Exteroceptive Reflexes in Stiff Person Syndrome, Other Neurological Conditions and Healthy Controls Authors: Belinda Cruse, Duncan Wilson, Hugo Morales Briceno, Pamela Garcia, Victor Zhang, Jessica Qiu, Tomas Kalincik, Neil Mahant, Victor Shue Chung Fung Published: 2026 ABSTRACT AbstractBackgroundStiff person syndrome (SPS) is a rare disabling neurological condition with overlapping symptomatology with more common neurological conditions. Exaggerated exteroceptive reflex mechanisms are thought to play a role in its pathophysiology.ObjectivesWe aim to describe the results of neurophysiological exteroceptive reflex testing in subjects with SPS, healthy controls, and other neurological conditions causing axial and lower limb muscle stiffness, pain and spasms.MethodsWe performed a prospective cohort study, of 59 participants with a variety of neurological conditions, healthy controls and 11 with SPS. Retrospective data from participants with SPS who had prior exteroceptive reflex testing performed in an equivalent manner were also included in our analysis. Reflex responses were recorded by simultaneous surface electromyography (sEMG) of axial and lower limb muscles following repeated stimulation of the median and tibial nerves.ResultsLatencies following repeated activation of median and tibial nerves demonstrated a moderate to very strong correlation at most EMG recording sites within individuals. Using a mixed effects logistic model shorter mean latencies were observed in those with Parkinson's disease, healthy controls, low back pain and serotonergic/noradrenergic antidepressant use compared with SPS following initial stimulation and when the shortest latency at each site within individual participants were analyzed. No study group differed significantly from SPS in the duration of responses following stimulation nor the presence or absence of after‐bursts.ConclusionThe use of exteroceptive testing as performed in this study did not differentiate SPS from other neurological conditions or healthy controls. Introduction Stiff person syndrome (SPS) is a rare neurological condition characterized by involuntary muscle activation and superimposed stimulus‐induced spasms. 1 The most affected regions are typically the axial and proximal limbs however the condition exists on a continuum with more limited, focal forms (stiff limb syndrome), and more severe forms associated with encephalomyelitis and myoclonus (PERM). 2 Clinical observation, detailed neurophysiological interrogation of affected individuals, and later the recognition of detectable antibodies in some sufferers have contributed to the current understanding of the pathophysiology of SPS. While not specific for the condition, inhibitory antibodies have been described against glutamic acid decarboxylase (GAD) in 70–80% cases, glycine in about 10% cases, amphiphysin in about 5% cases, and dipeptidylpeptidase‐like protein‐6 (DPPX) in about 3% cases. 3 It is now widely accepted that SPS and associated syndromes are caused by immune‐mediated inhibition of (inhibitory) central nervous system GABAergic pathways, ultimately resulting in the disinhibition of central motor pathways. 4 , 5 , 6 , 7 SPS remains a challenging diagnosis however; many of the presenting symptoms are non‐specific, about 20% of cases in specialty clinics do not demonstrate a known pathogenic antibody, 8 described autoantibodies are non‐specific when present especially at low or intermediate titers, and there is a lack of disease‐specific biomarkers. Exaggerated exteroceptive (cutaneomuscular) reflex mechanisms were first thought to play a pathological role in SPS due to the uniformity of observed spasms in affected patients following acoustic and somatosensory stimuli. 5 , 9 , 10 , 11 A neurophysiological method of evaluating exteroceptive reflexes has been described, measuring axial motor responses using intramuscular electromyography (EMG) following trains of electrical stimuli to the median nerve at the wrist and medial plantar nerve of the foot. 5 , 10 , 12 Stereotyped jerky spasms were precipitated by peripheral nerve stimulation, corresponding to synchronized bursts on EMG of axial and proximal limb muscles at latencies of around 70, 120 and 170 milliseconds following stimulation, then waning. This pattern was termed spasmodic reflex myoclonus, 10 and was observed in all eight SPS participants and three of eight participants with prominent parkinsonism but not in any other neurological condition studied nor in any normal participants. 10 In this study we perform neurophysiological exteroceptive reflex testing using surface EMG in subjects with SPS, other neurological conditions (some of which sharing symptoms with SPS), and healthy controls to evaluate the applicability of this neurophysiological test in clinical practice. We also evaluate the reproducibility of responses within individual subjects and consider the physiologic mechanisms underlying these reflexes. Methods Participants were recruited from the Royal Melbourne Hospital Neurophysiology Department and the Westmead Hospital Movement Disorders Units between June 2023 and November 2024. Participants over 18 years of age were included from the following categories: GAD antibody positive (GAD Ab+) SPS, Parkinson's Disease (PD), myelopathy causing lower limb spasticity (due to multiple sclerosis (MS) or hereditary spastic paraparesis (HSP)), lower limb dystonia, back pain due to radiologically confirmed spondylosis, functional movement disorders with axial jerks, myoclonus. These categories were chosen due to overlapping symptomatology with