The healing of nerves comprises a particular area of interest to head and neck surgeons because (1) much of head and neck surgery, including oncologic head and neck surgery, neurotologic surgery, and skull base surgery, places cranial nerves at risk for injury and (2) the healing of nerves has functional as well as cosmetic implications. As with other areas of wound healing, neural regeneration has seen an explosion in the understanding of mechanisms of repair as well as the birth of potential for manipulation of this repair process.
Interest in nerve repair dates to 1800 years ago. Historians believe that Galen was the first physician to discuss the possibility of nerve regeneration, in 200 AD. The first documented suture repair is attributed to Paul of Argina in 600 AD. Waller's work in the 1850s led to the rediscovery of the nature of peripheral nerve degeneration and subsequent regeneration after several hundred years of belief that nerves did not regenerate.
Seddon is credited with introducing the use of cable nerve grafts to overcome the poor results achieved when loss of significant amounts of nerve length required end-to-end neurorrhaphy under tension.  This concept was taken a step further in the form of vascularized nerve grafts.
The introduction of the operating microscope in peripheral nerve repair generally has produced improved results, which are reflected in the reports of numerous authors. Highet and Sanders were among the earliest to stress the importance of tension in peripheral nerve repair, and this has been investigated by many authors since that time. 
Sunderland made several contributions to the understanding of nerve physiology and nerve repair.  His description of intraneural topography and its complexity led many authors to pursue fascicular nerve repairs but left others doubting the ability of perineural suture repairs to restore the native axonal configuration.
The controversy regarding placement of sutures persists. Langley and Hashimoto are credited with the introduction of perineurial or fascicular suture repair of severed nerves, while Millesi et al first reported the interfascicular nerve repair in 1972. [4, 5]
Other approaches to nerve injury include internal neurolysis, which was first discussed by Babcock, and tubulization, which was introduced by Weiss in 1944. [6, 7] Materials for tubulization have included Millipore, silastic, Surgicel, collagen, and polyglycolic acid (PGA).
The variety of nerve injury repair techniques available and the continued active investigation are testimony to the facts that improvement certainly is necessary and no singular approach clearly is superior.
Sunderland described 3 fundamental types of nerve injury: (1) a transient interruption of nerve conduction without loss of axonal continuity (ie, neurapraxia, conduction block), (2) transection of axons (or conditions leading to loss of axonal integrity) with preservation of the endoneurium during Wallerian degeneration (ie, axonotmesis), and (3) complete disruption of the nerve fiber with loss of the normal architecture (ie, neurotmesis). The third level of injury can be further subdivided to include perineurial disruption (class IV injury) or epineurial transection (class V injury); all injuries that include neurotmesis may result in aberrant regrowth of axons into the wrong endoneurial tubes.
The response of the injured nerve in the first 12-48 hours includes Wallerian degeneration, which is degeneration of the distal axon to the motor endplate and of the proximal axon to the first node of Ranvier; axonal edema; and retraction of myelin. From 48-72 hours, the axons break into twisted fragments, and by the second week after injury, all traces of the axon are usually lost. The distal nerve fibers can be stimulated for approximately 72 hours after injury, an essential timeframe to consider when contemplating exploration of traumatic nerve injuries. Macrophages are mobilized to phagocytize debris along the nerve, and Schwann cells contribute to this activity. However, the main role of the Schwann cells is to guide regeneration by forming dense cellular cords (ie, Büngner bands) along the site of the degenerating axon. These bands provide conduits for axons once regeneration ensues.
Axonal regeneration generally occurs in 4 phases. In the initial phase, the neuron recovers, the axonal growth commences, and the axons reach the injured zone. In the second phase (scar delay), the axons must traverse the scar tissue at the site of injury. During the third phase, the axons propagate beyond the site of injury to reach the peripheral target where functional recovery (fourth phase) occurs with restoration of normal patterns of conduction.
