Abstract: Objectives. To assess the diagnostic performance of angiography-derived fractional flow reserve (FFRangio) measurements in patients with stable coronary artery disease when used online in the catheterization laboratory during routine coronary angiography. Background. FFR, an index of the hemodynamic severity of coronary stenosis, is derived from invasive measurements using a pressure-monitoring guidewire and hyperemic stimulus. While FFR is the gold standard, it remains under-utilized. FFRangio may have several advantages owing to the reduced operator time, no wire-related or procedural complications, and no need for administration of vasodilators. Methods. FFRangio is a novel technology that uses a patient’s hemodynamic data and routine angiograms to generate FFR values at each point along the coronary tree. We present the online application of the system where FFRangio was successfully used in the catheterization laboratory during routine coronary angiography and compared to invasive FFR. Fifty-three patients (79% men) and 60 coronary lesions were analyzed. Results. Values derived using FFRangio ranged from 0.58-0.96 and correlated closely (Pearson’s correlation coefficient, r=0.91; P<.001) with the invasive FFR measurements (range, 0.52-0.97). The 95% limits of agreement between invasive and non-invasive FFR ranged from -0.065 to 0.07 using Bland-Altman analysis. For an FFR cut-off value of 0.80, the sensitivity, specificity, and diagnostic accuracy of FFRangio were 86%, 100%, and 95%, respectively. Conclusions. This is the first report of using the online application of the FFRangio system. In this single-center experience, FFRangio values showed high correlation rates to invasive FFR.
J INVASIVE CARDIOL 2018;30(6):224-229. Epub 2018 March 15.
KEY WORDS: coronary pressure, computational fluid dynamics (CFD), image-based FFR
Fractional flow reserve (FFR) is an index that quantifies the hemodynamic impact of epicardial stenosis. It is defined as the ratio of hyperemic myocardial flow in the presence of stenosis, to the hyperemic flow in its absence, and is obtained by measuring the distal coronary pressure and the aortic pressure, respectively, using pressure-measuring guidewires during maximal hyperemia.1-3 FFR is considered the benchmark for clinical decision making. Clinical outcome studies have shown that medical therapy should be preferred for non-hemodynamically significant lesions (FFR >0.80), while coronary revascularization should be offered in cases of significant functional stenosis (FFR ≤0.80).4-11 Accordingly, both the United States and European revascularization guidelines recommend using FFR to guide the proper treatment strategy in stable coronary lesions.12,13
Nevertheless, for a variety of practical reasons, FFR measurements remain under-utilized. Therefore, the ability to derive FFR values from routinely performed coronary angiograms, without the need for a pressure guidewire or hyperemic stimulus, could have an important impact on daily clinical practice.14-16
Several image-based FFR methodologies have recently been introduced. Computational fluid dynamics simulation applied to cardiac computed tomography images as well as to flat-detector angiograms for the evaluation of non-invasive FFR have been proposed.14,16-21 The computational complexity of such simulations requires manual interaction and considerable processing time.
FFRangio is a novel technology based on multilayered image-processing techniques and blood-flow calculations from routine diagnostic angiography images. Two or more angiograms from different projections are reconstructed into a qualitative three-dimensional (3D) model of the demonstrated coronary arteries. The generated 3D model is then analyzed for stenosis detection. The FFR values are then calculated and presented at each point along the vessel. The accuracy of the offsite FFRangio system was tested in a recent publication,22 and further validated on over 200 coronary lesions, where it exhibited high accuracy, with sensitivity and specificity of 88% and 95%, respectively.23
In the present study, we evaluate the accuracy of online FFRangio in comparison to invasive pressure-wire based FFR, where the FFRangio was computed online in the catheterization laboratory amid the angiogram, in patients with suspected coronary artery disease referred for cardiac catheterization.
