Computed Tomography Perfusion in Acute Ischemic Stroke
Is It Ready for Prime Time?
Computed Tomography Perfusion Is Beyond Prime Time
David S. Liebeskind, Mark W. Parsons, and Max Wintermark
Computed tomography perfusion (CTP) is beyond prime time, ready for use to select patients with acute ischemic stroke for intravenous and endovascular reperfusion therapies in routine clinical practice. After 2 decades of fervent stroke research using advanced imaging and revascularization techniques, recent stroke trials astutely combined imaging and therapeutic expertise, likely the recipe for their resounding success. Multimodal CT, including CTP, was a key element in these trials that selected individuals with favorable collateral profiles and resultant improved outcomes after effective reperfusion.1 Academic discourse about optimal thresholds of ischemic core and mismatch volumes comprise late news, beyond prime time.
Literature of only a few years ago aptly predicted the pivotal role of CTP.2 Lev2 eloquently noted the impact of collateral perfusion, critical data from a snapshot of hemodynamics, and the essential imaging goals of delineating core to avert hemorrhage while measuring salvageable tissue at-risk that need not be perfect. Prior trials failed to leverage advanced imaging or alternatively, entwined imaging and reperfusion inefficiently. In 2015 and beyond, triage of acute ischemic stroke for potential reperfusion should rapidly and efficiently identify optimal candidates for thrombolysis and thrombectomy across the broadest population. Multimodal imaging can swiftly address ischemic injury in the brain, arterial occlusion, collateral status, and the topography of perfusion that map the risk of hemorrhagic transformation and nutritive reperfusion. As noted by Lev,2 multimodal imaging is brain and may improve outcomes and optimize costs.
The stroke community has historically been polarized between minimalists and extremists regarding the role and nature of imaging for acute stroke evaluation. Minimalists have argued that noncontrast CT may suffice; yet this approach will no longer work in the endovascular era when presence/absence and location of arterial occlusion and collateral profile are pivotal. In addition, telemedicine supplant refutes the minimalists’ argument that extra imaging is needless if one does not know how to interpret it. At the other extreme, some insist on the need for physiologically perfect measurements, when recent successful endovascular trials have shown that a sound operational concept is sufficient. These trials, including Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN), Extending the Time for Thrombolysis in Emergency Neurological Deficits–Intra-Arterial (EXTEND-IA), Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion With Emphasis on Minimizing CT to Recanalization Times (ESCAPE), and Solitaire With the Intention for Thrombectomy as Primary Endovascular Treatment Trial (SWIFT PRIME), used a considerable proportion of CTP.3–6 They all demonstrated relatively low rates of hemorrhagic transformation and dramatically improved outcomes with effective reperfusion. These studies follow the success of CTP selection with intravenous thrombolysis in Thrombus Aspiration in ST-Elevation Myocardial Infarction in Scandinavia (TASTE).7 It is time for both the imaging minimalists and extremists to learn from recent trials rather than perpetuate archaic arguments that impede progress and promote further trials purely for research perpetuation.
CTP rapidly provides an accurate measure of ischemic core that simultaneously averts risk of hemorrhagic transformation and potential benefit of reperfusion. Even when chronic collaterals may cause extensive delay in time parameters on CTP maps, cerebral blood volume and cerebral blood flow maps can readily demarcate that extent of core and tissue at-risk. Malignant collateral profiles that portend hemorrhagic transformation after reperfusion may be avoided and all other cases may be streamlined toward the now established paradigm for acute ischemic stroke. CTP may short of perfect, yet close enough. CTP core measurements may resolve the current noise or wide variability in reperfusion outcomes of stroke patients with low Alberta Stroke Program Early CT Scores (ASPECTS), as suggested by EXTEND-IA.3 CTP core and mismatch measures may be accurately acquired across different time epochs, with reduced contrast and limited radiation, avoidance of motion artifact, recognizing the impact of technical parameters, delineation of collaterals, automated arterial input and venous output functions with rapid postprocessing techniques, expert interpretation, and application anywhere a patient with stroke may be triaged. CTP is straightforward to apply, even at the community level. It is quick and does not delay treatment. It integrates easily with noncontrast CT and CTA, providing essential data in 1 simple, short imaging study. Multimodal CT can therefore be used for triage as part of both intravenous and endovascular reperfusion therapies.