The duration of the regenerative process varies and may require 6-18 months, depending on the length of the nerve and the site of the lesion. Although the commonly quoted regeneration rate is 1 mm/d, this figure varies considerably and can be used only as a rough estimate. Occasionally, very early signs of recovery may be present, which are thought to be due to so-called pioneer axons, which quickly navigate the pathway to the target tissue ahead of most nerve fibers.
A review of relevant issues in nerve repair arrived at several generalizations in the approach to nerve repair. These generalizations of the relevant issues are discussed in detail below.
Timing of Repair
Several authors in the past decade have solidified the notion purported since World War II that the earlier an injured nerve is repaired, the better. In the 1970s, investigation by McCabe and others in nerve physiology suggested that the ability of a nerve cell to regenerate is maximal at approximately 3 weeks postinjury. This information helped popularize the belief that neurorrhaphy should be delayed for 3 weeks. More recent data suggest that the sooner the integrity of the nerve is reestablished, the better the long-term outcome will be.
As indicated by Millesi and others, primary repair by neurorrhaphy is the treatment of choice following a transection.  An early secondary repair is recommended only when the amount of nerve damage is unable to be assessed.
In the repair of severed peripheral nerves, the epineurial suture repair remains the criterion standard to which other innovative techniques must be compared. However, this dictum has received many challenges, particularly since the introduction of perineurial repairs and the so-called interfascicular repair. The concept supporting these repairs was engendered in part by the work of Sunderland, who defined the topography of peripheral nerves.
Additional controversy surrounds the type, caliber, and number of sutures to be used. The optimal material for suturing peripheral nerves elicits the least foreign body reaction. Several materials, including human hair, silk, and particularly stainless steel, have demonstrated superiority over more reactive substances (eg, catgut). Subsequent research has indicated that monofilament nylon has reactivity similar to stainless steel, but it is more convenient for use in clinical practice. Several authors also have demonstrated acceptable results with PGA suture repair. However, this material has failed to gain widespread use thus far.
Despite numerous attempts at defining the issue, the most efficacious method of nerve reapproximation remains unproven. The proponents of fascicular and interfascicular repair contend that precise approximation of fascicles yields improved realignment of axons and the advantage of repairing only the damaged fascicles. The potential disadvantages include enhanced scarring required by the additional dissection and injury to the blood supply.
Aldskogius attempted to define a difference between epineurial and fascicular repairs by using horseradish peroxidase to delineate the motor neuron cell distribution corresponding to repaired nerves.  Aldskogius found that the normal pattern of motor axon innervation was not restored even after fascicular repair. Indeed, a number of authors conclude that no difference is found between the 2 types of repair as they currently exist. The prevailing sentiment is that the field must await some new facet with proven superiority to the fascicular repair before changing the current approach to the injured nerve.
Levinthal and his colleagues performed a direct comparison of the 3 techniques in a canine anterior tibial nerve model and found that epineurial and fascicular repairs produced comparable distal axonal regeneration, while interfascicular repair was inferior. These data are similar to the clinical results reported by Tupper on 93 patients with 109 digital nerve repairs (81 epineurial, 28 fascicular) and on 13 patients with 14 median nerve repairs (9 epineurial, 5 perineurial). 
One of the more recent additions to the pool of suture techniques includes the freeze-trim technique, introduced by de Medinaceli and reproduced by Wikholm with satisfactory results.  However, Wikholm's results for epineurial suture repair were surprisingly poor.
Kayikcioglu et al suggested another technique that incorporates an external metallic circle to achieve a larger coaptation area.  In their study, this technique resulted in less functional impairment than conventional epineural nerve repair.
Szal and colleagues helped establish the technical aspects of the ideal suture neurorrhaphy.  They examined only 10-0 nylon sutures, but they compared a continuous suture closure with 3 and 7 interrupted epineurial stitches in a rabbit facial nerve model. The best results were achieved in the group with 7 epineurial stitches.
More recent data, based on studies on cadaveric median nerves, suggest that 10-0 nylon is too fine, while 8-0 sutures have a tendency to pull out of nerve tissue. Giddens et al conclude that 9-0 nylon was the suture size of choice.
Interestingly, little work has compared different suture materials. Although early surgeons used fine silk to repair nerves, little deviation has occurred in recent decades from the use of monofilament nylon suture.
