Autologous Nerve Grafting Versus Multiple Nerve Transfers for Upper Trunk Brachial Plexus Injury: A Rat Model Study

Summary

By establishing rat models of root-level left upper-trunk brachial plexus injury repaired with either nerve grafting or nerve transfer, we found that both groups achieved similar muscle strength; the grafted animals showed superior movement coordination.

Abstract

Upper trunk (C5-C6) injuries of the brachial plexus impair shoulder abduction and elbow flexion. For C5-C6 ruptures with preserved proximal stumps, nerve grafting (NG) remains the conventional approach. Nerve transfers (NT) represent the sole viable option in C5-C6 avulsions. The current global standard for NT involves triple transfer procedures: spinal accessory nerve (SAN) to suprascapular nerve(SSN), Oberlin procedure, and Leechavengvongs procedure. Recent clinical trends indicate a broadening of NT indications to include root rupture scenarios, driven by the rationale that distal NT significantly shortens axonal regeneration distance. To enable direct comparison, this study established two rat models simulating these clinical strategies: 1) an NG model where the C5-C6 stumps were grafted to the SSN and upper trunk; 2) a multiple NT model comprising the above three procedures. Functional recovery (Ochiai score, Barth foot-fault test, and Terzis grooming test), electrophysiological parameters and regenerated sensory axon counts were evaluated at 8 and 16 weeks postoperatively. At 8 weeks, no statistically significant differences existed in functional outcomes between groups. By 16 weeks, while Ochiai scores and Terzis results remained comparable, NG demonstrated significantly superior performance on the Barth test. Electrophysiologically, NT exhibited significantly better recovery rates for musculocutaneous, axillary and suprascapular nerves at 8 weeks. However, these electrophysiological advantages were no longer statistically significant by 16 weeks. At 8 weeks, NG already carried more regenerated sensory axons in musculocutaneous and axillary nerves, while SSN counts were similar. By 16 weeks, NG surpassed NT in sensory axon numbers for all three nerves. These findings in a rat model of sharp C5-C6 transection indicate that, within a 16-week post-operative period, NG achieves comparable functional and electrophysiological recovery to NT, while yielding superior restoration of sensory and coordinated motor function. However, further studies with longer follow-up are needed to determine the translational relevance of these findings.

Introduction

Brachial plexus injuries are generally classified into upper-trunk injury (C5–C6, with or without C7 involvement), lower-trunk injury (C8–T1), and pan-plexus injury (C5–T1). A compressive force directed at the head and shoulder typically produces an upper-trunk injury. Because the superior transforaminal ligaments at the C5–C7 root levels partially dissipate upward transmission of the traumatic force, rupture of these roots—especially C5—at the intervertebral foramen is not uncommon¹. However, severe violence can still result in root avulsions and may even extend to the lower trunk. When the upper limb is forcibly distracted in the horizontal plane, lower-trunk root avulsions predominate; with greater force, complete brachial plexus avulsion may occur. Owing to the absence of superior transforaminal ligaments protecting the C8–T1 roots, virtually all traumatic lower-trunk injuries are root avulsions². Therefore, regarding the pattern of root injury, C5–C6 roots are more likely to rupture, whereas C7–C8–T1 roots are more prone to avulsion. These pathoanatomical characteristics are critical for selecting the appropriate surgical strategy.

Upper-trunk injuries are characterized clinically by loss of shoulder abduction and elbow flexion. In patients with C5–C6 ruptures in whom viable proximal root stumps are still present, the conventional strategy has been nerve grafting (NG): the residual C5 and C6 roots are bridged—usually with sural nerve autografts—to the anterior and posterior divisions of the upper trunk and to the suprascapular nerve³. When C5 and C6 roots are avulsed, nerve transfer (NT) becomes the only reconstructive option. The operative algorithm adopted by most centers worldwide combines three procedures: the spinal accessory nerve to the suprascapular nerve, one fascicle of the ulnar nerve to the biceps branch of the musculocutaneous nerve (Oberlin procedure), and the long head branch of the triceps from the radial nerve to the anterior branch of the axillary nerve (Leechavengvongs procedure)⁴.

