International Journal of Cardiology

Transcription

1 International Journal of Cardiology 168 (2013) Contents lists available at ScienceDirect International Journal of Cardiology journal homepage: Diagnostic accuracy of quantitative angiographic and intravascular ultrasound parameters predicting the functional significance of single de novo lesions Shao-Liang Chen a,,boxu b, Jack B. Chen c, Tian Xu a, Fei Ye a, Jun-Jie Zhang a, Tak W. Kwan d, Nai-Liang Tian a, Zhi-Zhong Liu a, Song Lin a a Nanjing First Hospital, Nanjing Medical University, Nanjing, China b Beijing Fuwai Cardiovascular Hospital, Beijing, China c Saint Joseph's Heart and Vascular Institute, Atlanta, Georgia d Beth Israel Hospital, New York, NY, USA article info abstract Article history: Received 9 January 2012 Received in revised form 19 September 2012 Accepted 5 December 2012 Available online 26 December 2012 Keywords: Quantitative coronary analysis Intravascular ultrasound Fractional flow reserve Receiver operator curve Diagnostic value Objectives: The current study aimed at determining the best cutoff value of angiographic and intravascular ultrasound (IVUS) parameters for defining fractional flow reserve (FFR) b0.8 in patients with single coronary artery lesion. Background: The correlation between angiographic or IVUS variables and FFR in patients with single coronary artery lesions has not been studied yet. Methods: Quantitative coronary analysis and IVUS and FFR measurements were used in 323 patients with a single lesion. The best angiographic and IVUS cutoff values and their predictive value for FFRb0.8 were compared using area under the receiver operator characteristic curve (AUC). Results: FFRb0.8 was in 54.2%. Minimal lumen area (MLA), plaque burden (PB), lesion length (LL) and lesion at left anterior descending artery (LAD) were four predictors of FFRb0.8. LL had less value in predicting FFRb0.8. The cutoff values of PB and MLA for FFRb0.8 were 72.7% and 2.97 mm 2. MLA and PB had similar high diagnostic value for vessel size 3 mm (cutoff values: 3.02 mm 2 and 80.7%), proximal LAD lesion (cutoff values: 3.04 mm 2 and 76.5%) and unstable angina (2.82 mm 2 and 71.9%). Combination of MLA (2.82 mm 2 ) and PB (80.6%) had increased diagnostic value for distal LAD lesion. Only PB (71%) had higher diagnostic value for diabetic patients. MLA and PB could not predict FFRb0.8 for vessel sizeb3 mm, and non-lad lesion. Conclusion: Best cutoff value of MLA and PB for FFRb0.8 in patients with a single lesion is patient-, vessel size- and lesion location-oriented. PB has strengthened diagnostic accuracy for diabetic patients Elsevier Ireland Ltd. All rights reserved. 1. Introduction The visual estimation of severity of obstructive coronary artery disease is usually inaccurate; quantitative coronary analysis (QCA) could not provide the anatomic information in detail [1,2]. Intravascular ultrasound (IVUS) permits the direct assessment of vessel/plaque, and provides accurate information regarding the selection of devices and stent expansion [3], but optimal IVUS criteria to define the functional significance of lesions fractional flow reserve (FFR), a golden of functional ischemia [4], are not clearly known yet. Based on the cost-effect The authors certify that the paper is not under consideration elsewhere, that none of the paper's contents have been previously published, that all authors have read and approved the manuscript, and that there is full disclosure of any potential conflicts of interest. Corresponding author at: Nanjing First Hospital, Nanjing Medical University, Cardiological Department, 68 Changle Road, Nanjing , China. Tel./fax: address: chmengx@126.com (S.-L. Chen). analysis, the simultaneous use of IVUS and FFR is still not accepted by most interventionists. Therefore, clarifying the predictive value of IVUS for FFRb0.8 is thought to be meaningful in a cath lab. Recently, several studies have reported the good correlation of IVUS parameters (minimal lumen area, MLA) with either FFRb0.75 or FFRb0.8 [5 13]. However, the differences in inclusion criteria (including percentage of diameter stenosis, multiple lesions, baseline clinical characteristics, etc.) and imaging devices resulted in the different cutoff values of MLA for predicting FFRb0.75 or FFRb0.8. As a diagnostic test to establish the correlation of FFR with anatomical variables, the total number of patient and the prevalence of FFRb0.8 influence the real power of tested variables [14,15]. Previous studies included patient or lesion sample varied from 34 to 267 [5 12]; very lower prevalence of FFRb0.8 is obviously associated with insufficient prediction. Accordingly, the results from those studies might be underpowered. Furthermore, the correlation of FFR and anatomical variables was not yet studied in patients with a single lesion. Therefore, the present study aimed to analyze the predictive value of QCA and IVUS parameters for FFRb0.8 in a single lesion from patients with preciously calculated sample size /$ see front matter 2012 Elsevier Ireland Ltd. All rights reserved.

