Background Coronary steal is defined as a fall in blood flow toward a certain vascular region in favor of another area during arteriolar vasodilatation, ie, a coronary flow velocity reserve (CFVR) <1. The purpose of this study was to determine the frequency of steal in patients with a wide range of collateral supply to a vascular area of interest and to assess whether steal is associated with the amount of collateral flow.
Methods and Results One hundred patients 57±9 years old with a coronary artery stenosis to be dilated were examined with intracoronary (IC) Doppler guidewires. IC adenosine–induced CFVR<1 obtained distal to the stenosis was defined as steal. An index for collateral flow was determined by positioning the Doppler guidewire in the collateral-dependent vessel distal to the stenosis and measuring the flow velocity time integral during (Vioccl, cm) and after (Viø-occl) balloon occlusion. Vioccl/Viø-occl was determined without and with intravenous adenosine (140 μg · kg−1 · min−1). Coronary steal occurred in 10 of 100 patients. Patients with steal showed superior collaterals compared with those without steal: Vioccl/Viø-occl=0.65±0.24 in patients with steal versus 0.29±0.18 in those without steal (P=.0001). In all patients with steal, there was a reduction in collateral flow during intravenous adenosine–induced hyperemia, whereas in the majority (70%) of patients without steal, collateral flow increased or remained unchanged during hyperemia.
Conclusions Coronary steal assessed by intracoronary Doppler flow velocity measurements occurs in 10% of patients with a wide range of coronary collaterals to the vascular area from which blood flow is redistributed. There is a direct association between the presence of steal away from and the amount of collateral flow toward the region under investigation. Collateral flow to the vascular region studied decreases during adenosine-induced hyperemia, which indicates a mechanism of steal via the extensive collaterals.
The coronary collateral circulation is an alternative source of blood supply to the myocardium jeopardized by the failure of the original stenotic or occluded vessel to provide adequate blood flow to this region. The functional relevance of coronary collaterals in humans has been debated for many years.1–3 Numerous studies have demonstrated a protective role of collaterals in hearts with CAD, showing smaller infarcts,4 less ventricular aneurysm formation, improved ventricular function,4,5 and better survival3 compared with patients in whom collaterals were not visualized. Conversely, it has been stated that “coronary collaterals in man are more an indication of severe regional ischemia than a sign of biological ’compensation’ for a perfusion deficit.”6 Furthermore, some reports have suggested negative aspects of well-developed versus poorly developed coronary collaterals, such as more frequent restenosis after angioplasty of atherosclerotic lesions proximal to a vascular region with good collateralization.7 The beneficial effect of nifedipine on variables of myocardial ischemia has been demonstrated to be impaired in patients with “good” collaterals,8 and it has been suggested that a redistribution of blood via collaterals away from the myocardial area in need could be one of the mechanisms responsible for the proischemic complications related to dihydropyridine calcium channel blockers.9 These variations in regional blood flow under conditions of microvascular dilatation are called coronary artery steal, and they are estimated to occur in 10% to 30% of patients with CAD undergoing dipyridamole stress testing and nuclear perfusion imaging.10 Steal is defined as an absolute or relative fall in coronary blood flow to a certain vascular region in favor of another supply area under conditions of hyperemia, ie, a coronary flow reserve of <110,11 (Fig 1⇓). The mechanism for this redistribution of blood flow can be a fall in perfusion pressure at the origin of collateral vessels due to proximal stenoses of coronary arteries supplying the collaterals or due to proximal viscous friction developing at high-flow states even in normal coronary arteries from which the collaterals arise.12–14 There are different categories of coronary steal, ie, steal via collaterals, “vertical” steal between different layers of the myocardium,10 and coronary artery branch steal among adjacent vascular areas originating from a common branch bifurcation.11
So far, direct documentation of steal has been possible only experimentally.12–14 With the advent of small intravascular Doppler angioplasty guidewires (0.014 in, 0.33 mm in diameter), it has become feasible to directly verify the consistent but indirect and qualitative findings of noninvasive methods on pharmacologically induced steal phenomena.15,16 Thus, the purpose of this study was (1) to document coronary steal, (2) to determine its frequency in a patient population with CAD and a wide spectrum of collateral supply, and (3) to assess the relation between steal and the amount of collateral supply to the vascular region under investigation.