Over the past decade, the indications for NT in brachial plexus reconstruction have progressively widened: some authors now advocate transfer techniques even for root ruptures, arguing that distal transfers markedly shorten the regenerative distance and thereby accelerate motor recovery^5,6,7. To date, however, these reports have focused exclusively on post-operative muscle-strength outcomes; assessments of dexterity and coordinated motion, variables that critically determine the quality of functional recovery, remain absent.

Given that residual root stumps contain a larger proportion of sensory fibers⁸ and obviate the need for long-distance cortical remapping after reconstruction⁹, we hypothesize that, for C5–C6 post-ganglionic ruptures, NG using the viable C5–C6 root stumps as the source of regenerating axons yields a superior overall quality of functional recovery compared with NT. Because the rat is widely accepted as the ideal experimental model for brachial plexus injury and repair^10,11,12, we will establish two adult rat models of C5–C6 post-ganglionic lesions reconstructed with NG or NT, respectively. Behavioral, electrophysiological assessments, and regenerated sensory axon counts will be employed to compare the two groups and test the stated hypothesis.

Protocol

This protocol was approved by the Animal Welfare and Ethics Committee of the Laboratory Animal Science Department, Fudan University (Approval No.: 2019 Huashan Hospital JS-008).

Establishment of the rat model
Left-sided upper-trunk rupture injury
Room temperature was maintained at 23–25 °C. Sprague-Dawley rats were anesthetized with intraperitoneal 1% sodium pentobarbital (0.4 mL/ 100 g body weight). After induction of anesthesia, the rat was placed in the supine position on the operating board with the forelimbs abducted 90° from the trunk. The neck, chest, axilla, and, in NG animals only, both hind limbs were shaved. The skin was disinfected with povidone-iodine. A 3 cm midline cervical incision was made from the cricoid cartilage to the superior border of the manubrium sterni. Under an operating microscope, the subcutaneous tissue and deep fascia were dissected layer by layer to expose the sternohyoid, sternomastoid, pectoralis major, and pectoralis minor muscles. These four muscles were retracted using sterile, custom-made paper-clip retractors. The distal ends of the retractors were fixed to the operating board with rubber bands, and the tension was adjusted to obtain optimal exposure while protecting the internal jugular vein. The anterior and middle scalene muscles were identified, and the supraclavicular portion of the brachial plexus in the inter-scalene space was exposed (Figure 1A). The C5 and C6 roots were transected 5 mm distal to their respective intervertebral foramina, then a 5 mm segment was resected to create a defined gap (Figure 1B). In the NT group, the proximal stumps of C5 and C6 were sutured to adjacent soft tissue to prevent regenerating axons from entering the distal pathway.

Establishment of the NG model
An L-shaped incision was made along the dorsolateral aspect of the hind paw, across the lateral malleolus, and up the lateral aspect of the lower leg. Under an operating microscope, the subcutaneous tissue and deep fascia were dissected layer by layer, taking care to protect the small saphenous vein, then the sural nerve was exposed and harvested. The bilateral sural nerves yielded a total length of 8 cm. After trimming the perineural connective tissue, each nerve was divided into eight 1-cm segments. The suprascapular nerve was transected at its origin from the upper trunk. One graft was used to bridge from C5 to the suprascapular nerve; three grafts from C5 and four grafts from C6 to the upper trunk (Figure 1C).

Establishment of the NT model
The midline cervical incision was extended approximately 2.5 cm distally toward the upper arm.  At the point where the lateral pectoral cutaneous branch emerges, the pectoralis major muscle was incised along the fiber direction between the pectoralis major and the biceps. The deeper pectoralis minor was bluntly dissected to expose the brachial plexus cords running beneath the pectoralis major. The ulnar nerve arises from the medial cord and courses most medially within the cord segment (Figure 1D). The musculocutaneous nerve originates from the lateral cord, passes under the lower border of the teres major into the arm, and finally enters the biceps brachii muscle belly, running most laterally within the cord segment (Figure 1D). The radial nerve originates from the posterior cord, passes under the lower border of the latissimus dorsi tendon into the posterior arm, and gives off three branches supplying the long, medial, and lateral heads of the triceps. The axillary nerve also arises from the posterior cord, traverses the quadrangular space, and divides into anterior and posterior branches that innervate the deltoid muscle.

