Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Original Research n Cardiac
Brian S. Ko, MBBS (Hons), PhD Dennis T. L. Wong, MBBS (Hons), PhD James D. Cameron, MBBS, MD Darryl P. Leong, MBBS (Hons), PhD Siang Soh, MBBS Nitesh Nerlekar, MBBS (Hons) Ian T. Meredith, MBBS (Hons), PhD Sujith K. Seneviratne, MBBS
Purpose:
To identify computed tomographic (CT) coronary indexes independently associated with a fractional flow reserve (FFR) of 0.8 or less, to derive a score that combines CT indexes most predictive of an FFR of 0.8 or less, and to evaluate the diagnostic accuracy of the score in predicting an FFR of 0.8 or less.
Materials and Methods:
This retrospective study had institutional review board approval and waiver of the need to obtain informed consent. Consecutive patients who underwent CT coronary angiography and FFR assessment with one or more discrete lesion(s) of intermediate (30%–70%) severity at CT were included. Quantitative CT measurements were performed by using dedicated software. The CT indexes evaluated included the following: plaque burden, minimal luminal area and diameter, stenosis diameter, area of stenosis, lesion length, remodeling index, plaque morphology, calcification severity, and the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) score, which approximates the size of the myocardium subtended by a lesion. By using covariates independently associated with an FFR of 0.8 or less, a score was determined on the basis of modified Akaike information criteria, and the C statistics of individual and combined indexes were compared.
Results:
Eighty-five patients (mean age, 64.2 years; range, 48–88 years; 65.9% men; 124 lesions; 38 lesions with an FFR 0.8) were included. Area of stenosis, lesion length, and APPROACH score were the strongest predictors of an FFR of 0.8 or less and were used to derive the ASLA score. The optimism-adjusted Harrell C statistic for the combined score was 0.82, which was superior to that for area of stenosis (0.74), lesion length (0.75), and the APPROACH score (0.71) (P , .001 for trend). The corresponding incremental discrimination improvement indexes were 0.17, 0.11, and 0.19, respectively (P , .001 for all), suggesting that the score improves reclassification compared with any one angiographic index. The average time required for score derivation was 102.6 seconds.
Conclusion:
The ASLA score, which accounts for CT-derived area of stenosis, lesion length, and APPROACH score, may conveniently improve the prediction, beyond individual indexes, of functionally significant intermediate coronary lesions.
1
From the Monash Cardiovascular Research Centre, Department of Medicine (Monash Medical Centre), Monash University and Monash Heart, Monash Health, 246 Clayton Road, Clayton, 3168 VIC, Australia (B.S.K., D.T.L.W., J.D.C., S.S., N.N., I.T.M., S.K.S.); Discipline of Medicine, University of Adelaide, Adelaide, Australia (D.T.L.W., D.P.L.); and Discipline of Medicine, Flinders University, Adelaide, Australia (D.P.L.). Received June 19, 2014; revision requested August 4; revision received November 11; accepted November 12; final version accepted December 10. B.S.K. and D.T.L.W. supported by the National Heart Foundation of Australia and the Robertson Family Scholarship. Address correspondence to B.S.K. (e-mail: [email protected]). RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Imaging
The ASLA Score: A CT Angiographic Index to Predict Functionally Significant Coronary Stenoses in Lesions with Intermediate Severity—Diagnostic Accuracy1
RSNA, 2015
q
Online supplemental material is available for this article. 1
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
C
omputed tomographic (CT) coronary angiography is recommended as an initial investigation to assess symptomatic patients with low to intermediate risk of coronary artery disease (1). In its current form, however, CT coronary angiography has limited specificity in determining the functional significance of coronary stenoses (2,3). For this reason, patients with moderate stenoses may require referral for noninvasive functional imaging to confirm the presence of ischemia before the decision is made to proceed to conventional angiography. A number of novel CT techniques have recently been described that may improve the ability of CT coronary angiography to predict ischemia. These include the use of adenosine stress CT myocardial perfusion imaging (4,5), the assessment of the transluminal attenuation gradient across coronary stenoses on the basis of resting CT coronary angiography images (6), and the noninvasive prediction of fractional flow reserve (FFR) by the application of computational fluid dynamics to resting CT coronary angiography images (3,7). Each technique has been demonstrated to improve the specificity, positive predictive value, and overall accuracy of CT coronary angiography in the prediction of functionally significant coronary
Advances in Knowledge nn Combined CT assessment of area of stenosis, lesion length, and size of the subtended myocardium with use of the ASLA score improves the prediction of functionally significant coronary stenosis beyond that of individual indexes alone (C statistics for combined assessment, area of stenosis, lesion length, and Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease score were 0.82, 0.74, 0.75, and 0.71 respectively; P , .001 for trend). nn The ASLA score can be conveniently applied in 102.6 seconds 6 37.5 (standard deviation). 2
Ko et al
stenoses as assessed by invasive FFR measurement (FFR 0.8) (3–7). Each novel technique in its current form, however, may have limitations. Stress perfusion imaging with CT requires additional imaging and hence is associated with increased radiation exposure and iodinated contrast material usage. CT-derived FFR can be computed currently only by the use of a supercomputer, which currently requires at least hours of processing time (3). While the assessment of transluminal attenuation gradient is promising and convenient, it requires manual vessel tracking with dedicated research software, and the technique remains unvalidated on images acquired by using narrow-detector CT scanners (6). Hence, there remains an unmet need for a simple and easily accessible approach for the CT angiographer to predict the functional significance of coronary stenoses—in particular, intermediate coronary stenoses identified at CT—to assist in clinical decisions regarding referral for conventional angiography. Results of recent studies (8,9) demonstrate that CT angiographic predictors of FFR may include minimal luminal diameter, area of stenosis, and lesion length—which by themselves remain insufficient predictors of an FFR of 0.8 or less. Moreover, the myocardium subtended by the lesion has also been demonstrated to strongly correlate with ischemic burden as assessed with cardiac magnetic resonance (MR) imaging and invasive FFR measurement (10,11), although data on the use of these scores with CT coronary angiography remain unknown. We hypothesize that the combined use of angiographic indexes may improve prediction of reduced FFR compared with the use of individual indexes alone. Our study was therefore performed to identify CT indexes independently associated with an FFR of 0.8 or
Implication for Patient Care nn Use of this score may decrease the need for further functional testing and conventional angiography.
less, to derive a score that combines CT indexes most predictive of an FFR of 0.8 or less, and to evaluate the diagnostic accuracy of the score in predicting an FFR of 0.8 or less.
Materials and Methods This retrospective study had approval from the institutional review board of Monash Health, with waiver of the need to obtain informed consent. The authors who have no relationships with industry had control of inclusion of all data and information in this article. We included consecutive patients who underwent clinically mandated CT coronary angiography and nonurgent conventional coronary angiography with FFR assessment performed in at least one discrete lesion of intermediate severity (30%–70%) as visually assessed at CT coronary angiography in our institution, Monash Heart, between October 2009 and July 2012. Conventional angiography and FFR measurement were performed a mean of 11 days after CT coronary angiography. Patients were excluded in cases of left ventricular dysfunction, if the time between CT coronary angiography and FFR measurement was longer than 6 Published online before print 10.1148/radiol.15141231 Content codes: Radiology 2015; 000:1–11 Abbreviations: APPROACH = Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease ASLA = area of stenosis, lesion length, and APPROACH score CI = confidence interval FFR = fractional flow reserve Author contributions: Guarantors of integrity of entire study, B.S.K., N.N., S.K.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, B.S.K., D.T.L.W., I.T.M.; clinical studies, B.S.K., S.S., N.N., I.T.M., S.K.S.; experimental studies, B.S.K., I.T.M.; statistical analysis, B.S.K., J.D.C., D.P.L., I.T.M.; and manuscript editing, B.S.K., D.T.L.W., J.D.C., D.P.L., I.T.M., S.K.S. Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
months, if there were adverse cardiac events or revascularization during this time interval, if the acute coronary syndrome was present in the 3 months prior to the CT examination, and if they had bypass graft lesions or significant left main artery stenoses. Among the patients who met the inclusion criteria, those with uninterpretable vessels at CT and those with stenosis greater than 70% or less than 30%, poor CT image quality, small vessel diameter (,2 mm), intracoronary stents, excessive calcifications, and/or myocardial bridges were excluded.
CT Coronary Angiography Patients underwent CT coronary angiography performed by using a 320–detector row CT scanner (Aquilion ONE; Toshiba Medical Systems, Nasu, Japan). β-Blockers and nitroglycerin were administered to achieve a prescanning heart rate of less than 65 beats per minute and to optimize vasodilation in accordance with previously published guidelines (12). A bolus of 55 mL of 100% iohexol 56.6 g/75 mL (Omnipaque 350; GE Healthcare, Princeton, NJ) was injected into an antecubital vein at a flow rate of 5 mL/sec, followed by 20 mL of a 30:70 mixture of contrast material and saline, followed by 30 mL of saline. Scanning parameters were as follows: detector collimation, 320 3 0.5 mm; tube current, 300–500 mA (depending on body mass index); tube voltage, 120 kV; gantry rotation time, 350 msec; and temporal resolution, 175 msec. Prospective electrocardiographic gating was used, covering phases 70%– 80% of the R-R interval. For images acquired at heart rates of 65 beats per minute or slower, scanning was completed with a single R-R interval utilizing a 180° segment. In patients with a heart rate greater than 65 beats per minute, data segments from two consecutive beats were used for multisegment reconstruction with an improved temporal resolution of 87 msec. Analysis of CT Coronary Angiograms Analysis was performed in coronary vessels that were greater than 2 mm in diameter. All data were transferred
to an external workstation (Vitrea 6, version 3.0; Vital Images, Minnetonka, Minn) for further analysis. Two angiographers with 5 years of experience in CT coronary angiography interpretation (B.S.K. and D.T.L.W.), who were blinded to the results of corresponding conventional angiography and FFR measurement, performed the analysis independently. All included vessels were deemed to be suitable for analysis of the coronary artery and plaque characteristics by the CT interpreters. Coronary plaques were classified as noncalcified, calcified, or mixed. Plaques with CT attenuation lower than that of the luminal contrast material were defined as noncalcified, and plaques with high CT attenuation that could be visualized separately from the contrast material in the lumen were defined as calcified. Mixed plaque was defined as the presence of calcified and noncalcified plaque components within the same lesion. The extent of coronary calcium was categorized as none, as less than 50% of the cross-sectional area, or as 50% or greater of the cross-sectional area. Quantification of luminal narrowing and vessel remodeling in the target lesion was performed by using a dedicated software tool (Sure Plaque, Vitrea 6, version 3.0; Vital Images and Toshiba Medical Systems) that has been previously validated against intravascular ultrasonography (13). Automatic vessel tracking was then used to locate the vessel centerline on the basis of the opacification of the lumen. After manual determination of the coronary lesion, the minimal lumen diameter (in millimeters), minimal lumen area (in square millimeters), and lesion length (in millimeters) were computed by the software. The reference diameter and area were determined as an average of the vessel dimensions immediately proximal and distal to the lesion, where minimal or no plaque could be detected. In ostial lesions, only one reference was used. The diameter of stenosis, as a percentage, was defined as 100% · (reference diameter 2 minimal lumen diameter)/ reference diameter. The area of stenosis, as a percentage, was defined as
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
100% · (reference lumen 2 minimal lumen area/reference lumen). Coronary plaque was defined as the area between the outer contour of the vessel and the lumen border, both of which were determined by the software (8). Measurements were obtained for each plaque and were defined as follows: plaque burden (in square millimeters) = (vessel area 2 lumen area)/vessel area, and remodeling index = outer vessel area (stenotic segment)/outer vessel area (reference segment).
