Evaluation of Spastic Muscle in Stroke Survivors Using Magnetic Resonance Imaging and Resistance to Passive Motion

Participants 

Six hemiplegic stroke survivors (5 men, 1 woman) were recruited from the outpatient Tone Management Program at University Hospital (Upstate Medical University) to participate in this study (table 1). None of the subjects was currently receiving pharmacologic or other forms of management therapy for the hemiplegia. The Syracuse University and Upstate Medical University institutional review boards approved the experimental protocol, and all subjects provided written informed consent before testing.

Table 1. Subject Demographic Data

	Subject	Weight (kg)	Site of Lesion	MAS Biceps	MAS Triceps	Age (y)	Months Since Stroke
	1	69.09	L MCA	2	1	51.2	72
	2	93.18	R thal	1+	2	55.8	125
	3	90.91	R MCA	1	2	62.6	10
	4	71.82	R MCA	1	0	57.4	35
	5	80.00	R MCA	1	2	52.9	115
	6	90.91	R MCA	1	3	55.7	33
	Mean	82.65				55.93	65
	SD	10.54				3.96	47.11

Abbreviations: L, left; MCA, middle cerebral artery; R, right; thal, thalamus; SD, standard deviation.

General Overview of the Experimental Design 

When subjects reported to the laboratory, T2-weighted magnetic resonance images were obtained from the upper arm of both the affected and unaffected sides. Next, each subject’s affected arm was passively moved through elbow extension and flexion at varying speeds via a motor-driven dynamometer, and the generated torque was recorded. The velocity-associated torque increase from the slowest speed (.087 radian/s) was calculated to quantify spasticity. In addition, subjects’ spasticity was assessed as the passive resistance to manual movement, and a score was assigned according to the MAS.10 After these tests were completed, the maximal voluntary contraction (MVC) strength for both elbow flexion and extension were determined, and subjects performed 3 sets of 10 repetitions of concentric-action elbow flexion and extension exercise at 33% of their MVC. Immediately (≈5min from end of exercise to beginning of scan) on completion of the exercise bouts subjects returned to the MRI scanner, and T2-weighted images of the upper arm were again obtained. Subsequently, this protocol was repeated for the unaffected side. Detailed information on all procedures is described below.

Magnetic Resonance Imaging 

Standard spin-echo magnetic resonance images of the right and left upper arms were obtained using a 1.5-T superconducting magnet.a These procedures were similar to those previously described.11, 12, 13, 14 Briefly, 10-mm–thick transaxial images (repetition time, 2000ms; echo time, 30ms; slice-to-slice interval, 12mm) were obtained along the entire length of the upper arm. During acquisition, subjects lay on the contralateral side so that the arm of interest was positioned along the midline of the body. After data collection the magnetic resonance images were saved to a disk for postprocessing (see Data Analysis).

Resistance to Passive Movement 

Resistance to passive movement was assessed in 2 ways: mechanically recorded torque response to motor-driven movements at various specific velocities6 and the more clinically used subjective grading based on an experienced investigator manually moving the arm at a rapid velocity (MAS) (see table 1).10

The generated torque of motor-driven movements was measured in a similar fashion to that previously described.6 Briefly, subjects were seated in a Biodex System 3 dynamometer.b The arm was positioned in a wrist-hand orthosis,c which was securely attached to a lever arm connected to the motor axis that housed a torque transducer, optical encoder, and potentiometer. The elbow axis of rotation was positioned directly over the motor axis, with a shoulder abduction angle of about 80° and shoulder flexion of 0°. Subjects’ arms were moved through a comfortable ROM via a direct-current servomotor at velocities of 0.087, 1.047, 1.571, and 2.094 radians/s (5°, 60°, 90°, and 120°/s, respectively). The slowest velocity was always tested first; if a subject tolerated the velocity, the next faster velocity was attempted. If a subject had not tolerated a given velocity, the testing would have been terminated at that speed; fortunately, all subjects tolerated even the fastest velocities. The average comfortable ROM was 92°±14°. Three trials at each speed were conducted for both the elbow flexors and extensors. Specifically, 3 trials of elbow extension and flexion were initially made at .087 radian/s, then 3 trials were assessed for elbow extension at 1.047 radians/s, during which the return flexion movement was at 0.087 radian/s. The extension velocity continually increased to 1.571 and 2.094 radians/s while the returning flexion movement was still maintained at the slow velocity (.087 radian/s). After the extension movements were completed, the various movement velocities were repeated, except with the faster velocities occurring during flexion, while the extension movements were kept slow (.087 radian/s) and constant. Torque, position, and velocity were amplified 500 times and sampled at 100Hz. These signals were subsequently averaged over 8 weighted epoch samples of a sliding window and saved on a disk for subsequent analysis (see Data Analysis).

