THE JOURNAL OF COMPARATIVE NEUROLOGY 311:197-209 (1991)

Central Course of Digital Axons Within the Median Nerve of Macaca Mulatta THOMAS M.E. BRUSHART Departments of Orthopaedics and Neurology and the Neuropathology Laboratories, Johns Hopkins Hospitals and The Raymond M. Curtis Hand Center, Baltimore, Maryland 21218

ABSTRACT The traditional view that axons are not functionally grouped within proximal human nerve is based on the interfascicular dissections of Sunderland (’45).However, microstimulation and microneurography (Schady et al., ’83a; Hallin, ’90) reveal proximal grouping of cutaneous sensory axons from small areas of skin. In the present studies, conjugates of horseradish peroxidase with wheat germ agglutinin (HRP-WGA) were used to trace the course of digital nerve axons within the median nerve of Macaca mulatta. The electrophysiologic findings were confirmed, suggesting the potential for precise surgical realignment of functionally related axons even after proximal nerve transection. Radial digital nerves were labeled in the thumb (bilateral, 1 animal), the index finger (unilateral, 2 animals), and the middle finger (bilateral, 1animal). Median nerve cross sections were cut at 1-cm intervals, treated with tetramethyl benzidine to demonstrate HRP-WGA within axons, and compiled to form maps of each digital nerve “territory” within the median nerve. These territories were limited to a single, densely labeled fascicle at the wrist level. They expanded somewhat in the forearm to encompass clusters of labeled axons within a matrix of unlabeled axon profiles. The clusters were more loosely packed in the arm, occupying 1/3 to 1/6 of the nerve cross section a t the entrance to the brachial plexus. The three digital nerve territories studied were widely separated at the wrist level. In the proximal arm, there was moderate intermingling of axons from adjacent digits, but those to the middle finger and thumb remained segregated. Territory configuration differed widely overall, but was moderately constant for each digit. The location of territories within the nerve was often strikingly similar from right to left and from animal to animal, with occasional prominent variations reflecting isolated rotation of one nerve. Key words: peripheral nerve, HRP-WGA, sensory system, somatotopy, axon tracing

Human peripheral nerve is readily divisible into fascicular subunits (Prochaska, 1779). These fascicles do not remain separate and parallel throughout the nerve but participate in frequent interconnections, forming an intraneural plexus. The extent to which individual fascicles represent discrete functions was closely examined at the beginning of this century (see Discussion). This work culminated in the suggestion by Langley and Hashimoto (’17) that fascicular subunits be individually sutured to better direct regenerating axons to appropriate distal pathways. At that time, epineurial suture was the most frequent means of nerve repair. The external epineurium, the outer covering that holds the fascicles together, was reapproximated and sutured; this occured without visualization, and thus without control, of internal fascicular alignment. The results obtained with this technique were often poor. Nearly 50 years would pass before development of the microsutures and microinstruments needed for suture of individual fascicles. By this time, Sunderland’s (’45) O

1991 WILEY-LISS, INC.

descriptions of intraneural complexity and disorganization had discouraged surgeons from opening the epineurium. Fascicular suture was tried in simple experimental models (Bora, ’67)but rarely applied in clinical practice. The extensive maps of intraneural topography prepared by Sunderland and coworkers (Sunderland, ’45; Sunderland and Ray, ’48; Sunderland et al., ’59) are accurate in their depiction of distal nerve segments. However, because of the constraints imposed upon dissection by interfascicular connections (see Discussion), these maps cannot precisely identify the proximal location of a peripherally defined group of mons. Instead, they represent the sum of all potential locations which these axons might occupy. Proximal grouping of peripheral axons could occur but escape detection by dissection technique. Discovery of a greater Accepted May 15,1991. Address reprint requests to Dr. Thomas M.E. Brushart, 1400 Front Ave., Lutherville,MD 21093.

Figure 1

MEDIAN NERVE MAPPING degree of localization would therefore refine rather than contradict Sunderland's findings. Intraneural recording techniques, which trace axons rather than fascicles, have in fact shown sensory axons from small cutaneous areas to be grouped together in the proximal human median nerve (Schady et al., '83b; Hallin, '90). If confirmed anatomically, these findings would suggest a possible role for precise alignment of functionally related axon groups during repair of proximal human nerve. Recent advances in HRP histochemistry (Mesulam, '78), combined with enhancement of HRP uptake through conjugation with WGA (Gonatas et al., '79; Brushart and Mesulam, '80), have provided an anatomical technique for tracing axons over long distances (Brushart, '86). The proximal location of a peripherally defined population of axons can be determined; these axons can be traced through intraneural plexi or located within fascicles, and their overall organization can be mapped. Anatomical confirmation of the electrophysiologic findings can now be sought. This study examines digital nerve projections in six median nerves of four juvenile Macaca mulatta. Central projections to the spinal cord and brainstem in these specimens have been described by Brown et al. ('89) and Culberson et al. ('89), and descriptions of the brachial plexi and dorsal root ganglia are in preparation. Digital nerve axons were found to occupy discrete "territories" within the median nerve, even at proximal levels. These results confirm the electrophysiologic findings of Schady et al. ('83a) and Hallin ('90). The long-accepted picture of intraneural chaos, which reflects the limitations of dissection technique, is thus replaced by a view of partial localization of distal function at proximal levels. This revised concept of intraneural anatomy is of potential importance to the surgeon. Appreciation of functional localization at proximal levels should lead to clinical evaluation of surgical techniques which provide more precise alignment of functionally related areas in proximal and distal nerve stumps.