Modified APPROACH Score The Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) score is based on the division of the left ventricle into regions in accordance with pathologic studies in humans and with correlation with cardiac catheterization results, which evaluate the proportion of the myocardium perfused by each artery (14–16). Table E1 (online) demonstrates the derivation of the APPROACH score, as has been previously described (17). Furthermore, it has been validated with cardiac MR imaging methods to quantify the myocardial area at risk after myocardial infarction (18). The modified score takes into account the location of the lesion (proximal, middle, or distal) and the dominance and size of the secondary branches and provides an estimate of the percentage of supplied myocardium beyond the considered coronary lesion (17). Conventional Coronary Angiography and FFR Measurement Conventional coronary angiography was performed as per standard clinical practice. The pressure wire (Pressure Wire Certus; St Jude Medical, St Paul, Minn) was calibrated and electronically equalized with the aortic pressure before being placed distal to the stenosis in the distal third of the coronary artery being interrogated. Intracoronary glyceryl trinitrate (100 µg) was injected to minimize vasospasm. Intravenous adenosine was administered at 140 µg/ kg/min through an intravenous line in the antecubital fossa. At steady-state 3
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
hyperemia, FFR was assessed by using the RadiAnalyser Xpress (St Jude Medical) and was calculated by dividing the mean coronary pressure measured with the pressure sensor placed distal to the stenosis by the mean aortic pressure measured through the guide catheter. An FFR of 0.8 or less was considered to define ischemia in the interrogated artery and its supplied territory (19,20). This threshold was chosen because it may determine vessel outcome and benefit from revascularization (21).
Statistical Analysis Continuous variables are expressed as means 6 standard deviations or with 95% confidence intervals (CIs). Categoric variables are expressed as percentages. Interobserver variability was assessed by using the intraclass coefficient. To identify the optimal combination of CT coronary angiography parameters for the estimation of FFR, linear mixedeffects modeling was performed. Patient identity was included as a random effect to account for the likelihood of greater correlation between different vessels in a particular patient than between vessels from different patients. To identify CT angiographic covariates that were independently associated with an FFR of 0.8 or less, we used generalized estimating equations. The following covariates were considered at a univariate level: plaque burden, minimal lumen area, minimal lumen diameter, diameter of stenosis, area of stenosis, lesion length, remodeling index, calcification, plaque morphology, and APPROACH score. Those covariates with a univariate P , .2 were included in the multivariate analysis, which was performed by using the “enter” approach. To ensure the most parsimonious model, the modified Akaike information criterion was used (22); models with a lower Akaike criterion are preferable to models with a higher Akaike criterion. The ability of the final model to predict an FFR of 0.8 or less was assessed by the optimism-adjusted Harrell C statistic on bootstrapped samples with 100 times replacement. The optimal score threshold is one that maximizes the sum of sensitivity and specificity with a 4
Ko et al
Figure 1
Figure 1: Patient (pt) and vessel flow diagram. ∗ = Numbers in parentheses are numbers of vessels rather than numbers of patients.
minimal sensitivity of 75%. Finally, the incremental utility of the score was assessed by using the integrated discrimination improvement index described by Pencina et al (23) as follows: integrated discrimination improvement = (ISnew 2 ISold) 2 (IPnew 2 IPold), where “new” refers to a model containing a novel diagnostic tool of interest in addition to conventional risk predictors, “old” pertains to the model containing only the conventional risk markers, and IS and IP are the integrals of sensitivity and (1 2 specificity), respectively. Statistical analysis was performed with SPSS 18.0 (SPSS, Chicago, Ill) and STATA 12.1 (Stata, College Station, Tex). P , .05 was considered to indicate a statistically significant difference.
Results There were 121 patients who underwent CT and FFR assessment in our institution during the study period
(Fig 1). Ten patients were excluded for the presence of left ventricular dysfunction (n = 4), a time between CT coronary angiography and FFR measurement of more than 6 months (n = 2), presence of acute coronary syndrome in the 3 months prior to CT coronary angiography (n = 1), presence of cardiac events or revascularization between CT angiography and FFR measurement (n = 1), presence of coronary artery bypass graft (n = 1), and left main disease (n = 1). Among the remaining 111 patients and 204 vessels, 80 vessels were excluded from analysis because of the presence of severe (.70%) or minor stenoses (n = 34), the presence of poor image quality as a result of excessive noise or motion (n = 17), vessel diameter of less than 2 mm (n = 11), intracoronary stent (n = 7), excessive calcification (n = 6), myocardial bridging (n = 2), or incomplete information (n = 3). The remaining 85 patients (mean age, 64.2 years 6 11.2; 66% men) with
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Table 1 Patient and Vessel Characteristics Characteristics Patient characteristics (n = 85) Age (y)* Age in male cohort (y)† Age in female cohort (y)† P value for difference in age between male and female cohorts Sex No. of men No. of women Body mass index (kg/m2)* Symptom Typical angina Atypical angina or noncardiac pain Suspected coronary artery disease Known coronary artery disease Previous myocardial infarction Previous percutaneous coronary intervention Cardiovascular risk factors Diabetes Hypertension Hypercholesterolemia Current smoker Family history of ischemic heart disease No. of vessels with a 50% stenosis as visually assessed at CT coronary angiography 0 1 2 3 FFR findings No. of vessels interrogated with FFR measurement 1 2 3 No. of vessels interrogated per patient* Vessels with FFR 0.8 per patient 0 1 2 No. of vessels with FFR 0.8 per patient* Lesion characteristics (n = 124) 50% diameter stenosis as visually assessed at CT coronary angiography Quantitative CT findings* Diameter stenosis (%) Area of stenosis (%) Minimal luminal diameter (mm) Minimal luminal area (mm2) Lesion length (mm) Plaque burden APPROACH score
Datum 64.2 6 11.2 63.9 (48–82) 64.7 (51–88) .76 56 (65.9) 29 (34.1) 28.1 6 5.1 65 (76.5) 20 (23.5) 49 (57.6) 36 (42.4) 9 (9.5) 12 (12.7) 13 (15.3) 63 (74.1) 63 (74.1) 13 (15.3) 34 (40.0)
36 (42.4) 42 (29.4) 5 (5.9) 2 (2.4)
55 (64.7) 21 (24.8) 9 (10.6) 1.6 6 0.8 48 (56.5) 35 (41.2) 2 (2.4) 0.46 6 0.55 58 (46.8) 39.4 6 20.9 47.8 6 24.0 1.8 6 0.8 5.0 6 3.5 19.5 6 11.4 71.5 6 15.6 26.5 6 11.9
Table 1 (continues)
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
124 vessels were included for analysis. Patient and vessel characteristics are summarized in Table 1. FFR measurement was performed in all 85 patients, of whom 55 (64.7%) underwent FFR interrogation in a single vessel; 21 (24.8%), in two vessels; and nine (10.6%), in three vessels. Overall, FFR ranged from 0.39 to 1.0 (mean, 0.84 6 0.11). Thirty-eight vessels (30.6%) were classified as having functionally significant stenosis, with an FFR of 0.8 or less. There were no adverse events during FFR interrogation.
Accuracy of Visual Assessment at CT in Predicting an FFR of 0.8 or Less On a per-vessel basis, 58 (48.6%) vessels had visual stenosis of 50% or greater at CT. The presence of a visual stenosis of 50% or greater at CT angiography was associated with a 78.9% sensitivity, 67.4% specificity, 51.7% positive predictive value, and 87.9% negative predictive value. The bootstrapped Harrell C statistic for visual stenosis in the diagnosis of an FFR of 0.8 or less was 0.74 (95% CI: 0.64, 0.83). Accuracy of Quantitative CT Parameters in Predicting an FFR of 0.8 or Less The results of univariate and multivariate analyses for predictors of FFR are presented in Table 2. According to univariate analysis, minimal lumen area, minimal lumen diameter, plaque burden, diameter of stenosis, area of stenosis, lesion length, remodeling index, presence of calcification, plaque morphology, and APPROACH score were all significant predictors of an FFR of 0.8 or less. According to results of multivariate analysis, plaque burden, area of stenosis, lesion length, and the APPROACH score remained significant predictors for lesion-specific ischemia. Accuracy of area of stenosis quantified at CT in predicting an FFR of 0.8 or less.—On a per-vessel basis, 56 (45.2%) vessels had an area of stenosis of 50% or greater as quantified at CT. The mean area of stenosis in lesions with an FFR of 0.8 or less was 61.5% 6 22.5, versus 41.8% 6 22.3 in vessels with an FFR greater than 0.8 (P , 5
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
mean plaque burden in lesions with an FFR of 0.8 or less was 69.3 6 14.5, versus 76.5 6 17.1 in vessels with an FFR greater than 0.8 (P , .0001). There was a statistically significant negative correlation between plaque burden and FFR (r = 20.27, P = .003) (Fig 2). The bootstrapped Harrell C statistic for plaque burden in the prediction of an FFR of 0.8 or less was 0.64 (95% CI: 0.54, 0.76).
Table 1 (continued) Patient and Vessel Characteristics Characteristics
Datum
Quantitative coronary angiography findings* Diameter stenosis (%) Minimal luminal diameter (mm) Vessel diameter (mm) FFR findings Mean FFR in all lesions* Lesions with FFR 0.8 Mean FFR in lesions with FFR 0.8* Lesion location Proximal right coronary artery Middle right coronary artery Distal right coronary artery Right posterior descending artery or posterolateral ventricular branch Proximal left anterior descending artery Middle left anterior descending artery Distal left anterior descending artery Diagonal artery Proximal left circumflex artery Ramus intermediate Obtuse marginal arteries Distal left circumflex artery
37.4 6 17.8 1.6 6 0.7 2.5 6 0.7 0.84 6 0.11 38 (30.6) 0.71 6 0.09 15 (12.1) 8 (6.5) 2 (1.6) 3 (2.4) 31 (25.0) 23 (18.5) 4 (3.2) 3 (2.4) 24 (20.2) 1 (0.8) 7 (5.6) 2 (1.6)
Note.—Unless otherwise specified, data are numbers of patients or lesions, with percentages in parentheses. * Data are means 6 standard deviations. †
Data are means, with ranges in parentheses.
