J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

ª 2 0 1 7 B Y T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y F O U N D A T I O N

P U B L I S H E D B Y E L S E V I E R

I S S N 1 9 3 6 - 8 7 9 8 / $ 3 6 . 0 0

h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j c i n . 2 0 1 6 . 1 2 . 0 1 5

CORONARY

Diagnostic Performance of Resting andHyperemic Invasive Physiological Indicesto Define Myocardial IschemiaValidation With 13N-Ammonia Positron Emission Tomography

Doyeon Hwang, MD,a Ki-Hyun Jeon, MD,b Joo Myung Lee, MD, MPH, PHD,c Jonghanne Park, MD, PHD,a

Chee Hae Kim, MD,a Yaliang Tong, MD,d Jinlong Zhang, MD,a Ji-In Bang, MD,e Minseok Suh, MD,e

Jin Chul Paeng, MD, PHD,e Sang-Hoon Na, MD, PHD,f,g Gi Jeong Cheon, MD, PHD,e Christopher M. Cook, MBBS,h

Justin E. Davies, MBBS, PHD,h Bon-Kwon Koo, MD, PHDa,g

ABSTRACT

FrobD

Me

Ch

Em

Ko

Im

OBJECTIVES The authors sought to compare the diagnostic performance of fractional flow reserve (FFR),

instantaneous wave-free ratio (iFR), and resting distal coronary artery pressure/aortic pressure (Pd/Pa) using13N-ammonia positron emission tomography (PET).

BACKGROUND The diagnostic performance of invasive physiological indices was reported to be different according

to the reference to define the presence of myocardial ischemia.

METHODS A total of 115 consecutive patients with left anterior descending artery stenosis who underwent both13N-ammonia PET and invasive physiological measurement were included. Optimal cutoff values and diagnostic

performance of FFR, iFR, and resting Pd/Pa were assessed using PET-derived coronary flow reserve (CFR) and relative

flow reserve (RFR) as references. To compare discrimination and reclassification ability, each index was compared with

integrated discrimination improvement (IDI) and category-free net reclassification index (NRI).

RESULTS All invasive physiological indices correlated with CFR and RFR (all p values <0.001). The overall diagnostic

accuracies of FFR, iFR, and resting Pd/Pa were not different for CFR <2.0 (FFR 69.6%, iFR 73.9%, and resting Pd/Pa

70.4%) and RFR <0.75 (FFR 73.9%, iFR 71.3%, and resting Pd/Pa 74.8%). Discrimination and reclassification abilities of

invasive physiological indices were comparable for CFR. For RFR, FFR showed better discrimination and reclassification

ability than resting indices (IDI ¼ 0.170 and category-free NRI ¼ 0.971 for iFR; IDI ¼ 0.183 and category-free NRI ¼ 1.058

for resting Pd/Pa; all p values <0.001).

CONCLUSIONS The diagnostic performance of invasive physiological indices showed no differences in the prediction of

myocardial ischemia defined by CFR. Using RFR as a reference, FFR showed a better discrimination and reclassification

ability than resting indices. (J Am Coll Cardiol Intv 2017;10:751–60) © 2017 by the American College of Cardiology

Foundation.

m the aDepartment of Internal Medicine and Cardiovascular Center, Seoul National University Hospital, Seoul, Korea;

epartment of Internal Medicine, Sejong General Hospital, Bucheon, Korea; cDivision of Cardiology, Department of Internal

dicine, Heart Vascular Stroke Institute, Samsung Medical Center, Seoul, Korea; dChina-Japan Union Hospital of Jilin University,

ina; eDepartment of Nuclear Medicine, Seoul National University Hospital, Seoul, Korea; fDepartment of Internal Medicine and

ergency Medical Center, Seoul National University Hospital, Seoul, Korea; gInstitute of Aging, Seoul National University, Seoul,

rea; and the hInternational Centre for Circulatory Health, National Heart and Lung Institute, Imperial College London and

perial College Healthcare National Health Service Trust, London, United Kingdom. Dr. Cook has received speakers fees from

ABBR EV I A T I ON S

AND ACRONYMS

AUC = area under curve

CFR = coronary flow reserve

CI = confidence interval

FFR = fractional flow reserve

IDI = integrated discrimination

improvement

iFR = instantaneous wave-free

ratio

MBF = myocardial blood flow

NRI = net reclassification index

Pd/Pa = distal coronary artery

pressure/aortic pressure

PET = positron emission

tomography

RFR = relative flow reserve

Philips Volc

rights perta

authors ha

contributed

Manuscript

Hwang et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

Diagnostic Performance of Invasive Physiological Indices A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0

752

P revious studies demonstrated thatpercutaneous coronary interventionfor coronary artery disease is only

beneficial in patients with myocardialischemia (1,2). Among invasive physiologicalindices, fractional flow reserve (FFR) hasbeen a standard invasive method to detectlesion-specific myocardial ischemia and hasbeen commonly used in daily clinical prac-tice (3–6). In recent years, resting indicessuch as instantaneous wave-free ratio (iFR)and resting distal coronary artery pressure/aortic pressure (Pd/Pa) were introduced as asimple invasive index to define myocardialischemia. Three large clinical studies investi-gated the diagnostic performance of restingindex against FFR and reported variousranges of diagnostic accuracy, from 60% to

90% (7–9). However, the diagnostic performance ofFFR and iFR was comparable when an independentreference test was used to define the presence ofmyocardial ischemia (10,11).

