Diagnostic accuracy of computed tomography perfusion in the prediction of haemorrhagic transformation and patient outcome in acute ischaemic stroke: A systematic review and meta-analysis.

PURPOSE: The aim of this systematic review and meta-analysis is to determine the diagnostic accuracy of computed tomography brain perfusion in the prediction of haemorrhagic transformation and patient outcome in acute ischaemic stroke.
METHOD: Electronic databases and grey literature published over the last 10 years related to healthcare and radiology were searched using the key terms: 'computed tomography perfusion', 'haemorrhagic transformation', 'acute ischaemic stroke', 'functional outcome' and their synonyms using both UK and American spellings. Inclusion criteria were: sample size at least 30 patients, original research, evaluate ability of computed tomography perfusion to predict haemorrhagic transformation, reports diagnostic accuracy or provide relevant data for a 2 × 2 contingency table, use follow-up non-contrast computed tomography (NCCT) or magnetic resonance imaging as reference standard.
FINDINGS: Twelve studies were included in the review; studies cover a total of 808 patients. Haemorrhagic transformation occurred in 30.2% of patients. Pooled sensitivity and specificity were 85.9% (95% CI; 65-97%), 73.9% (95% CI; 45-92%) and accuracy of 79.1% (95% CI; 57-98%). Pooled NPV was 92.9% with a high false positive rate (19.8%), which could be explained in terms of outcome classification, acquisition artefact and computed tomography perfusion processing algorithms. DISCUSSION: This review evaluated the importance of using pre-defined threshold measurement for optimal prediction of HT, the relevance of patient pre-treatment clinical parameters to HT occurrence, the CTP parameters and the measurements that are independent predictors of HT, the significance of rtPA rather as an exacerbator of HT and the impact of both minor and major HT/PH on patient 20 functional outcome.
CONCLUSION: Computed tomography perfusion has a high sensitivity and moderately high specificity for prediction of haemorrhagic transformation in acute ischaemic stroke. Pre-treatment clinical decision making requires consideration of clinical factors in addition to imaging findings. This systematic review and meta-analysis highlights that pre-treatment computed tomography perfusion adds to clinical confidence by predicting potential for haemorrhage, both in thrombolysed and un-thrombolysed patients, and also influences decisions about alternative treatments for acute ischaemic stroke patients. © European Stroke Organisation 2019.