Evaluation and management of peripheral nerve injuries requires a thorough knowledge of neuroanatomy, neurophysiology, and electrodiagnostic medicine. The purpose of this article is not to describe the clinical features of every conceivable nerve injury. This type of information is well presented in other publications.[1] Instead, this article emphasizes the use of various electrodiagnostic techniques in the evaluation and management of nerve injuries in general.
Nerve injuries can be classified on the basis of completeness and predominant pathophysiology.
Nerve injuries first should be classified as complete or incomplete.
Complete injuries disrupt all the neurons traversing the injured segment, causing total loss of distal motor or sensory function.
Incomplete lesions disrupt some neurons but leave others unaffected, with some sparing of distal motor or sensory function. An incomplete nerve injury implies that at least part of the nerve remains in continuity; this has important therapeutic implications.
Although peripheral nerves may be injured in various ways, pathophysiologic responses to trauma at the neuronal level comprise only 2—demyelination and axonal loss.
Segmental demyelination (ie, neurapraxia): A mild stretch or compression injury may disrupt or distort the myelin sheath at the injury site, resulting in focal demyelination and leaving the axons intact. This causes a transient state of disrupted conduction along the injured segment—conduction slowing or block. Because the axons remain intact, function can be restored by focal remyelination, usually within a matter of days to weeks. This type of nerve injury is known as neurapraxia and is best considered the peripheral nervous system equivalent of "concussion."
Axonal injury and wallerian degeneration: Injured axons undergo a highly stereotyped process known as wallerian degeneration. Axonal function is disrupted immediately after the injury, although the disconnected distal segment initially survives and conducts externally applied stimuli; over the course of the next 5-7 days, however, the distal axonal segment slowly degenerates in a centrifugal fashion and eventually becomes inexcitable. The neuron may recover subsequently by axonal regeneration from the intact cell body, which is a slow process occurring at a rate of about 1 mm/day.
Axonal injuries that spare the supporting perineural connective tissue sheath are known as axonotmetic. The intact perineural connective tissue sheaths provide a conduit for axonal regeneration from the cell body to the target muscle, facilitating recovery. Injuries that disrupt the whole nerve, affecting both the axon and supporting connective tissue, are known as neurotmetic. These injuries are less likely to recover by axonal regeneration; they more often require surgical repair.
Mixed injuries: Individual axons can exhibit only one of these types of pathophysiologic change; however, one injured nerve is composed of thousands of axons, and a mixed pattern of segmental demyelination and axonal loss is manifested frequently. Moreover, some axons may be affected by different pathophysiologic processes at various points along their courses. This can make assessing the type of injury very difficult, even with electrodiagnostic methods, thus confounding management (see Case study 1 in Medical/Legal Pitfalls). Recovery from mixed lesions is usually biphasic. The neurapraxic component of the injury recovers quickly by remyelination and the axonal component of the injury recovers slowly by axonal regeneration.
Magnetic resonance neurography (MRN) uses a high resolution fast spin echo T2 imaging technique to demonstrate abnormal signal within nerve trunks at sites of major nerve injuries and compression. This technique can be particularly useful for identifying nerve injuries, such as piriformis syndrome and brachial plexus injuries, at sites inaccessible to conventional nerve conduction studies.
View Image
MRN of the brachial plexus. a: Abnormal signal in the brachial plexus elements on the affected (right) side. Compare to b: normal plexus on the unaffe....
The same technique can be used to demonstrate pseudomeningoceles in the cervical spine from traumatic nerve root avulsions.[2, 3]
View Image
MRN image through the cervical spine showing pseudomengocele (arrows) at the site of a cervical root avulsion in a patient with traumatic brachial ple....
Neuromuscular ultrasound can be used to differentiate neural continuity from discontinuity in cases of complete axonal nerve inury, and when combined with electrodiagnostic testing can provide additional information helpful for surgical planning, but is subject to technical limitations in areas that cannot be easily visualized.[4]
A carefully planned electrodiagnostic study is critical for determining the completeness and pathophysiology of all nerve injuries.
The completeness of a nerve injury can be determined any time after the injury. The presence of voluntary motor unit potentials on needle electromyography (EMG) examination of a clinically paralyzed muscle always indicates that the nerve injury, at least the branch or fascicle supplying that individual muscle, is partial.
