An expert synthesis of peripheral nerve injury — the biology of degeneration and regeneration, the Seddon and Sunderland classifications, electrodiagnostic and ultrasound assessment, timing, direct repair, the management of gaps, nerve transfers, prognosis and rehabilitation — for FESSH candidates and consultant peers.
Peripheral nerve injury remains the reconstructive problem in which our understanding of the biology has run furthest ahead of our results. Fewer than half of patients who undergo nerve repair regain good motor or sensory function, and the surgical techniques in routine use are recognisably those Sunderland described more than sixty years ago1. Most traumatic nerve injuries are in the upper limb, and incomplete injuries are the most frequent23. This article follows the sequence a candidate is examined on: what happens to the nerve, how the injury is classified and assessed, when and how it is repaired, what to do about a gap, when a nerve transfer is the better operation, and what determines the outcome.

§ 01 · ANATOMY AND PHYSIOLOGY

The connective-tissue framework

Each axon is enclosed by ; the perineurium bundles axons into fascicles; and the epineurium, with an internal component between fascicles and an external component defining the nerve trunk, surrounds the whole45. The external epineurium is the layer that holds sutures, which is why epineurial repair is the default technique1. These sheaths give the nerve its mechanical tolerance to the tensile, shear and compressive stresses of limb movement, producing excursion, strain and transverse contraction rather than injury under physiological loads6.

Degeneration after injury

When an axon is interrupted, the segment distal to the lesion loses contact with the cell body and undergoes , beginning within 24–48 hours; the proximal segment dies back to the nearest node of Ranvier, from which regrowth later begins1. Degeneration is not passive decay but a programmed process with its own molecular machinery, including SARM1 and NMNAT2, which has made it a drug target7. Denervated Schwann cells shed myelin, proliferate, phagocytose debris and release cytokines that recruit blood-borne macrophages, which take over the bulk of clearance within days and leave the nerve once remyelination has begun8. Two practical corollaries follow. The distal stump continues to conduct action potentials for up to two days after transection, so very early conduction studies can mislead9. And the growth-supportive state of the distal Schwann cells is transient: their growth-associated gene expression "progressively fails to support axon regeneration" if axons do not arrive in time98.

Regeneration

Surviving Schwann cells line the empty endoneurial tubes in longitudinal columns, the , along which regenerating sprouts advance1. Crossing the repair site is staggered: in experimental repairs it takes about a month for all motoneurons to regenerate their axons across the suture line into the distal stump9. Thereafter axons advance at roughly 1–3 mm a day, faster in proximal segments and slower distally, a figure measured in humans by Seddon, Medawar and Smith and by Sunderland in the 1940s41011. At the wrist, a median or ulnar nerve must regenerate about 100 mm to reach many of the hand muscles, so recovery takes at least 100 days at the average rate of 1 mm a day1. Regeneration is more reliable when the endoneurial tube is intact to guide each axon back to its target; without it, sprouts are misdirected and a neuroma forms, and empty tubes shrink within about four months4. Misdirection of regenerating fibres to targets they never supplied remains one of the largest obstacles to functional recovery9. In partial injuries in which only a fifth to a third of axons are lost, recovery proceeds largely by collateral sprouting from intact neighbours over three to six months4.

Chronic axotomy and chronic denervation

The biology of delay has been dissected experimentally by Gordon's group. When the motoneuron is kept axotomised for more than three months before repair, the number of motor units it can regenerate falls to about 35 per cent of control, although enlargement of the surviving units masks the loss in force terms12. When the distal stump and muscle are kept denervated, regeneration collapses further — a mean of 15 motor units beyond six months against 137 after immediate suture — and the cause is deterioration of the intramuscular nerve sheaths rather than an inability of muscle to accept reinnervation13. Chronic denervation of the distal stump is the dominant factor, but the denervated muscle contributes independently14, and the critical time point after which regeneration becomes too poor appears to be about three months in the rat15.