In summary, despite some suggestions to the contrary in the literature, epineurial suture repair of the peripheral nerve remains the accepted criterion standard, and fine (9-0 or 10-0) monofilament nylon remains the suture of choice.
Investigation into agents that have the ability to enhance the regenerative capacity of nerve tissue has coincided with development of the optimal method of reconnecting nerves after injury. The mechanism of action of these substances includes prevention of the reparative response of fibroblasts, facilitation of the regenerative process of the injured nerve cell, and inhibition of protease activity.
Some of the factors that have shown promise in nerve repair in the laboratory include triamcinolone, alpha–melanocyte-stimulating hormone (alpha-MSH) and its derivatives, leupeptin, and apolipoproteins. Nerve growth factor has yielded less encouraging results.
Data accumulated by Lipton et al in a rodent model suggest that local treatment with a steroid (eg, triamcinolone) leads to a greater regenerative index (ie, compound action potential through the graft divided by the compound action potential proximal to the graft) than the control treatment or treatment with systemic steroids; however, no statistically significant difference was found in the muscle twitch strength.  In a related experiment, time-release systemic triamcinolone treatment of rodent nerve repairs with 0.5-mg pellets led to increases in twitch strength and regenerative index when delivered over 21 days but not when delivered over 60 days. The hypothesis of decreasing scar formation and allowing axonal regeneration via the antifibroblast activity of steroids is intriguing, but it remains unproven. Greater definition of appropriate dosage and method of delivery, followed by confirmation in a primate model, is needed.
The peptide hormone alpha-MSH and its analogs have been used locally and systemically in an effort to enhance nerve regeneration. Teare et al have suggested that this is due to the anti-inflammatory properties of alpha-MSH. Working with a rodent sensory nerve model, Edwards and colleagues applied alpha-MSH locally via a microporous polypropylene tube used in repair of a severed sciatic nerve.  Animals demonstrated shortened times to sensory recovery when the nerves were repaired with tubulization compared with sutures alone, and this return was even more rapid when alpha-MSH–impregnated tubes were used.
ORG.2766, an adrenocorticotropic hormone (ACTH) analog, administered systemically in rodents with crush injury of a tail nerve appeared to shorten the time to return of sensation.
Another peptide factor is leupeptin, a tripeptide that appears to have the ability to deter Wallerian degeneration and inhibit the degradation of the muscle tissue. Its mechanism of action is via inhibition of calcium-activated protease and cathepsins B and C.
In a rodent model, Hurst demonstrated that intramuscular injections of leupeptin enhanced lower extremity muscle weight, myofiber diameter, and sciatic nerve axonal count after dividing the sciatic nerve.  This work was repeated by the same group in a primate model. The initial small number of animals (4) yielded encouraging results in median nerve repair and was then extended to 10 animals in a subsequent study. Compared with controls, the leupeptin-treated animals manifested increased sensory axon myelin sheath width, increased axon number, and faster motor nerve conduction. These investigations continue.
The final category of adjunctive agents that show promise in peripheral nerve repair are the apolipoproteins. They appear to play an essential role in the transport and utilization of cholesterol. Spreyer's work suggested the importance of apolipoprotein activity in nerve repair by inference from the 40-fold increase in messenger RNA (mRNA) in injured nerves compared with mature uninjured nerves. 
Muller and colleagues demonstrated that after a crush injury to rodent sciatic nerves, several proteins are induced in the distal, but not the proximal, segment of the injured nerve, particularly a 37-kd protein that represents a sialylated form of the plasma protein apolipoprotein E (apoE).  In a series of experiments, the same group presented convincing evidence that apoE participates in lipid transport in the injured nerve, providing the cholesterol for membrane biogenesis during axon regeneration and remyelination.
The capacity of increased levels of these transport proteins to enhance the regeneration of injured peripheral nerves remains unexplored.