In general, sensory responses are affected earlier and more severely than motor responses in peripheral nerve injuries. A reduction in sensory response amplitude of 50% or more, compared to the other (unaffected) side, is the most sensitive indication of peripheral nerve injury. Normal sensory responses are seen with nerve root injuries, even from clinically anesthetic regions, because the injured nerve segment is proximal to the dorsal root ganglion.
The physician performing the EMG must be fully cognizant of the time course of wallerian degeneration when performing nerve conduction studies to differentiate demyelination from axonal loss. The dissociation between the rates of degeneration of motor and sensory fibers can be a particular source of problem for the novice. A nerve conduction study performed 3-7 days after a peripheral nerve injury may show low-amplitude evoked compound muscle action potential (CMAP) with normal amplitude sensory nerve action potential (SNAP), a pattern usually interpreted as nerve root injury/avulsion.
Needle EMG findings correlate poorly with the degree of axonal loss. Denervation potentials do not appear for as long as 21 days after the nerve injury; the delay depends on the distance between the nerve injury and affected muscle. Moreover, the density of denervation potentials cannot be extrapolated to indicate the severity of axonal loss. Denervation potentials should be absent even 21 days after a pure demyelinating injury. However, most nerve injuries are mixed, and even predominantly demyelinating lesions suffer some secondary loss, often resulting in surprisingly profuse denervation potentials.
The amplitude of distal evoked CMAP and SNAP responses yields the maximum information regarding the degree of axonal loss that has occurred in motor and sensory fibers 10 or more days after a nerve injury. Evoked amplitudes must be compared to either a baseline study (immediately after the injury) or to the response evoked on the contralateral (normal) side. Adequate assessment of nerve injuries may necessitate the use of nonconventional nerve conduction studies.
The motor nerves used conventionally in conduction studies of the upper extremity, the median and ulnar, are both derived from the lower cord and medial trunk of the brachial plexus. A musculocutaneous motor nerve conduction study is required to assess the degree of axonal loss in cases of upper trunk plexus injuries.
The presence of a relatively preserved distal CMAP response amplitude in a paralyzed muscle more than 7-10 days after a nerve injury always should suggest more proximal conduction block. In most cases, the conduction block will be determined readily by comparing evoked CMAP response amplitudes from stimulation proximal and distal to the injury site.
In some instances, however, the conduction block may be too proximal to be demonstrated reliably by conventional motor nerve conduction studies (eg, conduction block at the nerve root level). In these instances, F-wave responses may be absent despite the presence of more normal distal evoked CMAP responses. Additionally, somatosensory-evoked potential (SEP) testing and/or nerve root stimulation may be used to demonstrate proximal conduction block even at the nerve root level.
In summary, a carefully planned and executed electrodiagnostic study is paramount in the evaluation of nerve injuries.[5] Needle EMG can demonstrate whether the injury is complete or incomplete at any time after injury. Nerve conduction studies are required to differentiate demyelination from axon loss; they yield the maximal information in this regard approximately 10 days after the injury. Nerve conduction studies should be bilateral to allow side-to-side comparisons of amplitude. Some types of injuries may necessitate the use of unconventional studies to adequately assess the degree of axon loss to each individual nerve branch or fascicle.
Decisions regarding surgical intervention must take into account both the mechanism of injury and completeness of the nerve injury.
Incomplete injuries
Incompletely injured nerves remain in (at least partial) continuity; therefore, they are likely to recover spontaneously. In general, patients with incomplete nerve injuries should be treated conservatively. Lesions are judged to be partial when some residual motor or sensory function is noted in the distribution of the injured nerve segment.
Needle EMG examination can be used to confirm that a nerve injury is partial by demonstrating the presence of some recruited voluntary motor unit potentials or signs of reinnervation even in clinically paralyzed muscles. However, note that in some cases of mixed or multiple nerve injuries in which some branches or fascicles are injured incompletely, some are likely to recover while others are not (see Case study 3 in Medical/Legal Pitfalls). These cases are best managed as complete lesions.
Complete injuries
Complete nerve lesions caused by lacerations or penetrating injuries should be referred for early surgical exploration and direct end-to-end repair.
Management of other complete nerve injuries depends on whether the pathophysiology of injury is thought to be neurapraxic, axonotmetic, or neurotmetic. This underscores the importance of an appropriately and carefully timed electrodiagnostic study in the evaluation of all these cases.