FFRangio. The FFRangio methodology was described elsewhere.22 In short, FFRangio builds a 3D model of the coronary tree based on the geometry of ≥2 angiographic projections, and displays a functional angiogram, with a 3D angiography of the coronary tree superimposed with color-coded FFR values at any point along the tree.
Study population. In this single-center Institutional Review Board (IRB)-approved study, we included subjects above the age of 18 years who were diagnosed with either stable angina, unstable angina pectoris, or non-ST elevation myocardial infarction (NSTEMI) who were referred to coronary angiography. FFR measurements were performed for clinical reasons in at least one coronary artery. Patients with left main stenosis, ostial stenosis, in-stent restenosis at the target vessel, previous bypass surgery, and/or diffuse coronary disease were excluded. In all cases, the stenosis was clearly delineated on the angiogram.
Coronary angiography. Catheterization was performed using a 5 or 6 Fr diagnostic coronary catheter according to local practice and using either the radial or the femoral approach. After administration of intracoronary nitrates, at least two projections of the vessel to be measured were acquired at the smallest magnification possible. The exact inclination of the radiographic tube was left to the operator’s discretion. Contrast medium (5-6 mL) was injected at a rate of 4 mL/sec at 600 psi for adequate opacification. Coronary cineangiograms were recorded at 15 frames/sec using the Siemens AXIOM-Artis (n = 28) or the Philips Healthcare AlluraXper (n = 25).
Invasive FFR measurements. Invasive FFR measurements were performed using 6 Fr guide catheters and St. Jude/Abbott (n = 35) or Volcano/Philips (n = 18) pressure wires. Inva-sive FFR measurements were performed using 6 Fr guide catheters and St. Jude/Abbott (n = 35) or Volcano/Philips (n = 18) pressure wires. Hyperemia was ob-tained by intracoronary adminis-tration of adenosine using a 20 mL syringe with a power injec-tor (200 µg for the left coronary tree and 100 µg for the right coronary tree). Mea-surements were repeated twice, with the mean value used for analyses.
FFRangio computation. FFRangio computations were performed online amid the angiography procedure in the catheterization room. The analysis was performed by operators blinded to the invasive FFR results. The digital imaging and communications in medicine (DICOM) cineangiography sequences (with contrast, not adenosine) were transferred directly from the c-arm system and processed along with the patient’s mean aortic pressure obtained at the time of the angiogram. User interaction was required to guide automatic processing and included a temporal analysis of each cine angiogram, allowing for verification of cardiac phase synchronization even in the absence of an electrocardiographic signal, as well as proper extraction of vessel centerlines, which is performed automatically based on convolutional neural networks adapted to detection of vasculature in two-dimensional (2D) angiograms. At any stage, the user could review and verify the automatic processes, and edit and refine as required. The automatic processing consisted of the 3D coronary tree reconstruction and flow estimation (a case example is presented in Figure 1). The location of the pressure sensor used for hyperemic measure was made known to the FFRangio user, and FFRangio was then measured at the identical location.
Statistical approach. Standard summary statistical tests were used. Descriptive statistics and Pearson’s correlation were performed. In order to explore the agreement between invasive and non-invasive FFR estimates, Bland-Altman analyses were plotted. Estimated bias (defined as the mean difference between the two methods ± standard deviation) was calculated and one-sample t-test was used to evaluate whether it differed significantly from zero. The receiver operating characteristic (ROC) curve was used to evaluate the diagnostic accuracy of the FFRangio, visual estimation of diameter stenosis, and 2D and 3D quantitative coronary analysis (QCA). The area under the curve (AUC) of each ROC curve was used to estimate the accuracy of each method. All statistical analyses were performed with MedCalc version 17.2 (MedCalc Software).
