Examining short interval intracortical inhibition with different transcranial magnetic stimulation-induced current directions in ALS

Highlights • SICI measured using posterior-to-anterior induced current direction (SICIPA) has been reported to be lower in ALS.• We find both SICIPA and SICI recorded with anterior-to-posterior induced current direction (SICIAP) are lower in ALS.• SICIPA and SICIAP do not correlate in ALS or controls, indicating that they reflect distinct physiological processes.


Introduction
While changes in electromyography (EMG) can provide objective biomarkers of lower motor neuron degeneration in ALS (de Carvalho, 2020), there is no universally accepted quantitative biomarker of upper motor neuron dysfunction.Transcranial magnetic stimulation (TMS) is the non-invasive application of a brief magnetic pulse over the scalp which can activate upper motor neurons indirectly, leading to motor unit action potentials, recorded via EMG as motor evoked potentials (MEPs) (McMackin et al., 2019).''Paired pulse" TMS protocols apply a conditioning stimulus (CS) prior to the corticospinal volley-generating test stimulus (TS), with the CS-activated circuitry inducing a net inhibitory or facilitatory influence upon the excitability of upper motor neurons.This leads to a change in the average MEP amplitude evoked by a TS of a specific intensity.As a result, measuring the difference in TS intensity required to evoke a desired MEP amplitude when the CS is present versus absent can be used to quantify motor cortical excitation or inhibition, and the effects of ALS thereon.
All studies of SICI in ALS to date have used a posterior-to-anterior (PA) coil orientation.This refers to the induction of electrical currents directed perpendicularly across the precentral gyrus, from posterior to anterior, when the TMS pulse is applied over the motor cortex (Janssen et al., 2015).However, SICI 3ms is found to have greater magnitude when an anterior-to-posterior (AP) coil orientation (SICI AP ) is used (Cerins et al., 2022;Cirillo and Byblow, 2016).It has therefore been proposed that SICI AP could be a more sensitive measure of cortical ALS pathology (Cirillo and Byblow, 2016).
The aims of this study were to determine whether the use of an AP coil orientation could provide additional sensitivity in the detection of ALS pathology compared to the PA orientation, in addition to determining if we could replicate prior reports of sensitive discrimination of people with ALS and controls on the basis of SICI and ICF measures.

Ethical approval
Ethical approval was obtained from the ethics committee of St. James's Hospital (REC reference: 2017-02).All participants were over 18 years of age and could provide informed written or verbal consent (in the presence of two witnesses) prior to participation.Except for preregistration, all work was performed in accordance with the Declaration of Helsinki.

Recruitment
Participants were screened according to the TMS screening questionnaire of Rossi et al. (Rossi et al., 2011) and excluded if contraindications to TMS were identified.Those in the ALS cohort were diagnosed with Possible, Probable or Definite ALS in accordance with the El Escorial Revised Diagnostic Criteria.Healthy controls included neurologically normal individuals recruited from an existing population-based cohort of individuals who registered interest in volunteering for neurological research.A total of 35 people with ALS and 39 healthy controls participated.

Electromyography
Participants were seated upright in a sofa-style chair with wide arm rests and asked to place their arms in their lap or on the arm rest, with the elbow at an angle of approximately 90-120°.EMG activity was recorded at a sampling frequency of 10 kHz from the abductor pollicis brevis (APB) of the dominant hand.For further details see ''Electromyography recording protocol" in Supplementary Methods).