One hundred patients (57±9 years old; 21 men, 79 women) with one- or two-vessel CAD were included in this study. All of them underwent PTCA because of CAD-related symptoms of at least one stenotic lesion.
The present investigation was approved by the institutional ethics committee, and the patients gave informed consent to participate in the study.
The study population was divided into two groups according to the presence of coronary steal distal to the coronary stenosis undergoing PTCA (“steal” group, ie, CFVR determined by intracoronary Doppler flow wires <1 obtained three times repetitively; Fig 2⇓) and according to absent coronary steal (“no steal” group, ie, CFVR≥1).
Cardiac Catheterization and Coronary Angiography
Patients underwent left heart catheterization for diagnostic purposes. Premedication consisted of 10 mg chlordiazepoxide administered orally 1 hour before the procedure. Aortic pressure was measured with the PTCA guiding catheter. Biplane left ventricular angiography was performed in all patients, followed by diagnostic coronary angiography. Coronary artery stenosis severities were estimated qualitatively as percent diameter stenosis. Angiographic collateral degrees (0 to 3) were determined according to the extent of epicardial coronary artery filling via collaterals with contrast medium from the contralateral side: 0, no filling of the ipsilateral vessel via collaterals from the contralateral side; 1, small side branches filled; 2, major side branches of the main epicardial vessel filled; and 3, main epicardial vessel filled by collaterals from the contralateral side.17
Intracoronary Doppler Flow Velocity Measurements
IC Doppler flow velocity measurements were performed with a 0.014-in (0.33-mm diameter) PTCA Doppler guidewire with a 12-MHz piezoelectric crystal at its tip (FloWire, formerly Cardiometrics Inc, now Endosonics). This Doppler guidewire has recently been validated and, compared with electromagnetic flowmeters, has been shown to measure phasic flow velocity patterns accurately and to track changes in flow rate linearly.18
CFVR values were determined by dividing maximum hyperemic peak flow velocity averaged over three consecutive cardiac cycles (APV) by APV during resting conditions. Hyperemia was induced pharmacologically with either IC bolus injection of 18 μg adenosine for the left coronary artery and 12 μg adenosine for the right coronary artery or intravenous infusion at a rate of 140 μg · min−1 · kg−1 body mass.19 An index of coronary collateral flow to the balloon-occluded vascular region of interest relative to normal resting flow during vessel patency, ie, after completed revascularization, was determined as the ratio of flow velocity time integral distal to the occluded stenosis (Vioccl, cm) divided by that obtained at the identical location after PTCA (ie, not occluded, Viø-occl, cm): Vioccl/Viø-occl (Fig 3⇓).20 In patients who showed bidirectional flow velocity signals (Fig 3⇓), the ratio between antegrade and retrograde flow velocities was identified as the absolute flow velocity.
After diagnostic coronary angiography, an interval of at least 10 minutes was allowed for dissipation of the effect of the nonionic contrast medium (iopamidol 755 mg/mL) on coronary flow velocity and vasomotion. An IC bolus of 0.2 mg nitroglycerin was given to maintain epicardial coronary artery calibers constant and thus to prevent the influence of changing epicardial vessel diameters on CFVR measurements.21 CFVR measurements were obtained proximal and at least three times distal to the stenosis to be dilated. Repetitive distal CFVR measurements were averaged. After distal CFVR measurements, the distal flow velocity time integral, Vioccl, was determined repetitively during balloon occlusion without intravenous adenosine. During balloon inflation, the occurrence of chest pain and ischemic changes on the IC ECG22 were observed. After balloon deflation and cessation of reactive hyperemia, occlusive distal Vioccl was determined during adenosine infusion after a 10 mm Hg systolic blood pressure decrease. Blood pressure and heart rate were recorded continuously during all flow velocity measurements, including nonocclusive, “normal” flow velocity time integral Viø-occl at the same distal location as Vioccl, the former of which was recorded after completion of PTCA and after cessation of reactive hyperemia.