Spinal accessory nerve transfer to the suprascapular nerve
The sternocleidomastoid muscle was retracted within the midline cervical incision to expose the spinal accessory nerve. The nerve emerges from beneath the digastric muscle, reaches the middle third of the sternocleidomastoid medial border, gives off the sternocleidomastoid branch, continues along the posterior border of that muscle, and descends laterally in the neck to innervate the trapezius. The sternocleidomastoid branch was transected to eliminate traction on the accessory nerve that could otherwise disrupt the coaptation site after transfer. The trapezius branch was divided at its entry point into the muscle. The suprascapular nerve was transected at its origin distal to the upper trunk. The trapezius branch of the accessory nerve was coapted to the suprascapular nerve with a single 10-0 synthetic polypropylene suture (Figure 1E).

Ulnar nerve fascicle transfer to the biceps motor branch of the musculocutaneous nerve (Oberlin procedure)
The main trunk of the musculocutaneous nerve was mobilized distally until the entire biceps motor branch was fully exposed. At the level of the middle third of the biceps motor branch, a 6 mm fascicle from the lateral aspect of the ulnar nerve was harvested, and its distal end was sharply transected. The biceps motor branch was transected at its origin from the musculocutaneous nerve trunk. The mobilized ulnar fascicle was coapted to the biceps motor branch with a single 10-0 synthetic polypropylene suture. Complete exposure of the biceps motor branch serves two purposes: (1) to identify the exact site for fascicle harvest from the ulnar nerve, and (2) to ensure a tension-free coaptation after transfer (Figure 1F).

Transfer of the radial long-head triceps branch to the anterior branch of the axillary nerve (Leechavengvongs procedure)
The rat was repositioned in the prone position with the forelimb abducted 90° to the trunk. The dorsal aspect of the left upper arm was shaved and disinfected. A 3.5 cm longitudinal incision was made on the dorsal aspect of the left upper arm, extending from the shoulder joint to the mid-humeral level. Under the operating microscope, the subcutaneous tissue and deep fascia were dissected layer by layer. The deltoid muscle was retracted to expose the underlying quadrangular space, where the main trunk of the axillary nerve and the accompanying posterior circumflex humeral vessels were seen exiting the space. The axillary nerve was mobilized distally until the entire anterior and posterior divisions were fully exposed just before they entered the deltoid muscle belly. Within the same incision, the main trunk of the radial nerve was identified between the long and lateral heads of the triceps. The radial nerve was mobilized distally until the entire motor branch to the long head of the triceps was clearly visualized (Figure 1G). The long-head triceps branch was transected at its entry into the muscle, and it was reflected proximally to approximate the anterior branch of the axillary nerve. The anterior branch of the axillary nerve was transected at its origin from the main trunk, and it was reflected distally. The long-head triceps branch to the anterior branch of the axillary nerve was coapted with a single 10-0 synthetic polypropylene suture (Figure 1H).

After completion of all nerve reconstructions, each surgical field was irrigated with sterile saline, and every coaptation site was meticulously inspected to confirm integrity. All incisions were closed in layers: the pectoralis major muscle was approximated with 5-0 non-absorbable silk sutures (required only in the cervical incision of the NT group), followed by the subcutaneous fascia and skin. Because every nerve coaptation was performed without tension, post-operative upper-limb immobilization was unnecessary.

Behavioral assessment
Ochiai score¹³
The rat was lifted by the tail so that its body was suspended 30 cm above the table, and the posture of both forelimbs was observed. The rat was placed in a confined runway (50 cm long × 10 cm wide) on the table to observe its forelimb gait while walking. The entire runway was covered with white paper, and the starting end was moistened with black ink. As the rat walks forward, clear footprints are obtained. 0 points – when suspended, both forelimbs hang straight toward the floor in a relaxed manner (Figure 2A-1); during walking, neither forelimb is dragged (Figure 2A-2). 1 point – when suspended, the affected forelimb is slightly adducted and flexed (Figure 2B-1); while walking, mild dragging due to limited elbow flexion is seen (Figure 2B-2). 2 points – when suspended, the affected forelimb is fully flexed and adducted (Figure 2C-1); during walking, severe dragging is evident (Figure 2C-2).

