Biomechanical Muscle Stiffness Measures of Extensor Digitorum Explain Potential Mechanism of McArdle Sign

Abstract

Background:

McArdle sign is a phenomenon of impaired gait and muscle weakness that occurs with neck flexion, immediately reversible with neck extension. A recent report measured the specificity of this sign for multiple sclerosis by measuring differences in peak torque of the extensor digitorum between neck extension and flexion.

Methods:

This substudy included 73 participants (29 multiple sclerosis, 20 non-multiple sclerosis myelopathies, 5 peripheral nerve disorders, and 19 healthy controls). The effect of neck position was assessed on muscle stiffness and neuromechanical error of the extensor digitorum.

Findings:

Patients with multiple sclerosis had greater neuromechanical error (sum of squared error of prediction) compared to controls (P=0.023) and non-multiple sclerosis myelopathies (P=0.003). Neuromechanical error also provided improved sensitivity/specificity of McArdle sign. Peak torque, muscle stiffness, and neuromechanical error could distinguish multiple sclerosis from other myelopathies with 80% specificity and 97% sensitivity (AUC=0.95).

Interpretation:

A decrease in muscle stiffness and neuromechanical error in neck flexion compared to extension are additional indicators for a diagnosis of multiple sclerosis. Analysis of muscle stiffness may provide insights into the pathophysiology of this specific clinical sign for multiple sclerosis. Furthermore, muscle stiffness may provide an additional accurate, simple assessment to evaluate multiple sclerosis therapeutic interventions and disease progression.

Introduction

In 1987, O’Neill et al. described a clinical sign in a single patient with multiple sclerosis (MS), who developed difficulty walking during neck flexion and walked with their neck hyperextended (McArdle, 1988; O’Neill et al., 1987). M. J. McArdle taught but had not published this observation, but O’Neill et al. proposed McArdle sign as an eponym (O’Neill et al., 1987). Until recently, this sign has not been further studied to determine its specificity or clinical significance for diagnosis of MS (Savoldi et al., 2019).

Our group recently conducted a blinded study to assess the frequency and specificity of this sign for MS. The strength of the finger extensors (extensor digitorum), a sensitive and convenient muscle group in which the McArdle sign can be detected, was measured with a torque measurement device to detect changes with neck flexion compared to extension (Savoldi et al., 2019). McArdle sign was additionally detectable in other muscle groups sensitive to upper motor neuron lesions. The blinded cross-sectional study demonstrated that McArdle sign, defined as 10% reduction in strength with flexion, was 100% specific and 63% sensitive comparing patients with a variety of other non-MS myelopathies using a receiver operator curve (ROC) analysis (Savoldi et al., 2019). O’Neill et al. who reported this finding in a single patient did not recognize the high frequency of McArdle sign in patients with MS (O’Neill et al., 1987). Despite its high prevalence in patients with MS, this sign is largely asymptomatic and is clinically manifest only in patients with moderate to severe motor impairments. Nonetheless, detection of the sign by clinical examination can be a helpful way of supporting a diagnosis of demyelination in patients with a myelopathy of uncertain cause. We found a strong correlation between clinically and instrument-based measurements (Savoldi et al., 2019).

Our prior core study concentrated on peak strength. In this report, we evaluated the biomechanics of McArdle sign by assessment of neuromotor stiffness of the metacarpophalangeal joints generated by the extensor digitorum as a result of head position (i.e. neck flexion, neck extension) to better understand the pathogenic mechanisms of McArdle sign and determine whether additional testing might improve sensitivity of detection. Biomechanical stiffness is a measure of resistance of elastic materials to deformation. Measured as torque (τ) over angular displacement (θ), muscle stiffness is an intrinsic property produced by both passive and active musculature. Passive stiffness is created with actin-myosin cross-bridges and actively with gamma- and alpha-motor neuron co-activation impulses that generates muscle tone and contraction, respectively (Johansson, 1991; Li et al., 2015; Sjölander et al., 2002). Furthermore, joints are stabilized by the parallel synergy of the joint capsule, ligaments, muscles, sensory receptors, and the spinal/cortical neural connections (Riemann and Lephart, 2002; Solomonow and Krogsgaard, 2001) and is the mechanism by which the joint functions through physiologic motion and protection from external perturbation. Of these synergistic mechanisms, muscle stiffness is the most adaptable, quantifiable and useful for this analysis. Muscle stiffness is produced actively by rapid sensorimotor neuromuscular communication, which could be deficient or poorly integrated in patients with MS (Li et al., 2015; Loram et al., 2009). Previous studies that evaluate muscle stiffness have either utilized subjective interpretation or manual muscle strength tests and has not been widely performed due to issues with reproducibility or reliability (Caruso and Leisman, 2000). Furthermore, alterations in joint proprioception and neuromuscular control have inconsistently correlated with measures of joint stability and function (Needle et al., 2013). However, manual muscle strength tests are regularly performed and interpreted regularly as part of the neurological examination of myotome function (Krisa et al., 2013; Magee, 2005). In addition, recent work of the lower extremity has assessed the proprioceptive sensorimotor integration and suggests that dynamic evaluations (e.g. high velocity threshold to detect passive motion, force steadiness) may be more sensitive to sensorimotor evaluation than static measures (e.g. isometrics, joint position sense) (Li et al., 2015; Nagai et al., 2018). Since sensorimotor control of muscle stiffness as it relates to clinical examination and human movement control is inadequately understood, we analyzed McArdle sign by determining how it could be understood in terms of effects on muscle stiffness.