2 S.-L. Chen et al. / International Journal of Cardiology 168 (2013) Methods 2.1. Patient population This study was a prospective registry study conducted in four-center. Patients who underwent a diagnostic catheterization and had a single de novo stenosis ( 40% by visual estimation) with TIMI flow Grade 3 were consecutively enrolled. Simultaneous IVUS and FFR were recorded for all patients. Exclusion criteria were multi-vessel disease, multiple lesions ( 2lesion with diameter stenosis 40% by visual estimation in target vessel), left main disease (>20% by visual estimation), ostial side branch (diagonal, obtuse marginal, right posterior lateral or posterior descending artery) stenosis, restenotic lesions, thrombus-containing lesions by angiography, previous PCI or coronary artery bypass graft for any vessel, myocardial infarction (MI) at any stage, left ventricular eject fractionb40%, the presence of collateral circulation, co-existed valvular or pericardial disease, patients having permanent or temporary pacemaker, intolerable to adenosine. The study protocol was approved by Ethic Committee of participating centers. Written consent was obtained from patients FFR measurement Trans-radial approach was recommended for all patients, unless it is not available. After baseline angiography, a 6 F guiding catheter without side holes engaged into the ostium of the coronary artery. Repeat angiography was acquired after intra-coronary injection μg of nitroglycerin. A pressure wire (Premier, Volcano Therapeutics, Inc., Rancho Cordova, California, Volcano, US, or Pressure Wire, Radi Medical Systems, Uppsala, Sweden) was advanced and positioned at least 10-mm distal to the target lesion. When the transducer was just outside the guiding catheter, the ratio of pressure from the tip of the guiding catheter and from transducer onto the pressure wire was set at 1.0. Before measurement of FFR, additional μg of nitroglycerin was injected into coronary artery at bolus, then, FFR was measured at maximal hyperemia induced by intravenous continuous infusion of adenosine, admitted at 180 μg/kg/min [13]. The stenosis was considered functionally significant when the FFR was b IVUS imaging and quantitative coronary angiography analysis After FFR assessment, IVUS imaging was performed after intracoronary administration of μg nitroglycerin using motorized transducer pullback (0.5 mm/s) and a commercial scanner (Boston Scientific/SCIMED, Minneapolis, MN) consisting of a rotating 40-MHz transducer within a 3.2 F imaging sheath. Using computerized planimetry (EchoPlaque 3.0, Indec Systems, MountainView, CA), off-line quantitative IVUS analysis was performed by two independent observers at a core laboratory at the Nanjing Heart Center, who were blinded to the study design and FFR and QCA data. The proximal and distal reference segments were selected within 5 mm proximal and distal to the lesion without an intervening side branch. Averaged proximal and distal reference external elastic membrane (EEM) and reference lumen areas and the mean reference lumen diameter were obtained. At the site of the narrowest lumen, MLA and EEM areas were measured. Plaque burden (PB) at the MLA site was calculated as (EEM area lumen area)/eem area 100. The vessel remodeling index was defined as the ratio of the vessel area at the site of minimum lumen area and that of reference site. Remodeling was categorized as positive when the remodeling index was >1.05 and negative when it was b0.95. Angiographic images at an end-diastolic cycle were required at a best injection which could clearly be separated coronary artery branches. QCA was also performed by two experienced observers at an independent core laboratory at CCRF (Beijing, China), who was blinded to the FFR value and IVUS findings. By using the guiding catheter for calibration and an edge detection system (CAAS QCA system, Pie Medical, Maastricht, The Netherlands), the reference diameters, minimum lumen diameter (MLD) and lesion length were measured, and the percentage diameter stenosis was calculated. Lesions in LAD were classified by proximal (from branching point of left main to the orifice of first septal or diagonal branches) and distal (after the take-off of first large branch) Estimating sample size and statistical power The current study was a diagnostic test, using angiographic or IVUS parameters as test variables, and FFR as the Golden. According to the previous studies, the average of sensitivity and specificity of MLA, plaque burden and lesion length