.0001). Area of stenosis measurements were reproducible, with an interobserver intraclass correlation coefficient of 0.98 (95% CI: 0.91, 0.99). There was a statistically significant negative correlation between area of stenosis and FFR (r = 20.37, P , .0001) (Fig 2). The presence of an area of stenosis of 50% or greater was associated with 65.9% sensitivity, 62.1% specificity, 44.6% positive predictive value, and 79.7% negative predictive value. The bootstrapped Harrell C statistic for area of stenosis in the prediction of an FFR of 0.8 or less was 0.74 (95% CI: 0.65, 0.83). Accuracy of lesion length in predicting an FFR of 0.8 or less.—The mean lesion length in vessels with an FFR of 0.8 or less was 27.2 mm 6 12.9, versus 16.1 mm 6 8.9 in vessels with an FFR of greater than 0.8 (P , .0001). There was a statistically significant negative correlation between lesion length and FFR (r = 20.45, P , .0001) (Fig 6
2). The bootstrapped Harrell C statistic for lesion length in the prediction of an FFR of 0.8 or less was 0.75 (95% CI: 0.66, 0.85). The lesion length measurements were highly reproducible, with an interobserver intraclass correlation coefficient of 0.95 (95% CI: 0.83, 0.99). Accuracy of APPROACH score in predicting an FFR of 0.8 or less.—The mean APPROACH score in lesions with an FFR of 0.8 or less was 32.7 6 11.6, versus 23.7 6 11.0 in vessels with an FFR of greater than 0.8 (P , .0001). There was a statistically significant negative correlation between APPROACH score and FFR (r = 20.34, P = .0002) (Fig 2). The bootstrapped Harrell C statistic for the APPROACH score in the prediction of an FFR of 0.8 or less was 0.71 (95% CI: 0.61, 0.81). The APPROACH score demonstrated high reproducibility, with an interobserver intraclass correlation coefficient of 1.0 (95% CI: 1.0, 1.0). Accuracy of plaque burden in predicting an FFR of 0.8 or less.—The
ASLA SCORE The combined score was derived by means of multivariate generalized estimating equation regression. It was named the ASLA score, as it included CT-quantified area of stenosis and lesion length and the APPROACH score as ordinal variables categorized by using the cut points described in Table 3. Each was important for the model, as indicated by a likelihood ratio test P value of less than .02 when each parameter was dropped sequentially from the model. Plaque burden was not taken into account in the score because it did not lower the model’s modified Akaike criterion substantially and so was thought not to contribute to the model’s clinical performance. An exchangeable correlation structure was selected for the regression, as this minimized the modified Akaike criterion compared with other correlation structures. Figure 3 illustrates the application of the ASLA score. Accuracy of ASLA score in predicting an FFR of 0.8 or less.—The mean ASLA score in vessels with an FFR of 0.8 or less was 10.1 6 4.77, compared with 4.43 6 3.62 in vessels with an FFR greater than 0.8 (P , .001) (Fig 3). The bootstrapped Harrell C statistic of the ASLA score in predicting significant FFR was 0.82 (95% CI: 0.74, 0.91). The ASLA score was a better discriminator of an FFR of 0.8 or less than area of stenosis, lesion length, and APPROACH score (respective C statistics, 0.82 [95% CI: 0.74, 0.91], 0.74 [95% CI: 0.65, 0.83], 0.75 [95% CI: 0.66, 0.85], and 0.71 [95% CI: 0.61, 0.81]; P , .001 for trend). The comparison of areas under the receiver operating characteristic curve is presented in Figure 4. A representation of sensitivity
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
Table 2 Univariate and Multivariate Predictors of an FFR of 0.8 or Less Predictor Plaque burden Minimal luminal area Minimal luminal diameter Diameter of stenosis Area of stenosis Lesion length Remodeling index Calcification Plaque morphology APPROACH score
Univariate Coefficient
P Value
Multivariate Coefficient
P Value
0.033 (0.0050, 0.061) 20.22 (20.67, 20.070) 20.81 (21.4, 20.25) 0.029 (0.0093, 0.049) 0.038 (0.019, 0.056) 0.090 (0.051, 0.13) 20.0068 (20.018, 0.0044) 0.59 (0.023, 1.2) 0.31 (20.16, 0.77) 0.069 (0.035, 0.10)
.02 .004 .004 .004 ,.001 ,.001 .2 .04 .2 ,.001
20.071 (20.13, 20.013) 20.065 (20.37, 0.24) 21.2 (22.9, 0.41) 20.044 (20.10, 0.014) 0.067 (0.015, 0.12) 0.091 (0.036, 0.15) ... 20.068 (20.85, 0.71)
.02 .7 .1 .1 .01 .001 ... .9
0.053 (0059, 0.10)
.03
Note.—Data in parentheses are 95% CIs.
Figure 2
Figure 2: Scatterplots of FFR and CT angiographic parameters, including, A, area of stenosis, B, lesion length, C, APPROACH score, and, D, plaque burden. There was a statistically significant inverse relationship between CT-derived area of stenosis and FFR (R = 20.37, P , .0001), lesion length (R = 20.45, P , .0001), APPROACH score (R = 20.34, P = .0002), and plaque burden (R = 20.26, P = .003).
and specificity as functional cutoff values over a spectrum of ASLA scores is presented in Figure 5. A threshold score of 7 provided 76.3% (95% CI: 59.4%, 88.0%) sensitivity, 76.7% (95% CI:
66.2%, 85.0%) specificity, 59.2% (95% CI: 44.3%, 72.7%) positive predictive value, 88.0% (95% CI: 78.0%, 94.0%) negative predictive value, and 76.6% overall accuracy. A score of 3 or lower
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
was associated with 92% sensitivity and accounted for 50 (40.3%) lesions. Similarly, a score of 11 or higher was associated with 86% specificity and accounted for 21 lesions (16.9%) (Fig 6). 7
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Among the 50 lesions with a score of 3 or lower, four (8%) had an FFR of 0.8 or less (range, 0.68–0.78), including two lesions in the left anterior descending artery, one in a marginal branch, and one in the right coronary artery. Among the 21 lesions with a score of 11 or greater, four (19%) had an FFR of greater than 0.8 (range, 0.81–0.95),
Table 3 The ASLA Score Variable and Value
Points Assigned
Area of stenosis (%) .63 47–63 31–46 ,31 Lesion length (mm) .28 10.8–28 ,10.8 APPROACH score .44 25.1–44 18–25 ,18 Total
7 2 1 0 6 1 0 5 2 1 0 Maximum = 18
Ko et al
including three lesions in the left anterior descending artery and one in the left circumflex artery. We estimated the integrated discrimination improvement indexes for the score over and above each of its constituents (area of stenosis, lesion length, and APPROACH score) by using the approach of Pencina et al (23) (Table 4). The integrated discrimination improvement indexes were 0.17 (P , .001), 0.11 (P , .001), and 0.19 (P , .001), suggesting that the score improves reclassification of estimated FFR compared with any one angiographic index. Time required to perform ASLA scoring.—The average time required to perform ASLA scoring per lesion was 102.6 seconds 6 37.5 (25th percentile, 70.1 seconds; 75th percentile, 131.8 seconds).
Discussion This study derived a score on the basis of information quantified at CT to predict the hemodynamic significance of coronary artery stenoses. Our results demonstrate that the ASLA score, which takes into account area of stenosis, lesion length, and the APPROACH
score, is a superior predictor of functionally significant stenoses as assessed by using invasive FFR measurement, beyond individual CT indexes alone or visually assessed stenosis of 50% or greater at CT angiography. The ASLA score can be conveniently applied in less than 2 minutes. On the basis of multivariate analysis, area of stenosis and lesion length were found to be important predictors of an FFR of 0.8 or less. Our findings are in agreement with those of Kristensen et al (8), who found, using a similar CT-based method in a smaller cohort of 42 patients with 56 intermediate stenoses, that area of stenosis and lesion length were the strongest determinants of an abnormal FFR and were superior to plaque volume, minimal luminal diameter, and minimal luminal area. The incremental value of lesion length when added to standard angiographic measurement of diameter of stenosis in the prediction of FFR significant lesions has been reported (24,25). However, as with the results of the previous studies, our results highlight the limited predictive value of individual CT-derived parameters for an FFR of 0.8 or less (8,9).
Figure 3
Figure 3: A–D, Curved images from quantitative CT examination performed in a patient with an intermediate (50%–70%) lesion in the left circumflex artery. The lesion length was 12.4 cm (on B ), the area of stenosis was 77% on axial images (in C ), and the APPROACH score was 24.5 (on D ). The ASLA score was 9 (1 for lesion length plus 7 for area of stenosis plus 1 for lesion length). The FFR in the left circumflex artery was 0.76. 8
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Figure 4
Figure 4: Receiver operating characteristic curves of ASLA score, APPROACH score, area of stenosis, and lesion length.
Figure 5
Figure 5: Graph shows sensitivity and specificity as functional cutoff values over a spectrum of ASLA scores.
It is increasingly recognized that FFR is highly dependent on the area of myocardium supplied beyond the lesion in question (11,26). The APPROACH score has been used to estimate the myocardium at risk during myocardial infarction and was found to be inversely correlated with FFR (11). This study describes the correlation of a CT-derived APPROACH score with FFR. Our study results demonstrate modest but similar correlation (r = 20.34, P , .0001) with FFR as had been described for conventional angiography (11). We also demonstrated that the score is highly reproducible, with minimal interobserver variability, which is in agreement with previous reports (11,18).