Between-group comparison of demographic, angiographic, hemodynamic, and Doppler flow velocity data were performed by an unpaired two-sided Student’s t test. Hemodynamic data during different time points among patients of the same group were analyzed with a paired t test. A χ2 test was used for comparison of categorical variables between the two study groups. Linear regression analysis was applied for analysis of an association between collateral flow indices and intravenous adenosine–induced change in the indices. Statistical significance was defined at a value of P<.05.
The entire study population comprised 100 patients 57±9 years old: 10 were in the group with coronary steal, and 90 showed no steal. There were no statistically significant differences among the two groups regarding the patients’ age, sex, hemodynamic variables during diagnostic cardiac catheterization such as heart rate and blood pressure, number of coronary arteries with stenotic lesions on the coronary angiogram, left ventricular angiographic ejection fraction, and the coronary artery with the stenotic lesion to be dilated (ie, the index stenosis; Table 1⇓).
Patients revealing a coronary steal phenomenon had a more severe index stenosis than those without steal: 91±10% versus 79±16% diameter stenosis (P=.04; Table 1⇑). Among patients with a two-vessel CAD, the diameter stenosis of the lesion of the contralateral, collateral-supplying vessel was 50±55% in the group with steal and 31±65% in the group without steal (P=NS). Complete occlusion of the stenoses under consideration occurred more often in patients with steal than in those without (Table 1⇑). Proximal location of the stenosis to be dilated was found in 7 of 10 patients with steal and in 40 of 90 patients without steal (P<.05). During balloon occlusion of the stenosis to be dilated, patients with coronary steal suffered from angina pectoris significantly less often and showed signs of myocardial ischemia on IC ECG less often than those without steal (Table 2⇓). Angiographic collateral degree was 2.4±0.6 in the group with steal and 1.0±0.8 in the group without steal (P=.0001).
Doppler Flow Velocity Data
In agreement with the definition of coronary steal (CFVR<1), CFVR measured via the Doppler flow velocity guidewire distal to the index stenosis before PTCA was 0.7±0.5 among patients with steal and 1.9±0.7 among those without steal (P=.0001; Table 2⇑). Doppler-derived IC collateral flow index (Vioccl/Viø-occl) at rest (ie, without intravenous adenosine–induced hyperemia or microvascular dilatation) obtained distal to the stenosis and calculated as the flow velocity time integral (Vi, cm) during balloon occlusion of the stenosis (Vioccl, cm) divided by Vi after completed PTCA and after cessation of postocclusive hyperemia (Viø-occl, cm) was significantly higher in patients with steal than in those without steal (Fig 4⇓). The Vioccl/Viø-occl values at rest measured in the groups with and without steal indicate that collaterals from the contralateral side to the vascular region of interest provide 65% and 29%, respectively, of the normal blood flow via the revascularized ipsilateral coronary artery.
Association Between Collateral Flow Index and Steal
Determination of collateral flow index during intravenous adenosine–induced hyperemia compared with resting conditions showed reduced values of Vioccl/Viø-occl in all the patients with steal but increased or constant Vioccl/Viø-occl in 70% of the patients without steal (Fig 4⇑). The standard for steal defined as CFVR<1 could be predicted by a fall in Vioccl/Viø-occl during hyperemia with 100% sensitivity and 70% specificity. Presence of steal, ie, CFVR<1, was significantly associated with a fall in the flow index via the collaterals, and absence of steal was, on average, associated with an unchanged collateral flow index (Fig 4⇑ and Table 3⇓). Hemodynamic variables such as mean aortic pressure and heart rate changed similarly in response to adenosine infusion between the two study groups (Table 3⇓). In both groups, there was an ≈10 mm Hg decrease of mean aortic pressure (P<.03) but no significant increase in heart rate.
Fig 5⇓ illustrates that there is an inverse relation between the collateral flow index at resting conditions (ie, without intravenous adenosine) and its capacity to increase during intravenous adenosine–induced hyperemia.
