Coronary Thermodilution to Assess Flow Reserve
Validation in Humans
Background— Guide wire–based simultaneous measurement of fractional flow reserve (FFR) and coronary flow reserve (CFR) is important to understand microvascular disease of the heart. The aim of this study was to investigate the feasibility of simultaneous measurement of FFR and CFR by one pressure-temperature sensor-tipped guide wire with the use of coronary thermodilution and to compare CFR by thermodilution (CFRthermo) with simultaneously measured Doppler CFR (CFRDoppl).
Methods and Results— In 103 coronary arteries in 50 patients, a pressure-temperature sensor-tipped 0.014-inch floppy guide wire and a 0.014-inch Doppler guide wire were introduced. Both normal vessels and a wide range of stenotic vessels were included. With 3 mL of saline at room temperature used as an indicator, by hand-injection, thermodilution curves in the coronary artery were obtained in triplicate, both at baseline and at intravenous adenosine-induced maximum hyperemia. After adequate curve-fitting, CFRthermo was calculated from the ratio of inverse mean transit times and compared with CFRDoppl calculated by velocities at hyperemia and baseline. Adequate sets of thermodilution curves and corresponding CFRthermo could be obtained in 87% of the arteries versus 91% for Doppler CFR and 100% for FFR. CFRthermo correlated fairly well to CFRDoppl (CFRthermo=0.84 CFRDoppl+0.17; r=0.80; P<0.001), although individual differences of >20% between both indexes were seen in a quarter of all arteries.
Conclusions— This study shows the feasibility of simultaneous measurement of FFR (by coronary pressure) and CFR (by coronary thermodilution) in humans by one single guide wire in a practical and straightforward way and will facilitate assessment of microvascular disease.
Received December 27, 2001; revision received March 18, 2002; accepted March 21, 2002.
Guide wire–based measurement of fractional flow reserve (FFR) and coronary flow reserve (CFR) has become increasingly important to understand the physiological significance of coronary artery disease.1–9⇓⇓⇓⇓⇓⇓⇓⇓
Fractional flow reserve, calculated from coronary pressure measurement, is an accurate and specific index of epicardial stenosis severity.3,5,8,9⇓⇓⇓ It can be simply determined by a pressure wire and enables decision-making with respect to the need for an intervention and evaluation of PTCA or coronary stenting.8–12⇓⇓⇓⇓ However, it does not account for increased microvascular resistance.
Coronary flow reserve, on the other hand, mostly measured so far by the Doppler wire, investigates both epicardial and microvascular disease but does not allow discrimination between these entities.2,6,13⇓⇓ Therefore, simultaneous measurement of CFR and FFR should give the clinician better insight in the respective contribution of the epicardial vessel and microvasculature to total resistance to myocardial blood flow.2,6,13–15⇓⇓⇓⇓
Up to now, two different sensor-tipped guide wires are necessary to measure CFR and FFR, making such a procedure expensive and time-consuming. It would be an advantage to measure both indexes at the same time by using one single guide wire.
Recently, we validated experimentally a novel technology by using coronary thermodilution to assess CFR in conjunction with pressure-derived FFR by one single guide wire.15
The aim of this study was to investigate the feasibility of such simultaneous measurement of FFR and CFR in humans by one pressure-temperature sensor-tipped guide wire and to compare CFR measurement by thermodilution with Doppler CFR measurement.
Pressure-Temperature Guide Wire
In this study, a commercially available 0.014-inch floppy pressure guide wire (PressureWire-3, Radi Medical Systems) was used with modified software. This wire has a microsensor at a location 3 cm from the floppy tip, which enables simultaneous recording of high-fidelity coronary pressure measurement as well as temperature measurement at the location of that sensor, with an accuracy of 0.02°C.9 The shaft of this wire, acting as an additional electric resistance, can be used as a second thermistor, providing the input signal at the coronary ostium of any fluid injection with a temperature different from blood (Figure 1). All signals can be displayed on the regular catheter laboratory recording system or at a suitable interface (Radi-Analyzer, Radi Medical Systems), enabling on-line analysis as described below. Pressure and temperature are sampled with a frequency of 500 Hz.