The ASLA score, which takes into account area of stenosis, lesion length, and APPROACH score, was found to be superior and incremental to individual indexes alone or visual CT assessment in the prediction of an FFR of 0.8 or less. Visually assessed diameter of stenosis at CT currently plays an important role in the decision as to whether a patient may require further examination with noninvasive stress testing or invasive angiography (27). Rossi et al (9) demonstrated that area of stenosis as derived at quantitative CT may be superior in specificity to visually assessed diameter of stenosis in a cohort of 99 symptomatic patients. Our results extend these results and emphasize the importance of accounting for
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
the myocardium supplied distal to the lesion and the length of the lesion, in addition to the area of stenosis. Taking into account all three in the form of a score may improve diagnostic accuracy in the prediction of an FFR of 0.8 or less. The Poiseuille equation states that the pressure gradient across a coronary lesion is directly proportional to the coronary blood flow, blood viscosity, and lesion length and is inversely proportional to the 4th power of the radius. It would be intuitive for a score to take into account each of these key elements. Accordingly, the ASLA score includes the APPROACH score, which may act as a surrogate for myocardial blood flow (11); minimal area of stenosis, which represents the square of the radius of the vessel at the lesion; and lesion length. The ASLA score can be derived by using standard rest CT angiography images, without additional radiation exposure, contrast material, medications, reconstruction, or cost. The score can also be readily calculated with software available on most currently available reporting workstations during CT angiography interpretation. The technique is convenient and requires a short processing time. In our cohort, 57% of intermediate lesions had scores that were 3 or lower or that were 11 or higher. In the former, 92% were not FFR significant and in the latter, 81% were FFR significant. This suggests that the score may have value in differentiating lesions that may or may not require further functional assessment and/or conventional angiography. In contrast, in lesions with scores between 4 and 10, our results suggest that it will still be important to perform functional assessment or FFR measurement. Notably, given the sample size and the number of false-positive and falsenegative results at the extremes of the ASLA score, our results represent a preliminary report on the described technique. Larger studies with methodologic improvements are required to improve the diagnostic performance of the technique and to assess whether 9
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
in this study, as all patients, including the nine patients who had previously experienced myocardial infarction, had documented normal left ventricular ejection fraction at transthoracic echocardiography or conventional angiography, as per the inclusion criteria. However, the applicability of the study findings to patients with a sizeable loss of viable myocardial mass has not been studied. Last, we acknowledge that measurements of area of stenosis and lesion length in this study were quantified by using dedicated software provided by a sole vendor. The applicability of the technique using equipment from other vendors remains unknown. In conclusion, combined CT angiographic assessment with the ASLA score, which includes area of stenosis, lesion length, and APPROACH score, provides incremental predictive value over individual indexes alone for the detection of functionally significant coronary artery stenoses.
Figure 6
Figure 6: Scatterplot of FFR and ASLA score. A total of 124 lesions were assessed. In the 50 lesions with an ASLA score of 3 or lower, four (8%) had an FFR of 0.8 or less (range, 0.68–0.78), including two lesions in the left anterior descending artery, one in a marginal branch, and one in the right coronary artery. In the 21 lesions with an ASLA score of 11 or higher, four (19%) had an FFR greater than 0.8 (range, 0.81–0.95), including three lesions in the left anterior descending artery and one in the left circumflex artery.
Table 4 Lesion Characteristics Quantified according to CT and Interobserver Variability Characteristic Area of stenosis (%) Lesion length (mm) APPROACH score
Mean 6 Standard Deviation
Interobserver Intraclass Correlation Coefficient
47.8 6 24.0 19.5 6 11.4 26.5 6 11.9
0.977 (0.91, 0.99) 0.954 (0.824, 0.988) 1.0 (1.0, 1.0)
Note.—Data in parentheses are 95% CIs.
the use of the score may save costs and decrease downstream testing. There were limitations to our study. Our results represent a retrospective single-center experience involving 85 patients and hence require confirmation with larger studies. The findings are limited to intermediate stenoses alone, and the applicability of the score to vessels with tandem lesions or severe stenoses, in bypass graft lesions, and in patients with impaired left ventricular function is not known. Analysis was restricted to a pervessel basis alone. In a proportion of patients who underwent multivessel FFR interrogation, only one vessel 10
was included in the final analysis, as the remaining vessels met exclusion criteria. For this reason, the accuracy of the applied technique cannot be extended to the patient per se, but remains true for the vessel. Hence, perpatient analysis was not performed. The incremental value of the technique on pretest information, including symptom severity and results of functional testing, has also not been evaluated. We acknowledge that in patients with prior myocardial infarction, FFR may be elevated because of the loss of viable myocardial mass (26). The effect of this may have been minimized
Disclosures of Conflicts of Interest: B.S.K. disclosed no relevant relationships. D.T.L.W. disclosed no relevant relationships. J.D.C. disclosed no relevant relationships. D.P.L. disclosed no relevant relationships. S.S. disclosed no relevant relationships. N.N. disclosed no relevant relationships. I.T.M. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is a consultant for Boston Scientific and Medtronic. Other relationships: disclosed no relevant relationships. S.K.S. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is on the speakers bureau of Toshiba. Other relationhips: disclosed no relevant relationships.
References 1. Taylor AJ, Cerqueira M, Hodgson JM, et al. ACCF/SCCT/ACR/AHA/ASE/ASNC/NASCI/ SCAI/SCMR 2010 Appropriate Use Criteria for Cardiac Computed Tomography. A Report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, the Society of Cardiovascular Computed Tomography, the American College of Radiology, the American Heart Association, the American Society of Echocardiography, the American Society of Nuclear Cardiology, the North American Society for Cardiovascular Imaging, the Society for Cardiovascular Angiography and Interventions,
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
and the Society for Cardiovascular Magnetic Resonance. Circulation 2010;122(21):e525– e555. 2. Dorbala S, Hachamovitch R, Di Carli MF. Myocardial perfusion imaging and multidetector computed tomographic coronary angiography: appropriate for all patients with suspected coronary artery disease? J Am Coll Cardiol 2006;48(12):2515–2517. 3. Koo BK, Erglis A, Doh JH, et al. Diagnosis of ischemia-causing coronary stenoses by noninvasive fractional flow reserve computed from coronary computed tomographic angiograms: results from the prospective multicenter DISCOVER-FLOW (Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve) study. J Am Coll Cardiol 2011;58(19):1989– 1997. 4. Ko BS, Cameron JD, Leung M, et al. Combined CT coronary angiography and stress myocardial perfusion imaging for hemodynamically significant stenoses in patients with suspected coronary artery disease: a comparison with fractional flow reserve. JACC Cardiovasc Imaging 2012;5(11):1097– 1111. 5. Bettencourt N, Chiribiri A, Schuster A, et al. Direct comparison of cardiac magnetic resonance and multidetector computed tomography stress-rest perfusion imaging for detection of coronary artery disease. J Am Coll Cardiol 2013;61(10):1099–1107. 6. Wong DT, Ko BS, Cameron JD, et al. Transluminal attenuation gradient in coronary computed tomography angiography is a novel noninvasive approach to the identification of functionally significant coronary artery stenosis: a comparison with fractional flow reserve. J Am Coll Cardiol 2013;61(12):1271–1279. 7. Min JK, Leipsic J, Pencina MJ, et al. Diagnostic accuracy of fractional flow reserve from anatomic CT angiography. JAMA 2012;308(12):1237–1245. 8. Kristensen TS, Engstrøm T, Kelbæk H, von der Recke P, Nielsen MB, Kofoed KF. Correlation between coronary computed tomographic angiography and fractional
flow reserve. Int J Cardiol 2010;144(2):200– 205. 9. Rossi A, Papadopoulou SL, Pugliese F, et al. Quantitative computed tomographic coronary angiography: does it predict functionally significant coronary stenoses? Circ Cardiovasc Imaging 2014;7(1):43–51. 10. Jogiya R, Kozerke S, Morton G, et al. Validation of dynamic 3-dimensional whole heart magnetic resonance myocardial perfusion imaging against fractional flow reserve for the detection of significant coronary artery disease. J Am Coll Cardiol 2012;60(8):756– 765. 11. Leone AM, De Caterina AR, Basile E, et al. Influence of the amount of myocardium subtended by a stenosis on fractional flow reserve. Circ Cardiovasc Interv 2013;6(1):29– 36. 12. Abbara S, Arbab-Zadeh A, Callister TQ, et al. SCCT guidelines for performance of coronary computed tomographic angiography: a report of the Society of Cardiovascular Computed Tomography Guidelines Committee. J Cardiovasc Comput Tomogr 2009;3(3):190–204. 13. Sun J, Zhang Z, Lu B, et al. Identification and quantification of coronary atherosclerotic plaques: a comparison of 64-MDCT and intravascular ultrasound. AJR Am J Roentgenol 2008;190(3):748–754. 14. Lee JT, Ideker RE, Reimer KA. Myocardial infarct size and location in relation to the coronary vascular bed at risk in man. Circulation 1981;64(3):526–534. 15. Brandt PW, Partridge JB, Wattie WJ. Coronary arteriography; method of presentation of the arteriogram report and a scoring system. Clin Radiol 1977;28(4):361–365. 16. Graham MM, Faris PD, Ghali WA, et al. Validation of three myocardial jeopardy scores in a population-based cardiac catheterization cohort. Am Heart J 2001;142(2):254– 261. 17. Ortiz-Pérez JT, Meyers SN, Lee DC, et al. Angiographic estimates of myocardium at risk during acute myocardial infarction: validation study using cardiac magnetic resonance imaging. Eur Heart J 2007;28(14):1750–1758.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
18. Moral S, Rodríguez-Palomares JF, Descalzo M, et al. Quantification of myocardial area at risk: validation of coronary angiographic scores with cardiovascular magnetic resonance methods. Rev Esp Cardiol (Engl Ed) 2012;65(11):1010–1017. 19. Kern MJ, Samady H. Current concepts of integrated coronary physiology in the catheterization laboratory. J Am Coll Cardiol 2010;55(3):173–185. 20. Tonino PA, De Bruyne B, Pijls NH, et al. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 2009;360(3):213–224. 21. Johnson NP, Tóth GG, Lai D, et al. Prognostic value of fractional flow reserve: linking physiologic severity to clinical outcomes. J Am Coll Cardiol 2014;64(16):1641–1654. 22. Pan W. Akaike’s information criterion in generalized estimating equations. Biometrics 2001;57(1):120–125. 23. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27(2):157–172; discussion 207–212. 24. Jaffe R, Halon DA, Roguin A, Rubinshtein R, Lewis BS. A Poiseuille-based coronary angiographic index for prediction of fractional flow reserve. Int J Cardiol 2013;167(3):862– 865. 25. Brosh D, Higano ST, Lennon RJ, Holmes DR Jr, Lerman A. Effect of lesion length on fractional flow reserve in intermediate coronary lesions. Am Heart J 2005;150(2):338– 343. 26. De Bruyne B, Pijls NH, Bartunek J, et al. Fractional flow reserve in patients with prior myocardial infarction. Circulation 2001;104(2):157–162. 27. Shaw LJ, Hausleiter J, Achenbach S, et al. Coronary computed tomographic angiography as a gatekeeper to invasive diagnostic and surgical procedures: results from the multicenter CONFIRM (Coronary CT Angiography Evaluation for Clinical Outcomes: an International Multicenter) registry. J Am Coll Cardiol 2012;60(20):2103–2114.