SPS (ie, axial and lower limb muscle stiffness, pain and/or spasms). In addition, people with no neurological conditions taking selective serotonin‐ (+/− norepinephrine) reuptake inhibitors (SSRI/SNRI) (noting a theoretical enhancement of spinal reflexes), chorea and tics (noting hyperkinetic features with no expectation of enhanced reflexes), healthy controls with no known medical conditions and demyelinating nerve root pathology (expected to prolong reflex responses) were also included for comparison. Retrospective data from participants with GAD Ab+ SPS who had undergone exteroceptive reflex testing were also included if comparable neurophysiological techniques were used. Participants were defined as having GAD Ab+ SPS if they demonstrated GAD‐antibodies in serum +/− cerebrospinal fluid (CSF), and at the time of performance of the test clinical features were consistent with proposed diagnostic criterion: 8 stiffness of the axial muscles, superimposed painful spasms triggered by anxiety, tactile or auditory stimuli, and absence of other neurological findings to suggest alternative diagnosis. These clinical criteria were confirmed by investigators for prospective participants at the time of testing or in the case of retrospectively performed testing present on contemporaneous clinical records. The study was approved by the Royal Melbourne Hospital Ethics Committee and all participants provided informed consent. Age, sex at birth, and where applicable presence (and titre) of antibodies against GAD, glycine, amphiphysin and DPPX were recorded. Exteroceptive reflex testing was performed on participants using a protocol adapted from published case series. 5 , 10 Activation of the median and tibial nerves were performed with stimulus‐trains of square‐wave pulses of 0.2 ms duration each at 200 Hz as in prior publications. 5 , 10 Stimulus‐trains of four pulses (total duration of the stimulation train 10 ms) were used to stimulate median nerve at the wrist and trains of eight pulses (duration of stimulus train 20 ms) of the tibial nerve at the ankle. Each site was stimulated up to eight times depending on participant tolerance at each site at the maximally tolerated stimulus amplitude (milliamps, mA). At least 2.3 times the minimal intensity required to generate visible muscle contraction (the motor threshold, T) was the goal of stimulation, if tolerated, in line with previous publications to ensure adequate stimulation. 5 , 10 The stimulation protocol used in this study was unchanged from earlier publications except for the site of tibial nerve stimulation; to improve tolerability we stimulated the tibial nerve posterior to the medial malleolus rather than the more distal part of the nerve, the medial plantar nerve, used in earlier publications. Surface EMG responses following repeated stimulation of the median then tibial nerves were recorded. Multichannel surface electromyography (sEMG) was performed using silver chloride recording cup‐electrodes. Skin at each site and its reference were prepared by gently exfoliating the skin. Conductive gel is placed underneath the electrode and the electrode fixed using 3M Transpore™ or 3M Micropore™ tape. Participants with prior allergy to medical adhesives or gels were not included. Previous publications 5 , 10 recorded responses of peripheral and paraspinal muscles using intramuscular needle electrodes rather than sEMG. Eight sites were simultaneously recorded ipsilateral to the side of the nerve stimulated: midpoint of the sternocleidomastoid (SCM) (reference electrode over the mastoid process), paraspinal muscles at the 7th cervical, 4th 10th and 12th thoracic, 4th lumbar spinal levels (C7, T4, T10, T12, L4 respectively), the midpoint of rectus femoris (RF, reference electrode over the patella), and midpoint of tibialis anterior (TA, reference electrode over the patella). Paraspinal muscles were recorded 1.5 cm lateral to the spinous process at the respective level and reference electrode over the spinal process of the same level. To avoid contamination of results by preceding activation or habituation 5 , 10 the patient was allowed to fully relax between stimuli, and the stimulus was given at random intervals at least 30 seconds apart without forewarning. The studies were performed using a Natus® Nicolet EDX® EMG system at the Royal Melbourne Hospital. At Westmead, a Digitimer Ltd constant current stimulator (Model DS7A) was used for the exteroceptive reflex study. An 8‐channel (Contact precision instruments) amplifier was used for EMG recording. An in‐house user interface was designed using LabVIEW (version 8.0), and MATLAB (R2006b version) was used to analyze the recording. The amplifier (Contact precision instruments). Filter settings were set at high pass 3 Hz, low pass 10 kHz with 50 Hz notch filter. The following data was documented following median and tibial nerve stimulation in all participants: latency of sEMG activity in each channel following each stimulation, duration of initial recordable muscle activity, and presence of after discharge(s) (defined as recordable motor activation after initial motor activity has ceased) at each site of recording. Neurophysiologic testing was performed by a clinical neurophysiologist or experienced neurophysiology scientist trained in the protocol at each site. Markings for analysis of onset, duration and presence or absence of after discharges were performed on the raw sEMG traces following each stimulation by a neurophysiologist or movement disorders specialist from the alternative study site and blinded to the underlying condition (or control group) of the participant. Data were not included if accurate markings were not possible due to background motor activation or artifact as judged by the blinded assessor. The neurophysiologist/scientist performing the test chose the right or left side based on neurophysiological equipment available and participant convenience, except in the case of the dystonia or PD participants in whom the most affected side was studied. Retrospective sEMG data were used only from equivalent recording sites to those performed prospectively and when the stimulus parameters were equivalent. Markings of these data for analysis were performed in the same way as for the studies performed prospectively, with data marked by a blinded neurophysiologist or movement disorders specialist from the alternative site. Baseline demographic and electrophysiological characteristics are summarized descriptively (Table 1). No statistical comparisons were performed between groups, as such tests are not appropriate for baseline characteristics in observational studies. Instead, data are presented descriptively and visually to illustrate the distribution and spread of values. To easily illustrate latency and durations between the groups, we took the median values from all muscle groups from the first run. Rostrocaudal patterns are illustrated visually in graphs using the mean latencies at each muscle in each condition. For each muscle group, the data were structured such that the latency or amplitude value from the first electrophysiological run was treated as the independent variable (X‐axis), and the corresponding value from the repeated run for the same individual was treated as the dependent variable (Y‐axis). Pearson's r was employed to verify that the sequential trials maintained a stable, highly predictable linear relationship without significant random variance between runs before pooling or utilizing the data. We have additionally evaluated the absolute consistency of the repeated measures using Cronbach's alpha across the sequential trials for each muscle group. TABLE 1Demographic information, disease category, and antibody status of studied participantsAll n = 59SPS n = 11PD n = 5Spast (MS, HSP) n = 9Dyst n = 5LBP n = 7SSRI/SNRI n = 2Healthy control n = 6Functn n = 9CIDP n = 1Chorea/Tics n = 2Myocl n = 2Age median55 (39 to 65)45 (37 to 64)59 (58 to 62)61 (45 to 63)67 (55 to 70)68 (55 to 73)43 (34 to 51)37 (32 to 42)62 (38 to 65)62 (62 to 62)51 (47 to 55)52 (41 to 63)Sex (Female%)30 (50)4 (31)1 (20)5 (56)3 (60)3 (43)2 (100)2 (33)6 (67)1 (100)2 (100)1 (50)Immunotherapy (%)7 (12)4 (33)0 (0)3 (33)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)Benzodiazepine (%)7 (12)7 (58)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)Equiv. mean diazepam dose/day (mg)1818N/AN/AN/AN/AN/AN/AN/AN/AN/AN/AGAD antibodies12 (52) n = 2311 (100) n = 110 (0)0 (0)N/AN/AN/AN/A2 (40)1 (100)0 (0)0 (0)GAD antibody titre (IU/mL)>2000 (911 to >2000) n = 8N/AN/AN/AN/AN/AN/A1040 (80 to >2000) n = 2N/AN/AN/AAbbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; Dyst, Dystonia; Funct, functional axial spasms; GAD, glutamic acid decarboxylase; LBP, low back pain (spondylosis); PD, Parkinson's disease; Spast, lower limb spasticity; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI, Myocl, myoclonus. Demographic information, disease category, and antibody status of studied participants Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; Dyst, Dystonia; Funct, functional axial spasms; GAD, glutamic acid decarboxylase; LBP, low back pain (spondylosis); PD, Parkinson's disease; Spast, lower limb spasticity; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI, Myocl, myoclonus. To statistically compare differences in latency, duration and after‐bursts we undertook mixed linear models (latency and duration) or mixed logistic models (after‐bursts). We selected a priori variables to go into the model (age and sex as fixed effects, and center as a random effect). We undertook all statistical analysis using STATA version 15.1. Results Fifty‐nine participants were recruited (Table 1). Eleven participants had a diagnosis of stiff person syndrome (Table 2). TABLE 2Characteristics of SPS participantsParticipant1234567891011Age6557804629353162404473Sex (M, F)1MMFMMFMMFMDuration of SPS symptoms (yrs)7325114unk10.591Immunotherapy (any time)NilIVIgNilNilNilNilIVIgIVIgNilIVIg, corticosteroids, rituximab, MMF, autologous HCSTNilImmunotherapy (at time of testing)NilIVIgNilNilNilNilIVIgIVIgNilAutologous HCSTNilSymptomatic therapy (at time of testing)NilNilBZD, BacBZDNilGabBZDBZDBZDBZDNilBenzodiazepine: Equiv. mean oral diazepam dose/day (mg) (at time of testing)0051000606030500Anti‐GAD Ab titre (IU/mL) Serum>2000* 338* >2000* >2000* >2000* >2000* >2000* >2000* >2000* 1483*, 52** >2000* Anti‐GAD Ab titre (IU/mL) CSFUnkunkunkunkunkunkunkunkunkunk303Mean latency following run 1 (ms)94.9697.65192.65184.45106.4130.32140.478464.3594.79111.83Mean latency following run 1 (ms)172.5114.71149.47103.13116.83130.4129.491107.02158.87165.17Abbreviations: CSF, cerebrospinal fluid; Bac, baclofen; BZD, benzodiazepines; Gab, gabapentin; GAD, glutamic acid decarboxylase; HCST, hematopoietic stem cell transplant; IVIg, intravenous immunoglobulin; MMF, mycophenolate mofetil; SPS, Stiff‐person syndrome; unk, unknown.*Ref range negative <5.0 U/mL, ELISA.