Fifty-three patients (79% men) were enrolled, comprising 60 coronary lesions, most of which (93%) were in the left tree. A total of 65% of lesions were in the proximal or mid left anterior descending (15 and 24 lesions, respectively). Approximately one-third of the lesions were positive when assessed with invasive FFR (≤0.80; 22 lesions [37.0%]) and FFR measurements ranged from 0.52-0.97 (mean, 0.82). The mean percent diameter stenosis was 45.0 ± 10.3% as measured by 2D-QCA, and 59.8 ± 14.8% by visual estimation (range, 20%-95%). Demographic and procedural details are shown in Table 1 and Table 2, respectively.
FFRangio was measured onsite amid the procedures, and all FFRangio analyses were completed within a time frame of 10 min from the end of coronary angiogram to obtaining the color-coded FFRangio display. FFRangio measurements ranged from 0.58-0.96 (mean, 0.82) and 19 lesions (32%) were significant (≤0.80) when assessed by FFRangio. The correlation between the independent wire-based FFR values and the FFRangio values (dependent) is presented in Figure 2, with r=0.91 (intercept=0.09 and slope=0.89). The bias estimated for the mean values of FFRangio and invasive FFR was 0.0027, indicating that FFRangio values do not under-estimate or over-estimate invasive FFR (95% limits of agreement, -0.065-0.070) (Figure 2 and Table 2).
Using a cut-off of 0.80, the comparison between invasive FFR and FFRangio, visual estimation of diameter stenosis, and QCA are presented in Figure 3. The concordance between values obtained using invasive FFR and FFRangio was high (area under ROC, 0.95) as compared with the lower correlation obtained for visual lesion severity assessment, 2D-QCA, and 3D-QCA (area under ROC, 0.86, 0.88, and 0.82, respectively). The sensitivity, specificity, and diagnostic accuracy of FFRangio was 86%, 100%, and 95%, respectively. The diameter stenosis through visual estimation and 2D-QCA when done by the site was compared to FFRangio QCA (Table 3).
Main findings. In the present single-center study, we successfully performed blinded online analysis of FFR values derived from angiography (FFRangio) in 53 patients and compared the results to simultaneously obtained invasive FFR measurements. When continuous FFR values were dichotomized using the standard clinical cut-off value of 0.80, FFRangio achieved a diagnostic accuracy of 95%. Forty-seven (78%) of the lesions analyzed had invasive FFR values of 0.70-0.90, and 23 of the lesions (38%) had invasive FFR values between 0.75-0.85, ie, adjacent to the cut-off value. This is the first time that an online analysis of angiography-derived FFR has been described. When comparing the online results presented here with earlier published offline results, a similar level of high diagnostic accuracy is apparent (Table 4).23
Other angiography-based methods to derive FFR. While other groups have attempted to use angiographic data in order to simulate invasive FFR measurements,14,16-21,24 the methods are laborious, require additional analysis, and often involve long processing times. Some are limited to offline analysis by a core laboratory. Morris et al used a computational fluid dynamics method based on angiographic studies and FFR measurements in 19 patients with stable coronary artery disease, but this system requires a rotational coronary angiogram (which is not universally available), prior knowledge in computational fluid dynamics calculation is required, and the processing time is long.21 Tu et al used the quantitative flow ratio (QFR) method derived from three different flow models (Medis) based on 3D-QCA of vessel segments and the flow moving through the stenosis.14 For 84 vessels in 73 patients, the investigators reported a correlation of fQFR, cQFR, and aQFR to invasive FFR as r=0.69, r=0.77, and r=0.72, respectively, and overall accuracy of 80% (fQFR), 86% (cQFR), and 87% (aQFR). This system relies on 3D-QCA combined with additional TIMI frame count from high-quality images (15-30 frames/sec for cQFR) for the calculation of mean volumetric flow rate at hyperemia, and still requires induced hyperemic conditions for one of the methods (aQFR). In addition, this method provides assessment of the main vessel of interest without providing the side branches. Trobs et al also used computational fluid dynamics in order to retrospectively calculate angiography-based FFR in 73 patients with coronary artery disease. Bland-Altman analysis was 0.0082, with a standard deviation of -0.117 to 0.134.18
While considerable effort is being invested in similar systems, the availability of other solutions allowing for a real-time calculation of hemodynamic measurements is limited. Thus, our report shows great promise due to its accuracy and full integration within the routine catheterization laboratory workflow.