The MAS assessment was conducted by an experienced physical therapist who conducts spasticity assessments on a regular basis. The grades of the MAS range from 0 to 4 (includes 1+), with 0 indicating no increase in muscle tone and 4 indicating that the limb is rigid in flexion or extension. For both measures of resistance to passive movement the unaffected and affected sides for the extensors and flexors were assessed.

MVC and Resistance Exercise 

During the voluntary exertions subjects were seated in a Biodex System 3 dynamometer arranged to assess elbow flexion and extension forces. The forearms were positioned neutrally (half-way between pronation and supination), and the lower portion of the upper arm was rested on an arm rest at an angle of approximately 45° lateral from and vertical to the torso. Straps were used across the pelvis and torso regions to minimize movement and synergistic contributions. One subject was unable to comfortably perform the exercise in this anatomic position because of shoulder pain; thus this person was positioned with a lesser angle from the torso for both the affected and unaffected sides. If subjects were unable to effectively grasp the handle their arm was rigidly attached to the lever arm via the orthosis.

To assess MVC strength, each subject’s arm was maintained at an elbow flexion angle of 90°, and a minimum of 3 trials were provided for both flexion and extension. During the strength assessments subjects were asked to gradually increase torque and perform a maximal contraction for 5 seconds. During testing, strong verbal encouragement was provided by the investigators. The highest torques recorded for flexion and extension were considered the MVCs. The torque signal was amplified 500 times, sampled at 100Hz, and averaged over 8 weighted epoch samples of a sliding window.

Resistance exercise was performed using the isotonic mode of the Biodex System 3 dynamometer. This allows for repetitive movements of the concentric action only (concentric elbow flexion, then extension). Subjects performed 3 sets of 10 repetitions at an intensity set at 33% of the MVC. A 2-minute rest was provided between the sets.

Data Analysis 

The magnetic resonance images were transferred to a computer and analyzed with ImageJ software.d From the magnetic resonance images 2 variables were calculated: (1) muscle CSA and (2) T2. Muscle CSA was calculated from the pre-exercise (resting) images for the biceps and triceps brachii. This calculation was based on an average CSA over 5 slices on both the affected and unaffected sides, which were matched based on anatomic markers. Muscle T2 was calculated on a pixel-by-pixel basis and averaged over the entire muscle using the same 5 slices in the biceps brachii and triceps brachii for each respective side from the pre-exercise and postexercise magnetic resonance images.

Spastic hypertonia was quantified by calculating the reflex torque in a similar manner as previously described.6 Briefly, the reflex torque was considered the peak torque recorded at 1.571 radians/s between a constant-velocity segment of movement (elbow flexors, 100°−120°; elbow extensors, 80°−90°) after subtracting the passive torque recorded at the slowest velocity (.087 radian/s). The movement velocity speed of 1.571 radians/s was chosen for this analysis because it evoked the largest reflex torque responses, and an average of the 3 trials was used in the calculation. We calculated the torque in this fashion because by subtracting the torque generated during a slow movement, the passive tension generated from noncontractile elements (ie, tendon) is removed; thus the calculated outcome (reflex torque) should primarily be due to the evoked stretch reflex response and represent hypertonicity. Data analysis was performed with the System 3 Advantage Software (version 3.29).a

Statistical Analysis 

Mixed-model analysis-of-variance techniques with least significant differences post hoc tests were conducted to evaluate the differences in the dependent variables (passive resistance torque, reflex torque, CSA, T2, strength) with respect to the independent variables (ie, side [affected, unaffected], movement velocity [0.087, 1.047, 1.571, 2.094 radians/s], exercise state [pre-exercise, postexercise]). The SPSS statistical softwaree was used for all analyses. A preset α level of significance was established at .05.

Passive Resistance to Movement 

MAS scores for the elbow extensors ranged from 0 to 3 and for the flexors ranged from 1 to 1+. On the Biodex dynamometer test both the elbow extensors and flexors had a velocity-related increase in torque as movement velocity increased when compared with the unaffected side (P<.05) (figs 1A, 1B). Overall, the extensors had a greater degree of spastic hypertonia compared with the flexors (reflex torque, 5.89±1.65Nm vs 0.84±0.44Nm; P=.04). Interestingly, in comparing the measured values of spastic hypertonia obtained via mechanical recordings (reflex torque) versus manual assessment (MAS), it appears that the mechanical recordings provide a considerably wider range of values for the reflex torque compared with MAS (figs 2A, 2B).

