- Email: [email protected]
Verification of the median-to-ulnar and ulnar-to-median nerve motor fiber anastomosis in the forearm: an electrophysiological study
Clinical Neurophysiology 114 (2003) 94–98 www.elsevier.com/locate/clinph
Verification of the median-to-ulnar and ulnar-to-median nerve motor fiber anastomosis in the forearm: an electrophysiological study Georgios Amoiridis a,*, Ioannis G. Vlachonikolis b a
Department of Neurology, School of Health Sciences, University of Crete, P.O. Box 1393, 71110 Heraklion, Greece Department of Biostatistics, School of Health Sciences, University of Crete, P.O. Box 1393, 71110 Heraklion, Greece
b
Accepted 9 October 2002
Abstract Objective: To estimate the real occurrence of the motor median-to-ulnar nerve anastomosis in the proximal forearm (Martin–Gruber anastomosis, MGA), as its frequency varies between 6 and 44% in the literature and to investigate the incidence of the ulnar-to-median nerve anastomosis in the distal forearm. Methods: Compound muscle action potentials (CMAP) recorded over thenar, hypothenar, and first dorsal interosseus muscle on median or ulnar nerve stimulation at wrist and elbow and collision blocks of the median and ulnar nerve were compared in a group of 50 healthy volunteers. Particular precautions were undertaken in order to avoid false positive results due to stimulus spread to the neighboring nerve. Cases of uncertain MGA were classified as either MGA or non-MGA on the basis of posterior probabilities estimated by discriminant analysis. Results: The estimated frequency of MGA was 54% using the potential comparison method and 46% using the collision technique. An ulnar-to-median nerve anastomosis was not found in any subject. Conclusions: While the MGA is very common, the ulnar-to-median nerve anastomosis is a rarity. Standard nerve conduction studies of the median nerve with CMAP recordings solely over thenar will detect less than 14% of MGA cases. q 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Median-to-ulnar nerve anastomosis; Martin–Gruber anastomosis; Ulnar-to-median nerve anastomosis; Innervation anomaly; Collision block; Compound muscle action potential analysis
1. Introduction The first description of an anastomosis between the median and ulnar nerve in the proximal forearm was in 1763 by the Swedish anatomist Martin (1763). Gruber (1870) estimated its frequency as 15.2% in 250 arms. This median-to-ulnar nerve anastomosis in the proximal forearm was later referred to as Martin–Gruber anastomosis (MGA). Further anatomical studies have shown a frequency of MGA between 44 (Sokolow, 1925, cited by Hirasawa, 1931) and 21.3% (Nakashima, 1993). There are no anatomical studies dealing with the ulnar-to-median nerve anastomosis in the distal forearm. Electrophysiological studies in combination with pharmacological studies (Hopf and Hense, 1974) or collision blocks (Kimura et al., 1976) have revealed a frequency of 6 and 17% for the MGA, respectively. Electrophysiological methods analyzing the compound muscle action potentials * Corresponding author. Tel.: 130-810-392457; fax: 130-810-392474. E-mail address: [email protected] (G. Amoiridis).
(CMAP) over thenar, hypothenar, and first dorsal interosseous muscle (FDI) on median and ulnar nerve stimulation at wrist and elbow have shown MGA in 28 (Crutchfield and Gutmann, 1980) and 34% (Sun and Streib, 1983) of cases. In summary, the frequency of MGA varies between 6 and 44% in the literature. The motor ulnar-to-median nerve anastomosis in the distal forearm is extremely rare (Streib, 1979, Kimura et al., 1976, Amoiridis, 1992). The communications between the median and ulnar nerves in the forearm provide for variations in the innervation of the intrinsic hand muscles, as proved by nerve conduction studies (Kimura et al., 1976, Amoiridis, 1992). The knowledge of these variations is of major importance for a correct interpretation of needle electromyographic (EMG) and nerve conduction studies in cases of suspected median and ulnar nerve lesions. In this study, an attempt was made to estimate the occurrence of the MGA and the ulnar-to-median nerve anastomosis using two methods: CMAP comparison and collision blocks.
1388-2457/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88- 2457(02)0032 8-0
CLINPH 2002646
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
95
2. Materials and methods The left upper extremity of 50 healthy volunteers was examined (30 men, 20 women). Relative to the question of the frequency of the MGA, the age of the subjects was disregarded. Recordings were performed using a 4-channel device (Counterpoint Electromyograph of Dantec, Denmark), which enables a 4-channel recording using two stimulators. The band pass setting was 10–20 kHz. Stimulation was performed using surface electrodes (SE, Dantec, type 13L36). The stimulus intensity was gradually increased in steps of 5 mA, until the CMAP reached its maximal amplitude. After this, the stimulus was increased to another 5–10 mA (stimulus duration 0.2 ms, usually needed stimulus intensity 25 mA at wrist and 40 mA at elbow). Initially, the median and ulnar nerve, were successively stimulated at wrist and elbow (ulnar nerve 1 cm proximal to epicondylus medialis humeri). Subsequently, each of the nerves was simultaneously stimulated at wrist and elbow (collision block). In a second attempt, the stimulus at the elbow was delivered some milliseconds later than the one at the wrist. When criteria for MGA were fulfilled, the stimulating electrode on proximal median nerve stimulation was moved medially towards the medial epicondyle in steps of about 0.5 cm. Disappearance of the MGA-criteria by this maneuver excluded a stimulus spread to the ulnar nerve. When MGA-criteria persisted, teflon-insulated unipolar needle electrodes were used to stimulate the median nerve at elbow. In cases of MGA recognized on screen, simultaneous stimulation of the median nerve at the elbow and the ulnar nerve at wrist was obtained, in order to block the anastomotic fibers. Recordings of the CMAP were obtained by means of SE over thenar (sensitivity 2–10 mV/division), hypothenar (sensitivity 0.5–2 mV/division), and FDI (sensitivity 0.5– 2 mV). In the following discussion, thenar and abductor pollicis brevis (APB) will be used as identical recording sites, and the same will be the case for hypothenar and abductor digiti minimi (ADM). A concentric needle electrode (CNE) was inserted in the flexor carpi ulnaris muscle (FCU, sensitivity 0.5–10 mV), detecting a possible stimulus spread to the ulnar nerve on median nerve stimulation at elbow (Fig. 1). The correct position of the CNE in the FCU was proven by stimulating the median and ulnar nerves at elbow. Only ulnar nerve stimulation at elbow might evoke a potential with a sharp deflection from the base line. If criteria for the existence of a MGA were fulfilled when the CNE already recorded a potential in the FCU, selective-near nerve stimulation of the median nerve at elbow with teflon-insulated unipolar needle electrodes was performed. The amplitudes were measured from the base line to the negative peak. For comparisons of the CMAP, criteria for a MGA were the same as in a previous study (Amoiridis, 1992). In brief, the presence of a MGA was assumed when the CMAP over
Fig. 1. Electrode positioning and stimulating sites. FDI, first dorsal interosseous; CNE, concentric needle electrode; FCU, flexor carpi ulnaris muscle; S, stimulus; M, median; U, ulnar; 1, wrist; 2, elbow.