Study Population and Catheterization Protocol
Fifty patients were studied (103 coronary arteries), referred for physiological assessment of at least one coronary stenosis. The institutional review boards of both hospitals approved the study, and informed consent was obtained from all patients before the study. Cardiac catheterization was performed as usual; 10 000 U of heparin was administered, a 7F guiding catheter was advanced into the coronary ostium, and 300 μg of nitroglycerin was administered, after which reference images were made.
The sensor-tipped guide wire was calibrated for the pressure recording as usual and advanced to the tip of the guiding catheter to check equality of pressure signals. The temperature signal was then calibrated at that location, which means that the temperature at the coronary ostium was taken as reference temperature for further measurements. Next, the wire was introduced into the coronary artery and advanced across the stenosis in the case of a stenotic artery or to the distal third part of the artery in the case of a nonstenotic vessel. Thereafter, a 0.014-inch Doppler wire (Flowire, Endosonics Inc) was also introduced through the guiding catheter into the same coronary artery and advanced 1 to 2 cm distal to the tip of the pressure wire.
Fractional flow reserve (measured by the PressureWire), coronary flow reserve by thermodilution (CFRthermo, also measured by the PressureWire), and coronary flow velocity reserve (CFRDoppl, measured by the Flowire) were determined as described below.
All stenoses with FFR ≤0.75 were subsequently dilated and/or stented, after which the simultaneous measurement of FFR and CFR was repeated in a similar way. Further treatment of the patients was according to local routine.
Measurement of FFR and CFRDoppl
For all arteries, simultaneous measurements were performed of mean aortic pressure (Pa, by the guiding catheter), mean distal coronary pressure (Pd, by the PressureWire), and average peak velocity (APV, by the Flowire), both at baseline and at steady-state maximum coronary hyperemia induced by intravenous administration of 140 μg/kg per minute adenosine or by intracoronary administration of 15 to 20 mg papaverine.
FFR was calculated by the ratio Pd/Pa at maximum hyperemia. CFRDoppl was calculated by the ratio of APV at hyperemia and at baseline, according to regular clinical practice.
Thermodilution curves in the coronary artery were obtained by short manual injections of 3 mL of saline at room temperature in a similar way as in the animal study.15
It was checked carefully at the monitor that blood flow velocity was not influenced by the injection itself. As already described, the rapid injection was recorded by the input signal at the ostium of the coronary artery (Figure 2) derived from the temperature of the shaft of the wire. Measurements were performed 3 times at baseline and 3 times at hyperemia, after which curve analysis was performed as described below. Care was taken not to advance or pull back the wire during such series of measurements. An example of a registration of Pa, Pd, flow velocity, and temperature is presented in Figure 3.
Analysis of Coronary Thermodilution Curves
According to theory,16–18⇓⇓ flow (F) equals V/Tmn, where V represents the vascular volume between injection site of the indicator (ie, the tip of the guiding catheter) and location of the sensor, and where Tmn is the mean transit time and calculated by
where C(t) is the thermodilution curve registered by the distal thermistor and t=0 is defined as the time halfway the injection, indicated by the temperature change at the ostium of the catheter as described above (Figure 2).
To test the quality of the thermodilution curves objectively, after a semilogarithmic transformation of the acquired data, the second part of the curve was fitted to the best linear fit. Measurements of Tmn were only accepted if the mean square error (MSE) of the fitted curve was ≤0.05; if the time for analysis of the thermodilution curve equaled at least one heart cycle; and if the variability between the 3 values obtained for Tmn from the 3 measurements was <20%. CFRthermo was then calculated as the average Tmn at baseline divided by average Tmn at hyperemia. According to general indicator dilution theory, Tmn is independent of the amount of indicator as long as the indicator does not physically influence blood flow.16–18⇓⇓ Importantly, the injection of saline as described here did not influence coronary blood flow itself, as could be checked by the Doppler signal (Figure 3). The criterion of at least one heart cycle was taken to avoid influence of systolic-diastolic differences in blood flow and the necessity of triggered saline injection.
Baseline Characteristics and Procedural Outcome
One hundred three arteries were studied in 50 patients.
