, there is increasingly a clear divergence of more recent 'structural' and 'functional' cortical brain atlases (Glasser et al., 2016).While CST hyperintensity and the presence of a motor band are not specific to MND (Turner, 2005), being described in other neurological diseases and healthy populations (Ngai et al., 2007), a recent metaanalyses reported increased prevalence of both MRI features in MND patients when compared to controls, suggesting utility as a diagnostic aid (Zejlon et al., 2022).
The pathognomonic atrophy of M1 on structural imaging parallels the well-documented loss of cortical motor pyramidal neurons histopathologically in MND (Nihei et al., 1993), but imaging has not yet captured whether there is a uniform distribution of pathological involvement across the cortical motor homunculus or if this first presents as focal regions of abnormality modulated by clinical onset and disease stage, as has been recently suggested with TMS in the early stages of disease (Dharmadasa et al., 2020, Menon et al., 2020).A single-center study of spatial patterns in motor cortical thickness, parcellated into relative body regions, highlighted an interesting clinical dissociation (Nitert et al., 2022).Notably, cortical thickness of bilateral upper limb motor areas was reduced irrespective of clinical UMN signs, while bulbar and lower limb motor areas were progressively impacted by increasing clinical UMN involvement (Fig. 5).Longitudinal change in cortical thickness demonstrated greater sensitivity relative to clinical examination.These findings require prospective multi-center validation but present a promising direction for clinical translation of precise volumetric MRI markers in MND.
The presence of extra-motor atrophy is associated with a more rapid rate of functional decline (Menke et al., 2018, Senda et al., 2017, Steinbach et al., 2020), likely mediated by sub-clinical neuropsychiatric and cognitive dysfunction (Mioshi et al., 2013).The spatial pattern of extra-motor atrophy extending to frontal and temporal lobe regions bears striking similarity to FTD, which holds significant pathogenic overlap with ALS (Kiernan et al., 2011).Cerebellar atrophy has also been correlated with neuropsychiatric and motor features in MND-FTD phenotypes.Targeted analysis of the thalamus demonstrated distinct profiles of regional abnormality, mirroring established patterns of cortical dysfunction (Westeneng et al., 2016), suggesting a role for subcortical change as a holistic marker of evolving cortical dysfunction in MND (Bocchetta et al., 2022, Tu et al., 2018).
The proportion of MND patients that display structural motor abnormalities on MRI varies substantially across publications, ranging from 0% to 95% (Boll et al., 2019, Thorns et al., 2013).This is in part due to cohort heterogeneity, protocol differences in MRI methodology (T2-weighted vs T2-weighted-FLAIR vs susceptibility weighted imaging, SWI) and field strengths (Roeben et al., 2019, Zejlon et al., 2022).Emerging literature in ultra-high field imaging (7-Tesla, 7 T) suggests a higher sensitivity for detecting the structural changes (Obusez et al., 2018).

Microstructural white matter integrity (diffusion MRI)
Diffusion MRI techniques use the principles of water diffusion to evaluate white matter integrity.Diffusion tensor imaging (DTI) remains one of the most widely utilized diffusion approaches.Within the DTI framework, four quantitative measures are typically assessed that reflect different aspects of tissue microstructure: (i) fractional anisotropy (FA) describes the strength of directionality of water diffusion within the tissue; (ii) radial diffusivity; the average amount of diffusion along secondary and tertiary diffusion axes; (iii) axial diffusivity, the amount of diffusion along the principal axis of diffusion; and (iv) mean diffusivity, reflecting the average water displacement along each of the three axes (independent to direction).These parameters are commonly used as MRI biomarkers, with several analysis approaches adopted that non-invasively (i) map white matter pathways in the brain (Alexander et al., 2007) by using probabilistic (Behrens et al., 2007) and/or deterministic (Mori et al., 1999) tractography approaches, (ii) evaluate whole-brain connectivity pathways using vowel-wise analyses or (iii) Localize specific regions of interest (ROI) in the brain.FA (scaled between 0 and 1) has high test-retest and cross-scanner reliability and is the most commonly reported parameter, and can be obtained with a relatively small number of diffusion-encoding directions [!6 orthogonal directions] (Luque Laguna et al., 2020).In the setting of intact white matter, water diffusion is restricted along the direction of the fiber (higher FA), while damaged white matter fibers are less restrictive to diffusion (lower FA).Biophysical interpretations of these parameters are however problematic in the presence of complex white matter fiber geometries (particularly crossing fibres) and microstructural changes (e.g axonal swelling) which represent well established limitations of this technique, with various acquisition and analysis methods now developed in an attempt to resolve these (Jeurissen et al., 2013).