thenar or FDI was higher on elbow than on wrist stimulation of the median nerve (thenar or FDI presentation) or a potential with a different shape, on elbow stimulation compared with wrist stimulation of the median nerve, was evoked over ADM (hypothenar presentation). For comparisons of the CMAP, the presence of an ulnarto-median anastomosis was assumed when the CMAP over one of the recording sites (thenar, FDI or hypothenar) was higher on elbow than on wrist stimulation of the ulnar nerve. For the collision block technique, a MGA was assumed when on median nerve collision block a CMAP was recorded at least over one of the recording sites (thenar, hypothenar, FDI) with latency similar to the latency of the CMAP evoked by the elbow stimulation of the median nerve. Furthermore, elimination of this CMAP on simultaneous median elbow and ulnar wrist stimulation was required. For the collision block technique, an ulnar-to-median nerve anastomosis was assumed when on ulnar nerve collision block a CMAP was recorded at least over one of the recording sites (thenar, hypothenar, FDI), with latency similar to the latency of the CMAP evoked by the elbow stimulation of the ulnar nerve.
2.1. Statistics The differences of the CMAPs evoked over APB and FDI muscles on wrist (1) and elbow (2) stimulation (S) of the median (M) and ulnar (U) nerve were estimated for each volunteer (SM1-SM2 and SU1-SU2 over APB and FDI, see also Fig. 1). Their distribution was tested for normality by the Kolmogorov–Smirnov and Saphiro–Wilk tests (Armitage and Berry, 1994). Only SM1-SM2 over APB had a normal distribution; the other 3 were not normal (P , 0:05). Their correlation was therefore assessed by means of the non-parametric Spearman’s correlation coefficient rs (Armitage and Berry, 1994). Cases of uncertain MGA were classified as either MGA or non-MGA on the basis of posterior probabilities estimated by discriminant analysis (DA); the latter was carried out by two methods,
96
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
Fig. 2. Potentials recorded on wrist (A) and elbow (B) stimulation of the median nerve (M) in a volunteer without MGA. APB, abductor pollicis brevis (thenar); ADM, abductor digiti minimi (hypothenar); FDI, first dorsal interosseous; FCU, flexor carpi ulnaris. Same potential shape on wrist (A) and elbow stimulation (B). H-reflex in FCU on median nerve stimulation at elbow.
Fisher’s linear DA and logistic discrimination (Vlachonikolis and Marriott, 1982).
3. Results In 25 subjects, the CNE in FCU recorded a potential with a latency of 17–20 ms on median nerve elbow stimulation (Figs. 2 and 3). In 5 subjects, it was recorded also on median wrist stimulation. This potential represents a heteronymous H-reflex, with its afferent fibers in median and the efferent in ulnar nerve (Panizza et al., 1989). The potentials over APB, ADM, FDI, and FCU in one subject without MGA are shown in Fig. 2. The SE over ADM records a positive potential on wrist and elbow median nerve stimulation, which represents the volume conduction of the thenar CMAP (Kimura et al., 1976). Furthermore, the same potential is recorded over FDI as volume conduction potential with a negative deflection, as
Fig. 3. Potentials recorded on wrist (A) and elbow (B) stimulation of the median nerve (M) in a volunteer with MGA. Larger area over APB, negative potential over ADM and higher potential over FDI on elbow stimulation (B) of the median nerve. H-reflex in FCU on median nerve stimulation at elbow.
its origin, the thenar muscles, are located vertically under the SE over FDI. The potentials over APB, ADM, FDI, and FCU in one subject with MGA are shown in Fig. 3. The area of the CMAP over APB after elbow median nerve stimulation is slightly larger than on wrist stimulation of the same nerve. Furthermore, on median nerve stimulation at elbow, a negative CMAP is recorded over ADM and the CMAP over FDI is higher than the CMAP over the same muscle on median wrist stimulation. The collision block of the median nerve and of the anastomotic fibers in a subject with MGA are shown in Fig. 4. The stimulus at the elbow was given 3 ms (Fig. 4A) and 4 ms (Fig. 4B) later than at wrist. In the first 3 traces, a potential with a short latency (about 3.5 ms) is recorded, representing the result of the wrist stimulation of the median nerve (SM1, Fig. 4A). Another potential with a latency of about 9 ms is recorded over ADM and FDI (SM2 in Fig. 4A), representing the potential of the anastomotic fibers, which escape the collision block of the median nerve. These fibers are blocked when stimulating the median nerve at elbow and the ulnar nerve at wrist (Fig. 4B). The positive sharp deflection in the 4th trace (FCU) with latency of 3 ms (Fig. 4A) and 4 ms (Fig. 4B) is the stimulus artifact of the elbow stimulation of the median nerve. 3.1. Frequency of MGA comparing the CMAPs on wrist and elbow stimulation A thenar presentation was found in 15 (30%), a hypothenar presentation in 22 (44%) and a FDI presentation in 29 (58%) cases. In 15 of the 22 hypothenar presentations, a negative CMAP was recorded over ADM on elbow median nerve stimulation and in the remaining 7 cases, a different potential shape, between wrist and elbow median nerve
Fig. 4. Collision block of median nerve (A) in a case of MGA with hypothenar (ADM) and FDI presentation. SM1, potentials due to the wrist stimulation of the median nerve. Stimulation of the median nerve at elbow (SM2) elicited no potential over APB, as the MGA had no thenar presentation. SM2 over ADM and FDI (A) represents the potential of the anastomotic fibers. In (B), stimulating the median nerve at elbow (SM2) and the ulnar nerve at wrist (SU1) blocks the anastomotic fibers (disappearance of the SM2 potentials over ADM and FDI). The stimulus at the elbow was delivered 3 ms later than at wrist in A and 4 ms in B.
