ADC-values for prediction of postischaemic secondary cerebral haemorrhage

Introduction: Secondary haemorrhagic transformation (HT) and especially symptomatic haemorrhage is a major concern in stroke management. Haemorrhage within the infarcted area occurs both as a spontaneous event, with a peak occurrence during the first week after stroke, and, with an apparent higher frequency, in patients receiving thrombolytic therapy. Previous studies suggest that early MR-imaging within the first hours of stroke onset can identify patients who are at a greater risk to develop a secondary haemorrhagic transformation by analyzing the apparent diffusion coefficient (ADC). It has been suggested that low ADC-values within an ischemic area can predict an increased risk of HT in patients with acute ischaemic stroke.

Methods: We retrospectively screened patients suffering from acute infarction in the middle cerebral artery (MCA) territory initially diagnosed with a MRI including ADC-mapping. Of these we analysed all patients having developed a secondary haemorrhagic transformation or secondary intracranial haemorrhage (ICH) of the infarcted tissue. 7 Patients met our inclusion criteria, for comparison 9 patients without secondary ICH but similar infarction size were analysed. Analysis was performed by a specially developed software. For the classification of the voxel intensities in the ischaemic lesion, a histogram has been set up arbitrarily consisting of 10 equal intensity steps ranging from the minimum voxel intensity in the lesion to the maximum voxel intensity in the lesion. This allows to compare different measurements between patients since these histograms rule out the influence of different basic signal intensities as well as different ranges of signal intensities between patients.

Results: In the voxel-based analysis of the ischaemic lesions of our patients with secondary haemorrhagic transformation (sHT) a higher proportion of low ADC-values are found, than in our patients without sHT. The distribution curve of ADC-values is shifted towards a greater number of voxels with low ADC-values in HT-patients

Discussion: Our findings support the hypothesis, that indeed the size of an ischaemic lesion with low ADC-values correlates with a higher risk of haemorrhagic transformation following stroke.