Introduction of diffusion-weighted imaging (DWI), perfusion imaging (PI), and magnetic resonance angiography (MRA) into a comprehensive acute stroke evaluation paradigm is a formidable challenge in most centers' busy imaging schedules. If time is brain, the minutes required to identify the acute stroke patient, perform a CT scan, transport the patient from the emergency department to the MR scanner, wait for the scanner to become available, transport the patient to the scanning table, and then scan, process, and interpret the results, are precious minutes indeed. The time to accomplish these tasks can be shortened, as shown in the quality improvement document by Schellinger et al (page 1184) in this issue of the AJNR. The authors confirm that a concentrated effort to decrease the time until imaging can lead to considerably shortened MR examination time. These results challenge us all to reconsider our opinions and prejudices about our ability to perform MR imaging for early stroke.
That DWI/PI/MRA can contribute significant knowledge in acute stroke evaluation is unquestioned. The triad of a small wedge of diffusion abnormality, followed by a larger zone of perfusion delay (DWI/PI “mismatch”), accompanied by middle cerebral artery (MCA) occlusion on MRA, speaks volumes regarding the lengths we have come in stroke imaging and its promise in patient selection. It still does not, however, clarify the dilemmas why, where, when, and who to scan.
Why? Presumably, to direct performance of some intervention, or to recommend against performance of the intervention, in order to improve outcomes in the acute stroke patient. IV Activase in the 0–3-hour window, or even beyond the currently approved 3-hour window, might be administered only if appropriate DWI/PI criteria are present (such as a “mismatch”), but no MR data as yet suggest delaying its administration improves safety by identifying those patients who cannot improve, or reveals circumstances that may render the patient more susceptible to hemorrhage. Such circumstances might include:
(a) DWI has been suggested to show multiple infarcts in 10% of patients, perhaps from a central cardiac source, which need not be of identical age. It may be more than coincidence that 10% of hemorrhages in the National Institute of Neurological Disorders and Stroke (NINDS) trial occurred distant from the principal infarct. Thus, DWI might eliminate the risk of tissue plasminogen activator (TPA) administration when an older, silent infarct exists.
(b) If a DWI/PI “matched” defect was apparent prior to, or even during, IV TPA administration, the infusion could be stopped, presumably to reduce the likelihood of a hemorrhage when little benefit is expected.
(c) Failure to reveal a diffusion abnormality may obviate treatment. Or should it? False-negative DWI findings has been reported in up to 15% of posterior circulation infarcts shown by 48 hours. Are these patients who might still benefit from early thrombolytic therapy? Perfusion problems may be resolved with therapy in the absence of diffusion abnormality.
Again, the data to support these philosophies in the 0–3-hour treatment window are lacking.
Intraarterial (IA) recanalization efforts might be employed in any time window if a large perfusion abnormality and a small or absent diffusion abnormality (DWI/PI “mismatch”) were present. Yet, MR studies to date suggest “matched” defects, which might eliminate recanalization therapy, occur in approximately 20% of patients selected for scanning. If treatment is delayed to identify these matches in the 20% of patients who might not benefit or who might be susceptible to hemorrhage, are the other 80% of treatment candidates subjected to important delays in therapy? Research is needed to define the greater risk.
Where and when? Presuming ultrarapid MR imaging becomes a standard of imaging, must every emergency facility be DWI/PI/MRA-capable? One might argue that in the acute treatment setting this capability might be necessary only if acute therapy can be administered at the site. If a site is not therapy-capable, is it really prepared to perform and interpret the requisite images? And if a site is neither treatment- nor performance/interpretation–capable, can the delay inherent in its participation in stroke imaging be warranted? Should stroke suspects be identified by emergency management teams and triaged to acute stroke treatment centers that have the capabilities to work rapidly, as Shellinger has accomplished? The presumption must be that faster is better, and Shellinger et al point out that major centers can become faster.
Again, is the time required to identify, transport, scan, interpret, and plan therapy time well spent, or is it lost time that only diminishes good outcomes? The evidence that delays have a negative impact is compelling. Marler demonstrated in the NINDS trial that even with IV TPA, delaying therapy 20–30 minutes may diminish the likelihood of favorable outcomes by 10–20% (1). Kanter et al reviewed the post-NINDS experience in Cincinnati and reported similar findings (2).
In addition, a number of recent IA thrombolytic therapy reports also suggest that earlier treatment leads to better outcomes. Certainly one would not take a patient with an arteriographic thromboembolic occlusion of the M1 segment (the ultimate, ultra-early treatment candidate) off of the angiographic table to perform a DWI/PI scan. At the other end of the time spectrum, the Prolyse in Acute Cerebral Thromboembolism Trial (PROACT) II achieved good outcomes in 40% of patients when IA therapy was begun at 5.3 hours. Suarez reported good outcomes in 56% of patients at 4.75 hours in a more heterogeneous population (3). Bendszus reported good outcomes in 66% of patients with MCA occlusion treated in less than 4 hours (4). The Emergency Management of Stroke (EMS) Pilot Trial achieved good outcomes in 66% of patients with M1 and M2 occlusions when IA treatment was begun at 4.2 hours (5). Our soon-to-be-reported 1999 experience with 20 patients who had carotid-distribution occlusive disease (mean National Institutes of Health Stroke Scale Score [NIHSSS] = 20), treated with IV TPA within 3 hours, followed by arteriography, then treated with IA TPA at a mean of 3.3 hours in 16 of the 20 patients, achieved good outcomes in 65% of patients (R. Ernst, personal communication). These multi-study data points also suggest a 10–20% decrease in good outcomes with 20 minutes of delay until IA therapy. Elimination of how many genuine treatment candidates by DWI/PI/MRA scanning warrants even a 20-minute delay to rapid recanalization efforts at the latter early time windows? If good outcomes can be achieved in two thirds of the patients within 4 hours with recanalization efforts anyway, should we delay treatment to scan? Will DWI/PI/MRA somehow eliminate the other third from treatment in the first place? How will clot removal devices, and immediate recanalization, affect this management paradigm?