METHODS Experiments were performed on the upper extremities of four female Rhesus monkeys (Macaca mulatta), ages 9-12 months. These animals were housed and surgeries were performed at the Johns Hopkins Medical Institutions primate facilities in strict compliance with NIH guidelines. Central projections were studied from the radial digital nerves of the thumb (3), index finger (31, and middle finger

Fig. 1. The median nerve at the carpal tunnel level (1cm distal to the radial styloid) after exposure of the radial digital nerve of the middle finger to HRP-WGA. A. Photomicrograph of an uncounterstained 80-p,mcross section of the entire nerve. The volar surface of the nerve is superior and the radial edge is on the left. The nerve has divided into multiple fascicles at this level; labeled axons are confined to the ulnar portion of a single fascicle near the ulnar edge of the nerve. Bar, 0.5 mm. B.Photomicrograph at higher power of the labeled fascicle shown in A. HRP-WGA reaction product is present in both large and small myelinated axons. Focusing through the depth of the section reveals that almost all myelinated axons within the bounds of the digital nerve territory are labeled. Ill-defined areas of reaction product not within myelinated axon profiles are thought to correspond to HRP-WGA within unmyelinated axons (vide infra). Bar, 25 km. C. Electronmicrograph prepared from a section adjacent to that in A and B. Crystaline HRP-WGA reaction product is identified within a smaller myelinated axon. Bar, 1 p,m. D. HRP-WGA reaction product within an unmyelinated axon. Bar, 1km.

199 (2). Animals were anaesthetized with ketamine hydrochloride (Ketaset, Bristow Labs), 10 mg/Kg/hr, delivered intramuscularly, and all surgery was carried out under sterile conditions. The limbs were exsanguinated with an elastic bandage and a pediatric blood pressure cuff on the upper arm was inflated to 150 mm of mercury to provide a bloodless surgical field. The digital nerve under study was then isolated at the proximal interphalyngeal joint level in preparation for exposure to HRP-WGA. Nerve branches supplying adjacent cutaneous territories were approached through separate, proximal incisions and ligated to prevent factitious uptake and transport of HRP-WGA through unintended pathways. This denervation was not sufficient to interrupt feeding and grooming activities or to cause the animal to injure the denervated area. The tourniquet was deflated after 20-30 minutes, hemostasis was obtained with a bipolar cautery, and the proximal wounds were closed. The previously exposed digital nerve was then transected adjacent to the proximal interphalyngeal joint, intraneural hemostasis was obtained with topical thrombin (Thrombostat, Parke Davis), and the cut nerve was encased in a Vaseline well and exposed to a 20% solution of HRP-WGA (Mesulam, '82) for 2-3 hours. The same digital nerve was exposed to HRP-WGA on 3 successive days, after which 48 hours were allowed for additional central transport of the enzyme. At the termination of the experiment, animals were deeply anaesthetized with Ketamine and immobilized supine with the shoulders abducted go", the elbow fully extended, and the wrist in neutral position. The chest was opened, the descending thoracic aorta was clamped, and the animal was perfused through the left ventricle with 250 ml of warm normal saline, followed by 2L of 1/4 strength Karnovsky's fixative in 0.1 M Sorenson's phosphate buffer over 30 minutes and 2L of chilled 15% sucrose in 0.1 M Sorenson's phosphate over an additional 30 minutes. The volar surface of the median nerve was then exposed from palm to axilla, noting relationships between nerve and adjacent fascia, tendon and muscle. An 8-0 running suture was placed within the epineurium on the volar surface of the nerve to allow precise orientation of tissue sections. Nerve branches were identified and their location relative to the radial styloid was measured. A photographic record was made of each dissection before the nerves were removed and embeded in albumidgelatin. Eighty-micron frozen sections were cut at 1-cm intervals and were reacted with H,O, and tetramethyl benzidine (Mesulam, '82) to demonstrate HRP-WGA within axons. Sections from each level were mounted on albumin subbed slides and prepared by two separate methods. Thorough air drying followed by neutral red counterstain, dehydration in graded alcohols, and xylene-permount coverslipping produced a stable HRP reaction product, but also resulted in marked shrinking of neural tissue. Alternatively, rapid alcohol dehydration of moist tissue sections without counterstain minimized tissue shrinking, but produced an unstable reaction product. Tissue from one median nerve was cut at 80 km on a vibratome, embeded in Epon-Araldite, and thin sections were examined and photographed through an Hitachi H600 electron microscope to identify the size classes of axons containing HRP reaction product. Nerve cross sections were examined and photographed through a Nikon Optiphot microscope. At each level all sections were evaluated, and that with the best labeling and least distortion was selected for mapping. A drawing tube,

T.M.E. BRUSHART

200

Figure 2

MEDIAN NERVE MAPPING total magnification 150x, was used to map the areas containing HRP reaction product within myelinated axons. Labeling was considered to be “dense” if labeled axons were separated from each other by a maximum of three unlabeled axon profiles.Additional “sparse” labelingwas mapped separately if axons in a group were separated by more than three unlabeled axon profiles, or if three or fewer labeled axons lay in a group separated from the nearest other axons by more than three unlabeled profiles. The aggregate of densely and sparsely labeled areas was considered to be the “digital nerve territory” within the median nerve. Drawings from the left median nerve were reversed to facilitate comparison of data from both arms.