Transcranial magnetic stimulation
Monophasic magnetic stimuli were delivered to the scalp via a DuoMag MP Dual stimulator (Deymed Diagnostics s.r.o., Hronov, Czech Republic), equipped with a 50 mm mid-diameter figure-ofeight coil.The axis of intersection between the two loops was oriented at 45°to the sagittal plane to induce PA current flow across the precentral gyrus.Once the scalp hotspot for motor cortex stimulation was identified (see ''TMS hotspotting" in Supplementary Methods"), resting motor threshold (RMT, the percentage of maximum stimulator output (%MSO) at which 50 % of stimuli elicit an MEP of at least 50 lV) and threshold hunting target (THT, the %MSO at which 50 % of stimuli elicit an MEP of at least 200 lV) were measured using fully automated parameter estimation by sequential testing (PEST, see ''Adaptive threshold tracking" in Supplementary Methods) with PA current flow (PA-RMT and PA-THT).This was followed by rotation of the coil through 180°to achieve AP current flow across M1.The RMT and THT for the AP orientation (AP-RMT and AP-THT) were then obtained at the same position.
Next, paired pulse protocol variants were measured.A 10 ms ISI was used to measure ICF as this ISI was previously reported to provide the greatest ICF-based discrimination between those with ALS and controls (Vucic and Kiernan, 2006), SICI was measured at both 1 and 3 ms ISIs.While PA coil orientation was used for all paired pulse measures of interest, the AP coil orientation also was used to elicit SICI 3ms .This decision was based on previous studies wherein SICI 3ms is found to differ demonstrably when AP and PA orientations are compared, whereas no significant difference was demonstrated for SICI 1ms (Cirillo et al., 2018;Cirillo and Byblow, 2016;Mooney et al., 2018;Wessel et al., 2019).The paired pulse protocols investigated were: SICI 1ms-PA , SICI 3ms-PA , SICI 3ms-AP , and ICF 10ms-PA .For all paired pulse protocols, conditioning pulses were applied at 70 % of the RMT recorded using matching coil orientation.Paired pulse protocols were delivered in random order with rotation of the coil as required between recording of measures employing different orientations.For all TMS protocols, MEP peak-to-peak amplitudes were measured within the 15-50 ms window following test stimulation.
A neuronavigation system was not available for live tracking of coil position relative to target locations.Such navigation systems are very costly and not widely available in research and clinical settings.Further, such tracking systems require participants to wear a head tracker (e.g. a headband or glasses secured to the face) which can be uncomfortable and limit relaxation when worn for the duration of a recording session.We therefore employed a landmark-based system to ensure that pulses were consistently delivered to the same location for both AP and PA orientations (see ''Coil positioning protocol" in Supplementary Methods).

Maximum compound muscle action potentials
Maximum compound muscle action potentials (CMAPs) in the dominant hand APB were evoked using electrical stimulation delivered over the median nerve by a Digitimer DS7A stimulator (Digitimer Ltd., Welwyn Garden City, UK) at either the elbow or the wrist (for further information see ''Compound muscle action potential measurement" in Supplementary Methods).

Paired pulse TMS analysis
Facilitation/inhibition was defined as the percentage change in test stimulator output necessary to evoke an MEP of target amplitude (200 lV) in the presence of the conditioning stimulus (i.e., the conditioned threshold target, CTT) compared to in its absence (i.e., the THT) as follows: with positive values indicating inhibition and negative values indicating facilitation.To draw comparison to prior studies of ''averaged SICI" as an ALS biomarker, mean SICI values were calculated across ISI for PA orientation, and separately across orientations and ISIs, for those individuals for whom all measures could be obtained.

CMAP analysis
For all CMAP-associated signal analyses, latency and maximum peak-to-peak amplitude were examined within the 2-30 ms window following test stimulation.For further details see ''Compound muscle action potential analysis" in Supplementary Methods.

Statistics
Age and sex differences Ratio of male to female sex in each cohort was compared by chisquared test.Ages of each cohort's participants were compared by Mann-Whitney U test.