as a predictor of FFRb0.8 was 80% and 70%, 80% and 75%, 90% and 50%, respectively. Thus, according to the formula [14,15]: N=V a 2 p(1 P)/ 2, where, V a =1.96, p=sensitivity or specificity, =variation(5.5% for sensitivity and 9% for specificity). A total sample size=patient number in test group (calculated according to sensitivity) pluses patient number in control group (calculated according to specificity). If the prevalence of FFRb0.8 was >50% among studied lesions, the calculated patient sample was 303( ) by MLA, 292(203+89) by plaque burden, and 232( ) by lesion length, respectively. Finally, a total of 303 lesions (highest calculated sample size) were required. Because of the uncertainty in the quality of IVUS images (6%) [1 10], we decided to extend the lesion sample to 323 in order to get a power>80% Statistical analysis Data are presented as mean±sd for continuous variables and frequency for categorical variables. Comparison of continuous variables was performed using the Student t test or nonparametric Mann Whitney test. Analysis of discrete variables was performed using the chi-square test. Binary logistic regression analysis was performed to find the determinants of FFR b0.8. A simple Scatter Dot model was created to identify the correlation of FFR value with QCA and IVUS parameters, and the coefficients with p value were calculated by the bivariate correlation model. All analyses were performed using SPSS 15.0 (SPSS Co., Chicago, IL, US). Analyze-It was used to create and compare the receiver operator characteristic (ROC) curve in order to examine the angiographic and IVUS parameters as a predictor of FFRb0.8. The resulting area under the ROC curve (AUC), 95% CI, SE, Z and p value were calculated simultaneously. The comparison of AUCs between different ROCs (Z test) was then performed; a p value b0.05 indicated one AUC better over another. From this analysis, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, positive likelihood ratio (PLR) and negative likelihood ratio (NLR) were calculated. The best cutoff value was determined by the maximum sum of sensitivity and specificity. A p value b0.05 was considered significantly. 3. Results Between May 2009 and June 2011, 388 patients were enrolled. Among them, 65 patients were excluded due to lesions in the left main (n=12), distal right coronary artery lesions (n=26), collateral feeders (n=15), or poor IVUS images (n=12). Finally, a total 323 lesions in 323 patients were included in this study. Of them, the prevalence of FFRb0.8 was 54.2% (n=175). These met all criteria of study design. Both QCA and IVUS measurements were the averaged results of two observers. In order to keep the quality of data, intraobserver and interobserver coefficients of variation for FFR, QCA and IVUS analysis between two observers were calculated in 20 consecutive patients. The target variation was b5% Baseline characteristics The baseline clinical characteristics were comparable between FFRb0.8 vs. FFR 0.8 groups (Table 1), with exception for gender (83.4% vs. 66.2%, p b0.001) and current smoking (48.6% vs. 31.1%, p=0.006) Angiographic and IVUS parameters 169 (52.3%) lesions were seen in LAD, without significant difference between FFRb0.8 vs. FFR 0.8 groups (Table 2). However, FFRb0.8 group had more proximal LAD (78.8%) and tortuous lesions (25.7%), compared to FFR 0.8 group (47.1% and 12.8%, pb0.001 and p=0.005, respectively), leading to larger vessel area in FFRb0.8 group (p=0.001). FFRb0.8 group was characterized by longer lesion length, smaller MLD and lumen area, and increased DS and plaque burden, when compared with patients in FFR 0.8 group. Remodeling index was not correlated with FFR value (p>0.05) Identifying the predictors of FFRb0.8 By logistic regression, MLA, plaque burden, lesion length and lesion at LAD were four predictors of FFRb0.8 (Table 3) from 322 lesions Receiver operator characteristic curve and correlation analysis ROC curve analysis was used to assess the diagnostic accuracy and cutoff value of lesion length, plaque burden and MLA for FFRb0.8 from overall 323 lesions (Fig. 1). By Z test, plaque burden and MLA had similar predictive value (p=0.1758), compared to lesion length (pb and p=0.0001, respectively, Fig. 1 and Table 4). The cutoff values of lesion length, plaque burden and MLA for FFRb0.8 were mm, 72.7% and 2.97 mm 2 (Table 4), respectively, and their correlation with FFR was shown in Fig. 2A C. Although the sensitivity of MLA for FFRb0.8 was 82.9%, however, its specificity was still below 70%.

