- Email: [email protected]
Ischemic Stroke in Evolution: Predictive Value of Perfusion Computed Tomography
Ischemic Stroke in Evolution: Predictive Value of Perfusion Computed Tomography Amir Kheradmand, MD,* Marc Fisher, MD,† and David Paydarfar, MD†
Background: Various perfusion computed tomography (PCT) parameters have been used to identify tissue at risk of infarction in the setting of acute stroke. The purpose of this study was to examine predictive value of the PCT parameters commonly used in clinical practice to define ischemic penumbra. The patient selection criterion aimed to exclude the effect of thrombolysis from the imaging data. Methods: Consecutive acute stroke patients were screened and a total of 18 patients who initially underwent PCT and CT angiogram (CTA) on presentation but did not qualify to receive thrombolytic therapy were selected. The PCT images were postprocessed using a delay-sensitive deconvolution algorithm. All the patients had follow-up noncontrast CT or magnetic resonance imaging to delineate the extent of their infarction. The extent of lesions on PCT maps calculated from mean transit time (MTT), time to peak (TTP), cerebral blood flow, and cerebral blood volume were compared and correlated with the final infarct size. A collateral grading score was used to measure collateral blood supply on the CTA studies. Results: The average size of MTT lesions was larger than infarct lesions (P , .05). The correlation coefficient of TTP/infarct lesions (r 5 .95) was better than MTT/infarct lesions (r 5 .66) (P 5.004). Conclusions: A widely accepted threshold to define MTT lesions overestimates the ischemic penumbra. In this setting, TTP with appropriate threshold is a better predictor of infarct in acute stroke patients. The MTT/TTP mismatch correlates with the status of collateral blood supply to the tissue at risk of infarction. Key Words: Computed tomography—stroke—perfusion—mean transit time—time to peak. Ó 2013 by National Stroke Association
Introduction Perfusion computed tomography (PCT) has been increasingly advocated to guide optimal management strategies in acute stroke. PCT is widely accessible in emergency rooms, and its advantages over perfusion magnetic resonance imaging (MRI) include shorter imaging times and a lower cost.1 PCT allows rapid evaluation From the *Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, Maryland; and †Department of Neurology, University of Massachusetts Medical School, Worcester, Massachusetts. Received April 10, 2013; accepted July 7, 2013. Disclosures: None. Address correspondence to Amir Kheradmand, MD, The Johns Hopkins Hospital, Department of Neurology, Path 2-210, 600 N Wolfe St, Baltimore, MD 21287. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2013 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2013.07.014
of cerebral perfusion by generating maps of cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), and time to peak (TTP). These parameters are derived from PCT source data based on central volume principles.2-6 PCT parameters are used to distinguish irreversibly damaged infarct core tissue from potentially reversible ischemic tissue (ie, the ischemic penumbra).7 The presence and extent of salvageable ischemic penumbra is the mainstay of appropriate patient selection for thrombolytic therapy especially beyond the 3-hour time window.8-13 In recent years, various methods have been proposed to estimate the amount of at-risk ischemic tissue, using different CBF, MTT, TTP, and CBV thresholds.14-26 One proposed method uses relative MTT values above 145% to define at-risk ischemic tissue.20 MTT values are increased in at-risk ischemic tissue because of reduced blood flow, and CBV values are lowered within the infarct core because of loss of autoregulation.20
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2013: pp 1-8
1
A. KHERADMAND ET AL.
2
Current methods used for MTT calculation have recently been scrutinized regarding their accuracy.27-29 The sensitivity of processing algorithms to contrast delay is a potential pitfall that could result in overestimation of the ischemic penumbra by some software packages.28,29 In the present study, we examined the predictive value of PCT parameters for identifying the evolution of at-risk tissue using a commercially available delaysensitive software. For this purpose, we selected a group of acute stroke patients who underwent PCT imaging and did not qualify for thrombolytic therapy. This selection criterion aimed to represent the natural history of infarct evolution and exclude the effect of thrombolysis from the imaging data. For each patient, we compared the size of ischemic lesions derived by various PCT parameters to the size of the infarct on follow-up imaging studies (noncontrast CT or MRI performed in 1-7 days later).
Materials and Methods Patients PCT imaging data were obtained as part of standard clinical stroke care and were retrospectively reviewed with the approval of institutional review board at the University of Massachusetts. Patients with suspected acute stroke and no known renal insufficiency or allergy to contrast agent underwent the following imaging protocol on presentation: noncontrast CT, a single PCT slab of 4 cm width, and computed tomographic angiogram (CTA) of the cervical and intracranial vessels. This study was based on patients who presented from January to December 2009. We excluded all patients who received tissue plasminogen activator. A total of 18 acute stroke patients were found who initially underwent PCT and CTA on presentation but later did not qualify to receive thrombolytic therapy. All these patients had follow-up noncontrast CT or MRI to delineate the extent of their acute infarct.
Imaging and Data Processing The PCT images were obtained using a Brilliance 64channel scanner (Philips Medical Systems) and consisted of 4 axial 10-mm-thick sections obtained above the orbits toward the vertex to protect the lenses. Images were acquired and reconstructed at a temporal sampling rate of 1 image per second, resulting in a series of 45 images for each assessed section. According to our PCT imaging protocol, a 40-mL bolus of nonionic iodinated contrast agent (Isovue-370; Bracco Inc., Monroe Township, NJ) were used in all patients, administered into an antecubital vein by a power injector at an injection rate of 4-5 mL/sec. The acquisition parameters were 80 kVp and 100150 mAs. CT scanning was initiated 6-7 seconds after start of the contrast bolus injection. Follow-up imaging used in
this study was noncontrast CT images (5-mm-thick sections obtained by Brilliance 64-channel CT scanner; Philips Medical Systems) or diffusion-weighted MR sequence (5-mm-thick sections obtained by a General Electric HDxt 1.5-T MR imaging system). The follow-up imaging for each patient (CT or MRI) and the time it was obtained from onset of stroke symptoms are listed in Table 1. PCT data were analyzed using a deconvolution software developed by Philips Medical Systems (EBW version 3.0.1.3200). For each axial section, the software generates color-coded maps of different PCT parameters including MTT, CBF, CBV, and TTP. CBV is measured in units of milliliters of contrast material per 100 g of brain and is defined as the volume of blood for a given volume of brain. MTT and TTP are measured in seconds and defined, respectively, as the time contrast material takes to transit through a given volume of brain and achieve maximum enhancement. CBF is measured in units of milliliters of contrast per 100 g of brain tissue per minute and is defined as the volume of blood moving through a given volume of brain in a specific amount of time (CBF 5 CBV/MTT). The PCT maps along with followup CT (or MR) images were imported into Image J (Image processing software; version 1.43 m; National Institutes of Health, Bethesda, MD) for further analysis. All the imported images were in a 512 3 512 pixel matrix. The CT (or MR) images were coregistered to the corresponding PCT images by using ImageJ volume viewer plugin. The process included (1) volumetric conversion of the CT and MR images and (2) 3D adjustment of the crosssection plane orientation to match the PCT images. The selected slices were visually inspected for proper anatomical orientation by overlapping to corresponding PCT images. Infarct lesion in each section was defined as areas of hypodensity in CT and hyperintensity in diffusionweighted images (confirmed by decrease in the apparent diffusion coefficient). These areas were demarcated manually by ImageJ freehand selection tool (measured in pixels) and subsequently converted to square centimeter. For PCT parameters, lesion area was identified as the color pixels representing the maximum 20% range of values on MTT and TTP color-coded maps and the minimum 20% range of values on CBF color-coded maps. The MTT, TTP, and CBF lesions were then defined relative to the contralateral normal hemisphere. For this purpose, (1) areas in the corresponding regions of the contralateral hemisphere were subtracted from the lesion if they had values similar to the lesion area and (2) remaining values within the lesion were expressed as percentage of values from the corresponding regions of the contralateral (normal) hemisphere. These values were above 145% for MTT lesions, above 113% for TTP lesions, and below 20% for CBF lesions in all the patients. The CBV lesions were defined by their absolute size on the CBV maps, and the values obtained were below 2 mL/100 g in all the
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
3
Table 1. Characteristics of the 18 patients with acute stroke who underwent CT perfusion on admission but did not qualify to receive thrombolytic therapy Time of Arterial occlusion Carotid artery Patients Age Sex NIHSS PCT evaluation EF% per CTA stenosis per CTA 1 2 3 4 5
44 55 54 53 50
M M M M M
3 4 10 18 8
Within 3 h Within 3 h Within 3 h .3 h .3 h (wake up)
55 60 60 60 65
Right M1 Left M3 Right M1 Left M1 Left M2
6 7 8 9
73 46 85 76
M F M F
4 23 34 16
Within 3 h .3 h .3 h ?