SEE PAGE 761

Noninvasivemyocardial perfusion imaging plays animportant role in determining a therapeutic plan forpatients with coronary artery disease. Positron emis-sion tomography (PET) has been considered the mostaccurate noninvasive myocardial perfusion imaging todefine myocardial ischemia (12). In addition to abso-lute myocardial blood flow (MBF), perfusion PET scanscan provide coronary flow reserve (CFR) and relativeflow reserve (RFR) (13,14). PET-derived CFR and RFRhave been regarded as some of the gold standardmethods to define myocardial ischemia (13–18). Weperformed this study to compare the diagnosticperformance of FFR, iFR, and resting Pd/Pa usingPET-derived CFR and RFR as reference standards.

METHODS

STUDY POPULATION. The study population wasselected from the IRIS FFR (Study of the Natural His-tory of FFR Guided Percutaneous Coronary Interven-tion; NCT01366404) registry. The IRIS FFR registry is aKorean multicenter registry enrolling consecutivepatients who underwent FFR measurement for any

ano. Dr. Davies is a consultant for and receives research funding

ining to pressure wire technology. Dr. Koo has received an institu

ve reported that they have no relationships relevant to the con

equally to this work.

received October 9, 2016; revised manuscript received Decembe

major epicardial coronary artery. The exclusioncriteria were stenosis with Thrombolysis In Myocar-dial Infarction flow grade of<3, graft vessel, depressedleft ventricular systolic function (ejection fraction<30%), and stenosis that was technically not suitablefor FFR evaluation. From June 2011 to September 2015,144 consecutive patients with available 13N-ammoniaPET within 3 months of FFR measurement in the leftanterior descending coronary artery were included inthis study. Fifteen patients with poor image qualityand 14 patients with unavailable iFR measurementwere excluded. All patients were enrolled from SeoulNational University Hospital. The study protocol wasapproved by the institutional review board and wasconducted in accordance with the Declaration of Hel-sinki. All patients provided written informed consentbefore enrollment.

13N-AMMONIA PET PROTOCOL. The 13N-ammoniaPET images were acquired during resting and stressstate by continuous intravenous infusion of adeno-sine (140 mg/kg/min). Adenosine was administered3 min before the stress scan, and low-dose computedtomographic scans were used to correct scatterand attenuation (19). All patients were informed torefrain from any caffeine- or xanthine-containingproducts for 24 h before scanning, and vasodilatingmedications including nitrate, beta-blocker, and cal-cium channel blocker were also stopped for 24 hbefore PET acquisition. The 370 MBq of 13N-ammoniawas administrated in resting and stress states intoa peripheral vein, and then a list mode dynamicscan was performed with a Siemens Biograph-40PET/CT scanner (Siemens Medical Solutions, Erlan-gen, Germany). For image analysis and quantificationof resting and stress absolute MBF in millilitersper minute per gram of tissue image acquisition,Carimas TM software, version 2.8 (Turku PET Centre,Finland) was used (20).

QUANTIFICATION OF ABSOLUTE MBF AND

PHYSIOLOGICAL INDICES FROM 13N-AMMONIA PET.

A 2-compartment model was applied to quantifyabsolute MBF (ml/min/g). The absolute MBF andphysiological indices of a target segment werecalculated from PET scans as described previously(21). The 6 basal segments in PET images were notquantified due to low counts in membranous

from Philips Volcano; and holds intellectual property

tional research grant from St. Jude Medical. All other

tents of this paper to disclose. The first 2 authors

r 1, 2016, accepted December 15, 2016.

TABLE 1 Baseline Patient and Lesion Characteristics

Patient characteristics (N ¼ 115)

Age, yrs 63.6 � 9.0

Male 103 (89.6%)

Body mass index, kg/m2 24.6 � 2.3

Hypertension 79 (68.7%)

Diabetes mellitus 38 (33.0%)

Hypercholesterolemia 105 (91.3%)

Current smoker 19 (16.5%)

Family history of coronary artery disease 20 (17.4%)

Prior myocardial infarction 15 (13.0%)

Left ventricular ejection fraction, % 60.1 � 6.0

Left anterior descending artery (N ¼ 115)

Quantitative coronary angiography

Reference vessel diameter, mm 3.0 � 0.4

Minimum lumen diameter, mm 1.4 � 0.5

Diameter stenosis, % 46.7 � 16.0

Lesion length, mm 16.0 � 9.7

Invasive physiologic indices

Fractional flow reserve 0.81 (0.73–0.85)

Instantaneous wave-free ratio 0.92 (0.87–0.94)

Resting Pd/Pa 0.93 (0.90–0.95)

PET parameters

Resting myocardial blood flow, ml/min/g 0.92 � 0.20

Stress myocardial blood flow, ml/min/g 1.80 � 0.43

Coronary flow reserve 2.13 � 0.58

Relative flow reserve 0.77 � 0.09

Values are mean � SD, n (%), median (interquartile range).

