A patient-specific computer model to predict outcomes of the balloon occlusion test

Abstract

Balloon occlusion tests (BOTs) are performed to identify patients who are at risk for ischemia and stroke following permanent internal carotid artery (ICA) occlusion. The object of this work was to determine whether patient-specific blood flow modeling can be used to identify patients in whom the BOT would not be tolerated. The test was performed in 16 patients who underwent BOT with continuous neurological and electroencephalographic monitoring, followed by a hypotensive challenge. During hypotension a tracer was injected so that single-photon emission tomography (SPECT) scans could be obtained. Each individual brain circulation was modeled using information gained from phase-contrast magnetic resonance (MR) angiography and digital subtraction (DS) angiography, and the predicted effect of the BOT was evaluated. Six patients did not tolerate the BOT; in these patients, decreases in middle cerebral artery (M1 segment) blood flow of 41 +/- 27% (mean +/- standard deviation), anterior cerebral artery (A3 segment) flow of 56 +/- 33%, and posterior cerebral artery (P2 segment) flow of 4 +/- 13% ipsilateral to the site of occlusion were found with modeling; these changes were significantly greater than the percentage of changes measured in the contralateral hemisphere (p < 0.05). Ten patients who tolerated the BOT well had calculated decreases in ipsilateral flows of only 9 +/- 6% for the M1 segment, 12 +/- 40% for the A3 segment, and 17 +/- 21% for the P2 segment during BOT modeling. A decrease in blood flow in both the ipsilateral M1 and A3 segments that was greater than 20%, calculated by flow modeling of the BOT, was 100% sensitive and 100% specific in identifying patients who could not tolerate the BOT. Blood flow modeling, coupled with DS angiography and noninvasive phase-contrast MR angiography measurements to make calculations patient specific, can be used to identify patients who have an elevated risk of ischemia during the BOT.