A severed ventral root—without an avulsion injury—is a peripheral nerve injury with good possibilities to regenerate (11). A severed dorsal root—without an avulsion injury—is considered a type of central nervous injury as sensory nerve fibre regrowth cannot occur across the root–spinal cord junction or the PNS–CNS transitional region. Because of this, it is considered the most common spinal cord injury (SCI) (12).

Root avulsions interrupt transverse segmental nerve fibres in the spinal cord, causing a “longitudinal spinal cord injury.” If left untreated, the affected spinal cord segments can deteriorate within approximately 1 month. There is the disintegration of neuronal networks and the death of motor (13), secondary sensory (14), and autonomic (15) neurons, leading to paralysis, loss of sensation, and excruciating, almost unbearable, severe intractable pain (16).

The diagnosis of a complete intraspinal nerve plexus avulsion injury by clinical and/or auxiliary means is difficult. Lesions can be situated simultaneously at multiple levels in both the spinal root and the peripheral nerve. There can be a mixture of motor and sensory root injuries with continuity severed in one or the other. Clinical assessment and neurophysiological testing can lead to erroneous diagnoses. Even the most sensitive MRI techniques provide false results in approximately 50% of cases. Computed tomography (CT) myelogram is still considered to be the gold standard for the preoperative evaluation of root avulsion (10).

Intraspinal endoscopy can provide more information regarding the remaining intradural stumps or full root avulsions (10). Direct inspection of the subdural space after laminectomy gives an accurate diagnosis of the extent of the root rupture or avulsion (17).

In a long series of animal experiments, a direct, curative surgical technique for nerve root avulsions for clinical application in spinal cord surgeries was developed.

An original observation was made that spinal cord motoneurons, after ventral root avulsion and implantation, are reintegrated into local spinal cord reflex circuits and are able to regenerate and reinnervate muscles. Motoneurons have been observed by means of intracellular neurophysiology and electron microscopy to respond to injury by extending axons and dendrites (dendraxons) (18, 19).

Sensory function cannot be regained by implanting dorsal roots into the spinal cord. Dorsal root ganglia (DRG) neurons are unable to regrow axons into the CNS unless adjuvant molecules have been used (20). Like spinal cord motoneurons, spinal cord secondary sensory neurons can elongate new processes after injury. Provoked by an implanted PNS conduit, they can extend axons into a nerve graft after dorsal root ganglionectomy (10). Such recruited secondary sensory spinal cord neurons, many located in the lateral spinal nucleus (LSN), have connections for proprioception. MRI tractography demonstrated new neurite growth from the LSN into the periphery of importance for proprioception, as shown by electrophysiology (21). These findings prompted further experimental surgery and, eventually, human clinical application (10).

A surgical approach in the lateral triangle of the neck and the cervical spine allows access to both the intra- and extraspinal parts of the brachial plexus [for surgical details, see (22)]. The intraspinal parts of the lumbosacral plexus and the cauda equina are exposed with the patient in the prone position [see (23)].

Medullary implantation of peripheral nerve grafts or reimplantation of roots into the spinal cord is performed through small slits in the pia mater in the ventro-lateral or dorso-lateral aspect of the spinal cord (17). The superficial position of the implanted nerve or root just below the pia mater is retained by means of tissue glue. During this procedure, spinal cord monitoring is performed to avoid any medullary dysfunction caused by the spinal cord surgery. If performed within 1 month after the injury, this surgery can promote the regrowth of new motor, sensory, or autonomic axons through spinal cord tissue (CNS regeneration). This regenerative effect is crucial for restoring nerve fibre trajectories and, hence, for the possible recovery of function and alleviation of pain (24). Reimplantation into the spinal cord, however, should not be considered in cases later than 1 month after injury (25).

Longitudinal assessments of restored activity were performed by means of functional MRI, transcranial magnetic stimulation, electromyography, and quantitative sensory testing of all sensory modalities, including sudomotor function. Recovery of motor function after spinal cord replantation occurs in a proximodistal direction and is strongest in the shoulder, upper arm muscles, and hip and thigh muscles after repair of intradural lumbar nerve roots (10). This postoperative functional gain in daily life refers to the arm being used as a prop for bilateral activities and the leg for standing and independent locomotion (10). The best recovery was noted in preadolescent children. These patients showed useful distal return of independent function in the forearm and hand (26). Functional outcomes after surgery can be misinterpreted and attributed to other conditions, such as plasticity, collateral sprouting, or anatomical variances, rather than from spinal cord regeneration, even when a complete brachial plexus avulsion injury had been verified at surgery. Prefixed or postfixed plexus anomalies are rare and would be unlikely to affect the outcome in a case where the entire brachial plexus is dislodged from the cervical spine and pulled underneath the clavicle. Experimental studies on collateral sprouting from adjacent intact nerves have failed to demonstrate convincing growth following experimental injuries (27).