RESULTS Rapid perfusion fixation of intact extremities facilitated correlation of the external contour of the median nerve with the location of musculotendinous structures impinging upon it. Cross sections of the nerve were obtained at 10-mm intervals from the radial styloid (0 mm), a constant bony landmark on the radial aspect of the wrist (see Figs. 3-6). The most distal sections (-10 mm) were inside the carpal tunnel, where terminal sensory and motor fascicles were encased within a single epineurium. Proximal to the carpal tunnel, the median nerve lay directly beneath the skin before passing deep between the tendons of the ring finger sublimis and the flexor carpi radialis (10-20 mm), which moulded it into a triangular configuration. It then coursed proximally in the plane between the superficialis and profundus muscles (30-50 mm). The nerve was progressively flattened while entering the antecubital fossa between pronator teres and brachialis muscles, simultaneously rotating from the anterior forearm to the medial arm (60-90 mm). It regained an oval or triangular shape at the proximal margin of the antecubital fossa, but was again flattened in the dorso-volar plane between triceps and brachialis muscles. A more rounded configuration was achieved in the proximal arm as the nerve joined the brachial plexus. The median nerve branches between the carpal tunnel and brachial plexus were similar in all specimens, varying

Fig. 2. Photomicrographs of uncounterstained 80-pm cross sections of the median nerve shown in Figure 1.The radial digital nerve of the middle finger has been exposed to HRP-WGA. A. Section from the mid forearm, 40 mm proximal to the radial styloid. Labeled axons are clustered within the digital nerve territory on the dorso-ulnar aspect of the nerve. Black circle = cross section of orienting suture within the epineurium. Bar, 0.25 mm. B.Magnified view of a portion of the digital nerve territory seen in A. HRP-WGA reaction product occupies both large and small myelinated axon profiles. Labeled axons are grouped in tight clusters, with irregular layers of unlabeled axons among the groupings. Bar, 10 pm. C. Section from the proximal forearm, 60 mm from the radial styloid. The digital nerve territory is now an elongated strip oriented from radio-volar to dorso-ulnar. Same magnification as A. D. Enlargement of a portion of the digital nerve territory in C. The number and density of labeled axons and the degree of clustering are similar to that at 40 mm (A and B). Same magnification as B. E.Section from the proximal elbow (90 mm), with several daughter fascicles destined to become forearm motor branches and the communicating branch to the u l n a nerve. Labeled axons within the core of the digital nerve territory are less densely packed, and there is a large area of sparse labeling about the periphery. Same magnification as A. F. Section at the entrance to the brachial plexus (140 mm). Still more axonal dispersion has occurred, yet the digital nerve territory remains well-defined.Same magnificationas A.

201 slightly in their level of departure from the nerve and in the distribution of motor branches in the antecubital fossa. The location of each branch was measured in millimeters from the radial styloid. The branches in a characteristic specimen (see Fig. 4R)were: 33 mm-palmar cutaneous branch; 62 mm-branch to profundus muscle; 75-90 mm, radial aspect of nerve, 2 branches to pronator teres: ulnar aspectanterior interosseous, ulnar communicating branch, branches to sublimis and flexor carpi radialis; 125 mmmusculocutaneous communicating branch. Six of the eight nerves exposed were suitable for axon mapping. HRP reaction product was clearly visualized within myelinated axons of all sizes with the light microscope (Figs. lB, 2B, D), and within small myelinated (Fig. 1C) and unmyelinated (Fig. 1D) axons with the electron microscope. Clusters of unmyelinated axons seen electron microscopicallycorresponded to dark aggregates of reaction product seen among but outside myelinated axon profiles at the light microscopic level (Fig. 2B, D). Mapping of digital nerve territories was based on the distribution of reaction product within myelinated profiles, but these always included the areas of suspected unmyelinated axon labeling. At the wrist, 3-4 cm proximal to the site of enzyme exposure, labeled axons occupied an entire fascicle (radial digital nerve of thumb, RDT) or a discrete portion of a fascicle (radial digital nerve of index, RDI, and radial digital nerve of middle, RDM). Comparison of serial sections at this level revealed that virtually all myelinated profiles within a digital nerve territory were labeled. Slight axon dispersion occurred in the distal forearm (Fig. 2A), with myelinated and possible unmyelinated axons containing HRP-WGA grouped in clusters separated by unlabeled profiles. Labeled axons at the periphery of the digital nerve territory were more dispersed than those located centrally. Though changes in the contour and location of the digital nerve territory were seen in the forearm, the spacing of labeled axons remained relatively constant. The median nerve almost doubled in size at the elbow as motor branches to the forearm were added (Fig. 2E). As these fascicles became integrated within the nerve, both densely and sparsely labeled components of the digital nerve territory also increased in size. Variable degrees of further axon dispersion occurred in the arm; at the entrance to the brachial plexus, 1/6 to ‘/4 of the total nerve cross section was occupied by dense labeling with a variable amount of additional sparse labeling. Radial digital nerves were labeled in the thumb (bilateral, 1 animal), the index finger (unilateral, 2 animals), and the middle finger (bilateral, 1animal). Labeled axons from the right RDT (Fig. 3) occupied a dorso-radial position within the median nerve at the carpal tunnel level. The long axis of the nerve rotated 90” clockwise in the distal forearm (10 mm), bringing the digital nerve territory into a volar-radial position. Subsequent counterclockwise derotation of the axis was accompanied by a progressive shift in this territory to radial (30 mm) and dorsal (40 mm-70 mm) positions. This dorsal location was maintained throughout the antecubital fossa (80-90 mm), with a gradual extension of sparse (110 mm) and then dense (130 mm) labeling volarly on the radial side of the nerve, so that the digital nerve territory was in the straight radial position at the entrance to the brachial plexus. Axons from the contralateral RDT of the same animal were slightly more posterior within the carpal tunnel. However, both changes in nerve configuration and in the location of the digital nerve territory were strikingly