**Ref range negative <8.3 IU/mL, CLIA. Characteristics of SPS participants Abbreviations: CSF, cerebrospinal fluid; Bac, baclofen; BZD, benzodiazepines; Gab, gabapentin; GAD, glutamic acid decarboxylase; HCST, hematopoietic stem cell transplant; IVIg, intravenous immunoglobulin; MMF, mycophenolate mofetil; SPS, Stiff‐person syndrome; unk, unknown. Ref range negative <5.0 U/mL, ELISA. Ref range negative <8.3 IU/mL, CLIA. All participants were stimulated at their maximally tolerated level. Motor thresholds were available for 31 prospective participants; among these the mean motor threshold for median stimulation was 11.6 mA (range 5.88–22.7 mA) and mean stimulation intensity 43.1 mA (range 12.5‐100 mA). The mean motor threshold for tibial stimulation was 19.11 mA (range 6.67–57 mA) and mean stimulation intensity 49.4 mA (range 15.7–100 mA). Stimulation was repeated in the median nerve on average 6.0 times and tibial nerve 5.7 times. Within each individual studied, the onset/latency of muscle activation at each recording site following repeated stimulation of each nerve demonstrated a fair to excellent correlation (Alpha correlation 0.64–0.93), demonstrating [good / excellent] test–retest reliability and confirming that the repeated measures are highly stable and usable except for the T4 and T8 channels following tibial nerve stimulation (Table 3), This shows our results are robust and argues against a habituation in latency of muscle activation with sequential/repeated peripheral nerve stimulation at most sites. At recording sites where a response was present following the first median or tibial nerve stimulation, there were no participants in which subsequent responses were all absent (ie, complete habituation). Figures 1 and 2 are examples of responses in a healthy control following stimulation of the median and tibial nerve respectively. TABLE 3Pearson's correlations between repeated measures (latency of muscle activation at each recording site following sequential/repeated nerve stimulation)MusclePearson's coefficientsCronbach's AlphaSequential median nerve stimulationSCM0.9, 0.7, 0.8, 0.8. 0.9, 0.70.93 (Excellent reliability)C70.9, 0.7, 0.7, 0.9, 0.9, 0.60.93 (Excellent reliability)T40.6, 0.5, 0.9,0.9, 0.7, 0.60.93 (Excellent reliability)T80.7, 0.7, 0.8, 0.6,0.6, 0.80.91 (Excellent reliability)T120.9, 0.8, 0.8, 0.8, 0.8, 0.60.87 (Good reliability)L40.6, 0.4, 0.5, 0.4, 0.6, 0.20.69 (Fair reliability)RF0.8, 0.7, 0.7, 0.8, 0.7 0.70.86 (Good reliability)TA0.5, 0.8, 0.8, 0.6, 0.8, 0.80.84 (Good reliability)Sequential tibial nerve stimulationSCM0.7,0.8, 0.7, 0.7. 0.7, 0.90.90 (Excellent reliability)C70.8, 0.6, 0.8, 0.8, 0.8, 0.90.72 (Acceptable reliability)T40.4, 0.9, 0.4,0.9, 0.3, 0.90.30 (Poor reliability)T80.6, 0.8, 0.4, 0.0,0.5, 0.00.31 (Poor reliability)T120.7, 0.9, 0.8, 0.9, 0.9, 0.90.89 (Good reliability)L41.0, 0.9, 0.9, 0.9, 0.9, 1.00.64 (Fair reliability)RF0.9, 0.8, 0.7, 0.7, 0.6 0.90.94 (Excellent reliability)TA0.9, 0.7, 0.6, 0.7, 0.6, 1.00.92 (Excellent reliability) Note: Interpretation:0.9 to 1.0 = very strong, 0.7 to 0.9 = strong, 0.5 to 0.7 = moderate, 0.3 to 0.5 = weak, 0 to 0.3 = negligible. SCM = sternocleidomastoid, paraspinal muscles at the 7th cervical (C7), 4th, 8th, 12th thoracic (T4, T8, T12) and 4th lumbar (L4) levels.Abbreviations: RF, rectus femoris; TA, tibialis anterior. Pearson's correlations between repeated measures (latency of muscle activation at each recording site following sequential/repeated nerve stimulation) Note: Interpretation:0.9 to 1.0 = very strong, 0.7 to 0.9 = strong, 0.5 to 0.7 = moderate, 0.3 to 0.5 = weak, 0 to 0.3 = negligible. SCM = sternocleidomastoid, paraspinal muscles at the 7th cervical (C7), 4th, 8th, 12th thoracic (T4, T8, T12) and 4th lumbar (L4) levels. Abbreviations: RF, rectus femoris; TA, tibialis anterior. Example: muscle activation following single stimulus‐train stimulation of the ipsilateral median nerve in a healthy participant. Waveforms shown are following a single stimulus‐train. SCM = sternocleidomastoid, paraspinal muscles at the 7th cervical (C7), 4th, 8th, 12th thoracic (T4, T8, T12) and 4th lumbar (L4) levels, RF = rectus femoris, TA = tibialis anterior. NR = no response. AB = after burst, * = cardiac artifact, ** = stimulus artifact. Example: muscle activation following single stimulus‐train stimulation of the ipsilateral tibial nerve stimulation in a healthy participant. SCM = sternocleidomastoid, paraspinal muscles at the 7th cervical (C7), 4th, 8th, 12th thoracic (T4, T8, T12) and 4th lumbar (L4) levels, RF = rectus femoris, TA = tibialis anterior. NR = no response. AB = after burst, ** = stimulus artifact. To avoid any confounding due to a theoretical conditioning effect of prior stimuli, the analyses comparing responses between individuals and participant groups, a single trace (unrectified) for each site following the first median and tibial stimulus train in each individual were used. Mean latency of muscle activation following first median and tibial nerve stimulation differed between populations however ranges overlapped (Table 4, Figs. 3 and 4, Tables S1 and S5). TABLE 4Mean responses following the first stimulus‐train to the median/tibial nerve by subject groupAllSPSPDSpastDystLBPSSRI/SNRIHCFunctCIDPChorea/TicsMyocl n 59115957269122Median nerve stimulationMean latency* (ms)97 (76 to 124)99 (90 to 130)55 (48 to 92)90 (76 to 117)125 (106 to 143)88 (86 to 122)71 (71 to 71)65 (54 to 75)115 (98 to 127)143 (143 to 143)97 (90 to 103)99 (76 to 123)Mean burst duration* (ms)149 (105 to 258)130 (79 to 266)94 (94 to 110)164 (131 to 208)98 (78 to 403)146 (108 to 217)155 (155 to 155)122 (94 to 153)273 (136 to 381)60 (60 to 60)198 (137 to 258)404 (360 to 447)Afterbursts detected** (n (%))31 (61) n = 516 (67)2 (67)5 (56)1 (25)3 (60)1 (100)5 (83)5 (56)1 (100)1 (50)1 (50)Tibial nerve stimulationMean latency* (ms)122 (81 to 150) n = 47129 (105 to 150)78 (78 to 78) n = 1133 (97 to 163)168 (140 to 180)81 (77 to 86)68 (55 to 81)72 (65 to 75)143 (139 to 150)171 (171 to 171)140 (125 to 156)117 (110 to 124)Mean burst duration* (ms)188 (109 to 149)150 (126 to 202)112 (90 to 152)253 (245 to 293)210 (99 to 210)114 (109 to 120)129 (102 to 154)158 (108 to 229)305 (245 to 372)347 (347 to 347)195 (185 to 206)368 (275 to 460)Afterbursts detected** (n (%))2 (7) n = 270 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)1 (100)0 (0)1 (50)Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; Dyst, Dystonia; Funct, functional axial spasms; HC, Healthy Control; LBP, low back pain (spondylosis); Myocl, myoclonus; n, number; PD, Parkinson's disease; Spast, lower limb spasticity; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI.*Latencies and durations across all recording sites are averaged (ms).**One or more afterburst detected at any recording site. Mean responses following the first stimulus‐train to the median/tibial nerve by subject group Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; Dyst, Dystonia; Funct, functional axial spasms; HC, Healthy Control; LBP, low back pain (spondylosis); Myocl, myoclonus; n, number; PD, Parkinson's disease; Spast, lower limb spasticity; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI. Latencies and durations across all recording sites are averaged (ms). One or more afterburst detected at any recording site. Mean latency of muscle activation following median nerve stimulation (ms). CIDP, chronic inflammatory demyelinating polyneuropathy; LBP, low back pain (spondylosis); PD, Parkinson's disease; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI, Functional, functional axial spasms. Mean latency of muscle activation following tibial nerve stimulation (ms). CIDP, chronic inflammatory demyelinating polyneuropathy; LBP, low back pain (spondylosis); PD, Parkinson's disease; SPS, Stiff‐person syndrome; SSRI/SNRI, participants with no neurological conditions taking SSRI/SNRI, Functional, functional axial spasms. Following median nerve stimulation, the mean latency following first stimulation in all groups was 97 ms (inter‐quartile range (IQR) 76 to 104). This was shortest in the PD group (55 ms, IQR 48 to 92) and longest in the patient with CIDP (143 ms (143 to 143) (Table 4, S1, Fig. 3). Following tibial nerve stimulation, the mean latency in all groups was 122 ms (inter‐quartile range (IQR) 81 to 150). This was shortest in the healthy control group (72 ms, IQR 65 to 75) and longest in the patient with CIDP (171 ms (171 to 171) (Table 4, S5, Fig. 4). Mixed effects models for latency using SPS as the reference show there was no statistical difference in mean latency between SPS, spasticity, dystonia, functional, CIDP, chorea/tic or myoclonus patients following first (run 1) median or tibial nerve stimulation for each individual or when analyzing the earliest latency (of sequential runs) at each recording site in each individual. Healthy controls were statistically shorter than SPS in both analyses following median and tibial stimulation (Tables S1, S4, S5, S8). Those with Parkinson's disease were statistically shorter than SPS in both analyses following median nerve stimulation and when earliest latencies analyzed following tibial nerve stimulation (Tables S1, S4, S8). Low back pain and SSRI/SNRI exposure groups were also statistically shorter than SPS when earliest latencies analyzed following median nerve stimulation and in one analysis each following tibial nerve stimulation (Tables S1, S4, S5, S8). While not statistically significant, the recording site(s) with the shortest mean latency following first median nerve stimulation (run 1) differed between participant groups (Fig. 3). Healthy controls and those with stiff person syndrome, low back pain, and SNRI/SSRI intake demonstrated the earliest activation at a thoracic paraspinal level. Participants with Parkinson's disease and functional disorder demonstrated earliest onset at the C7 paraspinal level and those with lower limb dystonia at the L4 paraspinal level. There was a trend towards a rostrocaudal pattern of muscle activation (ie, muscle activation occurring sequentially later in levels rostral and causal to the earliest site of onset) in some but not all participant groups. This was best observed in participants with low back pain group, functional disorder, and healthy controls (Fig. 3). Following first tibial nerve stimulation the pattern differed, with the shortest mean latency at tibialis anterior in those with Parkinson's disease, spasticity, dystonia, low back pain, SNRI/SSRI intake, and healthy controls (Fig. 4). Except for sequential activation of tibialis anterior then rectus femoris in most groups (low back pain, SNRI/SSRI intake, Parkinson's disease, spasticity, healthy controls, Fig. 4), no clear pattern of activation was observed. The mean duration of muscle activity following first stimulation (run 1) of each nerve was also similar except for functional and myoclonus groups who had statistically longer durations than SPS (Tables S2 and S6). There was no statistical difference between any of the groups in the presence of after‐bursts at any recording site (Tables S3 and S7). We observed no difference in mean latency of responses (overall or at any recording site), when comparing SPS patients with or without benzodiazepines nor among those with or without immunotherapy (Table 2). Discussion In this study, there was no pattern of axial and lower limb motor responses (as measured by surface EMG) following