To rule out sprouting in humans, recovery of function was assessed 3 years after spinal cord root implantation surgery by CT-guided root block of neighbouring non-avulsed roots. This study verified new axon growth through the spinal cord and into the root implants (8).

The central nervous system can become structurally and functionally reorganised from a severe brachial plexus injury and spinal cord surgery. After ventral root to spinal cord replantation, functional MRI (fMRI) revealed expected motor cortex activity in the contralateral hemisphere of the operated limb. When the affected arm was used, activity was observed also on the ipsilateral side (Figure 1), which was not seen when the normal (intact) side was used (26). The ipsilateral cortex activity is most likely to be the result of cortical plasticity caused by post-injury diminished transcallosal inhibition (28), as has been noted in some CNS disorders (29). Only contralateral motor cortex plasticity has previously been observed on fMRI after peripheral nerve surgery (30). fMRI results also revealed that activity was observed in sensory cortex areas when the affected arm was used (9), despite the absence of sensory function. This probably occurs through pre-existing cortical sensory control in the associated sensory cortex areas (31) or from activation of some motoneurons located in the sensory cortex area. Spontaneous contractions of limb muscles in synchrony with respiration, i.e., “the breathing arm,” were noted in some operated cases after spinal cord repair of root avulsions. Such synkinesis was observed only during inspiration and verified by electromyography (EMG). The sources of this activity are the phrenic nerve motor neurons, which are situated at the C3–C5 spinal cord segments, in a discrete nucleus in the most medial part of the ventral horn, adjacent to the motoneurons supplying the shoulder and upper part of the arm. The implantation of a PNS conduit into the ventral part of the C-5 spinal cord segment could thus attract neurite growth from the phrenic nerve motor neurons within this segment to the arm instead of the diaphragm. This phenomenon demonstrates the intramedullary and spinal-cord-to-peripheral-nerve neurite growth produced by this surgery (32).

The quality of muscle function in many operated cases is disturbed by muscle co-contractions and synkinesis, probably due to absent sensory feedback and ineffective proprioception. In cases subjected to medullary connection of sensory neurons to the periphery, patients showed more controlled movements without the co-contractions and synkinesis only after the motor conduits had been repaired. Proprioception can be demonstrated by a bicep tendon reflex and electrophysiological demonstration of an H or Hoffman reflex (17). Functional MRI showed extended sensory motor cortical activity during active and passive arm movements in the primary somatosensory cortex S1 (9). The patient had perception of their movements but, in contrast to proprioception, there was no exteroception that could be demonstrated by objective sensory or cortical registration of contact evoked potential stimulation (9).

Dorsal root avulsion often generates spontaneous intractable deafferentation pain that parallels the generation of abnormal activity within the dorsal horn of the spinal cord. The magnitude of this pain is related to the number of avulsed roots (33). Pain provoked by light touch or allodynia experienced at the border zone of avulsed and non-avulsed dermatomes is probably due to reorganisation in the dorsal horn, for instance, inhibition of spinal afferent input, leading to spinal cord hyper-excitability (34). Reactive gliosis due to microglia and astrocyte activation (35) and metabolic changes, as revealed by proton magnetic resonance spectroscopy (36), in affected and neighbouring spinal cord segments could also cause the avulsion-induced pain. Relief of pain was frequently reported after timely repair of avulsed roots with reimplantation (33). Interestingly, alleviation of pain was also noted in conjunction with the return of motor function, probably due to the re-establishment of inhibitory cortical connections (26).

A majority of patients with dorsal root avulsions reported a referral of sensation that was perceived to be abnormal or at remote sites. Reorganisation of the somatosensory cortex by expansion of the adjacent sensory area into the deafferented cortex or in the thalamus may explain referred touch sensation from the face to the denervated arm (37). Sensation perceived in the deafferented arm when the trunk or the face is touched has been termed “wrong way of referral of sensation.” This type of sensation is considered to be related to plasticity in the cortex and is different from the “right way” of referred sensation, which can occur following peripheral nerve regeneration and is perceived at the origin of a lesioned nerve (33).