3 1366 S.-L. Chen et al. / International Journal of Cardiology 168 (2013) Table 1 Baseline characteristics. Overall FFR 0.8(n=148) FFRb0.8(n=175) p Male, n 244(75.5) 98(66.2) 146(83.4) b0.001 Age (yr.) 63.5± ± ± Height (cm) 167.4± ± ± Weight (kg) 70.1± 68.9± ± Hypertension, n 222(68.7) 108(73.0) 114(65.1) Diabetes, n 85(26.3) 42(28.4) 43(24.6) Hyperlipidemia, n 39(12.1) 18(12.2) 21(12.0) Baseline TCh, mmol/l 4.16± ± ± Baseline LDL, mmol/l 2.59± ± ± Baseline TG, mmol/l 1.55± ± ± Baseline HDL, mmol/l 1.04± ± ± Smoking, n 131(40.5) 46(31.1) 85(48.6) Strokeb6 months, n 24(7.4) 14(9.5) 10(5.7) GI bleedingb6 months, n 3(0.9) 0 3(1.7) Peripheral artery disease, n 21(7.0) 5(3.7) 16(9.6) Stable angina, n 222(68.7) 108(73.0) 114(65.1) Unstable angina, n 101(21.3) 40(27.0) 61(34.9) LVEF 58.9± ± ± Red blood cell, x10 9 /L 4.36± ± ± White blood cell, x10 9 /L 6.38± ± ± Platelet, x10 9 /L 185.0± ± ± Hemoglobin, x10 9 /L 134.5± ± ± Serum creatinine, mmol/l 79.10± ± ± CK-MB, U/L 15.38± ± ± Glucose, mmol/l 6.37± ± ± C-reactive protein, mg/dl 2.74± ± ± Interleukin-6, ng/dl 9.53± ± ± TCH, total cholesterol; LDL, low density lipoprotein; TG, triglycerin; HDL, high density lipoprotein; GI, gestrointestine; LVEF, left ventricular eject fraction Diagnostic accuracy of selected parameters in subgroups In each subgroup, ROC curves were regenerated again for lesion length, plaque burden and MLA, and were compared using Z test (Table 4). The sensitivity, specificity, accuracy, MPV, NPV, PLR, NLR and Youden Index were shown in Table 5. Generally, anatomical variables had different cutoff values for predicting FFRb0.8 among subgroups. For RVD 3 mm, LAD lesion, proximal or distal LAD lesion, unstable angina subgroups, MLA (cutoff values: 3.02 mm 2, 3.03 mm 2, 3.04 mm 2, 2.82 mm 2, 2.82 mm 2 ) and plaque burden (cutoff values: 80.7%, 75.4%, 76.5%, 80.6%, 71.9%) had equivalent diagnostic value (all p by Z test>0.05), both were superior to lesion length (cutoff values mm, mm, mm, mm, mm, all p by Z test b0.05). Combination of plaque burden and MLA had strengthened predictive value for lesion at distal LAD. Unfortunately, all three parameters had lower diagnostic value (Youden indexb0.50) when lesion was at non-lad vessel, although plaque burden was better than lesion length (p by Z testb0.05). In lesion with RVDb3 mm or in patients with stable angina, there was no significant difference in diagnostic accuracy between lesion length (cutoff values: mm and mm), plaque burden (73.8% and 71.7%) and MLA (2.49 mm 2, and 2.92 mm 2, all p by Z test>0.05). For patients with diabetes, plaque burden (cutoff value: 71%) was superior to either lesion length (cutoff value: mm, p by Z testb0.0001) or MLA (cutoff value: 3.04 mm 2, p by Z test=0.0438) in predicting FFRb0.8. Lesion length had lowest accuracy (Tables 4 and 5). 4. Discussion The current study was the first study having larger number of patients with a single lesion. The major findings are as follows: (1) the cutoff value of angiographic or IVUS parameters for FFRb0.8 is patient-, vessel size- or lesion location-dependent, (2) plaque burden and MLA had equivalent and increased diagnostic value for FFRb0.8 in lesion with RVD 3 mm, LAD lesion, proximal LAD lesion, and in patients with unstable angina, when compared with lesion length, and (3) plaque burden was superior to MLA for FFRb0.8 for diabetic patients Insight analysis into the study design All studies to establish the correlation of FFR with anatomical variables are classified by diagnostic test. Two factors influence the real power of tested variables [14,15]: the total number of patient and Table 2 Angiographic and IVUS parameters. Overall (n=323) FFR 0.8 (n=148) FFRb0.8 (n=175) Angiographic parameters Target vessels, n Non-LAD 154(47.7) 78(52.7) 76(43.4) Proximal lesion 101(65.6) 55(70.5) 46(60.5) Distal lesion 53(34.4) 23(29.5) 30(39.5) LAD 169(52.3) 70(47.3) 99(56.6) Proximal lesion 112(66.3) 33(47.1) 79(78.8) b0.001 Distal lesion 57(33.7) 37(52.9) 20(21.2) b0.001 Tortuous 64(19.8) 19(12.8) 45(25.7) Mild moderate 51(15.8) 18(12.2) 23(18.9) calcification RVD (mm) 3.12± ± ± b3 mm, n 141(43.7) 68(46.9) 73(41.2) MLD (mm) 1.09± ± ±0.40 b0.001 DS 64.2± ± ±11.9 b0.001 Lesion length (mm) 18.42± ± ±12.23 b0.001 IVUS parameters MLD (mm) 1.69± ± ±0.26 b0.001 Vessel area (mm 2 ) 11.52± ± ± Lumen area (mm 2 ) 2.85± ± ±0.80 b0.001 Plaque area (mm 2 ) 8.67± ± ±5.11 b0.001 Plaque burden 72.16± ± ±12.19 b0.001 Negative remodeling, 235(72.8) 105(70.9) 130(74.3) n Positive remodeling, n 15(4.6) 3(2.0) 12(6.9) LAD, left anterior descending artery; RVD, reference vessel diameter; MLD, minimal lumen diameter; DS, diameter stenosis. p