R

L

R

98

100

110

120

(Q

60

70

130

8

Figure 3

140

MEDIAN NERVE MAPPING similar from side to side in the forearm and antecubital fossa. The radial shift in dense labeling occurred more rapidly in the left arm (110 mm), but the remainder of the axon course was similar. Exposure of the right RDI (Fig. 4) to HRP-WGA produced reaction product confined to a single midvolar fascicle within the carpal tunnel. This digital nerve territory shifted to a more radial position in the mid forearm (40 mm) as the nerve rotated counterclockwise. Within the antecubital fossa (70 mm), the territory became elongated to form a strip extending from volar-ulnar to dorso-radial across the central portion of the nerve. More proximally in the antecubital fossa labeled axons were again found on the dorso-radial surface of the nerve, where they remained throughout the arm. Axons from the right RDI of a different animal (Fig. 4)were identically located within the carpal tunnel. However, the longitudinal axis of the nerve rotated clockwiseto assume a dorso-volar orientation in the distal forearm (10 mm), and this rotation correspondedto a shift in the the digital nerve territory to a dorso-ulnar position (30 mm). The long axis of the nerve derotated in a counter-clockwise direction upon entering the antecubital fossa (60-70 mm), restoring the digital nerve territory to a location identical to that of the other animal for the remainder of its proximal course. Exposure of the right RDM labeled a digital nerve territory near the ulnar edge of the nerve at the carpal tunnel level (Fig. 5 ) . This territory assumed a triangular configuration on the dorso-ulnar aspect of the nerve throughout most of the forearm ( 1 0 4 0 mm). It then elongated to become a longitudinal strip, which ran from radio-volar to dorso-ulnar across the center of the nerve (50-70 mm). This strip was compressed centrally within the antecubital fossa and continued proximally in a predominately central location with occasional lateral extensions similar to those observed more distally (90 mm, 130 mm). The territory of the contralateral left RDM was strikingly similar at carpal tunnel and forearm levels. However, as it entered the antecubital fossa, the nerve underwent a clockwiserotation, which was accompaniedby a corresponding shift in the digital nerve territory. The longitudinal strip now extended from dorso-radial to volar-ulnar, and maintained this orientation throughout the remainder of the nerve.

DISCUSSION The internal architecture of peripheral nerve was first described by Prochaska (1779). He dissected human nerves into their fascicular subunits, recognizing that fascicles do not remain separate throughout the nerve but form an

Fig. 3. Camera lucida drawings of median nerve cross-sections. In both right and left upper extremities the radial digital nerve of the thumb was exposed to HRP-WGA. Sections were taken at 10-mm intervals with reference to the radial styloid, and those from the left arm have been reversed to facilitate comparison of left and right sides. Volar is up and radial is to the left, as if one were looking proximally at the cut nerve surface in the right arm. Solid black areas represent "dense" labeling, where HRP-containing myelinated profiles are separated by no more than three unlabeled axons. Stippled areas represent "sparse" labeling, where axons in a group are separated by more than three unlabeled axons, or where three or fewer labeled axons lie in a group separated from the nearest other axons by more than three unlabeled profiles. Bar, 1 mm.

