864

JOURNAL OF NEUROTRAUMAVolume 24, Number 5, 2007© Mary Ann Liebert, Inc.Pp. 864–875DOI: 10.1089/neu.2006.0226

Quantitative Characterization of Regenerating Axons after End-to-Side and End-to-End Coaptation in a Rat

Brachial Plexus Model: A Retrograde Tracer Study

KANIT SANANPANICH,1 MARY P. GALEA,2 WAYNE A. MORRISON,1,3

and AURORA MESSINA1

ABSTRACT

The efficacy of end-to-side repair as a method of nerve reconstruction has been questioned, andmost studies that characterize the mode of re-innervation are marred by inappropriate experimen-tal design and lack quantitative analysis. This makes characterization of re-innervating neurons con-fusing and consequently controversy remains as to the extent and source of reinnervating axons. Inan experimental brachial plexus rat model, we transected the musculocutaneous nerve, labeled itsneuron pool with Fast-Blue and joined the distal stump to the side of the intact ulnar nerve, or tothe proximal stump of the divided ulnar nerve, to characterize neurons that reinnervate the recip-ient nerve. Tetramethyl-rhodamine dextran (TMRD) or fluoro-gold was used to map the reinner-vating motor and sensory neurons at 12 weeks post-transection. No neurons originally labeled frommusculocutaneous nerve were subsequently labeled with TMRD or fluoro-gold, showing that thisoriginal neuron pool does not contribute to re-innervation of the distal musculocutaneous nerve, butthat reinnervation occurs solely by ulnar nerve motor and sensory axons. In the end-to-side group,16.4% of the motor and 7% of the sensory donor ulnar nerve neurons re-innervated the musculo-cutaneous nerve exclusively, and a further 10% motor and 11.6% sensory innervated the musculo-cutaneous nerve by collateral sprouting of their axons. This compared to re-innervation by 62.6%of motor and 70.4% of ulnar nerve sensory neurons in the positive control that underwent end-to-end repair. Our results confirm the concept of collateral sprouting and support the use of end-to-side repair.

Key words: collateral sprouting; motor neuron; optical disector; sensory neuron; tracer

1Bernard O’Brien Institute of Microsurgery, St. Vincent’s Hospital, Melbourne, Fitzroy, Victoria, Australia.2School of Physiotherapy, and 3Department of Surgery, St. Vincent’s Hospital, University of Melbourne, Parkville, Victoria, Australia.

INTRODUCTION

FOLLOWING NERVE INJURY where the proximal portionof an injured nerve is not available for repair, the dis-

tal nerve stump may be sutured to the side of an adjacent

intact nerve as an alternative method of nerve recon-struction to restore functional recovery of sensory andmotor target organs (Mennen, 1999; Viterbo et al., 1992;Zhang et al., 2002). Although end-to-side repair has beensuccessful in clinical and experimental settings, the un-

predictable clinical outcomes and conflicting experimen-tal results make its efficacy questionable (Al Qattan,2001; Rovak et al., 2001; Zhang et al., 2002). In order togain wider acceptance as a method of nerve repair andbe amenable to therapeutic intervention, then an accurateunderstanding of the axon source and the extent of thesensory and motor axon participation are desirable.

It is hypothesized that nerve re-innervation observedfollowing end-to-side repair occurs by collateral sprout-ing of donor nerve axons, but this has been inadequatelyinvestigated. Retrograde tracers have been used to mapthe parent neurons of both the donor and recipient nervesor of both their targets (Adelson et al., 2004; Lutz et al.,2000; Xiong et al., 2003; Zhang et al., 1999). While dou-ble-labeled neurons were observed in these studies, con-firming that collateral sprouts from one neuron innervateboth donor and recipient targets, this approach does notdetermine whether the reinnervating axons arise from theoriginal or donor nerve pools. In the models used, therecipient nerve is transected and sutured to a window inthe side of its parent nerve at a short distance distal tothe original branch point, permitting both the originalcut axons and the donor nerve axons access to the re-cipient nerve targets (Chen et al., 1998; McCallister etal., 1999). To resolve this ambiguity, others attemptedto label the original neurons from the proximal nervestump, 8 weeks post-transection, at the same time thatthe distal recipient nerve neurons were labeled (Tara-sidis et al., 1997, 1998). The lack of double-labeled sen-sory neurons in these studies was interpreted as an in-dication that none of the reinnervating axons werederived from the original stump. These results, however,are open to other interpretations: (i) the original axons arelikely to have retracted from the proximal stump once theyinnervated another target and hence could not be labeledfrom the original nerve stump (Rovak et al., 2000), and(ii) competitive inhibition of retrograde transport betweentwo tracers, applied at the same time to one neuron, mayhave precluded double labeling (Puigdellivol-Sanchez etal., 2002). Later experiments utilized a model in which anunrelated nerve graft was attached to the side of the donornerve to act as the recipient nerve (Goheen-Robillard etal., 2002; Hayashi et al., 2004; Kanje et al., 2000). Whilethis model shows more convincing evidence of collateralsprouting from the donor nerve axons, it is heavily disad-vantaged because there is no target (Kanje et al., 2000) orno functional recovery (Goheen-Robillard et al., 2002),and the growing axons must cross a second coaptation siteto reach a target (Hayashi et al., 2004). Thus, the extent towhich these sprouts mediate functional recovery cannot bedetermined.

