Coronary Collateral Size, Flow Capacity, and Growth
Estimates From the Angiogram in Patients With Obstructive Coronary Disease
Background— Stimulation of coronary collateral growth has potential clinical value, yet techniques to assess such growth in patients are limited.
Methods and Results— A cineangiographic approach to classify the dominant collaterals and to quantify their lumen caliber and flow capacity was developed and validated. For measurement of 0.4- to 1.5-mm-diameter phantoms, mean error ranged from −0.01 to +0.02 mm. To illustrate the utility of such a method, 52 collateral pathways were measured in 13 patients with 17 occluded arteries before and after 10 years of intensive lipid therapy. The mean variance, ς, of 9 separate measurements of each collateral was ±0.101 mm. At pretreatment, collateral diameter averaged 0.50±0.11 mm (SD) (range, 0.3 to 1.4 mm) without tapering or central narrowing. Over 10 years, mean increase in diameter was +16% (P=0.028); in area, +64% (P=0.015); and in estimated flow capacity, +214% (P=0.009). Certain lipoprotein characteristics tended to predict collateral growth. Patients for whom angina disappeared during 10 years had a greater increase in flow capacity than those for whom it persisted (+331% versus 4%; P=0.05).
Conclusions— Coronary collateral diameter can be estimated with a precision of 0.10 mm. Flow capacity of the network is well approximated by measurement of the 2 or 3 largest connections serving an occluded artery. Initial studies with this method show that disappearance of angina is significantly associated with growth in collateral flow capacity. Collateral growth tends to associate with lipid therapy and with certain in-treatment lipid measures.
Received July 17, 2001; revision received October 22, 2001; accepted November 2, 2001.
Coronary collaterals are a network of nascent microvessels without apparent function in the healthy heart. But on coronary occlusion, they may grow within 1 week to perfuse viable distal myocardium.1–9 In vivo methods to estimate human collateral10 could expand our understanding of the rate and regulation of collateral growth and of the impact of angiogenic, vasoactive, or vasoprotective therapies. On first reflection, the prospects of estimating collateral capacity and collateral growth from the coronary arteriogram appear dim. There may be hundreds of such pathways7; most are well below the limit of radiographic resolution. However, the physics of flow through small tubes suggests that those few connections that do grow to become visible (and measurable) carry virtually all of the blood flow into the distal bed. In this study, we examine this idea and describe a new arteriographic technique to measure the caliber and estimate the flow capacity of the dominant coronary collateral connections. We provide examples of how this approach may be used to study collateral growth.
Classification by Type, Location, Relative Size, and Image Quality
We distinguish 4 types of collaterals: septal (SE), atrial (AT), branch-branch in ventricular free walls (BR), and bridging across lesions (BL) (Figure 1). The arterial source and recipient are classified anatomically; for example, l indicates left anterior descending coronary artery; r, right coronary artery (RCA); c, circumflex; m, marginal; d, diagonal; a, acute marginal; and n, atrioventricular nodal branch. Connections, so classified, whose paths can be clearly tracked from source to recipient branch in one or more cine frames and that are deemed measurable are rated as 1, 2, and 3 on the basis of relative size. Thus, a collateral described as BR, d-m, 2 is a branch type, connecting a diagonal source and marginal recipient of secondary relative size, and measurable.
Collateral Measurements: Methods
Collateral images are initially reviewed in an overhead cineangiographic projection system (Vanguard M35C with digital frame counter) at 5-fold optical magnification of true scale. The view and frames demonstrating the course of the individual collateral are carefully selected. This pathway is traced on paper, a visual guide for future reference. Three recognizable locations are specified at roughly equal intervals along this segment and one each on the source and recipient branches, as illustrated in Figure 4A. Three cine frames with representative, best-quality images are carefully selected at each of the 5 locations. Images are selected in diastole5 if they are of good quality.