In this study, we utilized the custom-built torque device to evaluate finger extensor muscle stiffness and compared its sensitivity and diagnostic properties as an adjunct to peak torque, which we had evaluated in the previous study (Savoldi et al., 2019). As McArdle sign has not been previously studied by others, we evaluated multiple groups including healthy controls (CTRL) or patients with peripheral nerve pathology that affected the muscle groups tested to address issues of specificity to central nervous system pathology. Further, we enrolled patients with clinical myelopathy for whom MS was in the differential diagnosis, in order to show the diagnostic value of the McArdle sign in clinically relevant patients. We hypothesized that: 1) MS patients would exhibit less biomechanical muscle stiffness with the neck in flexion vs extension when compared to CTRL and other non-MS myelopathies (OM), 2) patients with MS would demonstrate greater variation (error) in generated muscle stiffness than CTRL, and 3) muscle stiffness will enhance the ability to detect McArdle sign compared to peak torque. Further, we were interested in the principal components that allow an examiner to determine the severity of McArdle sign.

Methods

Human Subjects

The research protocol was carried out in accordance with the Declaration of Helsinki and was approved as a minimal risk study by Mayo Clinic’s internal review board (IRB #15-009112). All patients signed an informed consent and privacy of human subjects was maintained. From the entire cohort of 106 patients evaluated in the parent study, 79 participants participated in this substudy, of whom 73 patients had interpretable data [age 47.9 years (12.5)]. Of the 73 patients, 19 were CTRL; of those with finger extensor weakness, 29 were patients diagnosed with MS, 20 with OM, and 5 with peripheral nerve lesions (PNL) associated with finger extensor weakness (Table 1). The principal inclusion criterion was detectable weakness and at least antigravity strength in finger extensors when evaluated with the neck in the neutral position. Patients were not selected based on presence of McArdle sign, but some degree of detectable finger extensor weakness, even if mild, was an additional requirement. The OM group was comprised of myelopathy patients of different etiologies (Table 1). The PNL group was comprised of patients with peripheral nerve lesions or radiculopathies (C7 radiculopathy or other peripheral nerve lesions causing finger extensor weakness; Table 1). This group was selected to control for other non-CNS causes of finger extensor weakness. Enrollment was monitored by diagnosis to ensure that the desired number of patients in each diagnostic category was included. The study population was a convenience sample, although we attempted to enroll consecutive patients who met study criteria. All procedures performed were the same for all groups.

Table 1.

Clinical Demographics by Group.