To further investigate end-to-side repair, we reportedan experimental model that differs from the aforemen-

tioned tracer studies in a number of important aspects(Sananpanich et al., 2002). Cervico-thoracic nerveswere used that originate from discrete neuron poolsrather than lumbar nerves that originate from the sameneuron pool. Tracer was applied to the original proxi-mal nerve stump at the time of transection, rather thanafter recovery and the application of tracers to the dis-tal donor and recipient nerves, was separated tempo-rally to avoid competitive inhibition of retrogradetransport between tracers taken up by one neuron(Puigdellivol et al., 2000; K. Sananpanich, unpublisheddata). Qualitative data was collected using relativelyunbiased counting techniques (Gundersen et al., 1988).We validated this experimental end-to-side repairmodel demonstrating substantial recovery of the max-imum tetanic contraction tension of the recipientnerves’ dependent muscle (Sananpanich et al., 2002) aswell as efficient labeling of motor neurons. We nowextend this study by quantitatively characterizing therelative contribution of donor versus original neurons,terminal versus collateral axon sprouting and motorversus sensory neuron involvement and compare theoutcome to that following end-to-end repair.

METHODS

Animals

Thirteen male Sprague-Dawley rats weighing 200–250g were separated into end-to-end and end-to-side exper-imental groups. They were anesthetized with intraperi-toneal pentobarbitone sodium (Nembutal 50 mg/kg;Boehringer, Artarmon, Australia) prior to all surgical pro-cedures and operated on using aseptic technique and inaccordance with the National Health and Medical Re-search Council Australia guidelines and the approval ofthe St. Vincent’s Hospital Animal Ethics Committee.

Surgical Procedures

Nerve coaptation and labeling original musculocuta-neous neurons. The surgical procedures for nerves coap-tations have been previously described (Sananpanich etal., 2002). Briefly, with the rat supine, a longitudinal in-cision was made from mid-clavicle to elbow on the rightforelimb and both musculocutaneous and ulnar nervesidentified. With the aid of an operating microscope, themusculocutaneous nerve was exposed and transected atthe humeral neck level. Fast-Blue (EMS-POLYLOYGmbH, Gro�-Umstadt, Germany) crystals/powder wereapplied to the cut end of the proximal nerve stump using5.0 watchmakers forceps, and then the nerve plus adheredtracer was incubated in a plastic well, placed in situ, for

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

865

45 min (Sangster et al., 1999). On completion, the wellwas removed, and the stump gently wiped and rinsed withsaline, taking care not to allow the tracer to contact sur-rounding tissue, before it was turned back and sutured tothe proximal pectoral muscle. Fast-Blue labels the par-ent musculocutaneous neuron pool (Fig. 1) and remainsvisible in the neuronal cytoplasm without significant lossof fluorescence nor is it toxic to neurons for up to 6months (Novikova et al., 1997; Puigdellivol-Sanchez etal., 2000; Richmond et al., 1994).

For end-to-end coaptation (group A, n � 5), the ulnarnerve was then divided at mid-arm level and its proxi-mal stump sutured to the distal portion of the musculo-cutaneous nerve in an end-to-end fashion with three

stitches of 10-0 nylon (Dermalon; Sherwood, Davis andGeck, St. Louis MO).

For end-to-side coaptation (group B, n � 8), anepineural window 0.7 mm in length was made by gentlytouching the side of the ulnar nerve at mid-arm level withthe tip of a number-11 scalpel blade. This breached theepineurium and resulted in a bulging of the nerve below.The distal stump of the musculocutaneous nerve was su-tured to this epineural window with three stitches of 10-0 nylon in an end-to-side fashion.

Following surgery, all wounds were closed with con-tinuous 4-0 silk (Dysilk Dynek Pty. Ltd., Australia) andthe rats allowed to recover before being housed individ-ually in boxes, in an approved animal house facility un-der the supervision of a veterinarian.

Labeling reinnervating musculocutaneous and ulnarneurons. At 12 weeks post-transection, the original inci-sion was reopened and the recipient distal musculocuta-neous nerve was located, divided 10 mm distal to thecoaptation site and tracers II and III applied, as describedfor Fast-Blue.

For the end-to-end coaptation group either fluoro-gold(n � 2; Chemicon, Temecula, CA) or tetramethyl-rho-damine dextran (TMRD; n � 3, 10,000 MW lysine fix-able; Molecular Probes Inc., Portland, OR) was appliedto the proximal stump (Fig. 1A, tracer II). These tracerslabel motor and sensory neurons for up to 14 days, andare easily distinguished from Fast-Blue and each other.However, they have different efficacies of labeling andwere thus used in reverse order in approximately half therats, to ensure broad labeling of all neuron pools (Rich-mond et al., 1994). The contralateral ulnar and musculo-cutaneous nerves were then located, transected at the cor-responding level, and the respective neuron poolsretrogradely labeled with the same tracer used on thesenerves from the operated side (Fig. 1A, tracer II).

For the end-to-side coaptation group, either fluoro-gold (n � 5) or TMRD (n � 3) was applied to the prox-imal musculocutaneous nerve stump (Fig. 1B, tracer II).The rats’ wounds were closed, and they were allowed torecover. Two days later the wound was re-opened andthe ulnar nerve divided 10 mm distal to the repair site forapplication of tracer III (Fig. 1B, tracer III). In rats wherethe regenerated musculocutaneous nerve was mapped byfluoro-gold, the proximal stump of the ipsilateral ulnarnerve was labeled with TMRD (n � 5) and vice versa(n � 3). This delay in the timing of tracer III applicationto the distal ulnar nerve was made in order to eliminatepotential cross contamination of tracers and minimizecompetitive inhibition of retrograde transport betweenthese tracers (Puigdellivol-Sanchez et al., 2000; K.Sananpanich, unpublished data). The contralateral ulnar

SANANPANICH ET AL.