4 S.-L. Chen et al. / International Journal of Cardiology 168 (2013) Table 3 Predictors of physiological significant ischemia (FFRb0.8). OR 95% CI p Minimal lumen area Lesion length Plaque burden b0.001 Lesion at LAD LAD, left anterior descending artery. the prevalence of disease. The sensitivity and specificity may vary in different clinical populations, and prevalence is a marker for such differences [14]. Previous studies on predicting a FFRb0.8 included patient or lesion sample varied from 34 to 267 [5 12], FFRb0.8 was seen in 88 (32.9%) of 267 lesion by Koo et al. [12] and only 21% by Kang et al. [11] compared to >50% in our study. Furthermore, LAD lesion is thought to have higher prevalence of FFRb0.8; the percentage of LAD lesion might influence the results. The ratio of LAD/non-LAD lesions was 2.87:1 by Koo et al. [12] 1.99:1 by Kang et al. [11], significantly different to 1.09:1 by our study. Thus, very lower prevalence of FFRb0.8 coupled with inadequate sample, is obviously associated with insufficient prediction. The current study, appreciating for studies done earlier, firstly calculated sample number needed to strengthen the predictive power. Therefore, the present study might have increased predictive value for FFRb Assessment of diagnostic value The diagnostic test generated several parameters to assess the predictive value of a tested variable from different aspects. Sensitivity and specificity, two commonly used indexes, are always paired in two opposite directions: the higher sensitivity the lower specificity. Therefore, Youden index is the composite value of sensitivity and specificity and can represent the predictive value of a diagnostic test. For example, Kang et al. [11] reported that the sensitivity and specificity by MLA b2.4 mm 2 for FFRb0.8 were 90% and 60%, respectively, however, their Youden index was 0.5. Similarly, for RVD=3.0 mm from Koo et al. [12], the sensitivity and specificity by MLAb3.0 mm 2 were 76% and 62%, respectively, with Youden Index only When Youden index was classified by lower ( 0.5), intermediate ( ) and higher ( 0.70), MLA b3.04 mm 2 for FFRb0.8 for proximal LAD lesion has higher diagnostic value because of Youden index>0.70 by our study. In contrast, MLAb2.82 mm2 for distal LAD lesion has accuracy 82.5%, but Youden index is only Based on this analysis, we suggested that Youden index might be a better index for assessing the diagnostic value. On the other hand, compared to previous studies, we firstly introduced the Z test to compare several ROCs, this allowed the direct comparison of AUC by parameters Factors influence the FFR value Although FFR is independent of heart rate, blood pressure and myocardial contractility [16], FFR, as the sum of total atherosclerosis burden, is definitely influenced by the severity of stenosis and the area of myocardium provided by the target vessel [17]. To be working as a tree, different segment of coronary artery has its own area to supply with blood. Similar to IVUS study, traditional cutoff value of MLA for stenting is 6 mm 2 for left main and 4 mm 2 for LAD [3,10]. Koo et al. [12] firstly reported the different cutoff values of MLA for FFRb0.8 for lesions at proximal or distal LAD. Unfortunately, those studies had lower Youden index of different cutoff values of MLA, possibly because of the lower prevalence of FFRb0.8 and the presence of multi-vessel disease. The current study enrolled only a single lesion (but not only single vessel), and could diminish the effect of opposite (RCA or LCX) lesions on the FFR for LAD lesion. Using regression analysis, LAD lesion, lesion