		MS (n=29)	OM (n=20)	PNL (n=5)	CTRL (n=19)	P-value
Age	• Mean Years (SD)	48 (11)	53 (14)	48 (10)	42 (12)	0.061a
Side Tested	• Left	14	12	4	8	0.422b
	• Right	15	8	1	11
Sex	• Male	5	8	0	6	0.053b
	• Female	24	12	5	13
EDSS Score [Range 0-10]	• Median Score (25th, 75th)	4 (2, 7)	3 (2, 6)	NA	NA	0.246c
Disease Duration	• Median Years (25th, 75th)	15 (4, 22)	2 (1, 6)	0 (0, 4)	NA	<0.001c
Lhermitte’s Sign	• Present (%)	8 (28%)	1 (5%)	0 (0%)	NA	0.091b
Cervical Spine MRI	• Quantity	29	17	4	NA	0.054b
	• Time from Test, Months (25th, 75th)	2 (0, 13)	1 (0, 6)	1 (0, 13)		0.487c
Location of T2-weighted Cervical Cord Lesion to Weakness*	• None	0	5	4	NA	<0.001b
	• Ipsilateral	3	1	0
	• Contralateral	2	0	0
	• Bilateral	24	11	0
Atrophy in Cervical Spine*	• None	27	11	4	NA	0.250b
	• Ipsilateral	1	1	0
	• Contralateral	0	1	0
	• Bilateral	1	4	0
MS Phenotype	• Clinically Isolated Syndrome (CIS)	1	NA	NA	NA	NA
	• Relapsing-Remitting MS (RRMS)	12
	• Primary Progressive MS (PPMS)	7
	• Secondary Progressive MS (SPMS)	9
Types of Other Myelopathies	• Compressive myelopathy	NA	5	NA	NA	NA
	• Amyotrophic lateral sclerosis		3
	• Hereditary myelopathy		1
	• Intrinsic spinal cord tumor		1
	• Neuromyelitis optica		3
	• Neurosarcoidosis		2
	• Other myelopathy		5

EDSS = Expanded Disability Status Scale [0=normal to 10=death]; MS = Multiple Sclerosis; OM = Other Myelopathies; CTRL = healthy controls; PNL = Peripheral Nerve Lesion; NA = Not Applicable; Statistical tests:

^aAnalysis of Variance

^bFisher Exact

^cKruskal-Wallis

^*Note: T2-Weighted Lesions and Atrophy in Cervical Spine are lower n-values than total subject numbers for OM and PNL as cervical spine MRIs were only available for a subset of patients.

McArdle Sign Evaluation

For clinical evaluation of McArdle sign, the practitioner attempted to overcome maximal resistance of finger extension in 2 to 3 successive trials of neck extension and flexion. A custom-built finger extension device was designed to mimic the clinical evaluation of McArdle sign. The device secured the forearm and hand with Velcro straps and a torque cell (RTS-500, Transducer Techniques; Temecula, CA, USA) was placed in-line with the axis of the metacarpophalangeal joints with an attached, adjustable comfort bar (Fig. 1). The comfort bar was positioned above and immediately adjacent to the fingers so that finger extension would immediately register torque production. The fingers were positioned near full extension for each trial. A technician, blinded to the clinical diagnosis, performed the McArdle sign evaluation by two methods: 1) clinically (by hand), and 2) with the device in which downward pressure was placed on the lever that depressed the bar positioned above the fingers. The lever that allowed the technician to apply a downward pressure for the test was affixed at the base of the torque cell outside the sensors of the torque cell, thus allowing all torque measured to be attributed to the patient (Fig. 1). The score of the McArdle sign was recorded (graded with discrete values 0 – 3; 0 = no apparent difference between neck extension and flexion; 3 = marked difference between neck extension and flexion).

Figure 1. McArdle Biomechanical Test Setup.

The subject’s arm was secured in the custom-built device for measurement of finger extensor torque production. The hand was positioned so that the axes of rotation of the metacarpophalangeal joints were in-line with the rotation axis of the torque cell. The torque cell (blue arrow) measured torque as the subject pushed against a padded comfort bar. Two gyroscopes in the headband measured neck flexion/extension (yellow arrow). The lever was utilized by the examiner to apply a downward force that the subject was asked to resist (red arrow).

Patients were fitted with a 2-axis gyroscopic headband to measure accurate flexion/extension position of the neck (Fig. 1). The weakest arm was evaluated for all participants (not determined for CTRLs); in those with equal weakness of the two arms or in controls, one or other arm was selected randomly. All gyroscopic and torque cell data were collected at 500 Hz via proprietary LabVIEW programmed software (National Instruments, Austin, TX, USA) and sampled simultaneously through the same data acquisition device (NI USB-6009, National Instruments).

Patients performed five trials of isoinertial contractions (defined as a constant resistance force applied to the fingers by the examiner) with the neck either in full flexion or extension (Savoldi et al., 2019). Each subject was tested with the neck extended and then alternated between flexion and extension with 40 seconds of rest between each trial. Based on empirical testing of recovery after different intervals, an inter-trial interval of 40 seconds allowed satisfactory recovery of muscle force and prevented fatigue. This methodology was designed to recapitulate the neurological examination for muscle strength performed clinically to detect McArdle sign.