View Large Image Download to PowerPoint

Fig 1. Generated torque in response to passive movement of spastic (affected side) and nonspastic (unaffected side) arm movements at varying velocities for the (A) elbow extensors and (B) flexors in 6 hemiplegic stroke survivors. *Affected greater than unaffected at respective velocity. †Affected extensors: 1.571>0.087 radians/s, affected flexors: 1.571>1.047>0.087 radians/s.

View Large Image Download to PowerPoint

Fig 2. Variation in reflex torque with respect to subjective scoring based on the MAS. Note the more sensitive detection of spastic hypertonia (reflex torque) via mechanically detected force to passive movement in (A) the elbow extensors of subjects with varying degrees of MAS scores and the variation in reflex torque in (B) the elbow flexors despite all subjects having an MAS score of 1 or 1+.

Muscle Strength and Size 

Muscle strength was drastically lower on the affected side compared with the unaffected side, with a 65% and 61% lower strength observed for the flexors and extensors, respectively (P<.05) (fig 3A). Muscle CSA was 25% less in the extensors (P<.05) and only 5% lower in the flexors (P<.05) (fig 3B).

View Large Image Download to PowerPoint

Fig 3. (A) Muscle strength and (B) CSA of the affected and unaffected sides. *Unaffected greater than affected.

Resting and Postexercise T2 

The T2 of the resting elbow flexor muscles (pre-exercise) had a tendency to be slightly higher on the affected side (30.71±0.79ms vs 29.94±0.47ms, P=.07; η2 effect size, .48) (Fig 4, Fig 5). However, the extensors did not show a trend toward having a higher resting T2 value on the affected side (30.57±1.88ms vs 30.4±0.72ms, P=.95) (see fig 5). For both the extensors and flexors the contraction-induced increases in signal intensity were significantly blunted for the affected sides: the exercise-induced change in T2 was 1.9 and 2.9ms less for the flexors and extensors, respectively (P<.05).

View Large Image Download to PowerPoint

Fig 4. A representative example of an MRI scan of the unaffected and affected upper arm. Note the reduced muscle size in the elbow extensors and the increased signal intensity of the elbow flexors on the affected side.

View Large Image Download to PowerPoint

Fig 5. MRI T2 values of spastic (affected) and nonspastic (unaffected) skeletal muscle at rest and after resistance exercise. *Significant increase from postexercise value. †Affected postexercise greater than unaffected postexercise.

Feasibility of Testing 

Initially we had several concerns about the feasibility of the study, mostly relating to the positioning and ROM of subjects. For example, the MRI scanning required subjects to remain still with the arm outstretched above the head for about 10 minutes. Subjects had trouble with this on the affected side, and there was not a standard positioning that worked for all subjects. We were required to use pads inside the MRI device to position the upper arm appropriately in a manner that could be held for the duration of the scan. Typically it took about 30 minutes to find a suitable positioning for each subject. We were also concerned about how quickly we could transport subjects from the exercise facility to the MRI scanner in the next room. Subjects were able to be unstrapped from the Biodex and moved quickly into the scanner; it was critical, however, that we could rapidly position them in the scanner. On average it took about 5 minutes from the end of exercise to the start of the scan. It was important that we not disrupt the arrangement of the pads set up on the initial scan. We also had a concern about the ROM and the velocity of testing on the passive motion dynamometer tests. We were careful to use a comfortable ROM, which inherently varied among subjects. This comfortable zone ranged from 71° to 115° and averaged 91°. Accordingly, we were concerned about what velocities of movement should be used. The faster the velocity the higher the likelihood of detecting resistance, but higher velocities also have the potential to exacerbate the spasticity and be uncomfortable for subjects. For the upper arm we found that velocities ranging from 5° to 120°/s were well tolerated. The fact that we observed a wide range of reflex torque scores suggests that this velocity range would be suitable for further study. A faster velocity could probably be included as well.

Muscle Size and Strength 

In terms of muscle size and strength, it is not surprising that we observed lower maximal voluntary strength in the affected biceps and triceps (≈60%−65%). What is more surprising is the 25% lower muscle CSA in the triceps on the affected side compared with the unaffected side. This disproportionately large decrease in strength is probably due to the ability (or lack thereof) to centrally activate the muscle.

It has widely been reported that upper motoneuron lesions such as those occurring with stroke result in minimal or no muscle atrophy.15 This notion has been supported by the observation that in people with spinal cord injury muscle spasticity actually plays a protective role in the preservation of muscle mass.16 However, there are also reports showing atrophy after stroke. For example, a 3% to 4% difference in thigh mass between the affected and unaffected sides after stroke has previously been reported,17 which is similar to what we observed in the less-affected biceps muscles. Others, however, have shown much greater atrophy: Metoki et al18 reported a 25% difference in thigh muscle volume between sides. In our subject population, the triceps were more severely affected, and we too report a 25% difference in muscle CSA between sides. Therefore, it seems plausible that the extent of atrophy might be related to the severity of spasticity, although we did not observe significant correlations between muscle CSA and MAS score or CSA and reflex torque.

