Nerve defects present a reconstructive dilemma for microsurgeons. For large
nerve defects, cable grafting is widely accepted as the "gold"
standard1-6.
However, for shorter nerve defects, three techniques are commonly used: (1)
immobilization in a joint position that permits primary neurorrhaphy, (2)
nerve-grafting3, and
(3) nerve conduits7.
None of these techniques consistently achieves optimal results without
associated morbidity. Articulated external fixators offer promising
possibilities for patients with nerve defects by providing joint positioning
for end-to-end, tension-free nerve repair followed by controlled extension of
the affected extremity through the arc of the fixator. The use of an
articulated external fixator to flex the affected limb in order to allow for
an end-to-end nerve repair, followed by gradual extension of the joint, may
decrease the need for nerve-grafting and may improve functional outcome as
compared with traditional repair
techniques8.
However, the functional outcome of a repair in which the nerve is elongated by
gradual distraction within the confines of an external fixator has not
previously been compared with that of other repair techniques.
The hypothesis of the present study was that an articulated external
fixator can be used to facilitate tension-free end-to-end repair of nerve gaps
and may result in electrophysiologic and muscle force outcomes that are equal
to or better than those associated with repairs in which the joint is fixed in
immobilization to permit primary neurorrhaphy or nerve-grafting.
Electrophysiologic measurements can be used to determine when nerve-healing
occurs, and muscle force studies can be used to determine when and to what
degree reinnervation of the muscle occurs. The purpose of the present study
was to use both electrophysiologic and muscle force studies to compare the
results of three different nerve-repair techniques.
The experimental protocol described in the present study was approved by
the institutional animal care and use committee in accordance with National
Institutes of Health and United States Department of Agriculture guidelines
for the use and care of laboratory animals. Thirty-six skeletally mature male
New Zealand White rabbits supplied by Robinson Services (Winston-Salem, North
Carolina) were studied. The animals were housed in the animal vivarium in a
room with controlled temperature (20°C to 22°C) on a twelve-hour
light-dark cycle and were provided rabbit chow and water ad libitum. At the
end of the experimental protocol, each rabbit was anesthetized and killed with
an intravenous injection of sodium pentobarbital (100 mg/kg).
Surgical Technique
The rabbits were anesthetized with an intramuscular injection of ketamine
(35 mg/kg) and xylazine (5 mg/kg), with maintenance doses administered as
needed. The experimental forelimb was shaved into the axilla and was cleaned
with povidoneiodine. With use of aseptic microsurgical technique and a
double-headed operating microscope (Superlux 40; Carl Zeiss, Oberkochen,
Germany), an incision was made on the medial aspect of the proximal part of
the forelimb, and the median nerve was exposed. A nerve gap was created by
resecting a segment equal to four times the diameter of the median nerve on
the experimental side, thereby creating a nerve gap measuring approximately 7
mm in length. In a pilot study involving four repairs, a direct and
reproducible nerve repair was not possible after the resection of >7 mm
(approximately four times the diameter of the nerve) because tension across
the repair site resulted in immediate partial or complete rupture of the
repair following mobilization of the elbow.
The gap that was created in the present study (measuring four times the
diameter of the nerve) was a substantial one. The length of the rabbit arm
(from the shoulder to the elbow) is 7 cm; therefore, a 7-mm gap is quite large
(measuring approximately 10% of the length of the nerve) and cannot be
repaired primarily without elbow hyperflexion. However, in a human, a 10%
nerve gap would be considered to be too large to allow a primary repair,
regardless of positioning. Although elongation techniques to overcome gap
formation are recommended for gaps of as large as 8% to
10%4, a 10% nerve
defect would be treated by most hand surgeons or microsurgeons with cable
grafting.
