Article Text

Download PDFPDF

Emerging impact of CTA/perfusion CT on acute stroke thrombolysis in a community hospital
  1. M N Brant-Zawadzki,
  2. D M Brown,
  3. L A Whitaker,
  4. W W Peck
  1. Department of Radiology, Hoag Memorial Hospital Neurosciences Center, Newport Beach, California, USA
  1. Correspondence to Michael N Brant-Zawadzki, Department of Radiology, Hoag Memorial Hospital Neurosciences Center, One Hoag Drive, PO Box 6100, Newport Beach, California 92658, USA; mbrant{at}


Our objective was to retrospectively review the emerging role of CT, CTA, and perfusion CT (pCT) in the hyperacute stroke population of a community hospital. We reviewed 50 consecutive patients' records and imaging studies, who were treated with thrombolytic therapy within 6 h of symptom onset. Multidetector CT, CTA, and pCT studies were evaluated. Subsequent CT, magnetic resonance, or angiographic studies when available were correlated. Patients' clinical data at admission and outcomes at discharge were evaluated. Complications were tabulated. Of the 50 patients treated with thrombolytics, 37 had CT/CTA/pCT, the others non-contrast CT only. CT blood volume defect was present in a total of 14 patients, presaging permanent infarct in all. Arterial clot was seen in 28/37 CTAs (carotid “T” 6, MCA 16, vertebrobasilar 6). Viable penumbra was shown in 20/37; rescued penumbra was depicted after treatment in 14.

39 patients were treated with intravenous, nine with intra-arterial, two with both forms of thrombolysis. Modified Rankin score showed clinical improvement in 58%, three patients had complete recovery. Subsequent bleed was shown in two (4%), symptomatic in one (2%). Two patients died. Our experience suggests advanced CT is more sensitive to ischemia than routine CT, that salvageable penumbra can be identified, and that triage of patients with acute stroke for thrombolysis with CT/CTA/pCT is more robust than routine CT alone, and may improve outcomes in the community hospital setting.

  • Stroke
  • Thrombolysis

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Based on the guidelines developed from evidence of the National Institute of Neurologic Disorder and Stroke (NINDS) study,1 intravenous thrombolytic therapy is accepted treatment of acute ischemia in the first 3 h in the appropriate clinical context and a non-contrast CT that excludes a large infarct or bleed. Nevertheless, <5% of all eligible patients are treated, the benefits of widespread thrombolytic therapy in the community setting having been questioned, and its use remains inconsistent.2 Appropriate patient selection with limited data in the acute stage of a non-specific clinical syndrome, and fear of hemorrhage are two of the issues which seem to hinder acceptance of more widespread use.

In recent years, magnetic resonance (MR) imaging has assumed a primary role in evaluating patients with suspected cerebral infarction, utilizing diffusion-weighted and perfusion-weighted techniques, coupled with MR angiography.3 4 5 Contemporaneously, multi-detector CT (MDCT) development has produced instruments capable of sub-second temporal resolution of contrast transit, post-processing capabilities which provide brain perfusion data more sensitive to ischemia than routine CT, and depiction of arterial structure rivaling that of conventional angiography. Indeed, the advances of both MR and CT have prompted even academically based proponents of evidence-based guidelines to champion these newest techniques to influence treatment decisions in the acute stroke setting despite the absence of high-level evidence.6

The pragmatic advantage of CT's ability to produce such datasets in less than 10 minutes of combined study/post-processing time even in uncooperative patients has promoted use of this technology over MR (and its environmental obstacles) in our community hospital setting for evaluating acute stroke, verifying ischemia, identifying site of occlusion, and initiating thrombolytic therapy. Also, MR services are not available around the clock, and even during the day, when MR is sufficiently busy such that emergency patients cannot easily be studied at a moment's notice.

We have been routinely using MDCT, CTA, and pCT scanning in evaluating patients with hyperacute stroke (witnessed onset within 6 h) and triaging appropriate patients to intravenous or intra-arterial thrombolytic therapy aided by the results of the study. The purpose of this review was to retrospectively evaluate the findings provided by modern CT technology, and their impact on management of patients with acute ischemic stroke, within a community hospital setting.