stimulation of the median and tibial nerves that differentiated any neurological condition studied nor the healthy state. Following median nerve stimulation, the shortest mean latency was observed in those with PD, and following tibial nerve stimulation the shortest mean latency was observed in healthy controls. The participant with CIDP had a longer mean latency following median and tibial nerve stimulation than other groups; this is not an unexpected finding given peripheral nerve demyelination is expected to cause slower peripheral conduction time. While not included in comparisons to SPS, we included this participant as an internal control, noting our technique for exteroceptive testing has been modified from prior published series. Our findings differ from previously published series evaluating exteroceptive reflexes which demonstrated a distinct pattern of muscle activation (termed spasmodic reflex myoclonus) in those with SPS. 5 , 10 , 12 In comparing our findings with prior publications several points require consideration. Firstly, some of our comparator groups have conditions causing central disinhibition at the spinal level and therefore could enhance exteroceptive reflexes. Those with lower limb spasticity have pathology affecting corticospinal neurons, and thus disinhibition of lower motor neurons. Parkinson's disease is associated with increased sensitivity of long latency stretch reflexes 13 and exteroceptive reflex latencies were shown to be similar to SPS patients in a prior study. 10 Serotonin plays a role in neuromodulation of spinal motor neurons; increased activation occurs with increased serotonin levels, and as such those with SSRI/SNRI administration could also have enhanced exteroceptive reflexes. 14 This would not explain however our observation that our healthy control and functional neurological disorder groups were not differentiated from those with SPS, and furthermore in some of these patients we observed latencies similar to those described in SPS in previous publications (70 ms or less) (Figs. 1, 2, 3). Secondly, there are differences in our neurophysiologic method of evaluating exteroceptive reflexes when compared to prior publications. Previous studies used intramuscular needle electrodes (needle EMG), whereas we have measured responses using surface electrodes. There are important differences between the techniques which must be considered, but which alone do not account for the discrepancy between our findings and prior literature. Needle EMG provides accurate recording of specific motor units in very close proximity to the needle electrode. 15 The approach is inherently invasive (with attendant, albeit low, risks posing ethical implications in a research context) and more technically demanding potentially limiting the use of the technique diagnostically in routine clinical practice. Surface EMG records a summation of active motor units underlying the surface electrodes. 15 The paraspinal muscles consistent of several muscles, most notably multifidus, longissimus and erector spinae muscle groups, which themselves are made up of many individual muscles spanning between vertebrae 16 , 17 working as a functional group. The original publications describing the technique do not specify which specific paraspinal muscle(s) were targeted for analysis 5 and later publication specify recording the erector spinae in some participants. 10 We chose sEMG in this study as it offers the potential advantage of detecting a summative representation of active motor units within a broader portion of the paraspinal muscle group, is more accessible and practical in routine clinical practice and poses lower risk to participants in the context of research. 15 Further supporting this approach, original publications describing the neurophysiologic findings in spontaneous spasms and myoclonic attacks affecting SPS patients used either sEMG or intramuscular EMG of the paraspinal and limb muscles 10 and sEMG has an established role in evaluation of paraspinal muscle activity functionally. 16 There is limited literature which directly compares the techniques in detecting activation of individual muscles (including the paraspinal muscles). One study reports high correlation between simultaneous sEMG and intramuscular EMG 17 of paraspinal muscles, and it has been demonstrated that multifidus is less represented than more superficial paraspinal muscles by sEMG. 18 To our knowledge there are no studies comparing the two EMG techniques with respect to the precise onset of recordable activity in paraspinal or other muscles with voluntary or reflex activity. Regardless however our use of sEMG cannot be solely responsible for differences in our findings compared to prior literature. We observed the latency of paraspinal responses were in the ranges described previously in those with SPS (70 ms or less (Figs. 1, 2, 3)) in some healthy controls; as the paraspinal group represents the most proximal muscle group this observation cannot be due to cross‐talk from muscles expected to be activated earlier. Thirdly, our method of evaluating exteroceptive reflex activity differed from that of prior publications in the way EMG activity was analyzed. Prior publications analyzed rectified data at each recording site whereas we analyzed the raw EMG trace. Rectifying sEMG data refers to removing all negative values of EMG signal before further processing; negative values can be removed from the data set (half‐wave rectification) or the negative are inverted to positive to contribute to the magnitude of the response (full‐wave rectification). 