Computerized Image Analysis
Measurements are made with a Power Macintosh 7100 computer-based system, developed and validated in our laboratory, that makes 480×512 pixel images from optically magnified sections of the cine frame. System software includes the NIH Image package to measure vessel size, using the catheter as a scaling factor and adjusting for image distortion attributable to out-of-plane and pincushion effects11,12 and adjusting local contrast for background uniformity (important to avoid burn out of small vessels; see Figure 2). Cine films are mounted in a Sony SME 3500 digital cine projector and advanced to the various frames selected for collateral, or scaling catheter, measurement. Using a computer-controlled 4-fold optical zoom and 2-fold digital image magnification, a collateral of 0.5-mm true diameter is visualized as 4 mm on the video screen, with a pixel resolution of 0.03 mm.
Nine separate collateral diameter measures are averaged to obtain a single estimate. Accuracy and variability of this 2-point diameter measurement were determined.
Length, necessary for estimation of flow capacity, is estimated in the overhead projector by stepwise caliper summation with catheter scaling using the least-foreshortened angiographic view.
Phantom Measurements: Accuracy and Variability
Short segments of precision steel wire, 0.2, 0.4, 0.6, 0.8, 1.0, and 1.5 mm in diameter, were filmed together with an 8-French (2.66-mm) catheter, 8 inches from the image intensifier in a Philips Optimus N200 Poly-C cine system at 7-inch field of view. These images were measured 3 times each in 3 locations, simulating the 9-measurement method for collateral diameter estimation.
Collateral Network Flow Capacity: Mathematical Model
A chronically occluded artery is typically supplied, distally, by several well-visualized collateral pathways. The flow resistance, R, in mm Hg/mL per min, along each of these separate parallel pathways, is estimated by the well-established Poiseuille relationship13 for laminar tube flow: R=0.5 · μ · L/d4, where μ is blood viscosity (0.03 g/cm per second), L is estimated length (mm), and d is diameter (mm). Collateral perfusion capacity, GT (mL/min per mm Hg), is conductance in a parallel network of linear resistors: GT=1/RT=Σi1/Ri, where Ri is the individual resistance estimate.
Patient Population: Collateral Growth Study
Thirteen patients studied were originally in the FATS trial14 comparing lovastatin plus colestipol (n=5) or niacin plus colestipol (n=5) with conventional (placebo) therapy (n=3) over a 2.5-year treatment period. A consecutive group of 75 patients subsequently continued on the following triple-drug regimen: lovastatin 20 mg BID, colestipol 10 g BID, and niacin 1 g BID or TID (mean 2.4 g/day); 60 patients continued for 10 years. Angiograms were obtained at FATS baseline, 2.5 years later, and again after 2.5 and 7.5 years of triple therapy. All patients signed IRB-approved consent documents. These 13 patients were selected from the group of 60 for at least one persistent total arterial occlusion with measurable collaterals; 19 others had collaterals at baseline but were excluded because of subsequent recanalization of the occlusion (n=4), poor image quality (n=1), or subtotal proximal occlusion at baseline (n=14). Low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol, triglyceride, lipoprotein(a), and apolipoprotein E were recorded at baseline and averaged from 7 measurements during the ensuing 10 years.
Measurement of Collateral Growth From Serial Angiograms
Baseline and 10-year films were viewed side by side by two observers blinded to patient identity, therapy, and film temporal sequence. Using the methods described above, collateral diameters were measured at the same locations in comparable views.
The significance of changes in lipid parameters and in collateral size and flow capacity over 10 years was estimated by Student’s t test.15 Group average estimates are expressed as mean±SD. The relationship between changes in categorical variables (eg, angina class) and estimated collateral flow capacity was assessed by a 2-sample (Mann-Whitney) rank sum test.15 All comparisons were 2-tailed, with P≤0.05 considered significant. For patient-specific independent variables, comparisons were made among subgroups of the n=13 patients; for region-specific variables, among the n=17 recipient arteries.
Collateral Measurement Variability
Within the diameter range of 0.4 to 1.5 mm, the accuracy (ie, the 9-measurement mean error) ranged between −0.01 and +0.02 mm, and the variance (ς) averaged ±0.03 mm, without size-related trends. The 0.2-mm phantom was consistently overestimated by 0.06 mm.