75 55 60 35
Right M1 Left M1 Right M1 Left M1
10 11 12 13 14 15 16 17 18
59 43 53 76 17 93 76 38 41
F M M M F M F F M
4 17 14 9 3 23 12 2 6
.3 h .3 h (wake up) .3 h (wake up) ? (wake up) Within 3 h .3 h .3 h Within 3 h .3 h (wake up)
50 60 60 30 ? 45 55 ? 30
None Left M2 Left M1 Right M2 Right M2 Left M2 Right M1 Right M1 Left M2
None None RICA 90% LICA 90% LICA 100% RICA 90% None None None LICA 100% RICA 40% None LICA 50% None None None LICA 80% None RICA 100% RICA 40%
Reason tPA not given
Follow-up imaging
Low NIH score Low NIH score Post CABG Out of window Out of window
MRI, 24 h MRI, 24 h CT, 48 h CT, 48 h MRI, 5 d
Low NIH score Out of window Out of window Hypodensity on CT
MRI, 24 h CT, 48 h MRI, 24 h CT, 7 d
Out of window Out of window Out of window Out of window Symptoms resolved Out of window, SDH Out of window Symptoms resolved Out of window
MRI, 24 h CT, 7 d CT, 48 h CT, 72 h MRI, 24 h CT, 72 h CT, 5 d MRI, 24 h MRI, 72 h
Abbreviations: CABG, coronary artery bypass graft; CT, computed tomography; CTA, computed tomographic angiogram; EF, ejection fraction; LICA, left internal carotid artery; RICA, right internal carotid artery; SDH, subdural hematoma; M1-M3, segments of middle cerebral artery; NIHSS, National Institutes of Health Stroke Scale; PCT, perfusion computed tomography; tPA, tissue plasminogen activator.
patients. The lesion area on the PCT maps was also measured using the freehand selection tool and subsequently converted to square centimeter. A grading system with the scale of 1-3 was used to measure collateral blood supply on the CTA studies.30 A score of 0 indicated absent collateral supply to the occluded middle cerebral artery (MCA) territory. A score of 1 indicated collateral supply filling less than 50% but more than 0% of the occluded MCA territory. A score of 2 was given for collateral supply filling more than 50% but less than 100% of the occluded MCA territory. A score of 3 was given for 100% collateral supply of the occluded MCA territory. The collateral flow grading was scored by a stroke neurologist who was blinded to the results of all other imaging studies.
Statistical Analysis The average of MTT, TTP, CBF, and infarct lesion areas were calculated from PCT maps and corresponding CT or MR images (4 axial sections per parameter) for each patient. The average values were then compared using 1-way analysis of variance with post hoc Dunnett test. Correlations between the lesion area for each PCT parameter, the infarct area on corresponding CT (or MR) images, and collateral grading score were evaluated by linear regression analysis with calculation of Pearson linear correlation coefficient.
Results All patients in this study were disqualified from thrombolytic therapy. Twelve patients had arrival time outside the therapeutic window, 5 patients had low National Institutes of Health Stroke Scale (NIHSS) score or rapidly improving symptoms, and 1 patient had cardiac surgery. Summary demographics data, NIHSS score on presentation, reason for disqualification from thrombolytic therapy, and level of arterial occlusion reported on CTA for each patient are listed in Table 1. A total of 360 axial sections including PCT maps and follow-up CT or MR images were analyzed. The average area of MTT, TTP, CBF, and infarct lesions were statistically different (P 5 .002, analysis of variance). In post hoc analysis, the average area of MTT lesions was larger than infarct lesions (P ,.05, Dunnett test), whereas the average area of TTP and infarct lesions or the average area of CBF and infarct lesions were not significantly different (P . .05, Dunnett test). All voxels within the MTT lesions had relative values above 145%, TTP lesions were above 113%, and CBF lesions were below 20%. The infarct area on follow-up imaging was significantly correlated with TTP (r2 5 .9, P , .0001), CBF (r2 5 .82, P , .0001), CBV (r2 5 .67, P , .0001), and also MTT lesion areas (r2 5 .44, P 5 .0025). The linear regression line for each PCT parameters is shown in
A. KHERADMAND ET AL.
4
Figure 1. Correlation between the final infarct area on CT (or MR) images (abscissa) and lesions on (A) TTP, (B) MTT, (C) CBF, (D) CBV, and (E) all the PCT parameters at the time of admission (ordinate). Error bars (A-D) show standard error of mean for infarct lesions (horizontal bars) and PCT parameters (vertical bars) in each subject. Regression line is drawn for each PCT parameter. Overall, the slopes of regression lines are not significantly different (pooled slope:.75, P 5 .14). Infarct lesions are overestimated by MTT (y intercept: 10), whereas TTP and CBF lesions are better predictors of final infarct size (y intercept: .3 and .4, respectively). CBV lesions underestimate final infarct size (y intercept: 21.7, x intercept: 3). Abbreviations: CBF, cerebral blood flow; CBV, cerebral blood volume; CT, computed tomography; MR, magnetic resonance; MTT, mean transit time; TTP, time to peak; PCT, perfusion computed tomography.
Figure 1. The correlation coefficient of TTP/infarct lesions (r 5 .95) was higher than MTT/infarct lesions (r 5 .66, P 5 .004). Both these correlation coefficients were not significantly different from correlation coefficients of CBV/infarct lesions (r 5 .82, P 5 .17) and CBF/infarct lesions (r 5 .9, P 5 .18). Correlation analysis was also done for MTT and TTP lesions in patient subgroups with follow-up MRI or CT imaging. The infarct area was not significantly correlated with MTT lesions in both subgroups (MRI group: r2 5 .38, P 5 .1, and CT group: r2 5 .26, P 5 .1), and correlation coefficients were not different (MRI group: r 5 .61 and CT group: r 5 .51, P 5 .8). The infarct area was significantly correlated with TTP lesions in both subgroups (MRI group: r2 5 .97, P , .0001 and CT group: r2 5 .9, P , .0001) and correlation coefficients were not different (MRI group: r 5 .99 and CT group r 5 .95, P 5 .2). There was a significant correlation between the final infarct area and collateral grading score on presentation (r 5 2.7, r2 5 .49, P 5 .001) (Fig 2, A). The collateral grading score was also significantly correlated with the mismatch area between MTT and TTP when adjusted to the size of MTT lesion (r 5 .7, r2 5 .55, P 5 .0004) (Fig 2, B).