This study in 100 patients with CAD provides direct intracoronary evidence of blood flow redistribution during hyperemia away from a collateralized vascular region toward a contralateral, collateral-supplying area, ie, it documents coronary steal in humans for the first time by intravascular Doppler measurements. Among those patients with a wide range of collateral supply to the vascular region studied, coronary steal occurred in 10%. A strong direct association between the presence and the degree of steal away from and the amount of collaterals toward the collateralized vascular territory was found.
Evidence and Occurrence of Steal
The phenomenon of coronary steal has been recognized for more than 25 years,12,23 and it has been studied experimentally,13–15,24,25 suspected clinically,8,26 modeled theoretically,27 and recently demonstrated by PET in humans.15 So far, direct intracoronary evidence has been provided only in two case reports.28,29
Data from the literature on the proischemic effect of the calcium antagonist nifedipine due to suspected coronary steal may serve to estimate the occurrence of this phenomenon found in 10% of the present study population: an increased nifedipine-induced exertional ST-segment depression or exacerbation of anginal symptoms was detected by Loos and Kaltenbach30 in 6% of the patients, by Kober et al31 in 20%, by Stone et al32 in 14%, and by Schulz and coworkers26 in 10% to 20%. Of 1100 cardiac PET studies in patients with collateralized, occluded vascular territories, 75 (ie, ≈7%) revealed coronary steal.10
The rather small difference in the frequency of coronary steal among this and other investigations is probably accidental but may, on the other hand, be explained by the widely varying, predominantly indirect methods to detect steal. Furthermore, it may reflect some of the controversy that existed and may be revived by the present data on the coronary morphological conditions necessary to cause steal.
Coronary Steal and the Collateral Circulation
Coronary steal can be defined as a fall in absolute coronary perfusion of collateralized myocardium after coronary arteriolar vasodilation, ie, myocardial perfusion reserve <1. The definition of steal we chose is slightly different: an index for vascular flow rate reserve, CFVR, of <1 established coronary steal. As long as the epicardial coronary artery caliber is maintained constant by nitroglycerin before the capacity to increase flow is measured, CFVR can be used instead of the CFR for the functional assessment of the coronary circulation without introducing a substantial error.21 However, Gould has asserted that CFR does not correspond to myocardial perfusion reserve at values <1, because “… arterial CFR of less than one does not occur… .”33 This investigation was designed in its present form at the moment when a CFVR<1 could be documented by an intracoronary Doppler guidewire positioned distal to a stenotic lesion after balloon dilatation and stent implantation, thus excluding the possibility of a collapsing stenosis as the cause of a CFVR<1.34 Nevertheless, the possibility cannot be disregarded that elasticity of some of the stenoses under investigation may have partly influenced the results of this study.
The coronary structural conditions for the occurrence of steal were controversial34,35 before Becker13 demonstrated in dogs that dipyridamole-induced coronary steal occurs if collateral vessels supply myocardium downstream of coronary occlusions and even more so, if the collateral-providing vessel is proximally stenotic. Patterson and Kirk14 corroborated those data by showing in dogs with an occluded LAD collateralized via the LCx that increments of vascular resistance proximal to the origin of the collaterals caused a linear increase in the magnitude of coronary steal. However, a certain degree of steal could be observed even in the absence of a stenotic lesion upstream of the collateral-supplying artery. In their analysis of a general network model simulating the collateralized coronary circulation (ie, a Wheatstone bridge model), Demer and coworkers27 predicted the occurrence of coronary steal, ie, myocardial perfusion reserve <1, provided that collaterals from the LCx (with a 60% proximal cross-sectional area stenosis) to an occluded LAD conducted >1% of blood relative to normal maximal coronary conductance. Furthermore, assuming a collateral capacity sufficient to provide normal resting coronary flow in the occluded LAD, ie, 20% collateral conductance, that study estimated that steal begins to take place in the presence of a ≥20% vascular lumen area reduction of the LCx supplying the collaterals.