Barth foot-fault test¹³
A horizontal metal grid (50 cm × 38 cm) was prepared elevated 11 cm above the table; each grid square measures 1.5 cm. The rat was placed on the grid; whenever the affected forelimb slips through a grid opening, it is recorded as a "foot fault" (Figure 3). The number of foot faults made by the affected limb in 2 min was counted.

Terzis grooming test¹⁴
The rat was placed in an open-top transparent observation box. Approximately 1–3 mL of normal saline was sprayed onto the rat's snout and face, and its attempts to remove the water droplets were observed.

Grade 0 – no response from the affected limb.
Grade 1 – the affected limb flexes at the elbow and reaches the mouth, but cannot reach the nose (Figure 4A).
Grade 2 – the affected limb reaches the nose (Figure 4B).
Grade 3 – the affected limb reaches above the nose but cannot reach the eye (Figure 4C).
Grade 4 – the affected limb reaches the eye but cannot reach the ear (Figure 4D).
Grade 5 – the affected limb reaches the ear or behind it (Figure 4E).

Compound muscle action potential (CMAP)
The rat was anesthetized, immobilized, and prepared as described previously in the protocol, and the supraclavicular portion and cord segment of the brachial plexus were exposed. The black recording needle electrode was inserted into the belly of the target muscle (biceps brachii, deltoid, or infraspinatus). The red recording electrode was inserted into the tendon-muscle junction of the same muscle. The green reference electrode was placed between the black and red electrodes (Figure 5). All electrodes are inserted to a depth of 2 mm, and identical insertion sites are used on both sides. The nerve was gently elevated with the stimulating electrode for testing. A single square-wave pulse of 0.5 mA intensity and 0.2 ms duration was delivered; CMAP latency and amplitude were recorded. The contralateral (healthy) side was used as the reference (assigned a value of 1). The latency delay rate ([latency on affected side/latency on healthy side] × 100 %) and the amplitude recovery rate ([amplitude on affected side/amplitude on healthy side] × 100 %) were calculated.

Immunofluorescence Staining
Myelinated fibers were labeled with anti-neurofilament 200 (NF200); among them, motor myelinated fibers were distinguished by co-labeling with anti-choline acetyltransferase (ChAT). Fibers positive for NF200 but negative for ChAT were therefore classified as sensory myelinated fibers¹⁵.

For the nerve-graft group, the main trunk of the musculocutaneous, axillary, and suprascapular nerves was harvested distal to the second suture site on the affected limb. For the nerve-transfer group, the biceps branch of the musculocutaneous nerve, the anterior branch of the axillary nerve, and the suprascapular nerve were taken distal to their respective coaptation sites. A 3-mm segment of each nerve was immersed in 4% paraformaldehyde and fixed at 4 °C for 24 h.

The specimens were oriented perpendicular to the longitudinal axis, embedded in paraffin, and serially sectioned at 3 µm. After deparaffinization, antigen retrieval was performed in 0.01 M EDTA buffer (pH 8.0) using a microwave oven. Sections were incubated overnight at 4 °C with primary antibodies diluted in PBS (ChAT, 1:200; NF200, 1:500), followed by PBS washes and 50-min incubation at room temperature with secondary antibodies (Cy3-conjugated goat anti-rabbit, 1:300; Alexa Fluor 488-conjugated goat anti-mouse, 1:400) protected from light. Nuclei were counterstained with DAPI. Whole cross-sections of each nerve were imaged under a fluorescence microscope equipped with a slide scanner.

NF200 was visualized with green fluorescence and ChAT with red fluorescence; NF200-positive/ChAT-negative profiles were counted as sensory myelinated fibers. Five randomly selected sections from each nerve were analyzed with ImageJ. The total number of sensory myelinated fibers in the entire cross-section was counted for every section, and the mean value of the five sections was taken as the final count for that nerve specimen.

Statistical analysis
Sample size (n = 8 per group) was prospectively calculated with G*Power using the sensory-axon-count pilot effect size (Cohen's d = 2.31) to yield ≥ 80 % power at α = 0.05. All statistical analyses were performed using GraphPad Prism. Data are presented as median (interquartile range) [M (p25–p75)]. Between-group comparisons were carried out with the Mann-Whitney U test. A P value < 0.05 was considered statistically significant.