In MND, diffusion-weighted MRI (dMRI) has long demonstrated the most consistent disease signature of UMN dysfunction.Alterations in DTI metrics along core disease white matter tracts, the CST and corpus callosum (CC), appear sensitive enough to differentiate clinical MND phenotypes based on the predominance of UMN involvement (Agosta et al., 2014, Ferraro et al., 2017, Steinbach et al., 2021a), as well as fast progressing lower-motor predominant MND patients (Rosenbohm et al., 2016).Moderate to strong correlations between reduced white matter integrity and progressive clinical decline have been reliably reproduced by global MND research groups longitudinally across independent patient cohorts (Agosta et al., 2014, Bede and Hardiman, 2018, Menke et al., 2018, Steinbach et al., 2021b, Tu et al., 2018), and appears to show high classification accuracy in differentiating UMN predominant MND from disease mimics (>87% accuracy) (Ferraro et al., 2017).More importantly, DWI studies have provided the basis for proposed models of MND disease propagation, including corpus callosal fibers facilitating the inter-hemispheric spread of clinical symptoms (Steinbach et al., 2021b, Tu et al., 2020), sequential tract of interest disease staging (i.e., Stage 1 -CST; Stage 2 -corticorubral/corticopontine; Stage 3 -corticostriatal; Stage 4 -perforant pathway) (Kassubek et al., 2014), and rostro-caudal directionality of cortical-spinal disease progression (Gabel et al., 2020, Tu et al., 2019).While significant debate remains over the temporal trajectory of white matter degeneration with respect to evolving clinical features in MND, dMRI provides relatively higher consistency and sensitivity to pursue this research direction over other imaging modalities.Of note, DWI analyses greatly improve in precision when progression modelling is used to reduce the inherent noise of clinical reference data (Fig. 6), indicating that diffusion MRI approaches are capable of detecting imprints of disease aggressiveness separately from disease accumulation potentially providing a tool for monitoring treatment effects in clinical trials (Steinbach et al., 2021a).

Cortical metabolite abnormality (magnetic resonance spectroscopy)
Beyond structural imaging markers, proton magnetic resonance spectroscopy (MRS) can be employed to quantify brain metabolite concentrations and to understand regional brain biochemistry using CEST MRI techniques.MRS remains a niche modality in MND imaging but has gained momentum with increased availability of ultra-high 7 T field-strength MRI scanners.MRS benefits significantly from increased field strength in the form of increased signal-to-noise ratio and chemical shift dispersion, resulting in greater precision for resolving complex metabolite peaks (Pradhan et al., 2015), including neurotransmitters of interest (i.e., glutamate, GABA) underlying disease mechanisms such as cortical hyperexcitability (Vucic et al., 2023b).Other marks that can be quantified include N-acetylaspartate (NAA) as a neuronal marker, choline (Cho) to characterise cell membrane integrity, and creatine (Cr), a marker of energy metabolism (Christidi et al., 2022, Kaufmann et al., 2004).
In MND, an early MRS study found reduced GABA in MND relative to healthy controls (Foerster et al., 2012, Foerster et al., 2013), drawing on extensive TMS findings of GABAergic interneuron dysfunction as a pathogenic mechanism of UMN dysfunction in MND (Menon et al., 2020, Vucic et al., 2008).This, however, appears to be an isolated finding that has yet to be reproduced at both standard 3 T (Blicher et al., 2019, Weerasekera et al., 2019) or 7 T MRI (Atassi et al., 2017).Interestingly, elevated cortical motor glutamate concentration appears to be a consistent disease signature in MND (Caldwell and Rothman, 2021).Reflecting both structural and diffusion MRI findings, reduced neural integrity in the primary motor cortex has also been corroborated by consistent reductions of NAA levels, NAA:Cr and NAA:Chol ratios (Block et al., 1998, Pohl et al., 2001, Rooney et al., 1998), with similar but less significant changes noted in bilateral frontoparietal areas and along the CST.

