Luciana Catanese, MD
Appropriate patient selection and expedient treatment times have had a pivotal role in the success of endovascular thrombectomy in acute ischemic stroke (AIS). Perfusion imaging may be a tool in further refining optimal patient selection for endovascular acute stroke therapy and broadening current treatment boundaries. In this regard, the authors aimed to determine the computed tomographic perfusion (CTP) thresholds and accuracy for the prediction of irreversibly injured brain tissue at 24-48 hours in patients achieving early reperfusion (<90 minutes from CTP), reperfusion within 90-180 minutes and in those who did not reperfuse acutely.
AIS patients that received endovascular therapy as part of a prospective, multi-center study (Prove-IT) who were >18 years old, presented within 12 hours of a known symptom onset time, and with complete arterial occlusion were eligible. Recanalization was assessed on conventional or CT angiography. All patients underwent CTP on admission and a non-contrast head CT (NCCT) or MRI- diffusion weighted imaging (DWI) at 24-48 hours to establish the core of infarct. Cerebral blood flow (CBF), cerebral blood volume (CBV) and Tmax for both grey (GM) and white matter (WM) were calculated. Patient data was initially stratified in two groups (those who underwent CTP within 180 minutes and those who did it later), which were subsequently categorized into three sub-groups based on time from CTP to reperfusion (<90 minutes, 90-180 minutes and no acute reperfusion). The area under receiver operator characteristic (ROC) curve and Youden’s method were used to determine the most accurate thresholds for prediction of final infarct core along with the respective sensitivities and specificities for each threshold.
Overall, 132 out of 146 initial patients were included in the analysis after 14 patients were excluded due to inadequate CTP scans. When looking at the group who underwent CTP within 180 minutes from symptom onset, the Tmax thresholds (>16 s average for both GM and WM) had the highest accuracy for core infarct estimation (sensitivity 0.85 and specificity 0.81) if successful reperfusion [TICI 2b-3] was achieved early (<90 minutes), whereas CBF (< 8.15 ml.min-1 .100g-1 average for GM and WM) and CBV had moderate (average sensitivity 0.79 and specificity 0.82) and low (average sensitivity 0.5 and specificity 0.6) discriminative ability, respectively. For the rest of the groups and its corresponding subgroups, Tmax and CBF had similar accuracies and CBV had the least discriminative power. Interestingly, the time from stroke symptom onset to CTP had no significant impact on the CTP thresholds.
In summary, the authors showed that the discriminative ability of CTP in infarct core estimation at 24-48 hours laid mostly on the Tmax and CBF and was inversely correlated with the time from imaging to effective reperfusion. CBV did not prove to be a reliable parameter for this purpose.
Although perfusion imaging may ultimately prove to be useful in patient selection for endovascular thrombectomy, there still is quite a long road ahead. Perfusion imaging in acute ischemic stroke remains a snapshot of intricate and dynamic vascular pathophysiology. Validation of these results in future studies that employ systematic DWI for defining infarct core, concomitant penumbra measurements and much larger sample sizes, will aid in better understanding the applicability of these findings to clinical practice.
With this we should be able to come up with an estimate of the rate of neuronal death during the neuronal cascade of death. Some amount less than 1.9 million a minute. That might focus research on solving that problem.
Yes, exactly! Isn't it really amazing that some day we could actually do that? We certainly need more research to determine most accurate software parameters (Tmax>6 or 9?) and perhaps even include measures of cerebral metabolic rate.