End organs

Muscle atrophy begins within weeks of denervation and plateaus after about four months, by which time 60–80 per cent of muscle volume has been lost; the can maintain its integrity for up to a year, but irreversible fibrosis is established by two years14. The conventional window for functional muscle reinnervation is therefore 12–18 months, a figure that reviews describe as accepted rather than proven1. Sensory end organs — Pacinian and Meissner corpuscles and Merkel cells — survive denervation for two to three years, so late sensory repair can still be worthwhile when motor recovery is no longer realistic4.
Examination notes. Wallerian degeneration 24–48 h; distal conduction persists ~2 days; regeneration ~1 mm/day distally (1–3 mm/day overall); motor end-plate window 12–18 months, sensory receptors 2–3 years; chronic denervation of the distal stump is the main limit after delayed repair, with the denervated muscle contributing.

§ 02 · CLASSIFICATION

Seddon

Seddon's three types — , and — were first published in 1942 and 1943161718. Neurapraxia is segmental demyelination with an intact axon, usually from compression, producing a focal conduction block that resolves within about twelve weeks once remyelination is complete; axonotmesis is axonal interruption with the connective-tissue framework preserved; neurotmesis is complete transection14. Contemporary reviews restate the as the working clinical scheme3.

Sunderland

Sunderland's five degrees, published in 1951, subdivide axonotmesis by the layer of connective tissue lost19. First-degree injury is neurapraxia. In second-degree injury the axon is damaged but the endoneurium is intact, so recovery is complete; third-degree injury adds endoneurial disruption and fourth-degree perineurial disruption, with progressively more scarring and incomplete recovery; fifth-degree injury is neurotmesis41. Fourth-degree injuries usually require surgery because the endoneurium and perineurium are disrupted and the scarred segment blocks regeneration14. The therefore links structure to prognosis: first and second degree recover completely, third partially, fourth and fifth need surgical intervention1.

The sixth degree and the limits of classification

Mackinnon and Dellon added a sixth degree for the mixed injury, in which different fascicles of the same nerve carry different degrees of damage, a pattern that better reflects clinical practice; reviews note that its use is not universally accepted14. The classification's clinical limitation is that most injuries are mixed and there is no diagnostic test that discriminates a second-degree from a fourth-degree lesion at the time of injury1. The dilemma of the — whether the scarred segment will conduct regenerating axons or must be resected and grafted — was framed by Mackinnon in 1989 as the surgical dilemma of the neuroma-in-continuity, and it is still resolved by time, examination and intraoperative recording rather than by any pre-operative test2021.

Grading recovery

Recovery is graded with the Medical Research Council scales: M0–M5 for motor power and S0–S4 for sensation, with S3 defined as the return of superficial pain and tactile sensibility without over-response and S4 as complete recovery2. In outcome research, "satisfactory" recovery is conventionally M4–M5 and S3+–S422. is the common currency of the literature, although is reported so variably that comparisons between series are unreliable2.
Examination notes. Seddon 1942/1943 (three types), Sunderland 1951 (five degrees), Mackinnon–Dellon sixth degree (mixed); second degree recovers fully because the endoneurial tube is intact; no test separates second from fourth degree at presentation.

§ 03 · CLINICAL ASSESSMENT

History, examination and the Tinel sign

The mechanism — penetrating, crush, stretch or ischaemic — predicts the likely degree of injury and the need for exploration21. Motor and sensory mapping against the expected territory of the nerve establishes the baseline. The , the tingling elicited by percussion over regenerating sensory sprouts, was described in 1915 and, as Davis and Chung's history records, not first by Tinel2324. Its interpretation is the point most often misunderstood: an advancing sign shows that regeneration is progressing along the distal stump, but "it only signals the progress of nerve regeneration" and is not a prelude to complete functional recovery24. The rate at which the sign advances is inferred from the regeneration rate, roughly a millimetre a day in the distal limb, so a sign that fails to move from the repair site over several months is the clinical signal of a blocked coaptation41.