Functional abnormalities (functional MRI)
Functional MRI (fMRI) uses the principles of magnetic fields to detect small changes in signal that are associated with neuronal activity.The most commonly used technique detects blood oxygen level dependent (BOLD) changes that occur during a specific activity ('task fMRI') or at rest ('resting state' or 'task-free' fMRI) to indirectly measure neuronal activity and connectivity, working under the assumption that energy and cerebral blood flow (and thus oxygenated hemoglobin levels) change locally in response to activated neurons (Smith, 2004).Activity produces a relatively high oxyghemoglobin: deoxyhaemoglobin ratio and given their differing magnetic properties (diamagnetic vs paramagnetic) the resultant fluctuations in magnetic field during activity can be dynamically and sensitively measured using a T2*-weighted MRI sequence (Logothetis, 2003).Resting state fMRI (rs-fMRI) detects low frequency fluctuations in BOLD signal and identifies synchronicity between different brain regions, as a measure of brain networks.Rs-fMRI acquisition techniques have been an area of growing focus and rapid development, with this activity free modality particularly suited to patients with physical limitations such as in MND, in contrast to task-based motor responses which are challenging to standardize in this context (Cole et al., 2010).Importantly, early rs-FMRI studies demonstrated that BOLD signals show motor cortex synchronicity between dominant and non-dominant hemispheres in healthy subjects, suggesting intrinsic connectivity between motor cortical regions.Unlike positron emission tomographic (PET) imaging, fMRI does not report absolute change.
In MND, functional connectivity abnormalities are consistently demonstrated within the sensorimotor network and default mode network (Agosta et al., 2011).While there have been methodological and cohort driven discrepancies in the direction of this abnormal functional activity, recent studies in earlier stages of the disease support an increased interconnected network in ALS, possibly representing compensatory mechanisms secondary to neurodegeneration.Like other imaging modalities, this connectivity appears to correlate with functional disease burden, clinical symp-  (Steinbach et al., 2021a)].FA refer to fractional anisotropy, MD to mean diffusivity, AD to axial diffusivity and RD to radial diffusivity.P < 0.001 was regarded as significant.
toms, and the presence of cognitive changes (Agosta et al., 2013).Notably, patterns of connectivity are similar between MND and FTD, supporting the pathophysiological links between these conditions (Trojsi et al., 2015).The most marked areas of functional network change also appear to mirror marked loss of white matter integrity on DTI studies, reinforcing the concept of interhemispheric inhibition and a multimodal network of dysfunction (Douaud et al., 2011).

Neural hypometabolism and hypermetabolism (PET/MRI)
Positron emission tomography (PET) imaging uses radioisotope tracers to localize molecular changes or identify specific neuroreceptor pools in the CNS in vivo, allowing sensitive multifaceted exploration of neural molecular metabolic activity.Tracers can detect and reconstruct neural tissue three-dimensionally, permitting visualization of a myriad of molecular activity including neurotransmitter metabolism, neuroreceptor binding, inflammation, cerebral blood flow and glucose metabolism.
In MND, the first PET study was conducted more than 30 years ago using the analogue tracer 18 F-flurodeoxyglucose ( 18 F-FDG), demonstrating diffuse brain hypometabolism in patients with clinical UMN involvement (Dalakas et al., 1987).While subsequent studies reported variable levels of hypometabolism (Hoffman et al., 1992, Ludolph et al., 1992), large cross-sectional studies in sporadic MND cohorts have reconciled the discrepancy by showing regional divergence in metabolism, with hypometabolism in premotor and frontal cortices and hypermetabolism in the brainstem (Pagani et al., 2014, Van Laere et al., 2014).Growing interest in recent methods of spinal cord imaging showing cervical hypermetabolism now warrants ongoing exploration to determine the metabolic differences from cortex to cord (Yamashita et al., 2017).
PET studies have also supported the presence of widespread (Turner et al., 2004) localized motor cortical (Zürcher et al., 2015) neuroinflammatory activity in the MND brain using tracers that are highly expressed on activated astrocytes and microglia (e.g., 18kD translocator protein, TSPO), while GABA A receptor ligands (e.g., [ 11 C]-Flumazenil) have explored altered cortical excitability in sporadic and familial (SOD1 D90A) cohorts, supporting the loss of cortical inhibition as demonstrated by TMS (Tu and Kiernan 2023).Importantly, integrated [ 11 C]-PBR28 PET-MRI stud-ies have confirmed that such areas of increased uptake on PET colocalise with structural (cortical thinning) and diffusion (reduced FA) metrics (Alshikho et al., 2016, Alshikho et al., 2018), and correlate with clinical and functional decline.Although stability of metabolite activity during disease progression has been proposed, this may be due to the inherent recruitment bias of slow progressors (Alshikho et al., 2018).PET and PET-MRI imaging thus holds significant potential to determine pharmacodynamic biomarkers and measure UMN dysfunction in MND, but the major limitations of small studies, limited longitudinal data, and contradictory clinical-radiological correlations in the research to date will need confirmation in collaborative, multicenter longitudinal studies.