Discussion This study reveals important caveats for using PCT parameters to determine ischemic penumbra. We compared lesion area on each PCT map with the infarct area on follow-up imaging in a group of stroke patients who had unimpeded infarct evolution. Our results demonstrate that, among all PCT parameters, MTT significantly overestimates the size of at-risk ischemic tissue and, compared with TTP, has a significantly lower correlation with final infarct area. This finding calls into question (1) the accuracy of algorithms used to derive PCT parameters and (2) the current criteria used to define ischemic penumbra on PCT maps.
Limitations of MTT as a Marker of Penumbra As shown by Kudo et al,28 imaging maps are different among commercially available PCT software particularly because of difference in tracer-delay sensitivity. Software with delay-sensitive algorithms overestimates at-risk ischemic tissue and consequently final infarct volume, whereas at-risk tissue estimated with delay-insensitive software correlates better with final infarct volume.26,28,29
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
Figure 2. Correlation between the final infarct area on CT (or MR) images and collateral grading score measured from CTA studies on presentation (r 5 2.7) (A). The collateral grading score was also significantly correlated with the mismatch area between average MTT and average TTP lesion areas when adjusted to the size of average MTT lesion area (r 5 .7) (B). The numbers next to the data points correspond to the patient numbers in Table 1. Abbreviations: MTT, mean transit time; TTP, time to peak.
5
of at-risk tissue using MTT was validated in a subgroup of patients who did not recanalize after thrombolytic therapy.20 This subgroup may only represent cases with poor collateral flow in which compensatory mechanisms have completely failed to maintain adequate blood flow within the penumbra, and therefore, MTT lesion would fully evolve into infarction. Conceivably, the perfusion characteristics of this group cannot be extended to the whole array of stroke patients presenting with different levels of compensatory mechanisms. The results of present study are consistent with this idea. Among our patients, who all had infarct evolution without receiving thrombolysis, MTT lesions better matched the final infarct size in those with larger infarcts (lesions above 10 cm2, Fig 3, A). This is likely because of the very poor collateral circulation in this group as a result of a major arterial occlusion, leading to complete failure of compensatory mechanisms and full evolution of the calculated MTT lesions into infarction. The large MTT lesions in patients with small infarct size suggest that MTT is a worse predictor of infarct in these cases (lesions below 10 cm2, Fig 3, A). The Bland–Altman plot from our data (Fig 3, C) also shows better agreement between MTT and TTP as the size of the lesions increase. Kamalian et al26 have recently shown that among different PCT parameters, appropriately thresholded absolute and relative MTT maps can distinguish at-risk tissue from benign oligemia in patients with acute largevessel occlusion. The TTP maps, however, were not included in their analysis. The relative MTT thresholds for defining at-risk ischemic tissue were, respectively, 249% for the delay-sensitive and 150% for the delay-corrected algorithms. The patients in this study did not receive thrombolysis and showed no radiographic evidence of reperfusion.
Predictive Value of TTP The delay-sensitive software use deconvolution analysis to calculate MTT. This mathematical process removes the effects of a major arterial input (ie, middle cerebral or intracranial internal carotid artery) from the brain tissue time–density concentration curve. It, however, does not take into account circulatory delay or dispersion of the contrast material that could occur as a result of extracranial or intracranial pathologies. Consequently, the calculated MTT lesions may include hypoperfused but otherwise functional brain tissue, reflecting regions with delay in contrast arrival time but without the degree of ischemia that leads to infarction (ie, benign oligemia). The delay-sensitive method has been used to validate currently accepted criteria for penumbral definition, and despite the above-mentioned algorithmic flaws, relative MTT threshold of 145% was identified as the optimal predictor of the tissue at risk of infarction.20 How can this discrepancy be explained? The current definition
Our results suggest that TTP when compared with MTT is a more accurate parameter to identify the at-risk tissue (Figs 1 and 3). TTP, unlike other PCT parameters, is calculated directly without using processing algorithms and, therefore, is not susceptible to overestimation caused by delay-sensitive software. Although influenced by external factors such as contrast injection time, TTP can provide a realistic estimate of salvageable ischemic tissue as it represents the status of vascular compensation considering both cerebral (eg, extent of collateral circulation) and extracerebral factors (eg, cardiac function or patency of carotid arteries). Figure 4, for instance, shows PCT maps from a patient in this study who presented with a right M1 occlusion but had a low NIHSS of 3 on presentation and did not develop a large infarct. CTA showed a good collateral blood flow, and likely for this reason, the patient did not develop a large infarct in the MCA territory. The MTT map in this patient shows a large lesion,
6
A. KHERADMAND ET AL.
Figure 3. Average lesion area on (A) MTT and (B) TTP maps is shown along with the final infarct size (all from 4 similar cross-sections). Numbers on abscissa correspond to the patient numbers in Table 1. Error bar show standard error of mean for each parameter. The Bland–Altman plot (C) shows the relationship between the average and differences of MTT and TTP lesions (expressed as the percentage of the average value) for each patient; there is less difference between MTT and TTP lesions as the average size of lesions increase. The upper and lower lines show 95% limits of agreement. The numbers again corresponds to the patient numbers in Table 1. Abbreviations: MTT, mean transit time; TTP, time to peak.
whereas the lesion on the TTP map is much smaller and corresponds better to the final infarct size (Fig 4, C,D). This example illustrates that the accepted cutoff for MTT lesions overestimates at-risk tissue by including hypoperfused tissue that did not evolve into infarction. As shown in Figure 2,B, the collateral grading score was significantly correlated with the difference between MTT and TTP lesion areas (adjusted to the size of MTT lesion). This finding suggests that the mismatch between MTT and TTP lesions can provide information about the
amount of collateral blood supply in patients with acute stroke. TTP lesions measured in our study had values within the maximum 20% range of the color-coded map and all had relative values above 113%. Our results are in line with Bivard et al,31 which in a large group of patients found that ‘‘delay time’’ threshold was an accurate marker for at-risk ischemic tissue. Currently, there is no widely-accepted optimal cutoff for TTP values to define tissue at risk of infarction in PCT studies.
Figure 4. A case of acute onset fluctuating left hemiparesis (patient 1 from Table 1). Computed tomography angiogram shows right M1 occlusion with reconstitution of distal branches (A). There is a large lesion (blue color) on the MTT map (B). On the TTP map (C), the lesion area (blue color) is significantly smaller and better corresponds to the size of lesion on diffusion-weighted MRI (D). Abbreviations: MTT, mean transit time; TTP, time to peak.