The nerve ends were débrided of perineurium so that individual
fascicles were extending beyond the perineurium. A perineurial repair was
performed in all cases with use of 100 nonabsorbable nylon microsuture
(Surgical Specialties, Reading, Pennsylvania) with use of one of three
surgical techniques: cable nerve-grafting, primary end-to-end neurorrhaphy, or
primary end-to-end neurorrhaphy with gradual distraction. The soft tissues and
skin were closed with 4.0 Vicryl sutures (polyglactin; Ethicon, Somerville,
New Jersey). All rabbits received buprenorphine (0.03 mg/kg every twelve hours
for three days postoperatively) for pain control. Antibiotic prophylaxis was
provided with the intramuscular injection of cefazolin intraoperatively and
for four days postoperatively. The rabbits were checked daily for two weeks
following surgery or until the external fixator was removed and all soft
tissues were healed. One-half of the rabbits in each group were killed at
three months following nerve repair, and the other half were killed at six
months.
Nerve Repair Techniques
Group 1 (Cable Graft)
The nerve defect was repaired with tension-free nerve interposition
grafting techniques with use of either two or three strands of the medial
antebrachial cutaneous nerve, harvested from the same incision. The
perineurium of these strands was sutured together at each end to form a cable
that was the approximate size of the transected nerve. With use of epineurial
microsuture techniques, the perineurium of the proximal and distal stumps of
the transected nerve were then sutured to the perineurium of the interposition
graft. Postoperatively, the rabbits in this group were allowed full range of
motion of the elbow. Six animals were killed at three months, and five were
killed at six months.
Group 2 (End-to-End Repair without Distraction)
An articulated minirail external fixator with a horizontal axis (model
M111; Orthofix, Richardson, Texas) was applied at the elbow. Two 1.6-mm pins
were inserted into the lateral part of the humerus and two were inserted into
the ulna, with the hinge directly over the lateral humeral epicondyle and with
the elbow locked in full flexion, so that a tension-free end-to-end repair
could be performed after exposure of the median nerve in the axilla. Fourteen
days after surgery, each rabbit was again anesthetized and the fixator was
removed with use of aseptic technique. The rabbits then were allowed full
range of motion of the elbow. Six rabbits were killed at three months, and
seven were killed at six months.
Group 3 (End-to-End Repair with Gradual Distraction)
An articulated minirail external fixator with a horizontal axis (model
M111; Orthofix) was applied to the elbow. Two 1.6-mm pins were inserted into
the lateral part of the humerus and two were inserted into the ulna, with the
hinge directly over the lateral humeral epicondyle and with the elbow locked
in full flexion, so that a tension-free end-to-end repair could be performed
after exposure of the median nerve in the axilla. After fourteen days, the
fixator was extended at a rate of 10° every other day and was removed when
the limb reached full extension (25° to 30°)
(Fig. 1). Two days after full
extension was achieved, the rabbit was again anesthetized, and the fixator was
removed with use of aseptic technique. Six rabbits were killed at three
months, and six were killed at six months.
Outcome Measures
Before data collection, the rabbits were anesthetized with an intramuscular
injection of ketamine (35 mg/kg) and xylazine (5 mg/kg). Both axillae and the
left part of the groin were shaved. A catheter was then placed in the femoral
vein to allow for the intravenous administration of urethane (1000 mg/kg body
weight) during data collection. The median nerve on both the contralateral
(control) side and the experimental side was carefully dissected free from the
surrounding tissue and was exposed in the axilla. The flexor digitorum
superficialis tendon was exposed at the wrist on both sides and was tied with
3.0 silk suture. Several outcome measures were assessed in both forelimbs of
each rabbit, as described below.
Range of Motion
Before the beginning of dissection, the range of motion of the elbow was
measured in the experimental and control forelimbs with a goniometer.
