Author + information
- aDepartment of Medicine, Veterans Administration Long Beach Health Care System, Long Beach, California
- bDepartment of Medicine, University of California, Irvine, Irvine, California
- cDepartment of Cardiology, Veterans Administration Long Beach Health Care System, Long Beach, California
- ↵∗Address for correspondence:
Dr. Morton J. Kern, Department of Medicine, Veterans Administration Long Beach Health Care System, 5901 East 7th Street, Long Beach, California 90822.
In 1974, in seminal animal experiments with progressively applied arterial constrictions, Gould and Lipscomb (1) demonstrated that coronary stenoses reduced resting blood flow only when they were very severe, exceeding >85% diameter narrowed. By contrast, maximal hyperemic blood flow was affected by less severe stenoses, with reductions in hyperemic flow beginning with as little as a 50% narrowing (1). Given the individual flow variability, Gould and Lipscomb normalized the hyperemic flow to the resting flow in the same artery, naming the ratio coronary flow reserve (CFR). By characterizing the relationship between stenosis severity and CFR, Gould and Lipscomb established one of the foundational concepts of myocardial ischemia. From this animal experiment, clinicians extended the concept of CFR to visually estimated angiographic stenoses in humans, erroneously presuming that all angiographic stenoses with >50% diameter narrowing represented clinically significant lesions. Unfortunately, although true in the experimental animals, the relationship between stenosis and CFR demonstrably failed in patients because of the significant limitations of 2-dimensional angiography (e.g., foreshortening, calcifications, and branch overlap, especially for eccentric atherosclerotic lesions), and microvascular dysfunction. Recognizing the limitations of CFR, translesional pressure measurements such as fractional flow reserve (FFR) became a standard for physiological lesion assessment. Nonetheless, knowledge of the CFR remains critically important for a complete understanding of a patient’s coronary circulation, especially the microcirculation, and the associated long-term clinical outcomes.
Over the years, CFR has been measured with multiple techniques, including radiolabeled microspheres (2), invasive and noninvasive coronary Doppler (3,4), thermodilution flow transit time recordings (5), magnetic resonance imaging (6), and recently, positron emission tomographic (PET)-measured perfusion using radiolabeled water, ammonia, or other diffusible blood flow tracers (7). All of these techniques have technical challenges; many are not widely available. The best PET tracers in particular require a nuclear cyclotron on site due to short half-lives. 15Oxygen-labeled water most reliably quantifies blood flow because extraction of this highly diffusible tracer is complete and linearly related to the flow rate (8). At this time, 15O-H2O PET perfusion imaging is considered the gold standard for quantification of myocardial perfusion and CFR.
In this issue of JACC: Cardiovascular Interventions, Everaars et al. (9) directly compared invasive Doppler flow velocity reserve (CFRDoppler) and thermodilution-derived coronary flow reserve (CFRthermo) against 15O-H2O PET CFR in 98 vessels from 40 consecutive patients. FFR, CFRDoppler and CFRthermo were measured using pressure/flow sensor guidewires following coronary angiography. 15O-H2O PET CFR was obtained within 2 weeks of the invasive study.
CFRDoppler correlated modestly with CFRthermo (r = 0.59; p < 0.001) and better than CFRthermo with CFRPET (r = 0.82 and 0.55, respectively; both p < 0.001). CFRthermo overestimated CFRPET at high values. The relationship between CFRthermo and CFRPET deteriorated when only vessels with FFR >0.80 were included, whereas the relationship between CFRDoppler and CFRPET remained constant. The quality of Doppler flow signals was worse than those of thermodilution, but Doppler flow velocity measurements showed lower intraobserver variability. Overall, this study, from a center of expertise in coronary Doppler and PET, suggests that CFRDoppler is better correlated with the PET gold standard than CFRthermo, but is harder to measure accurately.
Differences between Doppler velocity and Thermodilution flow
Prior comparisons of CFRDoppler and CFRthermo in both experimental animal (10) and human (11) studies suggested near equivalency. Fifteen years later, with the availability of PET as a CFR standard, Everaars et al. (9) now demonstrate superior agreement of CFRDoppler over CFRthermo with regard to CFRPET. Given that each technique uses a different method for measuring blood flow, perfect agreement amongst the different methods should not be expected. Moreover, CFR sounds easier to measure than it really is. CFR is not routinely acquired during clinical catheterization procedures due to multiple technical challenges that are worth reviewing in detail.