Collateral Diameter: Repeat Measurement Variance
The mean±SD (ς) of 9 individual measurements of diameter were averaged for each of the 52 individual collaterals measured at 2 time points in these 13 patients. The diameters ranged between 0.31 to 1.36 mm; their variance, ς, averaged 0.101 mm, also without size-related trends.
Diameter Variation Along Length of Collateral: Taper or Necking
The local diameter, averaged from 3 selected frames, was estimated at each of 3 locations (Figure 4A) and expressed as a percent of the overall, 9-measurement, mean diameter. The group (n=52) average of these percentages varied by a statistically insignificant amount at each of the 3 locations, no more than 4% of the overall mean diameter, both at baseline and 10 years. Thus, there was no longitudinal tapering or central necking that would invalidate the assumption of uniform collateral diameter essential to the simple estimation of flow capacity.
Relative Flow Capacity of Different Collaterals
The velocity of flow through each of the three septal connections (l, m, and s) in Figure 2 may be estimated roughly as the collateral length divided by the time of dye transit from source branch to recipient branch (frame count×0.033 seconds). In this case, tl=0.4 seconds, tm=0.9 seconds, and ts=2 seconds. Volume flow rate through each connection is (lumen area × velocity). By this method of calculation, assuming equal collateral lengths, relative volume flow rate through the 1-mm-diameter collateral, l, is 6.5 times that through the 0.6-mm channel, m, [(1.0)2/0.4 versus (0.6)2/0.9] and 32 times that through the 0.4-mm s. By this approach, the 2 or 3 largest of these connections in the angiographically visible collateral spectrum (0.3- to 1.4-mm diameter) carry the majority of flow.
A similar conclusion is reached theoretically by estimating the flow capacity (ie, conductance, the inverse of Poiseuille flow resistance) through tubes of different diameter and equal length, connecting in parallel between a source and a recipient vessel. Because the flow capacity in each tube is proportional to the fourth power of diameter, a tube of 1-mm-lumen diameter would have a capacity 7.7 times that of a 0.6-mm tube, 39 times that of a 0.4-mm tube, and 1960 times that of a 0.15-mm tube. By this line of reasoning, these 3 largest, angiographically visible connections would carry more flow than 2000 parallel, angiographically invisible connections of 0.15-mm diameter. Thus, both directly and theoretically, we conclude that an estimate of the flow capacity of a collateral network requires measuring only the 2 or 3 largest visible connections.
Mean lipid levels (n=13) at baseline were as follows: LDL-c, 206±50; HDL-c, 42±6; and triglycerides, 197±90 mg/dL. During therapy, these averaged 112±46, 51±6, and 149±48 mg/dL, respectively.
Collateral Growth Over 10 Years
The 13 subjects studied had, throughout the study, 17 totally occluded major coronary arteries supplied distally by an average of 3.1±1.9 measurable collaterals (range, 1 to 5).
Lumen Diameter and Area
Mean diameter at baseline for the several measured collaterals serving each of these 17 recipients averaged 0.50±0.11 mm. Diameter at 10 years was 0.57±0.15 mm (P=0.028), a +16% average percent change (Figures 4⇓⇓A, 4B, 5, and 6). Similarly, lumen area averaged 0.21±0.10 mm2 at baseline and 0.29±0.17 at 10 years (P=0.015), a 64% average change.
Source and Recipient Branches
Over 10 years, source branch diameter increased from 0.94 to 1.10 mm (P=0.02); recipient diameters increased from 0.80 to 0.92 mm (P=0.02).
Overall, the estimated flow capacity for collateral networks supplying 17 recipient arteries increased from 0.67 to 1.44 mL/min per mm Hg (P=0.009). The percent increase in capacity for the 17 recipients averaged 214±367%. Assuming a physiologically plausible 30 mm Hg pressure drop across such a theoretically modeled collateral bed, the average flow at baseline would predictably be 20.1 mL/min, and 10 years later, 43.2 mL/min, ≈75% of the normal resting flow demand for the left anterior descending coronary artery bed16 and nearly normal for the RCA.