203 intraneural plexus. Intraneural anatomy was then neglected until the early twentieth century, when dissection, Wallerian degeneration, and electrical stimulation techniques were used, along with clinical observation, to study the location of functionally related axons within peripheral nerve. Stoffel ('13, '15) described the nerve as a cable, with exact correspondence of proximal and distal fascicles. This view was shared by Putti ('16) and Barile ('17). However, plexiform fascicular interconnections were demonstrated within the median, ulnar, and sciatic nerves by Heinemann ('16), within the median, ulnar, and radial nerves by Dustin ('18), and within the sciatic nerve by Compton ('171, Langley and Hashimoto ('17), McKinley ('21), and Goldberg ('24). On the basis of these dissections,intraneural architecture was viewed as continuously changing, with grouping of functionally related axons only near distal branchpoints. The widespread distribution of axons undergoing Wallerian degeneration after partial high nerve section (McKinley, '21; Kilvington, '40) was similarly interpreted as evidence against intraneural localization. The results of electrical stimulation experiments performed during this period varied according to technique and proximddistal location. Stimulation around the circumference of intact human nerve (Marie et al., '15; Kraus and Ingham, '20) defined a reproducible localization of fibers destined for specific muscles even at proximal levels. Stimulation of single cat peroneal nerve fascicles at the knee level resulted in contraction of individual muscles (Langley and Hashimoto, '17), whereas stimulation of single peroneal or tibia1 fascicles at the ischial tuberosity resulted in mass contractions (McKinley, '21). Though interpreted respectively as evidence for and against intraneural localization, the fascicular stimulation results are consistent with progressive localization of axons as they course distally. Clinical study of partial nerve lesions suggested poor localization, as Sherren ('08)described no significant motor or sensory defecit after section of one-third of a proximal nerve trunk. More recent studies of intraneural organization with dissection and degeneration techniques have refined earlier observations. In addition, an entirely new perspective has been provided by techniques of intraneural microstimulation, microneurography, and horseradish peroxidase (HRP) axon tracing. The most detailed anatomic and histologic explorations of intraneural organization were performed by Sunderland ('45), Sunderland and Ray ('481, and Sunderland et al. ('59). They determined that, despite the changing plexiform nature of the fascicular pattern, axons of peripheral branches are well localized to individual fascicles or fascicular groups for variable, but often considerable distances proximal to their exit from the nerve. Sunderland's findings were confirmed by Tamura's ('69) less extensive analysis of peripheral nerves in Japanese and by Perotto and Delagi's ('79) clinical study of partial median nerve lacerations at the wrist level. More recently, Jabaley et al. ('80) and Chow et al. ('85) have documented distal axon localization over distances somewhat greater than those found by Sunderland ('45). However, intraneural plexus formation continues to defeat purely anatomical attempts to determine the proximal location of axons identified at the periphery. The results of degeneration studies have varied according to the proximo-distal level of nerve interruption. Spinal root fibers take a specific, well-localized course through the sciatic nerve of the dog (Ueyama, '78) and tend to be

R

R'

5O

60

Q

Fig. 4. Median nerve cross sections after exposure of the right radial digital nerve of the index finger t o HRP-WGA in two separate animals. Bar,1mm.

L

R -10

8o 0 10

20

n

30

40

50

60 130

%

140

70

Fig. 5. Median nerve cross sections after exposure of the radial digital nerve of the middle finger to HRP-WGA in right and left arms of a single animal. Bar, 1mm.

206

clustered together within the extramuscular portion of the rat phrenic nerve (Laskowski and Sanes, '87). Partial ligation of the horse recurrent laryngeal nerve, in contrast, resulted in widespread degeneration within the distal nerve (Dyer and Duncan, '87). This discrepancy may reflect differences in distribution between an anatomically defined ventral root and a randomly chosen axon population. Intraneural microstimulation and microneurography techniques have circumvented the limitations imposed on anatomical dissection by intraneural plexus formation. When stimulatingintact human median nerve axons with a tungsten electrode, Schady et al. ('83a) found that 42% of fascicles studied in the upper arm projected only to skin, and within these 67% of sensations elicited were confined to a single digital interspace. Axons with adjacent or overlapping cutaneous projections were not directly adjacent within the fascicle (Schady et al., '83b). Similar intrafascicular dispersion was found by Dyck et al. ('72), Hess et al. ('79), and Parry et al. ('81) in their studies of ischemic neuropathy. However, Hallin ('90) was able to demonstrate intrafascicular axon localization by performing microneurography with a concentric electrode rather than the tungsten electrode used by Schady et al. ('83a,b). Functionally related axon groups may thus be confined to individual fascicles at proximal levels and may even be localized within these fascicles. Early attempts at axon tracing with HRP were severely limited by reliance on electron microscopy to demonstrate the sparse HRP reaction product within axons (Malmgren et al., '77). Subsequent improvements in histochemical technique (Mesulam, '78) dramatically increased the amount of reaction product present, making it possible to trace the proximal course of peripheral axons over several centimeters with the light microscope (Thomander et al., '82). However, early attempts at continuous tracing of primate upper extremity axons with unconjugated HRP proved unsuccessful (Brushart, unpublished data). Conjugates of HRP with wheat germ agglutinin (HRP-WGA)(Gonatas et al., '79) were subsequently found to produce dramatic labeling of central sensory projections after previously fruitless application techniques such as intradermal injection (Brushart and Mesulam, '80). These observations stimulated renewed interest in primate axon tracing. Results with HRP-WGA in adult rhesus monkeys were improved but still unsatisfactory. In further studies, juvenile monkeys, with shorter limbs and thus shorter axons with smaller axoplasmic volumes, proved ideal. After repeated daily exposure of digital nerves to HRP-WGA, median nerve axons could be continuously labeled for up to 25 cm, with excellent labeling of central sensory projections (Brushart, '86). Axon transport techniques could, for the first time, identify the proximal location of an entire population of axons, which had been defined at the periphery. These axons could be traced through intraneural plexi or located within fascicles, and their overall organization could be precisely mapped. This study has examined several aspects of primate intraneural topography: (1) the spacing amongst individual myelinated digital nerve axons as they course proximally, (2) the degree to which digital nerve territories are localized throughout the nerve, (3)the relative location of territories from various digital nerves, and (4) the replicability of these locations from side to side and from animal to animal. Each aspect is discussed in turn. In the juvenile rhesus monkey, myelinated axons of a single digital nerve were tightly packed at the wrist level