Although reports suggest the effectiveness of neurotrophic factors in enhancing sciatic nerve recovery after both crush injuries and complete division, the use of purified nerve growth factor showed no improvement in nerve regeneration in a rodent sciatic nerve repair model using silastic channels. Further studies are necessary before investigation on primates or humans should be contemplated.
Despite the theoretical advantage of the absence of foreign body in the form of sutures at a nerve repair site, the use of tissue adhesives, such as fibrin glue (first introduced by Young and Medawar), has yielded disappointing results compared with conventional suture repair.
Nishihira and McCaffrey found that fibrin glue was no better than traditional 10-0 nylon suture approximation of severed rodent sciatic nerves and nerve grafts; fibrin glue was worse in some parameters.  A similar conclusion was reached by Moy and colleagues, who used a rabbit tibial neurorrhaphy model; they detected no advantage of repairs made with fibrin glue than with epineurial suture repair, with the exception of operative time, which was lower in glue repairs.  Clinically, electrophysiologically, and histologically, the suture repair was superior.
Histologic studies performed by Boedts demonstrated little reaction around perineurial nylon anastomoses of rodent sciatic nerves after 12 weeks, compared with neurorrhaphies approximated with fibrin glue and fibrin glue combined with a fenestrated collagen tube.  In the latter 2 repairs, a pronounced connective tissue proliferation produced stricture at the anastomosis site and neuromas in 3 of 10 cases. The addition of the collagen tube prevented compression of the anastomotic area but also resulted in a mass of connective tissue after 12 weeks.
However, recent studies have supported the use of fibrin glue as an alternative to sutures. Ornelas et al, using rodents, reported that fibrin sealant led to a faster return to the functional baseline, as well as to a better recovery of nerve conduction velocities and wave amplitudes as opposed to microsutures. [21, 22] Histologically, with fibrin glue, they found less of an inflammatory response and fibrosis, along with better axonal regeneration and fiber alignment. Lastly, they noted the ease and quickness of use. Martins et al also found better conduction velocity with fibrin glue versus 10-0 nylon sutures, yet they found no differences in functional or histomorphometric findings. 
Reports exist, such as those by Becker et al and Smahel et al, that extoll the ease of fibrin glue nerve repair and demonstrate comparable Schwann cell metabolism, electrophysiologic parameters, and histologic findings as suture-repaired controls. [24, 25] However, more studies similar to that by Medders et al are needed to define the particular properties of fibrin that, in general, appear to contribute to inferior results when compared with suture.  Medders and colleagues compared fibrin glue with no glue or sutures (simple reapproximation) in the repair of a rodent facial nerve severed in the intratemporal location. No difference in axon counts distal to the repair existed, suggesting that mechanical obstruction is not involved in fibrin-associated nerve repairs.
Currently, fibrin glue must be considered an inferior alternative to the criterion standard of epineurial suture repair. Even if it is capable of saving several minutes of operative time, the cost in terms of risk of AIDS or hepatitis for pooled blood products or expense for autologous fibrin makes the simplicity of suture repair a more viable choice.
The most immediate promising innovations include the use of tubulization techniques (absorbable conduits rather than silicone tubules) and the use of trophic substances such as neurocytokines, including nerve growth factor, brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factor, and neurotrophin-3, neurotrophin-4/5, and neurotrophin-6. The complex interactions between these proteins and their target cells are still being elucidated. Although the enhancement in recovery using neurotrophic substances has only been achieved in animal models, in some instances the differences are impressive. In a rodent sciatic nerve model, Utley et al reported statistically significant improvements in functional nerve recovery with the use of collagen tubulization or the delivery of BDNF to the nerve bed.
The best recovery was achieved in animals that received the BDNF by cross-linking it to the tubule. This enhancement has been confirmed in subsequent studies, and the effect is greater when the cytokines are delivered locally rather than systemically. Obviously, great expectation exists that the future will bring even more precise pharmacologic options for the enhancement of nerve regeneration. For the present, careful microsurgical suture reapproximation remains the best way to facilitate the normal nerve healing process.
The concept of providing a conduit through which nerve regeneration can be guided has attracted attention from numerous investigators, beginning as early as the 1940s. The field began with silicone tubes in an epineurial fashion, progressed to biocompatible substances (eg, PGA, vein, collagen), and has included attempts at fascicular tubulization.