11
Original Research n Cardiac
Brian S. Ko, MBBS (Hons), PhD Dennis T. L. Wong, MBBS (Hons), PhD James D. Cameron, MBBS, MD Darryl P. Leong, MBBS (Hons), PhD Siang Soh, MBBS Nitesh Nerlekar, MBBS (Hons) Ian T. Meredith, MBBS (Hons), PhD Sujith K. Seneviratne, MBBS
Purpose:
To identify computed tomographic (CT) coronary indexes independently associated with a fractional flow reserve (FFR) of 0.8 or less, to derive a score that combines CT indexes most predictive of an FFR of 0.8 or less, and to evaluate the diagnostic accuracy of the score in predicting an FFR of 0.8 or less.
Materials and Methods:
This retrospective study had institutional review board approval and waiver of the need to obtain informed consent. Consecutive patients who underwent CT coronary angiography and FFR assessment with one or more discrete lesion(s) of intermediate (30%–70%) severity at CT were included. Quantitative CT measurements were performed by using dedicated software. The CT indexes evaluated included the following: plaque burden, minimal luminal area and diameter, stenosis diameter, area of stenosis, lesion length, remodeling index, plaque morphology, calcification severity, and the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) score, which approximates the size of the myocardium subtended by a lesion. By using covariates independently associated with an FFR of 0.8 or less, a score was determined on the basis of modified Akaike information criteria, and the C statistics of individual and combined indexes were compared.
Results:
Eighty-five patients (mean age, 64.2 years; range, 48–88 years; 65.9% men; 124 lesions; 38 lesions with an FFR 0.8) were included. Area of stenosis, lesion length, and APPROACH score were the strongest predictors of an FFR of 0.8 or less and were used to derive the ASLA score. The optimism-adjusted Harrell C statistic for the combined score was 0.82, which was superior to that for area of stenosis (0.74), lesion length (0.75), and the APPROACH score (0.71) (P , .001 for trend). The corresponding incremental discrimination improvement indexes were 0.17, 0.11, and 0.19, respectively (P , .001 for all), suggesting that the score improves reclassification compared with any one angiographic index. The average time required for score derivation was 102.6 seconds.
Conclusion:
The ASLA score, which accounts for CT-derived area of stenosis, lesion length, and APPROACH score, may conveniently improve the prediction, beyond individual indexes, of functionally significant intermediate coronary lesions.
1
From the Monash Cardiovascular Research Centre, Department of Medicine (Monash Medical Centre), Monash University and Monash Heart, Monash Health, 246 Clayton Road, Clayton, 3168 VIC, Australia (B.S.K., D.T.L.W., J.D.C., S.S., N.N., I.T.M., S.K.S.); Discipline of Medicine, University of Adelaide, Adelaide, Australia (D.T.L.W., D.P.L.); and Discipline of Medicine, Flinders University, Adelaide, Australia (D.P.L.). Received June 19, 2014; revision requested August 4; revision received November 11; accepted November 12; final version accepted December 10. B.S.K. and D.T.L.W. supported by the National Heart Foundation of Australia and the Robertson Family Scholarship. Address correspondence to B.S.K. (e-mail: [email protected]). RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Imaging
The ASLA Score: A CT Angiographic Index to Predict Functionally Significant Coronary Stenoses in Lesions with Intermediate Severity—Diagnostic Accuracy1
RSNA, 2015
q
Online supplemental material is available for this article. 1
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
C
omputed tomographic (CT) coronary angiography is recommended as an initial investigation to assess symptomatic patients with low to intermediate risk of coronary artery disease (1). In its current form, however, CT coronary angiography has limited specificity in determining the functional significance of coronary stenoses (2,3). For this reason, patients with moderate stenoses may require referral for noninvasive functional imaging to confirm the presence of ischemia before the decision is made to proceed to conventional angiography. A number of novel CT techniques have recently been described that may improve the ability of CT coronary angiography to predict ischemia. These include the use of adenosine stress CT myocardial perfusion imaging (4,5), the assessment of the transluminal attenuation gradient across coronary stenoses on the basis of resting CT coronary angiography images (6), and the noninvasive prediction of fractional flow reserve (FFR) by the application of computational fluid dynamics to resting CT coronary angiography images (3,7). Each technique has been demonstrated to improve the specificity, positive predictive value, and overall accuracy of CT coronary angiography in the prediction of functionally significant coronary
Advances in Knowledge nn Combined CT assessment of area of stenosis, lesion length, and size of the subtended myocardium with use of the ASLA score improves the prediction of functionally significant coronary stenosis beyond that of individual indexes alone (C statistics for combined assessment, area of stenosis, lesion length, and Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease score were 0.82, 0.74, 0.75, and 0.71 respectively; P , .001 for trend). nn The ASLA score can be conveniently applied in 102.6 seconds 6 37.5 (standard deviation). 2
Ko et al
stenoses as assessed by invasive FFR measurement (FFR 0.8) (3–7). Each novel technique in its current form, however, may have limitations. Stress perfusion imaging with CT requires additional imaging and hence is associated with increased radiation exposure and iodinated contrast material usage. CT-derived FFR can be computed currently only by the use of a supercomputer, which currently requires at least hours of processing time (3). While the assessment of transluminal attenuation gradient is promising and convenient, it requires manual vessel tracking with dedicated research software, and the technique remains unvalidated on images acquired by using narrow-detector CT scanners (6). Hence, there remains an unmet need for a simple and easily accessible approach for the CT angiographer to predict the functional significance of coronary stenoses—in particular, intermediate coronary stenoses identified at CT—to assist in clinical decisions regarding referral for conventional angiography. Results of recent studies (8,9) demonstrate that CT angiographic predictors of FFR may include minimal luminal diameter, area of stenosis, and lesion length—which by themselves remain insufficient predictors of an FFR of 0.8 or less. Moreover, the myocardium subtended by the lesion has also been demonstrated to strongly correlate with ischemic burden as assessed with cardiac magnetic resonance (MR) imaging and invasive FFR measurement (10,11), although data on the use of these scores with CT coronary angiography remain unknown. We hypothesize that the combined use of angiographic indexes may improve prediction of reduced FFR compared with the use of individual indexes alone. Our study was therefore performed to identify CT indexes independently associated with an FFR of 0.8 or
Implication for Patient Care nn Use of this score may decrease the need for further functional testing and conventional angiography.
less, to derive a score that combines CT indexes most predictive of an FFR of 0.8 or less, and to evaluate the diagnostic accuracy of the score in predicting an FFR of 0.8 or less.
Materials and Methods This retrospective study had approval from the institutional review board of Monash Health, with waiver of the need to obtain informed consent. The authors who have no relationships with industry had control of inclusion of all data and information in this article. We included consecutive patients who underwent clinically mandated CT coronary angiography and nonurgent conventional coronary angiography with FFR assessment performed in at least one discrete lesion of intermediate severity (30%–70%) as visually assessed at CT coronary angiography in our institution, Monash Heart, between October 2009 and July 2012. Conventional angiography and FFR measurement were performed a mean of 11 days after CT coronary angiography. Patients were excluded in cases of left ventricular dysfunction, if the time between CT coronary angiography and FFR measurement was longer than 6 Published online before print 10.1148/radiol.15141231 Content codes: Radiology 2015; 000:1–11 Abbreviations: APPROACH = Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease ASLA = area of stenosis, lesion length, and APPROACH score CI = confidence interval FFR = fractional flow reserve Author contributions: Guarantors of integrity of entire study, B.S.K., N.N., S.K.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, B.S.K., D.T.L.W., I.T.M.; clinical studies, B.S.K., S.S., N.N., I.T.M., S.K.S.; experimental studies, B.S.K., I.T.M.; statistical analysis, B.S.K., J.D.C., D.P.L., I.T.M.; and manuscript editing, B.S.K., D.T.L.W., J.D.C., D.P.L., I.T.M., S.K.S. Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
months, if there were adverse cardiac events or revascularization during this time interval, if the acute coronary syndrome was present in the 3 months prior to the CT examination, and if they had bypass graft lesions or significant left main artery stenoses. Among the patients who met the inclusion criteria, those with uninterpretable vessels at CT and those with stenosis greater than 70% or less than 30%, poor CT image quality, small vessel diameter (,2 mm), intracoronary stents, excessive calcifications, and/or myocardial bridges were excluded.
CT Coronary Angiography Patients underwent CT coronary angiography performed by using a 320–detector row CT scanner (Aquilion ONE; Toshiba Medical Systems, Nasu, Japan). β-Blockers and nitroglycerin were administered to achieve a prescanning heart rate of less than 65 beats per minute and to optimize vasodilation in accordance with previously published guidelines (12). A bolus of 55 mL of 100% iohexol 56.6 g/75 mL (Omnipaque 350; GE Healthcare, Princeton, NJ) was injected into an antecubital vein at a flow rate of 5 mL/sec, followed by 20 mL of a 30:70 mixture of contrast material and saline, followed by 30 mL of saline. Scanning parameters were as follows: detector collimation, 320 3 0.5 mm; tube current, 300–500 mA (depending on body mass index); tube voltage, 120 kV; gantry rotation time, 350 msec; and temporal resolution, 175 msec. Prospective electrocardiographic gating was used, covering phases 70%– 80% of the R-R interval. For images acquired at heart rates of 65 beats per minute or slower, scanning was completed with a single R-R interval utilizing a 180° segment. In patients with a heart rate greater than 65 beats per minute, data segments from two consecutive beats were used for multisegment reconstruction with an improved temporal resolution of 87 msec. Analysis of CT Coronary Angiograms Analysis was performed in coronary vessels that were greater than 2 mm in diameter. All data were transferred
to an external workstation (Vitrea 6, version 3.0; Vital Images, Minnetonka, Minn) for further analysis. Two angiographers with 5 years of experience in CT coronary angiography interpretation (B.S.K. and D.T.L.W.), who were blinded to the results of corresponding conventional angiography and FFR measurement, performed the analysis independently. All included vessels were deemed to be suitable for analysis of the coronary artery and plaque characteristics by the CT interpreters. Coronary plaques were classified as noncalcified, calcified, or mixed. Plaques with CT attenuation lower than that of the luminal contrast material were defined as noncalcified, and plaques with high CT attenuation that could be visualized separately from the contrast material in the lumen were defined as calcified. Mixed plaque was defined as the presence of calcified and noncalcified plaque components within the same lesion. The extent of coronary calcium was categorized as none, as less than 50% of the cross-sectional area, or as 50% or greater of the cross-sectional area. Quantification of luminal narrowing and vessel remodeling in the target lesion was performed by using a dedicated software tool (Sure Plaque, Vitrea 6, version 3.0; Vital Images and Toshiba Medical Systems) that has been previously validated against intravascular ultrasonography (13). Automatic vessel tracking was then used to locate the vessel centerline on the basis of the opacification of the lumen. After manual determination of the coronary lesion, the minimal lumen diameter (in millimeters), minimal lumen area (in square millimeters), and lesion length (in millimeters) were computed by the software. The reference diameter and area were determined as an average of the vessel dimensions immediately proximal and distal to the lesion, where minimal or no plaque could be detected. In ostial lesions, only one reference was used. The diameter of stenosis, as a percentage, was defined as 100% · (reference diameter 2 minimal lumen diameter)/ reference diameter. The area of stenosis, as a percentage, was defined as
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
100% · (reference lumen 2 minimal lumen area/reference lumen). Coronary plaque was defined as the area between the outer contour of the vessel and the lumen border, both of which were determined by the software (8). Measurements were obtained for each plaque and were defined as follows: plaque burden (in square millimeters) = (vessel area 2 lumen area)/vessel area, and remodeling index = outer vessel area (stenotic segment)/outer vessel area (reference segment).
