Highly Localized Arterial Remodeling in Patients With Coronary Atherosclerosis
An Intravascular Ultrasound Study
Background Preservation of luminal area and symmetry in the presence of irregular plaques necessitates local expansion of the artery wall.
Methods and Results Cross-sectional dimensions of coronary arteries in 65 patients were measured with the use of intravascular ultrasound. A total of 104 arterial segments were studied, of which 88 had atherosclerosis; 16 served as nonatherosclerotic control segments. Three features of atherosclerotic arterial segments were classified: (1) plaque formation, (2) lumen shape, and (3) shape of arterial external elastic lamina. With our intravascular ultrasound–based three-level classification system, we identified three patterns that accounted for 89% of all atherosclerotic arterial segments: (1) concentric plaque with a circular lumen and a circular external elastic lamina (n=17), (2) eccentric plaque with a circular lumen and an oval external elastic lamina (n=35), and (3) eccentric plaque with an oval lumen and a circular external elastic lamina (n=26). A circular lumen was preserved in 66% of all atherosclerotic arterial segments. Arterial segments with a circular lumen in the presence of an eccentric plaque had a significantly larger lumen area than the other two main groups (P<.05).
Conclusions With our intravascular ultrasound–based classification, we provided information regarding the local remodeling response in the coronary artery wall. In a majority of cases, a circular lumen is maintained. Failure of this highly localized response to be operative may contribute to the development of stenotic lesions at a specific site in the artery.
Atherosclerotic plaques usually appear as eccentric lesions in arterial cross sections.1 2 However, the remaining lumen in a majority of atherosclerotic arteries is seen to be circular, oval, or, rarely, D shaped if arteries are distended to physiological pressures during histopathological preparation.3 Preservation of lumen area and symmetry in the presence of an eccentric plaque necessitates regional expansion of the arterial wall. Indeed, based on histopathological studies, the concept that compensatory enlargement in response to atherosclerosis may delay the development of functionally important lumen stenoses has been introduced and accepted.4 5 Using in vivo intravascular ultrasound (IVUS), we found that compensatory enlargement occurs in 55% of arteries and that in 26% of patients, inadequate compensatory enlargement of the artery at the most stenotic segment of the artery contributed to the development of human coronary artery stenosis.6 All these observations indicate that arterial remodeling is an important factor for preservation of the lumen in human atherosclerosis. With consideration of the irregular formation of atherosclerotic plaques, preservation of a round lumen requires a highly localized expansion of the affected portion of the artery wall. Such a local remodeling of the artery may be detectable in vivo by IVUS.
The purpose of this study was therefore to use IVUS to study the shape of the arterial lumen and the artery itself associated with different atherosclerotic plaque morphologies in native human coronary arteries.
The study group consisted of 61 patients (45 men [mean age, 63±11 years] and 16 women [mean age, 67±13 years]) in whom 88 atherosclerotic native coronary arterial segments were studied. Sixteen nonatherosclerotic control segments were studied from 16 patients; 12 of them were among the 61 patients exhibiting atherosclerotic segments. The study and control groups were selected from 176 consecutive patients undergoing IVUS. Inclusion criteria were (1) clear delineation of the lumen-intimal border and (2) visualization of the external elastic lamina (EEL) for ≥300° of the total circumference of the artery. Exclusion criteria were (1) intervention in the artery before intravascular imaging and (2) marked eccentricity of the IVUS imaging catheter within the coronary artery being studied. One hundred three patients were excluded because an intervention was performed in the artery before intravascular imaging. The remaining 73 patients had stable angina pectoris and underwent diagnostic coronary angiography for evaluation of the severity of coronary atherosclerosis. All patients were considered for percutaneous transluminal coronary angioplasty. Eight patients did not have arterial segments on IVUS fulfilling the inclusion criteria. The control and study groups are described in Table 1⇓. A written informed consent was obtained from each subject.
The IVUS imaging system consisted of an imaging catheter (Sonicath, Boston Scientific Corp) and an SONOS Intravascular System imaging console (Hewlett-Packard). The imaging catheter has a 30-MHz single-piezoelectric crystal transducer mechanically rotating at 1800 rpm within a 3.5F monorail over-the-wire catheter sheath.
The right or left femoral artery was punctured by Seldinger technique, an 8F or a 9F arterial introducer sheath was advanced retrograde over a guide wire, and the sheath was placed in the femoral artery. After angiography, the imaging catheter was introduced through an 8F to 9F coronary guiding catheter over a 0.014-in guide wire. The imaging catheter was advanced across the angiographic lesion to the distal portion of the artery under fluoroscopic guidance, and IVUS imaging was performed during the slow pullback (1 mm/s) of the imaging catheter. X-ray fluoroscopy was used to confirm the coaxiality of the imaging catheter at a region of interest in the coronary artery. IVUS images were recorded on 0.5-in Super-VHS videotape for subsequent review and quantitative analysis.