866

FIG. 1. Diagrammatic representation of the brachial plexusshowing the points of tracer application for end-to-end repair(A) and end-to-side repair (B) models. On the injured side, fastblue was applied to the proximal stump of the musculocuta-neous nerve (M) at the time of transection. Tracer II was ap-plied to the distal musculocutaneous nerve (m), 12 weeks post-transection, and Tracer III was applied to the ulnar nerve (U),2 days later (end-to-side repair only). The corresponding trac-ers were applied to the respective contralateral nerves 12 weekspost-transection.

and musculocutaneous nerves were located and tran-sected at this time, and the respective tracers were ap-plied to their proximal stumps to determine their base-line characteristics.

Tissue Preparation and Morphometric Analyses

Five days after application of the last tracer, rats werere-anesthetized, and a transcardial perfusion of 0.01Mphosphate-buffered saline (PBS) was followed by 350mL of 4% paraformaldehyde in PBS was performed. TheC4–T2 spinal cord segments were removed, and the junc-tion between levels were identified using the emergingnerve roots as a guide and marked by piercing with en-tomological pins. Specimens were post fixed overnight,cryoprotected by sequential incubation in 0.01M phos-phate buffer pH 7.4, containing 5%–10%–20%–30% su-crose, frozen, cut into 50-�m-thick longitudinal sectionswith a freezing microtome, collected onto glass slides,air-dried overnight, and mounted in DPX. Due to the rel-atively small number of retrogradely labeled motor neu-rons in the spinal cord, all sections were screened at 200�magnification for fluorescent motor neurons using anOlympus BX60 microscope fitted with a fluorescence at-tachment. Motor neuron profiles that contained nucleiwere counted and categorized according to tracer or trac-ers with which they were labeled, and their location wasmapped using a computer-linked digitizing system (MD3Microscope Digitizer) and version 4, MD-plot software(Minnesota Datametrics Corp., Minneapolis, MN). If lo-cation mapping and visualization in serial sections iden-tified the same motor neuron, it was counted and mappedonly once. Computer generated plots from single sec-tions, depicting the neuron labeling and location, wereoverlaid and composite MD-plot maps of each spinal cordprepared.

The C5–T1 dorsal root ganglia (DRG) were removedand post fixed overnight in the same fixative and pre-pared for embedding in methacrylate polymer (Immuno-Bed TM Kit, Polysciences Inc., Pittsburgh, PA) usingstandard technique (Sangster et al., 1999). The specimenswere dehydrated in 20% distilled water/catalyzed solu-tion A mix, impregnated with catalyzed solution A, andthen embedded in catalyzed solution A to which an ac-celerator is added. DRG-containing blocks were cut se-rially into 30-�m-thick sections using Ralph knives onan auto-cut rotary microtome (Microm HM 360; Wal-dorf, Germany). Every fourth section was viewed at1,000� magnification using an Olympus BH2 micro-scope fitted with a fluorescent attachment. The numberof labeled neurons/mm3 (Nv) for each ganglion was de-termined by the optical disector method (Gundersen etal., 1988) using a computerized Prior Stage Stepper (PriorScientific Instruments, Faulbourn, Cambridge, UK) and

a Heidenhain HD ND281B microcator (Traunreut, Ger-many). A minimum of 150 neurons was counted in eachDRG. The volume of each ganglion (Vganglion) was esti-mated using the Cavalieri method. The total number ofneurons per ganglion (N) was calculated as (N � Nv �Vganglion).

Baseline Data

The contralateral ulnar and musculocutaneous nerveswere retrogradely labeled with individual tracers to de-lineate the ventral horn and DRG location of their neu-rons and determine the degree of overlap between them.For the musculocutaneous nerve, we determined the base-line number of Fast-Blue–labeled neurons. No othertracer was used on this nerve. For the ulnar nerve we de-termined the baseline number of TMRD-labeled and offluoro-gold–labeled neurons. Data are presented asmean � SEM.

Experimental Data

The number of neurons labeled from the reinnervatedmusculocutaneous (RM) and the donor distal ulnar nerveswere expressed as a proportion of the mean number oflabeled control ulnar nerve neurons, with respect to therelevant tracer. The raw data and proportions, for eachrat, are presented in rows. The mean proportion of la-beled neurons (pooled tracers) is presented as mean �SEM, with n � number of nerves.

Double-Labeled Neurons

In the end-to-side repair group, the number ofTMRD/fluoro-gold–doubled-labeled neurons was ex-pressed as a proportion of the mean number of TMRD-labeled neurons in control ulnar nerve.

Calculation of the Extent to which the Musculocutaneous Is Reinnervated

Fluoro-gold and Fast-Blue label motor and sensoryneurons with similar efficacy (Novikova et al., 1997;Puigdellivol-Sanchez et al., 2002; Richmond et al., 1994).

The number of neurons innervating the musculocuta-neous nerve was determined by multiplying the numberof Fluoro-Gold–labeled neurons in the ulnar nerve by theproportion of single- and double-labeled ulnar nerve neu-rons innervating the musculocutaneous nerve. Data areexpressed as numbers and as a proportion of the originalmusculocutaneous nerve neuron number.

Statistical Analysis

Comparisons of the number of neurons labeled in con-trol nerves with respect for each tracer, and the propor-

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

867

tion of sensory and motor neurons that reinnervated therecipient musculocutaneous nerve between groups wasmade was made using a Students’ t-test (two-tailed, un-paired). Statistical significance was assigned for p �0.05.