Results

On post-operative day 1, both NG and NT rats exhibited complete loss of shoulder abduction and elbow flexion in the operated limb; when the dorsal aspect of the affected forelimb was pricked while the animal was restrained, neither group showed withdrawal or vocalization. At 8 weeks, shoulder and elbow function had improved in both groups; the same stimulus now elicited withdrawal. By 16 weeks, further—though still incomplete—recovery was observed. When the dorsal forelimb was pricked, NT rats struggled and squealed, whereas NG rats struggled more vigorously and emitted louder squeaks.

At 8 and 16 weeks, Ochiai scores did not differ between NG and NT (both P > 0.05). Foot-fault counts during the 2-min test were similar at 8 weeks (P > 0.05), but at 16 weeks, NG rats made significantly fewer faults than NT rats (P < 0.05). Terzis' grooming scores were statistically indistinguishable at both time points (both P > 0.05, Table 1).

At 8 weeks, latency delay and amplitude recovery rates for the musculocutaneous (biceps branch), axillary (anterior branch), and suprascapular nerves were significantly better in NT than in NG (all P < 0.05). By 16 weeks, however, these electrophysiological differences had disappeared (all P > 0.05, Table 2).

At 8 weeks, the number of regenerated sensory myelinated fibers in the musculocutaneous and axillary nerves of the nerve-graft group was significantly higher than that in the biceps branch of the musculocutaneous nerve and the anterior branch of the axillary nerve in the nerve-transfer group (both P < 0.05). The difference in the number of regenerated sensory myelinated fibers in the suprascapular nerve between the two groups was not statistically significant (P > 0.05). At 16 weeks, the number of regenerated sensory myelinated fibers in the musculocutaneous, axillary, and suprascapular nerves of the nerve-graft group was significantly higher than that in the corresponding nerves of the nerve-transfer group (all P < 0.01; Table 3, Figure 6).

DATA AVAILABILITY
All raw data supporting this study are publicly available via Zenodo at https://doi.org/10.5281/zenodo.17921733.

Figure 1: Intra-operative photographs illustrating the creation of the nerve injury and reconstruction rat model. (A) Exposure of the left whole brachial plexus at the proximal level. (B) Left upper-trunk rupture model. (C) Nerve-graft repair after left upper-trunk root injury. (D) Supine-approach exposure of the left brachial-plexus cords. (E) Transfer of the left spinal accessory nerve to the suprascapular nerve. (F) Transfer of a fascicle of the ulnar nerve to the biceps branch of the musculocutaneous nerve (Oberlin procedure). (G) Prone-approach exposure of the left brachial-plexus cords (Leechavengvongs procedure). (H) Transfer of the long-head triceps branch of the radial nerve to the anterior branch of the axillary nerve. ➤: suture site. ADUT, anterior division of upper trunk; AN, axillary nerve; b., branch; C, caudal; L, left; MN, median nerve; McN, musculocutaneous nerve; R, right; RN, radial nerve; PDUT, posterior division of upper trunk; SAN, spinal accessory nerve; SSN, suprascapular nerve; UN, ulnar nerve; UT, upper trunk. Please click here to view a larger version of this figure.

Figure 2: Ochiai score. (A) 0 points: both forelimbs hang relaxed toward the floor (A-1); no dragging during walking (A-2). (B) 1 point: the affected forelimb is slightly adducted and flexed (B-1, →); mild dragging while walking (B-2, →). (C) 2 points: the affected forelimb is fully adducted and flexed (C-1, →); severe dragging while walking (C-2, →). Please click here to view a larger version of this figure.

Figure 3: Barth foot-fault test. (A) The affected forelimb of the rat does not "fault" and successfully supports itself on the grid. (B) The affected forelimb of the rat commits a "fault" and falls through the grid (→). Please click here to view a larger version of this figure.

Figure 4: Terzis grooming test. (A) Grade 1: the affected forelimb is below the nose but can reach the mouth (→). (B) Grade 2: the affected forelimb can touch the nose (→). (C) Grade 3: the affected forelimb is above the nose but below the eye (→). (D) Grade 4: the affected forelimb can touch the eye (→). (E) Grade 5: the affected forelimb can touch the ear (→). Please click here to view a larger version of this figure.

