Impact of shoulder subluxation on peripheral nerve conduction and function of hemiplegic upper extremity in stroke patients: A retrospective, matched-pair study

ABSTRACT Purpose: To investigate the impact of shoulder subluxation (SS) on peripheral nerve conduction and function of the hemiplegic upper extremity (HUE) in poststroke patients. Methods: Thirty post-stroke patients were selected (SS group: 15 patients, non-SS group: 15 patients, respectively). Evaluation of nerve conduction in upper limbs: the compound muscle action potential (CMAP) amplitude and latency of suprascapular, axillary, musculocutaneous, radial, median, and ulnar nerves; the motor and sensory conduction velocity and the sensory nerve action potential (SNAP) amplitude of median, ulnar, and radial nerves. The Brunnstrom stage scale was used to evaluate the HUE motor function. Results: Compared with the healthy side, the CMAP and SNAP amplitudes of tested nerves on the HUE in both groups were lower, and the CMAP latency of the suprascapular, axillary and musculocutaneous nerves on the HUE in the SS group was longer (P < 0.05). Compared with the HUE in non-SS group, the CMAP amplitude of tested nerves (except ulnar) was decreased more (P < 0.05), the motor conduction velocity of the median nerve was lower (P < 0.05), and the Brunnstrom stage of the HUE was lower in SS group (P < 0.05). Conclusions: Stroke may lead to extensive abnormal nerve conduction on the HUE, and SS may aggravate the abnormality, which may disturb the recovery of upper limb function.


Introduction
Shoulder subluxation (SS) or glenohumeral subluxation is one of the common complications in poststroke setting, with an incidence of approximately 17%-81% [1,2]. After a stroke, the hemiplegic upper extremity (HUE) could be affected by muscle weakness, gravity traction, incorrect posture transfer, or peripheral nerve/rotator cuff injury, so the humeral head partially detaches from the glenoid, which leads to the SS [1,3]. After a SS, the shoulder joint capsule, the surrounding tendons, and the brachial plexus nerves could be pulled by the HUE [4,5]. Then, the degree of SS and upper limb dysfunction could be further aggravated [1,3].
Electrodiagnosis and ultrasound medicine studies show that, in post-stroke patients, the peripheral nerves of the HUE could exist different degrees of impairment which seriously delays the function recovery of the HUE [6][7][8]. Meanwhile, because the soft tissue of the shoulder is stretched by long-term gravity, and the hemiplegic shoulder is not under comprehensive protection, SS may result in soft-tissue injuries and shoulder pain, which may negatively impact the motor function recovery of the HUE [9]. However, the impacts of SS on the HUE peripheral nerve conduction in post-stroke are rarely reported.
In the 1980s, Chino et al. [5] observed the variations of peripheral nerve conduction (suprascapular nerve, axillary nerve, musculocutaneous nerve, and radial nerve) in the HUE among post-stroke patients with an occurrence of SS, rather than comparison with healthy upper limb and the HUE without SS. Tsur et al. [10] investigated the difference of the axillary nerve between the HUE after SS and its healthy upper limb, other nerves were not analyzed. So far, no study has comprehensively assessed the degree of peripheral nerve injury in HUE after SS.
Therefore, it is necessary to further analyze the variations of peripheral nerve conduction and function in HUE after SS. The main purpose of this study was to investigate the impact of SS on peripheral nerve conduction and function of the HUE in stroke patients through the comparison of peripheral nerve conduction parameters and motor function between the HUE with SS and without SS.

Study design
This was a retrospective, matched-pair single-center study based on data collected from the Rehabilitation Medicine Center, the Affiliated Suzhou Science and Technology town Hospital of Nanjing Medical University.

Participants
Thirty post-stroke patients were selected from 158 stroke patients who were hospitalized from June 2018 to May 2020, and their peripheral nerves of bilateral upper limbs had been undergone by the electroneuromyographic examination. The characteristics of patients, such as age, gender, type of stroke, hemiplegic side, handedness, height, weight, and disease duration were recorded and paired in terms of retrospective, matched-pair study design [11].
Inclusion criteria: 1) age 18-75 years old; 2) the first onset of unilateral basal ganglia hemorrhage/ischemia stroke confirmed by MRI/CT and with a course of more than 1 month; 3) be able to accept electroneuromyographic examination of the bilateral upper limbs peripheral nerves; 4) the existence or inexistence of the SS determined by palpation or X-Ray measurements [10,12] (Figure 1 has shown the SS of HUE determined by X-Ray). Exclusion criteria: 1) medical history of upper limb peripheral nerve injury or upper limb surgery/trauma in pre-stroke; 2) diagnosis of the complex regional pain syndrome [13]; 3) diagnosis of upper limb peripheral nerve entrapment syndrome [14]; 4) diagnosis of diabetes [15].
This study was approved by the Medical Ethics Committee of the Affiliated Suzhou Science and Technology town Hospital of Nanjing Medical University (approval number: IRB2019062).