Electrophysiological Studies
Peripheral nerve-conduction studies (including nerve-conduction velocity,
latency, and amplitude) were performed with use of a Nicolet Viking IIe
electrodiagnostic system (Nicolet Instrument, Madison, Wisconsin). The
stimulating electrode was placed 5 cm proximal to the insertion of the median
nerve into the flexor digitorum superficialis muscle. The active recording
electrode was placed on the flexor digitorum superficialis muscle belly, and
the reference recording electrode was placed on the flexor digitorum
superficialis tendon. The nerve then was stimulated with a nonrecurring
stimulus at a level of 1.0 mA for a duration of 0.1 ms. This study was
repeated three times, and the latency, amplitude, and nerve-conduction
velocity measurements for the three trials were averaged.
Muscle Contraction Force
The limb was immobilized with use of Kirschner wires through the humerus
and the radius to fix the arm to a wooden table to prevent motion. The flexor
digitorum superficialis was left in situ until force generation studies were
started. The tendon then was transected distal to the suture, and the suture
was attached to a force transducer (model FT03; Grass, Quincy, Massachusetts),
which was connected to a force transducer amplifier (model 13-G4615-50; Gould,
Cleveland, Ohio). The repaired or control median nerve was directly stimulated
just proximal to the repair site (model SD9; Grass) with increasing voltage
until a maximum isometric single-twitch force was obtained. The frequency of
stimulation then was increased until a maximum tetanic contractile force was
generated. The same procedure was repeated on the contralateral arm. Responses
were recorded with a calibrated recording oscillograph (model RS 3800; Gould)
that was linked to the force transducer.
Muscle Mass
Following the studies described above, the experimental and control flexor
digitorum superficialis muscles were dissected free from the surrounding
tissues, blotted, and weighed on a balance (Mettler, Toledo, Ohio).
Histological Analysis
Histological analysis was performed on nerves that were removed from the
rabbits that were killed at six months. Myelinated nerve fiber morphology was
evaluated on cross section, and the findings in the sections from the distal
and proximal nerve stumps were compared. The median nerve was transected 1 cm
proximal and 1 cm distal to the anastomotic site, and this segment of nerve
was pinned to hardened silicone to prevent shrinkage. The nerves were then
covered in 4% paraformaldehyde for twenty-four to forty-eight hours and then
were washed overnight with phosphate buffer solution (0.15 mol/L). The nerves
were placed in phosphate buffer saline solution, dehydrated in increasing
concentrations of ethanol (70% to 100%), transferred to propylene oxide, and
embedded in Epon. The proximal and distal nerve stumps were embedded
separately. Semithin (1-µm) sections that had been cut from each pretrimmed
block of tissue were stained with toluidine blue and were mounted on slides
for analysis of myelinated fibers at the light microscopic level with use of
SigmaScan Pro 5.0 (SPSS Science, Chicago, Illinois). Digitized images of
sections at magnification powers of 4× and 40× under oil immersion
were analyzed. The portion of the nerve distal to the anastomosis was measured
with regard to the total number of myelinated axons and the diameter and area
of the myelinated axons. The percentage of myelinated axons in each of three
size categories (small [<4 µm], medium [4 to 8 µm], and large [>8
µm]) was
calculated9. The
nerve was first viewed at a power of 4× to determine the area and then
at a power of 40× to determine the density. Three fields of
20,000-µm2 sections of the portion of the nerve distal to the
anastomosis were randomly chosen, and the density of axons was quantitated.
The myelinated axon morphology in both the proximal and the distal nerve stump
was calculated. The measurements from the proximal stump served as the
control.
Morphometry of Neuromuscular Junction
Strips from the central part of the flexor digitorum superficialis around
the insertion of the median nerve from both sides were harvested and stretched
by an additional 10% of the muscle fiber's resting length, pinned to a solid
agarose gel, and immersed for one hour in TRITC-a-bungarotoxin (25
µg/mL) (Molecular Probe, Eugene, Oregon, distributed by Cambridge
BioScience, Cambridge, United Kingdom) to localize acetyl-choline receptors at
the postsynaptic side of the neuromuscular junctions.