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
Implication of Results Although a fairly small number of patients were included in this study, our results revealed a statistically strong distinction between the TTP and MTT lesions in predicting the final infarct size. Other PCT parameters including CBF and CBV also showed good correlations with final infarct size in this study. Like MTT, CBF is affected by tracer delay, and the calculated lesions overestimate final infarct size.28,29 CBF is, however, more specific than MTT for infarct prediction as MTT values can also be prolonged in transient ischemic attack patients.15 The high CBV/infarct correlation in our result could be related to the selection criteria in this study (ie, patients who did not qualify to receive thrombolysis): as the bulk of our patients had arrival time outside the therapeutic window with relatively established infarct core at the time of PCT imaging. The rest had small final infarcts and, therefore, no significant difference between the size of CBV and infarct lesions. Other potential limitation in this study is the lack of knowledge about spontaneous reperfusion that may occur in some patients. Our data, however, show a significant correlation between the collateral grading score from the initial CTA and the final infarct area (Fig 2, A) that suggests spontaneous reperfusion is not likely a confounding factor in this study. In addition, all patients with small infarct size and large MTT lesions (infarct lesions below 10 cm2, Fig 3, A) had low NIHSS on presentation (below 6, Table 1) showing that despite the large MTT lesion in each patient, the clinical findings on presentation and the final infarct were not consistent with a large stroke.
Summary In conclusion, our result shows that by using a widely accepted threshold value for MTT, at-risk ischemic tissue was overestimated. This discrepancy could be because of tracer-delay effect in PCT software using delay-sensitive algorithms for MTT measurement. Also, the threshold used to define MTT lesions—validated in a group of stroke patients who did not recanalize after thrombolysis—may not be applicable to all stroke patients presenting with different levels of collateral flow compensation. In this setting, TTP appears to be a better predictor of infarct evolution and MTT/TTP mismatch a good marker of collateral circulation in acute stroke patients. More studies are needed to verify these findings in a larger group of patients and identify proper thresholds for PCT parameters.
References 1. Jackson D, Earnshaw SR, Farkouh R, et al. Cost-effectiveness of CT perfusion for selecting patients for intravenous thrombolysis: a US hospital perspective. AJNR 2010;31:1669-1674.
7 2. Wintermark M, Maeder P, Thiran JP, et al. Quantitative assessment of regional cerebral blood flows by perfusion CT studies at low injection rates: a critical review of the underlying theoretical models. Eur Radiol 2001; 11:1220-1230. 3. Nabavi DG, Cenic A, Craen RA, et al. CT assessment of cerebral perfusion: experimental validation and initial clinical experience. Radiology 1999;213:141-149. 4. Cenic A, Nabavi DG, Craen RA, et al. Dynamic CT measurement of cerebral blood flow: a validation study. AJNR 1999;20:63-73. 5. Konstas AA, Goldmakher GV, Lee TY, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: theoretic basis. AJNR 2009;30:662-668. 6. Harrigan MR, Leonardo J, Gibbons KJ, et al. CT perfusion cerebral blood flow imaging in neurological critical care. Neurocrit Care 2005;2:352-366. 7. Konstas AA, Wintermark M, Lev MH. CT perfusion imaging in acute stroke. Neuroimaging Clin N Am 2011; 21:215-238. 8. Schellinger PD, Fiebach JB, Hacke W. Imaging-based decision making in thrombolytic therapy for ischemic stroke: present status. Stroke 2003;34:575-583. 9. Kumar G, Goyal MK, Sahota PK, et al. Penumbra, the basis of neuroimaging in acute stroke treatment: current evidence. J Neurol Sci 2010;288:13-24. 10. Ramos-Cabrer P, Campos F, Sobrino T, et al. Targeting the ischemic penumbra. Stroke 2011;42:S7-S11. 11. Obach V, Oleaga L, Urra X, et al. Multimodal CT-assisted thrombolysis in patients with acute stroke: a cohort study. Stroke 2011;42:1129-1131. 12. Michel P, Ntaios G, Reichhart M, et al. Perfusion-CT guided intravenous thrombolysis in patients with unknown-onset stroke: a randomized, double-blind, placebo-controlled, pilot feasibility trial. Neuroradiol 2012; 54:579-588. 13. Silva GS, Lima FO, Camargo EC, et al. Wake-up stroke: clinical and neuroimaging characteristics. Cerebrovasc Dis 2010;29:336-342. 14. Mayer TE, Hamann GF, Baranczyk J, et al. Dynamic CT perfusion imaging of acute stroke. AJNR 2000; 21:1441-1449. 15. Wintermark M, Fischbein NJ, Smith WS, et al. Accuracy of dynamic perfusion CT with deconvolution in detecting acute hemispheric stroke. AJNR 2005;26:104-112. 16. Wintermark M, Reichhart M, Thiran JP, et al. Prognostic accuracy of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol 2002;51:417-432. 17. Schramm P, Schellinger PD, Klotz E, et al. Comparison of perfusion computed tomography and computed tomography angiography source images with perfusionweighted imaging and diffusion-weighted imaging in patients with acute stroke of less than 6 hours’ duration. Stroke 2004;35:1652-1658. 18. Reichenbach JR, Rother J, Jonetz-Mentzel L, et al. Acute stroke evaluated by time-to-peak mapping during initial and early follow-up perfusion CT studies. AJNR 1999; 20:1842-1850. 19. Eastwood JD, Lev MH, Azhari T, et al. CT perfusion scanning with deconvolution analysis: pilot study in patients with acute middle cerebral artery stroke. Radiology 2002; 222:227-236. 20. Wintermark M, Flanders AE, Velthuis B, et al. Perfusion-CT assessment of infarct core and penumbra: receiver
A. KHERADMAND ET AL.
8
21.
22.
23.
24.
25.
26.
operating characteristic curve analysis in 130 patients suspected of acute hemispheric stroke. Stroke 2006;37:979-985. Campbell BC, Christensen S, Levi CR, et al. Cerebral blood flow is the optimal CT perfusion parameter for assessing infarct core. Stroke 2011;42:3435-3440. Bivard A, McElduff P, Spratt N, et al. Defining the extent of irreversible brain ischemia using perfusion computed tomography. Cerebrovasc Dis 2011;31:238-245. Hamberg LM, Hunter GJ, Maynard KI, et al. Functional CT perfusion imaging in predicting the extent of cerebral infarction from a 3-hour middle cerebral arterial occlusion in a primate stroke model. AJNR 2002;23:1013-1021. Kamalian S, Maas MB, Goldmacher GV, et al. CT cerebral blood flow maps optimally correlate with admission diffusion-weighted imaging in acute stroke but thresholds vary by postprocessing platform. Stroke 2011;42:1923-1928. Galvez M, York GE II, Eastwood JD. CT perfusion parameter values in regions of diffusion abnormalities. AJNR 2004;25:1205-1210. Kamalian S, Kamalian S, Konstas AA, et al. CT perfusion mean transit time maps optimally distinguish benign olige-
27.
28.
29.
30.
31.
mia from true "at-risk" ischemic penumbra, but thresholds vary by postprocessing technique. AJNR 2012;33:545-549. Zussman BM, Boghosian G, Gorniak RJ, et al. The relative effect of vendor variability in CT perfusion results: a method comparison study. Am J Roentgenol 2011; 197:468-473. Kudo K, Sasaki M, Yamada K, et al. Differences in CT perfusion maps generated by different commercial software: quantitative analysis by using identical source data of acute stroke patients. Radiology 2010;254:200-209. Konstas AA, Lev MH. CT perfusion imaging of acute stroke: the need for arrival time, delay insensitive, and standardized postprocessing algorithms? Radiology 2010;254:22-25. Tan IY, Demchuk AM, Hopyan J, et al. CT angiography clot burden score and collateral score: correlation with clinical and radiologic outcomes in acute middle cerebral artery infarct. AJNR 2009;30:525-531. Bivard A, Spratt N, Levi C, et al. Perfusion computer tomography: imaging and clinical validation in acute ischaemic stroke. Brain 2011;134:3408-3416.