Complete nerve injuries that are predominantly neurapraxic can be expected to recover favorably over the course of weeks to months. When such cases do not recover as expected, patients should undergo follow-up electrodiagnostic testing, which may show the presence of significant secondary axonal loss suggesting that the initial testing was done too early, before the electrophysiologic abnormalities had fully evolved (see Case study 2 in Medical/Legal Pitfalls). However, if the follow-up study shows persistent conduction block across the injury site, then the patient should be evaluated carefully for an ongoing compressive lesion (eg, hematoma) by appropriate imaging studies.
Complete lesions with electrophysiologic evidence of axonal loss may be axonotmetic or neurotmetic. Axonotmetic injuries are more likely to recover spontaneously. Neurotmetic injuries often require surgical repair for adequate recovery. The only way to differentiate these injury types noninvasively is to monitor the patient for signs of recovery. However, the chances of successful surgical repair begin to decline by 6 months after the injury. By 18-24 months, the denervated muscles usually are replaced by fatty connective tissue, making functional recovery impossible. In most cases, close clinical observation is warranted for 3-6 months after this type of nerve injury. If no clinical or electrophysiologic evidence of recovery is noted during this period, these patients should be referred for surgical exploration.
Symptomatic management of patients with nerve injury
Many patients develop neuropathic pain in addition to motor and sensory deficits from nerve injury. The author uses an escalating drug regimen for symptomatic control of neuropathic pain.
Some patients with very mild pain can be treated effectively with long-acting nonsteroidal anti-inflammatory drugs (NSAIDs).
Topical lidocaine patches are very useful or patients with small areas of cutaneous pain, eg, pain in the lateral foot after a sural nerve biopsy or other injury.
Patients with moderately severe pain usually respond to low-dose tricyclic agents such as nortriptyline or antiepileptic drugs such as gabapentin (Neurontin) and lamotrigine (Lamictal).
Patients with severe neuropathic pain, unresponsive to these agents, may require narcotic analgesia. The author usually begins with tramadol (Ultram). If and when this becomes ineffective, oxycodone (OxyContin) is used with increasing doses. The author uses fentanyl patches for patients who are allergic to codeine, morphine sulfate (MS Contin) and methadone for patients with severe pain.
Spinal cord stimulators may be useful for patients with segmental neuropathic pain.
Patients with weakness and deformity after nerve injury should be considered for physical and occupational therapy evaluation. Function may be improved significantly by the use of the appropriate assistive devices such as cock-up wrist splints (for radial nerve injuries) and AFO splints (for foot drop with peroneal or sciatic nerve injuries). Additionally, consider tendon transfer to improve residual function, depending on the precise pattern of residual injury and functional limitation.[6]
Indication for surgical exploration and (nerve graft) repair
Complete nerve lesions caused by lacerations or penetrating injuries should be referred for early surgical exploration and direct end-to-end repair.[7]
Other significant nerve injuries with no clinical or electrophysiologic evidence of recovery after 3-6 months of clinical observation are also indications for surgical exploration.
Intraoperative nerve conduction testing and surgical repair
At the time of surgical exploration, the injured nerve may be obviously severed, in which case the injured segment should be resected and an end-to-end anastomosis (usually with an intervening nerve graft) performed. If the injured nerve segment appears to remain in continuity, intraoperative nerve conduction studies can differentiate axonotmetic from neurotmetic injury.
Sterile bipolar hook electrodes are used to stimulate and record nerve action potentials (NAPs) from surgically exposed nerve segments. Low stimulus intensities and durations should be used to avoid further iatrogenic nerve injury. Responses are recorded directly from nerves, so the patients can be paralyzed pharmacologically. Lifting the electrodes and nerve out of the operative field during testing is important to avoid current spread through blood and other fluids.
The presence of an evoked NAP across the injured segment indicates that the lesion is axonotmetic and recovering spontaneously. Surgical intervention should be limited to external neurolysis in these cases;[8, 9, 10] however, note that normal (or "super normal") NAPs can also be recorded from the brachial plexus sensory fibers in cases of root avulsion.
View Image
A 25-year-old man had a "flail" right arm after injury in a motorcycle accident (Case study 4). Left panel: Somatosensory evoked potentials (SEPs) rec....