Figure 5: Compound muscle action potential recording. Using the right biceps brachii as an example, the three recording electrodes (→) are inserted into the muscle belly, the myotendinous junction, and the interval between them, while the stimulating electrode (➤) gently elevates the musculocutaneous nerve. H, head; R, right. Please click here to view a larger version of this figure.

Figure 6: Immunolabeling of myelinated fibers in the suprascapular nerve. (A–L) Immunolabeling of myelinated fibers. Myelinated fibers appear green after NF200 labeling (A, D, G, J). Motor myelinated fibers are red after ChAT labeling (B, E, H, K). Following double-labeling (C, F, I, L), motor fibers appear yellow because red and green signals overlap, whereas sensory myelinated fibers remain green (C, F, I, L). At 8 weeks, the total number of myelinated fibers and the number of motor myelinated fibers distal to the second suture site in the nerve-graft group (A, B) were lower than those in the nerve-transfer group (D, E); however, the number of sensory myelinated fibers did not differ noticeably between the two groups (C, F). At 16 weeks, the nerve-graft group (G, H, I) exhibited greater numbers of total myelinated fibers, motor myelinated fibers, and sensory myelinated fibers than the nerve-transfer group (J, K, L). NG, nerve graft; NT, nerve transfer. Scale bar = 20 µm. Please click here to view a larger version of this figure.

		NG	NT	Mann-Whitney U	P
Ochiai score	8 weeks	1.00 (1.00–2.00)	1.00 (0.75–1.00)	22.5	0.3888
	16 weeks	0.00 (0.00–0.25)	0.50 (0.00–1.00)	23	0.4452
Barth foot-fault test	8 weeks	10.50 (9.00–12.00)	9.00 (8.00–10.00)	16.5	0.1136
	16 weeks	5.50 (4.00–6.00)	7.00 (6.00–8.25)	12	0.0378
Terzis grooming test	8 weeks	2.00 (2.00–3.00)	2.50 (2.00–3.00)	30.5	0.9347
	16 weeks	4.00 (4.00–5.00)	3.00 (3.00–4.00)	15	0.0816

Table 1: Behavioral outcomes in the nerve-graft (NG) and nerve-transfer (NT) groups (n = 8).

		NG	NT	Mann-Whitney U	P
McN latency delay rate(%)	8 weeks	120.56 (118.21–128.69)	115.28 (110.57–118.78)	10	0.0207
	16 weeks	110.12 (106.27–112.88)	105.75 (104.07–108.80)	16	0.1049
McN amplitude recovery rate(%)	8 weeks	87.91 (87.31–89.04)	91.08 (89.87–92.58)	5	0.003
	16 weeks	92.88 (91.77–95.31)	95.63 (94.05–96.55)	15	0.083
AN latency delay rate(%)	8 weeks	123.84 (122.59–125.41)	121.62 (120.45–123.34)	12	0.0379
	16 weeks	103.20 (102.29–104.04)	102.45 (101.99–103.93)	26	0.5737
AN amplitude recovery rate(%)	8 weeks	81.34 (79.90–82.06)	85.05 (83.56–85.74)	5	0.003
	16 weeks	92.31 (91.82–93.69)	94.16 (92.61–95.65)	19	0.1851
SSN latency delay rate(%)	8 weeks	117.00 (115.72–118.98)	113.80 (112.63–114.80)	8	0.0104
	16 weeks	107.39 (106.49–108.74)	108.01 (106.37–109.03)	27.5	0.6657
SSN amplitude recovery rate(%)	8 weeks	82.32 (81.05–82.99)	85.18 (82.84–86.21)	12	0.0379
	16 weeks	92.84 (91.74–93.58)	91.48 (90.51–92.47)	21	0.2786

Table 2: Compound muscle action potential findings in the nerve-graft (NG) and nerve-transfer (NT) groups (n = 8).