Study limitations. This study is not without limitations. The design included a small sample size from a single center, and interoperator and intraoperator variabilities were not collected. Patients with coronary artery bypass graft surgery, diffuse disease, ostial lesions, and in-stent restenosis were not included; the findings of this study are therefore limited. A larger study on “real-world” consecutive patients is warranted. Other limitations related to the technology include the following: (1) high-quality angiograms are needed whereby the stenosis is clearly depicted, and the vessel of interest is well delineated; and (2) since the spatial resolution of the angiogram is close to the dimensions of the minimal luminal diameter in very severe stenosis, the FFRangio values below 0.50 were truncated to 0.50. This has likely decreased the degree of correlation. However, in clinical practice, this is of little importance as decisions made on the basis of such low values will not change.
This study reports for the first time the use of FFRangio when used online. The analysis showed a high concordance with invasive FFR and can be obtained within minutes in the setting of a regular coronary angiogram. If confirmed in a larger study, FFRangio appears to be a simple means of integrating anatomy and physiology with high spatial resolution in the catheterization laboratory. This, in turn, may facilitate the adoption of FFR-based clinical decision making regarding proper coronary revascularization.
1. Pijls NH, van Son JA, Kirkeeide RL, De Bruyne B, Gould KL. Experimental basis of determining maximum coronary, myocardial, and collateral blood flow by pressure measurements for assessing functional stenosis severity before and after percutaneous transluminal coronary angioplasty. Circulation. 1993;87:1354-1367.
2. De Bruyne B, Pijls NH, Paulus WJ, et al. Transstenotic coronary pressure gradient measurement in humans: in vitro and in vivo evaluation of a new pressure monitoring angioplasty guide wire. J Am Coll Cardiol. 1993;22:119-126.
3. De Bruyne B, Baudhuin T, Melin JA, et al. Coronary flow reserve calculated from pressure measurements in humans. Validation with positron emission tomography. Circulation. 1994;89:1013-1022.
4. Zimmermann FM, Ferrara A, Johnson NP, et al. Deferral vs. performance of percutaneous coronary intervention of functionally non-significant coronary stenosis: 15-year follow-up of the DEFER trial. Eur Heart J. 2015;36:3182-3188.
5. van Nunen LX, Zimmermann FM, Tonino PA, et al. Fractional flow reserve versus angiography for guidance of PCI in patients with multivessel coronary artery disease (FAME): 5-year follow-up of a randomised controlled trial. Lancet. 2015;386:1853-1860.
6. Tonino PA, De Bruyne B, Pijls NH, et al. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med. 2009;360:213-224.
7. Puymirat E, Peace A, Mangiacapra F, et al. Long-term clinical outcome after fractional flow reserve-guided percutaneous coronary revascularization in patients with small-vessel disease. Circ Cardiovasc Interv. 2012;5:62-68.
8. Pijls NH, van Schaardenburgh P, Manoharan G, et al. Percutaneous coronary intervention of functionally nonsignificant stenosis: 5-year follow-up of the DEFER study. J Am Coll Cardiol. 2007;49:2105-2111.
9. Pijls NH, 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;334:1703-1708.
10. Muller O, Mangiacapra F, Ntalianis A, et al. Long-term follow-up after fractional flow reserve-guided treatment strategy in patients with an isolated proximal left anterior descending coronary artery stenosis. JACC Cardiovasc Interv. 2011;4:1175-1182.
11. De Bruyne B, Fearon WF, Pijls NH, et al. Fractional flow reserve-guided PCI for stable coronary artery disease. N Engl J Med. 2014;371:1208-1217.