Computed Tomography Perfusion Is Not Ready for Prime Time
Ramón G. González and Michael H. Lev
The EXTEND-IA3 and tenecteplase8 trials have shown that intervention produces favorable outcomes when patients with stroke are selected using CTP. Does that mean that CTP is ready for prime time? If prime time refers to the use of CTP to decide whether an individual patient should undergo treatment, the answer is no. CTP is simply too imprecise to reliably measure the infarct core—the critical parameter for excluding from therapy patients who are at greatest risk of hemorrhagic complications and are unlikely to benefit. Moreover, there is a more precise alternative, diffusion magnetic resonance imaging (MRI).
CTP and diffusion-weighted imaging (DWI) are different. The inherently poor signal-to-noise ratio of postprocessed CTP images is a fundamental weakness of the technique. Low signal-to-noise ratio measurements may be useful if repeated and a mean calculated; this cannot be done for individual patients. That a strong linear correlation exists between CTP- and DWI-derived ischemic lesion volumes is not surprising because both result from the same arterial occlusion. High correlation in a population, however, does not confer high measurement accuracy in an individual. As Bland and Altman9 pointed out that almost 30 years ago, regression analyses are inappropriate to judge the validity of a quantitative clinical test. More appropriate are difference tests that establish the 95% confidence limits. As shown by Schaefer et al,10 a cerebral blood flow core measurement of 70 mL could actually range from 11 to 124 mL within the 95% confidence limits; other papers in the CTP literature reveal similar variability.
Although this large variability does not preclude using CTP to enroll patients into clinical trials, it does make such selection inherently less efficient compared with using reference standard DWI. Indeed, power calculations show that, for a simulated treatment study designed to detect a 20 mL improvement in final infarct volume, using CTP instead of DWI would require at least twice as many patients to reach significance.10
Given CTP’s relative inaccuracy in delineating core, to what, then, do we attribute EXTEND-IA’s phenomenal 71% good outcome rate using a CTP-based selection strategy? The answer lies in its patient selection criteria. EXTEND-IA—compared with the other successful intra-arterial therapy trials4,5—used a highly conservative selection strategy, cherry picking the best patients with small cores who were likely to do well even with alteplase alone. Targeting small cores minimizes the effects of large measurement errors, at the cost of excluding many who might benefit. The median core volume at <4.5 hours presentation for the EXTEND-IA patients was 12 mL for intra-arterially+intravenously treated and 18 mL for intravenous-only treated, helping to explain their striking 40% good outcome rate among the control group. EXTEND-IA screened 1044 patients to enroll only 70. Moreover, as noted in the discussion of tenecteplase from the recent Alteplase Versus Tenecteplase for Thrombolysis After Ischaemic Stroke (ATTEST) study,11 Parsons et al8 imaging selection criteria resulted in exclusion of 79% (477/604) of patients eligible for intravenous recombinant tissue plasminogen activator, and ultimately only 12% of treatment-eligible patients were randomly assigned, compared with 66% of patients in our study; 124 of 477 (26%) were excluded based solely on CTP criteria.
The recent success of intra-arterial therapy trials3–5 can be attributed to advances in both thrombectomy technology and imaging selection. Positive clinical trials have demonstrated the efficacy of new devices. It is time to move toward the optimal selection of individual patients using the most precise imaging method that is widely available. Our patients deserve nothing less.
Comments by Drs Selim and Molina
The preceding debate leaves us no room for further comments. Our gladiators (3 versus 2) fiercely point out why or why not CTP should be used to identify acute stroke patients who might be candidates for reperfusion therapy in the extended time window. They resort to used-car salesmen’s tactics to make us ponder if close enough is good enough. Is it? The widespread availability of CTP and faster imaging time, compared with MRI, and its successful application to select patients in multiple recent endovascular intervention trials lead us to conclude that using any mode of transportation to get to your destination faster is better than waiting endlessly to save for a Lamborghini! Using CTP is and should be used as an important tool in our armamentarium to identify stroke patients who are likely to benefit from thrombolysis, BUT this should not prohibit us from continuing our efforts to find better and improved tools in the future, if possible.