Background: Various perfusion computed tomography (PCT) parameters have been used to identify tissue at risk of infarction in the setting of acute stroke. The purpose of this study was to examine predictive value of the PCT parameters commonly used in clinical practice to define ischemic penumbra. The patient selection criterion aimed to exclude the effect of thrombolysis from the imaging data. Methods: Consecutive acute stroke patients were screened and a total of 18 patients who initially underwent PCT and CT angiogram (CTA) on presentation but did not qualify to receive thrombolytic therapy were selected. The PCT images were postprocessed using a delay-sensitive deconvolution algorithm. All the patients had follow-up noncontrast CT or magnetic resonance imaging to delineate the extent of their infarction. The extent of lesions on PCT maps calculated from mean transit time (MTT), time to peak (TTP), cerebral blood flow, and cerebral blood volume were compared and correlated with the final infarct size. A collateral grading score was used to measure collateral blood supply on the CTA studies. Results: The average size of MTT lesions was larger than infarct lesions (P , .05). The correlation coefficient of TTP/infarct lesions (r 5 .95) was better than MTT/infarct lesions (r 5 .66) (P 5.004). Conclusions: A widely accepted threshold to define MTT lesions overestimates the ischemic penumbra. In this setting, TTP with appropriate threshold is a better predictor of infarct in acute stroke patients. The MTT/TTP mismatch correlates with the status of collateral blood supply to the tissue at risk of infarction. Key Words: Computed tomography—stroke—perfusion—mean transit time—time to peak. Ó 2013 by National Stroke Association
Introduction Perfusion computed tomography (PCT) has been increasingly advocated to guide optimal management strategies in acute stroke. PCT is widely accessible in emergency rooms, and its advantages over perfusion magnetic resonance imaging (MRI) include shorter imaging times and a lower cost.1 PCT allows rapid evaluation From the *Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, Maryland; and †Department of Neurology, University of Massachusetts Medical School, Worcester, Massachusetts. Received April 10, 2013; accepted July 7, 2013. Disclosures: None. Address correspondence to Amir Kheradmand, MD, The Johns Hopkins Hospital, Department of Neurology, Path 2-210, 600 N Wolfe St, Baltimore, MD 21287. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2013 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2013.07.014
of cerebral perfusion by generating maps of cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), and time to peak (TTP). These parameters are derived from PCT source data based on central volume principles.2-6 PCT parameters are used to distinguish irreversibly damaged infarct core tissue from potentially reversible ischemic tissue (ie, the ischemic penumbra).7 The presence and extent of salvageable ischemic penumbra is the mainstay of appropriate patient selection for thrombolytic therapy especially beyond the 3-hour time window.8-13 In recent years, various methods have been proposed to estimate the amount of at-risk ischemic tissue, using different CBF, MTT, TTP, and CBV thresholds.14-26 One proposed method uses relative MTT values above 145% to define at-risk ischemic tissue.20 MTT values are increased in at-risk ischemic tissue because of reduced blood flow, and CBV values are lowered within the infarct core because of loss of autoregulation.20
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2013: pp 1-8
1
A. KHERADMAND ET AL.
2
Current methods used for MTT calculation have recently been scrutinized regarding their accuracy.27-29 The sensitivity of processing algorithms to contrast delay is a potential pitfall that could result in overestimation of the ischemic penumbra by some software packages.28,29 In the present study, we examined the predictive value of PCT parameters for identifying the evolution of at-risk tissue using a commercially available delaysensitive software. For this purpose, we selected a group of acute stroke patients who underwent PCT imaging and did not qualify for thrombolytic therapy. This selection criterion aimed to represent the natural history of infarct evolution and exclude the effect of thrombolysis from the imaging data. For each patient, we compared the size of ischemic lesions derived by various PCT parameters to the size of the infarct on follow-up imaging studies (noncontrast CT or MRI performed in 1-7 days later).
Materials and Methods Patients PCT imaging data were obtained as part of standard clinical stroke care and were retrospectively reviewed with the approval of institutional review board at the University of Massachusetts. Patients with suspected acute stroke and no known renal insufficiency or allergy to contrast agent underwent the following imaging protocol on presentation: noncontrast CT, a single PCT slab of 4 cm width, and computed tomographic angiogram (CTA) of the cervical and intracranial vessels. This study was based on patients who presented from January to December 2009. We excluded all patients who received tissue plasminogen activator. A total of 18 acute stroke patients were found who initially underwent PCT and CTA on presentation but later did not qualify to receive thrombolytic therapy. All these patients had follow-up noncontrast CT or MRI to delineate the extent of their acute infarct.
Imaging and Data Processing The PCT images were obtained using a Brilliance 64channel scanner (Philips Medical Systems) and consisted of 4 axial 10-mm-thick sections obtained above the orbits toward the vertex to protect the lenses. Images were acquired and reconstructed at a temporal sampling rate of 1 image per second, resulting in a series of 45 images for each assessed section. According to our PCT imaging protocol, a 40-mL bolus of nonionic iodinated contrast agent (Isovue-370; Bracco Inc., Monroe Township, NJ) were used in all patients, administered into an antecubital vein by a power injector at an injection rate of 4-5 mL/sec. The acquisition parameters were 80 kVp and 100150 mAs. CT scanning was initiated 6-7 seconds after start of the contrast bolus injection. Follow-up imaging used in
this study was noncontrast CT images (5-mm-thick sections obtained by Brilliance 64-channel CT scanner; Philips Medical Systems) or diffusion-weighted MR sequence (5-mm-thick sections obtained by a General Electric HDxt 1.5-T MR imaging system). The follow-up imaging for each patient (CT or MRI) and the time it was obtained from onset of stroke symptoms are listed in Table 1. PCT data were analyzed using a deconvolution software developed by Philips Medical Systems (EBW version 3.0.1.3200). For each axial section, the software generates color-coded maps of different PCT parameters including MTT, CBF, CBV, and TTP. CBV is measured in units of milliliters of contrast material per 100 g of brain and is defined as the volume of blood for a given volume of brain. MTT and TTP are measured in seconds and defined, respectively, as the time contrast material takes to transit through a given volume of brain and achieve maximum enhancement. CBF is measured in units of milliliters of contrast per 100 g of brain tissue per minute and is defined as the volume of blood moving through a given volume of brain in a specific amount of time (CBF 5 CBV/MTT). The PCT maps along with followup CT (or MR) images were imported into Image J (Image processing software; version 1.43 m; National Institutes of Health, Bethesda, MD) for further analysis. All the imported images were in a 512 3 512 pixel matrix. The CT (or MR) images were coregistered to the corresponding PCT images by using ImageJ volume viewer plugin. The process included (1) volumetric conversion of the CT and MR images and (2) 3D adjustment of the crosssection plane orientation to match the PCT images. The selected slices were visually inspected for proper anatomical orientation by overlapping to corresponding PCT images. Infarct lesion in each section was defined as areas of hypodensity in CT and hyperintensity in diffusionweighted images (confirmed by decrease in the apparent diffusion coefficient). These areas were demarcated manually by ImageJ freehand selection tool (measured in pixels) and subsequently converted to square centimeter. For PCT parameters, lesion area was identified as the color pixels representing the maximum 20% range of values on MTT and TTP color-coded maps and the minimum 20% range of values on CBF color-coded maps. The MTT, TTP, and CBF lesions were then defined relative to the contralateral normal hemisphere. For this purpose, (1) areas in the corresponding regions of the contralateral hemisphere were subtracted from the lesion if they had values similar to the lesion area and (2) remaining values within the lesion were expressed as percentage of values from the corresponding regions of the contralateral (normal) hemisphere. These values were above 145% for MTT lesions, above 113% for TTP lesions, and below 20% for CBF lesions in all the patients. The CBV lesions were defined by their absolute size on the CBV maps, and the values obtained were below 2 mL/100 g in all the
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
3
Table 1. Characteristics of the 18 patients with acute stroke who underwent CT perfusion on admission but did not qualify to receive thrombolytic therapy Time of Arterial occlusion Carotid artery Patients Age Sex NIHSS PCT evaluation EF% per CTA stenosis per CTA 1 2 3 4 5
44 55 54 53 50
M M M M M
3 4 10 18 8
Within 3 h Within 3 h Within 3 h .3 h .3 h (wake up)
55 60 60 60 65
Right M1 Left M3 Right M1 Left M1 Left M2
6 7 8 9
73 46 85 76
M F M F
4 23 34 16
Within 3 h .3 h .3 h ?