PET ¼ positron emission tomography; resting Pd/Pa ¼ resting distal coronaryartery pressure/aortic pressure.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7 Hwang et al.A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0 Diagnostic Performance of Invasive Physiological Indices

753

interventricular septum and to artifacts. CFR wascalculated as the ratio of stress MBF to resting MBF intarget segments (13). RFR was calculated as the ratioof stress MBF in target myocardial segments to that ofreference myocardial segments (14,15,21). Parametricstress MBF polar maps were used to delineate defectareas in target myocardial segments and to obtainMBF values in those areas (18). The averaged stressMBF in 3 segments with the highest MBF was used asreference hyperemic MBF (Online Figure 1). In orderto compare the diagnostic performance of invasivephysiological indices, CFR <2.0 and RFR <0.75 wereused as reference standards to define the presence ofmyocardial ischemia (13–15,22).

INVASIVE CORONARY ANGIOGRAPHY ANDMEASUREMENT

OF PHYSIOLOGICAL INDICES. Coronary angiography wasperformed by standard techniques. Angiographic viewswere obtained following the administration of intra-coronary nitrate (100 or 200 mg). All angiograms wereanalyzed at a core laboratory (Seoul National UniversityHospital) in a blinded fashion. Quantitative coronaryangiographywas performed in optimal projectionswithvalidated software (CAAS II, Pie Medical System,Maastricht, the Netherlands). The minimal lumendiameter, reference vessel size, and lesion length weremeasured, and % diameter stenosis was calculated.

All coronary physiological measurements were ob-tained after diagnostic angiography as previouslydescribed (3). Briefly, a 5-F to 7-F guide catheterwithout side holes was used to engage the coronaryartery. The pressure–temperature sensor guidewire(St. Jude Medical, St. Paul, Minnesota) was zeroed andequalized to aortic pressure, and then the pressuresensor was positioned at the distal segment of a targetvessel. Intracoronary nitrate (100 or 200 mg) wasadministered before each physiological measurement.Resting Pd/Pa was calculated as the ratio of meandistal coronary artery pressure to mean aortic pressurein the resting state. Continuous infusion of adenosine(140 mg/kg/min) was used to induce hyperemia. Hy-peremic distal coronary artery pressure and aorticpressure were obtained during sustained hyperemia,and FFR was calculated by mean distal coronary arterypressure/aortic pressure during hyperemia. Aftermeasurements, the pressure wire was pulled back tothe guide catheter and the presence of pressure driftwas checked. All FFR readings were collected andvalidated at the core laboratory (Seoul National Uni-versity Hospital) in a blinded fashion. iFR was calcu-lated as the mean pressure distal to the stenosisdivided by the mean aortic pressure during the dia-stolic wave-free period. The baseline tracing data ofmore than 5 heart beats were extracted and then

anonymized and coded as an ASCII text file. Those datawere sent to the iFR core laboratory (Imperial College,London, United Kingdom) where iFR was calculatedusing fully automated algorithms acting over thewave-free period over a minimum of 5 beats (9).

STATISTICAL ANALYSIS. Categorical variables werepresented as numbers and relative frequencies.Continuous variables were presented as mean � SD ormedian with interquartile range according to their dis-tributions, which were checked by the Kolmogorov-Smirnov test. Spearman correlation coefficients werecalculated to estimate the correlations between inva-sive physiological indices and PET-derived parametersdue to the non-normal distributions of FFR, iFR, andresting Pd/Pa. The Pearson correlation coefficient wasused to estimate the correlations between PET-derivedCFR andRFR. The differences of correlation coefficientswere tested by the Fisher r-to-z transformation.

The optimal cutoff values of invasive physiologicalindices for defining myocardial ischemia were calcu-lated on the basis of maximizing the sum of sensitivityand specificity of each index. Comparison of the areaunder curve (AUC) from receiver-operating character-istic curve analysis was performed with the DeLongmethod (23). Diagnostic performance of invasive

FIGURE 1 Distributions of FFR, iFR, Resting Pd/Pa, and Angiographic % Diameter

Stenosis

Histograms of the distributions of FFR (A), iFR (B), resting Pd/Pa (C), and diameter

stenosis (D) are shown. FFR ¼ fractional flow reserve; iFR ¼ instantaneous wave-free

ratio; Pd/Pa ¼ distal coronary artery pressure/aortic pressure.

Hwang et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

Diagnostic Performance of Invasive Physiological Indices A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0

754

physiological indices was presented with sensitivity,specificity, positive predictive value, negative pre-dictive value, and diagnostic accuracy. Diagnostic ac-curacies of FFR, iFR, and resting Pd/Pa were comparedusing the McNemar test. To compare discriminationand reclassification ability, each index was comparedby absolute and relative integrated discriminationimprovement (IDI) index, as well as category-free netreclassification index (NRI) (24).

All probability values were 2-sided, and ap value <0.05 was considered statistically signifi-cant. The statistical package SPSS version 22.0(SPSS, Chicago, Illinois) and SAS software, version 9.3(SAS Institute, Cary, North Carolina) were used forstatistical analyses.