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
97
stimulation, was observed. At least one of the criteria for a MGA was fulfilled in 33 of the 50 cases (66%). The 22 hypothenar presentations and one case with FDI presentation (23 cases) were recognized on screen and the stimulus spread to the ulnar nerve on median nerve stimulation at elbow was excluded. The remaining 10 cases were seen later and no stimulus spread control was performed for them (3 cases with solely thenar presentation, 6 with solely FDI presentation, and one case with both thenar and FDI presentation). These 10 cases could be MGA with few fibers or a result of stimulus spread. The correlation coefficient between the potential differences SM1-SM2 and SU1-SU2 over APB in the MGA group (n ¼ 33, 23 certain and 10 possible) was negative but nonsignificant (rs ¼ 20:231, P ¼ 0:197) while in the nonMGA group, it was practically zero (rs ¼ 20:049, P ¼ 0:852). In contrast, the correlation coefficient between the SM1-SM2 and SU1-SU2 over FDI in the MGA group was negative and highly significant (rs ¼ 20:855, P , 0:001), while in the non-MGA group, it was positive and non-significant (rs ¼ 0:310, P ¼ 0:226). The 23 (definitely MGA) and the 17 (non-MGA) cases were used as training samples in DA for the estimation of a classification rule of cases to one or the other group. The variables used in DA were the 4 potential differences SM1SM2 and SU1-SU2 over APB and FDI. The two methods of DA gave identical results: 4 were classified as MGA, while 6 were classified as non-MGA cases. If the findings for these 10 cases were indeed compatible with the groups to which they were classified, the above correlation results for the new groups of MGA and nonMGA cases (n ¼ 27 and 23, respectively) would have to be either confirmed or strengthened. Indeed, the correlation coefficient between the differences in SM1-SM2 and SU1SU2 over APB in this new MGA group was negative and almost significant (rs ¼ 20:372, P ¼ 0:054) while in the non-MGA group (n ¼ 23), it was positive but non-significant (rs ¼ 0:109, P ¼ 0:619). Similarly, the correlation coefficient between SM1-SM2 and SU1-SU2 over FDI in the MGA group was negative and highly significant (rs ¼ 20:644, P , 0:001), while in the non-MGA group, it was practically zero (rs ¼ 0:099, P ¼ 0:654). The above results confirm the existence of negative correlation coefficients between SM1-SM2 and SU1-SU2 over APB and FDI in the group of so-defined MGA cases (n ¼ 27). The result of a marked, just outside the 5% level of significance correlation over APB (rs ¼ 20:372, P ¼ 0:054) is indeed significant in view of the small sample size (27 cases). The power of detecting this as a significant result (at the 5% level of significance) is only 48%. Hence, the estimated real incidence of MGA was 54% (27/50).
than on wrist stimulation of the ulnar nerve indicating an ulnar-to-median nerve anastomosis. Statistical analysis in these 12 cases could not demonstrate a negative correlation for the potential differences in SM1-SM2 and SU1-SU2 over APB (rs ¼ 0:378, P ¼ 0:226). In 3 of these 12 cases, the difference between wrist and elbow CMAP of thenar was more than 20% (21.6, 46.7 and 61.5%, respectively). Excluding these 3 cases, the correlation coefficients between SM1-SM2 and SU1-SU2 over APB (rs ¼ 0:227, P ¼ 0:557) remained non-significant. Moreover, the findings of these 3 cases could not be reproduced by supramaximal stimulation of the ulnar nerve at elbow by means of teflon-insulated unipolar needle electrodes. Therefore, the cause of proximal greater than distal CMAP of thenar upon ulnar nerve stimulation was a stimulus spread to the median nerve at the proximal site.