Evaluation of nerve conduction [16]
The Keypoint electromyograph/evoked potential device (Keypoint G4, Dantec medical, Denmark) were operated by a five years qualified physician to administer evaluation of nerve conduction on all participants.

Evaluation of upper limb motor nerve conduction
The patient was asked to perform a supine position, and the skin temperature was no less than 32°C. The saddle-shape electrode was used as a stimulating electrode, the superficial electrode was used as a recording and reference electrode, and the ground electrode was placed between the stimulating electrode and the recording electrode. The placement of each electrode is outlined in Table 1.
When motor nerve conduction was assessed, the intensity of electrical stimulation gradually increased until the compound muscle action potential (CMAP) amplitude reached its maximum, and the amplitude was the distance between the baseline and the negative peak.

Evaluation of upper limb sensory nerve conduction
The saddle-shape electrode was used as a stimulating electrode, the ring-shaped electrode or superficial electrode was used as a recording and reference electrode. The sensory nerve action potential (SNAP)/conduction velocity of the median nerve, ulnar nerve, and radial nerve was recorded. The placement of each electrode is outlined in Table 2.
When sensory nerve conduction was assessed, the stimulus intensity gradually increased until the baseline of the waveform was stable and the maximum amplitude arrived, and the waveform was superimposed 30 times. The SNAP amplitude is the distance from the baseline to negative peak, and conduction velocity is the distance between the stimulating electrode and the recording electrode divided by latency.

The variation rate of nerve conduction parameters on HUE
The nerve conduction parameters (CMAP amplitude, and latency, SNAP amplitude, motor, and sensory conduction velocity) on the unaffected side were used as a reference to evaluate the variation rate of the nerve conduction parameters on the HUE. The formula for calculating the variation rate is as follows: The positive value of the change rate represents the percentage of increase compared with the healthy side, and the negative value represents the percentage of decrease compared with the healthy side.

Upper limb function assessment
The Brunnstrom stage scale was used to evaluate the motor function of the upper limb and hand on the hemiplegic side [17].

Statistical analysis
According to the study of İsnaç F et al. [6], the sample size was calculated using G*Power (3.1.7). In the design of matched-pair t-test, if the effect size d = 0.5, alpha error = 0.05, and power (1-beta) = 0.8, at least 27 patients were required.
Statistical analysis was performed using SPSS 22.0 software (IBM Corporation, Armonk, New York). Measurement data were expressed as X±S. All measurement data were tested for normality with the Kolmogorov-Smirnov test and all these data obeyed normal distribution. Hence, the paired t-test was utilized for the comparison between the two groups and between the paretic/nonparetic extremity. And the Brunnstrom stage between the two groups was analyzed using the Wilcoxon Signed Rank test. P < 0.05 was set as a statistically significant difference.

Demographic characteristics
A total of 30 patients was selected and 15 patients with SS were allocated to the SS group, the other 15 without SS were allocated to the Non-SS group (N-SS group).
There were no significant differences in age, height, weight, and clinical course between the SS group and the N-SS group (P > 0.05, Table 3).

CMAP amplitude
No significant difference in the CMAP amplitude of each nerve on the healthy upper limb between the two groups was evident (P > 0.05). The CMAP amplitude of each nerve in the affected upper limb of both groups was significantly lower than that in the healthy side (P < 0.05). Compared with the affected side of the N-SS group, CMAP amplitude of suprascapular nerve, axillary nerve, musculocutaneous nerve, radial nerve, and median nerve in the HUE of the SS group were significantly decreased (P < 0.05). No significant difference in the CMAP amplitude of the ulnar nerve  on the HUE between the two groups was found (P > 0.05). The CMAP amplitude variation rate of the suprascapular nerve, the axillary nerve, the musculocutaneous nerve, and the median nerve in the HUE of the SS group significantly decreased more than that of in the N-SS group (P < 0.05). There was no significant difference in the CMAP amplitude variation rate between the SS group and the N-SS group in the radial nerve and the ulnar nerve of the HUE (P > 0.05, Table 4).