RESULTS

Fluorescence Morphology

Fast-Blue–labeled neurons had a characteristic bluefluorescent granular cytoplasm and relatively unlabelednucleus. No double-labeled Fast-Blue neurons were ob-served in this study. Neurons labeled with fluoro-goldcontained a bright fluorescent white cytoplasm and nu-cleus (Fig. 2A,C). TMRD labeled the cytoplasm with acharacteristic red fluorescence (Figure 2B). Double-la-beled TMRD/fluoro-gold typically comprised of a whiteneuron (UV filter) superimposed on a red neuron (Rho-damine filter; Fig. 2B,C).

Control

Motor neurons. Retrograde labeling mapped the mo-tor neurons of the unoperated musculocutaneous and ul-nar nerves to almost exclusive cervical (C) and thoracic(T) spinal cord levels (Fig. 3, contralateral ulnar and orig-inal musculocutaneous). Fast-Blue labeled 277.3 � 13.7(n � 9) motor neurons in the musculocutaneous pool, andthese were located in C4 to C7 spinal cord levels withthe highest number in C5. TMRD labeled 226.2 � 7.5(n � 6) and fluoro-gold labeled 339.6 � 10.4 (n � 7;p � 0.0001) motor neurons in the ulnar nerve pool (Fig.4A). These were located in C7 to T1 spinal cord levelswith highest number in T1. The spatial distribution of flu-oro-gold– and TMRD–labeled motor neurons was simi-lar. In some cords, up to six motor neurons from eachnerve were housed at the C7 level.

Sensory neurons. Retrograde labeling mapped the sen-sory neurons of the unoperated musculocutaneous and ul-nar nerves to almost exclusive dorsal root ganglias(DRGs; Fig. 5). Fast-Blue labeled 3,200 � 190.7 (n � 9)musculocutaneous nerve sensory neurons, and these werelocated in C5 to C7 level DRG’s with the highest num-ber located in C6. TMRD labeled 3741 � 158.9 (n � 5)and fluoro-gold–labeled (6098 � 296.8; n � 7; p �0.0001) sensory neurons in the ulnar nerve pool (Fig. 4B).These were located in C7 to T1 DRGs, with the highestnumber located in C8. Both nerves were represented inthe C7 DRGs, which housed a mean of 26% (832) and2% (120) fluoro-gold–labeled ulnar and musculocuta-neous sensory neurons, respectively (Fig. 5).

End-to-End

When the distal musculocutaneous nerve was labeled12 weeks after transection and end-to-end coaptation, thereinnervating motor and sensory neurons were mapped

SANANPANICH ET AL.

868

FIG. 2. Transverse sections of dorsal root ganglia retro-gradely labeled with fast blue (B), fluoro-gold (G), or TMRD(R). (A) Neurons labeled exclusively with either fast blue orfluoro-gold. (B,C) Same field observed with rhodamine (B) orUV (C) fluorescence, containing neurons labeled exclusivelyor double-labeled (*) with TMRD or fluoro-gold.

A

B

C

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

869

FIG. 3. Longitudinal sections of the cervico-thoracic spinal cord (C4–T2) showing composite plots of motor-neurons, from onerepresentative rat, labeled by application of tracers at the time of transection and 12 weeks post-transection following end-to-endrepair (A) or end-to-side repair (B). Fast blue was applied to the proximal stump of the musculocutaneous nerve at the time oftransection. Tracer II was applied to the regenerated musculocutaneous nerve 12 weeks post-transection, and tracer III was ap-plied to the distal ulnar nerve (end-to-side repair only) 2 days later. The contralateral ulnar nerve neuron pool was labeled withtracer II (end-to-end repair) and tracer III (end-to-side repair).

FIG. 4. Number of motor (A) and sensory (B) neurons retrogradely labeled with fast-blue (FB), TMRD, or fluoro-gold fromthe control musculocutaneous (MC) or ulnar nerves. #p � 0.01 for FG versus TMRD.

to the donor ulnar nerve neuron pool level (C7–T1). Noneof the original musculocutaneous Fast-Blue–labeled neu-rons were labeled with TMRD or fluoro-gold tracers ap-plied to the distal regenerated musculocutaneous nerve,confirming that the reinnervating neurons originate ex-clusively from the donor neurons.

Motor neurons. Motor neuron maps showed a clear de-marcation between the original musculocutaneous neu-ron pool and the re-innervated musculocutaneous nerveneuron pool (Fig. 3A). A mean of 62.6% of the respec-tively mapped control ulnar nerve motor neurons regen-erated across the coaptation site (Table 1).

Sensory neurons. A mean of 70.4% of the respectivelymapped control ulnar nerve sensory neurons regeneratedacross the coaptation site (Table 1).

End-to-Side

Four categories of labeled neurons were generatedwhen the original musculocutaneous neurons were la-beled with Fast-Blue at the time of transection and thedistal musculocutaneous, and ulnar nerves were labeledwith tracers II and III, respectively, at 12 weeks after tran-section and end-to-side repair:

Tracer Category Origin of axons

1. FB Single Proximal musculocutaneous2. Tracer II Single Distal musculocutaneous3. Tracer III Single Ulnar4. Tracer II/III Double Distal musculocutaneous � ulnar

As observed in the end-to-end group, none of the Fast-Blue neurons labeled from the original musculocutaneouswere subsequently labeled with tracers applied to the dis-tal musculocutaneous or ulnar nerves.