Baseline and angiographic characteristics of the patients are presented in the Table. The studies were uneventful in all patients. FFR could be successfully measured in all arteries, CFRDoppl in 94 arteries, and CFRthermo in 92 arteries. Both adequate Doppler and thermo registrations were obtained in 86 arteries. In 36 stenoses, FFR was ≤0.75, and all these vessels were subsequently dilated and/or stented, after which similar measurements of FFR, CFRDoppl, and CFRthermo were performed successfully in 33 of them. Therefore, complete data with respect to CFRthermo, CFRDoppl, and FFR were available from 119 arteries.
Feasibility and Variability of Thermodilution Curves
The average time needed to perform a complete set of 3 baseline and 3 hyperemic thermodilution curves was 6.4±2.3 minutes.
The variability of Tmn within a set of 3 measurements was 10.5±8.2% for the baseline measurements and 8.0±6.4% at hyperemia (P<0.05).
The duration of injection of the indicator was 0.49±0.10 seconds and ranged between 0.4 and 0.6 seconds in 94% of the curves.
Adequate thermodilution curves were obtained in 733 of 846 studies, corresponding with 87% of all curves. The main reason for not obtaining adequate curves was an interval too short for adequate curve fitting (<1 cycle), mostly accompanied by a variability of >20% in 104 curves.
This latter problem mainly occurred in normal coronary arteries at hyperemia when the course of the artery was short and the sensor could not be placed sufficiently distally. This observation is in accordance with in vitro observations on the relation between curve quality and distance between injection site and measuring site as published by De Bruyne et al15 recently.
Comparison of CFRthermo With CFRDoppl
A fair correlation was found between CFRthermo and CFRDoppl, as presented in Figure 4A:
CFRthermo=0.84× CFRDoppl+ 0.17 (r=0.80; P<0.0001). The average absolute difference between both indexes was 17±14% (range, 0% to 51%). In 26% of all studies, the difference between both parameters was >20%. The corresponding Bland-Altman diagram is shown in Figure 4B.
In a number of patients with coronary artery disease, not only epicardial coronary stenosis is present, but the microvasculature may be involved. Although in such cases FFR still reliably indicates the functional severity of the epicardial stenosis (given that particular state of the microvasculature) and therefore indicates if and to what extent a patient may benefit from mechanical revascularization (PTCA or CABG) of the respective epicardial stenosis, for correctly estimating the contribution of microvascular disease to inducible ischemia, the combination of pressure and flow measurement is mandatory.2,6,13,14⇓⇓⇓ Theoretically both parameters together should yield necessary information to discriminate between increased epicardial or microvascular resistance and their respective contribution to myocardial hypoperfusion.2,7,12,13⇓⇓⇓ Especially in patients with diffuse epicardial disease or a mild or moderate stenosis but without significant pressure drop along the epicardial artery, a low CFR in addition to nonischemic FFR confirms the involvement of the microvasculature and the need for aggressive medical treatment.9 Also, in the subacute or chronic phase after previous myocardial infarction, combination of FFR and CFR will clarify whether it is the residual epicardial stenosis or the damaged microvasculature that is responsible for low flow, with implications for the decision to dilate or to treat the patient medically.8
Therefore, methodology to assess FFR and CFR simultaneously by one single guide wire is important and feasible, as demonstrated in this study. CFR, determined by the thermodilution method, showed a fair correlation to Doppler CFR measurement.
According to thermodilution theory, flow equals vascular volume between injection site and measuring site (V), divided by mean transit time.16 In our study, 300 μg of nitroglycerin was administered at the start of the procedure to offset changes in epicardial vascular volume by flow-mediated vasodilation during hyperemia. Therefore, the vascular volume of the coronary arterial segment can be assumed to remain constant, flow is proportional to inverse Tmn, and CFR can be assessed by the ratio of Tmn at baseline and hyperemia.19 As a matter of fact, it is required that the wire is not moved forward or backward between baseline and hyperemic measurements. In contrary to that part of thermodilution theory used for cardiac output measurement, the approach used in this study is independent of the amount of indicator,16,19⇓ which was unknown in this study because of heating up of the saline in the guiding catheter and loss of indicator in the aortic root and in side branches.