CMAP amplitude and latency
Spearman rank correlation analysis was performed to initially test for the relationship between age and CMAP median latency or maximum amplitude, as groups differed slightly in age distribution.Bonferroni correction was applied, such that differences with p values below 0.025 were deemed significant.As age showed significant correlation with CMAP median latency (rho = 0.42, p = 0.005), CMAP characteristics were compared between groups using linear mixed effects models, with group, site of stimulation, and their interaction, as well as age, incorporated as fixed effects as follows (according to Wilkinson notation): 'Median Latency = Group*Site + Age+(Group*Site|Subject)' and 'Maximum Amplitude = Group*Site + Age+(Group*Site|Subject)' On the basis that Cohen's d is calculated as group mean difference divided by pooled standard deviation when simply comparing groups (Cohen, 1988), Cohen's d values for Site [i.e., location of stimulation] and Group were estimated as the coefficient values (corresponding to the mean difference) divided by the standard deviation of the residuals.A 95 % confidence interval for this value was calculated within the fitlme function of MATLAB R2016a.

Conditioning effects and group differences
Shapiro-Wilk (Shapiro and Wilk, 1965) and Anderson-Darling (Anderson and Darling, 1952) tests indicated that some TMS measures did not conform to a normal distribution, therefore, nonparametric statistics were employed.To test if the expected SICI/ ICF effects were reliably evoked, sign rank testing of control inhibition/facilitation (%) values (compared to zero) was performed for each paired pulse protocol.To account for multiple comparisons, these four p values were corrected to obtain a false discovery rate (FDR) below 5 %, determined by the Benjamini-Hochberg method (Benjamini, 2010), and only those corrected p values below 0.05 were considered significant.To test for abnormalities in SICI/ICF in ALS, Mann-Whitney U testing was performed, comparing inhibition/facilitation values between those with ALS and controls for each single or paired pulse protocol.To account for multiple comparisons, p values were corrected to obtain a positive FDR below 5 % by the Benjamini-Hochberg method (Benjamini, 2010), and only corrected p values below 0.05 were considered significant.Brown-Forsythe testing (Brown and Forsythe, 1974), which is robust to non-normality, was performed for all TMS measures and in the case of all measures a p > 0.05 indicates that the null hypothesis of equal variance between cohorts could not be rejected.

Potential ceiling effect bias
In some participants, a ceiling effect occurred where it was not possible to apply sufficiently high stimulus intensities (i.e., >100 % MSO) to counteract an inhibitory conditioning effect.In these participants, some SICI measures could not be measured directly.Inability to record SICI in these individuals could potentially introduce bias to comparative analyses of people with ALS and controls by excluding or underestimating SICI values of those with high thresholds.In most of these participants, MEPs > 50 lV but < 200 lV were repeatedly evoked at 100 % MSO during application of SICI protocols.Further, the difference between these individuals' threshold stimulation intensities for 200 lV and 50 lV target amplitudes was known in the absence of inhibitory conditioning (i.e., THT minus RMT).We therefore approximated SICI values in these individuals based on the assumption that the difference between the threshold stimulation intensities required to evoke 50 lV and 200 lV does not change with the introduction of the inhibitory conditioning stimulus.Under this assumption, we approximated the true conditioned threshold in these individuals as: CTT ¼ 100% MSO þ ðTHT À RMTÞ and used this CTT value to estimate %Inhibition/Facilitation (as per formula above) in these participants.Our group comparison statistical analyses were then repeated, including these approximate values for most of those who were excluded from the primary analysis.This additional, supplementary analysis was performed purely to investigate if the apparent ALS-related abnormalities observed in this study could be driven purely by bias caused by this ceiling effect.However, the assumption upon which we have based these SICI estimates has not yet been proven, and cannot be tested using on this dataset due to lack of SICI measures tracked with a target amplitude of 50 lV.Therefore, we do not recommend the use of this SICI approximation method beyond this purpose.

Discrimination ability and effect size
To quantify the potential of paired pulse TMS measures to discriminate people with ALS from controls, area under the receiver operating characteristic curve (AUROC) was calculated for each TMS parameter for each ISI and coil orientation used.Effect size was measured by Hedge's g, a measure similar to Cohen's d which includes a correction factor for potential bias caused by smaller sample sizes (Lin and Aloe, 2021).Values and confidence intervals for Hedge's g and AUROC values were calculated via the ''Measures of Effect Size" MATLAB toolbox of Hentschke and Stüttgen (Hentschke and Stüttgen, 2011) applying bootstrapping 1000 times.

