Impact of Wrist and Finger Extension on Passive Joint Torques and Their Association with Forearm Flexor Muscle Structure in Hemiparetic Cerebral Palsy: Early Findings

Abstract

This study aimed to characterize interlimb differences in passive torques (PT) at the wrist and metacarpophalangeal (MCP) joints in individuals with hemiparetic cerebral palsy (HCP), and to relate them to previously acquired diffusion tensor imaging (DTI)-derived diffusivity measures in forearm flexor muscles. In 10 individuals (5F, 18.14±8.7 years) with HCP, a custom-built device was used to measure the PTs produced by the wrist and MCP joints of both limbs in 16 discrete hand configurations. Interlimb differences in wrist and finger PT showed significant increases with extreme wrist and finger extension and severity of HCP. Furthermore, increased interlimb differences in PT during extreme wrist and finger extension were significantly correlated with DTI-derived diffusivity estimates, indicating that underlying structural muscle changes are related to increased PTs in extended hand configurations.

I. Introduction

Cerebral palsy (CP) results in reduced passive range of motion (PROM) and increased resistance to passive muscle elongation, known as passive stiffness. This phenomenon occurs more at distal joints such as the wrist and fingers and can progress to permanent shortening of the musculotendinous complex, clinically termed a ‘contracture’ [1]. However, the underlying structural mechanisms by which contractures develop are unclear. It has been suggested that altered passive properties are caused by intrinsic muscle changes in addition to altered neural drive [2]. Therefore, a better understanding of the relationship between altered passive properties and root muscle adaptations is necessary for the prevention of contracture development in CP [3].

Existing studies have reported a decrease in PROM and an increase in passive stiffness in the joints of the upper and lower extremities in CP [4, 5]. However, existing investigations have methodological limitations that confound the understanding of passive properties in CP. First, manual implementation of passive movement at a given speed may potentially lead to inconsistent stretch velocities throughout the PROM. Additionally, the use of continuous motion, even at low velocities, may elicit hyperactive stretch reflexes that invalidate passive torque measurements. Finally, existing work has not considered the heterogeneity of the CP population, which includes varying subtypes and etiologies that affect resulting motor impairments [6].

Here, we implemented a custom-built device [7] to measure PTs around the wrist and MCP joints, the proximal finger knuckles, at discrete joint angles. Measurement at discrete hand postures, in addition to EMG activity measurement, ensures that muscles are fully relaxed thus measuring truly passive torques. Additionally, this study focused on individuals with HCP, which causes motor impairments on only one (paretic) side of the body. The other (nonparetic) was used as a within-subject control for the normalization of measurements. Finally, we compared passive torque measures in the wrist and MCP to previously acquired DTI-derived estimates averaged across two forearm flexor (flexor carpi radialis and flexor digitorum profundus) muscles in the same participants. DTI has the potential to non-invasively estimate musculoskeletal pathological changes, such as extracellular matrix (ECM) proliferation or muscle fascicle shortening [8], which are likely to contribute to increased PTs following CP [9].

II. Methods

A. Study Cohort

This study included 10 individuals (5F, 18.14±8.7 years) with mild to severe HCP. All individuals were between 12 and 40 years old and had motor impairment confined to one side. All participants or their guardians provided informed, written consent and assent to participate in this study, which was approved by the institutional review board of Northwestern University.

B. Experimental Setup

Participants were seated in a Biodex experimental chair (Mirion Technologies Inc., San Ramon, CA) with shoulder and lap straps to minimize trunk and shoulder motion. Muscle activity was measured at 1kHz using a surface EMG system (16-channel Bagnoli EMG System^™; Delsys, Inc., Natick, MA). EMG sensors were placed over the flexor carpi radialis, flexor digitorum superficialis, flexor digitorum profundus, extensor carpi radialis, and extensor digitorum communis muscles. After placement of the EMG sensors, the forearm and hand were secured in a custom-built non-motorized device [7] with the wrist and MCP axes of rotation each aligned with one degree of freedom load cells on the device (Fig. 1A). The arm was positioned such that shoulder abduction was 15° and elbow flexion was 100°, and the forearm was mechanically coupled to the arm of the Biodex chair using custom 3D printed rings. The setup was replicated for the second arm.