length, MLA and plaque burden were four predictors of FFRb0.8 from our study, similar to report by Kang et al. [11]. MLA represents the absolute lumen area, but plaque burden might be more comprehensive than MLA because it is the weighted calculation from vessel/lumen/ plaque areas. Theoretically, plaque burden should at least equal to MLA in diagnosing FFRb0.8. Our results confirmed that MLA and plaque burden had similar Youden index for RVD 3 mm 2, LAD (either proximal or distal) lesion, or unstable lesion. For RVDb3 mm or non-lad lesion, MLA and plaque burden had lower predictive value, partially because of small area in jeopardy of ischemia Diagnostic value of plaque burden for patients with diabetes Except for the lesion length and lesion location serving as confounding factors of FFRb0.8 based on previous studies [11,12] and our results, the diagnostic value of anatomical variables for FFRb0.8 in patients with diabetes was also studied separately because one of the features in diabetic patient is the dysfunction of microvascular bed [18], which affects the distal vessel dilation to adenosine [20], leading to increased FFR. In contrast, Sahinarslan et al. [19] did not find the difference in percentage of FFR b0.8 between patients with diabetes vs. non-diabetes. Our results showed that MLAb3.04 mm 2 had significantly lower diagnostic value when compared to plaque burden, which implied that the reduction of predictive power by MLA might be the limitation of FFR because diabetes induces dysfunction of microvascular beds. Actually, similar to results by Kang et al. [11], the predictive value of the plaque burden was stable and did not differ among the subgroups. As discussed above, plaque burden was the ratio of plaque area and vessel area. In a given vessel and a lesion with 3 mm of RVD (plaque area=7.07 mm 2 ), if plaque area was 4.57 mm 2 and the calculated plaque burden was only 64.6% (MLA was 2.5 mm 2 ). Thus, even MLA was quite small, however, plaque area was still lower, which probably could explain why plaque burden had increased predictive value for FFRb0.8 when compared with MLA Limitations Fig. 1. Comparison of receiver operator curve in overall patients. Pijis et al. [16] reported that an FFR myo b0.75 yielded a specificity rate for detection of inducible ischemia of 100%, a sensitivity rate of 88% and a diagnostic accuracy rate of 93%. We did not analyze the

5 1368 S.-L. Chen et al. / International Journal of Cardiology 168 (2013) Table 4 Area under the ROC curve for lesion length, minimal lumen area and plaque burden in overall patients and subgroups. AUC 95% CI SE p Cutoff value p by Z test a Overall lesions Lesion length b mm b MLA b mm Plaque burden b % b RVD 3 mm Lesion length mm MLA b mm 2 b Plaque burden b % b RVDb3 mm Lesion length b mm b MLA b mm Plaque burden b % LAD Lesion length b mm b MLA b mm 2 b Plaque burden b % b Proximal LAD Lesion length mm b MLA b mm Plaque burden b % Distal LAD Lesion length mm b MLA b mm Plaque burden b % Non-LAD Lesion length mm MLA b mm Plaque burden b % Diabetic patients Lesion length mm b MLA b mm Plaque burden b % b Unstable angina Lesion length mm b MLA b mm 2 b Plaque burden b % b Stable angina Lesion length mm b MLA mm Plaque burden b % AUC, area under the curve; RVD, reference vessel diameter; MLA, minimal lumen area; LAD, left anterior descending artery. a Z test indicated the comparison of the area under the curve between MLA or plaque burden and lesion length. b Indicated the comparison of the area under the curve between MLA and plaque burden. diagnostic value of anatomical variables for predicting FFRb0.75. Finally, Bruyne et al. [20] reported that the coefficient of variation for FFR was found to be 4.8% in 13 patients with normal left ventricles and no diabetes during varying hemodynamic conditions. We tested the reproducibility of FFR in patients with a single lesion, but we did not pay attention to the left ventricular function and diabetes. Finally, the number of lesions in subgroups as that with RVDb3, diabetics or distal LAD is obviously less than adequate for providing cutoff values Pathophysiological