The absence of a recordable NAP across the injured nerve segment more than 2-3 months after injury suggests that the injury is neurotmetic, necessitating nerve graft repair. In this instance, a normal nerve segment should always be tested as a positive control to confirm the integrity of the stimulating and recording apparatus. Furthermore, if a tourniquet was used during surgery, it should be released for at least 30 minutes prior to testing, as ischemia may attenuate normal NAP responses.
Selective use of intraoperative NAPs with either neurolysis or graft repair, depending on results, has been shown to improve postoperative outcome in these cases.[11, 12]
Intraoperative somatosensory-evoked potential testing and surgical repair
Brachial plexus injuries may be intraspinal (eg, root avulsions). In these cases, a NAP cannot be conducted across the injured segment to test continuity without performing very extensive surgery (eg, multilevel laminectomies). Intraoperative SEP testing may be very helpful in this regard.[8, 13]
A handheld bipolar stimulator is used to electrically activate the most proximally exposed region of the plexus with recordings made from surface electrodes placed over the contralateral scalp. The absence of cortical SEP responses suggests more proximal nerve root avulsion. However, cortical SEP responses can also be absent in the presence of high doses of volatile anesthetic agents, so testing a normal plexus element as a positive control is always important (see Case study 4 in Medical/Legal Pitfalls).
Nerve root avulsions can only be repaired by neurotization from adjacent nerves, such as the spinal accessory nerve, cervical plexus, or intercostal nerves. However, in cases of complete brachial plexus avulsion, these nerves cannot provide adequate donor neurones for adequate repair, and cross-chest nerve root transfer has been used with encouraging results.[14] There is only minimal postoperative deficit in the contralateral limb, provided certain precautions are taken.[14, 15]
For further information, please see Brachial Plexus Injuries, Traumatic and Facial Nerve Repair.
As outlined in the text, a wide variety of analgesic medications may be effective in the treatment of neuralgic pain. These include both narcotic and nonnarcotic medications.
A review of opioid equivalents and conversions may be found here.
Clinical Context:
Potent narcotic analgesic with much shorter half-life than morphine sulfate. DOC for conscious sedation analgesia. Ideal for analgesic action of short duration during anesthesia and immediate postoperative period.
Excellent choice for pain management and sedation with short duration (30-60 min) and easy to titrate.
Easily and quickly reversed by naloxone.
After initial dose, subsequent doses should not be titrated more frequently than q3h or q6h thereafter.
When using transdermal dosage form, pain in majority of patients controlled with 72-h dosing intervals; however, some patients require dosing intervals of 48 h.
Clinical Context:
DOC for analgesia because of reliable and predictable effects, safety profile, and ease of reversibility with naloxone.
Various IV doses used; commonly titrated until desired effect attained.
For chronic severe pain unremitting to alternative therapy, oral immediate–release and extended-release morphine sulfate may be warranted. Arymo ER is a morphine sulfate abuse-deterrent formulation.
Pain control is essential to quality patient care. Analgesics ensure patient comfort, promote pulmonary toilet, and have sedating properties, which are beneficial for patients who have sustained trauma or injuries.
Clinical Context:
Inhibits ascending pain pathways, altering perception of and response to pain; also inhibits reuptake of norepinephrine and serotonin.
Pain control is essential to quality patient care. Analgesics ensure patient comfort, promote pulmonary toilet, and have sedating properties, which are beneficial for patients who have sustained trauma or injuries.
Clinical Context:
By inhibiting re-uptake of serotonin and/or norepinephrine by presynaptic neuronal membrane, may increase synaptic concentration in CNS.
Useful as analgesic for certain chronic and neuropathic pain.
Clinical Context:
Has demonstrated effectiveness in treatment of chronic pain.
By inhibiting reuptake of serotonin and/or norepinephrine by presynaptic neuronal membrane, may increase synaptic concentration in CNS.
Pharmacodynamic effects, such as desensitization of adenyl cyclase and down-regulation of beta-adrenergic receptors and serotonin receptors, also appear to play role in its mechanisms of action.
These agents are a complex group of drugs that have central and peripheral anticholinergic effects as well as sedative effects. They have central effects on pain transmission and block the active re-uptake of norepinephrine and serotonin.
Clinical Context:
Has properties common to other anticonvulsants and has antineuralgic effects. Exact mechanism of action not known. Structurally related to GABA but does not interact with GABA receptors.