T.M.E. BRUSHART (Fig. lA,B).Within the forearm, labeled axons were grouped in clusters separated by unlabeled axon profiles (Fig. 2B,D). In the arm, labeling of smaller myelinated and potentially unmyelinated axons became attenuated, leaving progressively looser clusters of large myelinated axons which persisted to the brachial plexus (Fig. 2E,F). The digital nerve territory can thus be thought of as an elongated cone with distal apex and proximal base. As the cross-sectional area of this cone increases in the arm, the axons that define it are progressively dispersed. This dispersion allows for overlap of adjacent digital nerve territories at proximal levels. Complementary electrophysiologic data have been provided by Schady et al. ('83b1, who stimulated small groups of adjacent axons in the median nerves of awake humans with a tungsten electrode. They found the cutaneous projections of these fibers to be more scattered about the hand than one would expect were the nerve organized on a strictly somatotopic basis. However, this scatter was usually confined within the innervation territory of one or two digital nerves and was less prominent at the wrist than in the upper arm. Greater intrafascicular localization was found when similar studies were performed with a concentric electrode (Hallin, '90). This concept of partial intraneural organization is also consistent with the more general information provided by nerve infarction studies (Dyck et al., '72; Hess et al. '79; Parry et al., '81). Similarly, proximal nerve section (Sherren, '08) could partially delete several overlapping territories while leaving enough axons within each to permit useful function. Digital nerve territories as defined by HRP-WGA were well localized throughout the median nerve, occupying only % to 1/6 of the nerve cross-section in the upper arm. Axon labeling at proximal levels did not correspond quantitatively to that at the wrist, largely due to decreased labeling of smaller fibers. In proximal portions of the extremity, selective loss of labeling within a specific fiber population might artificially shrink a digital nerve territory. However, because of the heterogeneity of axon caliber throughout the territory, selective loss of smaller fibers should alter the density of labeling within a territory rather than its overall dimensions. The degree of localization seen with HRPWGA contrasts markedly with the findings of fascicular dissection studies. Sunderland ('45) could only isolate the fibers serving RDT for 2 cm proximal to the radial styloid. Proximal to this their specific identity is lost, and they are mapped as a group with all motor and sensory fibers to the thumb and first webspace. The location of this larger group is in turn lost at the elbow. This pessimistic view results from assumptions which are forced upon the dissector. He functionally identifies a fascicle by its distal termination, then works proximally separating the fascicle from its neighbors. However, when an interfascicular plexus is encountered, he must assume that all proximal components contribute equally to the single distal fascicle. As he repeatedly encounters fascicular interconnections, he is soon tracing a large number of fascicles, most of which do not actually contribute to the fascicle under study. In this way fascicular identity is rapidly lost. Intrafascicular microstimulation studies, which trace axons rather than fascicles, reveal greater correspondence between dstal fascicles and those more proximal in the nerve. Schady et al. ('83a) found that 42%of human median nerve fascicles in the upper arm projected only to skin, and of these 67% projected only to a single digital interspace. At the wrist level, 87% of fascicles projected even more discretely to a single digital nerve. This evidence strongly suggests that axons terminating in a

MEDIAN NERVE MAPPING

RDT-L

207

RDT-R

RDI-R

RDI-R’

RDM-L RDM-R

CARPAL TUNNEL

Q Q

Q Q DISTAL

FOREARM

FOREARM

PROXIMAL

-

DISTAL

Q

ARM

PROXIMAL

ARM

Fig. 6. Camera lucida drawings taken from Figures 3,4,and 5 to compare the location of various digital nerve territories at specific levels of the extremity. Because of limb length differences, the sections within each group are not all equidistant from the radial styloid. Bar, 1mm.

peripheral human median nerve fascicle travel near one another even at proximal levels; they are not widely separated by interfascicular plexus formation, as suggested by Sunderland. Human and nonhuman primates differ in that unbranched nerve segments are multifascicular in the

former and monofascicular in the latter (Brushart, unpublished data). The present HRP-WGA studies thus produce no specific information on the relationship between monkey digital nerve “territories” and human nerve fascicles. However, the overall degree of localization found in mon-