19 This process does not alter the latency of signal onset/offset as analyzed in this study, however is required if amplitudes of EMG signals are to be quantified or averaged 19 . Amplitudes were not analyzed in our study; the amplitude of sEMG is affected by skin and tissue impedance, and the distance between the surface and underlying muscles, 16 which inherently vary between individuals, confounding any comparison of this parameter between participants. Averaged EMG data obtained (ie from several successive stimulations) was analyzed in prior publications, whereas we analyzed findings following each stimulation within an individual, and for intergroup comparisons we compared results following the first stimulation and the earliest recorded latency at each recording site for each individual. Doing so enabled further neurophysiological interrogation; we assessed correlation between latencies following sequential stimulations to assess whether these reflex circuits are reproducible within an individual. Comparing between groups using the minimum latency observed at each recording site per participant allowed for comparison of our findings with prior publications. It is understood that the clinical manifestations of SPS are caused by immune‐mediated inhibition of (inhibitory) central nervous system GABAergic pathways, and thus it is not surprising that benzodiazepines (which potentiate the binding of endogenous GABA, and thereby act as GABA agonists) 20 reduce stiffness and spasms in SPS. At sufficient doses benzodiazepines have been observed to abolish exteroceptive reflex responses, 5 , 10 however the effect on the latency or duration of responses, or the presence of after‐bursts is not known. In this study it was not possible to completely avoid benzodiazepine exposure in participants due to the use of retrospective data and some participants’ intolerance of periods without treatment; mean diazepam (or equivalent) dose among those with SPS in our study was 19 mg/day (Table 1). Additionally, four out of the 12 participants with SPS were receiving immunotherapy (Table 1) at the time of testing. While most patients included in prior studies were treated with benzodiazepines and half with plasma exchange,5,10 it is not detailed whether immunotherapy or benzodiazepines were used at the time of exteroceptive testing; comparison between studies is not therefore possible in this respect. Peripheral nerve stimulation must be sufficiently high to produce exteroceptive reflex responses. 5 , 10 , 12 Due to individual differences in skin impedance and tolerance it is not possible to ensure precisely uniform intensity of peripheral nerve stimulation between participants, however it is unlikely that subthreshold stimulation is responsible for the differences between our findings and those of prior authors. Our study protocol allowed for stimulation at a level comparable to those of prior publications. All participants were stimulated at their maximum tolerated, and 31 participants had motor thresholds documented confirming stimulation at least 2.3 times the amplitude of stimulation needed to produce a visible motor twitch. Previous authors have described habituation of responses. 5 , 10 We took precaution to avoid this by randomly stimulating without warning with at least 30 seconds between stimuli. Furthermore, we compared results following the first stimulation between patient groups; each participant's first set of motor responses was used in an analysis (Table 4). Any modification of sequential responses due to habituation or conditioning did therefore not influence our findings. In 12 out of our 59 participants the earliest site of detectable muscle activation following median nerve stimulation was the thoracic paraspinal muscles, and in 7 the cervical paraspinal muscles. Median nerve somatosensory fibers arise from the 5th–8th cervical (and in some people 1st thoracic) spinal level, so it is of interest that the motor activation measured in a significant proportion of our cohort commences in the thoracic spine, caudal to the median nerve's spinal segment. Thoracic onset of motor activation was also observed in a prior study in two out of eight patients with SPS (46–68 ms) and in three of eight onset was in the cervical paraspinal muscles (52–81 ms). 10 Based on this finding and other elegant neurophysiological studies these authors hypothesized that excitatory volleys are transmitted within the cord using intraspinal ascending and descending pathways such as the propriospinal system; authors postulated that the earliest onset of motor activation (cervical versus thoracic) depends on the synaptic gain within each segment, and may represent the site of primary excitatory pathology in those with SPS. 10 Our findings however suggest activation of this polysynaptic spinal reflex is not unique to SPS, rather is a physiologic reflex that is demonstrable in other neurological conditions and healthy controls. Following tibial nerve stimulation, the pattern of motor activation we observed differed from that following median nerve stimulation in most participants (Figs. 3 and 4). Early activation of the tibialis anterior and rectus femoris was observed, in a pattern consistent with flexor reflex/withdrawal reflexes. 