Pathogenic insights provided by UMN biomarkers in MND
At a pathophysiological level, TMS and neuroimaging have provided novel insights into MND pathogenesis.The importance of UMN dysfunction in MND pathogenesis was originally proposed by Charcot, with identification of upper and lower motor neuron degeneration on post-mortem human pathological studies (Charcot and Joffroy, 1869).Over a 100 years later, Eisen and colleagues proposed the dying forward hypothesis whereby corticomotoneuronal hyperexcitability mediates motor neuron degeneration via an anterograde glutaminergic process [Fig.7] (Eisen et al., 1992).Degeneration of inhibitory interneuronal populations along with and enhanced activity of facilitatory circuits in the M1 were proposed as putative mechanisms in mediating cortical hyperexcitability in MND (Cavarsan et al., 2023, Clark et al., 2021, Clark et al., 2017, Clark et al., 2023, Nihei et al., 1993, Van den Bos et al., 2018a, Zhang et al., 2016).
Threshold tracking TMS has provided support for a dyingforward process, with cortical hyperexcitability emerging as a central tenet of MND pathogenesis (Vucic et al., 2023b).As demonstrated above, cortical hyperexcitability (reflected by SICI reduction) was shown to precede the onset of LMN dysfunction in MND (Menon et al., 2015b, Vucic et al., 2008), to correlate with biomarkers of LMN degeneration (Vucic and Kiernan, 2007, 2010, Vucic et al., 2008) and be associated with development of specific clinical features (Bae et al., 2014, Menon et al., 2014).The development of cortical hyperexcitability appears to be a specific feature of MND when compared to mimicking disorders (Menon et al., 2015a), and has been associated with patterns of disease progression in MND (Dharmadasa et al., 2020, Menon et al., 2017).Additionally, MND patients exhibiting cortical dysfunction had a more malignant disease trajectory, with faster rate of progression, greater functional disability, and reduced survival, particularly if cortical dysfunction occurred in early stages of the disease (Dharmadasa et al., 2021, Menon et al., 2019, Shibuya et al., 2016).Increased activity of high threshold facilitatory interneuronal populations, as reflected by increased SICF, was also established in MND, correlating with a greater level of UMN dysfunction and functional decline.Of further relevance, dysfunction of transcallosal inhibitory circuits also appear to contribute to the development of hyperexcitability in MND, its pathogenicity suggested through correlation with disease progression rate (van den Bos et al., 2021).
More recently, it was reported that the degree of SICI reduction was more prominent in MND patients with absent or less prominent UMN signs (Tankisi et al., 2021b, Tankisi et al., 2023), thereby questioning the pathogenicity of cortical hyperexcitability in typical MND phenotypes.A possible explanation could relate to methodological differences, namely the use of a parallelpseudorandom versus the conventional serial ascending ordering of ISIs.Additionally, the tracking methodology was different, whereby the original threshold tracking approach used 1% changes T. Dharmadasa, N. Pavey, S. Tu et al. Clinical Neurophysiology 163 (2024) 68-89 in TMS intensity output compared to the 4-2-1 approach (proportional tracking) in the more recent method.Alternatively, it could be argued that the dominant abnormality in the assessed MND patients may relate to more prominent activity of high threshold facilitatory circuits rather than reduced cortical inhibition, which would still be consistent with an excitotoxicity hypothesis.Taken together, threshold tracking TMS has provided evidence for the importance of cortical hyperexcitability in MND.In keeping with the multistep hypothesis of MND (Al-Chalabi et al., 2014, Vucic et al., 2020, Vucic et al., 2019), the development of cortical hyperexcitability may represent a crucial step prior to onset of LMN degeneration.It should be stressed, however, that MND is a complex neurodegenerative disorder, mediated by a complex interaction of genetic, molecular, and environmental mechanisms.At a molecular level, glutamate excitotoxicity, oxidative stress, impaired RNA metabolism, autophagy and proteasome dysfunction have been well described (Feldman et al., 2022, Kiernan et al., 2020).Additionally, genetic mutations in specific MND related genes (SOD-1, TDP-43, and FUS) lead to protein aggregation, further contributing to pathogenesis.Mitochondrial dysfunction, neuroinflammation, and impaired axonal transport also play an important role, as does prion-like spread of protein aggregates.Whether development of cortical hyperexcitability is an initiating event or occurs as part of the multi-step process in MND needs to be further elucidated.