Electrodiagnostic studies

and needle localise the lesion, quantify axon loss and contribute to prognosis25. In neurapraxia the compound muscle and sensory action potentials on stimulation distal to the lesion are maintained indefinitely, while stimulation above the lesion reveals partial or complete conduction block; in axonotmesis and neurotmesis the picture depends on the time since injury, because the distal responses fall only as Wallerian degeneration proceeds21. Conventional teaching holds that studies should wait until about three weeks after injury, but Campbell argues that a great deal of important information can be obtained in the first week, when a conduction block can already be shown21. , the electrical marker of denervated muscle, appear one to four weeks after axonal injury and later for proximal lesions; Willmott's tracing of that guideline found it rests almost entirely on mid-twentieth-century animal work261. Fibrillation amplitude falls with time — a mean of 612 µV in the first two months, 320 µV at five to six months and none above 100 µV after the first year — so it can date an injury of uncertain age27.

Ultrasound

High-resolution ultrasound answers the question electrodiagnosis cannot: whether the nerve is in continuity. In Padua's series of 112 nerves, ultrasound strongly modified the diagnostic and therapeutic path in 58 per cent of cases, chiefly by distinguishing neurotmesis from axonotmesis, identifying the cause and showing multiple sites of damage28. Zhu's prospective series classified injuries into seven sonographic types with 93 per cent accuracy against operative findings29. A 2026 individual-participant meta-analysis found disruption of fascicular echotexture in 74 per cent of high-grade and 9 per cent of low-grade lesions-in-continuity, independently associated with high-grade injury (adjusted odds ratio 21.6), although prospective validation is still awaited3031.
Examination notes. Neurapraxia = distal CMAP/SNAP preserved with conduction block across the lesion; fibrillations 1–4 weeks (later proximally); an advancing Tinel sign proves regeneration, not recovery; ultrasound is the test that shows continuity.

§ 04 · TIMING AND THE DECISION TO OPERATE

A clean transection is repaired primarily. In Birch and Raji's series of 108 median and ulnar nerve repairs, thirteen of sixty secondary repairs or grafts failed but no primary repair failed completely, and the few excellent results followed primary distal repair in younger patients32. Closed injuries in continuity are observed for clinical or electrodiagnostic evidence of regeneration; the absence of recovery after three to six months, depending on the level, warrants exploration1. Proximal injuries force the decision earlier, because the distance to the hand muscles makes it difficult to reinnervate them before irreversible change, and intraoperative nerve action potential recording across the lesion is often used to decide between neurolysis and resection with grafting21.
Delay is a measurable, independent predictor. In the individual-patient meta-analysis of 623 median and ulnar nerve injuries, each month of delay reduced the odds of satisfactory motor recovery (odds ratio 0.94 per month) and of satisfactory sensory recovery (0.92 per month)22. Experimental and clinical data "clearly point toward the advantage of early nerve repair", and considering nerve transfers early can shorten the time to reinnervation of muscle targets33. The consequences extend beyond the hand: after forearm nerve injury a quarter of patients had lost their work at a mean of eighteen months, with poor recovery, proximal level and heavier work predicting work loss and hand therapy protecting against it34.
Examination notes. Primary repair of clean transections; explore closed injuries without recovery at 3–6 months; each month of delay costs 6 per cent of the odds of motor recovery (Ruijs 2005).

§ 05 · DIRECT REPAIR

Adherence to a few well-established principles of evaluation and repair optimises the result of even complex injuries35. Tension-free epineurial repair with fine nylon in a well-vascularised bed is the standard31. Fascicles are aligned by their pattern and by the epineurial vessels. Grouped fascicular repair is anatomically attractive, but the randomised comparison of fascicular and epineurial digital nerve repairs by Young, Wray and Weeks in 1981 is the trial behind the conclusion, restated by Grinsell and Keating, that it is no better than epineurial repair in functional outcome; the added intraneural dissection and suture material cause scar that offsets the theoretical gain361. Lundborg's twenty-five-year perspective concluded that techniques for approximating nerve ends had reached an optimal technical refinement and that further gains would come from biology, not from suture37. Fibrin glue has performed equal to, or better than, microsuture in animal and cadaveric studies, with less granulomatous inflammation, but not a single well-controlled human trial exists38. A useful rule of thumb is that about half of the axons are lost at each coaptation: roughly 50 per cent regenerate through a primary repair and 25 per cent through a graft with two coaptation sites1.

