Leaving the Black Box Approach: Individualized Prediction of Recanalization Benefit by Advanced Imaging in Acute Stroke

The enthusiasm about mechanical recanalization in acute stroke has experienced a ‘state of hibernation’ due to inconclusive studies several years ago but has now been ‘reperfused’ by overwhelming evidence from recent randomized trials.^1–3 On the one hand, the short message delivered by these studies is the following: recanalization is beneficial if early and complete. This is not a surprising finding for the stroke community but it is well embedded in a clinical setting. On the other hand, these studies open the door to treatment efforts in a late time window (that still awaits clear definitions), ⁴ and to the use of advanced imaging biomarkers as for example collateral flow assessment or imaging of the tissue at risk. ⁵

This leads to a crucial issue inherent in stroke imaging studies: acute neuroimaging is a snapshot and reflects morphological changes between stroke onset and the time point of imaging. The following time span from acute imaging to (potential) reperfusion/ recanalization—spontaneously, by thrombolysis or by mechanical recanalization—is usually not captured by and hemodynamic repeated imaging. However, it may cause additional metabolic and hemodynamic alterations and may hamper the predictive power of acute stroke imaging, even if advanced techniques and definitions of the tissue at risk are applied. Moreover, the response to ischemic challenge might be different on a patient level due to individual factors. Therefore, the tissue outcome may be different in two patients even though the initial imaging signature and the recanalization characteristics are similar.

In this respect, the study by Kemmling et al ⁶ in the current issue of JCBFM presents a timely and very important contribution. In a sophisticated retrospective approach to a large multicentric database of patients receiving mechanical recanalization, pre-therapeutic computed tomography perfusion data sets were used to calculate infarct risk on a voxel level. In contrast to many other imaging studies, two important levels of information were included: time/degree of recanalization and patient-specific data. They demonstrate the overall time-dependent effect of recanalization, an average infarct reduction of about 50% in recanalization within 5 hours but also a possible tissue benefit up to 15 hours in individual patients. This cross-sectional study design with advanced modeling yields a good approximation to serial imaging that is hardly feasible in the clinical setting.

The study also confirms that advanced stroke imaging can offer additional information and can support stratification in stroke trials. In the near future, the proposed model has to be translated into clinical decision making in order to move acute stroke therapy one step closer to a tailored therapy instead of the ‘one fits all’ approach.