Data Analysis

Torque and displacement data were post-processed in customized LabVIEW software to extract a linear fit of the stiffness slope (torque over displacement) during the epoch of active finger extension torque generation (Fig. 2). The muscle contraction observed was not isometric, isotonic, or isokinetic. Rather, the movement was termed isoinertial (Savoldi et al., 2019) in which there was a change in which the fingers were displaced by force sufficient to ‘break’ the maximum strength the patient could exert in finger extension. (Kendall et al., 1993) Thus, the average torque and standard deviation could not be accurately extracted to calculate a coefficient of variation as these outcomes were always variable. Rather, the additional variables extracted during the process of linear fit were 1) R-squared and 2) sum of squared error of prediction (SSE). These extracted values of the variance of the best fit stiffness slope during active torque development were analyzed as additional indicators of neuromechanical error. For example, a linear trend with significant error from the actual slope could provide evidence of either a sensorimotor or neuromotor deficit with an inability to maintain consistent muscle stiffness (Fig. 2). As signal dependent noise increases with increased neural output, SSE was normalized to peak torque (NormSSE).

Figure 2. Sample raw data plots of torque vs displacement with fitted lines of stiffness.

Representative curves of active stiffness (torque / displacement) of 3 individual patients. The examiner depressed a lever that provided resistance to the patient’s fingers with extension, similar to a clinical muscle strength test. A) A CTRL patient demonstrated a similar stiffness slope in both neck extension and flexion and typically had a linear trend with minimal error (i.e. high R-square value, low SSE). B) Patient with MS demonstrated differences in both stiffness and torque magnitude with the neck in extension and flexion. Further, additional error from the linear trend line is observed by increased oscillations. C) OM demonstrated similar stiffness slopes compared to CTRL, but more oscillations from the fitted line of stiffness similar to MS patients. (Note: Y- and X-axes vary based on distributions of the data in the various groups; CTRL = healthy control; MS = multiple sclerosis; OM = other myelopathies;)

Data were exported into JMP 14 Pro (SAS Institute Inc., Cary, NC) for statistical analysis. For nonparametric data, a Log transform was applied as appropriate. Variables of interest (Peak torque, Stiffness, R-square, SSE) were calculated as the difference (Δ) between neck extension and flexion (Ext – Flex). The first two trials were ignored to mitigate lack of familiarity with the task in the initial trials; thus, the mean value of the final three of five trials performed in each state of neck flexion/extension was analyzed. For outcomes with parametric distribution, one-way ANOVAs were utilized to compare means and further analyzed with Tukey’s post hoc comparison with the independent variable of group. When data was non-parametric, Kruskal-Wallis analysis was utilized with Steel-Dwass post hoc comparison. Further analysis was performed with principal component analysis (PCA) for dimension reduction of the data and to maximize the contributions of the important outcome variables of neuromechanical error. This was of interest to determine the principal components that allow an examiner to identify and grade McArdle sign. A receiver operator characteristic (ROC) curve logistic regression was performed for determination of specificity and sensitivity of muscle stiffness measures for MS diagnosis compared to CTRL, OM, and PNL groups. A Kappa Agreement test was also utilized for assessment of ‘clinical’ and ‘device’ categorization of McArdle sign. Statistical significance was set a priori at an alpha of <0.05.

Results

Peak Torque and Muscle Stiffness

The Δ Peak Torque of neck extension and flexion was greater in MS patients (F_3,72=12.570, P<0.001) than CTRL (P<0.001), OM (P<0.001), and PNL (P=0.004; Fig. 3A and Table 2). Δ Stiffness of neck extension and flexion (Fig. 3B; χ²=6.334, P=0.096) did not differ between groups. [Omitting two outliers of Δ stiffness (positive value of CTRL and negative value for MS, Fig. 3B) resulted in a difference between groups (χ²=9.621, P=0.022), but still no differences from Steel-Dwass posthoc.]

Figure 3. Difference of Peak Torque, Stiffness, R-square, and SSE between neck extension and flexion.

The final three trials of each neck condition were averaged (MEDext – MEDflex or AVGext – AVGflex). A value of zero designates no difference of extensor digitorum values between neck extension and flexion. A) MS patients have lower peak torque values in neck flexion vs extension (value above zero) compared to CTRL, other myelopathies, and peripheral nerve lesions. B) MS and other patients have increased stiffness compared to CTRL of the extensor digitorum (values below zero). C) MS patients have higher variability of R-square values with neck flexion vs extension compared to CTRL, OM, and PNL (values below zero). D) MS patients have increased normalized SSE (values below zero) compared to CTRL, OM, and PNL. (Significance of P<0.05 denoted by *, P<0.01 denoted by **, and P<0.001 by ***; Black diamonds represent 95% confidence interval; CTRL = Healthy Control; MS = Multiple Sclerosis; OM = Other Myelopathies; PNL = Peripheral Nerve Lesion.)