Rebuttal by Drs Liebeskind, Parsons, and Wintermark
The dwindling debate arguments that oppose CTP implementation for the selection of revascularization candidates in acute ischemic stroke teeter on increasingly archaic and irrelevant descriptions of acute stroke care in 2015. Four specific points are conjured to deny recent advances in clinical research, jeopardizing the ultimate goal of improving the outcome of our stroke patients; the definition of prime time, precision in stroke medicine, the role of DWI, and the nature of future stroke trials. These counter arguments ignore the groundbreaking progress in the use of CTP within the recent EXTEND-IA and tenecteplase trials for endovascular and intravenous revascularization. Several phase III randomized controlled trials have now definitely established the value of CTP for patient selection, heralding the essential prime time validation of such technology in phase IV registries already underway. Feasibility, generalizability, and pivotal details of implementation in routine clinical practice are key goals of such registries that will overwhelm the amount of data just recently acquired in randomized controlled trials. As previously noted, CTP selection should be empirically defined by precision of rendering a logical decision in the care of an individual stroke patient. There is no need to resort to the bland argument cited from 30 years ago about the limitations of population-based imaging assessments, as precision stroke medicine, including the use of imaging for an individual patient now shares the prime time stage with CTP. Arguments about DWI are irrelevant in a debate on CTP and impractical for large-scale stroke imaging initiatives. If patients with stroke could be treated only at sites that offer emergent MRI 24/7 without delaying treatment, many more patients would be deprived compared with the 26% of patients excluded by CTP in EXTEND-IA as waved. The academic deliberations of CT versus MRI should not detract from our shared goal of improving stroke outcomes. Our patients deserve much more!
Rebuttal by Drs González and Lev
Both sides agree that clinical trials have demonstrated that thrombolysis and thrombectomy are effective treatments for stroke. Both agree that identifying a target occlusion is important, and both agree that measurement of the infarct-core is critical. The controversy centers on whether core measurement by CTP is sufficiently precise to be used for treatment selection in INDIVIDUAL patients? A wealth of theoretical, experimental, and clinical evidence suggests the answer is no.
CTP measures hemodynamics, not tissue status. Hence, although a marker for irreversible injury absent timely reperfusion, CTP—which reflects a snapshot-in-time—is not a marker for treatment futility. Not surprisingly, validation studies in animal models are sparse and have not been reproduced. All published clinical data are consistent: CTP core estimates have high error. Although CTP may be adequate for selection of patients with small cores, where large measurement errors are of little consequence, the cost is exclusion of many with a high likelihood of treatment benefit.12 CTP core lesions segmented using automated software offer the illusion of quantitative accuracy that simply does not exist.
Our pro-CTP friends argue that CTP may short of perfect, yet close enough. Traumatic brain injury imagers recognize that fractional anisotropy maps are too inaccurate for individual patient assessment. Would an internist accept a blood glucose or international normalized ratio measurement with >50% error as close enough? No, she would not. Why, then, should stroke neurologists accept a core measurement error of >50% as close enough? Clearly, they should not—especially when a more accurate alternative is readily available.
Sources of Funding
Drs Liebeskind and Wintermark are funded by National Institutes of Health–National Institute of Neurological Disorders and Stroke awards (NIH/NINDS) K24NS072272, R01NS077706, and R13NS089280.
Dr Parsons receives grant support from National Health and Medical Research Council of Australia. Dr Lev serves as a consultant for GE Healthcare and Millenium Pharmaceuticals and receives grant support from GE Healthcare. The other authors report no conflicts.
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
- © 2015 American Heart Association, Inc.
- Saver JL,
- Goyal M,
- Bonafe A,
- Diener HC,
- Levy EI,
- Pereira VM,
- et al
- Schaefer PW,
- Souza L,
- Kamalian S,
- Hirsch JA,
- Yoo AJ,
- Kamalian S,
- et al
- Huang X,
- Cheripelli BK,
- Lloyd SM,
- Kalladka D,
- Moreton FC,
- Siddiqui A,
- et al
- Bivard A,
- Levi C,
- Krishnamurthy V,
- McElduff P,
- Miteff F,
- Spratt NJ,
- et al