Effects of Lipid Therapy
Because all were intensively treated during most of the 10 years, it is unclear whether lipid therapy contributed to growth. Three patients with 4 recipient arteries were treated with placebo (n=2) or with colestipol only (n=1) for 2.5 years before starting triple therapy. In this small control group for the effects of lipid therapy, the annualized mean percentage rate of diameter change during the first 2.5 years was −5.8±5.7% decrease versus +9.3±7.3% annual increase over the subsequent 7.5 years (P=0.1). Consistent with this small subgroup trend, a single patient with familial hypercholesterolemia who stopped triple therapy for 3.5 years had substantial collateral shrinkage (Figure 4B).
Effects of Certain In-Treatment Lipid Levels
The average percentage growth of regional collateral flow capacity was compared for groupings of patients with specific lipid measurements above or below the 13-patient median values. Although not statistically significant, those with higher HDL-c and apolipoprotein E17 and with lower LDL-c/HDL-c ratio and lipoprotein(a) tended toward greater growth (Table).
Effects of Baseline Regional Ischemia or Infarction
The presence of Q-waves in baseline ECG or ischemic ST depression (≥1 mm) in baseline exercise test was examined as a potential correlate of collateral growth (Table). Mean estimated regional flow capacity increased approximately equally for those with and without Q-waves or ischemia.
Effects of Collateral Growth on Angina
Angina was classified quarterly for 10 years, using NYHA criteria (I through IV).18 None of the 13 patients had worsening of angina. Six remained in class I and two in class II. Angina improved by one class in 3 patients and by two classes in 2 patients. Estimated flow capacity increased (Table), on average, to nearly four times baseline among those with improved angina (P=0.02) but by only 49% among those without change (between-group P=0.01). For 7 patients with angina at baseline, change in flow capacity correlated with change in angina class (P=0.05, Mann-Whitney).
Effects of Collateral Type
Flow capacity and its growth for collaterals of different type (septal, atrial, branch, and bridging) were estimated (Table). Atrial collaterals were the largest. All collateral types increased capacity, although not significantly for any of these small subgroups.
Angiographic estimation of the size and capacity of human coronary collaterals is described. With this system, the variance of 9 repeated measures of lumen diameter averaged 0.101 mm. The typical occluded recipient artery receives 3 measurable collaterals of 0.3- to 1.4-mm diameter, averaging 0.50±0.11 (SD) mm at baseline and increasing by 16% (P=0.028) after 10 years of intensive lipid therapy. Estimated flow capacity to recipient arteries tripled, on average, over 10 years. Lipid therapy seemed to contribute to growth, but this requires additional confirmation. Improvement in angina was significantly associated with collateral growth.
Study limitations include retrospective analysis of prospectively obtained data, small sample size, possible length underestimation attributable to projection foreshortening and 3D tortuosity, and dependence on the high resolution of cine film. Prospectively defined analyses in a larger sample are needed to confirm observed trends and relationships. Estimates of flow capacity are based on physically appropriate assumptions of laminar flow at low Reynolds number and, as measured, uniform diameter; they could be inaccurate because of deviations from these assumptions.
Mechanisms and determinants of collateral growth have long been studied.19–21 Lipid therapy may increase production or availability of nitric oxide22–24 or of growth factors (eg, vascular endothelial growth factor).20 Shear stress and ischemia have been debated as growth mediators.25–27 The lengthwise uniformity of collateral diameter seems to support shear stress. Were ischemia a primary stimulant, diameter should be largest distally, which was not the case (Table). Shear stress, inversely related to diameter, provides an adaptive mechanism for growth regulation, the result being evolution toward uniform diameter, as is observed. This method has potential use for studying the natural history of collateral growth and the short-term effects of vasoactive stimuli (eg, exercise and drugs) and of agents infused to stimulate collateral growth.28–31
These cine-film collateral images were projected at high optical magnification, with 0.03-mm/pixel resolution. It is unclear whether the CD-ROM technology (≈500×500-pixel density; ≈0.4 mm/pixel for 7-inch image field) will have adequate resolution for this application. Resolution could be improved to 0.14 mm/pixel by imaging at 5-inch field at 1000×1000-pixel density.
Finally, the typical visible collateral averages 0.5-mm diameter. Thus, some collateral connections could be crossed with a 0.36-mm guidewire, setting the stage for mechanical intervention or collateral-specific growth stimulation to enlarge the perfusion pathway.