Clinical Context:
Triazine derivative used in neuralgia. Inhibits release of glutamate and inhibits voltage-sensitive sodium channels, leading to stabilization of neuronal membrane.
Follow manufacturer's recommendation for dose adjustments.
Clinical Context:
Structural derivative of GABA. Mechanism of action unknown. Binds with high affinity to alpha2-delta site (a calcium channel subunit). In vitro, reduces calcium-dependent release of several neurotransmitters, possibly by modulating calcium channel function. FDA approved for neuropathic pain associated with diabetic peripheral neuropathy or postherpetic neuralgia and as adjunctive therapy in partial-onset seizures.
These agents stabilize the neuronal membrane so the neuron is less permeable to ions. This prevents the initiation and transmission of nerve impulses, thereby producing the local anesthetic action.
An attempt should be made to classify all nerve injuries according to the completeness of the injury and the predominant pathophysiologic process involved: however, recognize that individual fascicles can be affected differently.
The results of nerve conduction studies may be difficult to interpret during the first 10 days after nerve injury until the effects of wallerian degeneration have had a chance to fully evolve in both motor and sensory fibers.
The best measure of axonal loss is the amplitude of the evoked CMAP response (compared to the other side) in a weak muscle from nerve stimulation distal to the injury site at least 7 days after the injury.
The density of denervation potentials in weak muscles is a poor measure of axonal loss. Denervation potentials may be absent for as long as 14-21 days after nerve injuries with severe axonal loss. Denervation potentials may be "profuse" in mixed injuries, even if the predominant pathophysiologic process is neurapraxia.
The presence of voluntary motor unit potentials in a clinically paralyzed muscle indicates that the nerve injury is partial, even if the distal CMAP response is absent.
Intraoperative nerve conduction testing often is required to differentiate axonotmesis from neurotmesis in closed nerve injuries that appear continuous. However, beware of "super normal" NAPs with more proximal nerve root avulsions.
Neil R Holland, MBBS, MBA, FAAN, Director of Neurology, Geisinger Health System;
Clinical Professor of Medicine (Neurology), The Commonwealth Medical College
Disclosure: Nothing to disclose.
Specialty Editors
Francisco Talavera, PharmD, PhD, Adjunct Assistant Professor, University of Nebraska Medical Center College of Pharmacy; Editor-in-Chief, Medscape Drug Reference
Disclosure: Received salary from Medscape for employment. for: Medscape.
Neil A Busis, MD, Chief of Neurology and Director of Neurodiagnostic Laboratory, UPMC Shadyside; Clinical Professor of Neurology and Director of Community Neurology, Department of Neurology, University of Pittsburgh Physicians
Disclosure: Serve(d) as a director, officer, partner, employee, advisor, consultant or trustee for: American Academy of Neurology<br/>Serve(d) as a speaker or a member of a speakers bureau for: American Academy of Neurology<br/>Received income in an amount equal to or greater than $250 from: American Academy of Neurology.
Chief Editor
Nicholas Lorenzo, MD, MHA, CPE, Co-Founder and Former Chief Publishing Officer, eMedicine and eMedicine Health, Founding Editor-in-Chief, eMedicine Neurology; Founder and Former Chairman and CEO, Pearlsreview; Founder and CEO/CMO, PHLT Consultants; Chief Medical Officer, MeMD Inc; Chief Strategy Officer, Discourse LLC
Disclosure: Nothing to disclose.
Additional Contributors
Milind J Kothari, DO, Professor, Department of Neurology, Pennsylvania State University College of Medicine; Consulting Staff, Department of Neurology, Penn State Milton S Hershey Medical Center
Disclosure: Nothing to disclose.
References
Stewart JD. Focal Peripheral Neuropathies. New York: Raven Press. 1993.
Kliot M, Slimp J. Techniques for assessment of peripheral nerve function at surgery. In: Loftus CM, Traynelis VC, eds. Intraoperative Monitoring Techniques in Neurosurgery. New York: McGraw-Hill Inc;. 1994:275-85.
Wilbourn AJ. Assessment of the brachial plexus and the phrenic nerve. In: Johnson EW, Pease WS, eds. Practical Electromyography. Baltimore: Williams & Wilkins. 1997:273-310.