Intracoronary Doppler ultrasound measures red cell velocities assuming a parabolic laminar flow profile. Velocity is displayed as a spectral waveform and quantitated by the average peak (instantaneous) velocity (APV) value. Coronary flow velocity reserve (CFVR), APVhyperemia/APVrest is not always identical to other CFR measurements calculated by volumetric ratios. For CFVR, volumetric flow can be calculated as the product vessel cross-sectional area (CSA) × APV/2. The CSA at the interrogation site is presumed to remain constant over the measurement, thus CFVR may more closely approximate CFR by volume. Unfortunately, hyperemic flow-mediated vasodilation and changes in CSA are rarely incorporated, a factor contributing to the variance between CFRDoppler and other CFR measures. Lastly, accurate signal acquisition is challenging as the Doppler ultrasound beam must be maintained in a stable position within the narrow flow field, and acute sensor angulation relative to the flow vector will reduce the measured velocity.
In contrast to Doppler velocity flow, thermodilution coronary flow is measured by the mean transit time of the room temperature saline indicator as detected by the temperature change from the sensor on the coronary guidewire. Flow is calculated as 1/mean transit time (1/Tmn), with signals from several small saline bolus injections through the guide catheter averaged. CFRthermo is calculated as [hyperemic 1/Tmn]/[resting 1/Tmn]. The 1/Tmn is affected by quality of the injections, sensor wire position, adequate guide seating without damping, and the coronary flow rate. The technique of manual bolus injection accounts for some variance. The thermodilution method becomes less accurate at high flow rates.
To their credit, Everaars et al. (9) accepted only good-quality signals. Acceptable Doppler signals were those with clearly identifiable systolic and diastolic phases with early diastolic peak flow having a gradual decline. Good-quality thermodilution curves were those with a unimodal shape without distortion. Low-quality tracings for both techniques were excluded. Intraobserver variability of flow signals was higher with 1/Tmn at rest and at hyperemia (11.5 ± 7% and 14.6 ± 9%, respectively; p < 0.016) than APV (4.8 ± 3% and 5.3 ± 4%, respectively; p = ns). Variability of CFRDoppler and CFRthermo was 6.6 ± 5% and 18.8 ± 11%, respectively.
In contrast to Everaars et al. (9), earlier studies suggested a closer agreement between CFRDoppler and CFRthermo (10,11). The authors (9) postulate that the discrepancy between these studies is due, in part, to technical differences (e.g., intravenous vs. intracoronary adenosine, accurate wire position relative to side branches, etc.). Everaars et al. (9) note that when examined in normal or minimally narrowed vessels with FFR >0.80, the higher flow rates produced an even great disparity between CFRthermo with CFRPET. Lastly, patient selection may have contributed to variance of CFR measurements, as only a minority of patients had Doppler recordings of sufficient quality for comparison in the earlier studies. Advocates for the thermodilution technique would undoubtedly disagree (10,11).
Which technique to choose?
Everaars et al. (9) remind us that both current invasive CFR techniques have different strengths and weaknesses. The variance of measurements and only modest correlations to CFRPET reinforce a longstanding need for improved techniques, technology, and equipment to measure flow. Without effective tools to measure flow, large multicenter studies that would make CFR a routine cath lab practice will not be performed. Having used both techniques, we appreciate the technical challenges and errors of each. At the present time, the conflicting comparative CFR data suggest that the best technique to use is still the one with which the operator and lab have the most expertise.
↵∗ Editorials published in JACC: Cardiovascular Interventions reflect the views of the authors and do not necessarily represent the views of JACC: Cardiovascular Interventions or the American College of Cardiology.
Dr. Kern has been a consultant and speaker for Abbott/St. Jude Medical, Philips/Volcano, Acist Medical, Opsens, and Heartflow. Dr. Seto has received research funding from Philips/Volcano; and has been a speaker for ACIST Medical Systems, Boston Scientific, and Philips/Volcano.
- Florenciano-Sánchez R.,
- de la Morena-Valenzuela G.,
- Villegas-García M.,
- et al.
- Brown L.A.E.,
- Onciul S.C.,
- Broadbent D.A.,
- et al.
- Danad I.,
- Uusitalo V.,
- Kero T.,
- et al.
- Bergmann S.R.,
- Fox K.A.,
- Rand A.L.,
- et al.
- Everaars H.,
- de Waard G.A.,
- Driessen R.S.,
- et al.
- Fearon W.F.,
- Farouque O.,
- Balsam L.B.,
- et al.
- Pijls N.H.J.,
- De Bruyne B.,
- Smith L.,
- et al.