Figure 1:

Pipeline of passive torque data acquisition and analysis. (A) Experimental setup of the forearm and hand positioned in custom-built, non-motorized wrist and torque sensing device. (B) Passive torque hand configurations tested included the combination of four MCP postures and four wrist postures, ranging from 0° neutral to 45° extension (E), in 15° increments, for a total of 16 configurations. (C) Example of passive torque analysis, including torque-angle curves for both nonparetic (NP, green) and paretic (P, pink) MCP joints, as well as schematic of ΔPT_MCP calculation at various hand configurations. Data point for P_{wrist 15°E, MCP 45°E} is missing due to the participant’s inability to reach that posture.

C. Data Acquisition

First, the participant’s wrist and fingers were passively ranged until fully relaxed. Then, both angle positions were unlocked, and participants were told to rest their wrist and fingers at their resting posture, which was recorded to the closest 15° increment. With the wrist and finger position unlocked, three 15-second baseline trials were collected to find EMG activity threshold and torque offset values. Finally, PTs were measured across the combination of four MCP postures and four wrist postures, resulting in a total of 16 hand configurations. Both MCP and wrist postures ranged from 0° (neutral) to 45° extension in 15° increments (Fig. 1B). The interphalangeal joints were held at neutral for all postures. The order of these 16 hand configurations was randomized for each participant. After an extensive stretching protocol and under conditions that encouraged participants to fully relax [10], the wrist and fingers were moved into each configuration manually, and the posture was held until all evidence of muscle activity (observed by the EMG signal) and/or physical movement stopped. At this point, the passive torque produced by the wrist and MCP joints was measured for 15 seconds at a sampling frequency of 1 kHz. The same data acquisition protocol was followed for the second arm.

D. Data Processing and Analysis

All data was processed in MATLAB version 9.13.0 (The MathWorks Inc., Natick, Massachusetts). Torque data was analyzed according to methods previously established [10]. For all trials, the torque and rectified EMG data were filtered in forward and reverse directions with a 4^th-order Butterworth low-pass filter. From the baseline trials, the torque offsets for both joints were calculated according to

$$\mathit{\text{Torque}}{\mathit{\text{Offset}}}_{\mathit{\text{joint}}}={\mu }_{\mathit{\text{joint}},\mathit{\text{baseline}}},$$ (1)

where μ_{joint,baseline} is the average torque produced by the defined joint across the baseline trials, during which both joints were at their resting posture. The EMG threshold for each muscle (EMG_m,thresh) was calculated according to

$${\mathit{\text{EMG}}}_{\text{m},\text{thresh}}={\mu }_{m,\mathit{\text{EMG rect}}}+3*{\sigma }_{m,\mathit{\text{EMG rect}}},$$ (2)

where μ_{m,EMG rect} is the average rectified EMG value for the defined muscle across the baseline trials, and σ_{m,EMG rect} is the standard deviation of the rectified EMG value for the defined muscle across the baseline trials. For each passive torque trial, the torque offset was subtracted from the torque data for each joint. The torque and EMG data were then divided into one-second bins for a total of 15 bins per trial. Bins that had an EMG value that exceeded the EMG threshold for any muscle or had a torque value that deviated more than 5% from the mode across the entire trial were eliminated. Then, the average torque across all bins was measured to determine the passive torque produced at that hand configuration. Each passive torque was divided by the cross-sectional area of the forearm to normalize torque measures to arm size to account for potential interlimb differences in size. Finally, the interlimb difference in passive torque for each configuration and joint, ΔPT_joint,config, was determined according to

$${\Delta PT}_{\mathit{\text{joint}},\mathit{\text{config}}}={PT}_{P,\mathit{\text{joint}},\mathit{\text{config}}}–{PT}_{NP,\mathit{\text{joint}},\mathit{\text{config}}},$$ (3)

where PT_{P,joint,config,} and PT_{NP,joint,config} represent the average PT produced by each joint at each hand configuration in the paretic and nonparetic limbs, respectively (Fig. 1C).