RESULTS

BASELINE CHARACTERISTICS AND CORONARY

PHYSIOLOGY DATA. Table 1 shows baseline patientand lesion characteristics. Themean agewas 63.6�9.0years, and 103 patients (89.6%) were male. The meandiameter stenosis was 46.7 � 16.0%, and 58.5% of thelesions had an intermediate degree of stenosis. The

median (interquartile range) values FFR, iFR, andresting Pd/Pa values were 0.81 (0.73 to 0.85), 0.92(0.87 to 0.94), and 0.93 (0.90 to 0.95), respectively.The distributions of invasive physiological indices andangiographic lesion severity are shown in Figure 1. Themean values of CFR and RFR by PET were 2.13 � 0.58and 0.77 � 0.09, respectively (Table 1).

CORRELATIONS BETWEEN INVASIVE PHYSIOLOGICAL

INDICES AND PET-DERIVED CFR AND RFR. Theinvasive physiological indices showed positive corre-lations with both PET-derived CFR and RFR (Figure 2).The degree of correlation among the FFR, iFR, andresting Pd/Pa was not different for both CFR and RFR(Online Table 1). The trend was the same with therelationship between stress MBF and invasive physi-ological indices (Online Figure 2, Online Table 1).

OPTIMAL CUTOFF VALUES AND DIAGNOSTIC

ACCURACIES OF INVASIVE PHYSIOLOGICAL

INDICES. The optimal cutoff values of FFR, iFR, andresting Pd/Pa for defining myocardial ischemia werecalculated using CFR (<2.0) and RFR (<0.75) asreference standards. The optimal cutoff values ofFFR, iFR, and resting Pd/Pa were 0.79, 0.92, and 0.93using CFR as a reference standard. The optimal cutoffvalues of FFR, iFR, and resting Pd/Pa for RFR <0.75were the same as CFR <2.0.

Figure 3 shows diagnostic performance of FFR, iFR,and resting Pd/Pa using ischemic cutoff values calcu-lated from this study. With CFR as a reference, thesensitivity, specificity, positive predictive value,negative predictive value, and diagnostic accuracy ofFFR were 64.7%, 73.4%, 66.0%, 72.3%, and 69.6%,respectively. Those of iFR were 72.5%, 75.0%, 69.8%,77.4%, and 73.9%, and those of resting Pd/Pa were64.7%, 75.0%, 67.3%, 72.7%, and 70.4%, respectively(Figure 3A). With RFR as a reference, the diagnosticaccuracies of FFR, iFR, and resting Pd/Pa were 73.9%,71.3%, and 74.8%, respectively (Figure 3B). There wasno difference in diagnostic accuracies among 3 physi-ological indices for both CFR (p¼ 0.359 for FFR vs. iFR,p¼ 1.000 for FFR vs. resting Pd/Pa) and RFR (p¼ 0.648for FFR vs. iFR, p ¼ 1.000 for FFR vs. resting Pd/Pa).

The diagnostic performance of invasive physio-logical indices was not changed significantly whenthe previously well-defined cutoff values (cutoffvalue 0.80 for FFR, 0.90 for iFR and 0.92 for restingPd/Pa) were used (Online Figure 3).

DISCRIMINATION AND RECLASSIFICATION ABILITIES OF

INVASIVE PHYSIOLOGICAL INDICES. There was no dif-ference in AUC among the 3 invasive indices to predictCFR <2.0 (0.716 for FFR, 0.762 for iFR, and 0.761 forresting Pd/Pa) (Figure 4A). The AUC for RFR <0.75 was

FIGURE 2 Correlations Between Invasive Physiological Indices and PET-Derived Parameters

Scatter plots show that CFR has significant correlations with FFR, iFR and resting Pd/Pa (A). And also, for RFR, there are significant correlations with FFR, iFR and resting

Pd/Pa (B). CFR ¼ coronary flow reserve; CI ¼ confidence interval; PET ¼ positron emission tomography; RFR ¼ relative flow reserve; rs ¼ Spearman correlation

coefficient; other abbreviations as in Figure 1.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7 Hwang et al.A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0 Diagnostic Performance of Invasive Physiological Indices

755

0.826 (95% confidence interval [CI]: 0.749 to 0.903),0.771 (95%CI: 0.684 to 0.858), and 0.774 (95%CI: 0.684to 0.864) for FFR, iFR, and resting Pd/Pa, respectively.The AUC of FFR was higher than that of iFR (p ¼ 0.047for comparison) (Figure 4B).

Compared with iFR and resting Pd/Pa, FFR showedcomparable discrimination and reclassification abilityfor determining myocardial ischemia defined by CFR<2.0 (IDI ¼ �0.029 and NRI ¼ �0.357 with iFR;IDI ¼ �0.036 and NRI ¼ �0.317 with resting Pd/Pa)(Table 2). As for RFR <0.75, FFR showed improvementof discrimination and reclassification ability for deter-mining myocardial ischemia compared with restingindices (IDI ¼ 0.170 and NRI ¼ 0.971 with iFR; IDI ¼0.183 and NRI ¼ 1.058 with resting Pd/Pa) (Table 3).