Motor neurons. Motor neuron maps showed a clear de-marcation between the original musculocutaneous neu-ron pool and the reinnervated musculocutaneous neuronpool (Fig. 3B). Tracer II labeled 16.4% of neurons, in-dicating that they projected axons exclusively down themusculocutaneous nerve; tracer III labeled 76.3% of ul-nar motor neurons, indicating that they projected axonsexclusively down the ulnar nerve; and tracers II and IIIdouble-labeled 10% of neurons, indicating that they pro-jected axons into both these nerves (Table 2 and Fig. 6).

Sensory neurons. Tracer II labeled 7% of neurons, in-dicating that they projected axons exclusively down themusculocutaneous nerve; tracer III labeled 93% of neu-rons, indicating that they projected axons exclusively

SANANPANICH ET AL.

870

TABLE 1. RAW DATA SHOWING THE NUMBER OF MOTOR

AND SENSORY NEURONS RETROGRADELY LABELED

WITH TMRD (BOLD TYPE) OR FLUORO-GOLD

FROM THE REINNERVATED MUSCULOCUTANEOUS

NERVE, 12 WEEKS AFTER END-TO-END REPAIR

Rat n %

Motora

1 127 562 130 573 155 684 240 715 208 61

Mean 62.6SEM 3.0Sensoryb

1 2120 572 2680 723 2768 744 4548 755 4496 74

Mean 70.4SEM 3.4

Each number is expressed as a proportion of the mean numberof labeled ulnar nerve neurons for the respective tracer (thirdcolumn).

aNumber and proportion of motor neurons labeled from there-innervated musculocutaneous nerve 12 weeks after end-to-end repair.

bNumber and proportion of sensory neurons labeled from there-innervated musculocutaneous nerve 12 weeks after end-to-end repair.

FIG. 5. Mean numerical distribution of fluoro-gold mappedcontrol dorsal root ganglia sensory neurons of the control mus-culocutaneous and ulnar nerves. Note the small overlap at theC7 DRG level.

down the ulnar nerve; and tracers II and III double-la-beled 11.6% of TMRD-labeled ulnar nerve neurons, in-dicating that they projected axons into both these nerves(Table 2 and Fig. 6). This distribution accounts for111.6% of the ulnar nerve neurons present in respectivelylabeled control nerves, indicating an increased labelingefficacy or a contribution of non-ulnar nerve sensory neu-rons.

End-to-End versus End-to-Side Repair

Overall, the proportion of motor and sensory neuronsthat projected axons into the distal musculocutaneousnerve was significantly greater following end-to-end thanend-to-side repair: (62.60% � 2.9) vs. (26.4% � 1.4),p � .0001, for motor neurons; and (70.4% � 3.39) vs.(18.6% � 1.3), p � 0.0001 for sensory neurons (Fig. 6).

However, the donor ulnar nerve was sacrificed for end-to-end repair but spared for end-to-side repair.

The total of number of neurons innervating the distalmusculocutaneous after end-to-end and end-to-side repairwas calculated and is presented in Figure 7. When ex-pressed as a proportion of the original musculocutaneousnerve neurons 32% of motor and 35% of sensory mus-culocutaneous nerve neuron innervation was restored af-ter end-to-side repair compared to 76% motor and 134%sensory after end-to-end repair.

DISCUSSION

Using an experimental brachial plexus rat model ofend-to-side repair, we have quantitatively measured therelative contribution of donor versus original neurons,

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

871

TABLE 2. RAW DATA SHOWING THE NUMBER OF MOTOR AND SENSORY NEURONS

RETROGRADELY LABELED WITH TMRD (BOLD TYPE) OR FLUORO-GOLD

RM RM

ULNAR, III II II/III ULNAR, III II II/III

Rat n n n % % %

Motor neuronsa

1 24 112 145 56 23 64 17 103 156 62 19 69 18 84 242 35 9 71 15 45 244 44 15 72 19 76 231 67 27 102 20 127 180 72 29 80 21 138 34 37 10 16

Mean 76.3 16.4 10SEM 5.5 1.4 1.5Sensory neuronsb

1 152 423 2864 712 512 77 12 144 6616 164 436 108 4 125 4984 224 252 82 6 76 3872 644 516 103 11 147 3556 332 404 95 5 118 400 7

Mean 93.0 7.0 11.6SEM 5.9 1.2 1.3

RM, reinnervated musculocutaneous.aNumber and proportion of motor neurons labeled 12 weeks after end-to-end repair.bNumber and proportion of sensory neurons labeled 12 weeks after end-to-end repair.Tracer II was applied to the reinnervated musculocutaneous nerve and Tracer III was applied to the donor ulnar nerve 12 weeks

after end to side repair. Each number is expressed as a proportion of the mean number of labeled ulnar nerve neurons for the respec-tive tracer (columns 5–7).

terminal versus collateral axon sprouting and motor ver-sus sensory neuron involvement and compared the out-come to that following end-to-end repair. This is the firstcomprehensive quantitative study of the musculocuta-neous and ulnar nerve neuron pool and of the neuronsthat mediate end-to-side and end-to-end repair. It con-firms and extends our previous observations that it is pos-sible to achieve functional motor re-innervation of the bi-ceps brachii muscle from the ulnar nerve motor axonsfollowing end-to-side repair of the musculocutaneousnerve (Sananpanich et al., 2002). Innervation of the mus-culocutaneous nerve by sprouting of sensory and motorulnar nerve axons was achieved with minimal loss ofdonor nerve innervation and no loss of the ulnar nerveneuron pool. While significantly more motor and sensoryneurons innervate the musculocutaneous nerve after end-to-end repair, this comes with 100% loss of donor nerveinnervation, as well as 38% motor and 30% sensory neu-ron loss. From a recipient nerve aspect, end-to-side re-pair restored less motor and sensory innervation com-pared to end-to-end repair. However, in addition to thecentral neuron and peripheral axon loss, end-to-end re-pair also resulted in an over-abundance of sensory neu-ron innervation of the musculocutaneous nerve. All theseanomalies have been implicated in neuropathic pain sen-sation (Woolf et al., 1992; Ruocco et al., 2000).