Table 2:

Summary of outcome variables between groups with each neck condition (extension, flexion, and extension-flexion).

		CTRL	MS	OM	PNL	Significance
Torque (N-m)	Ext	1.39	1.12	1.03	0.84	0.013a
	Flex	1.39	0.96	1.00	0.85	0.001a
	*Ext-Flex	0.00	0.16	0.03	−0.01	<0.001a
Median Muscle Stiffness	Ext	1.2	1.2	1.4	1.0	0.857b
	Flex	1.5	0.8	1.1	1.1	0.510b
	Ext-Flex	−0.1	0.2	0.1	−0.1	0.096b
Log SSE	*Ext	9.6	10.9	9.1	9.5	<0.001a
	Flex	9.8	10.3	9.4	9.6	0.217a
	* Ext-Flex	−0.2	0.6	−0.3	−0.1	0.002a
Norm. Log SSE	Ext	7.3	12.5	10.1	13.5	0.034a
	Flex	7.3	15.6	10.7	13.7	0.005a
	Ext-Flex	0.0	−3.1	−0.6	−0.2	0.017a
Median R-square	Ext	0.94	0.85	0.97	0.97	0.004b
	Flex	0.93	0.82	0.94	0.94	0.018b
	Ext-Flex	0.01	0.01	0.00	0.00	0.994b

^*= variables determined as primary contributors for logistic regression; SSE: sum of squared error of prediction; Statistical tests:

^aAnalysis of Variance

^bKruskal-Wallis; (CTRL = Healthy Control; MS = Multiple Sclerosis; OM = Other Myelopathies; PNL = Peripheral Nerve Lesion)

Stiffness Variation

Δ R-square values of neck extension and flexion did not differ between groups (χ²=0.082, P=0.994; Fig. 3C). However, the R-square value was lower for MS patients than all other groups in extension (P=0.004) and flexion (P=0.018; Table 2). Δ SSE was higher for MS patients (F_3,72=5.343, P=0.002) compared to CTRL (P=0.023) and OM (P=0.003). Δ NormSSE (χ²=10.428, P=0.015) was greater in neck flexion than extension for MS vs CTRL (P=0.023; Fig. 3D, Table 2).

McArdle Inter-rater Agreement

Kappa Agreement (assessed between −1 and 1) between the ‘device’ and ‘clinical’ categorization of McArdle sign was moderate (P<0.001), with a Kappa value of 0.48 (0.33, 0.63).

Principal Component of McArdle Sign

The variables of Δ Peak Torque, Δ Stiffness, Δ NormSSE, and Δ R-square were studied for covariance PCA to determine their contribution to McArdle sign. One principal component (PC1) described 97.2% of the signal variance. PC1 is described by the following formula:

$$-0.02\ast \Delta PeakTorque-0.02\ast \Delta Stiffness+1.00\ast \Delta NormSSE+0.00\Delta RSquare+1.4$$

PC1 by ‘device’ McArdle categorization demonstrated differences between McArdle sign grades (χ²=16.095, P=0.001; Fig. 4A) with Grade 3 lower than Grade 0 (P=0.001), Grade 1 (P=0.015), and Grade 2 (P=0.038). PC1 also demonstrated differences for ‘clinical’ McArdle categorization (χ²=14.601, P=0.002; Fig. 4B) with Grade 3 higher than Grade 0 (P=0.004).

Figure 4. McArdle Sign Categorization by Principal Component Analysis from Differences Between Neck Extension and Flexion.

PC1 described 97.2% of the covariance of the variables: Δ Peak Torque, Δ Stiffness, and Δ NormSSE. This demonstrates the influential variables that describe McArdle sign. A) PC1 demonstrates a decrease in value with those characterized with a graded device McArdle sign. B) PC1 demonstrates a decrease in value with those characterized with a graded clinical McArdle sign. Device and Clinical categorization have moderate (0.48) agreement. (Significance of P<0.05 denoted by * and P<0.001 by ***; Black diamonds represent 95% confidence interval.)