§ 06 · NERVE GAPS

Autograft

Millesi's interfascicular grafting, developed in 1964 and reported in 1972 and 1976, established that a tension-free cable of sural gives results "at least as good as those after epineural nerve suture under ideal conditions and better than those after neurorrhaphy under tension"; useful motor recovery (M3 or better) was achieved in 82 per cent of median nerve lesions, and the poor results clustered in older patients, severe trauma and long delay3940. The autograft remains the standard for gaps over 3 cm, proximal injuries and critical nerves1. The sural nerve is the commonest donor despite requiring a second operative limb and leaving numbness of the lateral foot; Poppler and Mackinnon's series of 109 grafts drew 71 from the upper limb — the medial and lateral antebrachial cutaneous nerves, the third web-space branch of the median nerve, the palmar cutaneous branch and the dorsal cutaneous branch of the ulnar nerve — and avoided a second operative limb in two thirds of patients41. Grafts are reversed to funnel regenerating axons distally, and grafts proximal to the elbow, longer than 7 cm, in older patients or after longer delay do worse1. Sunderland's five rules, restated after forty years, summarise the field: early is better than late, coaptation is better than grafting, young do better than old, distal is better than proximal, and short grafts do better than long1.

Conduits

The rests on the observation that a short gap enclosed in a tube can bridge itself. Lundborg's randomised trial enclosed the injury zone of median and ulnar nerves in the distal forearm in a silicone tube with a deliberate 3–4 mm gap; there was no difference from conventional repair at one year, and at five years the only significant difference was less cold intolerance with the tube, with functional sensibility still improving throughout the five years4243. Re-exploration showed a new nerve structure bridging the former gap and only a limited tissue reaction, although four of seven re-explored patients had the tube removed for local discomfort44. Weber's multicentre randomised trial of polyglycolic acid conduits for 136 digital nerve transections found no overall difference from standard repair, but nerves with gaps of 4 mm or less had better moving two-point discrimination with the conduit than with end-to-end suture (3.7 versus 6.1 mm), and gaps of 8 mm or more also favoured the conduit over graft, on smaller numbers45. Autogenous vein conduits were inferior to direct repair in Chiu and Strauch's prospective comparison and are reserved for gaps of 3 cm or less in non-essential sensory nerves46. Collagen conduits and processed allografts gave statistically similar results for digital gaps under 2.5 cm in a single-institution series, with a trend to fewer poor results with allograft47. The biological ceiling is Schwann-cell migration, which is insufficient beyond about 2 cm, and no conduit has demonstrated outcomes equivalent to autograft for gaps greater than 3 cm1.