CMAP latency
No significant difference in the healthy upper limb CMAP latency between the two groups was evident (P > 0.05). In the SS group, the CMAP latency of the suprascapular, the axillary, and the musculocutaneous nerves in the HUE were significantly longer than that of the healthy side (P < 0.05), while the CMAP latency of the radial nerve, median nerve, and ulnar nerve were not significantly different from that of on the healthy side (P > 0.05). In the N-SS group, there was no significant difference in corresponding nerve CMAP latency between the HUE and the healthy upper limb (P > 0.05). Compared with the N-SS group, the axillary nerve CMAP latency in the SS group was significantly prolonged more (P < 0.05), and there was no significant difference in CMAP latency between the two groups in the remaining nerves of the HUE (P > 0.05). In the SS group, the variation rate of CMAP latency of the axillary and the suprascapular nerves on the HUE were significantly greater than that of in the N-SS group (P < 0.05), and the difference of CMAP amplitude variation rate between the two groups on the remaining nerves of the HUE was not statistically significant (P > 0.05, Table 5).

Motor nerve conduction velocity
There was no significant difference in motor nerve conduction velocity on the healthy upper limb between the two groups (P > 0.05). The motor conduction velocity of the median nerve on the HUE in the SS group was significantly lower than that of the HUE in the N-SS group (P < 0.05), also, there was no significant difference in the conduction velocity of the radial nerve and the median nerve in the HUE between the two groups (P > 0.05). Besides, there was no significant difference in the variation rate of motor conduction velocity of radial, median, and ulnar nerves in the HUE between the two groups (P > 0.05, Table 6).

SNAP amplitude and sensory conduction velocity
There was no significant difference in SNAP amplitude of radial, median, and ulnar nerves on the healthy upper limb between the two groups (P > 0.05). The SNAP amplitudes of the nerves in the HUE in both groups were significantly lower than that in the healthy side (P < 0.05). There was no significant difference in SNAP amplitude and SNAP amplitude variation rate of the HUE between the two groups (P > 0.05, Table 7).

Sensory nerve conduction velocity
There was no significant difference in sensory conduction velocity of radial, median, and ulnar nerves on the healthy upper limb between the two groups (P > 0.05).
There was no significant difference in sensory conduction velocity of the HUE and the healthy side of the same group in the two groups (P > 0.05). There was no significant difference in SNAP velocity and SNAP velocity variation rates of radial, median, and ulnar nerves in the HUE between the two groups (P > 0.05, Table 8).

Brunnstrom stage
Compared with the N-SS group, the Brunnstrom stage of hemiplegic upper limbs and hands in the SS group was lower, and the difference was statistically significant (P < 0.05, Table 9, 10).