Neuroimaging studies have also corroborated a pathogenic neurobiochemical environment of altered cortical excitability in the MND brain, with advanced modalities such as PET, MRS and rs-fMRI highlighting a compromised GABA-ergic system and loss of inhibitory interneurons (Turner et al., 2012).The level of GABA binding in the motor regions also appears to be implicated in progression rate, suggesting that this may be critical aetiological mechanism of MND pathogenesis.The clinical and mechanistic implications, as well as longitudinal trajectory of cortical motor glutamate abnormality in MND remains largely unexplored through these imaging modalities, and a pressing question is the association with clinical disease trajectory and whether the dosage of riluzole (as a glutamate antagonist) can be further optimized.
It could be argued that cortical hyperexcitability forms a unifying pathogenic basis for the 'apparent' differences in patterns of disease spread in MND, whereby hyperexcitability of specific corticomotoneuronal tracts represent a conduit for disease spread.A contiguous horizontal pattern of spread (limb-to-limb) is most frequently reported (Gargiulo-Monachelli et al., 2012, Korner et al., 2011, Menon et al., 2019), with concordance between UMN and LMN dysfunction within the affected body region as well as a concordance between site of disease onset and limb dominance (Devine et al., 2014, Turner et al., 2011).Given the higher density of corticospinal tracts on the dominant side and their function in mediating complex finer tasks (Rose et al., 2012), the established concordance between limb dominance and site of disease onsetcould be explained by a greater corticomotoneuronal hyperexcitability to the affected region.Further, the spread of disease to contiguous or non-contiguous regions could be explained by development of hyperexcitability in corticomotoneuronal pathways destined for specific body regions, resulting in a 'river of hyperexcitability'.Support for this notion is provided by TMS studies disclosing a focal onset of cortical hyperexcitability within the M1 corresponding to site of disease onset (Dharmadasa et al., 2020, Menon et al., 2019).More recently, additional support for the pathogenic importance of cortical hyperexcitability has been provided by transgenic mouse model studies reporting that mislocalisation of TDP-43 in the UMNs increases excitatory inputs to the spinal motor neurons and induces disease progression through a dying forward process (Reale et al., 2023).Therapeutic interven-tions aimed at modulating this corticofugal process may be of therapeutic utility in MND.
Neuroimaging studies also provide compelling support towards the underlying network of UMN dysfunction in MND, lending validation to the concept of MND as a multisystem syndrome.The pathophysiological nature of underlying T2 signal intensity change along the CST on structural MRI shows congruence with axonal loss (Zejlon et al., 2022).Similarly, altered DTI metrics may also represent primary axonal degeneration and/or secondary microglial activation, highlighting the multiple aetiological pathways that may play a role in this complex disease.Distinct patterns of regional brain involvement using advanced multimodal imaging techniques have also shown utility in discriminating between phenotypes offering further some pathogenic insights underlying clinical disease heterogeneity (Grolez et al., 2016).A specific differentiating feature between clinical MND phenotypes appears to be the greater degree and spread of motor abnormalities in bulbar-onset MND and patients with concomitant FTD demonstrating compared to patients with limb-onset disease as reflected by structural (increased atrophy), diffusion (greater reduction in FA and increase in MD), and spectroscopy (lower NAA levels) imaging techniques (Ellis et al., 1999, 1998, Kim et al., 2017).This may offer a pathophysiological explanation towards the more malignant prognostic trajectory for these clinical phenotypes, and a multimodal clinical tool for objectively mapping clinical heterogeneity.

Utility of UMN biomarkers in diagnosis and disease monitoring
The diagnosis of MND relies on identifying progressive clinical upper and lower motor neuron signs and the exclusion of potential mimicking disorders (Goutman et al., 2022, Kiernan et al., 2011, Kiernan et al., 2020).Clinically based criteria have evolved to facilitate an earlier and more definitive MND diagnosis in the absence of a diagnostic test (Brooks, 1994, Brooks et al., 2000), critical for guiding appropriate clinical management and recruitment into clinical trials.The original criteria (El-Escorial and revised El-Escorial criteria) were complex with multiple levels of diagnostic certainty (definite, probable, possible) and exhibited poor sensitivity, particularly in early stages of the disease process (Costa et al., 2012, Geevasinga et al., 2016a, Turner et al., 2009).The limitation of sensitivity pertained to reliance on clinical assessment of UMN dysfunction, with difficulties identifying UMN signs in MND patients with prominent LMN dysfunction (Higashihara et al., 2012, Swash, 2012), including atypical phenotypes, such as the flail arm variant (Vucic and Kiernan, 2007).Additionally, poor interrater reproducibility was also reported and related to diagnostic complexity (Costa et al., 2012, Geevasinga et al., 2016a, Johnsen et al., 2019, Turner et al., 2009).