3.2. Frequency of ulnar-to-median nerve anastomosis comparing the CMAPs on wrist and elbow stimulation
The incidence of the MGA was 54% using the potential comparison method and 46% using the collision technique. The 4 cases (8%) of MGA not seen on the screen using the potential comparison method were also not recognized
In 12 subjects, the CMAP over APB was higher on elbow
3.3. Frequency of MGA using the collision technique A thenar presentation of MGA was found in 9 cases, a hypothenar presentation in 22 cases and a FDI presentation in 22 cases. In all 9 thenar presentations, criteria for hypothenar and FDI presentations were also fulfilled. In one case of hypothenar presentation, the FDI and the thenar presentation was negative, and in one case of FDI presentation, the hypothenar and the thenar presentations were negative. At least one of the criteria for MGA was fulfilled in 23 of the 50 subjects (46%), they being the same 23 MGA individuals of the potential comparison method recognized on the screen. In 6 of the 9 thenar presentations, the thenar presentation was not fulfilled using the comparison of the CMAP, but the FDI presentation was fulfilled using the comparison of the CMAP and also the collision technique. Thus it is likely that the potentials over APB in these 6 cases were volume conductions from FDI. All 23 cases were recognized on screen and control for stimulus spread was undertaken. 3.4. Frequency of ulnar-to-median nerve anastomosis using the collision technique In two subjects, a potential was recorded over APB on ulnar nerve collision block, indicating an ulnar-to-median nerve anastomosis. These were two of the 12 cases mentioned before. At reevaluation with teflon-insulated needle electrodes, it was observed that no ulnar-to-median nerve anastomosis but stimulus spread to the median nerve on elbow stimulation of the ulnar nerve was the cause of these findings. 4. Discussion
98
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
using the collision technique, because of the small number of the anastomotic fibers. The 23 cases (46%) recognized on the screen seem to be important for the correct interpretation of nerve conduction studies and EMG. The estimated incidence of MGA in this study exceeds the 44% of the anatomical study of Sokolow (cited by Hirasawa, 1931). Other anatomical studies found an even lower incidence (15.2% Gruber, 1870, 23% Nakashima, 1993) of MGA. As the anastomotic fibers rise from different branches of the anterior interosseous nerve (Nakashima, 1993), it is possible that small fiber bundles can escape the observation in anatomical studies. Contrasting this, as the activation of only one motor unit produces a potential of about 0.6– 1.5 mV in normal subjects (CNE recordings), it has to be expected that such small fiber bundles can easily be disclosed using electrophysiological recordings. Arbitrary limits for the amplitude differences between the potentials gained on wrist and elbow stimulation for the normal case can be the reason for the lower frequency of MGA in earlier electrophysiological studies (28%, Crutchfield and Gutmann, 1980 and 34%, Sun and Streib, 1983). Crutchfield and Gutmann required at least 1 mV for all recording sites (thenar, hypothenar, and FDI) and Sun and Streib required 25% for FDI. The lower rate (32%) of MGA in our previous report could be explained by the fact that patients with neuropathies were included in that study (Amoiridis, 1992). Stimulus spread to the ulnar nerve on median nerve stimulation at elbow could imitate the presence of MGA in any of the mentioned electrophysiological studies. Recordings in FCU by means of a CNE were used in one study to detect such stimulus spread (Amoiridis, 1992). As partial stimulation of a nerve stem is always possible, even the CNE in FCU cannot exclude a selective stimulus spread to the ulnar bundle for the intrinsic hand muscles. Apart from the CNE in FCU, moving the stimulating electrode towards the ulnar nerve in steps of 0.5 cm, and, if needed, near nerve needle electrode stimulation, excluded stimulus spread in all cases with at least one positive criterion for MGA seen on the screen. Slightly positive criteria for a MGA ascertained later (small differences between the potentials gained at wrist and elbow stimulation in 10 cases) were statistically investigated, 4 of them were classified to be a MGA and 6 a result of stimulus spread. The incidence of MGA using the collision block method was 46%, based on the study of the same individuals as the 23 MGA individuals of the potential comparison method recognized on the screen. The substantially lower incidence of MGA in the study of Kimura et al. (1976) has to be possibly attributed to the fact that they did not perform recordings over FDI. While in the present study, there was only one case with positive FDI and negative hypothenar
presentation of MGA, Sun and Streib (1983) found 18.5% of the FDI presentations associated with a negative hypothenar presentation. No case of motor ulnar-to-median nerve anastomosis could be found in this study. This type of anastomosis seems to be a rarity (Streib, 1979). Cases of false positive criteria for ulnar-to-median nerve anastomosis in this study have shown that stimulus spread to the median nerve on supramaximal stimulation of the ulnar nerve about 1 cm above the epicondylus medialis humeri is possible. Avoiding stimulus spread is more important than the method used for detection of communications between median and ulnar nerves. It can be ensured by precautions such as those used in this study. In a certain individual, comparison of CMAPs recorded over thenar, hypothenar, and FDI on median nerve stimulation at wrist and elbow is sufficient and simpler than collision blocks. In spite of the high frequency of MGA, median nerve conduction studies with recordings solely over thenar will detect a MGA in only a few cases, as thenar presentation is rare (in this study 14%).
References Amoiridis G. Median-ulnar nerve communications and anomalous innervation of the intrinsic hand muscles: an electrophysiological study. Muscle Nerve 1992;15:576–579. Armitage P, Berry G. Statistical methods in medical research, 3rd ed. Blackwell: Oxford, 1994. Crutchfield CA, Gutmann L. Hereditary aspects of median-ulnar nerve communications. J Neurol Neurosurg Psychiatry 1980;43:53–55. ¨ ber die Verbindung des Nervus medianus mit dem Nervus Gruber W. U ulnaris am Unterarme des Menschen und der Sa¨ ugethiere. Arch Anat Physiol Med Leipzig 1870;37:501–522. ¨ ber das periphere Nervensystem, Plexus Hirasawa K. Untersuchungen U brachialis und die Nerven der oberen Extremita¨ ten. Arch Anat Inst Kaiserlichen Univ Kyoto 1931;A2:135–137. Hopf HC, Hense W. Anomalien der motorischen Innervation an der Hand. Z EEG–EMG 1974;5:220–224. Kimura J, Murphy MJ, Varda DJ. Electrophysiological study of anomalous innervation of intrinsic hand muscles. Arch Neurol (Chic) 1976;33:842–844. Martin R. Tal om Nervers allmanna Egenskaper i Manniskans Kropp, Stockholm: L Salvius, 1763. Nakashima T. An anatomic study on the Martin–Gruber anastomosis. Surg Radiol Anat 1993;15:193–195. Panizza M, Lelli S, Hallet M. H-Reflex in non homonymous muscles in the human forearm. Neurology 1989;39:785–788. Streib EW. Ulnar-to-median nerve anastomosis in the forearm: electromyographic studies. Neurology 1979;29:1534–1537. Sun SF, Streib EW. Martin–Gruber anastomosis: electromyographic studies. Electromyogr Clin Neurophysiol 1983;23:271–285. Vlachonikolis IG, Marriott FHC. Discrimination with mixed binary and continuous data. Appl Stat 1982;31:23–31.