Surgery on spinal cord injuries sustained in conjunction with an intraspinal nerve injury has had promising results, demonstrating exceptional functional outcomes resulting from nerve fibre growth in the CNS (8). However, this recovery is limited and often disturbed by misdirected and erroneous connections caused by both the injury and surgery (33, 38). Restitution is particularly limited in the sensory system, occasionally leading to impeded motor functions, which has led to the current provisional method of implanting ganglionectomised dorsal roots. This outcome is, however, the best possible with surgery alone.

There is a need for non-surgical or adjuvant treatments to ameliorate outcomes after a spinal root avulsion in a longitudinal spinal cord lesion and in the treatment of a classical transverse spinal cord injury (39).

The development of retinoids as therapeutics has been curtailed because most hit compounds have poor drug-like properties (56, 57), and the few lead compounds that progress to toxicity studies exhibit adverse events (AEs), the majority of which are related to liver toxicity due to their accumulation as fat-soluble molecules (42). Of the six retinoids that have been used in clinical settings, four are used to treat acne. These are the gel formulations adapalene and tretinoin and two orally available retinoids, isotretinoin and acitretin (42). As for the other two, tamibarotene is used to treat acute promyelocytic leukaemia and bexarotene, the only RXR agonist, to treat cutaneous T-cell lymphoma (42). What is also apparent about these drugs is that they are non-selective for RARβ, which, when tested in models of CNS injury, means they have no clear benefit (39).

The challenge then was to develop a selective RARβ agonist with good drug-like properties. From a hit to a lead compound, and following a lead-optimisation campaign, KCL-286 (C286) was developed. It fulfils key drug-development criteria, including oral availability, a clean profile in 28-day toxicity studies at the no-adverse-effect level (NOAEL) in rodent proof-of-concept (POC) nerve-injury studies, and good solubility and brain penetration (57). A phase I study in healthy male participants showed that the drug engaged its receptor and was well tolerated, with no drug-driven AEs at a 100-mg daily dose, equivalent to a 3-mg/kg oral dose in rodent POC studies on SCI (58).

In rodent models of avulsion repair, in which severed sensory roots are reinserted into the spinal cord, treatment with KCL-286 administered orally every other day for 28 days (starting 1 day post-injury) successfully induced axons to form the correct connections within the dorsal horn (57). This functional reinnervation contrasts with human clinical outcomes, where sensory root implantation alone has failed to achieve similar restorative results (33, 38). The drug also resolved neuropathic pain in a spinal nerve ligation model (59). This efficacy stands in direct contrast to global RARα signalling, which has been identified as a mediator that increases neuropathic pain (60), thereby highlighting the necessity for specific RARβ agonists to treat nerve injuries. Furthermore, when using the same treatment regimen established for avulsion repair, C286 successfully prevented neuronal cell death and facilitated functional recovery in rodent models of spinal cord contusion, demonstrating its broader therapeutic applicability in various spinal cord injuries (39).

As predicted, KCL-286 is multifactorial for nerve repair (Figure 2). It modulates the extracellular matrix, such as chondroitin sulphate proteoglycans, to create a permissive growth environment by removing the glial scar (39, 61). Moreover, it regulates myelination (61), reduces inflammation (39), and regulates the mitochondrial dynamics required to extend neurites (62). A pathway analysis of injured rodent spinal cords treated with KCL-286 revealed a multimodal regulatory effect across both neuronal and non-neuronal domains (39). These pathways have been identified in other studies of nerve repair, and include modulation of nicotinic receptors (63, 64), regulation of phagocytosis (65) and the orchestration of immune responses essential for axonal pathfinding (66, 67). A significant therapeutic advantage of this signalling is the initiation of an RARβ2-driven feedback loop, which sustains the transcriptional programmes necessary for axonal regeneration (20, 61).

Therefore, KCL-286 fulfils the criteria of a master regulator for nerve regeneration and may also be relevant to the treatment of neurodegenerative diseases, including Parkinson's disease, Alzheimer's disease, and motoneuron disease, given the correlation between RARβ expression and the affected neurons (39).