E. Statistical Analyses

All statistical tests were performed in R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Paired t-tests were performed to determine significant differences in the ΔPT for each joint across hand configurations. Cumulative ΔPT at each joint was calculated as the sum of the ΔPT across that joint’s four configurations. Finally, for two hand configurations (neutral: MCP and wrist at 0°; extended: MCP and wrist at 45° extension,), Pearson correlation coefficients were calculated between the ΔPT and interlimb differences in average mean diffusivity (ΔMD) of forearm flexor muscles (flexor digitorum profundus and flexor carpi radialis), determined according to previously published methods to estimate muscle diffusivity properties [11]. In all statistical tests, the significance level was set at p<0.05.

III. Results

A. Interlimb Differences in PTs Across Hand Postures

Fig. 2 shows the ΔPT for the MCP and wrist joints for eight hand configurations: four MCP and two wrist postures (Fig. 2A) and four wrist and two MCP postures (Fig. 2B). When the MCP is at 45° extension, ΔPT_MCP torque was increased by 96.8% when the wrist posture changed from 0° to 45° extension (p=0.036) (Fig. 2A). There were no significant differences in ΔPT_MCP or ΔPT_wrist between any other hand configurations (p>0.05 for all).

Figure 2.

ΔPT (Nm/mm²) in MCP and wrist joints across hand configurations. (A) ΔPT_MCP across all four MCP postures and two wrist postures. (B) ΔPT_wrist across all four wrist postures and two MCP postures. A “ns” indicates no significant difference, and a * indicates p<0.05.

B. Interlimb Differences in PTs Across Severity Level

Fig. 3 shows the cumulative ΔPT_MCP (Fig. 3A) and ΔPT_wrist (Fig. 3B) across levels of impairment and hand postures. Compared to the mild case, the individual with severe impairment showed a 9-to-14-fold increase in cumulative ΔPT_MCP as wrist posture was altered (neutral wrist: 9.8-fold, extended wrist: 14.0-fold) (Fig. 3A). Meanwhile for wrist torques, compared to the mild case, the individual with severe impairment showed a 12-to-16-fold increase in cumulative ΔPT_wrist as the MCP posture was altered (neutral MCP: 15.7-fold, extended MCP: 12.3-fold) (Fig. 3B).

Figure 3:

Cumulative ΔPT (Nm/mm²) in MCP and wrist joints in one participant with mild HCP and one participant with severe HCP. (A) Cumulative ΔPT_MCP across neutral and extended wrist postures. (B) Cumulative ΔPT_wrist across neutral and extended MCP postures. Diagonal black lines represent estimated torque values (based on torque at achieved previous wrist posture) in the severe case, as this participant could not extend wrist past 15° extension.

C. Correlations Between PTs and Forearm Flexor Mean Diffusivity

Fig. 4 shows the relationship between ΔPT across two hand configurations (neutral and extended) and forearm flexor ΔMD values. The ΔPT_MCP values were not significantly correlated to ΔMD at any of the hand postures (p>0.05 for all) (Fig. 4A). Meanwhile, the ΔPT_wrist values during extreme wrist and finger extension were significantly correlated to ΔMD (r=0.80, p=0.017), but did not show significant correlations when the wrist and fingers were held at neutral (p>0.05) (Fig. 4B).

Figure 4:

ΔPT (Nm/mm²) versus ΔMD values (mm²/s) in forearm flexor muscles, at neutral and extended hand configurations. (A) ΔPT_MCP versus interlimb difference in MD. (B) ΔPT_wrist versus the interlimb difference in MD. Pearson correlation coefficients (r) and associated p-values are shown for each comparison. A “ns” indicates no significant difference and a * indicates p<0.05.

IV. Discussion

This study investigated the passive torques produced by the MCP and wrist joints at different hand configurations in individuals with HCP. We found that when the wrist or fingers were extremely extended, there was an increased interlimb difference in passive torques. Furthermore, interlimb discrepancies were even greater in the case of severe impairment. Finally, we found that interlimb differences in DTI-derived diffusivity measures were related to passive torques produced by the wrist when the wrist and fingers were extremely extended.