Processed nerve allograft

The decellularised supplies the basal lamina scaffold a hollow tube lacks. In a blinded pilot randomised trial of digital nerve gaps averaging 12 mm, allograft gave better and more consistent static two-point discrimination than hollow conduit, and every allograft repair regained some discrimination against 75 per cent of conduit repairs48. The industry-sponsored RANGER registry reports 82 per cent meaningful recovery (S3 or M3 and above) across its quantitatively assessed sensory, mixed and motor repairs with gaps up to 70 mm, with recovery falling as the gap lengthens (91 per cent below 15 mm against 69 per cent at 50–70 mm in the reported subgroup comparison), and 86 per cent S3 or better in fifty digital gaps of 25–50 mm — comparisons that rest on historical controls4950. Where the evidence sits is contested. The 2014 review confined allograft and conduit to sensory gaps under 3 cm1; Rbia and Shin found substitutes acceptable in motor and mixed nerves only for gaps under 6 mm and held that major segmental motor or mixed injury is optimally treated with cabled autograft51; the 2023 meta-analysis of 1,559 repairs found no significant difference in meaningful recovery between autograft and allograft across short and long gaps, with conduits significantly worse even in short sensory gaps (62 per cent against 82 and 87 per cent) and more painful52; and the 2022 Cochrane review of five randomised trials concluded that the evidence does not support the use of currently available repair devices over standard repair, with device removal required in 12 of 129 device repairs against none of 127 standard repairs53. Even the most recent algorithmic review concedes that surprisingly few studies have compared the methods of gap repair rigorously and that evidence-based guidelines for digital, motor and mixed nerve gaps are lacking54.
Examination notes. Millesi 1972/1976 (tension-free interfascicular grafting beats suture under tension); ~50 per cent axon loss per coaptation; conduits for short (≤ 3 cm, and realistically ≤ 2 cm) sensory gaps; PGA conduit beat end-to-end suture for gaps ≤ 4 mm (Weber 2000); silicone tube equivalent at 5 years with less cold intolerance (Lundborg 2004); allograft ≈ autograft in the 2023 meta-analysis but Cochrane 2022 finds no RCT support for devices over standard repair.

§ 07 · NERVE TRANSFERS

A sacrifices an expendable donor to reinnervate a more important recipient close to its target. Mackinnon and Novak stated the principle: good motor function requires a maximal number of motor axons reaching the motor end plate within a critical period, and transfers eliminate the graft by allowing a direct end-to-end repair without tension55. Against a graft, a transfer has one coaptation rather than two and minimises the distance over which the nerve must regenerate1; for proximal injuries and delayed presentations within the motor window, distal transfers often give a reconstruction superior to grafting5657.
The classic donors are worth knowing by name. The takes about 10 per cent of the bulk of the ulnar nerve — one or two fascicles — for direct suture to the motor nerve of the biceps in C5–C6 avulsion, with no significant impairment of hand function58. Its double fascicular extension adds a median nerve fascicle to the brachialis branch: clinical reinnervation appeared at a mean of 5.5 months and elbow flexion recovered to a mean of MRC 4+ in Mackinnon's first six patients, and in Ray's series of 29 cases 97 per cent regained elbow flexion (M5 in eight, M4 in fifteen, M3 in four) with no deficit in the donor distributions5960. For a high ulnar nerve injury the terminal anterior interosseous nerve branch to pronator quadratus is transferred to the deep motor branch; all eight of Novak and Mackinnon's patients reinnervated the intrinsic muscles with improved pinch and grip and no functional deficit in pronation61, and Brown's technical description adds two sensory transfers — the third web-space contribution of the median nerve to the volar ulnar sensory component, and end-to-side reinnervation of the dorsal cutaneous branch from the median sensory trunk62. The of the anterior interosseous nerve to the ulnar motor branch leaves the injured nerve in continuity and is intended for second- and third-degree axonotmetic injuries, to augment partial recovery or to "babysit" the motor end plates until native axons arrive63. Distal transfers from expendable median-innervated donors have also been described to restore wrist and finger extension after radial nerve palsy, as an alternative to grafting or tendon transfer64. When the reinnervation window has closed, tendon transfer remains the reconstruction of choice.
Examination notes. Transfer = one tension-free coaptation near the target; Oberlin 10 per cent of the ulnar nerve to biceps; double fascicular transfer M4+ in most; AIN (pronator quadratus branch) to deep motor ulnar without loss of pronation; SETS is for axonotmetic injuries, not complete transection.