12. Windecker S, Kolh P, Alfonso F. 2014 ESC/EACTS guidelines on myocardial revascularization: the task force on myocardial revascularization of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS) developed with the special contribution of the European Association of Percutaneous Cardiovascular Interventions (EAPCI). Eur Heart J. 2014;35:2541-2619.
13. Levine GN, Bates ER, Blankenship JC, et al. 2011 ACCF/AHA/SCAI guideline for percutaneous coronary intervention: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines and the Society for Cardiovascular Angiography and Interventions. Circulation. 2011;124:e574-e651.
14. Tu S, Bourantas CV, Norgaard BL, Kassab GS, Koo BK, Reiber JH. Image-based assessment of fractional flow reserve. EuroIntervention. 2015;11:V50-V54.
15. Ntalianis A, Trana C, Muller O, et al. Effective radiation dose, time, and contrast medium to measure fractional flow reserve. JACC Cardiovasc Interv. 2010;3:821-827.
16. Morris PD, van de Vosse FN, Lawford PV, Hose DR, Gunn JP. “Virtual” (computed) fractional flow reserve: current challenges and limitations. JACC Cardiovasc Interv. 2015;8:1009-1017.
17. Tu S, Westra J, Yang J, et al. Diagnostic accuracy of fast computational approaches to derive fractional flow reserve from diagnostic coronary angiography: the international multicenter FAVOR pilot study. JACC Cardiovasc Interv. 2016;9:2024-2035.
18. Trobs M, Achenbach S, Rother J, et al. Comparison of fractional flow reserve based on computational fluid dynamics modeling using coronary angiographic vessel morphology versus invasively measured fractional flow reserve. Am J Cardiol. 2016;117:29-35.
19. Taylor CA, Fonte TA, Min JK. Computational fluid dynamics applied to cardiac computed tomography for noninvasive quantification of fractional flow reserve: scientific basis. J Am Coll Cardiol. 2013;61:2233-2241.
20. Papafaklis MI, Muramatsu T, Ishibashi Y, et al. Fast virtual functional assessment of intermediate coronary lesions using routine angiographic data and blood flow simulation in humans: comparison with pressure wire – fractional flow reserve. EuroIntervention. 2014;10:574-583.
21. Morris PD, Ryan D, Morton AC, et al. Virtual fractional flow reserve from coronary angiography: modeling the significance of coronary lesions: results from the VIRTU-1 (VIRTUal Fractional Flow Reserve From Coronary Angiography) study. JACC Cardiovasc Interv. 2013;6:149-157.
22. Kornowski R, Lavi I, Pellicano M, et al. Fractional flow reserve derived from routine coronary angiograms. J Am Coll Cardiol. 2016;68:2235-2236.
23. Pellicano M, Lavi I, De Bruyne B, et al. Validation study of image-based fractional flow reserve during coronary angiography. Circ Cardiovasc Interv. 2017:10(9).
24. Chen SJ, Carroll JD. 3-D reconstruction of coronary arterial tree to optimize angiographic visualization. IEEE Trans Med Imaging. 2000;19:318-336.
From the 1Rabin Medical Center, Petach Tikva, Israel; and 2CathWorks, Ltd, Kfar Saba, Israel.
Funding: This study was funded by CathWorks, Ltd.
Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. Prof Kornowski and Dr Lavi are co-founders at CathWorks, Ltd (Kfar Saba, Israel). Prof Kornowski reports licensing and royalty income from the FFR-angiography patent. Drs Lavi and Valtzer are employees of CathWorks, Ltd. The remaining authors have no conflicts of interest regarding the content herein.
Manuscript submitted November 27, 2017, provisional acceptance given January 8, 2018, final version accepted January 17, 2018.
Address for correspondence: Ran Kornowski, MD, FACC, FESC, Rabin Medical Center, Ze’ev Jabotinsky Rd 39, Petah Tikva, 4941492, Israel. Email: firstname.lastname@example.org