A neurophysiologically based Awaji-Shima criteria were subsequently proposed (de Carvalho et al., 2008), whereby needle electromyography (EMG) findings of ongoing neurogenic changes (fibrillation potentials/positive sharp waves) or fasciculations, along with chronic neurogenic changes, were deemed equivalent to LMN signs.While the Awaji-Shima criteria exhibited greater sensitivity (Costa et al., 2012, Geevasinga et al., 2016a), the limitations relating to a clinical reliance on identification of UMN dysfunction and complexity resulting in poor-interrater reproducibility remained (Johnsen et al., 2019).More recently, the simpler Gold Coast criteria were proposed (Shefner et al., 2020), whereby diagnostic certainty levels were excluded and the presence of UMN and LMN dysfunction in one region, or LMN dysfunction in two regions, were deemed diagnostic of MND in the setting of disease progression and exclusion of other mimicking disorders.While the Gold Coast criteria exhibited increased sensi-tivity and a comparable specificity when contrasted with the previous criteria (Hannaford et al., 2021), the identification of UMN dysfunction remained clinically based.
TMS has proven to be a robust and objective biomarker of UMN dysfunction in MND (Geevasinga et al., 2014, Menon et al., 2015a, Vucic et al., 2011b).The reduction of SICI or presence of motor cortex inexcitability, reliably differentiates MND from neuromuscular mimicking disorders, while the subclinical identification of UMN involvement enables the diagnosis of MND to be made approximately 8 months earlier compared to the use of clinical criteria alone (Vucic et al., 2011b), improving the diagnostic utility of the Awaji criteria by $34% (Menon et al., 2015a).This has been of particular importance in atypical phenotypes, where the detection of sub-clinical UMN dysfunction has been of diagnostic utility (Menon et al., 2016, Vucic andKiernan, 2007).The prolongation of CMCT, particularly from proximal muscle groups in patients with no clinical UMN features, appears to be specific to MND (Di Lazarro et al., 1999) and offers a potential diagnostic biomarker (Eisen et al., 1990, Mills, 2003), although the sensitivity of this measure may be lower when recording from distal muscles (Di Lazzaro et al., 1999;Truffert et al., 2000;Menon et al., 2015).At present, the application of threshold tracking TMS into clinical practice has been limited to specialised centres.In part, this relates to requirement for specialised equipment (multiple pieces of hardware and licenced software), knowledge for assembling and integrating hardware-software functioning (often requiring working knowledge of computer programming), and extensive training requirements.Integration of TMS protocols, such as the threshold tracking paired-pulse methodology, into commercially neurophysiological devices will likely obviate some of these limitations and help clarify the diagnostic utility of TMS in MND.The TST technique may also be of diagnostic utility in differentiating MND from mimicking disorders, with diagnostic sensitivity varying between 54% and 100% (Kleine et al., 2010, Komissarow et al., 2004, Rösler et al., 2000, Wang et al., 2019).Specifically, the sensitivity of detecting conduction block is increased with TST when compared to conventional neurophysiological studies, particularly if the conduction block is proximal, as evidenced for multifocal motor neuropathy (Attarian et al., 2005b, Deroide et al., 2007), Guillain-Barré syndrome (Taieb et al., 2015) and chronic inflammatory demyelinating polyradiculoneuropathy (Attarian et al., 2015).When combined with MRI, this sensitivity is increased (Corazza et al., 2020).The confounding effects of sub-maximal peripheral stimulation, particularly at proximal sites, may impact specificity (Caranzano et al., 2021).The wider applicability of TST as a routine diagnostic technique is limited by patient tolerability (mainly related to patient discomfort), complexity of the test, and restriction to assessment intrinsic hand and foot muscles (Buhler et al., 2001, Magistris et al., 1999, Magistris et al., 1998, Rosler et al., 2002).The absence or a marked reduction of CMAP responses, which typically occur in the later disease stages of MND, may also preclude TST studies as a marker of disease progression.
