Automated CT Perfusion Ischemic Core Volume and Noncontrast CT ASPECTS (Alberta Stroke Program Early CT Score)
Correlation and Clinical Outcome Prediction in Large Vessel Stroke
Background and Purpose—The semiquantitative noncontrast CT Alberta Stroke Program Early CT Score (ASPECTS) and RAPID automated computed tomography (CT) perfusion (CTP) ischemic core volumetric measurements have been used to quantify infarct extent. We aim to determine the correlation between ASPECTS and CTP ischemic core, evaluate the variability of core volumes within ASPECTS strata, and assess the strength of their association with clinical outcomes.
Methods—Review of a prospective, single-center database of consecutive thrombectomies of middle cerebral or intracranial internal carotid artery occlusions with pretreatment CTP between September 2010 and September 2015. CTP was processed with RAPID software to identify ischemic core (relative cerebral blood flow<30% of normal tissue).
Results—Three hundred and thirty-two patients fulfilled inclusion criteria. Median age was 66 years (55–75), median ASPECTS was 8 (7–9), whereas median CTP ischemic core was 11 cc (2–27). Median time from last normal to groin puncture was 5.8 hours (3.9–8.8), and 90-day modified Rankin scale score 0 to 2 was observed in 54%. The correlation between CTP ischemic core and ASPECTS was fair (R=−0.36; P<0.01). Twenty-six patients (8%) had ASPECTS <6 and CTP core ≤50 cc (37% had modified Rankin scale score 0–2, whereas 29% were deceased at 90 days). Conversely, 27 patients (8%) had CTP core >50 cc and ASPECTS ≥6 (29% had modified Rankin scale 0–2, whereas 21% were deceased at 90 days). Moderate correlations between ASPECTS and final infarct volume (R=−0.42; P<0.01) and between CTP ischemic core and final infarct volume (R=0.50; P<0.01) were observed; coefficients were not significantly influenced by the time from stroke onset to presentation. Multivariable regression indicated ASPECTS ≥6 (odds ratio 4.10; 95% confidence interval, 1.47–11.46; P=0.01) and CTP core ≤50 cc (odds ratio 3.86; 95% confidence interval, 1.22–12.15; P=0.02) independently and comparably predictive of good outcome.
Conclusions—There is wide variability of CTP-derived core volumes within ASPECTS strata. Patient selection may be affected by the imaging selection method.
The quantification of infarct extent is a critical aspect in patient selection for reperfusion therapy in acute ischemic stroke. Among the available and validated modalities, the semiquantitative Alberta Stroke Program Early CT Score (ASPECTS) and the automated quantitative RAPID computed tomography (CT) perfusion (CTP) platform have been used to select patients for endovascular treatment of anterior circulation large vessel occlusion strokes (LVOS).1–5 However, the best imaging strategy to screen patients for thrombectomy and the level of agreement between these methods is uncertain. We aim to (1) establish the correlation between noncontrast CT (NCCT) ASPECTS and automated CTP ischemic core volumes; (2) evaluate the variability of ischemic core volumes within different ASPECTS strata; (3) assess the strength of association between 2 commonly used treatment exclusion thresholds (ASPECTS <6 versus CTP core >50 cc) and good clinical outcome.
This was a retrospective review of a prospectively collected LVOS intervention database from a tertiary care academic institution. Consecutive acute ischemic stroke cases undergoing endovascular thrombolysis between September 2010 and September 2015 for anterior circulation intracranial LVOS (middle cerebral artery M1, M2, and M3 segments or the internal carotid artery terminus), undergoing both baseline pretreatment NCCT, and technically adequate pretreatment CTP were included. Imaging acquisition parameters have been previously described.6 This study was approved by the local institutional review board.