perspectives and clinical implications The concept that FFR is a reliable hallmark of myocardial ischemia because of its higher reproducibility of measurements and good correlation with both invasive and non-invasive tests. Based on the basic theory, FFR was defined as the ratio of maximal blood flow to the myocardium in the presence of a stenosis to the theoretical maximal flow in the same myocardial bed without a stenosis. It is assumed that the myocardial resistance is constant, minimal and comparable. Thus, FFR has been calculated by Pd/Pa. Obviously, the myocardial resistance varies from different settings, and the pressure ratio without resistance taken into account tends to underestimate Fig. 2. A. Correlation of FFR with lesion length. B. Correlation of FFR with plaque burden. C. Correlation of FFR with minimal lumen area (MLA). the actual flow impediment [21]. Additional factors influence the FFR value include blood viscosity [22] (inducing pressure loss over a stenosis) and diseased microvascular beds [23]. Finally, FFR is definitely affected by the myocardium area as reflected by the vessel diameter and lesion length [17]. All these factors working together would induce the different anatomical cutoff values in different

6 S.-L. Chen et al. / International Journal of Cardiology 168 (2013) Table 5 Diagnostic accuracy of lesion length, minimal lumen area and plaque burden for FFRb0.8 in overall patients and subgroups. SeS SpC Accuracy clinical and angiographic settings. Thus, the combination of coronary flow reserve and FFR would be meaningful in further study. Plaque burden had stable diagnostic value in predicting FFRb0.8; this is mainly because the MLA could not directly represent the sum of atherosclerosis. For a given diseased vessel, the site of MLA is not usually the site with the heaviest plaque burden, because of the presence of tapering. Otherwise, the plaque burden is the ratio of plaque area and vessel area. This ratio indicates the real burden of atherosclerosis beyond this level. Therefore, plaque burden has sustained predictive power for FFRb0.8 in subgroups analyzed in the current study. In conclusion, from the current study with 323 single de novo lesions, different cutoff values of MLA and plaque burden for FFRb0.8 were identified in subgroups: plaque burden and MLA had equivalent predictive power of FFRb0.8 for vessel 3 mm, proximal LAD lesion and patients with unstable angina. Plaque burden was superior to MLA for predicting FFRb0.8 in diabetic patients. Of course, there was no need to find out the IVUS variables correlating with FFR for intermediate lesion, because FFRb0.8 was the sign of myocardial ischemia. Declaration of potential conflicts PPV NPV PLR NLR Youden index Overall Lesion length MLA Plaque burden RVD 3 mm Lesion length MLA Plaque burden RVDb3 mm Lesion length MLA Plaque burden LAD Lesion length MLA Plaque burden Proximal LAD Lesion length MLA Plaque burden Distal LAD Lesion length MLA Plaque burden Combined a Non-LAD Lesion length MLA Plaque burden Diabetic patients Lesion length MLA Plaque burden Unstable angina Lesion length MLA Plaque burden Stable angina Lesion length MLA Plaque burden RVD, reference vessel diameter; MLA, minimal lumen area; LAD, left anterior descending artery. a Indicated MLAb2.82 mm2 and plaque burden 80.6%. Acknowledgment We acknowledge the JPOMP Grant that supported the completion of this study. Moreover, we deeply appreciate Ms. Jing Kan, Ms. Hai-Mei Xu, and Ms. Ying-Ying Zhao for their great contributions to the measurements of QCA and IVUS. References [1] Tobis J, Azarbal B, Slavin L. Assessment of intermediate severity coronary lesions in the catheterization laboratory. J Am Coll Cardiol 2007;49: [2] Topol EJ, Nissen SE. Our preoccupation with coronary luminology: the dissociation between clinical and angiographic findings in ischemic heart disease. Circulation 1995;92: [3] Abizaid AS, Mintz GS, Mehran R, et al. Long-term follow-up after percutaneous transluminal coronary angioplasty was not performed based on intravascular ultrasound findings: importance of lumen dimensions. Circulation 