Specificity / Sensitivity of Stiffness and Variation

The variables described in Table 2 that demonstrated significance between groups were utilized in a response screening analysis utilizing a False Discovery Rate LogWorth > 2.0 to determine the optimal variables for use in a logistic regression to determine predictive power of MS from CTRL, OM, and PNL groups. The following predictive variables were determined: Δ Peak Torque, SSE (in neck extension), and Δ SSE (Table 2, denoted by *). The AUC comparing MS vs OM (χ²=36.940, P<0.001) was 0.95 with an overall 80% specificity and 97% sensitivity. The AUC comparing MS vs PNL (χ²=19.708, P<0.001) was 0.97 with an overall 100% specificity and 80% sensitivity. The AUC comparing MS vs CTRL (χ²=30.622, P<0.001) was 0.92 with an overall 79% specificity and 93% sensitivity (Fig. 5).

Figure 5. ROC curves of the predictive power of the difference of finger extension stiffness from neck extension and flexion for MS diagnosis A) vs OM B) vs CTRL and C) vs PNL.

Other neurological disorders are sometimes difficult to differentiate from MS. Muscle stiffness, as measured with neck extension and flexion reveals ‘excellent’ predictive power for MS vs OM, MS vs CTRL, and MS vs PNL. (CTRL = Healthy Control; MS = Multiple Sclerosis; OM = Other Myelopathies; PNL = Peripheral Nerve Lesion; Sp = Specificity; Sn = Sensitivity)

Discussion

Study Overview

This substudy demonstrated that many of the outcome variables assessed differed between MS patients vs CTRL, OM, and PNL (Table 2). Patients with MS had increased error (higher SSE and lower R-square) in generated muscle stiffness of the extensor digitorum compared to CTRL and OM. Median muscle stiffness was lower with MS patients vs CTRL, OM, and PNL when comparing values recorded in neck flexion. However, neuromechanical error of muscle stiffness, rather than muscle stiffness itself, enhanced detection of McArdle sign. Peak torque, stiffness, and SSE were all lower in patients with MS with neck flexion compared to neck extension (Figs. 2-3, Table 2). When comparing values recorded with the neck flexed versus extended, muscle stiffness and stiffness variability measurements distinguished between MS, CTRL, OM, and PNL groups (Table 2). Muscle stiffness and its variability (neuromechanical error), therefore, demonstrate an additional diagnostic tool for MS patients vs OM and PNL (Fig. 5A, C) and CTRL (Fig. 5B). Principal component analysis demonstrated that both the ‘clinical’ and the ‘device’ assessment could similarly distinguish the categorical grading of the McArdle sign (Fig. 3) as both had moderate agreement (Viera and Garrett, 2005). The variables utilized for the diagnostic accuracy were Δ Peak Torque, SSE (in neck extension), and Δ SSE (Table 2, denoted by *).

Potential Mechanisms of McArdle Sign

During clinical evaluation of strength and of difference in strength between neck extension and flexion (McArdle sign), clinicians sense both the torque production at the joint generated by a muscle as well as the ability of that muscle to maintain a consistent resistance with additional applied force by the examiner (stiffness). Muscle stiffness is determined by the interaction of all muscle components (Ettema and Huijing, 1994) and include the cross-bridges formed by different muscle fiber types, both passive and dynamic (Petit et al., 1990). The addition of neuromechanical noise from the increased error during the clinical test may be what is perceived as muscle ‘weakness’ by the examiner (Krisa et al., 2013; Magee, 2005). Thus, the designation of muscle ‘weakness’ may not only directly relate to ‘strength’ output of the muscle (peak value), but rather, as demonstrated by the principal component data (PC 1), is a combination of decreased peak torque, decreased muscle stiffness, and increased neuromechanical error with the difference between neck extension and flexion (Fig. 3).