Verification of the median-to-ulnar and ulnar-to-median nerve motor fiber anastomosis in the forearm: an electrophysiological study Georgios Amoiridis a,*, Ioannis G. Vlachonikolis b a
Department of Neurology, School of Health Sciences, University of Crete, P.O. Box 1393, 71110 Heraklion, Greece Department of Biostatistics, School of Health Sciences, University of Crete, P.O. Box 1393, 71110 Heraklion, Greece
b
Accepted 9 October 2002
Abstract Objective: To estimate the real occurrence of the motor median-to-ulnar nerve anastomosis in the proximal forearm (Martin–Gruber anastomosis, MGA), as its frequency varies between 6 and 44% in the literature and to investigate the incidence of the ulnar-to-median nerve anastomosis in the distal forearm. Methods: Compound muscle action potentials (CMAP) recorded over thenar, hypothenar, and first dorsal interosseus muscle on median or ulnar nerve stimulation at wrist and elbow and collision blocks of the median and ulnar nerve were compared in a group of 50 healthy volunteers. Particular precautions were undertaken in order to avoid false positive results due to stimulus spread to the neighboring nerve. Cases of uncertain MGA were classified as either MGA or non-MGA on the basis of posterior probabilities estimated by discriminant analysis. Results: The estimated frequency of MGA was 54% using the potential comparison method and 46% using the collision technique. An ulnar-to-median nerve anastomosis was not found in any subject. Conclusions: While the MGA is very common, the ulnar-to-median nerve anastomosis is a rarity. Standard nerve conduction studies of the median nerve with CMAP recordings solely over thenar will detect less than 14% of MGA cases. q 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Median-to-ulnar nerve anastomosis; Martin–Gruber anastomosis; Ulnar-to-median nerve anastomosis; Innervation anomaly; Collision block; Compound muscle action potential analysis
1. Introduction The first description of an anastomosis between the median and ulnar nerve in the proximal forearm was in 1763 by the Swedish anatomist Martin (1763). Gruber (1870) estimated its frequency as 15.2% in 250 arms. This median-to-ulnar nerve anastomosis in the proximal forearm was later referred to as Martin–Gruber anastomosis (MGA). Further anatomical studies have shown a frequency of MGA between 44 (Sokolow, 1925, cited by Hirasawa, 1931) and 21.3% (Nakashima, 1993). There are no anatomical studies dealing with the ulnar-to-median nerve anastomosis in the distal forearm. Electrophysiological studies in combination with pharmacological studies (Hopf and Hense, 1974) or collision blocks (Kimura et al., 1976) have revealed a frequency of 6 and 17% for the MGA, respectively. Electrophysiological methods analyzing the compound muscle action potentials * Corresponding author. Tel.: 130-810-392457; fax: 130-810-392474. E-mail address: [email protected] (G. Amoiridis).
(CMAP) over thenar, hypothenar, and first dorsal interosseous muscle (FDI) on median and ulnar nerve stimulation at wrist and elbow have shown MGA in 28 (Crutchfield and Gutmann, 1980) and 34% (Sun and Streib, 1983) of cases. In summary, the frequency of MGA varies between 6 and 44% in the literature. The motor ulnar-to-median nerve anastomosis in the distal forearm is extremely rare (Streib, 1979, Kimura et al., 1976, Amoiridis, 1992). The communications between the median and ulnar nerves in the forearm provide for variations in the innervation of the intrinsic hand muscles, as proved by nerve conduction studies (Kimura et al., 1976, Amoiridis, 1992). The knowledge of these variations is of major importance for a correct interpretation of needle electromyographic (EMG) and nerve conduction studies in cases of suspected median and ulnar nerve lesions. In this study, an attempt was made to estimate the occurrence of the MGA and the ulnar-to-median nerve anastomosis using two methods: CMAP comparison and collision blocks.
1388-2457/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88- 2457(02)0032 8-0
CLINPH 2002646
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
95
2. Materials and methods The left upper extremity of 50 healthy volunteers was examined (30 men, 20 women). Relative to the question of the frequency of the MGA, the age of the subjects was disregarded. Recordings were performed using a 4-channel device (Counterpoint Electromyograph of Dantec, Denmark), which enables a 4-channel recording using two stimulators. The band pass setting was 10–20 kHz. Stimulation was performed using surface electrodes (SE, Dantec, type 13L36). The stimulus intensity was gradually increased in steps of 5 mA, until the CMAP reached its maximal amplitude. After this, the stimulus was increased to another 5–10 mA (stimulus duration 0.2 ms, usually needed stimulus intensity 25 mA at wrist and 40 mA at elbow). Initially, the median and ulnar nerve, were successively stimulated at wrist and elbow (ulnar nerve 1 cm proximal to epicondylus medialis humeri). Subsequently, each of the nerves was simultaneously stimulated at wrist and elbow (collision block). In a second attempt, the stimulus at the elbow was delivered some milliseconds later than the one at the wrist. When criteria for MGA were fulfilled, the stimulating electrode on proximal median nerve stimulation was moved medially towards the medial epicondyle in steps of about 0.5 cm. Disappearance of the MGA-criteria by this maneuver excluded a stimulus spread to the ulnar nerve. When MGA-criteria persisted, teflon-insulated unipolar needle electrodes were used to stimulate the median nerve at elbow. In cases of MGA recognized on screen, simultaneous stimulation of the median nerve at the elbow and the ulnar nerve at wrist was obtained, in order to block the anastomotic fibers. Recordings of the CMAP were obtained by means of SE over thenar (sensitivity 2–10 mV/division), hypothenar (sensitivity 0.5–2 mV/division), and FDI (sensitivity 0.5– 2 mV). In the following discussion, thenar and abductor pollicis brevis (APB) will be used as identical recording sites, and the same will be the case for hypothenar and abductor digiti minimi (ADM). A concentric needle electrode (CNE) was inserted in the flexor carpi ulnaris muscle (FCU, sensitivity 0.5–10 mV), detecting a possible stimulus spread to the ulnar nerve on median nerve stimulation at elbow (Fig. 1). The correct position of the CNE in the FCU was proven by stimulating the median and ulnar nerves at elbow. Only ulnar nerve stimulation at elbow might evoke a potential with a sharp deflection from the base line. If criteria for the existence of a MGA were fulfilled when the CNE already recorded a potential in the FCU, selective-near nerve stimulation of the median nerve at elbow with teflon-insulated unipolar needle electrodes was performed. The amplitudes were measured from the base line to the negative peak. For comparisons of the CMAP, criteria for a MGA were the same as in a previous study (Amoiridis, 1992). In brief, the presence of a MGA was assumed when the CMAP over
Fig. 1. Electrode positioning and stimulating sites. FDI, first dorsal interosseous; CNE, concentric needle electrode; FCU, flexor carpi ulnaris muscle; S, stimulus; M, median; U, ulnar; 1, wrist; 2, elbow.