Modified APPROACH Score The Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) score is based on the division of the left ventricle into regions in accordance with pathologic studies in humans and with correlation with cardiac catheterization results, which evaluate the proportion of the myocardium perfused by each artery (14–16). Table E1 (online) demonstrates the derivation of the APPROACH score, as has been previously described (17). Furthermore, it has been validated with cardiac MR imaging methods to quantify the myocardial area at risk after myocardial infarction (18). The modified score takes into account the location of the lesion (proximal, middle, or distal) and the dominance and size of the secondary branches and provides an estimate of the percentage of supplied myocardium beyond the considered coronary lesion (17). Conventional Coronary Angiography and FFR Measurement Conventional coronary angiography was performed as per standard clinical practice. The pressure wire (Pressure Wire Certus; St Jude Medical, St Paul, Minn) was calibrated and electronically equalized with the aortic pressure before being placed distal to the stenosis in the distal third of the coronary artery being interrogated. Intracoronary glyceryl trinitrate (100 µg) was injected to minimize vasospasm. Intravenous adenosine was administered at 140 µg/ kg/min through an intravenous line in the antecubital fossa. At steady-state 3
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
hyperemia, FFR was assessed by using the RadiAnalyser Xpress (St Jude Medical) and was calculated by dividing the mean coronary pressure measured with the pressure sensor placed distal to the stenosis by the mean aortic pressure measured through the guide catheter. An FFR of 0.8 or less was considered to define ischemia in the interrogated artery and its supplied territory (19,20). This threshold was chosen because it may determine vessel outcome and benefit from revascularization (21).
Statistical Analysis Continuous variables are expressed as means 6 standard deviations or with 95% confidence intervals (CIs). Categoric variables are expressed as percentages. Interobserver variability was assessed by using the intraclass coefficient. To identify the optimal combination of CT coronary angiography parameters for the estimation of FFR, linear mixedeffects modeling was performed. Patient identity was included as a random effect to account for the likelihood of greater correlation between different vessels in a particular patient than between vessels from different patients. To identify CT angiographic covariates that were independently associated with an FFR of 0.8 or less, we used generalized estimating equations. The following covariates were considered at a univariate level: plaque burden, minimal lumen area, minimal lumen diameter, diameter of stenosis, area of stenosis, lesion length, remodeling index, calcification, plaque morphology, and APPROACH score. Those covariates with a univariate P , .2 were included in the multivariate analysis, which was performed by using the “enter” approach. To ensure the most parsimonious model, the modified Akaike information criterion was used (22); models with a lower Akaike criterion are preferable to models with a higher Akaike criterion. The ability of the final model to predict an FFR of 0.8 or less was assessed by the optimism-adjusted Harrell C statistic on bootstrapped samples with 100 times replacement. The optimal score threshold is one that maximizes the sum of sensitivity and specificity with a 4
Ko et al
Figure 1
Figure 1: Patient (pt) and vessel flow diagram. ∗ = Numbers in parentheses are numbers of vessels rather than numbers of patients.
minimal sensitivity of 75%. Finally, the incremental utility of the score was assessed by using the integrated discrimination improvement index described by Pencina et al (23) as follows: integrated discrimination improvement = (ISnew 2 ISold) 2 (IPnew 2 IPold), where “new” refers to a model containing a novel diagnostic tool of interest in addition to conventional risk predictors, “old” pertains to the model containing only the conventional risk markers, and IS and IP are the integrals of sensitivity and (1 2 specificity), respectively. Statistical analysis was performed with SPSS 18.0 (SPSS, Chicago, Ill) and STATA 12.1 (Stata, College Station, Tex). P , .05 was considered to indicate a statistically significant difference.
Results There were 121 patients who underwent CT and FFR assessment in our institution during the study period
(Fig 1). Ten patients were excluded for the presence of left ventricular dysfunction (n = 4), a time between CT coronary angiography and FFR measurement of more than 6 months (n = 2), presence of acute coronary syndrome in the 3 months prior to CT coronary angiography (n = 1), presence of cardiac events or revascularization between CT angiography and FFR measurement (n = 1), presence of coronary artery bypass graft (n = 1), and left main disease (n = 1). Among the remaining 111 patients and 204 vessels, 80 vessels were excluded from analysis because of the presence of severe (.70%) or minor stenoses (n = 34), the presence of poor image quality as a result of excessive noise or motion (n = 17), vessel diameter of less than 2 mm (n = 11), intracoronary stent (n = 7), excessive calcification (n = 6), myocardial bridging (n = 2), or incomplete information (n = 3). The remaining 85 patients (mean age, 64.2 years 6 11.2; 66% men) with
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Table 1 Patient and Vessel Characteristics Characteristics Patient characteristics (n = 85) Age (y)* Age in male cohort (y)† Age in female cohort (y)† P value for difference in age between male and female cohorts Sex No. of men No. of women Body mass index (kg/m2)* Symptom Typical angina Atypical angina or noncardiac pain Suspected coronary artery disease Known coronary artery disease Previous myocardial infarction Previous percutaneous coronary intervention Cardiovascular risk factors Diabetes Hypertension Hypercholesterolemia Current smoker Family history of ischemic heart disease No. of vessels with a 50% stenosis as visually assessed at CT coronary angiography 0 1 2 3 FFR findings No. of vessels interrogated with FFR measurement 1 2 3 No. of vessels interrogated per patient* Vessels with FFR 0.8 per patient 0 1 2 No. of vessels with FFR 0.8 per patient* Lesion characteristics (n = 124) 50% diameter stenosis as visually assessed at CT coronary angiography Quantitative CT findings* Diameter stenosis (%) Area of stenosis (%) Minimal luminal diameter (mm) Minimal luminal area (mm2) Lesion length (mm) Plaque burden APPROACH score
Datum 64.2 6 11.2 63.9 (48–82) 64.7 (51–88) .76 56 (65.9) 29 (34.1) 28.1 6 5.1 65 (76.5) 20 (23.5) 49 (57.6) 36 (42.4) 9 (9.5) 12 (12.7) 13 (15.3) 63 (74.1) 63 (74.1) 13 (15.3) 34 (40.0)
36 (42.4) 42 (29.4) 5 (5.9) 2 (2.4)
55 (64.7) 21 (24.8) 9 (10.6) 1.6 6 0.8 48 (56.5) 35 (41.2) 2 (2.4) 0.46 6 0.55 58 (46.8) 39.4 6 20.9 47.8 6 24.0 1.8 6 0.8 5.0 6 3.5 19.5 6 11.4 71.5 6 15.6 26.5 6 11.9
Table 1 (continues)
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
124 vessels were included for analysis. Patient and vessel characteristics are summarized in Table 1. FFR measurement was performed in all 85 patients, of whom 55 (64.7%) underwent FFR interrogation in a single vessel; 21 (24.8%), in two vessels; and nine (10.6%), in three vessels. Overall, FFR ranged from 0.39 to 1.0 (mean, 0.84 6 0.11). Thirty-eight vessels (30.6%) were classified as having functionally significant stenosis, with an FFR of 0.8 or less. There were no adverse events during FFR interrogation.
Accuracy of Visual Assessment at CT in Predicting an FFR of 0.8 or Less On a per-vessel basis, 58 (48.6%) vessels had visual stenosis of 50% or greater at CT. The presence of a visual stenosis of 50% or greater at CT angiography was associated with a 78.9% sensitivity, 67.4% specificity, 51.7% positive predictive value, and 87.9% negative predictive value. The bootstrapped Harrell C statistic for visual stenosis in the diagnosis of an FFR of 0.8 or less was 0.74 (95% CI: 0.64, 0.83). Accuracy of Quantitative CT Parameters in Predicting an FFR of 0.8 or Less The results of univariate and multivariate analyses for predictors of FFR are presented in Table 2. According to univariate analysis, minimal lumen area, minimal lumen diameter, plaque burden, diameter of stenosis, area of stenosis, lesion length, remodeling index, presence of calcification, plaque morphology, and APPROACH score were all significant predictors of an FFR of 0.8 or less. According to results of multivariate analysis, plaque burden, area of stenosis, lesion length, and the APPROACH score remained significant predictors for lesion-specific ischemia. Accuracy of area of stenosis quantified at CT in predicting an FFR of 0.8 or less.—On a per-vessel basis, 56 (45.2%) vessels had an area of stenosis of 50% or greater as quantified at CT. The mean area of stenosis in lesions with an FFR of 0.8 or less was 61.5% 6 22.5, versus 41.8% 6 22.3 in vessels with an FFR greater than 0.8 (P , 5
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
mean plaque burden in lesions with an FFR of 0.8 or less was 69.3 6 14.5, versus 76.5 6 17.1 in vessels with an FFR greater than 0.8 (P , .0001). There was a statistically significant negative correlation between plaque burden and FFR (r = 20.27, P = .003) (Fig 2). The bootstrapped Harrell C statistic for plaque burden in the prediction of an FFR of 0.8 or less was 0.64 (95% CI: 0.54, 0.76).
Table 1 (continued) Patient and Vessel Characteristics Characteristics
Datum
Quantitative coronary angiography findings* Diameter stenosis (%) Minimal luminal diameter (mm) Vessel diameter (mm) FFR findings Mean FFR in all lesions* Lesions with FFR 0.8 Mean FFR in lesions with FFR 0.8* Lesion location Proximal right coronary artery Middle right coronary artery Distal right coronary artery Right posterior descending artery or posterolateral ventricular branch Proximal left anterior descending artery Middle left anterior descending artery Distal left anterior descending artery Diagonal artery Proximal left circumflex artery Ramus intermediate Obtuse marginal arteries Distal left circumflex artery
37.4 6 17.8 1.6 6 0.7 2.5 6 0.7 0.84 6 0.11 38 (30.6) 0.71 6 0.09 15 (12.1) 8 (6.5) 2 (1.6) 3 (2.4) 31 (25.0) 23 (18.5) 4 (3.2) 3 (2.4) 24 (20.2) 1 (0.8) 7 (5.6) 2 (1.6)
Note.—Unless otherwise specified, data are numbers of patients or lesions, with percentages in parentheses. * Data are means 6 standard deviations. †
Data are means, with ranges in parentheses.