A review of biomechanics that influence the spinal cord may inform the understanding of changes in muscle stiffness and neuromechanical error induced by neck flexion and pathophysiology of McArdle sign. When the neck is flexed, the spinal cord is elongated (Kitahara et al., 1995; Lew et al., 1994; Rade et al., 2014). Neck flexion elicits greater spinal cord displacement than traction (Lew et al., 1994; Whedon and Glassey, 2009). The spinal cord is suspended within the spinal canal via denticulate ligaments and the filum terminale (both pia mater) (De Vloo et al., 2016), which stabilize the spinal cord and allow it to shift with spinal movement (e.g. neck and hip flexion / extension) (Whedon and Glassey, 2009). In fact, the filum terminale endures considerable strain with an applied load and reduces stretch load to the spinal cord neurons (De Vloo et al., 2016). These pia mater attachments fix the spinal cord to the dura mater, which is anchored to the spinal canal at the foramen magnum, C2 or C3, and the sacrum (Sinnatamby, 2011). These dural attachments and resultant stretch of the spinal cord are the substrate for the clinical signs (Kernig’s and Brudzinski’s) sometimes observed with meningitis or subarachnoid hemorrhage. These signs involve either a straight leg raise that elicits pain (Kernig’s) or flexion of the neck that elicits a reflexive flexion of the hip, potentially caused from a stretched spinal cord (Brudzinski’s) (De Vloo et al., 2016; Thomas et al., 2002). Interestingly, symptoms of a neurological disorder that is caused by a taut filum terminale – tethered cord syndrome – include loss of muscle strength and/or tone and abnormal gait (Selcuki et al., 2015; Yamada, 2009). These symptoms of a stretch-related clinical syndrome may be pertinent to McArdle sign (Savoldi et al., 2019). In the face of pre-existing conduction inefficiency associated with demyelination in patients with MS, elongation of the spinal cord (and subsequently the neurons) by neck flexion exacerbates conduction inefficiency and may cause conduction block.

Conduction block caused by an elongation of the neurons from neck flexion may also describe the increased error observed in the MS patient group (Figs. 3D, 6). Although there are important contributions of the spinal cord, brain stem, and motor cortex in the control of motion, the cerebellum is important for adaptation of movement patterns (Malone et al., 2012). The cerebellum receives afferents from the spinal cord and other brain regions and subsequently coordinates posture, balance, and coordination by directly targeting motor neurons (Grillner and El Manira, 2020; Schreck et al., 2018). Further, the cerebellum continuously computes state predictions based on efferent copies of the outgoing motor command to guide the movement and integrates multiple sensory estimates to form a single, more precise estimate of the body state (Bhanpuri et al., 2013). In effect, the cerebellum is an ‘error detector’ from perturbation and adapts motor control in the short- and long-term (Grillner and El Manira, 2020). With demyelination present in the central nervous system, MS subjects will have difficulty completing this cerebellar loop of error correction as efficiently (both ascending and descending pathways) as those without the demyelination (Fig. 6). This potential conduction block from demyelination would decrease error correction at the cerebellum and increase the error present in muscular contraction, as observed in this study (Fig. 7).

Figure 6. Comparison of stiffness variability between an MS and CTRL with similar torque and stiffness values.

The raw data between a MS and CTRL subject had similar peak torque and stiffness values (fitted stiffness slopes shown in black; ext = solid, flex = dashed). The noticeable difference between the two participants is the error from the slope.

Figure 7. Potential Biomechanical Mechanisms of McArdle Sign.

As described by the Results, the biomechanical effects may be caused by potential conduction block from stretch of the spinal cord with neck flexion, causing disturbances of signal transmission to effectors (i.e. muscle, cerebellum, basal ganglia, cortices, etc.) These disturbances to effectors would lead to the measures of decreased peak torque, stiffness, and increased neuromechanical error as described in this study (denoted above in parentheses). Norm SSE = Normalized Sum of Squared Error of Prediction.

The data represented from these experiments highlight a possible mechanism by which McArdle sign is exhibited biomechanically and interpreted clinically with neuromotor activation (Fig. 7). The extensor digitorum may represent an easily accessible and sensitive muscle group that is subject to rapid changes of muscle stiffness (Salmond et al., 2017; Savoldi et al., 2019). MS is a disease resulting in impaired central neural drive (Fimland et al., 2010). Voluntary muscle contraction is corrupted by signal-dependent noise – the greater the motor recruitment, the greater the variability in the resulting muscle force (or torque) (Salmond et al., 2017). As torque development is insufficient with neck flexion in the setting of demyelination (Fig. 4), larger motor units are recruited. Slow fiber motor units that are recruited first have higher stiffness than later fast-fatigue motor units (Petit et al., 1990). This insufficiency to recruit the typical small motor units from conduction block would result in the loss of muscle stiffness and peak torque in neck flexion with fewer cross-bridges formed.