thenar or FDI was higher on elbow than on wrist stimulation of the median nerve (thenar or FDI presentation) or a potential with a different shape, on elbow stimulation compared with wrist stimulation of the median nerve, was evoked over ADM (hypothenar presentation). For comparisons of the CMAP, the presence of an ulnarto-median anastomosis was assumed when the CMAP over one of the recording sites (thenar, FDI or hypothenar) was higher on elbow than on wrist stimulation of the ulnar nerve. For the collision block technique, a MGA was assumed when on median nerve collision block a CMAP was recorded at least over one of the recording sites (thenar, hypothenar, FDI) with latency similar to the latency of the CMAP evoked by the elbow stimulation of the median nerve. Furthermore, elimination of this CMAP on simultaneous median elbow and ulnar wrist stimulation was required. For the collision block technique, an ulnar-to-median nerve anastomosis was assumed when on ulnar nerve collision block a CMAP was recorded at least over one of the recording sites (thenar, hypothenar, FDI), with latency similar to the latency of the CMAP evoked by the elbow stimulation of the ulnar nerve.
2.1. Statistics The differences of the CMAPs evoked over APB and FDI muscles on wrist (1) and elbow (2) stimulation (S) of the median (M) and ulnar (U) nerve were estimated for each volunteer (SM1-SM2 and SU1-SU2 over APB and FDI, see also Fig. 1). Their distribution was tested for normality by the Kolmogorov–Smirnov and Saphiro–Wilk tests (Armitage and Berry, 1994). Only SM1-SM2 over APB had a normal distribution; the other 3 were not normal (P , 0:05). Their correlation was therefore assessed by means of the non-parametric Spearman’s correlation coefficient rs (Armitage and Berry, 1994). Cases of uncertain MGA were classified as either MGA or non-MGA on the basis of posterior probabilities estimated by discriminant analysis (DA); the latter was carried out by two methods,
96
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
Fig. 2. Potentials recorded on wrist (A) and elbow (B) stimulation of the median nerve (M) in a volunteer without MGA. APB, abductor pollicis brevis (thenar); ADM, abductor digiti minimi (hypothenar); FDI, first dorsal interosseous; FCU, flexor carpi ulnaris. Same potential shape on wrist (A) and elbow stimulation (B). H-reflex in FCU on median nerve stimulation at elbow.
Fisher’s linear DA and logistic discrimination (Vlachonikolis and Marriott, 1982).
3. Results In 25 subjects, the CNE in FCU recorded a potential with a latency of 17–20 ms on median nerve elbow stimulation (Figs. 2 and 3). In 5 subjects, it was recorded also on median wrist stimulation. This potential represents a heteronymous H-reflex, with its afferent fibers in median and the efferent in ulnar nerve (Panizza et al., 1989). The potentials over APB, ADM, FDI, and FCU in one subject without MGA are shown in Fig. 2. The SE over ADM records a positive potential on wrist and elbow median nerve stimulation, which represents the volume conduction of the thenar CMAP (Kimura et al., 1976). Furthermore, the same potential is recorded over FDI as volume conduction potential with a negative deflection, as
Fig. 3. Potentials recorded on wrist (A) and elbow (B) stimulation of the median nerve (M) in a volunteer with MGA. Larger area over APB, negative potential over ADM and higher potential over FDI on elbow stimulation (B) of the median nerve. H-reflex in FCU on median nerve stimulation at elbow.
its origin, the thenar muscles, are located vertically under the SE over FDI. The potentials over APB, ADM, FDI, and FCU in one subject with MGA are shown in Fig. 3. The area of the CMAP over APB after elbow median nerve stimulation is slightly larger than on wrist stimulation of the same nerve. Furthermore, on median nerve stimulation at elbow, a negative CMAP is recorded over ADM and the CMAP over FDI is higher than the CMAP over the same muscle on median wrist stimulation. The collision block of the median nerve and of the anastomotic fibers in a subject with MGA are shown in Fig. 4. The stimulus at the elbow was given 3 ms (Fig. 4A) and 4 ms (Fig. 4B) later than at wrist. In the first 3 traces, a potential with a short latency (about 3.5 ms) is recorded, representing the result of the wrist stimulation of the median nerve (SM1, Fig. 4A). Another potential with a latency of about 9 ms is recorded over ADM and FDI (SM2 in Fig. 4A), representing the potential of the anastomotic fibers, which escape the collision block of the median nerve. These fibers are blocked when stimulating the median nerve at elbow and the ulnar nerve at wrist (Fig. 4B). The positive sharp deflection in the 4th trace (FCU) with latency of 3 ms (Fig. 4A) and 4 ms (Fig. 4B) is the stimulus artifact of the elbow stimulation of the median nerve. 3.1. Frequency of MGA comparing the CMAPs on wrist and elbow stimulation A thenar presentation was found in 15 (30%), a hypothenar presentation in 22 (44%) and a FDI presentation in 29 (58%) cases. In 15 of the 22 hypothenar presentations, a negative CMAP was recorded over ADM on elbow median nerve stimulation and in the remaining 7 cases, a different potential shape, between wrist and elbow median nerve
Fig. 4. Collision block of median nerve (A) in a case of MGA with hypothenar (ADM) and FDI presentation. SM1, potentials due to the wrist stimulation of the median nerve. Stimulation of the median nerve at elbow (SM2) elicited no potential over APB, as the MGA had no thenar presentation. SM2 over ADM and FDI (A) represents the potential of the anastomotic fibers. In (B), stimulating the median nerve at elbow (SM2) and the ulnar nerve at wrist (SU1) blocks the anastomotic fibers (disappearance of the SM2 potentials over ADM and FDI). The stimulus at the elbow was delivered 3 ms later than at wrist in A and 4 ms in B.