Materials and methods

The medical records and imaging studies of 50 consecutive patients treated with thrombolytic therapy for acute ischemic insult were retrospectively reviewed. All patients were treated within 6 h of witnessed symptom onset, either by intravenous or intra-arterial thrombolytic therapy (or both). Intravenous treatment was with tissue plasminogen activator (0.9 mg/kg). Intra-arterial treatment was used only in patients who had documented large-vessel occlusion on CTA, and given at the site of the clot, using recombinant tissue plasminogen activator, administered by a combination of hand infusion and drip (up to eight units total). CT without contrast was performed in all of these patients; an additional 37 patients had CT combined with CTA and pCT scans and form the major focus of this review.

CT protocol

CT was performed initially with eight, and then 16 and 64 detector spiral CT scanners (Light Speed, Volume CT, GE, Milwaukee, Wisconsin, USA). Initially, 2 cm brain volumes were available for the perfusion study (eight and 16 detector scanners); subsequently, a 4 cm volume was obtainable (64-slice scanner). Contrast (Omnipaque 350 mg of iodine per ml; Amersham, Oslo, Norway, or iso osmolor Visapaque, 270 mg of iodine per ml; GE Healthcare), was used for the CTA pCT study. Injection rates varied from 2 to 4 ml/s and the anatomic scans utilized slice thickness of 0.625 mm, pitch of 1.375 or less. Slice acquisition was begun at the level of the aortic arch, and extended through the brain for the CT angiographic portion of the study. The total volume of injection was typically 70 cc for the angiographic portion of the study which covered the arch of aorta to the brain vertex, with an additional bolus of 40 cc for the perfusion component. The perfusion scans were centered on the foramina of Monro, utilizing typical external landmarks. Four adjacent 5 mm slices were analyzed, in the original patients studied with 8 and 16 slice scanners, eight 5 mm slices were processed when patients were scanned on the 64-slice scanner. Matrix with the acquisition data was 512 × 512; time density curves were calculated for each pixel and data analyzed using software provided by the vendor based on the central volume principle (Advantage Workstation 4.2. GE).7 Briefly, pCT measures transit time of intra-vascular contrast through a known volume (voxel matrix) of brain tissue. Color-coded pixel maps of blood volume are generated as slices. Subsecond sampling of change of CT attenuation values per voxel occurs by repetitive scanning of the volume of interest. Thus, transit time of contrast is presented in color-coded maps. Regional cerebral flow (CBF) can be calculated through the formula CBF = CBV/MTT (CBV = cerebral blood volume, MTT = mean transit time). Calculated perfusion scans are then generated. (For example, with complete loss of regional contrast inflow (indicating lack of collaterals too), CT will detect no change in Hounsfield units over the dynamic time frame of the first pass pCT study. All three color-coded maps will show a defect in such cases.) The vessel chosen for baseline arterial inflow data was that opposite the hemisphere of suspected insult. The sagittal sinus or torcula was used for venous outflow data point. It is important to note that we have trained our CT technicians to do the scans and the post-processing immediately afterwards, with average scan to data availability of 10 minutes.

Record review

The routine CT brain images were evaluated for the presence of acute ischemic changes or hemorrhage. CT angiographic studies were reviewed with attention to morphology of the major vessels on axial, coronal, and sagittal reconstructions including the carotid bifurcation, the vertebral arteries, and specifically the presence of any carotid siphon and intracranial major vessel occlusion or presence of filling defects. The perfusion maps were evaluated qualitatively (visually) for asymmetry in major vessel territory of any of the three following parameters: blood volume (the perfused intravascular space as measured by the Hounsfield number increase caused by the accumulation of iodinated contrast over baseline within the voxels sampled), transit time (time to peak of the Hounsfield values as measured by the repetitive scans following injection) and cerebral perfusion (calculated for each voxel based on the first two parameters). Penumbra was defined, as significant (greater than 50% by subjective estimate of the radiologist) territorial mismatch between blood volume abnormality (if any) and transit time slowing and perfusion map defect. Subsequent CT, MR, or angiographic studies were correlated when available. The patients' subsequent hospital course and clinical diagnosis were noted through record review.