Who? Some argue that the heterogeneity of vascular occlusive disease diminishes the significance of any clinical finding in the acute stroke patient, and rapid vascular imaging (eg, MRA) may be mandatory to begin the treatment triage. Many disease processes may mimic thromboembolic cerebrovascular disease, including migraine, seizures, and inflammatory disease. Should everyone with CNS symptoms/signs be urgently imposed into the daily scanning schedule of extremities and spines, even to the point of removing a patient from the scanner? Or are there clinical findings sufficiently predictive of the presence of acute major thromboembolic occlusion that warrant rapid intervention? Even in the most sophisticated centers, patients have been treated with IV TPA only to establish subsequently that the clinical problem was seizure activity. This, however, is the exception.
Some have argued that “angiograms in suspected stroke patients don't always show blocked vessels, so how do we know these patients had strokes? IA recanalization efforts shouldn't begin without that demonstration.” In the EMS trial (in which patients were treated with IV TPA or placebo, then underwent arteriography and treatment with IA TPA if an occlusive lesion was revealed), patients without vascular occlusion on arteriography subsequently demonstrated infarcts on imaging in 10 of 10 cases (5) with presumptive evidence of arterial occlusions. No occlusions at angiography were shown in approximately 20% of 480 PROACT II patients subjected to angiography (6), and we await the follow-up imaging and clinical outcomes in those patients.
The clinical description of stroke patients in the United States is defined by the NIHSSS. In the NINDS pilot trial, we recognized that patients with an NIHSSS < 10 seldom had a major clot at arteriography following therapy. Patients typically did well by 3 months, and we hypothesized that IV therapy alone might be sufficient in the majority of such patients (7). The latter observation subsequently was confirmed in the NINDS trial; 52% of such patients had excellent outcomes. Perhaps patients with an NIHSSS < 10 should be treated immediately with IV TPA, and then have a DWI/PI/MRA to determine larger vessel occlusion.
The NIHSSS in the acute stroke patient also predicts the presence of a major arterial occlusion and a potentially lysable thrombus. Patients with an NIHSSS > 10 have up to a 50% likelihood of MCA occlusion, as documented by the hyperdense MCA sign (HMCAS) (7). The EMS study revealed major occlusions in all patients with an NIHSSS > 14, and 78% of patients with an NIHSSS > 10. In PROACT II, only 15% of 180 patients with M1 or M2 occlusions had an NIHSSS < 10. Furthermore, the control group did as well as the treated patients with an NIHSSS < 10 on an overall-improvement basis. So, the clinical examination does tell a lot about the arterial occlusive process and outcome, despite the heterogeneity of the arterial occlusive lesion. Is MRA/DWI/PI needed in a 65-year-old patient with atrial fibrillation and an NIHSSS of 15, whose CT at 90–120 minutes postictus reveals an HMCAS before being taken to the angiographic suite to recanalize the vessel? If not, at what point does the benefit of imaging to exclude patients (perhaps 10%….or 20%…or 30%) outweigh the risks of inherent delay before attempted recanalization? To me, these are the real questions: what percent are eliminated, at what time, and at what risk of delay for imaging?
Once again, I have asked myself more questions than I have answered, with full realization that nihilism, prejudice, and bias lead to no new scientific knowledge. The image of the stopwatch ticking has been the poster child for acute ischemic stroke; however, perhaps it is time to change that poster image to one of a compass, to reorient us in the right direction, much like the refocusing 180-degree pulse of the spin-echo sequence. Which direction is right is the question. Certainly research efforts devoted to “how far out” we can treat patients is important, even if the number of treatment candidates is small, while the number of patients eliminated from treatment is large. A research effort to perform DWI/PI/MRA on all patients to determine why, who, when, and where is to be applauded and encouraged as we assimilate all the data being accumulated regarding stroke treatment. The effort exerted in diminishing times to MR imaging as reported by Schellinger et al, certainly demonstrates how we can improve, and efforts in that regard ultimately will pay dividends in better patient selection for therapies. Unquestionably, if the DWI/PI/MRA evaluation were instantaneous as the patient passed through the portals of the emergency department, its use would be a no-brainer. And some day it probably will.
Is the emphasis to be placed in the direction of MR scanner improvement, or should it be in the direction of universal patient education, faster EMS identification of stroke patients in the field, and more rapid patient triage, evaluation, and treatment, all with minimum delay? Efforts expended in the latter direction will reap the greatest immediate rewards for more patients in the early time window. In addition to our efforts in speeding up the imaging process itself, in order to help those individuals not seen until a later time window, it should be the responsibility of neuroimagers and neurointerventionalists to cooperate maximally with the stroke therapy infrastructure to achieve that end as well.
- Copyright © American Society of Neuroradiology