Calculation of Tmn requires temperature measurement in the distal coronary artery as well as registration of the injection signal for definition of t=0, obtained in this study by using the shaft of the wire as a second thermistor (Figure 2).
We did not use automatic and ECG-triggered injection of the indicator but manual injection. Therefore, the curve fit should be obtained from data from at least one complete heart cycle. As a consequence, we could not obtain reliable data in some arteries with very high flow, especially not if the sensor could not be introduced far enough to the distal coronary artery.
Theoretically, ECG-triggered injections could solve this problem but would make the procedure more complex, whereas in the experimental study triggered injections were not better than manual injections.15 For reasons of safety and simplicity, we chose saline at room temperature to be taken from the reservoir on the table. Because of the extreme sensitivity of the sensor, 3 mL of saline at room temperature was sufficient to obtain adequate curves in the majority of the studies.
The relation between CFRthermo and CFRDoppl was fair but not excellent, with an absolute difference of 17±14% between both values. It should be noted, however, that Doppler does not provide a true gold standard for CFR in humans and has an intrinsic variability of at least 15% to 20%. The use of PET could have yielded a better gold standard for CFR, but in that case, simultaneous acquisition of CFRthermo would not have been possible.
Advantages and Limitations of CFRthermo
The method of simultaneous measurement of FFR and CFR as described in this study is safe because only small amounts of saline are administered. It can be rapidly performed and is not associated with any extra costs compared with the present physiological measurement of FFR alone, and there is no need for extra hardware.
However, there are some limitations. First, the success rate was 87%, compared with a success rate of 91% for successful Doppler CFR measurement (NS) and a success rate of 100% for FFR measurement.
Second, the differences between CFRthermo and CFRDoppl were in the range of 20%.
Third, CFR by thermodilution may be overestimated in the case of large sidebranches, especially if such a sidebranch originates closely proximal to the stenosis and the sensor is located shortly distal of the stenosis. This limitation is not very prominent for proximal stenoses but may lead to overestimation of CFR by thermodilution in the case of stenoses in the mid or distal part of a coronary artery.
Another limitation compared with Doppler is that steady-state hyperemia should be present during at least 30 seconds to perform the manipulations necessary to obtain the hyperemic thermodilution curve. Therefore, unlike for Doppler CFR, intracoronary adenosine cannot be used and intravenous adenosine or intracoronary papaverine are mandatory. Although this is a limitation on one hand, the use of intravenous adenosine or intracoronary papaverine has the advantage of being more accurate to determine FFR and offers the possibility to analyze sequential and diffuse abnormalities along the complete coronary artery during a so-called pressure pull-back curve, once FFR and CFR have been determined.6,9,20⇓⇓
Next, intravenous adenosine is accompanied by a decrease of blood pressure of ≈10% to 15%, and therefore CFR may be underestimated by 10% to 15% if not corrected for these pressure changes. Such a correction can be made as described earlier.19 In the present validation study, such a correction was not necessary because CFRthermo and CFRDoppl were both measured with intravenous adenosine or intracoronary papaverine simultaneously and therefore were affected in the same way by hemodynamic variations.
Careful attention should be given to the position of the guiding catheter. On one hand, it needs to be sufficiently engaged in the coronary artery to guarantee adequate delivery of the indicator into the vessel; on the other hand, easy backflow into the aortic root is necessary to avoid mechanical influence of the injection on baseline blood flow, resulting in underestimation of CFR.
Finally, as a matter of fact and in contrast to FFR, correct interpretation of CFR may be difficult, and CFR determined by thermodilution has the same limitations inherent to the concept of CFR in general, like the aforementioned dependency on heart rate, blood pressure, age, interindividual variability between persons, and its dependency on true baseline flow, which is often hard to achieve in the catheterization laboratory.2,4⇓
Only qualitative assessment of microvascular disease by combination of pressure and flow data has been described so far.7,14⇓ Further studies are mandatory to develop suitable algorithms for true quantitative description of the microvasculature. Such studies may be facilitated by having available the present methodology.