Despite the rather convincing data discouraging the use of a nonabsorbable conduit, research continues to examine the use of silicon tubes. Merle et al reported on complications of the silicone chamber technique of nerve repair in 3 patients.  These authors also point out that several authors who had advocated silicon tubulization in the past have now abandoned it as a method of nerve repair, citing changes of chronic nerve compression proceeding to Wallerian degeneration even when the silicon polymer band does not constrict the nerve.
A potential use of nonbiocompatible chambers was suggested by Humphrey et al, who devised an observation chamber for nerve repair, which allowed repeated minimally invasive assessment of the sequence of nerve regeneration events.
Although the silicone tube appears to have little promise for clinical use, the concept of conduit-guided nerve repair should not be abandoned. Active research continues to isolate a material that will allow nutrients and blood supply to reach the regenerating nerve, prevent aberrant sprouting of regenerating axons, and resorb or dissolve when the regenerative process has been completed.
Walton and his colleagues reported results on 22 patients who underwent vein graft conduit repair of digital nerves; the results were comparable to those obtained in historical controls.  Although prospective randomized trials are needed, this report follows the work by Chiu and Rice in which an adequate environment for nerve regeneration was provided by short vein conduits.  Barcelos et al recently compared artery and vein grafts.  They found that vein grafts have closer-to-normal nerve organization than the artery grafts. However, overall, both types of grafts produced satisfactory nerve regeneration.
In a series of experiments, Rosen and colleagues developed an approach to nerve repair emphasizing the concepts of fascicular realignment, tubulization, and absorbable products in neurorrhaphy.  Although not recommending the technique as superior to traditional epineurial suture repair, they reported comparable quantitative histologic and physiologic parameters when comparing collagen fascicular tubulization repair to epineurial suture or fascicular suture repair in 38 cats.
More recently, Bertleff et al studied the use of tubulization in the transected sensory nerves of 17 patients versus the use of standard epineural suturing repair in 12 other patients.  Both methods returned good results, but they did not find that sensory recovery between the 2 methods significantly differed.
In a rodent model, Marshall et al used a PGA nerve coupler to perform fascicular tubulization repairs.  The authors thought that these sutureless repairs were quicker and easier to perform than the perineurial suture repairs, but again, the results were no better. In a similar study, Pham et al combined Avitene with a simple PGA tube for repair of rodent peroneal nerves.  They found no differences between the tubulization repairs and epineurial suture repairs in terms of tensile strength and histologic studies, but they did detect superior regeneration in the PGA group by electrophysiologic studies. Finally, in a primate model, the PGA tube was again found to be comparable to both perineurial and epineurial suture repairs.
Nerve injuries with a large gap may be treated more successfully with tubulization than with a nerve autograft. Tubulization has the advantages of no donor site morbidity, and it is thought to decrease the ingrowth of scar tissue. However, results have been mixed, with most studies demonstrating no significant differences using tubulization compared with autologous grafts.
The most useful aspect of the tubulization method ultimately may be its ability to allow localized prolonged delivery of exogenous trophic factors when these become defined and are available. Presently, tubulization appears to offer no advantage to traditional epineurial suture repair, even when fascicular alignment is undertaken.
Despite prior evidence to the contrary, injured nerves should be repaired as early as possible. Waiting until the metabolic environment has been maximized has no advantages. The current criterion standard technique of repair remains epineurial suture approximation of transected nerves with fine monofilament suture (eg, 9-0, 10-0) with interposition grafting using an autologous nerve if a tensionless repair cannot be achieved. What comprises tensionless remains debated; however, if the resected nerve segment is greater than 2 cm, most investigators advocate grafting. Techniques such as laser neurorrhaphy or the use of tissue adhesives (eg, fibrin glue) for nerve reapproximation are actively under study by various investigators. However, despite theoretical advantages of sutureless repair, these techniques, when subjected to rigorous scientific investigation, have no demonstrated advantage to date.