Correlation analyses
Spearman's rank correlation analysis was performed to determine the linear association between each TMS measure and 1) disease duration (time since self-reported first symptom onset) and 2) revised ALS functional rating scale (ALSFRS-R) total score.ALSFRS-R score recorded within two months of study participation was available for 26 people with ALS and scores for an additional 4 people with ALS were estimated based on interpolation of two ALSFRS-R measures taken before and after participation.To account for multiple comparisons, p values were corrected (across p values for each TMS measure, separately for ALSFRS-R scores and disease duration) to obtain a positive FDR below 5 % by the Benjamini-Hochberg method (Benjamini, 2010), and only corrected p values below 0.05 were considered significant.Spearman's rank correlation analysis was also used to investigate the relationship between the different paired pulse TMS measures.

Final cohort
The minimum requirement for the inclusion of an individual in this analysis was a measurable RMT PA , obtained below 100 % MSO.  1).The two cohorts were sex-(chi 2 = 0.0037, p = 0.951) but not age-(p = 0.041) matched.Spearman correlation analyses failed to reveal associations between age and any of the TMS-based measures.Maximum CMAP data were collected from 29 people with ALS and 31 controls who provided RMT PA data.The numbers of datapoints successfully collected for each measurement are given in Table 2.

Motor thresholds
When the PA orientation was used, the 50 lV RMTs of the ALS group were higher (in terms of %MSO) than those of the controls (Fig. 1, Table 2).

SICI
Sign rank testing demonstrated that significant inhibition was achieved in controls (SICI 1ms z = 5.03, p < 0.001; SICI 3ms-PA z = 4.85, p < 0.001; SICI 3ms-AP z = 3.73, p < 0.001) for all SICI measures (Fig. 2).Sign rank testing of paired SICI 3ms-PA and SICI 3ms-AP values in controls (n = 18) identified that SICI was of significantly greater magnitude when AP coil orientation was employed (p < 0.001).When SICI measures were compared between those with ALS and controls (Table 2), less inhibition was observed in those with ALS in the case of the SICI 3ms-AP (Hedge's g = -0.93)and SICI 3ms-PA (Hedge's g = -0.65)conditions.In respect of the SICI 1ms-PA , the magnitude of the difference was smaller (Hedge's g = -0.61)and did not satisfy the FDR adjusted criterion for statistical significance.Inclusion of estimated values for those with conditioned thresholds > 100 % MSO (Table S1), increasing the number of datasets compared, did not change our findings, with the exception that the inhibition arising from SICI 1ms-PA was deemed significantly lower in those with ALS than in controls (Figure S2, Table S1).When those taking baclofen or amitriptyline, or those for whom recordings were obtained from the nondominant hand, were excluded from the analysis of each SICI protocol, the pattern of outcomes did not change for SICI 3ms measures (there were no changes to which comparisons were deemed significant at a 5 % FDR).

Relationship between SICI and ICF measures
Across participants, significant positive correlations were observed between SICI 1ms-PA and SICI 3ms-PA , as well as between (PA) ICF and SICI 1ms-PA and between (PA) ICF and PA SICI 3ms .By contrast, no significant correlations were identified between SICI 3ms-AP and SICI 3ms-PA , SICI 1ms-PA or ICF (Table 3).In keeping with the absence of positive correlation between SICI PA and SICI 3ms-AP measures in people with ALS (rho = -0.36,p = 0.273), individuals with ALS displaying the least SICI 3ms-PA displayed relatively high SICI 3ms-AP values compared to the rest of the ALS cohort and vice versa (Fig. 3F, circled data points).