In conclusion, despite several conceptual limitations and issues to be solved, this study shows the feasibility of simultaneous measurement of FFR (by coronary pressure) and CFR (by coronary thermodilution) by a single guide wire in humans in a practical and straightforward way and will facilitate the assessment of microvascular disease.
This study was supported in part by a grant from Radi medical Systems, Uppsala, Sweden. The authors are indebted to Anne Hol for assistance in preparing the manuscript and to Ruud Bogers, Danny de Moor, Henk Kusters, Olivier Nelis, Marie-José Sanders, and Eduard Van Hagen for their assistance in performing the studies.
Dr Smith is an employee of Radi.
- ↵De Bruyne B, Bartunek J, Sys SK, et al. Simultaneous coronary pressure and flow velocity measurements in humans. Circulation. 1996; 94: 1842–1849.
- ↵De Bruyne B, Pijls NHJ, Heyndrickx GR, et al. Pressure-derived fractional flow reserve to assess serial epicardial stenoses: theoretical model and animal validation. Circulation. 2000; 101: 1840–1847.
- ↵De Bruyne B, Pijls NHJ, Bartunek J, et al. Fractional flow reserve in patients with prior myocardial infarction. Circulation. 2001; 104: 157–162.
- ↵De Bruyne B, Hersbach F, Pijls NHJ, et al. Abnormal epicardial coronary resistance in patients with diffuse atherosclerosis but normal coronary angiography. Circulation. 2001; 104: 2401–2406.
- ↵Pijls NHJ, van Son JAM, Kirkeeide RL, et al. Experimental basis of determining maximum coronary, myocardial, and collateral blood flow by pressure measurements for assessing functional stenosis severity before and after PTCA. Circulation. 1993; 87: 1354–1367.
- ↵Bech JWB, Pijls NHJ, De Bruyne B, et al. Usefulness of fractional flow reserve to predict clinical outcome after balloon angioplasty. Circulation. 1999; 99: 883–888.
- ↵Hanekamp C, Koolen JJ, Pijls NHJ, et al. Comparison of quantitative coronary angiography, intravascular ultrasound, and pressure-derived fractional flow reserve to assess optimal stent deployment. Circulation. 1999; 99: 1015–1021.
- ↵Di Mario C, Krams R, Gil R, et al. Slope of the instantaneous hyperemic diastolic coronary flow velocity-pressure relation: a new index for assessment of the physiological significance of coronary stenosis in humans. Circulation. 1994; 90: 1215–1224.
- ↵Meuwissen M, Chamuleau SAJ, Siebes M, et al. Role of variability in microvascular resistance on fractional flow reserve and coronary blood flow velocity reserve in intermediate coronary lesions. Circulation. 2001; 103: 184–187.
- ↵De Bruyne B, Pijls NHJ, Smith L, et al. Coronary thermodilution to assess flow reserve. Circulation. 2001; 104: 2003–2006.
- ↵Zierler KL. Circulation times and the theory of indicator dilution methods for determining blood flow and volumes.In: Handbook of Physiology. Washington, DC: American Physiological Society; 1962: 585–615.
- ↵Rutishauser W, Simon H, Stucky JP, et al. Evaluation of roentgen cinedensitometry for flow measurement in models and in the intact circulation. Circulation. 1967; 36: 951–963.
- ↵Pijls NHJ, Uijen GJH, Hoevelaken A, et al. Mean transmit time for the assessment of myocardial perfusion by videodensitometry. Circulation. 1990; 81: 1331–1340.
- ↵Pijls NHJ, Aengevaeren WRM, Uijen GJH, et al. The concept of maximal flow ratio of immediate evaluation of percutaneous transluminal coronary angioplasty results by videodensitometry. Circulation. 1991; 83: 854–865.
- ↵Pijls NHJ, De Bruyne B, Bech JWB, et al. Coronary pressure measurement to assess the hemodynamic significance of serial stenosis within one coronary artery. Validation in humans. Circulation. 2000; 102: 2371–2377.
- ↵Stefanadis C, Diamatopoulos L, Vlachopoulos C, et al. Thermal heterogeneity within human atherosclerotic coronary arteries detected in vivo. Circulation. 1999; 99: 1965–1971.