A number of nerve fibers supplying ulnar-innervated neuropathy, or to exclude abnormalities. Supramaximuscles cross over the median to the ulnar nerve in mal stimuli were given at distal and proximal sites of Martin-Gruber anastomosis (MGA). Anatomical and the median and ulnar nerves. Large electrodes 11,12 on surgical data indicate that it occurs in 5-40% of defined sites 10 were used to record CMAPs simultanearms. 3,9 Various neurographic diagnostic criteria for ously from thenar, first dorsal interosseous, and hy-MGA have been proposed, 1,2,[4][5][6]8 but, except for one, 5 pothenar muscles, allowing costimulation of nerves these were not based on empirical data and did not or inframaximal stimulation to be recognized easily. quantify its essential feature: in MGA, stimulation Reference electrodes were placed at the bases of the proximal or distal of the anastomosis does not excite proximal phalanges of the thumb, index, and fifth the same number of nerve fibers. The ulnar nerve finger. CMAP amplitudes were measured from basegains fibers, so the compound muscle action potential line to negative phase peak. (CMAP) following distal stimulation will have a
The loss of the median nerve was calculated by higher amplitude than following proximal stimulasubtracting the distally evoked CMAP amplitude tion. The number of fibers crossing over will affect from the proximally evoked one (AMPmed P minus the magnitude of the difference. Normal 10 or abnor-AMPmed D ). MGA results in a positive value, and normal temporal dispersion and conduction blocks also mal conduction in a small negative one. The gain of cause distal CMAP amplitudes to be higher than proxthe ulnar nerve was quantified by subtracting the imal ones, but it is not known which value of ampliproximally evoked amplitude from the distally tude decay distinguishes between (ab)normal conevoked one (AMPuln D minus AMPuln P ). MGA results duction and MGA. The median nerve loses fibers in in a positive value and normal conduction in a small the forearm, so the distally evoked CMAP amplitude positive one. The gain/loss voltage was calculated has a lower amplitude than the proximally evoked for each muscle group: (AMPmed P Ϫ AMPmed D ) one. 3 This pattern can be normal, depending on reϩ (AMPuln D Ϫ AMPuln P ). Normal conduction will cording site, 12 but it is again not known which value result in values close to zero, while MGA results in a distinguishes between normal findings and MGA. In large positive value. Ratios showing median nerve MGA, one nerve's loss must be the other nerve's gain (AMPmed P /AMPmed D ) and ulnar nerve loss gain, suggesting that a quantification of this gain/loss (AMPuln D /AMPuln P ) were also calculated. A value phenomenon might result in separate distribution of 1 was subtracted from each ratio, to ensure that peaks for MGA and normal conduction. If so, this it would normally lie close to zero. These were added could provide an empirical diagnostic criterion.
to form the gain/loss ratio [(AMPmed P /AMPmed D )
Ϫ 1] ϩ [(AMPuln D /AMPuln P ) Ϫ 1]. Distributions
Methods
were analyzed with standard descriptive statistics us-Data were collected from 150 arms, tested for carpal ing the SPSS package. 7 tunnel syndrome, compression ulnaropathy, poly-
Results
The study comprised 150 arms (83 right arms). The