Voxel-based mapping of irreversible ischaemic damage with PET in acute stroke

Abstract

Objective mapping of irreversible tissue damage in the acute stage of ischaemic stroke would be useful for prognosis and in assessing the efficacy of therapeutic manoeuvres in impeding extension of infarction. From our database of 30 patients studied with 15O-PET within 5-18 h after onset of first-ever middle cerebral artery territory stroke, we extracted a subgroup of 19 survivors (age 74.6 +/- 8.5 years) in whom late CT coregistered with PET was available to determine final infarct topography. By means of a voxel-based analysis of the PET data, we determined putative thresholds for irreversible tissue damage as the lower limit of the 95% confidence interval calculated from all voxels within the ultimately non-infarcted brain parenchyma ipsilateral to the insult. The following values were found: 8.43 ml/100 ml/min, 0.87 ml/100 ml/min, 1.64 ml/100 ml, 0.27 and 2.21/min, for cerebral blood flow (CBF), oxygen consumption (CMRO2), blood volume (CBV), oxygen extraction fraction and the ratio CBF : CBV, respectively. Voxels below these thresholds occurred significantly more frequently in the final infarct region than in the non-infarcted parenchyma for CBF and CMRO2 (P = 0.016 and P = 0.0045, respectively, Wilcoxon test), but not for the other PET variables. Furthermore, with both CBF and CMRO2, the percentage of irreversible tissue damage voxels in the affected hemisphere relative to the opposite hemisphere was significantly positively correlated to both the volume of final infarct and the neurological outcome at 2 months (all P < 0.005, Spearman ranked test). These findings validate our voxel-based CBF and CMRO2 thresholds for probabilistic mapping of irreversible tissue damage within the 5-18 h interval after stroke onset; however, whether they would be applicable to earlier intervals remains to be determined. Transfer of our procedure for determination of irreversible tissue damage thresholds to other imaging modalities such as single proton emission computed tomography and diffusion-weighted MRI should be straightforward.