.0001). Area of stenosis measurements were reproducible, with an interobserver intraclass correlation coefficient of 0.98 (95% CI: 0.91, 0.99). There was a statistically significant negative correlation between area of stenosis and FFR (r = 20.37, P , .0001) (Fig 2). The presence of an area of stenosis of 50% or greater was associated with 65.9% sensitivity, 62.1% specificity, 44.6% positive predictive value, and 79.7% negative predictive value. The bootstrapped Harrell C statistic for area of stenosis in the prediction of an FFR of 0.8 or less was 0.74 (95% CI: 0.65, 0.83). Accuracy of lesion length in predicting an FFR of 0.8 or less.—The mean lesion length in vessels with an FFR of 0.8 or less was 27.2 mm 6 12.9, versus 16.1 mm 6 8.9 in vessels with an FFR of greater than 0.8 (P , .0001). There was a statistically significant negative correlation between lesion length and FFR (r = 20.45, P , .0001) (Fig 6
2). The bootstrapped Harrell C statistic for lesion length in the prediction of an FFR of 0.8 or less was 0.75 (95% CI: 0.66, 0.85). The lesion length measurements were highly reproducible, with an interobserver intraclass correlation coefficient of 0.95 (95% CI: 0.83, 0.99). Accuracy of APPROACH score in predicting an FFR of 0.8 or less.—The mean APPROACH score in lesions with an FFR of 0.8 or less was 32.7 6 11.6, versus 23.7 6 11.0 in vessels with an FFR of greater than 0.8 (P , .0001). There was a statistically significant negative correlation between APPROACH score and FFR (r = 20.34, P = .0002) (Fig 2). The bootstrapped Harrell C statistic for the APPROACH score in the prediction of an FFR of 0.8 or less was 0.71 (95% CI: 0.61, 0.81). The APPROACH score demonstrated high reproducibility, with an interobserver intraclass correlation coefficient of 1.0 (95% CI: 1.0, 1.0). Accuracy of plaque burden in predicting an FFR of 0.8 or less.—The
ASLA SCORE The combined score was derived by means of multivariate generalized estimating equation regression. It was named the ASLA score, as it included CT-quantified area of stenosis and lesion length and the APPROACH score as ordinal variables categorized by using the cut points described in Table 3. Each was important for the model, as indicated by a likelihood ratio test P value of less than .02 when each parameter was dropped sequentially from the model. Plaque burden was not taken into account in the score because it did not lower the model’s modified Akaike criterion substantially and so was thought not to contribute to the model’s clinical performance. An exchangeable correlation structure was selected for the regression, as this minimized the modified Akaike criterion compared with other correlation structures. Figure 3 illustrates the application of the ASLA score. Accuracy of ASLA score in predicting an FFR of 0.8 or less.—The mean ASLA score in vessels with an FFR of 0.8 or less was 10.1 6 4.77, compared with 4.43 6 3.62 in vessels with an FFR greater than 0.8 (P , .001) (Fig 3). The bootstrapped Harrell C statistic of the ASLA score in predicting significant FFR was 0.82 (95% CI: 0.74, 0.91). The ASLA score was a better discriminator of an FFR of 0.8 or less than area of stenosis, lesion length, and APPROACH score (respective C statistics, 0.82 [95% CI: 0.74, 0.91], 0.74 [95% CI: 0.65, 0.83], 0.75 [95% CI: 0.66, 0.85], and 0.71 [95% CI: 0.61, 0.81]; P , .001 for trend). The comparison of areas under the receiver operating characteristic curve is presented in Figure 4. A representation of sensitivity
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
Table 2 Univariate and Multivariate Predictors of an FFR of 0.8 or Less Predictor Plaque burden Minimal luminal area Minimal luminal diameter Diameter of stenosis Area of stenosis Lesion length Remodeling index Calcification Plaque morphology APPROACH score
Univariate Coefficient
P Value
Multivariate Coefficient
P Value
0.033 (0.0050, 0.061) 20.22 (20.67, 20.070) 20.81 (21.4, 20.25) 0.029 (0.0093, 0.049) 0.038 (0.019, 0.056) 0.090 (0.051, 0.13) 20.0068 (20.018, 0.0044) 0.59 (0.023, 1.2) 0.31 (20.16, 0.77) 0.069 (0.035, 0.10)
.02 .004 .004 .004 ,.001 ,.001 .2 .04 .2 ,.001
20.071 (20.13, 20.013) 20.065 (20.37, 0.24) 21.2 (22.9, 0.41) 20.044 (20.10, 0.014) 0.067 (0.015, 0.12) 0.091 (0.036, 0.15) ... 20.068 (20.85, 0.71)
.02 .7 .1 .1 .01 .001 ... .9
0.053 (0059, 0.10)
.03
Note.—Data in parentheses are 95% CIs.
Figure 2
Figure 2: Scatterplots of FFR and CT angiographic parameters, including, A, area of stenosis, B, lesion length, C, APPROACH score, and, D, plaque burden. There was a statistically significant inverse relationship between CT-derived area of stenosis and FFR (R = 20.37, P , .0001), lesion length (R = 20.45, P , .0001), APPROACH score (R = 20.34, P = .0002), and plaque burden (R = 20.26, P = .003).
and specificity as functional cutoff values over a spectrum of ASLA scores is presented in Figure 5. A threshold score of 7 provided 76.3% (95% CI: 59.4%, 88.0%) sensitivity, 76.7% (95% CI:
66.2%, 85.0%) specificity, 59.2% (95% CI: 44.3%, 72.7%) positive predictive value, 88.0% (95% CI: 78.0%, 94.0%) negative predictive value, and 76.6% overall accuracy. A score of 3 or lower
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
was associated with 92% sensitivity and accounted for 50 (40.3%) lesions. Similarly, a score of 11 or higher was associated with 86% specificity and accounted for 21 lesions (16.9%) (Fig 6). 7
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Among the 50 lesions with a score of 3 or lower, four (8%) had an FFR of 0.8 or less (range, 0.68–0.78), including two lesions in the left anterior descending artery, one in a marginal branch, and one in the right coronary artery. Among the 21 lesions with a score of 11 or greater, four (19%) had an FFR of greater than 0.8 (range, 0.81–0.95),
Table 3 The ASLA Score Variable and Value
Points Assigned
Area of stenosis (%) .63 47–63 31–46 ,31 Lesion length (mm) .28 10.8–28 ,10.8 APPROACH score .44 25.1–44 18–25 ,18 Total
7 2 1 0 6 1 0 5 2 1 0 Maximum = 18
Ko et al
including three lesions in the left anterior descending artery and one in the left circumflex artery. We estimated the integrated discrimination improvement indexes for the score over and above each of its constituents (area of stenosis, lesion length, and APPROACH score) by using the approach of Pencina et al (23) (Table 4). The integrated discrimination improvement indexes were 0.17 (P , .001), 0.11 (P , .001), and 0.19 (P , .001), suggesting that the score improves reclassification of estimated FFR compared with any one angiographic index. Time required to perform ASLA scoring.—The average time required to perform ASLA scoring per lesion was 102.6 seconds 6 37.5 (25th percentile, 70.1 seconds; 75th percentile, 131.8 seconds).
Discussion This study derived a score on the basis of information quantified at CT to predict the hemodynamic significance of coronary artery stenoses. Our results demonstrate that the ASLA score, which takes into account area of stenosis, lesion length, and the APPROACH
score, is a superior predictor of functionally significant stenoses as assessed by using invasive FFR measurement, beyond individual CT indexes alone or visually assessed stenosis of 50% or greater at CT angiography. The ASLA score can be conveniently applied in less than 2 minutes. On the basis of multivariate analysis, area of stenosis and lesion length were found to be important predictors of an FFR of 0.8 or less. Our findings are in agreement with those of Kristensen et al (8), who found, using a similar CT-based method in a smaller cohort of 42 patients with 56 intermediate stenoses, that area of stenosis and lesion length were the strongest determinants of an abnormal FFR and were superior to plaque volume, minimal luminal diameter, and minimal luminal area. The incremental value of lesion length when added to standard angiographic measurement of diameter of stenosis in the prediction of FFR significant lesions has been reported (24,25). However, as with the results of the previous studies, our results highlight the limited predictive value of individual CT-derived parameters for an FFR of 0.8 or less (8,9).
Figure 3
Figure 3: A–D, Curved images from quantitative CT examination performed in a patient with an intermediate (50%–70%) lesion in the left circumflex artery. The lesion length was 12.4 cm (on B ), the area of stenosis was 77% on axial images (in C ), and the APPROACH score was 24.5 (on D ). The ASLA score was 9 (1 for lesion length plus 7 for area of stenosis plus 1 for lesion length). The FFR in the left circumflex artery was 0.76. 8
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Figure 4
Figure 4: Receiver operating characteristic curves of ASLA score, APPROACH score, area of stenosis, and lesion length.
Figure 5
Figure 5: Graph shows sensitivity and specificity as functional cutoff values over a spectrum of ASLA scores.
It is increasingly recognized that FFR is highly dependent on the area of myocardium supplied beyond the lesion in question (11,26). The APPROACH score has been used to estimate the myocardium at risk during myocardial infarction and was found to be inversely correlated with FFR (11). This study describes the correlation of a CT-derived APPROACH score with FFR. Our study results demonstrate modest but similar correlation (r = 20.34, P , .0001) with FFR as had been described for conventional angiography (11). We also demonstrated that the score is highly reproducible, with minimal interobserver variability, which is in agreement with previous reports (11,18).