Clinical correlations
A negative correlation was observed between ALSFRS-R 5a (cutting food and handling utensils) and THT recorded with PA orientation (rho = -0.64,p < 0.001).No significant correlations between TMS-based measures and ALSFRS-R total scores, other ALSFRS-R subscores, or self-reported disease duration were identified.

Differences in CMAP measures between people with ALS and controls
Median CMAP latency was longer in those with ALS compared to controls (Fig. 4 left, p = 0.035, d[95 % confidence interval] = 0.28[0.020-0.54]).CMAP amplitude was smaller in those with ALS (Fig. 4 right, p = 0.005; d[95 % confidence interval] = -0.38[-0.64-À0.12]).Group*stimulation site interactions were not statistically significant for amplitude (p = 0.491) or latency (p = 0.565), indicating that groups did not differ in how CMAP values were influenced by the location at which electrical stimulation was applied.Omitting those taking baclofen or amitriptyline or those for whom the non-dominant hand was tested did not affect these findings (values for those in these subcohorts are illustrated in Fig. 4), with the exception of latency not being found to be significantly longer in ALS if only dominant hand measurements were included (p = 0.060).All other p values and effect sizes for the Group coefficient remained similar when analysis was repeated with omission of individuals from any of these subcohorts (latency: p = 0.030-0.035,d = 0.25-0.30,amplitude: p = 0.006-0.009,d = -0.36-À0.38).

Discussion
We examined if the use of an AP coil orientation during SICI measurement could provide additional sensitivity in the detection of ALS pathology, compared to the PA orientation.We also examined if we could replicate prior findings of ALS-related changes in SICI and ICF using a novel, fully automated, PEST-based threshold tracking TMS protocol.
Table 2 Summary statistics for each TMS parameter recorded.Numbers of ALS and control datasets recorded are listed under Pn and Cn respectively.Group p (g) column values refer to p-value and Hedge's g for the comparison of ALS and control values.Bolded values represent those correlations deemed significant at a 5 % FDR.95 % CI -95 % confidence interval.Numbers of ALS and control datasets excluded due to CTT > 100 % MSO are listed under P > 100 and C > 100 respectively.P values listed are uncorrected.Mean -Mean inhibition values across 1 and 3 ms ISIs.Those coefficient values with corrected p values (at 5 % false discovery rate) < 0.05 are RMT -Resting motor threshold.THT -Threshold hunting target.SICI -Short-interval intracortical inhibition.ICF -Intracortical facilitation.ISI -Interstimulus interval.PA -Posterior-to-anterior.AP -Anterior-to-posterior.AUROC -Area under the receiver operating characteristic curve.# / ## Data for this measure not recorded from one participant/two participants respectively in this cohort due to difficulty relaxing/ time constraints.A number of studies in control cohorts have demonstrated that SICI 3ms is of greater magnitude when evoked with AP coil orientation compared to the typically-employed PA coil orientation (Cerins et al., 2022;Cirillo et al., 2018;Cirillo and Byblow, 2016;Sale et al., 2016).We therefore hypothesised that motor cortical disinhibition in ALS, measured with SICI 3ms , would be substantially more evident when measured with AP coil orientation compared to PA coil orientation.First, we replicated previous findings that SICI 3ms-AP is of significantly greater magnitude than SICI 3ms-PA in controls.However, we did not find disinhibition to be substantially more evident in ALS when employing AP coil orientation.The overlapping 95 % confidence intervals of the AUROC and Hedge's g values for these two measures indicated similar discrimination abilities.