In addition to diagnostic utility, TMS measures have shown promise as potential outcome biomarkers of therapeutic efficacy in clinical trials settings (Kiernan et al., 2020).Given the limitations of current clinical measures, there has been an increasing shift towards the use of objective biological and neurophysiological outcome biomarkers as primary endpoints in MND clinical trials (Goutman et al., 2022, Kiernan et al., 2020).Neurofilament light chain has been successfully used as an outcome biomarker in the VALOR study, providing evidence for the biologic effectiveness of the antisense oligonucleotide tofersen (in MND patients with mutations in the SOD-1 gene) prior to any clinical benefits (Miller et al., 2022).Additionally, motor unit number index has been used as a neurophysiological marker of LMN dysfunction in MND trials, most recently suggesting potential clinical effectiveness of the nanocrystalline gold compound CNM-Au8 (Vucic et al., 2023a).In keeping with an objective biomarker theme, SICI has also been shown as a promising outcome biomarker in the MND clinical trial that assessed effectiveness of the antiexcitotoxic agent retigabine (Wainger et al., 2021).Given that SICI reduction has been established as an adverse prognostic factor in MND (Shibuya et al., 2016), the utility of SICI as a biomarker for patient stratification, in addition to an outcome measure, should also be considered although this requires further confirmation.
Currently, imaging continues to play a supporting role in MND diagnosis with prospective global initiatives employing multicenter validation of diagnostic utility (Picher-Martel et al., 2023, Steinbach et al., 2018).Conventional structural MRI remains the most frequently utilized neuroimaging modality in MND (Zejlon et al., 2022), but the findings are non-specific (and often unremarkable), even in patients with significant motor disability, thereby limiting diagnostic utility.Specifically, the utility of CST T2 hyperintensity as a diagnostic biomarker is limited, with low sensitivity (<50%) and specificity (<80%) (Gupta et al., 2014, Waragai, 1997), but accuracy can be improved by employing quantitative analysis of mean signal intensity (Fabes et al., 2017).More advanced imaging techniques, including DTI, appear to demonstrate high classification accuracy in differentiating UMN predominant MND from disease mimics (>87% accuracy) (Ferraro et al., 2017).In particular, the utility of diffusivity metrics as a clinical marker of UMN dysfunction as the disease progresses shows great potential but requires: (i) prospective validation in real-world clinical settings, (ii) harmonized multi-center MRI acquisition protocols and analysis pipelines, and (iii) development of age-normalized percentile scores for physician interpretation.Methodological development to improve precision of diffusivity metrics is ongoing.Notable advances for consideration include the adoption of high-angular resolution DWI (i.e., !64 non-collinear gradient directions; bvalue !2000 s/mm 2 ) in combination with complex fiber orientation distribution modelling to account for crossing white matter fibers in the brain to target 'tract-specific' changes in diffusivity (Dhollander et al., 2021, Jeurissen et al., 2014, Steinbach et al., 2021b).The clinical and diagnostic utility of such approaches can also be seen through enhanced reconstruction and visualization of whole white matter fiber bundles.For example, to assess asymmetry of corticospinal pathways, improving diagnostic certainty for MND phenotypes such as Mills Syndrome (Huynh et al., 2021), and in quantitative neurography (D 'Souza et al., 2022) [Fig. 8].Assessment of NAA levels on MRS studies may also diagnostically differentiate MND from disease mimics such as spinal muscular atrophy (Pioro et al., 1994), and have captured longitudinal changes (decreasing NAA levels) that may provide a surrogate biomarker of disease progression (Rule et al., 2004, Unrath et al., 2007).Particularly with the increased availability of ultra-high 7 T field-strength MRI, this is an active area in the development of more robust and reliable pipelines to counter limitations of metabolite variability.
Currently however, any utility of brain changes as an imaging biomarker in MND remains challenging due to low sensitivity when translated to the patient level.This is reflected by the absence of their clinical adoption despite the availability of robust, fully automated analysis pipelines (Grosskreutz et al., 2006, Henschel et al., 2020), and ease of access to structural T1weighted MRI data.An important factor that needs to be addressed is the added value that can be provided by adopting imaging biomarkers for detecting subclinical UMN dysfunction.Demonstrating high classification accuracy of imaging metrics between MND and healthy controls does not address any practical issues for MND physicians.Efforts should therefore particularly focus on differential diagnosis of complex clinical presentations and regionally restricted MND phenotypes (Ferraro et al., 2017, Huynh et al., 2021, Tu et al., 2019, Tu et al., 2020).
Combining sophisticated neuroimaging techniques with TMS, as well as integrating multiple imaging biomarkers, may provide more optimal diagnostic utility in a clinical setting.Such approaches may yield novel diagnostic and prognostic biomarkers that would be of utility in the clinical care of MND patients, as well as being useful in a clinical trial setting in terms of patient stratification.Additionally, biomarkers to determine target engagement of novel compounds at an early stage of disease development may also be developed.

