The role of microembolisation in cerebral injury as defined by functional magnetic resonance imaging.

OBJECTIVE: Cerebral injury, in both overt and subtle forms, is common following cardiac surgery. Current methods of assessment, most commonly neuropsychological testing, have several limitations and do not accurately define the anatomical and functional injury that occurs. We have assessed the degree of cerebral injury following on-pump and off-pump cardiac surgery using functional magnetic resonance imaging and correlated this with the severity of microembolism as measured by transcranial Doppler ultrasound. METHODS: Sixteen patients undergoing cardiac surgery (8 off-pump coronary artery bypass grafting (CABG), 4 on-pump CABG and 4 open-heart surgery) underwent functional magnetic resonance imaging of the brain pre-operatively and 4 weeks post-operatively. The functional magnetic resonance images demonstrated brain activation during performance of a verbal working memory paradigm. Each patient had continuous transcranial Doppler monitoring intraoperatively using a recently validated technique (multirange, multifrequency Doppler) that allows rejection of artefacts and separation of gas and solid microemboli. Covariate analysis of pre- and post-operative functional magnetic resonance images was performed to correlate local mean signal intensity change with the extent of gas and solid microembolism. RESULTS: The median number of microemboli was 34 (range 10-176) in the off-pump group, 229 (range 127-314) in the on-pump CABG group, and 1220 (range 874-1261) in the open-heart group (P<0.05). The proportion of solid microemboli was significantly lower in the off-pump group in comparison to the on-pump CABG and open-heart groups (9 vs. 25 vs. 20%, respectively, P<0.01). Comparison of pre- and post-operative functional magnetic resonance images demonstrated an overall reduction in task-associated activation in the post-operative period. However, and paradoxically, in certain specific regions of interest there was an increase in the signal intensity which correlated with the total number of microemboli (r=0.9, P