Previously, using the experimental model described inthis paper, we explored the motor function recovery ofthe musculocutaneous nerve’s dependent biceps brachii

muscle (Sananpanich et al., 2002). At 12 weeks aftersurgery, end-to-side and end-to-end repair restored 43%and 65% of muscle function, measured as the maximumtetanic force developed by the biceps brachii muscle vianerve stimulation. Taken together with this study, the dataindicate that motor neuron innervation is somewhat moreefficacious after end-to-side compared to end-to-end re-pair restoring 1.3 versus 0.9 (muscle function percent-age) per (neuron percentage re-innervation). Overall, ourdata indicate that when the original proximal stump is notavailable, end-to-side repair achieves an acceptable levelof recipient nerve reinnervation without the potential sideeffects of end-to-end repair.

In a recent brachial plexus study, it was shown thatboth motor and sensory neurons cross the end-to-sidecoaptation site restoring up to 70% of the dependent mo-tor and sensory target function at 6 months post-transec-

SANANPANICH ET AL.

872

FIG. 6. Mean proportion of the control ulnar-nerve motor andsensory neurons that were labeled from the distal musculocu-taneous nerve (MC) following end-to end (EE) repair; and fromthe MC, ulnar and MC � ulnar nerves following end-to-side(ES) repair.

FIG. 7. Comparison of the motor (A) and sensory (B) neuronnumbers labeled from control and experimental nerves follow-ing end-to-end (EE) and end-to-side (ES) repair. R-MC, distalmusculocutaneous nerve.

tion (Bontioti et al., 2005). However, the methodologyof nerve selection, retrograde labeling, and neuron countsin this study differs from ours in number of significantways so that it is not possible to directly compare data.Compared to our study, these authors reported consider-ably fewer labeled neurons overall, a much larger pro-portion of donor nerve neurons that exclusively inner-vated the recipient nerve and very few double labeledneurons. In another brachial plexus model, all neuronsinnervating the recipient nerve were double labeled, andthese out-numbered the neurons that exclusively inner-vated the donor nerve (Lutz et al., 1999). In a peronealto tibial nerve model, where retrograde tracers were ap-plied to the donor and recipient nerves (Adelson et al.,2004; Xiong et al 2003.), or their targets (Zhang et al.,1999), only a few double-labeled neurons were detected.

Over the last decade, increasing numbers of end-to-side repairs have been carried out clinically with en-couraging, though variable, results. Case reports have de-scribed many satisfactory outcomes (Kostakoglu, 1999;Mennen, 1998, 2003; Yuksel et al., 2004) although somedisappointing functional results have also been obtained(Kayikciogl et al., 2000). Both positive and negative out-comes have also been reported experimentally (Al Qat-tan, 2001; Rovak et al., 2001; Zhang et al., 2002). Over-all sensory innervation shows superior recovery to motorinnervation however the variety of repair techniques,clinical presentation of nerve injuries and numerous ex-perimental models make direct comparison difficult.Taken together, this highlights the need for more studiesto improve the technique of end-to-side repair.

Controversy exists as to extent to which sensory andmotor neurons cross the coaptation site and there are lim-ited data in which both axon populations are studied atthe same time (Adelson et al., 2004; Chen et al., 1998;Lutz et al., 2000; Matsumoto et al., 1999; Sananpanichet al., 2002; Tarasidis et al., 1997). The behaviour of mo-tor and sensory neurons exposed to an end-to-side coap-tation protocols has been reported to differ. We observeda difference in the pattern of sensory and motor neuronplasticity when exposed to our end-to-side repair proto-col. In general, up to 16% of motor neurons relinquishedinnervation of their parent ulnar nerve compared to only7% of the sensory neurons. Furthermore, although 100%of the motor neuron population could be accurately ac-counted for, there appears to be over-abundance 111%(approximately 600 neurons) of sensory neuron involve-ment. This may point to involvement of sensory neuronsnot originally housed at the ulnar nerve coaptation site.Stimulated axon sprouts, derived from brachial nervebranches located proximal to the coaptation site, couldextend along the intervening nerve to reach the coapta-tion site. The capacity for proximal stump axons to rein-

nervate their distal stump by using the intervening nerveas a peri-epineural bridge has been reported in models ofend-to-side repair where the distal stump is sutured to theparent nerve up to 1.5 cm more distal (Chen et al., 1998;McCallister et al., 1999). This mechanism of re-innerva-tion by original axons or by axons not originally housedin the parent nerve should not be underestimated, butclearly requires further study.

Furthermore, it is not clear whether the recipient nervecan be innervated by neurons that reside at a differentspinal cord level (Matsumoto et al., 1999; Sananpanichet al., 2002; Zhang et al., 1998). The use of donor andrecipient neuron pools that do not reside in the samespinal cord and DRG levels when considering nerve forend-to-side repair also merits discussion. The use of suchexclusive neuron pools has been reported in other end-to-side studies with both positive (Kubek et al., 2004;Sananpanich et al., 2002) and negative (Zhang et al.,1998) results being reported. We show that there is con-siderable re-innervation of distal recipient musculocuta-neous nerves by neurons from the unrelated ulnar nervepools. This observation has considerable clinical rele-vance since it shows that different levels of donor andrecipient neuron pools can result in effective repair andit supports use of a broader choice of donor nerves in aclinical setting or a better motor/sensory function match-ing of donor and recipient nerves.