Fig. 2 .
Fig. 2. Cortical silent period duration.(A) The cortical silent period (CSP) duration refers to a period of electrical silence after a TMS stimulus to resumption of electromyography activity.The CSP duration increases with TMS intensity and is mediated by long latency inhibitory circuits acting via GABA B receptors.(B) The CSP duration is significantly reduced in motor neuron disease patients indicating dysfunction of the long latency inhibitory circuits.

Fig. 4 .
Fig. 4. MRI changes in MND.Age-and sex-matched case examples of standard clinical T1 and fluid attenuated inversion recovery (FLAIR) MRI images from healthy controls and MND patients.Visual atrophy and widening of the central sulcus on T1-weighted MRI (red and blue lines) are typically unremarkable and difficult to differentiate from age-related change.Minor hyperintensity of the corticospinal tract can be observed on FLAIR images in both ALS and healthy controls.Radiological signs of MND pathology demonstrate low sensitivity and are not specific to the disease (Figure reproduced from Tu and Kiernan.Amyotrophic Lateral Sclerosis.Advances in Magnetic Resonance Technology and Applications.2023;9:363-385, with permission of Elsevier).

Fig. 5 .
Fig. 5. Cortical thickness in MND.Consideration of functionally divergent cortical motor processing regions when examining volumetric change can result in more precise and clinically informative structural MRI biomarkers of MND disease progression.[Top] 3D reconstruction of the brain's cortical surface on an individual patient's MRI scan (Pial Surface [Left], White Matter Surface [Right]; Gyri [Green], Sulcus [Red]).[Bottom] Smoothed representation of the inflated gyri/sulci cortical surface.Primary motor cortex (Brodmann Area 4) labelled in blue.Broad anatomical parcellation into leg (green), arm (arm), and bulbar (red) regions.Functionally discrete motor regions potentially demonstrate dissociable levels of UMN degeneration and disease trajectories.

Fig. 6 .
Fig. 6.Diffusion tensor imaging of ALS related tract damage in radial diffusivity.Left panel: Whole brain group analyses.(I) Comparison of healthy controls (n = 60) versus ALS patients (n = 103) shows extensive tract damage in corticospinal tract (CST), central transcallosal fibres (cTF) and centro-frontal projections (CFP).(II) within the ALS group, disease accumulation, reported by the D50 model as rD50, is reflected in CFP but not in CST or cTF, indicating that radial diffusivity did not further increase during disease accumulation independent of overall disease aggressiveness.(III) within the ALS group the rate of D50 model calculated functional loss rate (cFL) at the time of MRI strongly associates with the CST and cTF, and extends into the brain stem and cerebellum, a pattern nor detectable in comparisons of ALS patients versus controls.This indicates changes radial diffusivity in tracts which are closely connected to disease aggressiveness.Right panel: number of significant voxels vs. significance levels of all DWI modalities.(A) regression with model based functional state (cFS) and (C) with overall aggressiveness (D50) demonstrate the improvement utilizing clinical progression modelling over traditional measures such as (B) ALSFRS-R and (D).its progression rate (PR) which shows no association with DWI [figure adapted from(Steinbach et al., 2021a)].FA refer to fractional anisotropy, MD to mean diffusivity, AD to axial diffusivity and RD to radial diffusivity.P < 0.001 was regarded as significant.

Fig. 7 .
Fig. 7. Site of MND onset.Three competing theories for MND onset have been proposed, including the dying forward, dying back and independent hypothesis.Cortical hyperexcitability is mediated by loss of inhibition and increased facilitation.Transcranial magnetic stimulation studies have provided evidence for a dying forward hypothesis (Kiernan et al., 2011).

Fig. 8 .
Fig. 8. Advances in MND neuroimaging.Advanced diffusion MRI methods for evaluating microstructural white-matter tract integrity.[Left] Conventional tract-based spatial statistics comparing group differences in white-matter integrity.Tracts are skeletonized to improve reproducibility and accuracy.[Middle] Tractography of the whole tract bundle associated with corticospinal and callosal motor white-matter fibres in an individual patient using high-angular resolution DWI acquisition in combination with constrained spherical deconvolution to model crossing white-matter fibres.[Right] Tract density imaging visualises the density of localised white-matter within tracts.The whole corticospinal white-matter tract bundle can be reliably visualised across individual patients (Top).Example of asymmetry in corticospinal tract density in an atypical MND patient diagnosed with Mill's Syndrome.