75 55 60 35
Right M1 Left M1 Right M1 Left M1
10 11 12 13 14 15 16 17 18
59 43 53 76 17 93 76 38 41
F M M M F M F F M
4 17 14 9 3 23 12 2 6
.3 h .3 h (wake up) .3 h (wake up) ? (wake up) Within 3 h .3 h .3 h Within 3 h .3 h (wake up)
50 60 60 30 ? 45 55 ? 30
None Left M2 Left M1 Right M2 Right M2 Left M2 Right M1 Right M1 Left M2
None None RICA 90% LICA 90% LICA 100% RICA 90% None None None LICA 100% RICA 40% None LICA 50% None None None LICA 80% None RICA 100% RICA 40%
Reason tPA not given
Follow-up imaging
Low NIH score Low NIH score Post CABG Out of window Out of window
MRI, 24 h MRI, 24 h CT, 48 h CT, 48 h MRI, 5 d
Low NIH score Out of window Out of window Hypodensity on CT
MRI, 24 h CT, 48 h MRI, 24 h CT, 7 d
Out of window Out of window Out of window Out of window Symptoms resolved Out of window, SDH Out of window Symptoms resolved Out of window
MRI, 24 h CT, 7 d CT, 48 h CT, 72 h MRI, 24 h CT, 72 h CT, 5 d MRI, 24 h MRI, 72 h
Abbreviations: CABG, coronary artery bypass graft; CT, computed tomography; CTA, computed tomographic angiogram; EF, ejection fraction; LICA, left internal carotid artery; RICA, right internal carotid artery; SDH, subdural hematoma; M1-M3, segments of middle cerebral artery; NIHSS, National Institutes of Health Stroke Scale; PCT, perfusion computed tomography; tPA, tissue plasminogen activator.
patients. The lesion area on the PCT maps was also measured using the freehand selection tool and subsequently converted to square centimeter. A grading system with the scale of 1-3 was used to measure collateral blood supply on the CTA studies.30 A score of 0 indicated absent collateral supply to the occluded middle cerebral artery (MCA) territory. A score of 1 indicated collateral supply filling less than 50% but more than 0% of the occluded MCA territory. A score of 2 was given for collateral supply filling more than 50% but less than 100% of the occluded MCA territory. A score of 3 was given for 100% collateral supply of the occluded MCA territory. The collateral flow grading was scored by a stroke neurologist who was blinded to the results of all other imaging studies.
Statistical Analysis The average of MTT, TTP, CBF, and infarct lesion areas were calculated from PCT maps and corresponding CT or MR images (4 axial sections per parameter) for each patient. The average values were then compared using 1-way analysis of variance with post hoc Dunnett test. Correlations between the lesion area for each PCT parameter, the infarct area on corresponding CT (or MR) images, and collateral grading score were evaluated by linear regression analysis with calculation of Pearson linear correlation coefficient.
Results All patients in this study were disqualified from thrombolytic therapy. Twelve patients had arrival time outside the therapeutic window, 5 patients had low National Institutes of Health Stroke Scale (NIHSS) score or rapidly improving symptoms, and 1 patient had cardiac surgery. Summary demographics data, NIHSS score on presentation, reason for disqualification from thrombolytic therapy, and level of arterial occlusion reported on CTA for each patient are listed in Table 1. A total of 360 axial sections including PCT maps and follow-up CT or MR images were analyzed. The average area of MTT, TTP, CBF, and infarct lesions were statistically different (P 5 .002, analysis of variance). In post hoc analysis, the average area of MTT lesions was larger than infarct lesions (P ,.05, Dunnett test), whereas the average area of TTP and infarct lesions or the average area of CBF and infarct lesions were not significantly different (P . .05, Dunnett test). All voxels within the MTT lesions had relative values above 145%, TTP lesions were above 113%, and CBF lesions were below 20%. The infarct area on follow-up imaging was significantly correlated with TTP (r2 5 .9, P , .0001), CBF (r2 5 .82, P , .0001), CBV (r2 5 .67, P , .0001), and also MTT lesion areas (r2 5 .44, P 5 .0025). The linear regression line for each PCT parameters is shown in
A. KHERADMAND ET AL.
4
Figure 1. Correlation between the final infarct area on CT (or MR) images (abscissa) and lesions on (A) TTP, (B) MTT, (C) CBF, (D) CBV, and (E) all the PCT parameters at the time of admission (ordinate). Error bars (A-D) show standard error of mean for infarct lesions (horizontal bars) and PCT parameters (vertical bars) in each subject. Regression line is drawn for each PCT parameter. Overall, the slopes of regression lines are not significantly different (pooled slope:.75, P 5 .14). Infarct lesions are overestimated by MTT (y intercept: 10), whereas TTP and CBF lesions are better predictors of final infarct size (y intercept: .3 and .4, respectively). CBV lesions underestimate final infarct size (y intercept: 21.7, x intercept: 3). Abbreviations: CBF, cerebral blood flow; CBV, cerebral blood volume; CT, computed tomography; MR, magnetic resonance; MTT, mean transit time; TTP, time to peak; PCT, perfusion computed tomography.
Figure 1. The correlation coefficient of TTP/infarct lesions (r 5 .95) was higher than MTT/infarct lesions (r 5 .66, P 5 .004). Both these correlation coefficients were not significantly different from correlation coefficients of CBV/infarct lesions (r 5 .82, P 5 .17) and CBF/infarct lesions (r 5 .9, P 5 .18). Correlation analysis was also done for MTT and TTP lesions in patient subgroups with follow-up MRI or CT imaging. The infarct area was not significantly correlated with MTT lesions in both subgroups (MRI group: r2 5 .38, P 5 .1, and CT group: r2 5 .26, P 5 .1), and correlation coefficients were not different (MRI group: r 5 .61 and CT group: r 5 .51, P 5 .8). The infarct area was significantly correlated with TTP lesions in both subgroups (MRI group: r2 5 .97, P , .0001 and CT group: r2 5 .9, P , .0001) and correlation coefficients were not different (MRI group: r 5 .99 and CT group r 5 .95, P 5 .2). There was a significant correlation between the final infarct area and collateral grading score on presentation (r 5 2.7, r2 5 .49, P 5 .001) (Fig 2, A). The collateral grading score was also significantly correlated with the mismatch area between MTT and TTP when adjusted to the size of MTT lesion (r 5 .7, r2 5 .55, P 5 .0004) (Fig 2, B).