Clinical Significance of Results

Misdiagnosis of MS is a serious problem that probably stems from overconfidence in the presence of radiologic dissemination in space and time even when pretest probability of MS is low based on clinical and laboratory data (Savoldi et al., 2019; Solomon and Weinshenker, 2013). Although objective testing of McArdle sign does not replace conventional methods for diagnosis (i.e. MRI), it could serve as a complementary approach to support that a lesion detected on MRI is a demyelinating lesion and also could be used to easily assess improvement of the disease severity via therapeutic interventions. Clinical signs are still an important aspect of diagnosis and misdiagnosis is a significant problem (Gaitán and Correale, 2019). McArdle’s sign may clinically refine the ability to distinguish myelopathy, a common syndrome of MS, from myelopathy due to other causes. Further, this objective diagnostic tool may facilitate diagnosis especially in atypical patients who do not fulfill typical diagnostic criteria (Gaitán and Correale, 2019; Savoldi et al., 2019).

Strengths and Limitations

This study utilized simple and effective biomechanical measurements with reliable data from a torque cell and head gyroscopic sensors. Additionally, the isoinertial paradigm utilized to assess finger extensor strength and stiffness simulated mirrored the clinical manual muscle test. Further analysis of the recorded measures utilized linear regression methods to determine best fit slope, R-square, and neuromechanical error. These extracted variables allowed for unique assessment of motor control variability that is clinically evaluated with the McArdle sign.

Our study was limited by potential errors in the isoinertial measure of stiffness with the subject/examiner interface on the finger bar. Although the same trained examiner always performed the displacement on the bar and the patients resisted the isoinertial maneuver in 1 to 2 seconds, variability in the rate at which the lever was displaced could contribute to error in the results. However, the measure of stiffness is not dependent on time, but rather torque and displacement. Further, the key differences were taken within subject for neck extension and flexion. Given the consistent results of torque, stiffness, and error (Fig. 3, Table 2), the concern of the examiner-subject variability is minimized. A hydraulic or magnetically driven movement of the bar attached to the torque cell would reduce human variability in the testing paradigm. Manual muscle strength tests are usually performed submaximally in the clinical setting (Conable and Rosner, 2011). In this study, in order to maintain consistency for data analysis, force was exerted by the trained practitioner until the subject could no longer overpower the applied force. Consequently, torque and displacement were measured across the full ROM. However, to minimize this error, the data during the active response of the subject was extracted and analyzed.

The current data did have fewer men than women enrolled, but this reflects the 2:1 ratio of women to men with MS in the general population. The size of the PNL group was smaller than the other groups. With this study, the main comparison was between MS and OM. We included PNL to ensure that McArdle sign was not present in non-CNS disorders that produce finger extension weakness, the primary inclusion criterion of the study. The presence of Lhermitte’s sign was provided in Table 1 and did not prove significant between groups. Lhermitte’s sign is usually transient in patients with MS, whereas McArdle sign seems to persist (based on our clinical experience).

This study was not designed to assess repeatability. Future studies should assess the repeatability of these measurements over time.

Conclusion

Evaluation of metrics related to muscle stiffness quantitation comparing values obtained with neck extended to flexed (McArdle sign) with a torque-cell device differentiates MS from other myelopathies with a 80% specificity and 97% sensitivity (AUC = 0.95). The sensitivity-specificity balance for McArdle sign utilizing muscle stiffness is superior to that previously reported based on analysis of peak torque alone (AUC of 0.84). Adding variables of neuromechanical error (noise) recorded from the same device increased the specificity and sensitivity for distinguishing MS from those with other myelopathies, peripheral nerve lesions, and healthy controls. The neuromechanical error is likely caused by a conduction block between the cerebellum and the motor neurons due to demyelination which is exacerbated by neck flexion. Adding measures of muscle stiffness to those quantitating peak strength to objective assessment of McArdle sign provides insight into the pathophysiology of MS and can potentially serve as an ancillary diagnostic tool that is worthy of further assessment to improve clinical diagnostic strategies.

Highlights.

1. Multiple Sclerosis has greater neuromechanical error than myelopathy with flexion.

2. Neuromechanical error shows high specificity/sensitivity for Multiple Sclerosis.

3. Neuromechanical error may describe spinal cord mechanism of McArdle sign.

4. McArdle sign has potential to translate clinical practice and reduce misdiagnosis.

Abbreviations:

AUC

Area Under the Curve

CTRL

Healthy controls

MS

Multiple Sclerosis

NormSSE

Normalized Sum of Squared Error of Prediction

OM

Other Myelopathies

PC

Principle Component

PCA

Principle Component Analysis

PNL

Peripheral Nerve Lesion

ROC

Receiver Operator Characteristic

SSE

Sum of Squared Error of Prediction