Prediction of midline shift after media ischemia using computed tomography perfusion

Decision-making about the indication for decompressive hemicraniectomy in cases with malignant infarction in the territory of the middle cerebral artery (MCA) is still a matter of debate. Some scores have been introduced and tested, most of them are midline-shift dependent. We introduce the Kinematics of malignant MCA infarction (KM) index, which can be calculated based on an initial computed tomography perfusion scan and the chosen therapy (lysis/thrombectomy/conservative) in order to estimate the maximum midline-shift in the subsequent 6 days.

We retrospectively analyzed patients with middle cerebral artery infarction who had a non-enhanced computed tomography (CT) scan, CT angiography and a CT perfusion scan in the acute setting and who presented in our emergency room between 2015 and 2019. 186 patients were included. Midline shift was measured on follow-up imaging between days 0 and 6 after stroke. We evaluated Pearson's correlation between the KM index and the amount of midline shift.

The mean KM index of all patients was 1.01 ± 0.09 (decompressive hemicraniectomy subgroup 1.13 ± 0.13; midline shift subgroup 1.18 ± 0.13). The correlation coefficient between the KM index and substantial midline-shift was 0.61, p < 0.01 and between KM index and decompressive hemicraniectomy or death 0.47; p < 0.05. KM index > 1.02 shows a sensitivity of 92% (22/24) and a specificity of 78% (126/162) for detecting midline shifts. The area under curve of the receiver operator characteristics was 91% for midline shifts and 86% for the occurrence of decompressive hemicraniectomy or death.

In this retrospective study, KM index shows a strong correlation with significant midline-shift. The KM index can be used for risk classification regarding herniation and the need of decompressive hemicraniectomy.