Discussion This study reveals important caveats for using PCT parameters to determine ischemic penumbra. We compared lesion area on each PCT map with the infarct area on follow-up imaging in a group of stroke patients who had unimpeded infarct evolution. Our results demonstrate that, among all PCT parameters, MTT significantly overestimates the size of at-risk ischemic tissue and, compared with TTP, has a significantly lower correlation with final infarct area. This finding calls into question (1) the accuracy of algorithms used to derive PCT parameters and (2) the current criteria used to define ischemic penumbra on PCT maps.
Limitations of MTT as a Marker of Penumbra As shown by Kudo et al,28 imaging maps are different among commercially available PCT software particularly because of difference in tracer-delay sensitivity. Software with delay-sensitive algorithms overestimates at-risk ischemic tissue and consequently final infarct volume, whereas at-risk tissue estimated with delay-insensitive software correlates better with final infarct volume.26,28,29
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
Figure 2. Correlation between the final infarct area on CT (or MR) images and collateral grading score measured from CTA studies on presentation (r 5 2.7) (A). The collateral grading score was also significantly correlated with the mismatch area between average MTT and average TTP lesion areas when adjusted to the size of average MTT lesion area (r 5 .7) (B). The numbers next to the data points correspond to the patient numbers in Table 1. Abbreviations: MTT, mean transit time; TTP, time to peak.
5
of at-risk tissue using MTT was validated in a subgroup of patients who did not recanalize after thrombolytic therapy.20 This subgroup may only represent cases with poor collateral flow in which compensatory mechanisms have completely failed to maintain adequate blood flow within the penumbra, and therefore, MTT lesion would fully evolve into infarction. Conceivably, the perfusion characteristics of this group cannot be extended to the whole array of stroke patients presenting with different levels of compensatory mechanisms. The results of present study are consistent with this idea. Among our patients, who all had infarct evolution without receiving thrombolysis, MTT lesions better matched the final infarct size in those with larger infarcts (lesions above 10 cm2, Fig 3, A). This is likely because of the very poor collateral circulation in this group as a result of a major arterial occlusion, leading to complete failure of compensatory mechanisms and full evolution of the calculated MTT lesions into infarction. The large MTT lesions in patients with small infarct size suggest that MTT is a worse predictor of infarct in these cases (lesions below 10 cm2, Fig 3, A). The Bland–Altman plot from our data (Fig 3, C) also shows better agreement between MTT and TTP as the size of the lesions increase. Kamalian et al26 have recently shown that among different PCT parameters, appropriately thresholded absolute and relative MTT maps can distinguish at-risk tissue from benign oligemia in patients with acute largevessel occlusion. The TTP maps, however, were not included in their analysis. The relative MTT thresholds for defining at-risk ischemic tissue were, respectively, 249% for the delay-sensitive and 150% for the delay-corrected algorithms. The patients in this study did not receive thrombolysis and showed no radiographic evidence of reperfusion.
Predictive Value of TTP The delay-sensitive software use deconvolution analysis to calculate MTT. This mathematical process removes the effects of a major arterial input (ie, middle cerebral or intracranial internal carotid artery) from the brain tissue time–density concentration curve. It, however, does not take into account circulatory delay or dispersion of the contrast material that could occur as a result of extracranial or intracranial pathologies. Consequently, the calculated MTT lesions may include hypoperfused but otherwise functional brain tissue, reflecting regions with delay in contrast arrival time but without the degree of ischemia that leads to infarction (ie, benign oligemia). The delay-sensitive method has been used to validate currently accepted criteria for penumbral definition, and despite the above-mentioned algorithmic flaws, relative MTT threshold of 145% was identified as the optimal predictor of the tissue at risk of infarction.20 How can this discrepancy be explained? The current definition
Our results suggest that TTP when compared with MTT is a more accurate parameter to identify the at-risk tissue (Figs 1 and 3). TTP, unlike other PCT parameters, is calculated directly without using processing algorithms and, therefore, is not susceptible to overestimation caused by delay-sensitive software. Although influenced by external factors such as contrast injection time, TTP can provide a realistic estimate of salvageable ischemic tissue as it represents the status of vascular compensation considering both cerebral (eg, extent of collateral circulation) and extracerebral factors (eg, cardiac function or patency of carotid arteries). Figure 4, for instance, shows PCT maps from a patient in this study who presented with a right M1 occlusion but had a low NIHSS of 3 on presentation and did not develop a large infarct. CTA showed a good collateral blood flow, and likely for this reason, the patient did not develop a large infarct in the MCA territory. The MTT map in this patient shows a large lesion,
6
A. KHERADMAND ET AL.
Figure 3. Average lesion area on (A) MTT and (B) TTP maps is shown along with the final infarct size (all from 4 similar cross-sections). Numbers on abscissa correspond to the patient numbers in Table 1. Error bar show standard error of mean for each parameter. The Bland–Altman plot (C) shows the relationship between the average and differences of MTT and TTP lesions (expressed as the percentage of the average value) for each patient; there is less difference between MTT and TTP lesions as the average size of lesions increase. The upper and lower lines show 95% limits of agreement. The numbers again corresponds to the patient numbers in Table 1. Abbreviations: MTT, mean transit time; TTP, time to peak.
whereas the lesion on the TTP map is much smaller and corresponds better to the final infarct size (Fig 4, C,D). This example illustrates that the accepted cutoff for MTT lesions overestimates at-risk tissue by including hypoperfused tissue that did not evolve into infarction. As shown in Figure 2,B, the collateral grading score was significantly correlated with the difference between MTT and TTP lesion areas (adjusted to the size of MTT lesion). This finding suggests that the mismatch between MTT and TTP lesions can provide information about the
amount of collateral blood supply in patients with acute stroke. TTP lesions measured in our study had values within the maximum 20% range of the color-coded map and all had relative values above 113%. Our results are in line with Bivard et al,31 which in a large group of patients found that ‘‘delay time’’ threshold was an accurate marker for at-risk ischemic tissue. Currently, there is no widely-accepted optimal cutoff for TTP values to define tissue at risk of infarction in PCT studies.
Figure 4. A case of acute onset fluctuating left hemiparesis (patient 1 from Table 1). Computed tomography angiogram shows right M1 occlusion with reconstitution of distal branches (A). There is a large lesion (blue color) on the MTT map (B). On the TTP map (C), the lesion area (blue color) is significantly smaller and better corresponds to the size of lesion on diffusion-weighted MRI (D). Abbreviations: MTT, mean transit time; TTP, time to peak.
ACUTE STROKE AND PREDICTIVE VALUE OF PCT
Implication of Results Although a fairly small number of patients were included in this study, our results revealed a statistically strong distinction between the TTP and MTT lesions in predicting the final infarct size. Other PCT parameters including CBF and CBV also showed good correlations with final infarct size in this study. Like MTT, CBF is affected by tracer delay, and the calculated lesions overestimate final infarct size.28,29 CBF is, however, more specific than MTT for infarct prediction as MTT values can also be prolonged in transient ischemic attack patients.15 The high CBV/infarct correlation in our result could be related to the selection criteria in this study (ie, patients who did not qualify to receive thrombolysis): as the bulk of our patients had arrival time outside the therapeutic window with relatively established infarct core at the time of PCT imaging. The rest had small final infarcts and, therefore, no significant difference between the size of CBV and infarct lesions. Other potential limitation in this study is the lack of knowledge about spontaneous reperfusion that may occur in some patients. Our data, however, show a significant correlation between the collateral grading score from the initial CTA and the final infarct area (Fig 2, A) that suggests spontaneous reperfusion is not likely a confounding factor in this study. In addition, all patients with small infarct size and large MTT lesions (infarct lesions below 10 cm2, Fig 3, A) had low NIHSS on presentation (below 6, Table 1) showing that despite the large MTT lesion in each patient, the clinical findings on presentation and the final infarct were not consistent with a large stroke.