Discussion
This is a retrospective, matched-pair study and it conducts the comparison in upper limb nerve conduction/motor function of the HUE between post-stroke patients with SS and without SS. The results showed that: 1) there was extensive abnormal peripheral nerve conduction on the HUE after stroke; 2) the occurrence of SS could lead to more serious abnormality to the suprascapular, axillary, musculocutaneous, radial, and median motor nerves of the HUE than patients without SS; 3) the occurrence of SS could lead to greater prolongation of the latency of the axillary and       suprascapular nerves on the HUE than patients without SS; 4) hemiplegic upper limb and hand function recovery may be worse in stroke patients with SS than in those without SS. Stroke can lead to extensive damage to peripheral nerves (demyelination changes, axonal damage, etc.) in the HUE [6,7,[18][19][20], as well as, contralesional hemispheric disinhibition, and trans-synaptic changes in peripheral nerves after stroke could reduce the excitability or conduction of peripheral nerves in the hemiplegic side [21][22][23], these may negatively influence the functional recovery of the HUE. Besides, other multiple factors may impact the research outcomes, such as research-operation involvement timing [23], muscle atrophy [24], and peripheral nerve structural changes [25]. Moreover, whether the lesion site of stroke (a motor tract of the motor cortex, internal capsule, or brain stem) affects the peripheral nerve conduction, which still needs further study.
The SS after stroke often manifests as the humeral head to downward subluxation [26]. In the early stage of stroke, due to the weak deltoid and supraspinatus on the HUE, combined with the effect of gravity, the humeral head could not be effectively fixed into the glenoid, which may lead to the occurrence of SS [1,27]. The results of this study suggest that, after stroke, SS may lead to more severe abnormal peripheral nerve conduction on the HUE compared with non-SS stroke patients. It has been confirmed that neurological complications were manifested in 5.4-55% among all shoulder dislocations [28]. In traumatic inferior shoulder dislocation, it has been reported that 29% of the patients experienced a neurological injury, and the axillary nerve is particularly often damaged, probably due to the overload as it goes across the quadrangular space [29]. However, the whole characteristics that the impacts of SS on the HUE peripheral nerves remain unclear.
Anatomically, the axillary, musculocutaneous, radial, and median nerves all travel/pass near the humerus [30]. Therefore, combined with the results of the current study and previous studies [5,10], the humeral head downward displacement after SS may cause persistent traction of the above nerves. Compared with other upper limb nerves, the ulnar nerve has the most mild of abnormal conduction which may be due to the farther anatomical relationship between the ulnar nerve and humerus [30]. However, the specific mechanism of upper limb nerve damage caused by SS after a stroke needs further study.
The suprascapular nerve originates from the brachial plexus and it crosses the suprascapular notch and the transverse scapular ligament to innervate the supraspinatus and infraspinatus [30]. Through the X-ray in comparing the position of bilateral scapulae and glenohumeral joints in stroke patients, Culham et al. [26] concluded that the scapulae on the hemiplegic side were generally downward rotation, downward displacement, and outward displacement during the period of flaccid paralysis, so the occurrence of the SS could be more easily. Anatomically, the downward rotation, downward displacement, and outward displacement of the scapula may cause the higher tension of the suprascapular nerve, then lead to pull injury. In the case of scapulothoracic dissociation, excessive traction force and prolonged pull on the infraclavicular brachial plexus may cause the injury of the axillary and suprascapular nerves, and have an adverse effect on the spontaneous recovery of the nerve lesions [31]. However, the exact injury mechanism is still unclear.
Tsur et al. [10] investigated the electrophysiological parameters of axillary nerve in patients with the SS after stroke and their parameters displayed that axillary nerve damage may occur. Chino et al. [5] found that in stroke patients with the SS, the latency of the suprascapular, axillary, musculocutaneous, and radial nerves on the HUE was longer than the reference values and the corresponding muscles had spontaneous potential. However, this study only investigated the latency, conduction velocity, and spontaneous potential of the relevant motor nerves, and, it did not conduct a comparison between the healthy side and the HUE. Besides, CMAP and SNAP parameters were not considered by this study.
In post-stroke, the recovery of the HUE function should be related to the location/degree of brain injury and the timing of rehabilitation intervention [32]. In addition, the results of the current study demonstrated that post-stroke patients with the occurrence of the SS may aggravate the abnormal nerve conduction on the HUE which could impact the functional recovery of the HUE. Previous studies [33,34] have suggested that long-term denervation can lead to muscular dystrophy and irreversible loss of neuromuscular junctions.
It has been reported that surface electrical stimulation or functional electrical stimulation (FES) can prevent the deterioration of denervated muscle atrophy [35,36]. Also, two micro-stimulators of electrical stimulation were implanted next to the axillary nerve and movement point of deltoid middle bundle fiber, respectively, which may be useful to reduce the shoulder pain and improve the shoulder function in stroke patients with SS [37,38]. Meanwhile, according to the systematic review of Arya et al. [1], applying FES on the supraspinatus and posterior deltoid muscles could be effective in reducing the subluxation during the early stage of stroke; however, the effectiveness of shoulder supports/orthosis, taping technique, robotic assessment therapy, or sling exercise in improving the SS of stroke patients still remains unclear. Therefore, combining with the results of the current study, the authors speculate that if the prevention strategies (i. e. FES) of the SS is carried out in the early stage of stroke, it might reduce the degree of the HUE nerve damage. However, this speculation still needs further study to confirm.

Limitations and future direction
Limitations of our preliminary study: 1) this is a single-center, retrospective study, the results of which may be biased due to the small sample size and matching criteria; 2) although the grouping of all patients was matched according to the type of stroke, it was difficult to accurately control the damage degree of central nervous system in both groups; 3) in this study, because not all patients underwent shoulder radiography, no correlation study between the degree of shoulder subluxation/the change of scapular position and the degree of upper limb nerves damage was performed.
Follow-up studies can explore the change rule of the impact of SS on nerve conduction in hemiplegia, investigate the relationship between the degree of shoulder subluxation/the change of scapular position and the degree of upper limb nerves damage, and study the effectiveness of different shoulder protection strategies on nerve conduction and function in the HUE after stroke.

Conclusions
The current study shows that stroke may lead to extensive abnormal nerve conduction of the HUE nerves, and the SS may aggravate the abnormality degree of the HUE nerves, especially in the axillary, the musculocutaneous, and the suprascapular nerves. The occurrence of the SS in post-stroke patients may impact the functional recovery of the HUE. Therefore, the protection of the shoulder joint on the HUE should be strengthened in the early stage of stroke.