Patient hospital course was reviewed initially for time of symptom onset, neurological stroke scale on admission NIHSS (National Institutes of Health Stroke Scale) when available, Rankin scale, type of thrombolytic therapy performed, and subsequent hospital course. Outcomes at discharge were evaluated, including discharge NIHSS and Rankin scales. Complications of therapy (eg, intracerebral bleed or death) were tabulated.


The 50 treated patients ranged in age from 29 to 97 years. Thirty-nine of these were treated with intravenous thrombolytic therapy, 9 with intra-arterial, 2 with both. All had initial routine CT; 13 were treated solely based on this, 37 of the patients additionally had contrast enhanced CT/CTA/pCT. Thirteen of all 50 patients demonstrated a small zone of ischemia on non-contrast CT, an additional 13 demonstrated ischemia on pCT (total 26 of 50). Arterial filling defects responsible for observed perfusion defects were seen in 28 of 37 patients (carotid “T” 6, middle cerebral artery 16, vertebrobasilar trunks 6). Posterior cerebral artery perfusion defects were seen in all six of the vertebrobasilar occlusions (cerebellum was outside the field of view). Viable penumbra (defined as transit time regional perfusion deficit with no or small blood volume defect) was visible in 20 of 37 patients. Rescue of significant component of the penumbra was documented in 14 patients either on the basis of follow-up CT perfusion study or MR (figures 1 and 2). All of these 14 patients showed clinical improvement. Two small lacunar infarcts in the basal ganglia subsequently seen in patients with larger territorial infarcts were missed on the CT/CTA/pCT study.

Figure 1

Elderly patient, 2 h of left hemiparesis, altered mental status, treated with intravenous tissue plasminogen activator. (A) Non-contrast CT scan demonstrates high density in region of right middle cerebral artery trifurcation (arrow). (B) Single slice from coronal reconstruction of CT angiographic study shows filling defect at the trifurcation of the right middle cerebral artery. (C) Three-dimensional reconstruction shows the trifurcation defect of right M1. (D) Cerebral blood volume, transit time, regional cerebral perfusion maps (left to right) at two adjacent levels demonstrate matched blood volume defect in right corpus striatum, but preservation of blood volume with transit time and perfusion defects in the distribution of the right middle cerebral artery. (E,F) Diffusion and fluid attenuated inversion recovery slices demonstrate acute ischemic insult of right corpus striatum, as predicted by CT blood volume data, but sparing of penumbra in right middle cerebral artery territory. Magnetic resonance angiography (not shown) demonstrated recannulization of M1 trifurcation (magnetic resonance imaging/MRA study obtained next day).

Figure 2

Patient in late 70s with acute aphasia, right hemiparesis onset, at the end of a negative coronary arteriogram. Plain CT (at 90 minutes) was negative, CTA showed a M2 branch filling defect (not shown). (A) Blood volume, transit time, regional cerebral perfusion maps demonstrate zone of penumbra in the anterior left middle cerebral artery distribution. (B) Selected frames from angiogram demonstrate filling defect in left M2 branch of left middle cerebral artery, and lack of anterior middle cerebral artery branches on the lateral view; note catheter for intra-arterial infusion on the lateral view. Third frame shows reappearance of the middle cerebral arcade after successful thrombolysis. (C) Routine CT, transit time and perfusion maps show no abnormality the next morning. Patient was asymptomatic and was discharged.

Figure 3

Middle-aged patient in emergency room with altered mental status and acute left arm paresis, was seen normal less than 3 h previously. Routine CT without and with contrast was negative. Patient was thought a candidate for thrombolytic therapy. (A) Blood volume, transit time, perfusion maps show findings that suggest increased blood volume, rapid transit time, and increased regional cerebral perfusion in the right posterior temporal lobe. Based on these data, EEG was performed and found consistent with right temporal status epilepticus. Clinical diagnosis of Todd's paralysis was made. (B) Magnetic resonance diffusion, FLAIR, and MRA study performed subsequently demonstrated no focal abnormality of the brain or middle cerebral artery.