The dissected muscle was fixed by immersion in 4% paraformaldehyde in 0.1-M
phosphate buffer solution (pH 7.4) at room temperature for one hour and then
was washed in phosphate buffer saline solution three times for fifteen minutes
each. Each TRITC-a-bungarotoxin-labeled muscle was divided
longitudinally into small bundles through the nerve-insertion site. The small
muscle bundles were teased further and were mounted onto microscopic slides.
The muscle bundles then were covered with Vectashield mountant and
coverslips.
Quantification and Morphometric Analysis
Images of the TRITC-a-bungarotoxin-labeled neuromuscular junctions
were made with a confocal microscope (LSM 510; Carl Zeiss) (objective,
60× or 100×) with use of filters. For each muscle-fiber bundle,
images of the neuromuscular junction were made by superimposing a series of as
many as eight optical sections that were separated by 1-µm increments. At
six months after nerve repair, neuromuscular junctions in each group were
analyzed. The measured parameters, which are associated with neuromuscular
junction maturation, included the surface area, length, width, circumference,
and gutter depth of the neuromuscular junctions, as described in our previous
report10.
Measurements from the contralateral side were used as the control.
Statistical Analysis
Raw data were expressed as a percentage of the control value (i.e., the
value for the experimental side divided by the value for the control side).
Repeated-measures analysis of variance was used to detect differences between
the three and six-month groups. When differences between the groups were
significant (p < 0.05), multiple pairwise comparisons were performed with
use of the Student-Newman-Keuls method. The final statistical model that was
chosen contained only the terms that were significant at the p < 0.05
level.
Range of Motion
All of the rabbits regained full active and passive range of motion of the
forelimb, with no significant differences in range of motion between any of
the three-month or six-month groups.
Electrophysiological Studies
Amplitude
At three months, there was no significant difference in median nerve
amplitude between any of the groups. At six months, the amplitude was
significantly higher in Group 3 (end-to-end repair with gradual distraction)
than in Group 1 (cable graft) (p = 0.05) and Group 2 (end-to-end repair
without distraction) (p = 0.05). No significant difference in amplitude was
noted between Groups 1 and 2 at six months
(Table I).
Nerve-Conduction Velocity
At three months, the nerve-conduction velocities in Group 2 (p = 0.05)
and Group 3 (p = 0.05) were significantly faster than that in Group 1. At
six months, the nerve-conduction velocity in Group 3 was significantly faster
than those in Group 1 (p = 0.01) and Group 2 (p = 0.05)
(Table I).
Latency
No significant differences in latency were noted among the groups at either
three or six months.
Muscle Force Generation
Tetanic Force
No significant differences in tetanic force generation were noted among the
groups at three months. At six months, the tetanic force of the flexor
digitorum superficialis in Group 3 was significantly greater than those in
Group 1 (p = 0.05) and Group 2 (p = 0.05)
(Table I).
Single-Twitch Force
No significant differences in single-twitch force generation were noted
among the groups at three or six months.
Muscle Mass
No significant differences in muscle mass were noted among the groups.
Histological Findings
At six months following nerve repair, the total number of myelinated nerve
fibers in the distal stump was not significantly different among the three
groups. The average axon diameter and cross-sectional area were smaller in
Group 1 than in Groups 2 and 3; however, these differences were not
significant. Group 3 demonstrated slightly larger values for both axon
diameter and area as compared with Group 2; however, these differences were
not significant. In all three repair groups, the distal stump had a smaller
percentage of large myelinated nerve fibers than did the corresponding
proximal stump, but no significant difference in the percentage of small,
medium, and large-diameter axons was found among the three treatment groups.
Although there was no significant difference among the groups, there was a
trend toward larger fiber size in Group 3 as compared with the other two
groups (Table II and
Fig. 2).
Morphometric Analysis of Neuromuscular Junction
The neuromuscular junctions in Group 3 had a significantly larger surface
area than did those in Group 1 (p = 0.002) and Group 2 (p = 0.034). The
surface area in Group 3 (end-to-end repair with gradual distraction) was
approximately the same as that in the control group (832.046 compared with
876.183 µm2; p = 0.955). There were no significant differences
among the three repair groups with regard to length, width, circumference, or
gutter depth (Table III and
Fig. 3).