1999;100: [4] Tonino PA, Fearon WF, De BB, et al. Angiographic versus functional severity of coronary artery stenoses in the FAME study fractional flow reserve versus angiography in multivessel evaluation. J Am Coll Cardiol 2010;55: [5] Takagi A, Tsurumi Y, Ishii Y, et al. Clinical potential of intravascular ultrasound for physiological assessment of coronary stenosis: relationship between quantitative ultrasound tomography and pressure-derived fractional flow reserve. Circulation 1999;100: [6] Briguori C, Anzuini A, Airoldi F, et al. Intravascular ultrasound criteria for the assessment of the functional significance of intermediate coronary artery stenoses and comparison with fractional flow reserve. Am J Cardiol 2001;87: [7] Jasti V, Ivan E, Yalamanchili V, Wongpraparut N, Leesar MA. Correlations between fractional flow reserve and intravascular ultrasound in patients with an ambiguous left main coronary artery stenosis. Circulation 2004;110: [8] Lee CH, Tai BC, Soon CY, et al. New set of intravascular ultrasound-derived anatomic criteria for defining functionally significant stenoses in small coronary arteries (results from Intravascular Ultrasound Diagnostic Evaluation of Atherosclerosis in Singapore [IDEAS] study). Am J Cardiol 2010;105: [9] Melikian N, De Bondt P, Tonino P, et al. Fractional flow reserve and myocardial perfusion imaging in patients with angiographic multivessel coronary artery disease. J Am Coll Cardiol Intv 2010;3: [10] Abizaid A, Mintz GS, Pichard AD, et al. Clinical, intravascular ultrasound, and quantitative angiographic determinants of the coronary flow reserve before and after percutaneous transluminal coronary angioplasty. Am J Cardiol 1998;82: [11] Koo BK, Yang HM, Doh JH, et al. Optimal intravascular ultrasound criteria and their accuracy for defining the functional significance of intermediate coronary stenoses of different locations. J Am Coll Cardiol Intv 2011;4: [12] Kang SJ, Lee JY, Ahn JM, et al. Validation of intravascular ultrasound-derived parameters with fractional flow reserve for assessment of coronary stenosis severity. Circ Cardiovasc Interv 2011;4: [13] Fischer JJ, Samady H, McPherson JA, et al. Comparison between visual assessment and quantitative angiography versus fractional flow reserve for native coronary narrowings of moderate severity. Am J Cardiol 2001;90: [14] Li J. Assessing the dependence of sensitivity and specificity on prevalence in meta-analysis. Biostatistics 2011;12: [15] Leeflang MMG, Bossuyt PMM, Irwig L. Diagnostic test accuracy may vary with prevalence: implications for evidence-based diagnosis. J Clin Epidemiol 2009;62(1):5 12. [16] Pijls NHJ, De Bruyne B, Peels K, et al. Measurement of fractional flow reserve to assess the functional severity of coronary-artery stenoses. N Engl J Med 1996;334: [17] White CW, Wright CB, Doty DB, et al. Does visual interpretation of the coronary arteriogram predict the physiologic importance of a coronary stenosis? N Engl J Med 1984;310: [18] Kamide K, Rakugi H, Higaki J, et al. The renin-angiotensin and adrenergic nervous system in cardiac hypertrophy in fructose-fed rats. Am J Hypertens 2002 Jan;15(1): [19] Sahinarslan A, Kocaman SA, Olgun H, et al. The reliability of fractional flow reserve measurement in patients with diabetes mellitus. Coron Artery Dis 2009;20(5): [20] De Bruyne B, Bartunek J, Sys SU, et al. Simultaneous coronary pressure and flow velocity measurements in humans: feasibility, reproducibility, and hemodynamic dependence of coronary flow velocity reserve, hyperemic flow versus pressure slope index, and fractional flow reserve. Circulation 1996;94: [21] Pijls NHL, De Bruyne B. Coronary pressure measurement and fractional flow reserve. Heart 1998;80: [22] Siebes M, Steven AJ, Chamuleau M, et al. Influences of hemodynamic conditions on fractional flow reserve: parametric analysis of underlying models. Am J Physiol 2002;283:H [23] Pijls NHL, De Bruyne B. Coronary pressure. Dordrecht, the Netherlands: Kluwer Academic Publishers; p This study was funded by the Jiangsu Provincial Outstanding Medical Program (JPOMP ). There are no relationships with industry.