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
97
stimulation, was observed. At least one of the criteria for a MGA was fulfilled in 33 of the 50 cases (66%). The 22 hypothenar presentations and one case with FDI presentation (23 cases) were recognized on screen and the stimulus spread to the ulnar nerve on median nerve stimulation at elbow was excluded. The remaining 10 cases were seen later and no stimulus spread control was performed for them (3 cases with solely thenar presentation, 6 with solely FDI presentation, and one case with both thenar and FDI presentation). These 10 cases could be MGA with few fibers or a result of stimulus spread. The correlation coefficient between the potential differences SM1-SM2 and SU1-SU2 over APB in the MGA group (n ¼ 33, 23 certain and 10 possible) was negative but nonsignificant (rs ¼ 20:231, P ¼ 0:197) while in the nonMGA group, it was practically zero (rs ¼ 20:049, P ¼ 0:852). In contrast, the correlation coefficient between the SM1-SM2 and SU1-SU2 over FDI in the MGA group was negative and highly significant (rs ¼ 20:855, P , 0:001), while in the non-MGA group, it was positive and non-significant (rs ¼ 0:310, P ¼ 0:226). The 23 (definitely MGA) and the 17 (non-MGA) cases were used as training samples in DA for the estimation of a classification rule of cases to one or the other group. The variables used in DA were the 4 potential differences SM1SM2 and SU1-SU2 over APB and FDI. The two methods of DA gave identical results: 4 were classified as MGA, while 6 were classified as non-MGA cases. If the findings for these 10 cases were indeed compatible with the groups to which they were classified, the above correlation results for the new groups of MGA and nonMGA cases (n ¼ 27 and 23, respectively) would have to be either confirmed or strengthened. Indeed, the correlation coefficient between the differences in SM1-SM2 and SU1SU2 over APB in this new MGA group was negative and almost significant (rs ¼ 20:372, P ¼ 0:054) while in the non-MGA group (n ¼ 23), it was positive but non-significant (rs ¼ 0:109, P ¼ 0:619). Similarly, the correlation coefficient between SM1-SM2 and SU1-SU2 over FDI in the MGA group was negative and highly significant (rs ¼ 20:644, P , 0:001), while in the non-MGA group, it was practically zero (rs ¼ 0:099, P ¼ 0:654). The above results confirm the existence of negative correlation coefficients between SM1-SM2 and SU1-SU2 over APB and FDI in the group of so-defined MGA cases (n ¼ 27). The result of a marked, just outside the 5% level of significance correlation over APB (rs ¼ 20:372, P ¼ 0:054) is indeed significant in view of the small sample size (27 cases). The power of detecting this as a significant result (at the 5% level of significance) is only 48%. Hence, the estimated real incidence of MGA was 54% (27/50).
than on wrist stimulation of the ulnar nerve indicating an ulnar-to-median nerve anastomosis. Statistical analysis in these 12 cases could not demonstrate a negative correlation for the potential differences in SM1-SM2 and SU1-SU2 over APB (rs ¼ 0:378, P ¼ 0:226). In 3 of these 12 cases, the difference between wrist and elbow CMAP of thenar was more than 20% (21.6, 46.7 and 61.5%, respectively). Excluding these 3 cases, the correlation coefficients between SM1-SM2 and SU1-SU2 over APB (rs ¼ 0:227, P ¼ 0:557) remained non-significant. Moreover, the findings of these 3 cases could not be reproduced by supramaximal stimulation of the ulnar nerve at elbow by means of teflon-insulated unipolar needle electrodes. Therefore, the cause of proximal greater than distal CMAP of thenar upon ulnar nerve stimulation was a stimulus spread to the median nerve at the proximal site.