However, only 1 longitudinal study19 has actually investigated the atrophy process in hemiplegic stroke survivors; it found about a 22% side difference in thigh muscle CSA on admission to a rehabilitation program. Both sides increased thigh CSA about 10% during the rehabilitation program, and thus the side differences remained the same after the program (mean rehabilitation length of stay, 104d).19 Studies of muscle CSA using muscle biopsies to evaluate the CSA of individual fibers also show mixed results, with some showing normal-sized fibers,20 others showing selective type II atrophy,21, 22, 23 and yet others showing type I and type II atrophy in hemiparetic stroke patients.24 Regardless of how atrophy is measured (muscle biopsy of individual fibers vs whole muscle volumes), there are dramatic differences among studies showing anywhere from no atrophy to large (25%) differences between sides in hemiparetic subjects. Our study used MRI to evaluate the CSA of individual muscles, whereas other studies have used computed tomography, which has lower spatial resolution and is used to identify whole muscle groups.19 This may explain some discrepancy, but more likely individual stroke survivors vary considerably in the extent of atrophy. This is an area that deserves further study, because it seems there are wide disparities among the extent of motor loss and strength compared with the amount of atrophy. There is clearly not a straightforward relation between muscle strength and size in stroke survivors, and it would be clinically useful to understand what factors influence or how to predict the extent of atrophy in a given patient, especially for the more severe cases of atrophy.

Resistance to Passive Movement and MRI to Assess Spasticity 

A number of laboratories have introduced the concept of measuring the resistance of spastic muscle to passive motion at varying speeds—essentially mechanically automating the MAS.3, 4, 5, 6, 7 Although previous reports indicate that this technique has promise,4, 6, 7 it has not been sufficiently studied to know what velocities to use for various muscle groups, if normative values can be obtained, or whether it is reliable. We confirm that resistance to passive motion can be detected with a commercially available device and that it yields values with a wider range than the MAS. We observed that the flexor reflex torque measurements showed a 3-fold difference among subjects who all received an MAS score of 1. Given our small sample size and relatively homogeneous subject pool, we are not able to identify correlations among the various methods of assessing spasticity, but this is suggested for future research.

Our MRI data are novel, because this is the first study to report on the MRI characteristics of spastic muscle. Muscle fMRI is widely used to evaluate muscle activation9, 13 and is known to correlate highly (≈.99) with muscle electromyographic activity and load lifted.25 Resting T2 is 28±3ms in normal resting muscle.11, 12, 13, 14 We hypothesized that resting T2 would be elevated in spastic muscle given its constant state of slight activation, and although not statistically significant (P=.07), we observed a large effect size (η2=.48) for the flexors, suggesting a trend toward higher resting T2 values (P=.07). However, a similar finding was not observed for the extensor muscles, which actually had higher spasticity than the flexors (as assessed from the higher MAS and reflex torque values); therefore, it is difficult to fully interpret these results, and future research is warranted to explore this issue.

The postexercise T2 is reflective of muscle activity during the prior exercise bout. We observed a failure of the spastic muscle to increase T2 in response to exercise, especially in the extensor muscle group, which was more spastic. Based on these data it is not possible to know why the spastic muscle displays an attenuated exercise-induced T2 response, although there are several plausible explanations. It could be as simple as the fact that spastic muscle cannot lift a heavy enough absolute load to cause the metabolic changes required for T2 to increase or that the spastic muscle is not able to be fully activated during the MVC (central activation failure) and, therefore, the relative loading is not a true relative for the spastic muscle. Alternatively, it could be more complex, such as an altered metabolic capacity of spastic muscle, either as a direct result of the neural lesion or, more likely, as a secondary adaptation. In any event, these MRI techniques indicate disparate adaptations between spastic and nonspastic muscle, and it is recommended that further experiments using these and similar MRI techniques such as the evaluation of the heterogeneity of T2-pixel values in higher resolution images be performed. It seems quite likely that muscle fMRI techniques hold promise for more precise spatial information regarding muscle spasticity.

Study Limitations 

Given that this is a pilot study, there were several limitations. There were a small number of subjects, mostly men, with a narrow age range and mild spasticity. This study is not meant to be generalized to a wide variety of clinical populations but rather to show that the exercise, resistance to passive motion, and MRI measurements can be performed on spastic arm muscle. The study is also intended to encourage other researchers to study spastic muscle itself and not just the nervous system.