21 We acknowledge several limitations of this study. The small sample size of participant categories limits reliability of analyses comparing responses between these groups; as such these specific analyses should be interpreted as exploratory. The latency of motor onset/offset as measured by surface EMG was marked by a movement disorders specialist or neurophysiologist from the other study site and blinded to the participant's clinical information. While this necessarily introduces the potential for inter‐rater variability between site data, we did not find significant differences in findings between centers/raters, nor an effect of differing neurophysiological testing equipment between centers. Indeed, we methodically conducted this study using different neurophysiologic equipment across 2 clinical sites to broaden the clinical applicability of our findings. It is important to emphasize that the neurophysiological evaluation performed in this study is not equivalent to the electrophysiologic evaluation of spontaneous axial spasms or myoclonus; such studies have known clinical applicability, for example in the differentiation of spinal and brainstem myoclonus. 22 Stiff person syndrome is a rare condition and remains a diagnostic challenge, due to the non‐specific nature of many of the symptoms, lack of specificity of anti‐GAD and other rarer autoantibodies even when present, and lack of other disease‐specific biomarkers. We performed neurophysiologic exteroceptive reflex testing in a cohort of participants with SPS, other conditions with overlapping clinical features with SPS and healthy controls and could not detect a pattern of response that distinguished any participant group from others studied. Previous authors interpreted exteroceptive reflex testing in combination with detailed neurophysiologic evaluation of spontaneous jerks, startle responses, segmental reflexes (eg, H reflexes, blink reflexes), and long‐loop reflexes, providing additional evidence of disinhibition at the level of the brainstem and invaluable insights into the pathophysiology of SPS and related conditions. 10 Our findings support the use of multiple neurophysiologic modalities in combination for evaluation of central disinhibition, rather than isolated use of exteroceptive testing following peripheral nerve stimulation, in the diagnosis of suspected cases of SPS. Further research is needed to build upon the elegant neurophysiologic studies of prior authors 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 to fully elucidate the pathophysiology of central motor disinhibition as seen in those with SPS and develop techniques to more readily evaluate and accurately differentiate SPS and related conditions from disease mimics in clinical practice. Author Roles (1) Research Project: A. Conception and Design, B. Organization, C. Data Acquisition and Analysis; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft and Figure Preparation, B. Review and Critique. B.C.: 1A, 1B, 1C, 2C, 3A D.W.: 1B, 1C, 2A, 2B, 2C, 3A, 3B H.M.B.: 1A, 1B, 1C, 2C, 3B N.M.: 1C, 2C, 3B P.G.: 1C V.Z.: 1C J.Q.: 1B, 1C T.K.: 2A, 2B V.F.: 1B, 3B Disclosures Ethical Compliance Statement: This research was approved by the Royal Melbourne Hospital Research Ethics Committee. Written consent was obtained for all participants. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines. Funding Sources and Conflict of Interest: No specific funding was received for this work. The authors declare there are no conflicts of interest relevant to this work. Financial Disclosures for the Previous 12 Months: BC served on scientific advisory boards or as a consultant for Abbvie, Alexion, Allergan, Argenx, CSL Behring, UCB, received conference travel support and/or speaker honoraria from Abbvie, CSL Behring, UCB. JQ has received conference travel support from NSW Health. TK served on scientific advisory boards or as a consultant for MS International Federation and World Health Organization, Therapeutic Goods Administration, BMS, Roche, Janssen, Genzyme, Novartis, Merck and Biogen, received conference travel support and/or speaker honoraria from WebMD Global, Merck, Sandoz, Novartis, Biogen, Roche, Eisai, Genzyme, Teva and BioCSL and received research or educational event support from Biogen, Novartis, Genzyme, Roche, Celgene and Merck. VF has received royalties or licenses from Health Press, Taylor and Francis Group, Oxford University Press, grants from Michael J. Fox Foundation, conference travel support from the International Movement Disorder Society, and serves as its President (2023–2025). DW, HMB, VZ, PG and NM report no disclosures. Author disclosures are available in the Supporting Information. Supporting information Table S1. Mixed effects linear model with fixed effects for age and sex, and random effects for study site: Median nerve run 1 burst onset latency. Table S2. Mixed effects linear model. Fixed effects for age and sex, random effects for study site: Median nerve run 1 average duration time. Table S3. Mixed effects logistic model. Fixed effects for age and sex, random effects for study site: Median nerve run 1 after‐bursts. Table S4. Mixed effects linear model with fixed effects for age and sex, and random effects for study site: Median nerve average latency of shortest latency. Table S5. Mixed effects linear model. Fixed effects for age and sex, random effects for study site: Tibial nerve run 1 average onset time. Table S6. Mixed effects linear model. Fixed effects for age and sex, random effects for study site: Tibial nerve run 1 average duration time. Table S7. Mixed effects logistic model. Fixed effects for age and sex, random effects for study site: Tibial nerve run 1 after‐bursts. Table S8. Mixed effects linear model with fixed effects for age and sex, and random effects for study site: Tibial nerve average latency of shortest latency. Supporting Information: COI_disclosure.