To understand the relationship between brainstem lesions and auditory
neurology in patients with multiple sclerosis, we compared behavioural
, electrophysiological and imaging data in 38 patients with probable o
r definite multiple sclerosis and normal or near normal hearing. Behav
ioural measures included (i) general hearing tests (audiogram, speech
discrimination) and (ii) hearing tests likely to be critically depende
nt upon brainstem processing (masking level difference, and level disc
rimination). Brainstem potentials provided the electrophysiological da
ta. Multiplanar high-resolution MRI of the brainstem provided the anat
omical data. Interaural time discrimination for high-frequency sounds
was by far the most sensitive of all tests with abnormalities in 71% o
f all subjects. Whenever any other test was abnormal this test was alw
ays abnormal. Interaural time discrimination for low-frequency sounds
and evoked potentials were closely related and next most sensitive wit
h abnormalities in similar to 40% of all subjects. Interaural level di
scrimination and masking level difference were least sensitive with ab
normalities in <10% of subjects. Speech discrimination scores correlat
ed significantly with the masking level differences, as well as with i
nteraural time discrimination for high-frequency sounds. Pontine lesio
ns were found in five of the 16 patients, in whom an objective method
for detecting magnetic resonance lesions could be applied. All four wi
th lesions involving the pontine auditory pathway had marked abnormali
ties in interaural time discrimination and evoked potentials. None of
the other 12 had evoked potentials abnormalities. We conclude that neu
rological tests requiring precise neural timing can reveal behavioural
deficits for multiple sclerosis lesions of the auditory pens that are
otherwise 'silent'. Of all neurological systems the auditory system a
t the level of the pens is probably the most sensitive to multiple scl
erosis lesions, because of its exceptional dependence upon neural timi
ng in the microsecond range and the lack of redundancy in the encoding
of high-frequency sounds. Precise neural timing may be critical for s
ome aspects of speech processing.