Baseline NCCT was used to grade the Alberta Stroke Program Early CT Score (ASPECTS) standardized 10-point scale by a stroke neurologist.7 CTP images were processed with a fully automated, commercially available software platform (RAPID version 4.5.0. iSchemaView Inc, Menlo Park, CA) to define tissue state, as previously described.8 Irreversibly injured tissue (ischemic core) was defined by reduction in the relative cerebral blood flow to <30% of that in normal tissue.4 Patients with nondiagnostic NCCT or CTP maps were excluded. Follow-up in all patients included magnetic resonance imaging (MRI) or NCCT to capture final infarct volumes (FIV) before hospital discharge. FIV were measured preferentially by MRI. All follow-up imaging was performed within the first 5 days of the treatment. Imaging processing has been previously described.9 Good outcome was defined as 90-day modified Rankin scale (mRS) score of 0 to 2 determined preferentially by an in-person follow-up appointment or via structured phone interview.10 Reperfusion was graded by the modified Treatment in Cerebral Ischemia scale.11 FIV were defined preferentially and predominantly by MRI (NCCT performed if contraindications precluded MRI) within 5 days of treatment. Diffusion weighted imaging was used when MRI obtained <72 hours of the stroke and T2-FLAIR (T2-fluid attenuation inversion recovery) thereafter. FIV were measured using a manual segmentation tool after export of raw DICOM (Digital Imaging and Communications in Medicine) data to the Fiji release of the ImageJ software platform (http://imagej.nih.gov/ij/).9
Continuous variables are reported as mean±SD or median (IQR) and categorical variables as proportions. Between groups, comparisons for continuous or ordinal variables were made with Student t test or Mann–Whitney U test. Categorical variables were compared by χ2 or Fisher exact test. Scatter plots and box plots were assessed to evaluate the distribution of CTP core volumes across ASPECTS grades. Correlation coefficient was calculated with Spearman ρ. Significance level was 0.05. Multivariable logistic regression analyses for predictors of good clinical outcome (mRS 0–2 at 90 days) were performed for variables significant at P<0.1 level in univariate analyses (Enter method). ASPECTS ≥6 and CTP ischemic core ≤50 cc were used as standard thresholds1 and were included independently in the model because of collinearity. Statistical analyses were performed using SPSS Statistics version 21 (IBM, Armonk, NY).
Out of 834 patients treated within the study period, 332 fulfilled inclusion criteria (Figure I in the online-only Data Supplement). Table indicates the baseline characteristics, procedural variables, and outcomes. Median age was 66 years (55–75) and 50% were men. Median ASPECTS was 8 (7–9), whereas median CTP ischemic core was 11 cc (2–27). The median time from last known normal to groin puncture was 5.8 hours (3.9–8.8). Stent-retrievers were used in most patients (77%) leading to high reperfusion rates (mTICI2b-3=89%). The rate of good clinical outcome was 54%, and median FIV was 26.9 cc (10.7–65.9). Follow-up MRI was performed in 79.6% of patients.
The correlation between NCCT ASPECTS with CTP ischemic core was fair (R=−0.36; P<0.01). Figure 1 shows a well-defined trend of decreasing median baseline CTP ischemic cores as ASPECTS grade increases and a notable degree of CTP ischemic core variability per ASPECTS strata. Figures 2 and 3 illustrate cases of discrepant results between NCCT ASPECTS and RAPID CTP.
The correlations between ASPECTS with FIV (R=−0.42; P<0.01) and CTP ischemic core with FIV (R=0.50; P<0.01) were moderate. When the data set was dichotomized into early presenting (≤4.5 hours; 37%) versus late presenting (>4.5 hours; 63%) patients, no clear impact of time from stroke onset to presentation on the degree of correlation was noted between ASPECTS and FIV (R=−0.52; P<0.01 and R=−0.45; P<0.01 for early and late presenting, respectively) and CTP ischemic core and FIV (R=0.48; P<0.01 and R=0.43; P<0.01 for early and late presenting, respectively). An evaluation of patients with full reperfusion (mTICI3) was performed to eliminate the impact of infarct growth in incompletely reperfused patients, and the correlations remained moderate (ASPECTS with FIV: R=−0.61; P<0.01 and CTP ischemic core with FIV: R=0.43; P<0.01).
When using current standard imaging thresholds (NCCT ASPECTS <6 or CTP core volume >50 cc), several patients in our registry would not have qualified for thrombectomy but still achieved good outcome. Twenty-six patients (8%), who would not have been intervened based on NCCT ASPECTS <6, were taken for thrombectomy based on CTP core imaging ≤50 cc, and 37% of these patients achieved a good outcome, whereas 29% died. Conversely, a total of 27 patients (8%), who would not have been treated because of CTP core >50 cc, were intervened based on an ASPECTS ≥6, and 29% achieved good outcome, whereas 21% died. Eight patients (2%) had ASPECTS <6 and CTP core >50 cc, having 1 (12%) achieved good outcome, whereas 2 (25%) died at 90 days.
In multivariable logistic regression analyses, we found that ASPECTS ≥6 (odds ratio [OR], 4.10; 95% confidence interval [CI], 1.47–11.46; P=0.01) and CTP ischemic core ≤50 cc (OR, 3.86; 95% CI, 1.22–12.15; P=0.02) were both independently and comparably predictive of good outcome. Additionally, age in years (OR, 0.97; 95% CI, 0.93–0.97; P<0.01), NIHSS (National Institutes of Health Stroke Scale) (OR, 0.87; 95% CI, 0.83–0.92; P<0.01), procedure length (OR, 0.99; 95% CI, 0.98–0.99; P<0.01), and parenchymal hemorrhage (OR, 0.11; 95% CI, 0.02–0.45; P<0.01) were independently associated with good outcome.