The ASLA score, which takes into account area of stenosis, lesion length, and APPROACH score, was found to be superior and incremental to individual indexes alone or visual CT assessment in the prediction of an FFR of 0.8 or less. Visually assessed diameter of stenosis at CT currently plays an important role in the decision as to whether a patient may require further examination with noninvasive stress testing or invasive angiography (27). Rossi et al (9) demonstrated that area of stenosis as derived at quantitative CT may be superior in specificity to visually assessed diameter of stenosis in a cohort of 99 symptomatic patients. Our results extend these results and emphasize the importance of accounting for
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
the myocardium supplied distal to the lesion and the length of the lesion, in addition to the area of stenosis. Taking into account all three in the form of a score may improve diagnostic accuracy in the prediction of an FFR of 0.8 or less. The Poiseuille equation states that the pressure gradient across a coronary lesion is directly proportional to the coronary blood flow, blood viscosity, and lesion length and is inversely proportional to the 4th power of the radius. It would be intuitive for a score to take into account each of these key elements. Accordingly, the ASLA score includes the APPROACH score, which may act as a surrogate for myocardial blood flow (11); minimal area of stenosis, which represents the square of the radius of the vessel at the lesion; and lesion length. The ASLA score can be derived by using standard rest CT angiography images, without additional radiation exposure, contrast material, medications, reconstruction, or cost. The score can also be readily calculated with software available on most currently available reporting workstations during CT angiography interpretation. The technique is convenient and requires a short processing time. In our cohort, 57% of intermediate lesions had scores that were 3 or lower or that were 11 or higher. In the former, 92% were not FFR significant and in the latter, 81% were FFR significant. This suggests that the score may have value in differentiating lesions that may or may not require further functional assessment and/or conventional angiography. In contrast, in lesions with scores between 4 and 10, our results suggest that it will still be important to perform functional assessment or FFR measurement. Notably, given the sample size and the number of false-positive and falsenegative results at the extremes of the ASLA score, our results represent a preliminary report on the described technique. Larger studies with methodologic improvements are required to improve the diagnostic performance of the technique and to assess whether 9
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
Ko et al
in this study, as all patients, including the nine patients who had previously experienced myocardial infarction, had documented normal left ventricular ejection fraction at transthoracic echocardiography or conventional angiography, as per the inclusion criteria. However, the applicability of the study findings to patients with a sizeable loss of viable myocardial mass has not been studied. Last, we acknowledge that measurements of area of stenosis and lesion length in this study were quantified by using dedicated software provided by a sole vendor. The applicability of the technique using equipment from other vendors remains unknown. In conclusion, combined CT angiographic assessment with the ASLA score, which includes area of stenosis, lesion length, and APPROACH score, provides incremental predictive value over individual indexes alone for the detection of functionally significant coronary artery stenoses.
Figure 6
Figure 6: Scatterplot of FFR and ASLA score. A total of 124 lesions were assessed. In the 50 lesions with an ASLA score of 3 or lower, four (8%) had an FFR of 0.8 or less (range, 0.68–0.78), including two lesions in the left anterior descending artery, one in a marginal branch, and one in the right coronary artery. In the 21 lesions with an ASLA score of 11 or higher, four (19%) had an FFR greater than 0.8 (range, 0.81–0.95), including three lesions in the left anterior descending artery and one in the left circumflex artery.
Table 4 Lesion Characteristics Quantified according to CT and Interobserver Variability Characteristic Area of stenosis (%) Lesion length (mm) APPROACH score
Mean 6 Standard Deviation
Interobserver Intraclass Correlation Coefficient
47.8 6 24.0 19.5 6 11.4 26.5 6 11.9
0.977 (0.91, 0.99) 0.954 (0.824, 0.988) 1.0 (1.0, 1.0)
Note.—Data in parentheses are 95% CIs.
the use of the score may save costs and decrease downstream testing. There were limitations to our study. Our results represent a retrospective single-center experience involving 85 patients and hence require confirmation with larger studies. The findings are limited to intermediate stenoses alone, and the applicability of the score to vessels with tandem lesions or severe stenoses, in bypass graft lesions, and in patients with impaired left ventricular function is not known. Analysis was restricted to a pervessel basis alone. In a proportion of patients who underwent multivessel FFR interrogation, only one vessel 10
was included in the final analysis, as the remaining vessels met exclusion criteria. For this reason, the accuracy of the applied technique cannot be extended to the patient per se, but remains true for the vessel. Hence, perpatient analysis was not performed. The incremental value of the technique on pretest information, including symptom severity and results of functional testing, has also not been evaluated. We acknowledge that in patients with prior myocardial infarction, FFR may be elevated because of the loss of viable myocardial mass (26). The effect of this may have been minimized
Disclosures of Conflicts of Interest: B.S.K. disclosed no relevant relationships. D.T.L.W. disclosed no relevant relationships. J.D.C. disclosed no relevant relationships. D.P.L. disclosed no relevant relationships. S.S. disclosed no relevant relationships. N.N. disclosed no relevant relationships. I.T.M. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is a consultant for Boston Scientific and Medtronic. Other relationships: disclosed no relevant relationships. S.K.S. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is on the speakers bureau of Toshiba. Other relationhips: disclosed no relevant relationships.
References 1. Taylor AJ, Cerqueira M, Hodgson JM, et al. ACCF/SCCT/ACR/AHA/ASE/ASNC/NASCI/ SCAI/SCMR 2010 Appropriate Use Criteria for Cardiac Computed Tomography. A Report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, the Society of Cardiovascular Computed Tomography, the American College of Radiology, the American Heart Association, the American Society of Echocardiography, the American Society of Nuclear Cardiology, the North American Society for Cardiovascular Imaging, the Society for Cardiovascular Angiography and Interventions,
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
CARDIAC IMAGING: Diagnostic Accuracy of ASLA Score in Stenoses with Intermediate Severity
and the Society for Cardiovascular Magnetic Resonance. Circulation 2010;122(21):e525– e555. 2. Dorbala S, Hachamovitch R, Di Carli MF. Myocardial perfusion imaging and multidetector computed tomographic coronary angiography: appropriate for all patients with suspected coronary artery disease? J Am Coll Cardiol 2006;48(12):2515–2517. 3. Koo BK, Erglis A, Doh JH, et al. Diagnosis of ischemia-causing coronary stenoses by noninvasive fractional flow reserve computed from coronary computed tomographic angiograms: results from the prospective multicenter DISCOVER-FLOW (Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve) study. J Am Coll Cardiol 2011;58(19):1989– 1997. 4. Ko BS, Cameron JD, Leung M, et al. Combined CT coronary angiography and stress myocardial perfusion imaging for hemodynamically significant stenoses in patients with suspected coronary artery disease: a comparison with fractional flow reserve. JACC Cardiovasc Imaging 2012;5(11):1097– 1111. 5. Bettencourt N, Chiribiri A, Schuster A, et al. Direct comparison of cardiac magnetic resonance and multidetector computed tomography stress-rest perfusion imaging for detection of coronary artery disease. J Am Coll Cardiol 2013;61(10):1099–1107. 6. Wong DT, Ko BS, Cameron JD, et al. Transluminal attenuation gradient in coronary computed tomography angiography is a novel noninvasive approach to the identification of functionally significant coronary artery stenosis: a comparison with fractional flow reserve. J Am Coll Cardiol 2013;61(12):1271–1279. 7. Min JK, Leipsic J, Pencina MJ, et al. Diagnostic accuracy of fractional flow reserve from anatomic CT angiography. JAMA 2012;308(12):1237–1245. 8. Kristensen TS, Engstrøm T, Kelbæk H, von der Recke P, Nielsen MB, Kofoed KF. Correlation between coronary computed tomographic angiography and fractional
flow reserve. Int J Cardiol 2010;144(2):200– 205. 9. Rossi A, Papadopoulou SL, Pugliese F, et al. Quantitative computed tomographic coronary angiography: does it predict functionally significant coronary stenoses? Circ Cardiovasc Imaging 2014;7(1):43–51. 10. Jogiya R, Kozerke S, Morton G, et al. Validation of dynamic 3-dimensional whole heart magnetic resonance myocardial perfusion imaging against fractional flow reserve for the detection of significant coronary artery disease. J Am Coll Cardiol 2012;60(8):756– 765. 11. Leone AM, De Caterina AR, Basile E, et al. Influence of the amount of myocardium subtended by a stenosis on fractional flow reserve. Circ Cardiovasc Interv 2013;6(1):29– 36. 12. Abbara S, Arbab-Zadeh A, Callister TQ, et al. SCCT guidelines for performance of coronary computed tomographic angiography: a report of the Society of Cardiovascular Computed Tomography Guidelines Committee. J Cardiovasc Comput Tomogr 2009;3(3):190–204. 13. Sun J, Zhang Z, Lu B, et al. Identification and quantification of coronary atherosclerotic plaques: a comparison of 64-MDCT and intravascular ultrasound. AJR Am J Roentgenol 2008;190(3):748–754. 14. Lee JT, Ideker RE, Reimer KA. Myocardial infarct size and location in relation to the coronary vascular bed at risk in man. Circulation 1981;64(3):526–534. 15. Brandt PW, Partridge JB, Wattie WJ. Coronary arteriography; method of presentation of the arteriogram report and a scoring system. Clin Radiol 1977;28(4):361–365. 16. Graham MM, Faris PD, Ghali WA, et al. Validation of three myocardial jeopardy scores in a population-based cardiac catheterization cohort. Am Heart J 2001;142(2):254– 261. 17. Ortiz-Pérez JT, Meyers SN, Lee DC, et al. Angiographic estimates of myocardium at risk during acute myocardial infarction: validation study using cardiac magnetic resonance imaging. Eur Heart J 2007;28(14):1750–1758.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Ko et al
18. Moral S, Rodríguez-Palomares JF, Descalzo M, et al. Quantification of myocardial area at risk: validation of coronary angiographic scores with cardiovascular magnetic resonance methods. Rev Esp Cardiol (Engl Ed) 2012;65(11):1010–1017. 19. Kern MJ, Samady H. Current concepts of integrated coronary physiology in the catheterization laboratory. J Am Coll Cardiol 2010;55(3):173–185. 20. Tonino PA, De Bruyne B, Pijls NH, et al. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 2009;360(3):213–224. 21. Johnson NP, Tóth GG, Lai D, et al. Prognostic value of fractional flow reserve: linking physiologic severity to clinical outcomes. J Am Coll Cardiol 2014;64(16):1641–1654. 22. Pan W. Akaike’s information criterion in generalized estimating equations. Biometrics 2001;57(1):120–125. 23. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27(2):157–172; discussion 207–212. 24. Jaffe R, Halon DA, Roguin A, Rubinshtein R, Lewis BS. A Poiseuille-based coronary angiographic index for prediction of fractional flow reserve. Int J Cardiol 2013;167(3):862– 865. 25. Brosh D, Higano ST, Lennon RJ, Holmes DR Jr, Lerman A. Effect of lesion length on fractional flow reserve in intermediate coronary lesions. Am Heart J 2005;150(2):338– 343. 26. De Bruyne B, Pijls NH, Bartunek J, et al. Fractional flow reserve in patients with prior myocardial infarction. Circulation 2001;104(2):157–162. 27. Shaw LJ, Hausleiter J, Achenbach S, et al. Coronary computed tomographic angiography as a gatekeeper to invasive diagnostic and surgical procedures: results from the multicenter CONFIRM (Coronary CT Angiography Evaluation for Clinical Outcomes: an International Multicenter) registry. J Am Coll Cardiol 2012;60(20):2103–2114.
11