When a sharp peripheral nerve transection occurs without the creation of a
nerve gap, an end-to-end anastomosis is the preferred technique of repair.
However, a more common scenario involves the loss of a segment of nerve that
results in the creation of a gap. Some authors have recommended repair with
use of a cable
graft3,11-13,
whereas others have recommended treatment with primary neurorrhaphy.
Nerve-grafting has the disadvantages of donor-site
morbidity5,6,14,15,
technical difficulty, increased surgical duration, and the requirement for
axon regeneration across two repair
sites16. With
primary neurorrhaphy, the anastomosis site may be placed under some degree of
tension and compromise the repair. Primary repair under elevated tension is
certainly
undesirable1,13,17.
The negative effects of tension at the repair site include compromised nerve
blood flow (at 15%
elongation)18 and
suture pull-out (at 16% to 17%
elongation)18, with
both scenarios leading to possible failure of the repair. Techniques such as
limb-shortening4,
nerve
transposition19 and
end-to-side
anastomoses3 have
been used to prevent these complications. Joint positioning presents another
option to avoid complications, but its use has been discouraged on the basis
of the assumption that, upon joint remobilization, tension across the repair
site may lead to failure of the
neurorrhaphy13,20.
Another option, nerve elongation, was first described in 1944 by Weiss, who
used elastic cuffs of
tantalum2l; other
methods of nerve elongation have been described since
then2,7,17,18,22-24.
However, complications associated with tension at the anastomotic site have
demonstrated that the role of nerve elongation in acute and delayed repairs is
limited18.
Nerve defects that are equal in size to one nerve diameter or less can be
repaired primarily under little
tension25. If the
gap is more than one nerve diameter, primary nerve repair must be aided by
mobilization of the
nerve26 or joint
positioning27.
However, even if such techniques are used, the nerve repair may still be under
tension if the defect is very large. In this case, nerve-grafting is
indicated. Gradual nerve elongation by means of an external fixator is a
promising alternative technique for the repair of nerve defects because it
allows for nerve elongation without disrupting the anastomosis or blood
supply. In the present study, a nerve segment that was equal in size to four
times the diameter of the median nerve was removed to create a 7-mm gap. This
size of defect could be repaired directly and without tension only with
prolonged joint hyperflexion.
Clinical experience suggests that very rapid elongation of the nerve
following repair will result in loss of nerve function. However, according to
Ilizarov's concept of a tension-stress effect, in which nerve-lengthening
occurs when bone and other soft tissues are gradually lengthened, it is
possible to lengthen peripheral nerves
safely28,29.
It is difficult to determine how much tension is actually applied across the
anastomotic site with use of this technique. However, several animal studies
on the effect of tension on nerve repair have shown that nerves that are
repaired under some tension perform as well as or better than nerve grafts
do27,30,31.
Clinical studies also have supported early extremity motion after nerve
repair32. It is
difficult to determine how much tension can be tolerated at the repair site
and how soon after nerve repair that motion can be initiated safely.
Therefore, most authors have continued to recommend nerve-grafting for larger
nerve
gaps1-6.
Despite acceptable clinical outcomes in association with the use of hinged
external fixation to facilitate primary neurorrhaphy as reported by
us8, it is difficult
to quantify clinical results because of the wide variety of factors that are
required for a successful nerve repair in the clinical setting. The nerve
defects that were evaluated in the present study were smaller than those that
were treated in our clinical study when expressed as a multiple of the
diameter of the affected nerve. In the present experiment, we studied the
rabbit median nerve and there was limited space within the forelimb to release
the nerve. In contrast, in our clinical study, a sciatic nerve defect was
treated with the described technique. A longer nerve segment can be freed when
the sciatic nerve is treated with this technique than is the case when the
median nerve is treated with this technique. The longer the nerve segment that
can be released, the larger the nerve defect that can be treated. Despite this
difference, the results of the present animal study suggest that the described
gradual-distraction technique could be used to treat nerve defects measuring
more than four times the diameter of the affected nerve.