It has recently been proposed that interaction of injuredand uninjured neurons within the DRGs is required forcollateral sprouting to occur (Bajrovic et al., 2002). How-ever, in our study, this interaction would be minimal sinceonly 2% of the donor nerve neurons shared the C7 DRGwith injured musculocutaneous nerve neurons but 12%of ulnar nerve neurons project axons to both ulnar andthe musculocutaneous nerve. Our data would indicate thatthis interaction is not essential for collateral sprouting tooccur but may contribute. The cues that initiate axonsprouting in circumstances such as end-to-side are largelyunknown and beyond the scope of this study, but theyare also likely to arise from in the distal stump and theinjured epineurium.

Using retrograde labeling and unbiased countingmethod, we showed that when the cut distal end of themusculocutaneous nerve is coapted to the side of the ip-silateral ulnar nerve approximately 30% of both its sen-sory and motor neuron population is restored. This oc-curs by collateral sprouting of donor nerve neurons in theabsence of any original proximal stump neurons and ir-respective of the fact the donor and recipient neurons re-side in exclusive spinal levels. However, sprouts fromother more proximal donor nerve branches may also con-tribute, and this has not been considered in previous stud-ies. The finding would strongly support the use of end-

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

873

to-side coaptation as an alternative method of nerve re-pair, when the original distal stump is unavailable, withpotential for a very positive outcome.

ACKNOWLEDGMENTS

We thank Anthony Jaworowski for reviewing the man-uscript. The Transport Accident Commission Victoriaand the Jack Brockhoff Foundation, Victoria, Australia,supported this work. Kanit Sananpanich was financiallysupported by the Faculty Development Fund of the Fac-ulty of Medicine, Chiang Mai University, Thailand.

REFERENCES

ADELSON, P.D., BONAROTI, E.A., THOMPSON, T.P.,TRAN, M., and NYSTROM, N.A. (2004). End-to-side neu-rorrhaphies in a rodent model of peripheral nerve injury: apreliminary report of a novel technique. J. Neurosurg. Pedi-atr. 101, 78–84.

AL QATTAN, M.M. (2001). Terminolateral neurorrhaphy: re-view of experimental and clinical studies. J. Reconstr. Mi-crosurg. 17, 99–108.

BAJROVIC, F., KOVACIC, U., PAVCNIK, M., andSKETELJ, J. (2002). Interneuronal signalling is involved ininduction of collateral sprouting of nociceptive axons. Neu-roscience 111, 587–596.

BONTIOTI, E., KANJE, M., LUNDBORG, G., and DAHLIN,L.B. (2005). End-to-side nerve repair in the upper extremityof rat. J. Peripher. Nerv. Syst. 10, 58–68.

CHEN, Y.G., and BRUSHART, T.M. (1998). The effect of den-ervated muscle and Schwann cells on axon collateral sprout-ing. J. Hand Surg. Am. 23, 1025–1033.

GOHEEN-ROBILLARD, B., MYCKATYN, T.M., MACKIN-NON, S.E., and HUNTER, D.A. (2002). End-to-side neur-orrhaphy and lateral axonal sprouting in a long graft ratmodel. Laryngoscope 112, 899–905.

GUNDERSEN, H.J., BAGGER, P., BENDTSEN, T.F., et al.(1988). The new stereological tools: disector, fractionator,nucleator and point sampled intercepts and their use in patho-logical research and diagnosis. APMIS 96, 857–881.

HAYASHI, A., YANAI, A., KOMURO, Y., NISHIDA, M., IN-OUE, M., and SEKI, T. (2004). Collateral sprouting occurs fol-lowing end-to-side neurorrhaphy. Plast. Reconstr. Surg. 114,129–137.

KANJE, M., ARAI, T., and LUNDBORG, G. (2000). Collat-eral sprouting from sensory and motor axons into an end toside attached nerve segment. Neuroreport 11, 2455–2459.

KAYIKCIOGLU, A., KARAMURSEL, S., AGAOGLU, G.,KECIK, A., CELIKER, R., and CETIN, A. (2000). End-to-

side neurorrhaphies of the ulnar and median nerves at thewrist: report of two cases without sensory or motor im-provement. Ann. Plast. Surg. 6, 641–643.

KOSTAKOGLU, A. (1999). Motor and sensory reinnervationof the hand in an end-to-side median to ulnar nerve coapta-tion in the forearm. Br. J. Plast. Surg. 52, 404–407.

KUBEK, T., KYR, M., HANINEC, P., SAMAL, F., andDUBOVY, P. (2004). Morphological evidence of collateralsprouting of intact afferent and motor axons of the rat ulnarnerve demonstrated by one type of tracer molecule. Ann.Anat. 186, 231–234.

LUTZ, B.S., CHUANG, D.C.C., HSU, J.C., MA, S.F., andWEI, F.C. (2000). Selection of donor nerves—an importantfactor in end-to-side neurorrhaphy. Br. J. Plast. Surg. 53,149–154.

MATSUMOTO, M., HIRATA, H., NISHIYAMA, M.,MORITA, A., SASAKI, H., and UCHIDA, A. (1999).Schwann cells can induce collateral sprouting from intact ax-ons: experimental study of end-to-side neurorrhaphy using aY-chamber model. J. Reconstr. Microsurg. 15, 281–286.

MCCALLISTER, W.V., TANG, P., and TRUMBLE, T.E.(1999). Is end-to-side neurorrhaphy effective? A study of ax-onal sprouting stimulated from intact nerves. J. Reconstr. Mi-crosurg. 15, 597–603.