208 keys with HRP-WGA corresponds well to that seen in humans with microstimulation techniques. The relative locations of digital nerve territories as revealed with HRP-WGA are shown in Figure 6. At the wrist level, RDT is radial and posterior, RDI is volar and central, and RDM is ulnar, contacting both dorsal and volar surfaces of the nerve; there is no overlap of territories. These territories lie within those shown by Sunderland ('45) for the appropriate digital interspaces, except that Sunderland's first webspace (containing RDI) is separated from the volar surface of the nerve by thenar motor fibers. In the proximal forearm, RDT is posterior, RDI extends centrally from the midradial aspect of the nerve, and RDM occupies a strip extending from volar-radial to dorso-ulnar. Varying external contour of the nerve at this and proximal levels prevents exact comparison of the locations of digital nerve territories. However, the only significant overlap in the proximal forearm is between one RDI, in a nerve which had undergone 180" rotation in relation to its counterpart in the opposite arm, and RDM. In the proximal arm RDT has a wide radial base and extends dorso-volarly, RDI extends centrally from a narrow radial base, and RDM is central with an ulnar extension. This sequence is consistent with innervation of the digits by progressively lower spinal contributions as one passes from radial to ulnar. There is virtually no overlap of RDT and RDM, but moderate overlap of adjacent territories in both directions. Comparable electrophysiologic data are provided by Schady and coworkers ('83a). They found 2/3 of proximal human cutaneous fascicles projecting to a single digital interspace, with the remaining % projecting to more than one interspace. The three digital nerves studied with HRP-WGA are all from separate interspaces. The moderate overlap of proximal digital nerve territories obtained with HRP-WGA is thus consistent with intermingling of axons destined for separate digital interspaces in ?/3 of proximal human sensory fascicles. The location of digital nerve territory was, with two exceptions, reasonably constant from side to side and from animal to animal. RDT (Fig. 3) was examined in both median nerves of one animal. It slowly moved from volarradial in the proximal forearm to dorso-ulnar at the antecubital fossa. This change in position corresponded to physical rotation of the nerve seen at the time of dissection, a process which occurred symetrically in both arms, so that both territories occupied strikingly similar locations at each level. RDT assumed a more radial position in the arm, a shift which was accomplished more rapidly on the left without apparent reason. RDI (Fig. 4)was examined in the right median nerves of two separate animals. In one, the territory gradually shifted in the forearm to lie in a radial-dorsal position. In the other, it was of similar configuration but underwent a dramatic change in location. Clockwise rotation of the nerve in the proximal forearm shifted the RDI territory to a dorso-ulnar location, 180" opposite to its counterpart, and counterclockwise derotation shifted it back at the entrance to the antecubital fossa. Within the arm both territories were similar in both configuration and location. RDM (Fig. 5 ) territories were symmetrical in both forearms of a single animal. However, they underwent different rotational changes at the elbow; a longitudinal band of axons traversed the nerve from radiovolar to dorso-ulnar on the right, and the opposite on the left. Overall, territories of a given digital nerve were quite similar in configuration; their locational differences often

T.M.E. BRUSHART corresponded to physical rotation of the entire nerve. Rotational and configurational changes in the nerve in response to the musculoskeletal environment are preserved by perfusion fixation; they may thus reflect points in a spectrum rather than permanent attributes of the nerve. Stretching and mounting of fresh nerve for immersion fixation and anatomical dissection appear to eliminate these variations and result in straighter axon trajectories as mapped by Sunderland ('45). The finding of proximal localization of distal function in this and other recent studies (Schady et al., '83a; Hallin, '90) should stimulate re-evaluation of nerve repair technique at proximal levels. One of the surgeon's primary functions is the generation of topographic specificity by approximating like axons in proximal and distal nerve stumps (Brushart, '91). This function is most readily accomplished by fascicular suture in areas of the nerve where axons subserving a given function are confined to specific fascicles. In chosing these areas surgeons most often refer to the classical studies of Sunderland ('45) as updated by Jabaly et al. ('80) and Chow et al. ('85). However, these dissection studies significantly underestimate functional localization at proximal levels for the reasons discussed previously. The evidence presented here suggests that fascicular alignment may be as important at proximal levels as it is distally.

ACKNOWLEDGMENTS The author thanks Mr. Philip Kessens and Mr. David Seiler for their superb technical assistance. Thanks are also due to Ms. Suzanne Merrick for preparation of camera lucida drawings, and to Professor John Griffin for helpful comments on the manuscript.

LITERATURE CITED Barile, C. (1917) Sul reale valore practico della topogrda fascicolare dei nervi periferici (second0 Stoffel) per l'esecuzione dell anastomosi dei nervi degli arti. Policlinico 24t1177-1180. Bora, W. (1967) Peripheral nerve repair in cats: the fascicular stitch. J. Bone Joint Surg. (Am.)49t659-666. Brown, P.B., T.M. Brushart, and L.A. Ritz (1989) Somatotopy of digital nerve projections to the dorsal horn in the monkey. Somatosens. Motor Res. 6:309-317. Brushart, T.M. (1986) The central course of digital axons within the median nerve of Macaca niulatta. SOC.Neurosci. Abst. 12108. Brushart, T.M. (1991) The mechanical and humoral control of specificity in nerve repair. In R.H. Gelberman (ed):Operative Nerve Repair and Reconstruction. Philadelphia: Lippincott (in press). Brushart, T.M., and M.-M. Mesulam (1980) Transganglionic demonstration of central sensory projections with HRP-lectin conjugates. Neurosci. Lett. I7:l-6. Chow, J.A., A.L. VanBeek, D.L. Meyer, and M.C. Johnson (1985) Surgical significance of the motor fascicular group of the ulnar nerve in the forearm. J. Hand Surg. lOAr867-872. Compton, A.T. (1917) The intrinsic anatomy of the large nerve trunks of limbs. J. Anat. 51:103-117. Culberson, J.L., and T.M. Brushart (1989) Somatotopy of digital nerve projections to the cuneate nucleus in the monkey. Somatosens. Motor Res. 6t319-330. Dustin, A.P. (1918) La fasciculation des nerfs. Ambul. Ocean 2:135-154. Dyck, P.J., D.L. Conn, and H. Okazaki (1972) Necrotizing angiopathic neuropathy. Three dimensional morphology of fiber degeneration related to sites of occluded vessels. Mayo Clinic Proc. 47:461-475. Dyer, K.R., and I.D. Duncan (1987) The intraneural distribution of myelinated fibers in the equine recurrent laryngeal nerve. Brain IlO:15311543.