Parameter
Notably, the magnitude of SICI 3ms-AP and SICI 3ms-PA values within individuals did not exhibit a positive correlation in either controls (rho = 0.04) or people with ALS (rho = -0.36).It appears, therefore, that these measures capture distinct facets of motor network activity induced by this paired pulse TMS protocol.A similar lack of correlation between SICI 3ms-PA and SICI 3ms-AP measures was recently reported in controls (Pavey et al., 2023).The study of Pavey et al., (2023) differs slightly from the control aspect of this study in that serial tracking of SICI measures across 1-7 ms ISIs was employed, as opposed to independent tracking of each measure (as employed here).Serial tracking has been found to bias SICI 3ms values towards the values of SICI 1-2.5ms (Tankisi et al., 2022), potentially blurring the distinction between these measures.This might explain why Pavey et al., did not find SICI 3ms-AP to be of significantly greater magnitude than SICI 3ms-PA .The study of Pavey et al., provides important additional insights as a broader range of ISIs were examined with AP, PA and lateromedial coil orientations.In keeping with the rationale underpinning the ISIs selected for our protocol, Pavey et al. found that SICI AP and SICI PA were of greatest magnitude when 1 and 3 ms ISIs were employed.
We subsequently averaged values across SICI measures recorded with different coil orientations and ISIs to determine whether the composite would improve the sensitivity with which people with ALS could be discriminated from controls.The AUROC for the composite was not greater than for SICI 3ms-AP alone (or markedly different from that of SICI 3ms-PA alone).This finding might be explained by ALS causing impairment to a physiological element common to AP and PA SICI 3ms , orthogonal to the dimension on which these measures are discriminated (i.e., accounting Table 3 Correlation statistics between SICI and ICF measures across cohorts.n -Cohort size (individuals for whom both values could be measured), 95% CI -Bootstrapped 95% confidence interval.Bolded values represent those correlations deemed significant at a 5% FDR.SICI -Short-interval intracortical inhibition.ICF -Intracortical facilitation.PA -Posterior-to-anterior orientation.AP -Anterior-to-posterior orientation.for their relatively low inter-correlation in controls and the overall cohort).However, if ALS were causing such an impairment, that is captured correspondingly by both measures, a positive correlation between the measures would be expected in those with ALS.Instead, we observed a weak negative correlation between the measures in people with ALS (rho = -0.36,p = 0.273).While our correlation analyses including AP-based measures are more limited in statistical power due to fewer available datapoints, even the combined cohort (n = 29) showed no correlation between these measures (rho = 0.19, p = 0.313).Further, those with ALS who exhibited the lowest degree of SICI 3ms-PA had relatively high SICI 3- ms-AP values and vice versa.Therefore, our results suggest that the nature of motor cortical dysfunction, and the TMS measures which best detect it, differs across individuals with ALS.As such, averaging across discordantly affected SICI measures may not produce a measure of improved discriminative power.Overall, our findings indicate that measuring SICI with both AP and PA coil orientations can provide additional measures of motor cortical pathophysiology in ALS, but these measures should be considered separately, and not averaged.Such discordant effects of ALS on SICI PA and SICI AP within an individual might be explained by differences in the contribution of I-wave facilitation (or short ICF, SICF) to these two measurements.When applying standard stimulation parameters used to evoke SICI, SICF can be concurrently evoked (Awiszus et al., 1999), despite a net inhibitory effect being observed.Therefore, lower SICI 3ms measurements may reflect a combination of greater facilitation and/or less inhibition.SICF PA has been shown to be elevated in ALS (van den Bos et al., 2018), such that little/absent SICI 3ms-PA in ALS may in part be a reflection of increased I-wave facilitation.However, Pavey et al., (2023) found that, in healthy controls, SICF 3ms is evoked only when PA, and not AP, coil orientation is applied.Therefore, enhanced I-wave facilitation might not contribute to a lower SICI 3ms-AP .If this is the case, individuals with ALS with increased I-wave facilitation would be expected to show lowering of SICI PA , not SICI AP .By contrast, individuals with relatively unaffected I-wave facilitation but decline in GABA-Aergic activity would be expected to show lowering of SICI AP to a greater extent than SICI PA .