Blood volume defect was shown in 14 patients, and predicted the portion of brain where permanent infarction was seen on subsequent CT or MR in all (figure 1). In general, the greater the zone of penumbra, the more profound the initial clinical deficit, but lack of good documentation from emergency room data precluded a formal analysis.

Outcome analysis demonstrated improvement in modified Rankin scores, as judged clinically in retrospect, in 58%. NIHSS stroke scale at time of admission was only available in 15 of our patients and improved greater than three points in seven of those (46%), three of whom had complete recovery of their presenting syndrome. Clinically significant hemorrhage occurred in one patient (2%) treated intra-arterially on the basis of CTA/pCT; a second patient had a clinically insignificant bleed. Two patients died as a direct result of their infarct, one of whom had a hemorrhagic conversion.


Our results document that pCT is more sensitive to hyperacute ischemia when compared with routine CT, that it shows viable penumbra as well as permanent infarction, and when combined with CTA, helps guide appropriate management with thrombolytics including endovascular intervention. However, pCT may miss small lacunar infarcts. The known pathophysiology of cerebral blood flow and ischemia helps explain the findings discussed. Normally, the brain requires 20% of the cardiac output, tissue perfusion being over 50 cc blood/100 g brain tissue/min. Blood volume, normally representing 4% of brain volume overall, drops in regions with flow occlusion but collateral pathways and autoregulation may compensate to a degree. With total loss of blood flow, brain cells die within minutes, but collaterals can allow a therapeutic window of several hours. At 20 cc/100 g/min of regional perfusion or less, electrical (and hence clinical) dysfunction begins. Levels below 10–12 cc/100 g/min for more than 30 minutes correlate with permanent infarction in experimental models.8

Although quantitative data are available with pCT, their reproducibility have been questioned, and visual analysis has been found effective in identifying penumbra and presaging subsequent infarct.9 Highly metabolic sites such as the basal ganglia or even deep white matter at the end zones of blood supply with no collaterals may experience brief (less than 1 h) periods of complete ischemia sufficient to kill the cells, yet reflow is seen into these on a CT study shortly thereafter, explaining one possible mechanism of missing a lacunar infarct with CT/pCT study. On the other hand, missing such lacunes in the early stages of patient management may be less important than detecting larger salvageable zones of partially ischemic brain tissue. In addition, a negative pCT does not preclude intravenous thrombolytic therapy when all other NINDS criteria apply, particularly as the pCT study does not cover the entire brain.

The ability to analyze cerebral circulation with dynamic computed tomography was proposed in the earliest days of CT.10 The ability to provide carotid angiographic display from CT data was demonstrated long ago.11 However, evolutionary advances in the temporal and spatial resolution of these techniques, have allowed the ability to evaluate the status of the blood vessels and any filling defects, together with the perfusion physiology of the brain with its greater sensitivity to ischemia.

Our growing experience in a community hospital setting suggests that modern CT technology can facilitate the adoption of acute thrombolytic therapy even by inexperienced neurologists, and yield outcomes as good as or better than those reported by others.12 These prior authors also found a wide variability in the use of such therapy across the country and speculated on the reluctance to provide a perceived high-risk therapy based solely on a single study which used non-contrast CT for triage. Indeed, our experience to date suggests that advanced CT can provide stronger justification and triage such patients efficiently and perhaps more appropriately for (or away from) thrombolytic therapy, especially when intra-arterial administration was contemplated, when compared with non-contrast CT. Although the small number of patients precludes comparison of outcomes between our patients treated based only on routine versus advanced CT, our overall percentage of improved patients exceeds that reported in the NINDS trial, thus suggesting a more appropriate selection of patients occurred. During the early period studied, we also offered thrombolytic therapy to those patients with large matched perfusion blood volume defects (despite no or small zones of ischemia on routine CT). The retrospective targeting of only treated patients did not allow for meaningful comparison of outcomes in treated versus non-treated groups.