		NG	NT	Mann-Whitney U	P
McN	8 weeks	1162 (1077–1200)	1054 (978–1105)	12	0.0379
	16 weeks	1606 (1515–1692)	1352 (1279–1439)	3	0.0011
AN	8 weeks	904 (860–954)	797 (755–855)	7	0.007
	16 weeks	1308 (1250–1348)	1047 (989–1106)	0	0.0002
SSN	8 weeks	501 (484–528)	503 (476–525)	30	0.8785
	16 weeks	833 (802–887)	691 (648–752)	3	0.0011

Table 3: Number of regenerated sensory myelinated fibers in the nerve-graft (NG) and nerve-transfer (NT) groups (n = 8).

Discussion

The objective of this study was to test our hypothesis that reconstruction of brachial plexus injuries using residual nerve-root grafting yields an overall superior outcome compared with nerve transfer. Adult rats with C5–C6 ruptures were divided into two groups: one underwent upper-trunk reconstruction via nerve grafting, and the other via transfers of the spinal accessory nerve, the long-head triceps branch of the radial nerve, and a fascicle of the ulnar nerve. Electrophysiologically, the nerve-graft group was not superior to the nerve-transfer group; in fact, early recovery in the graft group lagged behind that in the transfer group, because the regenerating axons in nerve grafting must traverse additional coaptation sites. By 16 weeks post-operatively, however, the nerve-graft animals performed significantly better on the Barth foot-fault test. These findings indicate that, although muscle strength recovery did not differ between the two strategies, the coordination of movements recovered after nerve grafting surpassed that achieved after nerve transfer.

In the present study, three classic behavioral assays were employed to obtain a comprehensive evaluation of overall forelimb function in the rat model. The Ochiai score primarily assesses the impact of limb impairment on gait pattern and stride length during locomotion¹³. The Barth foot-fault test quantifies aberrant movements of the affected limb and provides a numerical index of proximal forelimb muscle performance; it is essentially a sensorimotor task that relies on sensory feedback¹⁶. The Terzis grooming test evaluates the integrated function of shoulder elevation, abduction, and elbow flexion, while simultaneously allowing comparison with the contralateral intact limb¹⁴. At 8 weeks post-operatively, no significant differences were detected between the two groups in any of the three tests. By 16 weeks, the Ochiai scores and Terzis grooming grades remained statistically similar; however, the nerve-graft group outperformed the nerve-transfer group in the Barth foot-fault test. Thus, despite comparable compound muscle action potentials, the grafted animals exhibited superior coordination and dexterity. This suggests that the independence of the recovered movements—i.e., the extent to which their initiation and maintenance are free from the assistance of unrelated movements—is better in the graft group than in the transfer group. Collectively, these findings indicate that the overall quality of functional recovery is higher following nerve grafting than following nerve transfer.

Sensory nerve fibers are classified as unmyelinated or myelinated. Unmyelinated fibers conduct impulses slowly and mainly transmit nociceptive, thermal, and tactile-pressure information. Myelinated fibers not only convey nociceptive, thermal, and tactile-pressure sensation, but also innervate joint-capsule mechanoreceptors, Golgi tendon organ stretch receptors, and muscle-spindle proprioceptors. Owing to the marked increase in conduction velocity conferred by myelin, myelinated fibers constitute the primary pathway for sensory feedback that modulates motor function¹⁷. With respect to the number of regenerating axons available after conventional upper-trunk reconstruction, C5 and C6 contain 38,210 and 75,861 myelinated fibers, respectively⁸, whereas the spinal accessory nerve contains 4,239⁸, a fascicle of the ulnar nerve 1,318¹⁸, and the long-head triceps branch of the radial nerve 2,302¹⁹. Thus, a single root provides far more myelinated fibers than the combined total of these three donor nerves. Theoretically, therefore, sensory recovery should be superior after root-grafting. In this study, we found that at 8 weeks postoperatively, most evaluations, and at 16 weeks, all evaluations, showed significantly more sensory myelinated fibers in the nerve-graft group than in the nerve-transfer group. We therefore infer that, compared with the nerve-transfer group, the nerve-graft group had a larger population of sensory myelinated fibers contributing to motor-function modulation. This advantage clearly underlies its superior overall functional recovery, especially in behavioral tasks reliant on sensory feedback such as the Barth foot-fault test.