MRN of the brachial plexus. a: Abnormal signal in the brachial plexus elements on the affected (right) side. Compare to b: normal plexus on the unaffected (left) side.
MRN image through the cervical spine showing pseudomengocele (arrows) at the site of a cervical root avulsion in a patient with traumatic brachial plexopathy.
A 25-year-old man had a "flail" right arm after injury in a motorcycle accident (Case study 4). Left panel: Somatosensory evoked potentials (SEPs) recorded at the scalp from stimulation of the (healthy) middle trunk (gain = 0.2 mcV/div, time base = 10 milliseconds [ms]/div). Middle panel: SEPs recorded at the scalp from stimulation of the lower trunk—no reproducible responses present (gain = 0.2 mcV/div, time base = 10 ms/div). Right panel: "Super normal" nerve action potentials recorded at the lower trunk from stimulation of the medial cord (time base = 1.5 ms/div, gain = 20 mcV/div).
Large-amplitude compound muscle action potential (CMAP) response was recorded from the right biceps muscle after intraoperative direct bipolar stimulation of the proximal right musculocutaneous nerve at low stimulus intensities (3.9 mA). The time base shown is 10 milliseconds/div and the gain is 50 mcV/div.
Electrodiagnostic testing 1 day after the injury revealed the following: (Left) Right ulnar motor conduction study showed a normal distal amplitude with conduction block across the elbow segment (gain = 2 mV/div, time base = 2 milliseconds [ms]/div). (Second from left) Right ulnar sensory response was normal (gain = 20 mcV/div, time base = 2 ms/div). (Third from left) Right ulnar F-wave responses were absent. (Right) Needle electromyographic (EMG) examination of right abductor digiti minimi was quiet at rest but showed a single fast firing unit on attempted contraction (gain = 200 mcV/div, time base = 10 ms/div).
Electrodiagnostic testing 3 days after the injury revealed the following: (Left) Right distal ulnar motor response is of lower amplitude than on day 1, approximately 50% of baseline (gain = 2 mV/div, time base = 5 milliseconds [ms]/div) with persistent conduction block across the elbow. (Right) Right ulnar sensory response is still normal (gain = 20 mcV/div, time base =2 ms/div).
Electrodiagnostic testing 6 days after the injury revealed the following: (Left) Right distal ulnar motor response is less than 10% of baseline (gain = 2 mV/div, time base = 5 milliseconds [ms]/div) with persistent conduction block across the elbow. (Right) Right ulnar sensory response amplitude still is relatively preserved at 50% of baseline (gain = 20 mcV/div, time base = 1 ms/div).
Electrodiagnostic testing 10 days after the injury revealed the following: Right ulnar motor (middle) and sensory (right) responses are absent. Needle electromyography (EMG) of first dorsal interosseus shows sparse denervation potentials with 1 fast firing unit on attempted volitional activity.
Intraoperative nerve action potentials recorded from the lateral cord (point R) with successive stimulation (at points 1, 2, 3, 4, and 5) along the course of the musculocutaneous nerve (gain = 100 mcV/div, time base = 0.5 milliseconds [ms]/div). Normal responses are recorded from stimulation at points 1 and 2. A slight increase in latency and drop in amplitude are noted on stimulation at point 3 close to the nerve injury. Stimulation at points 4 and 5 (distal to the injury) fail to evoke a recordable response.
A 25-year-old man had a "flail" right arm after injury in a motorcycle accident (Case study 4). Left panel: Somatosensory evoked potentials (SEPs) recorded at the scalp from stimulation of the (healthy) middle trunk (gain = 0.2 mcV/div, time base = 10 milliseconds [ms]/div). Middle panel: SEPs recorded at the scalp from stimulation of the lower trunk—no reproducible responses present (gain = 0.2 mcV/div, time base = 10 ms/div). Right panel: "Super normal" nerve action potentials recorded at the lower trunk from stimulation of the medial cord (time base = 1.5 ms/div, gain = 20 mcV/div).
MRN of the brachial plexus. a: Abnormal signal in the brachial plexus elements on the affected (right) side. Compare to b: normal plexus on the unaffected (left) side.
MRN image through the cervical spine showing pseudomengocele (arrows) at the site of a cervical root avulsion in a patient with traumatic brachial plexopathy.
){kind=link}
){kind=link}
){kind=link}