Thus, it appears unambiguous that the following qualitative elements are indispensable for the occurrence of collateral steal: an occluded or at least highly obstructed coronary artery, coronary collaterals supplying this vessel with blood from a contralateral artery, and a collateral-supplying artery that itself is more or less obstructed proximal to the collaterals (see also Fig 1⇑). The results of the present intravascular Doppler study indicate the presence of all those elements in patients who show coronary steal. There is an association between the qualitative degree of collateral supply to the region of interest and the occurrence of steal from it, ie, patients with coronary steal suffered from angina pectoris less often and showed signs of myocardial ischemia on an intracoronary ECG during stenosis occlusion less often than individuals without steal. An intracoronary Doppler flow velocity–derived index of collateral flow also revealed a quantitative relation between the occurrence of steal and the resting flow through collaterals relative to the antegrade flow after the revascularization of the vessel studied (Fig 4⇑). The Doppler-derived collateral flow index has been demonstrated to accurately distinguish between collaterals to a balloon-occluded vascular region supplying sufficient or insufficient blood flow to this area to prevent myocardial ischemia.20 A reduction of ≈40% in (occlusive) collateral flow during intravenous adenosine in all the patients with steal (Fig 4⇑) further indicates a mechanism of steal via the extensive collaterals and not through local redistribution via a coronary branch adjacent to that under investigation taking off from a common bifurcation (ie, coronary artery branch steal11). The inverse association between collateral flow and the decrease of flow during adenosine-induced arteriolar dilatation (Fig 5⇑) corroborates the notion27 of increasing steal with augmented collateral conductance. However, some aspects of coronary structural characteristics among patients with steal vary from the cited literature, namely, the presence of steal in cases without complete occlusion of the coronary artery under investigation and the finding that steal may occur in patients without stenotic lesions of the collateral-supplying artery. In addition, the collateral flow necessary to serve as a condition for the occurrence of steal is probably much larger than that indicated above (1%),27 and it is possibly as high as 30% to 40% of normal resting flow. Thus, it may be that only patients with a collateral supply to a vascular region sufficient to prevent myocardial infarction at rest20 are at risk to develop coronary steal via these collaterals during hyperemia.
In principle, it is possible that flow velocity signals may not be recorded during coronary occlusion because of obstruction of collaterals through the inflated PTCA balloon or because of malpositioning of the Doppler guidewire with its tip directed against the vessel wall. In the present study, this may have occurred by chance more often during intravenous adenosine than during the first stenosis occlusion, thereby at least partly accounting for the wide variety of collateral flow velocity responses in the no-steal group. However, it would not have changed the incidence of steal, because steal was defined on the basis of CFVR and not of changes in collateral flow in response to adenosine. Conversely, the question can be raised in this context as to whether steal may be occurring in patients showing a nonartificial fall in collateral flow during hyperemia in the presence of a CFVR >1, a theoretical possibility that cannot be further elucidated based on the design of this study.
Clinical Implications of Steal
The presence of coronary steal most often indicates extensive collateral supply to the vascular region under examination and therefore most likely viable myocardium. Coronary steal in the absence of large collaterals is probably quite infrequent,10 and it may be due to hyperemia-induced phenomena of “vertical” blood flow redistribution between epicardial and endocardial vessels10 or to diversion of blood at a coronary artery bifurcation showing certain atherosclerotic morphological changes.11 By attesting to a well-developed collateral circulation, coronary steal may also suggest that after balloon dilatation of a stenosis, there could be flow via collaterals to the region distal to it competing for the antegrade flow and thus heightening the risk for restenosis.7 It can be speculated that collateral steal may be one of the reasons for an inadequate functional result after PTCA, because the impact of hyperemia-induced flow diversion may continue even after completed revascularization.29 In the context of collateral steal, the necessity or adequacy of PTCA, particularly in patients with complete occlusions, may even be questioned, and alternative treatment strategies can be considered, such as medical therapy with β-blocking agents, which have been demonstrated to reduce the incidence of myocardial ischemic episodes independent of coronary collateral flow.8
Selected Abbreviations and Acronyms
|CAD||=||coronary artery disease|
|CFR||=||coronary flow reserve|
|CFVR||=||coronary flow velocity reserve|
|LAD||=||left anterior descending coronary artery|
|LCx||=||left circumflex artery|
|PTCA||=||percutaneous transluminal coronary angioplasty|
This study was supported by a grant from the Swiss Heart Foundation and by a grant from the Swiss National Science Foundation, grant 32–49623.96.