Summary In conclusion, our result shows that by using a widely accepted threshold value for MTT, at-risk ischemic tissue was overestimated. This discrepancy could be because of tracer-delay effect in PCT software using delay-sensitive algorithms for MTT measurement. Also, the threshold used to define MTT lesions—validated in a group of stroke patients who did not recanalize after thrombolysis—may not be applicable to all stroke patients presenting with different levels of collateral flow compensation. In this setting, TTP appears to be a better predictor of infarct evolution and MTT/TTP mismatch a good marker of collateral circulation in acute stroke patients. More studies are needed to verify these findings in a larger group of patients and identify proper thresholds for PCT parameters.
References 1. Jackson D, Earnshaw SR, Farkouh R, et al. Cost-effectiveness of CT perfusion for selecting patients for intravenous thrombolysis: a US hospital perspective. AJNR 2010;31:1669-1674.
7 2. Wintermark M, Maeder P, Thiran JP, et al. Quantitative assessment of regional cerebral blood flows by perfusion CT studies at low injection rates: a critical review of the underlying theoretical models. Eur Radiol 2001; 11:1220-1230. 3. Nabavi DG, Cenic A, Craen RA, et al. CT assessment of cerebral perfusion: experimental validation and initial clinical experience. Radiology 1999;213:141-149. 4. Cenic A, Nabavi DG, Craen RA, et al. Dynamic CT measurement of cerebral blood flow: a validation study. AJNR 1999;20:63-73. 5. Konstas AA, Goldmakher GV, Lee TY, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: theoretic basis. AJNR 2009;30:662-668. 6. Harrigan MR, Leonardo J, Gibbons KJ, et al. CT perfusion cerebral blood flow imaging in neurological critical care. Neurocrit Care 2005;2:352-366. 7. Konstas AA, Wintermark M, Lev MH. CT perfusion imaging in acute stroke. Neuroimaging Clin N Am 2011; 21:215-238. 8. Schellinger PD, Fiebach JB, Hacke W. Imaging-based decision making in thrombolytic therapy for ischemic stroke: present status. Stroke 2003;34:575-583. 9. Kumar G, Goyal MK, Sahota PK, et al. Penumbra, the basis of neuroimaging in acute stroke treatment: current evidence. J Neurol Sci 2010;288:13-24. 10. Ramos-Cabrer P, Campos F, Sobrino T, et al. Targeting the ischemic penumbra. Stroke 2011;42:S7-S11. 11. Obach V, Oleaga L, Urra X, et al. Multimodal CT-assisted thrombolysis in patients with acute stroke: a cohort study. Stroke 2011;42:1129-1131. 12. Michel P, Ntaios G, Reichhart M, et al. Perfusion-CT guided intravenous thrombolysis in patients with unknown-onset stroke: a randomized, double-blind, placebo-controlled, pilot feasibility trial. Neuroradiol 2012; 54:579-588. 13. Silva GS, Lima FO, Camargo EC, et al. Wake-up stroke: clinical and neuroimaging characteristics. Cerebrovasc Dis 2010;29:336-342. 14. Mayer TE, Hamann GF, Baranczyk J, et al. Dynamic CT perfusion imaging of acute stroke. AJNR 2000; 21:1441-1449. 15. Wintermark M, Fischbein NJ, Smith WS, et al. Accuracy of dynamic perfusion CT with deconvolution in detecting acute hemispheric stroke. AJNR 2005;26:104-112. 16. Wintermark M, Reichhart M, Thiran JP, et al. Prognostic accuracy of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol 2002;51:417-432. 17. Schramm P, Schellinger PD, Klotz E, et al. Comparison of perfusion computed tomography and computed tomography angiography source images with perfusionweighted imaging and diffusion-weighted imaging in patients with acute stroke of less than 6 hours’ duration. Stroke 2004;35:1652-1658. 18. Reichenbach JR, Rother J, Jonetz-Mentzel L, et al. Acute stroke evaluated by time-to-peak mapping during initial and early follow-up perfusion CT studies. AJNR 1999; 20:1842-1850. 19. Eastwood JD, Lev MH, Azhari T, et al. CT perfusion scanning with deconvolution analysis: pilot study in patients with acute middle cerebral artery stroke. Radiology 2002; 222:227-236. 20. Wintermark M, Flanders AE, Velthuis B, et al. Perfusion-CT assessment of infarct core and penumbra: receiver
A. KHERADMAND ET AL.
8
21.
22.
23.
24.
25.
26.
operating characteristic curve analysis in 130 patients suspected of acute hemispheric stroke. Stroke 2006;37:979-985. Campbell BC, Christensen S, Levi CR, et al. Cerebral blood flow is the optimal CT perfusion parameter for assessing infarct core. Stroke 2011;42:3435-3440. Bivard A, McElduff P, Spratt N, et al. Defining the extent of irreversible brain ischemia using perfusion computed tomography. Cerebrovasc Dis 2011;31:238-245. Hamberg LM, Hunter GJ, Maynard KI, et al. Functional CT perfusion imaging in predicting the extent of cerebral infarction from a 3-hour middle cerebral arterial occlusion in a primate stroke model. AJNR 2002;23:1013-1021. Kamalian S, Maas MB, Goldmacher GV, et al. CT cerebral blood flow maps optimally correlate with admission diffusion-weighted imaging in acute stroke but thresholds vary by postprocessing platform. Stroke 2011;42:1923-1928. Galvez M, York GE II, Eastwood JD. CT perfusion parameter values in regions of diffusion abnormalities. AJNR 2004;25:1205-1210. Kamalian S, Kamalian S, Konstas AA, et al. CT perfusion mean transit time maps optimally distinguish benign olige-
27.
28.
29.
30.
31.
mia from true "at-risk" ischemic penumbra, but thresholds vary by postprocessing technique. AJNR 2012;33:545-549. Zussman BM, Boghosian G, Gorniak RJ, et al. The relative effect of vendor variability in CT perfusion results: a method comparison study. Am J Roentgenol 2011; 197:468-473. Kudo K, Sasaki M, Yamada K, et al. Differences in CT perfusion maps generated by different commercial software: quantitative analysis by using identical source data of acute stroke patients. Radiology 2010;254:200-209. Konstas AA, Lev MH. CT perfusion imaging of acute stroke: the need for arrival time, delay insensitive, and standardized postprocessing algorithms? Radiology 2010;254:22-25. Tan IY, Demchuk AM, Hopyan J, et al. CT angiography clot burden score and collateral score: correlation with clinical and radiologic outcomes in acute middle cerebral artery infarct. AJNR 2009;30:525-531. Bivard A, Spratt N, Levi C, et al. Perfusion computer tomography: imaging and clinical validation in acute ischaemic stroke. Brain 2011;134:3408-3416.