§ 08 · OUTCOMES AND PROGNOSTIC FACTORS

The individual-patient meta-analysis by Ruijs and colleagues remains the reference for what repair achieves: of 623 median and ulnar nerve injuries repaired with modern microsurgical technique (complete motor data for 281 and sensory data for 380), only about half achieved satisfactory motor recovery (M4–M5) and fewer than half satisfactory sensory recovery (S3+–S4) — 51.6 and 42.6 per cent in Grinsell and Keating's restatement — the two correlating (r = 0.62)221. Age was the strongest predictor — patients under sixteen had 4.3 times the odds of motor and 27 times the odds of sensory recovery of those over forty — followed by level (proximal odds ratio 0.46), the nerve (ulnar 71 per cent lower odds of motor recovery than median, odds ratio 0.29) and delay22. Millesi's grafted series found the same adverse circumstances: advanced age, severity of the original trauma and a long interval before repair40. Mackinnon and Dellon's forty-year compilation found that 20–40 per cent achieved very good (M4 S3+) recovery after direct coaptation but that few injuries recovered fully1.
is the symptom patients complain of longest. It affected 56 per cent of patients with a single nerve injury and 70 per cent with combined median and ulnar injury two to ten years after repair, correlated with poor sensory recovery, did not diminish over the years, and was not predicted by age, arterial injury, site, type of injury or hand dominance65. A Cold Intolerance Symptom Severity score above 30 is pathological, the upper 95 per cent confidence limit in a normative population662. Outcome documentation has moved from single measures to instruments that combine domains: the Rosén–Lundborg score separates sensory, motor and pain/discomfort domains and correlates with the MRC sensory scale67, and Wang, Sunitha and Chung's review sets out which measures are reliable and which, like two-point discrimination, are not standardised2.

§ 09 · REHABILITATION AND THE PLASTIC BRAIN

Nerve injury is also a brain injury. Deafferentation from a nerve transection produces very rapid cortical synaptic remodelling and a distorted cortical hand map; denervated regions shrink and adjacent regions in both hemispheres enlarge, and misdirected regeneration then presents the cortex with a scrambled input681. Lundborg's conclusion is blunt: no surgical technique can assure recovery of tactile discrimination in the adult hand after nerve repair, whereas young individuals usually attain complete functional sensibility, and the outcome depends mainly on central factors6837. , described by Dellon, Curtis and Edgerton in 1974, aims to refine the distorted receptive fields6968. Timing may matter: in a randomised multicentre trial, patients who began re-education immediately after median nerve repair with an audio-tactile sensor glove had better tactile gnosis at twelve months than those who started conventional re-education at three months70. The systematic-review evidence is thin — limited support for early and late programmes in 2012, and in 2021 promising effects for early re-education with mirror visual feedback and for classic re-education combined with topical anaesthesia, but weak evidence for conventional programmes alone7172. Hand therapy also carries an economic signal: it was the one modifiable factor that improved return to work after forearm nerve injury34.

§ 10 · SYNTHESIS FOR THE EXAMINATION

The candidate should be able to move from structure to decision in one breath. A conduction block with preserved distal responses is neurapraxia and will recover; axonal loss with fibrillations at three weeks is at least second degree, and whether the endoneurial tubes survived cannot be known — only time, an advancing Tinel sign, serial electrodiagnosis and, increasingly, ultrasound will tell. Clean transections are repaired at once, without tension, epineurially. A gap is bridged with autograft when the nerve is motor or mixed, proximal or longer than 3 cm, and with a conduit or processed allograft when it is a short sensory gap. Proximal injuries with long regeneration distances, or presentations late within the motor window, are the domain of nerve transfers, which trade an expendable donor for a single coaptation next to the target. And whatever is done, recovery is governed by age, level, the nerve involved and delay, is measured in months and years, and is completed not by the surgeon but by the plastic brain.
Examination notes. Ruijs 2005: 51.6 per cent satisfactory motor and 42.6 per cent sensory recovery; age, level, ulnar nerve and delay are the predictors. Cold intolerance persists and CISS > 30 is pathological. Early sensory re-education has randomised support; conventional programmes have weak evidence.