DISCUSSION

In this study, we compared the diagnostic perfor-mance of FFR, iFR and resting Pd/Pa for the prediction

of myocardial ischemia defined by PET-derived CFRand RFR. The main findings of this study were as fol-lows. First, all invasive physiological indices and PET-derived CFR and RFR showed significant correlations.Second, optimal cutoff values of FFR, iFR, and restingPd/Pa for defining myocardial ischemia defined byPET parameters were 0.79, 0.92, and 0.93, respec-tively. Third, there were no significant differences indiagnostic accuracies among FFR, iFR, and restingPd/Pa against CFR and RFR. Fourth, discriminationand reclassification ability of FFR to define low RFRwas better than those of resting indices. These findingsimply that the diagnostic abilities of these physiolog-ical indices can be different according to the referenceused for the comparison.

PHYSIOLOGICAL INDICES TO DEFINE MYOCARDIAL

ISCHEMIA. The presence of myocardial ischemia is thekey prognostic indicator in patientswith coronary arterydisease (1,2). Because coronary angiography has several

FIGURE 3 Diagnostic Performance of Invasive Physiological Indices

Diagnostic performance of FFR, iFR, and resting Pd/Pa were compared using CFR (A) and RFR (B) as reference standards for predicting

myocardial ischemia. NPV ¼ negative predictive value; PPV ¼ positive predictive value; other abbreviations as in Figures 1 and 2.

Hwang et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

Diagnostic Performance of Invasive Physiological Indices A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0

756

limitations to define myocardial ischemia, the use ofinvasive physiological studies has becomemore popular(13,16,17). The benefit of FFRhas been validated throughseveral clinical studies, and FFR is considered the goldstandard for defining lesion-specific myocardialischemia in daily practice (3–5,17). Recently, restingindices, such as iFR and resting Pd/Pa, that do notrequire hyperemia have been proposed as a simplealternative for FFR. Previous studies reported variousranges (60% to 90%) of diagnostic accuracies of iFR andresting Pd/Pa compared with FFR (7–9). However, thediagnostic performance canbedifferent according to thereference indexused to definemyocardial ischemia. Senet al. (10) reported comparable diagnostic agreement ofFFR and iFR when hyperemic stenosis resistance wasused as a reference to define myocardial ischemia. Pet-raco et al. (11) used coronary flow velocity reserve as areference and showed a better diagnostic discriminationof iFR than that of FFR (iFR AUC 0.82; FFR AUC 0.72;p< 0.001). In our study, PET-derived CFR and RFRwere

used as references to compare the diagnostic perfor-mance between FFR and resting indices. PET has beenconsidered the gold standard to measure myocardialblood flow, and PET-derived CFR and RFR have beenthoroughly investigated as noninvasive methods todefinemyocardial ischemia and prognostic indicators inpatients with coronary artery disease (13–15,22).

RATIONALE OF VALIDATING INVASIVE PHYSIOLOGICAL

INDICES USING PET-DERIVED CFR AND RFR. CFR is theratio of stress MBF and resting MBF, and representshow much MBF can be supplied in stress conditionscompared with that of a resting condition (13,16).RFR is the ratio of stress MBF in diseased segmentsand that in normal segments and means the degree ofhyperemic flow decrease due to the coronary arterystenosis; this is a noninvasive version of FFR (14,15).Although the concept of CFR and RFR are different,the prognostic values of both parameters have beenthoroughly investigated. CFR is the oldest and

FIGURE 4 ROC Curve Analysis Using PET-Derived Parameters as Reference Standards

Comparison of ROC curves of FFR, iFR and resting Pd/Pa to predict CFR<2.0 is shown and AUCs are presented (A). Also, with RFR<0.75 as a

reference standard to predict myocardial ischemia, ROC curves of FFR, iFR and resting Pd/Pa are compared (B). AUC ¼ area under curve;

ROC ¼ receiver-operating characteristic; other abbreviations as in Figures 1 and 2.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7 Hwang et al.A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0 Diagnostic Performance of Invasive Physiological Indices

757

extensively investigated physiological index, and theprognostic implication of CFR was consistentlyobserved regardless of the methods of measurement,such as invasive flow measurement, stress echocar-diography, and PET (25–28). The clinical relevance ofRFR can be inferred from the well-validated FFRstudies (3–5,15,26).

Each physiological index can represent differentaspects of myocardial ischemia and has its ownstrength and weakness (29). As for RFR, it is a moreepicardial stenosis–specific index and cannot reflectthe microvascular dysfunction of diseased myocardialterritory, because it is also influenced by microvas-cular disease state of normal reference myocardialterritory. By contrast, CFR reflects not only theseverity of epicardial stenosis, but also the microvas-cular disease status (29,30). These differences lead tothe 37.4% discordance between CFR and RFR in thisstudy, and this was similar to the previous results(Online Figure 4) (29).