Following peripheral nerve injury and repair, the central nervous system undergoes extensive functional reorganization²⁰. In the early phase, cortical territory originally corresponding to the injured nerve is encroached upon by adjacent areas that subserve other peripheral nerves²¹. Once target reinnervation is achieved, axonal sprouting from the donor cortical representation into the recipient territory establishes novel synaptic connections²². The extent to which the donor cortex can "reclaim" the previously lost territory critically depends on whether the repaired nerve regains independent function—that is, whether movements under its control can be initiated and maintained without the assistance of unrelated muscle groups^23,24. Socolovsky et al.⁹ investigated the relationship between independent functional recovery and cortical plasticity after brachial-plexus nerve transfers. They found that the Oberlin procedure outperformed transfers using the contralateral C7, intercostal, or phrenic nerves because the somatotopic distance between the cortical representations of the ulnar and musculocutaneous nerves is markedly shorter than that between the latter and the alternative donor nerves. This proximity accelerates the establishment of new cortical pathways and facilitates the reclamation of the lost cortical territory. Their findings indicate that the shorter the cortical distance required for post-operative reorganization, the more complete the recovery of independent function. After nerve grafting, cortical reorganization is largely confined to the original territory of the injured nerve, whereas nerve transfer necessitates the construction of a pathway between two distinct cortical representations. Consequently, the former is theoretically more conducive to efficient post-operative cortical plasticity.

This study has several limitations. First, the root transection performed in our model differs from the pathomechanism of clinical root ruptures, where the traumatic force is often more severe and may obscure the exact level of injury. Consequently, the experimental findings cannot be directly extrapolated to the clinical setting. Second, while the 16-week follow-up revealed superior motor coordination in the NG group, this represents a relatively short-term outcome in the rat lifespan. In clinical practice, patients undergoing nerve transfer often require a prolonged period of cortical re-learning to achieve optimal motor control, even after initial muscle reinnervation. Therefore, longer-term follow-up in animal models might reveal different comparative functional outcomes, potentially showing continued improvement in the NT group as cortical plasticity evolves. Future studies with extended observation periods are necessary to address this temporal factor. Third, we did not directly assess neural plasticity in the two surgical models. Therefore, the mechanisms underlying the superior motor coordination observed in the nerve-graft group remain to be elucidated by further investigations.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1× Citrate Antigen Retrieval Solution	Servicebio	G1219-1L
1× Tris-EDTA Antigen Retrieval Solution(ph 8.0)	Servicebio	G1207-1L
1× Tris-EDTA Antigen Retrieval Solution(ph 9.0)	Servicebio	G1218-1L
10-0 PROLENE	Johnson&Johnson Med Tech	W2790
5-0 MERSILK	Johnson&Johnson Med Tech	SA82G
Alexa Fluor 488-conjugated Goat Anti-Mouse IgG (H+L)	Servicebio	GB25301
Anti-160 kDa Neurofilament Medium Mouse mAb	Servicebio	GB12763-100
Anti-Choline Acetyltransferase Rabbit pAb	Servicebio	GB11070-1-100
Antifluorescence Quencher Coating Agent	Servicebio	G1401
Bovine Serum Albumin	Servicebio	GC305010-5g
ChAT	Servicebio	GB11070-1
Cy3 conjugated Goat Anti-Rabbit IgG (H+L)	Servicebio	GB21303
DAPI	Servicebio	G1012-10
Embedding machine	GET Electronics (Hui ZHOU) Co., Ltd.	JB-P5
Fluorescence microscope	Nikon	Eclipse C1
Four-channel electromyograph	Dantes	DT440
G*Power	Heinrich Heine University Düsseldorf	3.1.9.7
High-Speed Camera Controller	Nikon	DS-U3
ImageJ	National Institutes of Health	1.8.0
Microsurgical instruments	Ningbo Chenghe Micro-Instrument Factory	6804
Microwave oven	Galanz	P70D20TL-P4
Operating microscope	Shanghai Medical Optical Instrument Factory	SXP-1B
Pipette	Eppendorf	100221588457
Prism	GraphPad Software	10.1.2
Rotary Microtome	Leica	RM2016
Scanner	Epson	V300
Sprague Dawley Rat	Shanghai Sippe-Bk Lab Animal	OR-1001
Tissue Autofluorescence Quencher	Servicebio	G1221-5ML
Tissue Marking Pen‌	Gene Tech	GT1001