Pair of measures
However, a number of further studies are warranted to test if this hypothesis explains our observations.For example, pharmacological studies of the relationship between change in tracked SICI measures and GABA-Aergic function (e.g.employing benzodiazepines) remain to be performed in ALS, while epidural recording studies (albeit practically challenging to perform) would enable examination of the contribution of I-wave facilitation to SICI AP and SICI PA measures in ALS.Future TMS-based studies which concurrently examine SICF and SICI with both AP and PA coil orientations in ALS are also required.Finally, we found no correlations between SICI and metrics of disease progression (ALSFRS-R and disease duration) to indicate that disease stage explains differences in extent of SICI PA and SICI AP lowering between individuals with ALS.However, longitudinal examination of these measures is warranted in ALS to more definitively determine the contribution of disease stage to this heterogeneity.
It is important to note that when using AP stimulation, a higher intensity is required to evoke MEPs than when the orientation that induces PA current flow is used.As such, the elicitation of SICI with AP stimulation may not be possible for some people with ALS who exhibit high RMT values due to muscle wasting or cortical inexcitability.However, those at very early symptomatic stages of ALS, when diagnostic markers would be applied clinically, have been reported to have normal, or even reduced, motor thresholds (Huynh et al., 2019), such that SICI AP is likely to be feasibly measured at this early stage.

Replication of previously reported findings in ALS
While lower SICI 1ms-PA was observed in people with ALS, this was only observed when specific medications were excluded, or when estimated values for those with thresholds > 100 % MSO were included, indicative of the heterogeneity with which SICI 1ms is affected in ALS.Tankisi et al. (Tankisi et al., 2022) and Vucic et al. (Vucic et al., 2011) noted that SICI 3ms-PA provides marginally superior discrimination of people with ALS from controls compared to SICI 1- ms-PA .While we also report a greater AUROC for SICI 3ms-PA than SICI 1- ms-PA (0.70 and 0.66 respectively) the AUROC values for the two conditions show broad, considerably overlapping corresponding confidence intervals (see Table 1), indicating that the extent to which SICI 1ms and SICI 3ms are lowered is heterogenous between individuals with ALS.As these different SICI variants are thought to be associated with discriminable physiological processes (Fisher et al., 2002;Stagg et al., 2011;Ziemann et al., 2015), averaging may mask clinically or prognostically relevant differences in cortical pathophysiology that discriminate among individuals with ALS.Notably, we observed a significant, weak, positive correlation (rho = 0.37, p = 0.006) between SICI 3ms-PA and SICI 1ms-PA when cohorts were combined.A positive correlation between these measures has previously been identified in older, but not younger, healthy adults (Hermans et al., 2018).Therefore, this relationship may reflect an aspect of the aging process which similarly affects both measures, rather than a common physiological underpinning of both measures.It must be highlighted that the specific physiology reflected by SICI 1ms is difficult to interpret, as the literature surrounding its underpinnings remains limited and, in some cases, to be replicated (Fisher et al., 2002;Stagg et al., 2011).
ICF did not differ between people with ALS and controls.Although, when considered at group level the conditioning stimulus had facilitatory effect for the controls, this was not the case for every individual.Moreover, ICF 10ms was strongly correlated with SICI 3ms .This high degree of association and the inconsistent manifestation of ICF 10ms among healthy individuals accords with the supposition that the inhibitory networks generating SICI are also activated during ICF (Schwenkreis et al., 1999;Ziemann et al., 1998).This consideration, taken together with indications that ALS-related variations in ICF do not conform to a clear pattern (Geevasinga et al., 2014;Vucic and Kiernan, 2006), suggest that this specific measure is very unlikely to have diagnostic or prognostic utility.