3.2. Frequency of ulnar-to-median nerve anastomosis comparing the CMAPs on wrist and elbow stimulation
The incidence of the MGA was 54% using the potential comparison method and 46% using the collision technique. The 4 cases (8%) of MGA not seen on the screen using the potential comparison method were also not recognized
In 12 subjects, the CMAP over APB was higher on elbow
3.3. Frequency of MGA using the collision technique A thenar presentation of MGA was found in 9 cases, a hypothenar presentation in 22 cases and a FDI presentation in 22 cases. In all 9 thenar presentations, criteria for hypothenar and FDI presentations were also fulfilled. In one case of hypothenar presentation, the FDI and the thenar presentation was negative, and in one case of FDI presentation, the hypothenar and the thenar presentations were negative. At least one of the criteria for MGA was fulfilled in 23 of the 50 subjects (46%), they being the same 23 MGA individuals of the potential comparison method recognized on the screen. In 6 of the 9 thenar presentations, the thenar presentation was not fulfilled using the comparison of the CMAP, but the FDI presentation was fulfilled using the comparison of the CMAP and also the collision technique. Thus it is likely that the potentials over APB in these 6 cases were volume conductions from FDI. All 23 cases were recognized on screen and control for stimulus spread was undertaken. 3.4. Frequency of ulnar-to-median nerve anastomosis using the collision technique In two subjects, a potential was recorded over APB on ulnar nerve collision block, indicating an ulnar-to-median nerve anastomosis. These were two of the 12 cases mentioned before. At reevaluation with teflon-insulated needle electrodes, it was observed that no ulnar-to-median nerve anastomosis but stimulus spread to the median nerve on elbow stimulation of the ulnar nerve was the cause of these findings. 4. Discussion
98
G. Amoiridis, I.G. Vlachonikolis / Clinical Neurophysiology 114 (2003) 94–98
using the collision technique, because of the small number of the anastomotic fibers. The 23 cases (46%) recognized on the screen seem to be important for the correct interpretation of nerve conduction studies and EMG. The estimated incidence of MGA in this study exceeds the 44% of the anatomical study of Sokolow (cited by Hirasawa, 1931). Other anatomical studies found an even lower incidence (15.2% Gruber, 1870, 23% Nakashima, 1993) of MGA. As the anastomotic fibers rise from different branches of the anterior interosseous nerve (Nakashima, 1993), it is possible that small fiber bundles can escape the observation in anatomical studies. Contrasting this, as the activation of only one motor unit produces a potential of about 0.6– 1.5 mV in normal subjects (CNE recordings), it has to be expected that such small fiber bundles can easily be disclosed using electrophysiological recordings. Arbitrary limits for the amplitude differences between the potentials gained on wrist and elbow stimulation for the normal case can be the reason for the lower frequency of MGA in earlier electrophysiological studies (28%, Crutchfield and Gutmann, 1980 and 34%, Sun and Streib, 1983). Crutchfield and Gutmann required at least 1 mV for all recording sites (thenar, hypothenar, and FDI) and Sun and Streib required 25% for FDI. The lower rate (32%) of MGA in our previous report could be explained by the fact that patients with neuropathies were included in that study (Amoiridis, 1992). Stimulus spread to the ulnar nerve on median nerve stimulation at elbow could imitate the presence of MGA in any of the mentioned electrophysiological studies. Recordings in FCU by means of a CNE were used in one study to detect such stimulus spread (Amoiridis, 1992). As partial stimulation of a nerve stem is always possible, even the CNE in FCU cannot exclude a selective stimulus spread to the ulnar bundle for the intrinsic hand muscles. Apart from the CNE in FCU, moving the stimulating electrode towards the ulnar nerve in steps of 0.5 cm, and, if needed, near nerve needle electrode stimulation, excluded stimulus spread in all cases with at least one positive criterion for MGA seen on the screen. Slightly positive criteria for a MGA ascertained later (small differences between the potentials gained at wrist and elbow stimulation in 10 cases) were statistically investigated, 4 of them were classified to be a MGA and 6 a result of stimulus spread. The incidence of MGA using the collision block method was 46%, based on the study of the same individuals as the 23 MGA individuals of the potential comparison method recognized on the screen. The substantially lower incidence of MGA in the study of Kimura et al. (1976) has to be possibly attributed to the fact that they did not perform recordings over FDI. While in the present study, there was only one case with positive FDI and negative hypothenar
presentation of MGA, Sun and Streib (1983) found 18.5% of the FDI presentations associated with a negative hypothenar presentation. No case of motor ulnar-to-median nerve anastomosis could be found in this study. This type of anastomosis seems to be a rarity (Streib, 1979). Cases of false positive criteria for ulnar-to-median nerve anastomosis in this study have shown that stimulus spread to the median nerve on supramaximal stimulation of the ulnar nerve about 1 cm above the epicondylus medialis humeri is possible. Avoiding stimulus spread is more important than the method used for detection of communications between median and ulnar nerves. It can be ensured by precautions such as those used in this study. In a certain individual, comparison of CMAPs recorded over thenar, hypothenar, and FDI on median nerve stimulation at wrist and elbow is sufficient and simpler than collision blocks. In spite of the high frequency of MGA, median nerve conduction studies with recordings solely over thenar will detect a MGA in only a few cases, as thenar presentation is rare (in this study 14%).
References Amoiridis G. Median-ulnar nerve communications and anomalous innervation of the intrinsic hand muscles: an electrophysiological study. Muscle Nerve 1992;15:576–579. Armitage P, Berry G. Statistical methods in medical research, 3rd ed. Blackwell: Oxford, 1994. Crutchfield CA, Gutmann L. Hereditary aspects of median-ulnar nerve communications. J Neurol Neurosurg Psychiatry 1980;43:53–55. ¨ ber die Verbindung des Nervus medianus mit dem Nervus Gruber W. U ulnaris am Unterarme des Menschen und der Sa¨ ugethiere. Arch Anat Physiol Med Leipzig 1870;37:501–522. ¨ ber das periphere Nervensystem, Plexus Hirasawa K. Untersuchungen U brachialis und die Nerven der oberen Extremita¨ ten. Arch Anat Inst Kaiserlichen Univ Kyoto 1931;A2:135–137. Hopf HC, Hense W. Anomalien der motorischen Innervation an der Hand. Z EEG–EMG 1974;5:220–224. Kimura J, Murphy MJ, Varda DJ. Electrophysiological study of anomalous innervation of intrinsic hand muscles. Arch Neurol (Chic) 1976;33:842–844. Martin R. Tal om Nervers allmanna Egenskaper i Manniskans Kropp, Stockholm: L Salvius, 1763. Nakashima T. An anatomic study on the Martin–Gruber anastomosis. Surg Radiol Anat 1993;15:193–195. Panizza M, Lelli S, Hallet M. H-Reflex in non homonymous muscles in the human forearm. Neurology 1989;39:785–788. Streib EW. Ulnar-to-median nerve anastomosis in the forearm: electromyographic studies. Neurology 1979;29:1534–1537. Sun SF, Streib EW. Martin–Gruber anastomosis: electromyographic studies. Electromyogr Clin Neurophysiol 1983;23:271–285. Vlachonikolis IG, Marriott FHC. Discrimination with mixed binary and continuous data. Appl Stat 1982;31:23–31.