Our finding of a 96% increase in ΔPT_MCP as the wrist and fingers move into extreme extension is in line with previously reported ranges. Although only a handful of studies have investigated passive properties in the upper limbs of individuals with CP, von Walden et al. found a 2-fold increase in passive stiffness of the forearm flexor muscles [4] while another study reported a 32% increase in wrist flexor passive stiffness [12]. Our findings are consistent with previous reports of passive properties even after correcting for confounding factors such as heterogeneous CP type and differences in limb size, which were not accounted for in previous work.

Specifically, we observed that ΔPT_MCP only increased significantly when both joints moved from neutral to extreme extension. Therefore, the paretic joints produced more passive torque, as compared to the nonparetic joints, at the ends of the joints’ PROM, while there were fewer interlimb differences over the middle of the PROM. Previous literature has suggested that during extreme wrist extension, forearm flexor sarcomeres, or the smallest functional unit in skeletal muscle, are under extreme stress with less overlap between actin and myosin filaments, resulting in greater resistance to elongation and therefore increased PTs [4]. Future work could expand upon these early findings by investigating how sarcomere and fascicle lengths relate to wrist and finger PTs.

We observed decreased MD in the paretic forearm flexor muscles with respect to the nonparetic limb MD, and this was associated with increased paretic PT, with respect to the nonparetic limb, during extreme wrist and finger extension. However, in this limited cohort of primarily mildly impaired participants, there was no significant relationship at the neutral hand posture. Previous work has identified that decreased MD in skeletal muscle may indicate excessive accumulation of connective tissue, or ECM [8]. These findings further highlight that intrinsic muscle adaptations contribute to resistance to passive elongation at the ends, but not the middle, of the PROM.

The lack of significant correlations between ΔPT_MCP and ΔMD is likely due to the fact that the forearm muscles reported here cross the wrist joint but not the MCP joint [13]. Therefore, intrinsic muscle changes would contribute more to increased PT in the wrist than in the MCP. Future work should investigate the relationship between PT produced by the MCP and structural DTI-derived parameters in intrinsic hand muscles, rather than only extrinsic forearm flexor muscles.

When comparing ΔPT between a mild and severe case of HCP, we found a 9-to-14-fold increase in cumulative ΔPT_MCP and a 12-to-16-fold increase in cumulative ΔPT_wrist. Previous work has shown that ECM content increases with impairment level [14], so the stark increase in cumulative ΔPT in the severe case may be due to greater structural adaptations in the forearm flexor muscles. However, other studies have not reported significant correlations between passive stiffness measures and functional assessment scores [3]. More work is needed in a population of diverse levels of impairments to understand how passive biomechanical properties are altered with increasing severity of CP.

Several limitations of this study should be acknowledged. Here, we have established proof of concept that increased passive torques during extreme wrist and finger extension are related to adaptations in muscle structure. However, the sample size is small and includes only one participant with severe impairments, limiting the generalizability of the findings. Future work with a larger sample size and a study cohort with more heterogeneous impairment levels is needed to validate these findings and explore their broader applicability. Additionally, we only measured passive torques at joint angles between 0° and 45° extension. Though extreme wrist and finger extension may not appear to be functionally significant, the ends of the PROM are often where individuals with CP experience the most difficulty with fine motor control [6]. Future work can expand to 60° extension or more to understand the passive properties at the true ends of each participants’ PROM. Finally, the relationship between passive torques and age was not investigated in this cross-sectional study, but it has previously been reported that alterations in passive properties in CP are progressive [3]. Longitudinal studies could provide insights into how passive torque production, structural adaptations, and their relationship changes with age.

V. Conclusion

This study provides valuable insights into how passive properties are altered in the affected MCP and wrist joints at different wrist and finger postures in individuals with CP. The results shown here establish, for the first time in a CP cohort, that passive torques can be measured at discrete joint angles, without significant EMG activity, removing potential contributions from neural activity during measurement. Additionally, investigation of the relationships between passive torque and DTI-derived muscle diffusivity estimates, suggest complex underlying structural mechanisms. A comprehensive understanding of how intrinsic muscle changes alter passive biomechanical properties and has important implications for targeted intervention and therapy in individuals with CP.

Clinical Relevance—

These findings establish that intrinsic forearm muscle adaptations significantly and negatively influence passive biomechanical properties when individuals with HCP extend their wrist/fingers to the end of their passive range of motion.