We demonstrated significant variability of RAPID CTP-derived ischemic core volumes within different ASPECTS strata, as well as similar strengths of association between ASPECTS≥6 and CTP core ≤50 cc with good clinical outcome in a LVOS cohort that underwent thrombectomy.
NCCT ASPECTS is highly predictive of outcome.12 Importantly, studies have demonstrated that ASPECTS is more accurately determined by CTP as opposed to NCCT or CT angiography.13,14 However, the addition of more radiation, contrast media exposure, and potential treatment delays render the benefits indeterminate. The advantage of RAPID CTP software is the automatization and fast postprocessing, providing consistent and timely quantification of infarcted tissue and an estimation of at-risk tissue volumes. Both methods have been used and validated in recent clinical trials.1–5 The trials with advanced imaging selection techniques demonstrated higher rates of good clinical outcomes but with a potential cost of overselecting patients (as demonstrated by better outcomes in the control patients). We present a unique and large cohort of patients treated with broad pretreatment ASPECTS and CTP infarction core ranges, which allowed a pragmatic understanding of the strength of association of commonly used thresholds with outcome.
The ASPECTS regions are weighted unequally in regards to volume representation (semiquantitative method).15 Moreover, the specific ASPECTS regions may have interindividual variability. Therefore, our findings showing the significant variability of CTP-derived ischemic core volumes within the same ASPECTS grades and the lack of strong correlations between the methods were expected. Although higher ASPECTS scores are clearly related to good outcomes, segments of infarcted tissue may not be detectable in the early aftermath of infarction. Furthermore, borderline ASPECTS may not accurately reflect likelihood of favorable outcome.7,13 We demonstrate that patients with NCCT ASPECTS=6 may present with a wide variation of CTP ischemic core sizes. The addition of CTP in such scenario may aid in ruling out patients that may not benefit from reperfusion (large cores). When using ASPECTS <6 or CTP ischemic core >50 cc as cutoffs, patients who could benefit from thrombectomy may be missed; therefore, a complementary modality may help to define patient eligibility. Notably, we observed a similar strength of association of standard imaging thresholds (ASPECTS ≥6 and CTP ischemic core ≤50 cc) with good clinical outcome. Finally, a significant proportion of patients with CTP core >50 cc and ASPECTS <6 still achieved favorable outcomes, suggesting that these thresholds, interpreted individually, may be overly conservative.
This study has several limitations, mostly related to its retrospective nature. Importantly, this cohort is biased toward CT perfusion selection and toward patients treated with mechanical thrombectomy. The relatively later presentation compared with thrombolysis and thrombectomy clinical trials may have enhanced the relative accuracy of ASPECTS, although we did not find a significant impact of time from onset to presentation on the correlations of ASPECTS and FIV. Because of the fact that the study population was composed of confirmed LVOS, there is inevitably a selection bias toward patients with larger parenchymal ischemic cores. The study lacks analyses for predictors or correlations for poor outcomes, which is an important aspect of patient selection or exclusion for thrombectomy. Finally, we have not calculated interobserver coefficients for the different NCCT ASPECTS readers.
There is wide variability of RAPID CTP-derived ischemic core volumes within specific ASPECTS strata. Patient selection may be significantly affected by the imaging selection method. In clinical practice, a combined NCCT plus CTP (package approach) evaluation may better delineate patients who may not benefit from thrombectomy and avoid excluding patients who may benefit from it.
R.G. Nogueira declares having potential conflicts of interest with Stryker (PI:Trevo-2 PI/DAWN Trials), Covidien (SWIFT/SWIFT-PRIME Steering Committee, STAR Trial Core-Laboratory), and Penumbra (3-D Trial Executive Committee). The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.014117/-/DC1.
- Received May 18, 2016.
- Revision received July 13, 2016.
- Accepted July 18, 2016.
- © 2016 American Heart Association, Inc.
- Nogueira RG,
- Haussen DC,
- Dehkharghani S,
- Rebello LC,
- Lima A,
- Bowen M,
- et al
- Dehkharghani S,
- Bammer R,
- Straka M,
- Albin LS,
- Kass-Hout O,
- Allen JW,
- et al
- Bruno A,
- Akinwuntan AE,
- Lin C,
- Close B,
- Davis K,
- Baute V,
- et al
- Zaidat OO,
- Yoo AJ,
- Khatri P,
- Tomsick TA,
- von Kummer R,
- Saver JL,
- et al
- Dzialowski I,
- Hill MD,
- Coutts SB,
- Demchuk AM,
- Kent DM,
- Wunderlich O,
- et al
- Lin K,
- Rapalino O,
- Law M,
- Babb JS,
- Siller KA,
- Pramanik BK