External fixator-assisted primary neurorrhaphy is indicated for moderate to
large nerve defects, defined as those measuring four to fourteen times the
diameter of the affected
nerve8. The present
study involved the use of a validated animal model and a standardized protocol
(i.e., reproducible injury and repair techniques), thereby allowing
comparisons to be made between treatment groups. At three months following
nerve repair, the group that had been treated with an articulated external
fixator had no significant superiority in terms of electrophysiological
findings or muscle function as compared with the other two groups. We believe
that a three-month recovery period is probably too short for the effective
evaluation of functional recovery. For example, in the present study, the
distance between the nerve entry point into the flexor digitorum superficialis
muscle and the repair site was approximately 5 cm. Given a rate of nerve
regeneration of 1 to 2
mm/day33, almost
two months would be required for the axon to reach the neuromuscular junction.
Only then would the muscle fibers have started to recover. Therefore, the
three-month recovery period represented only one month of muscle functional
recovery at best. However, the findings at six months demonstrated that a more
complete reinnervation had occurred in the group that had been treated with
gradual distraction. The amplitude of the evoked action potentials to stimulus
was larger, the nerve conduction velocity was faster, the tetanic force
generation was greater, and the surface area of neuromuscular junctions was
larger. In addition, there was a trend toward a greater percentage of large,
mature myelinated axons in the gradual distraction group as compared with the
other two groups, although this difference was not significant. Furthermore,
morphometric analysis suggested that the neuromuscular junctions in the
gradual distraction group had a significantly larger surface area than did the
junctions in the other two groups. Increased neuromuscular junction size in
the external fixator group may have been associated with increased muscle
fiber size, implying that this procedure has a functional benefit in terms of
increased force
generation34. The
combination of these results (a tendency toward a higher percentage of mature
axons distal to the repair site and a significantly greater surface area of
neuromuscular junctions in the gradual distraction group) may have been
responsible for the significant superiority in electrophysiological findings
and muscle force generation in the gradual distraction group. The most likely
explanation for the fact that a significant difference was not observed
histologically is that the time-period needed to appreciate such differences
exceeds the six-month period of the present study. It should be noted that at
six months after treatment with the gradual distraction technique, the number
of large axons in the distal nerve stump was still <50% of the number found
in the proximal stump, indicating that the time required for maturation of the
regenerating axons exceeded the six-month period.
The limitations of the present study included diminished single-twitch
force generation in all three groups on the repaired side as compared with the
control side, with no significant differences observed among the groups. Other
authors also have found that single-twitch force generation does not return to
normal at twenty-four weeks following
repair35. Perhaps
differences in single-twitch force generation cannot be measured until after
the six-month period used in the present study. In addition, no significant
difference in latency was observed among the groups. Latency also may be a
parameter that is measured more appropriately at a later time-point. Another
limitation of the current study is that the repairs were performed immediately
following injury, a condition that often is not possible in the clinical
setting. The potential for complete recovery is higher when the nerve is
repaired immediately than is the case when there is a delay from the time of
injury to the time of repair.
In summary, the present study demonstrates the efficacy of gradual
distraction with use of an articulated external fixator following primary
neurorrhaphy of peripheral nerve defects created by clean transection.
However, additional clinical research is needed to determine the validity of
this technique following injury in humans. The optimal gap size for this type
of repair, the rate of joint extension, and the timing of joint mobilization
in order to prevent joint contracture remain to be determined. A clearer
understanding of the effects of tension on the nerve, the anastomotic site,
and neuromuscular junctions is crucial before articulated external fixators
may be used extensively to treat nerve defects in the clinical setting.