This study was supported in part by Grants PO1 HL 30086 and RO1 HL 49546 from the National Heart, Lung and Blood Institute, by the University of Washington Clinical Research Center (NIH # RR-37), and by a grant (DK-35816) to the Clinical Nutrition Research Unit from the National Institute of Diabetes, Digestive, and Kidney Disorders. The efforts of Heather Bruggman in preparing this manuscript are greatly appreciated.
Helfant RH, Vokonas PS, Gorlin R. Functional importance of the human coronary collateral circulation. N Engl J Med. 1971; 284: 1277–1281.
Sabia PJ, Powers ER, Jayaweera AR, et al. Functional significance of collateral blood flow in patients with recent acute myocardial infarction: a study using myocardial contrast echocardiography. Circulation. 1992; 85: 2080–2089.
Eckstein R, Gregg D, Pritchard W. The magnitude and time of development of the collateral circulation in occluded femoral, carotid and coronary arteries. Am J Physiol. 1941; 132: 351–361.
Fam WM, McGregor M. Effect of nitroglycerin and dipyridamole on regional coronary resistance. Circ Res. 1968; 22: 649–659.
White FC, Carroll SM, Magnet A, et al. Coronary collateral development in swine after coronary artery occlusion. Circ Res. 1992; 71: 1490–1500.
Habib GB, Heibig J, Brown BG, et al. Influence of coronary collateral vessels on myocardial infarct size in humans: results of phase I Thrombolysis in Myocardial Infarction (TIMI) trial. The TIMI Investigators. Circulation. 1991; 83: 739–746.
Rogers WJ, Hood WP Jr, Mantle JA, et al. Return of left ventricular function after reperfusion in patients with myocardial infarction: importance of subtotal stenoses or intact collaterals. Circulation. 1984; 69: 338–349.
Brown BG, Bolson E, Frimer M, et al. Quantitative coronary arteriography: estimation of dimensions, hemodynamic resistance, and atheroma mass of coronary artery lesions using the arteriogram and digital computation. Circulation. 1977; 55: 329–337.
Dodge JT Jr, Brown BG, Bolson EL, et al. Intrathoracic spatial location of specified coronary segments on the normal human heart: applications in quantitative arteriography, assessment of regional risk and contraction, and anatomic display. Circulation. 1988; 78: 1167–1180.
Schlichting H. Boundary Layer Theory. Vol. 69. New York, NY: McGraw Hill; 1960.
Fisher L, van Belle G. Biostatistics: A Methodology for the Health Sciences. New York, NY: Wiley-Interscience; 1993.
Ganz W, Tamura K, Marcus HS, et al. Measurement of coronary sinus blood flow by continuous thermodilution in man. Circulation. 1971; 44: 181–195.
Couffinhal T, Silver M, Kearney M, et al. Impaired collateral vessel development associated with reduced expression of vascular endothelial growth factor in ApoE−/− mice. Circulation. 1999; 99: 3188–3198.
Spalteholz W. Die Arterian der Herzwand. Leipzig, Germany: Hirzel; 1924.
Schaper W, Ito WD. Molecular mechanisms of coronary collateral vessel growth. Circ Res. 1996; 79: 911–919.
Ito WD, Arras M, Scholz D, et al. Angiogenesis but not collateral growth is associated with ischemia after femoral artery occlusion. Am J Physiol. 1997; 273: H1255–H1265.
Kersten JR, Pagel PS, Chilian WM, et al. Multifactorial basis for coronary collateralization: a complex adaptive response to ischemia. Cardiovasc Res. 1999; 43: 44–57.
Lazarous DF, Shou M, Scheinowitz M, et al. Comparative effects of basic fibroblast growth factor and vascular endothelial growth factor on coronary collateral development and the arterial response to injury. Circulation. 1996; 94: 1074–1082.
Losordo DW, Vale PR, Symes JF, et al. Gene therapy for myocardial angiogenesis: initial clinical results with direct myocardial injection of phVEGF165 as sole therapy for myocardial ischemia. Circulation. 1998; 98: 2800–2804.