COMPARISON BETWEEN INVASIVE PHYSIOLOGICAL

INDICES AND PET-DERIVED PARAMETERS. FFR,iFR, and resting Pd/Pa showed significantcorrelations with PET-derived parameters, butshowed different patterns to the CFR and RFR.With CFR, FFR demonstrated a numerically lowercorrelation coefficient than iFR and restingPd/Pa. This result is similar to a previous study byPetraco et al. (11) that reported a better correlationof iFR with coronary flow velocity reserve thanFFR (iFR r ¼ 0.68; FFR r ¼ 0.50; p forcomparison <0.001). In contrast to the CFR, thecorrelation coefficient of FFR with RFR was numeri-cally higher than those of resting indices. Thedegree of correlation between FFR and RFR fromour study was comparable to a previous study re-ported by Stuijfzand et al. (14) (FFR vs. RFRr ¼ 0.54; p < 0.01). Considering RFR is a hyperemicindex and the concept of RFR is more similar to FFRthan that of iFR and resting Pd/Pa, the better

TABLE 2 Comparison of Discrimination and Reclassification Abilities of the Invasive Physiologic Indices Using PET-Derived CFR (<2.0)

as a Reference Standard

Model 1(Reference)

Model 2(Testing)

AUC IDI NRI (Category-Free)

Model 1 Model 2 p Value Absolute Relative p Value Value p Value

Resting vs. hyperemia

iFR FFR 0.762 0.716 0.133 �0.029 �0.168 0.138 �0.357 0.057

Resting Pd/Pa FFR 0.761 0.716 0.183 �0.036 �0.201 0.051 �0.317 0.091

Resting vs. resting

Resting Pd/Pa iFR 0.761 0.762 0.932 �0.007 �0.040 0.348 �0.113 0.546

AUC ¼ area under the curve; CFR ¼ coronary flow reserve; FFR ¼ fractional flow reserve; IDI ¼ integrated discrimination improvement; iFR ¼ instantaneous wave-free ratio;NRI ¼ net reclassification index; other abbreviations as in Table 1.

Hwang et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

Diagnostic Performance of Invasive Physiological Indices A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0

758

correlation between FFR and RFR seems to be natu-ral. These different patterns of correlations with PET-derived parameters between resting and hyperemicphysiological indices suggest that the diagnosticperformance of invasive physiological indices can bedifferent according to the reference standard used forcomparison.

DIAGNOSTIC PERFORMANCE OF INVASIVE

PHYSIOLOGICAL INDICES. In this study, wecompared the diagnostic performance of resting andhyperemic indices using 2 different concepts ofmyocardial ischemia, PET-derived CFR and RFR. Likeprevious studies, our study results showed thedifferent diagnostic performance of resting and hy-peremic indices according to the reference (10,11).Although the overall diagnostic accuracy was notdifferent regardless of cutoff values of invasive phys-iological indices (6,7), the discrimination and reclas-sification ability of FFRwas better than those of restingindices when RFR was used as a reference. Eventhough it was not statistically significant, the restingindices showed numerically higher correlation andbetter diagnostic agreement with CFR, comparedwith those of FFR. These results are in line with theprevious study of Petraco et al. (11), which usedDoppler-measured coronary flow velocity reserve as areference standard. Because CFR reflects both macro-vascular and microvascular disease status and RFR is

TABLE 3 Comparison of Discrimination and Reclassification Abilities

as a Reference Standard

Model 1(Reference)

Model 2(Testing)

AUC

Model 1 Model 2 p Value

Resting vs. hyperemia

iFR FFR 0.771 0.826 0.047

Resting Pd/Pa FFR 0.774 0.826 0.093

Resting vs. resting

Resting Pd/Pa iFR 0.774 0.771 0.836

RFR ¼ relative flow reserve; other abbreviations as in Tables 1 and 2.

more epicardial stenosis–specific, this study impliesthat resting and hyperemic indices may representdifferent aspects of myocardial ischemia. Therefore,these differences need to be appreciated with cautionwhen the different invasive physiological index is usedin clinical practice.

Although it is beyond the scope of this study, thedemonstration of prognostic implication is the mostimportant aspect in the evaluation of clinicalrelevance of any diagnostic test. The benefit of aFFR-guided revascularization strategy has been well-demonstrated by several clinical studies (3–5).Ongoing clinical studies that compare the clinicaloutcomes of FFR-guided and iFR-guided strategies(DEFINE-FLAIR [Functional Lesion Assessment ofIntermediate Stenosis to Guide Revascularisation][NCT02053038], iFR SWEDEHEART [Evaluation ofiFR vs FFR in Stable Angina or Acute CoronarySyndrome] [NCT02166736]) will provide additionalinformation on the prognostic implication of iFR.

STUDY LIMITATIONS. First, reference standards usedin this study are also the surrogate for myocardialischemia. With the lack of a clinically available truegold standard, this limitation can be applied to allclinical studies. Second, PET segmentation by vascularterritory can be influenced by individual variations incoronary anatomy. Although the quantification ofmyocardial blood flow by PET is known to have low

of the Invasive Physiologic Indices Using PET-Derived RFR (<0.75)

IDI NRI (Category-Free)

Absolute Relative p Value Value p Value

0.170 1.332 <0.001 0.971 <0.001

0.183 1.592 <0.001 1.058 <0.001

0.013 0.111 0.069 0.159 0.402

PERSPECTIVES

WHAT IS KNOWN? The presence of myocardial ischemia is the

key prognostic indicator in patients with coronary artery disease,

and the use of invasive physiological studies has become more

popular. However, the diagnostic performance of invasive

physiological indices was reported to be different according

to the reference to define the presence of myocardial

ischemia.