Our study was meant to show the evolution of stroke management from the meager evidence based on a non-contrast CT to the advanced CT era, and cannot verify that patient outcomes will be better. Clearly, decisions that are based on more physiologic parameters should impact appropriate choice of therapy. It has been speculated that patients with large-vessel occlusions who undergo intra-arterial treatment represent a distinct population with potentially worse outcomes than those with small-vessel occlusions for whom intravenous versus thrombolytics may be more appropriate. Future studies made possible by CTA/pCT may tease those subgroups apart. Also, factors such as the percentage of salvageable brain and presence as well as robustness of collaterals will be worthwhile predictive factors to evaluate. Other authors have already provided data that suggest pCT can be used to identify salvageable penumbra and extend the window for intravenous thrombolysis beyond 3 h.13 14

The experience reported here further underscores the dissociation between clinical symptoms and physiology. Thus, a number of our patients not treated nor included in this analysis had transient symptoms which completely resolved within minutes of the study which demonstrated MTT and CBF defects. Because some patients were discharged with no treatment or follow-up it is not known if any of those suffered “silent” infarcts. Transient embolic events may explain some of these cases, as small clots can lyse with no permanent sequelae, the true transient ischemic attacks (TIAs). Studies have shown a rate of infarctions as high as 72% in patients with clinical TIAs, depending on their duration.15 16 Some have suggested the entity of “incomplete infarction” for this spectrum of pathophysiology.17

It has been speculated that some patients treated in the original NINDS tissue plasminogen activator trial on the basis of non-contrast CT may not have been appropriate candidates for thrombolysis. Several conditions such as dissection, post-ischemic hyperperfusion state, postictal state can produce or mimic acute ischemia but may be inappropriate for institution of thrombolytic therapy. We have personally seen patients in whom a transient ischemia-like syndrome occurred, was in the process of resolving during and completely resolved shortly after the CT study was performed, in whom no therapy was given because the study showed no arterial defects, and a normal perfusion picture.

Additionally, one of our patients who initially fit the criteria but was not treated, showed hyperperfusion on pCT with a matching syndrome (mimicking stroke) due to a silent seizure (figure 3). This patient would have been judged appropriate for intravenous tPA on clinical and plain CT criteria. A similar group of patients has been reported.18 We have also seen hyperperfusion syndromes in post-endarterectomy patients, which mimicked acute ischemia due to re-occlusion. Such patients suffer from loss of autoregulation in the distal vascular bed beyond a severely flow-restrictive lesion. Their dysfunctional arterioles are maximally dilated to “suck” blood in, but cannot quickly respond to the suddenly increased perfusion pressure following endarterectomy or angioplasty, thus the “breakthrough” edema and/or hemorrhage.19 20 21

CT does suffer from several major disadvantages when compared with MRI. First of all, as shown here, small-vessel or perforator lacunar insults can be missed. Fortunately, as mentioned, such lesions do not impact clinical management decisions in the first several hours following infarction. Next, given the current typical CT technique, only a portion of the brain anatomy can be covered for the perfusion analysis. However, others have documented complete brain coverage with the “toggle table technique” which we did not employ.22 The disadvantage of using doses of intravenous contrast up to 130 cc is potentially relevant; however, in none of our cases was there any documentation of subsequent renal failure and others have shown the safety of the protocol used in this regard.23

In summary, given modern CT's ability to triage patients for acute management of stroke syndromes, its ability to visualize large territorial perfusion and blood vessel pathology and its proven efficiency in evaluating acutely ill patients, its routine clinical application in acute stroke is likely to expand. Our initial clinical experience reported here led us to replace non-contrast CT with CT, CTA and pCT in evaluating patients with acute stroke for possible thrombolytic therapy. We felt that appropriate patient selection, and possibly superior outcomes could result with such an approach. Indeed, since this original review of our experience, our entire approach to acute stroke management has evolved to a comprehensive stroke team, with standardized pathways, including NIHSS metrics pretreatment and post-treatment on every patient, triage to earlier endovascular intervention (including mechanical techniques) and long-term outcome measures which should hopefully allow a more formal validation of the preliminary clinical data presented here.