Mismatch between baseline diffusion-weighted images and apparent diffusion coefficient maps predicts favourable penumbra: a retrospective study

Abstract

BackgroundHyperintense diffusion-weighted images (DWIs) were thought to represent infarct, whereas hypointense apparent diffusion coefficient (ADC) maps could represent both infarct and penumbra. The mechanism and implication of the inhomogeneous intensity distribution of DWI infarct are unknown. This study aimed to investigate whether imaging characteristics of baseline DWIs and ADC maps could guide decision making of thrombolysis. Methods70 consecutive patients with confirmed ischaemia and ictus within 9 h from four centres in China were prospectively recruited. On the basis of our previous work to approximate infarct and penumbra from baseline ADC maps, high-DWI regions were determined as voxels within the approximated infarct core and penumbra as areas with high DWI intensity (greater than average of the mean and maximum). Within the high-DWI regions, the proportion of approximated penumbra voxels in terms of ADC intensity was calculated and denoted as rDWI–ADC. Good clinical response was based on the responder analysis. Receiver operating characteristic analysis was used to derive the optimum threshold of rDWI–ADC to maximise the Youden index. Patients with rDWI–ADC greater than or equal to the optimum threshold were deemed to have DWI–ADC mismatch. We tested the hypothesis that patients with DWI–ADC mismatch would have a better clinical response from thrombolysis than those without the mismatch. The Institutional Review Boards of the four centres approved the study, and written informed consent was obtained from all patients. Findings20 (29%) patients had DWI–ADC mismatch. Thrombolysis was significantly associated with good clinical response in patients with DWI–ADC mismatch compared with no thrombolysis (odds ratio [OR] 35; p=0·002). Thrombolysis based on DWI-ADC mismatch yields a substantially better clinical response than that based on perfusion–diffusion mismatch (OR 8·3; p=0·04). InterpretationDWI–ADC mismatch could be used to identify patients purely from baseline imaging who are likely to benefit from intravenous thrombolysis. FundingChinese National Program on Key Basic Research Project (2013CB733800 and 2013CB733803), Chinese National Science and Technology Pillar Program during the Twelfth Five-Year Plan Period (2011BAI08B09), Shenzhen Key Technical Development Grant (CXZZ20140610151856719), and Shenzhen Basic Research Grant (JCYJ20140414170821262).