WHAT IS NEW? This study compared the diagnostic

performance of FFR, iFR and resting Pd/Pa using CFR and RFR

from 13N-ammonia PET. The study results showed that FFR, iFR,

and resting Pd/Pa had similar diagnostic performance when CRF

was used as a reference, and better discrimination and reclassi-

fication ability of FFR when RFR was used as a reference. Because

each physiological index can represent different aspects of

myocardial ischemia and has its own strength and weakness, the

user needs to understand these differences when applying an

invasive physiological index in clinical practice.

WHAT IS NEXT? Although we compared the diagnostic

performance of invasive physiological indices using PET, the

demonstration of prognostic implication is the most important

aspect in the evaluation of clinical relevance of any diagnostic

test. Therefore, further study is needed to demonstrate the

prognostic value of resting indices such as iFR and resting

Pd/Pa.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7 Hwang et al.A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0 Diagnostic Performance of Invasive Physiological Indices

759

intra-observer and interobserver variability, myocar-dial segmentation of the target vascular territory inPET images could be different by different observers.Third, our study used 13N-ammonia as a PET tracer,and the absolute MBF measured by different tracerscould be different. However, the flow ratio such as CFRor RFR was reported to be relatively constant amongdifferent tracers (31). Fourth, there is a possibility ofinsufficient statistical power due to the relativelysmall sample size of this study to detect significantdifference among physiological indices. Further studywith a larger sample size is warranted to clarify thisissue.

CONCLUSIONS

The diagnostic performance of invasive physiologicalindices showed no difference in the prediction ofmyocardial ischemia defined by CFR. Using RFR as areference standard, FFR showed higher discrimina-tion and reclassification ability than iFR or restingPd/Pa. The user needs to understand this differencewhen applying an invasive physiological index inclinical practice.

ADDRESS FOR CORRESPONDENCE: Dr. Bon-KwonKoo, Department of Internal Medicine and Cardio-vascular Center, Seoul National University Hospital,101 Daehang-ro, Chongno-gu, Seoul 03080, Korea.E-mail: bkkoo@snu.ac.kr.

RE F E RENCE S

1. Erne P, Schoenenberger AW, Burckhardt D, et al.Effects of percutaneous coronary interventions insilent ischemia after myocardial infarction: theSWISSI II randomized controlled trial. JAMA 2007;297:1985–91.

2. Shaw LJ, Berman DS, Maron DJ, et al. Optimalmedical therapy with or without percutaneous cor-onary intervention to reduce ischemicburden: resultsfrom the Clinical Outcomes Utilizing Revasculariza-tion andAggressiveDrugEvaluation (COURAGE) trialnuclear substudy. Circulation 2008;117:1283–91.

3. Pijls NH, van Schaardenburgh P, Manoharan G,et al. Percutaneous coronary intervention offunctionally nonsignificant stenosis: 5-yearfollow-up of the DEFER study. J Am Coll Cardiol2007;49:2105–11.

4. Tonino PA, De Bruyne B, Pijls NH, et al. Frac-tional flow reserve versus angiography for guidingpercutaneous coronary intervention. N Engl J Med2009;360:213–24.

5. De Bruyne B, Fearon WF, Pijls NH, et al. Frac-tional flow reserve-guided PCI for stable coronaryartery disease. N Engl J Med 2014;371:1208–17.

6. Windecker S, Kolh P, Alfonso F, et al. 2014 ESC/EACTS guidelines on myocardial revascularization.EuroIntervention 2015;10:1024–94.

7. Jeremias A, Maehara A, Genereux P, et al.Multicenter core laboratory comparison of theinstantaneous wave-free ratio and resting Pd/Pawith fractional flow reserve: the RESOLVE study.J Am Coll Cardiol 2014;63:1253–61.

8. Berry C, van ’t Veer M, Witt N, et al. VERIFY(VERification of Instantaneous Wave-Free Ratioand Fractional Flow Reserve for the Assessment ofCoronary Artery Stenosis Severity in EverydaYPractice): a multicenter study in consecutive pa-tients. J Am Coll Cardiol 2013;61:1421–7.

9. Sen S, Escaned J, Malik IS, et al. Developmentand validation of a new adenosine-independentindex of stenosis severity from coronary wave-intensity analysis: results of the ADVISE (ADeno-sine Vasodilator Independent Stenosis Evaluation)study. J Am Coll Cardiol 2012;59:1392–402.

10. Sen S, Asrress KN, Nijjer S, et al. Diagnosticclassification of the instantaneous wave-free ratiois equivalent to fractional flow reserve and is notimproved with adenosine administration. Resultsof CLARIFY (Classification Accuracy of Pressure-Only Ratios Against Indices Using Flow Study).J Am Coll Cardiol 2013;61:1409–20.

11. Petraco R, van de Hoef TP, Nijjer S, et al.Baseline instantaneous wave-free ratio as a

pressure-only estimation of underlying coronaryflow reserve: results of the JUSTIFY-CFR Study(Joined Coronary Pressure and Flow Analysisto Determine Diagnostic Characteristics of Basaland Hyperemic Indices of Functional LesionSeverity-Coronary Flow Reserve). Circ CardiovascInterv 2014;7:492–502.

12. Jaarsma C, Leiner T, Bekkers SC, et al. Diag-nostic performance of noninvasive myocardialperfusion imaging using single-photon emissioncomputed tomography, cardiac magnetic reso-nance, and positron emission tomography imagingfor the detection of obstructive coronary arterydisease: a meta-analysis. J Am Coll Cardiol 2012;59:1719–28.