Limitations
As participants in this study were also participating in further research studies on the same day, only specific combinations of ISI and coil orientation were tested to minimise participant fatigue.While averaging between SICI values was performed to provide some comparability to other studies of SICI in ALS, our ''average" SICI measures are not directly comparable to those calculated based on a more detailed spectrum of ISIs.Further, AP coil orientation was only tested for SICI 3ms , and not SICI 1ms (see Experimental Paradigm for the rationale of protocol selection).Going forward, further examination of SICI using a wider range of ISIs is warranted using AP coil orientation to determine the extent of additional information on cortical pathophysiology which can be obtained from use of this coil orientation.We did not use a neuronavigation device in this study, instead employing a comprehensive, landmark-based system to ensure consistency of coil positioning.Nonetheless, in the absence of a neuronavigation device, we cannot preclude the possibility that small deviations in coil positioning occurred between different measurements.
Most participants with ALS in this study were concurrently taking riluzole.Riluzole has no significant effects on CMAP, RMT or ICF (Sommer et al., 1999;Stefan et al., 2001), and only have transient effects on SICI (both with 1 and 3 ms ISIs) which return to baseline after 12 weeks (Geevasinga et al., 2016;Stefan et al., 2001).As all of those taking riluzole who were included in this study had been treated for longer than 12 weeks, we do not believe this treatment affects our findings.
Regarding heterogeneity between individuals, a recent study demonstrated that SICI PA normalises in ALS with greater upper motor neurone impairment (Tankisi et al., 2021b).As we did not collect a score specifically of upper motor neurone symptom severity, we cannot investigate how this factor may have contributed to heterogeneity of SICI in people with ALS.
Finally, it is important to highlight that, when using a threshold tracking approach, very good agreement (intraclass correlation of 0.75-0.9)has been shown between intraindividual SICI measurements.While this indicates that these measures can be employed reliably for research, caution is still warranted when applying these measures for clinical decision making purposes (Nielsen et al., 2021).

Fig. 1 .
Fig. 1.People with ALS show greater RMT, when PA orientation is used, compared to controls.Boxes represent the interquartile range, with horizontal black line within each box representing the median.Within the ALS cohort, cyan crosses represent values of those where the non-dominant hand was tested, magenta dots represent values of those taking baclofen, bright red dots represent values of those taking amitriptyline and bright blue dots represent values of those not taking riluzole.P values in subfigure titles are those for Mann-Whitney U test comparisons between cohorts.RMT -Resting motor threshold.THT -Threshold hunting target.PA -Posterior-to-anterior induced current direction.AP -Anterior-to-posterior induced current direction.%MSO -Percentage of maximum stimulator output.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2 .
Fig. 2. People with ALS show less SICI when an ISI of 3 ms is used.Boxes represent the interquartile range, with horizontal black line within each box representing the median.Within the ALS cohort, cyan crosses represent values of those where the non-dominant hand was tested, magenta dots represent values of those taking baclofen, bright red dots represent values of those taking amitriptyline and bright blue dots represent values of those not taking riluzole.P values in subfigure titles are those for Mann-Whitney U test comparisons between cohorts.SICI -Short-interval intracortical inhibition.ICF -Intracortical facilitation.PA -Posterior-to-anterior induced current direction.AP -Anterior-to-posterior induced current direction.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3 .
Fig. 3. Scatter plot illustrating the relationships between SICI and ICF values within people with ALS (red) and controls (blue).Circled values in (F) highlight that the two individuals with ALS with the least SICI 3ms-AP had SICI 3ms-PA values amongst the highest of the ALS cohort, and vice versa.Spearman's rank correlation p-values are given for controls only (p control ), people with ALS only (p ALS ) and the overall dataset (p total ).SICI -Short-interval intracortical inhibition.ICF -Intracortical facilitation.PA -Posterior-toanterior induced current direction.AP -Anterior-to-posterior induced current direction.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Thirty-five people with ALS (11 female, age median [range]: 64 [41-79] years, 2 left-handed) and thirty-nine controls (12 female, age median [range]: 59 [34-76] years, 1 left-handed) satisfied this requirement (Table

Table 1
Clinical characteristics of final cohort of 35 people with ALS.U-Upper limb, L-Lower limb, l-left side, r-right side, e-both sides affected.ALSFRS-R scores measured within 90 days of TMS recording were available for 30 out of 35 people with ALS.Record of El Escorial category at time of diagnosis was retrievable for 25 out of 35 people with ALS.