- Received June 5, 1997.
- Revision received August 22, 1997.
- Accepted September 7, 1997.
- Copyright © 1997 by American Heart Association
Helfant R, Vokonas P, Gorlin R. Functional importance of the human coronary collateral circulation. N Engl J Med. 1971;284:1277–1281.
Habib G, Heibig J, Forman S, Brown B, Roberts R, Terrin M, Bolli R, and the TIMI Investigators. Influence of coronary collaterals on myocardial infarct size in humans: results of phase I Thrombolysis in Myocardial Infarction (TIMI) Trial. Circulation. 1991;83:739–746.
Hirai T, Fujita M, Nakajima H, Asanoi H, Yamanishi K, Ohno A, Sasayama S. Importance of collateral circulation for prevention of left ventricular aneurysm formation in acute myocardial infarction. Circulation. 1989;79:791–796.
Gorlin R. Coronary Artery Disease. Philadelphia, Pa: WB Saunders Co; 1976.
Waters D. Proischemic complications of dihydropyridine calcium channel blockers. Circulation. 1991;84:2598–2600.
Gould KL. Coronary collateral function assessed by PET. In: Gould KL, ed. Coronary Artery Stenoses: A Textbook of Coronary Pathophysiology, Quantitative Coronary Arteriography, Perfusion Imaging and Reversal of Coronary Artery Disease. New York, NY: Elsevier Science Publishing Co Inc; 1991:169–178.
Seiler C. In: Nienaber C, Sechtem U, eds. Assessment of Functional Significance of the Stenotic Substrate by Doppler Flow Measurements: Imaging and Intervention in Cardiology. Dordrecht, Boston, London, Kluwer, 1996;pp 295–311.
Becker L. Conditions for vasodilator-induced coronary steal in experimental myocardial ischemia. Circulation. 1978;57:1103–1110.
Patterson R, Kirk E. Coronary steal mechanism in dogs with one-vessel occlusion and other arteries normal. Circulation. 1983;67:1009–1015.
Demer L, Gould K, Goldstein R, Kirkeeide R. Noninvasive assessment of coronary collaterals in man by PET perfusion imaging. J Nucl Med. 1990;31:259–270.
Doucette J, Corl D, Payne H, Flynn A, Goto M, Nassi M, Segal J. Validation of a Doppler guide wire for intravascular measurement of coronary artery flow velocity. Circulation. 1992;85:1899–1911.
Wilson R, Wyche K, Christensen B, Zimmer S, Laxson D. Effects of adenosine on human coronary arterial circulation. Circulation. 1990;82:1595–1606.
Seiler C, Fleisch M, Maier W, Meier B. Koronarkollateralfluss-Beurteilung mittels intravaskulärer Doppler-Flussgeschwindigkeitsmessungen. Schweiz Med Wochenschr. 1997;128–525. Abstract.
Ritter M, Vassalli G, Vieli A, Schuiki E, Amann F, Hess O, Jenni R. How accurately can coronary flow reserve be determined by Doppler velocimetry ? Eur Heart J. 1995;16:202. Abstract.
Meier B, Rutishauser W. Coronary pacing during percutaneous transluminal coronary angioplasty. Circulation. 1985;71:557–561.
Rowe G. Inequalities of myocardial perfusion in coronary artery disease (’coronary steal’). Circulation. 1970;42:193–200.
Fam W, McGregor M. Effect of nitroglycerin and dipyridamole on regional coronary resistance. Circ Res. 1968;22:649–660.
Kern M, Wolford T, Donohue T, Bach R, Aguirre F, Caracciolo E, Flynn M. Quantitative demonstration of dipyridamole-induced coronary steal and alteration by angioplasty in man: analysis by simultaneous, continuous dual Doppler spectral flow velocity. Cathet Cardiovasc Diagn. 1993;29:329–334.
Seiler C, Kaufmann U, Meier B. Intracoronary demonstration of adenosine-induced coronary collateral steal. Heart. 1997;77:78–81.
Becker L. Effect of nitroglycerin and dipyridamole on regional left ventricular blood flow during coronary artery occlusion. J Clin Invest. 1976;58:1287–1296.