13. Gould KL, Johnson NP, Bateman TM, et al.Anatomic versus physiological assessment of cor-onary artery disease. Role of coronary flowreserve, fractional flow reserve, and positronemission tomography imaging in revascularizationdecision-making. J Am Coll Cardiol 2013;62:1639–53.

14. Stuijfzand WJ, Uusitalo V, Kero T, et al. Rela-tive flow reserve derived from quantitativeperfusion imaging may not outperform stressmyocardial blood flow for identification of

Hwang et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 0 , N O . 8 , 2 0 1 7

Diagnostic Performance of Invasive Physiological Indices A P R I L 2 4 , 2 0 1 7 : 7 5 1 – 6 0

760

hemodynamically significant coronary artery dis-ease. Circ Cardiovasc Imaging 2015;8:e002400.

15. De Bruyne B, Baudhuin T, Melin JA, et al.Coronary flow reserve calculated from pressuremeasurements in humans. Validation with positronemission tomography. Circulation 1994;89:1013–22.

16. Naya M, Tamaki N, Tsutsui H. Coronary flowreserve estimated by positron emission tomogra-phy to diagnose significant coronary artery diseaseand predict cardiac events. Circ J 2015;79:15–23.

17. Berry C, Corcoran D, Hennigan B, Watkins S,Layland J, Oldroyd KG. Fractional flow reserve-guided management in stable coronary diseaseand acute myocardial infarction: recent de-velopments. Eur Heart J 2015;36:3155–64.

18. Danad I, Uusitalo V, Kero T, et al. Quantitativeassessment of myocardial perfusion in the detec-tion of significant coronary artery disease: cutoffvalues and diagnostic accuracy of quantitative[(15)O]H2O PET imaging. J Am Coll Cardiol 2014;64:1464–75.

19. Lubberink M, Harms HJ, Halbmeijer R, deHaan S, Knaapen P, Lammertsma AA. Low-dosequantitative myocardial blood flow imaging using15O-water and PET without attenuation correc-tion. J Nucl Med 2010;51:575–80.

20. El Fakhri G, Kardan A, Sitek A, et al. Repro-ducibility and accuracy of quantitative myocardialblood flow assessment with (82)Rb PET: compar-ison with (13)N-ammonia PET. J Nucl Med 2009;50:1062–71.

21. Lee JM, Kim CH, Koo BK, et al. Integratedmyocardial perfusion imaging diagnostics improvedetection of functionally significant coronary ar-tery stenosis by 13n-ammonia positron emissiontomography. Circ Cardiovasc Imaging 2016;9:e004768.

22. Pijls NH, De Bruyne B, Peels K, et al. Mea-surement of fractional flow reserve to assess thefunctional severity of coronary-artery stenoses.N Engl J Med 1996;334:1703–8.

23. DeLong ER, DeLong DM, Clarke-Pearson DL.Comparing the areas under two or more correlatedreceiver operating characteristic curves: anonparametric approach. Biometrics 1988;44:837–45.

24. PencinaMJ,D’AgostinoRBSr.,D’AgostinoRBJr.,VasanRS. Evaluating the added predictive ability of anew marker: from area under the ROC curve toreclassification and beyond. Stat Med 2008;27:157–72; discussion 207–12.

25. Chamuleau SA, Tio RA, de Cock CC, et al.Prognostic value of coronary blood flow velocityand myocardial perfusion in intermediate coronarynarrowings and multivessel disease. J Am CollCardiol 2002;39:852–8.

26. Marques KM, Knaapen P, Boellaard R,Lammertsma AA, Westerhof N, Visser FC. Micro-vascular function in viable myocardium afterchronic infarction does not influence fractionalflow reserve measurements. J Nucl Med 2007;48:1987–92.

27. Herzog BA, Husmann L, Valenta I, et al. Long-term prognostic value of 13N-ammonia myocardial

perfusion positron emission tomography addedvalue of coronary flow reserve. J Am Coll Cardiol2009;54:150–6.

28. Cortigiani L, Rigo F, Gherardi S, et al. Coronaryflow reserve during dipyridamole stress echocar-diography predicts mortality. J Am Coll CardiolImg 2012;5:1079–85.

29. Johnson NP, Kirkeeide RL, Gould KL. Isdiscordance of coronary flow reserve and frac-tional flow reserve due to methodology or clini-cally relevant coronary pathophysiology? J AmColl Cardiol Img 2012;5:193–202.

30. van de Hoef TP, Siebes M, Spaan JA, Piek JJ.Fundamentals in clinical coronary physiology: whycoronary flow is more important than coronarypressure. Eur Heart J 2015;36:3312–19a.

31. Tahari AK, Lee A, Rajaram M, et al. Absolutemyocardial flow quantification with (82)Rb PET/CT: comparison of different software packagesand methods. Eur J Nucl Med Mol Imaging 2014;41:126–35.

KEY WORDS 13N-ammonia positronemission tomography, coronary arterydisease, fractional flow reserve,instantaneous wave-free ratio,myocardial ischemia

APPENDIX For supplemental figures anda table, please see the online version ofthis article.