Mennen, U. (1998). End-to-side nerve suture in the human pa-tient. Hand Surg. 3, 7–15.

MENNEN, U. (1999). End-to-side nerve suture—a techniqueto repair peripheral nerve injury. S. Afr. Med. J. 89,1188–1194.

MENNEN, U. (2003). End-to-side nerve suture in clinical prac-tice. Hand Surg. 8, 33–42.

NOVIKOVA, L., NOVIKOV, L., and KELLERTH, J.O.(1997). Persistent neuronal labeling by retrograde fluorescenttracers: a comparison between Fast Blue, Fluoro-Gold andvarious dextran conjugates. J. Neurosci. Methods 74, 9–15.

PUIGDELLIVOL-SANCHEZ, A., PRATS-GALINO, A., RU-ANO-GIL, D., and MOLANDER, C. (2000). Fast blue anddiamidino yellow as retrograde tracers in peripheral nerves:efficacy of combined nerve injection and capsule applicationto transected nerves in the adult rat. J. Neurosci. Methods 95,103–110.

PUIGDELLIVOL-SANCHEZ, A., VALERO-CABRE, A.,PRATS-GALINO, A., NAVARRO, X., and MOLANDER,C. (2002). On the use of fast blue, fluoro-gold and diamidinoyellow for retrograde tracing after peripheral nerve injury:uptake, fading, dye interactions, and toxicity. J. Neurosci.Methods 115, 115–127.

RICHMOND, F.J., GLADDY, R., CREASY, J.L., KITA-MURA, S., SMITS, E., and THOMSON, D.B. (1994). Effi-cacy of seven retrograde tracers, compared in multiple-la-belling studies of feline motor neurones. J. Neurosci.Methods 53, 35–46.

SANANPANICH ET AL.

874

ROVAK, J.M., CEDERNA, P.S., MACIONIS, V., UR-BANCHEK, M.S., VAN DER MEULEN, J.H., and KUZON,W.M., JR. (2000). Termino-lateral neurorrhaphy: the func-tional axonal anatomy. Microsurgery 20, 6–14.

ROVAK, J.M., CEDERNA, P.S., and KUZON, W.M., JR.(2001). Terminolateral neurorrhaphy: a review of the litera-ture. J. Reconstr. Microsurg. 17, 615–624.

RUOCCO, I., CUELLO, A.A., and RIBEIRO-DA-SILVA, A.(2000). Peripheral nerve injury leads to a novel pattern ofsympathetic fibre innervation in the rat skin. J. Comp. Neu-rol. 422, 287–296.

SANANPANICH, K., MORRISON, W.A., and MESSINA, A.(2002). Physiologic and morphologic aspects of nerve re-generation after end-to-end or end-to-side coaptation in a ratmodel of brachial plexus injury. J. Hand Surg. Am. 27,133–142.

SANGSTER, C.L., GALEA, M.P., FAN, R., MORRISON,W.A., and MESSINA, A. (1999). A method for processingfluorescent labelled tissue into methacrylate: a qualitativecomparison of four tracers. J. Neurosci. Methods 89,159–165.

TARASIDIS, G., WATANABE, O., MACKINNON, S.E.,STRASBERG, S.R., HAUGHEY, B.H., and HUNTER, D.A.(1997). End-to-side neurorrhaphy resulting in limited sensoryaxonal regeneration in a rat model. Ann. Otol. Rhinol. Laryn-gol. 106, 506–512.

TARASIDIS, G., WATANABE, O., MACKINNON, S.E.,STRASBERG, S.R., HAUGHEY, B.H., and HUNTER, D.A.(1998). End-to-side neurorraphy: a long-term study of neuralregeneration in a rat model. Otolaryngol. Head Neck Surg.119, 337–341.

VITERBO, F., TRINDADE, J.C., HOSHINO, K., and MAZ-ZONI, A. (1992). Latero-terminal neurorrhaphy without re-moval of the epineurial sheath: experimental study in rats.Sao Paulo Med J. 267, 267–275.

WOOLF, C.J., SHORTLAND, P., and COGGESHALL, R.E.(1992). Peripheral nerve injurry triggers central sprouting ofmyelinated afferents. Nature 355, 75–78.

XIONG, G., LING, L., NAKAMURA, R., and SUGIURA, Y.(2003). Retrograde tracing and electrophysiological findingsof collateral sprouting after end-to-side neurorrhaphy. HandSurg. 8, 145–150.

YUKSEL, F., PEKER, F., and CELIKOZ, B. (2004). Two ap-plications of end-to-side nerve neurorrhaphy in severe upperextremity nerve injuries. Microsurgery 24, 363–368.

ZHANG, F., and FISCHER, K.A. (2002). End-to-side neuror-rhaphy. Microsurgery 22, 122–127.

ZHANG, F., CHENG, C., CHIN, B.T., et al. (1998). Results oftermino-lateral neurorrhaphy to original and adjacent nerves.Microsurgery 18, 276–281.

ZHANG, Z., SOUCACOS, P.N., BO, J., and BERIS, A.E.(1999). Evaluation of collateral sprouting after end-to-sidenerve coaptation using a fluorescent double-labeling tech-nique. Microsurgery 19, 281–286.

Address reprint requests to:Aurora Messina, Ph.D.

Bernard O’Brien Institute of Microsurgery42 Fitzroy Street

Fitzroy, 3065 Victoria, Australia

E-mail: messinaa@svhm.org.au

END-TO-SIDE NERVE REPAIR IN THE BRACHIAL PLEXUS

875