MEDIAN NERVE MAPPING Goldberg, I. (1924) The internal architecture of the tibial, peroneal and obturator nerves. Am. J. Anat. 3 2 4 4 7 4 6 5 . Gonatas, N.K., C. Harper, T. Mizutani, and J. Gonatas (1979) Superior sensitivity of conjugates of horseradish peroxidase with wheat germ agglutinin for studies of retrograde axonal transport. J. Histochem. Cytochem.27:728-734. Hallin, R.G. (1990) Microneurography in relation to intraneural topography: somatotopic organisation of median nerve fascicles in humans. J. Neurol. Neurosurg. Psychiatry 53:73&744. Heinemann, 0. (1916) Ueber Schutzverletzungen der peripheren Nerven. Nebst anatomischen Untersuchungen iiber den inneren Bau der grossen Nervenstamme. Arch. klin. Chir. 108:107-150. Hess, K., R.A. Eames, P. Darveniza, and R.W. Gilliatt (1979) Acute ischemic neuropathy in the rabbit. J. Neurol. Sci. 4431943. Jabaley, M.E., W.H. Wallace, and F.R. Heckler (1980) Internal topography of major nerves of the forearm and hand. J. Hand Surg. 51-18. Kilvington, B. (1940) A note on the distribution of myelinated nerve fibers within nerve bundles. Aust. 3. Exp. Biol. Med. Sci. 18:367-368. Kraus, W.M., and S.D. Ingham (1920) Electrical stimulation of peripheral nerves exposed at operation. JAMA 74:58&589. Langley, J.N., and M. Hashimoto (1917) On the suture of separate nerve bundles in a nerve trunk and on internal nerve plexuses. J. Physiol. Lond. 51:318-345. Laskowski, M.B., and J.R. Sanes (1987) Topographic mapping of motor pools onto skeletal muscles. J. Neurosci. 7952-260. McKinley, J.C. (1921) The intraneural plexus of fasciculi and fibers in the sciatic nerve. Archs. Neurol. Psychiat. (Chicago)6:377-399. Malmgren, L.T., M.J. Lyon, and R.R. Gacek (1977) Localization of abductor and abductor fibers in the kitten recurrent laryngeal nerve: Use of a variation of the horseradish peroxidase tracer technique. Exp. Neurol. 55:187-198. Marie, M.P., H. Meige, and A. Gosset (1915) Les localisations motrices dans les nerfs periphbriques. Bull. Acad. Med. 74:798-810. Mesulam, M.-M. (1978) Tetramethyl benzidine for horseradish peroxidase neurohistochemistry; a non-carcinogenic blue reaction product with

209 superior sensitivity for visualizing neural af€erents and efferents. J. Histochem. Cytochem. 26:10&117. Mesulam, M.-M. (1982) Tracing Neural Connections with Horseradish Peroxidase. New York: John Wiley & Sons, pp. 122-131. Parry, G.J., and M.J. Brown (1981) Arachidonate-induced experimental nerve infarction. J. Neurol. Sci. 50:123-133. Perotto, A.O., and E.F. Delagi (1979) Funicular localization in partial median nerve injury at the wrist. Arch. Phys. Med. Rehab. 6Or165-169. Prochaska, G. (1779) De Structura Nervorum. Vienna: R. Graeffer. Putti, V. (1916) Sulla topografia fascicolare dei nervi periferici e piu specialmente dello sciatic popliteo esterno. Clinica chir. 24r1021-1035. Schady, W., H.E. Torebjork, and J.L. Ochoa (1983b) Peripheral projections of nerve fibers in the human median nerve. Brain Res. 277949-261. Schady, W., J.L. Ochoa, H.E. Torebjork, and L.S. Chen (1983a) Peripheral projections of fascicles in the human median nerve. Brain 106t74~5760. Sherren, J. (1908) Injuries of Nerves and Their Treatment. London: James Nisbet, pp. 35-44. Stoffel, A. (1913) Beitrage zu einer rationeller Nervenchirurgie. Munch. med. Wschr. 60:175-179. Stoffel, A. (1915) Ueber die Behandlung verletzter Nerven im Kriege. Munch med. Wschr. 62:201-203. Sunderland, S. (1945) The internal topography of the radial, median and u l n a nerves. Brain 68.243-298. Sunderland, S., and L.J. Ray (1948) The intraneural topography of the sciatic nerve and its popliteal divisions in man. Brain 71.242-273. Sunderland, S., R.D. Marshall, and W.E. Swaney (1959) The intraneural topography of the circumflex, musculocutaneous, and obturator nerves. Brain 82:116-129. Tamura, K. (1969) The funicular pattern of Japanese peripheral nerves. Arch. Jap. Surg. 38:35-57. Thomander, L., H. Aldskogius, and G. Grant (1982) Motor fibre organization in the intratemporal portion of cat and rat facial nerve studied with the horseradish peroxidase technique. Acta. Otolaryngol. 93:397405. Ueyama, T. (1978) The topography of root fibers within the sciatic nerve trunk of the dog. J. Anat. 17:277-290.