IVUS images were analyzed off-line with an SONOS Intravascular System. The most stenotic cross section or the closest (either distal or proximal) cross section fulfilling the inclusion criteria within the area of the angiographic lesion was targeted for analysis. A second or third cross section was selected for analysis only if a different shape/pattern was observed. The EEL of the artery was defined as the outer border of the sonolucent zone adjacent to the echo-dense adventitia, as previously described.7 8 9 Images recorded with an imaging catheter that was markedly eccentric within the arterial lumen were not considered acceptable to avoid errors due to possible image distortions related to nonuniform transducer rotation.10 11 Of the 73 patients analyzed, 61 had at least one arterial segment meeting the inclusion and exclusion criteria of this study, and an additional 4 patients exhibited a segment without atherosclerosis.
Atherosclerosis was considered to be present if the maximal thickness of the echogenic endoluminal layer was >0.3 mm, the sonolucent zone thickness was >0.2 mm, or both.7 9 Plaque morphology was described as fibrous if less echo dense or fibrocalcific if more echo dense than the adventitia, according to previously described criteria.7 8 The plaque was defined as concentric if the degree of atherosclerosis was evenly distributed around the full circumference of the lumen. An eccentric atherosclerotic plaque was defined by three criteria: (1) marked wall thickening involving ≤180° of the lumen, (2) a thickest wall–to–opposite wall thickness ratio of ≥1.5, and (3) thickest wall plus opposite wall being greater than the perpendicular wall1 plus wall2 (Fig 1⇓). Wall thickness was measured as the distance from the EEL to the lumen-intimal border. As illustrated in Fig 1⇓, lumen diameters were measured along a line through the center of the lumen and the thickest portion of the arterial wall and as the maximum perpendicular diameter of the lumen. Artery diameters were measured from EEL to EEL along the same lines. Lumen and arterial diameter ratios were calculated as the diameter along the line including the thickest portion of the wall compared with the perpendicular diameter. To assess the arterial lumen area (mm2), the lumen-intimal border was traced. The area within this border and the area within the EEL were determined by the computerized planimetry function in the SONOS intravascular imaging console (Fig 2⇓).
Because a reliable delineation of the internal elastic lamina with IVUS is not feasible in atherosclerotic arteries, we used arterial wall area to indicate the degree of atherosclerosis.12 Arterial wall area was calculated as EEL area minus luminal area. Percent area stenosis was calculated as (artery wall area/EEL area)×100.
The definition of lumen and arterial shape was based on the assumption that the nonatherosclerotic artery would present with a circular lumen and artery, resulting in diameter ratios close to 1. To establish a variation of this measure using the present IVUS technique, with application of the inclusion and exclusion criteria, and using the definitions of this study, the diameter ratios were calculated in the 16 arterial segments without atherosclerosis.7 9 A ratio outside ±1.96 SDs of the ratios derived from our control group was considered to represent an oval lumen or artery.
Geometric Form Index
An independent form index indicative of “roundness” was measured using a Micro-Plan II Image Analysis System (Laboratory Computer Systems) to describe the degree of asymmetry of a circumference. With this device, the contours of the cross section of the lumen and the EEL (representing the shape of the arterial outer circumference) were traced from photographs of each of the 104 segments. This form index is a dimensionless ratio of the enclosed area (A) to the perimeter (P) normalized so that for a perfect circle the index is 1, and for a line, it is 0. The relationship is expressed as Form Index=4/P2. A form index outside ±1.96 SDs of the indexes derived from our control group was considered to represent an oval lumen or artery. The greater the deviation from 1, the less circular and more oval was the artery.
Classification of Atherosclerotic Coronary Arteries
The IVUS definitions of plaque formation and the lumen and EEL diameter ratios were subsequently used to describe the cross section of the atherosclerotic artery segments. Each plaque was classified as concentric or eccentric. Lumen and arterial shape were classified as circular or oval, creating eight possible combinations. The same classification was applied using the geometric form index for lumen and artery symmetry.
Reproducibility of IVUS Measurements
Mean intraobserver and interobserver variabilities of cross-sectional area determinations by planimetry in our laboratory were 2.6±1.9% and 2.4±2.6%, respectively. Mean intraobserver and interobserver variabilities of lumen diameter determinations were 1.7±2.0% and 2.2±1.7%, respectively.
Data are given as mean±SD. Differences between groups in continuous variables were tested with one-way ANOVA with Fisher’s Protected LSD as posthoc test. A χ2 test followed by Fisher’s exact test was used for categorical data for comparison between groups. Correlations between variables were tested with linear regression. A value of P<.05 was considered significant.
Nonatherosclerotic coronary artery control segments (n=16) had a smaller EEL area than the atherosclerotic segments (n=88): 12.7±5.3 and 15.1±4.1 mm2, respectively (P=.043). The nonatherosclerotic lumen was bigger than in the atherosclerotic arteries (8.9±3.5 versus 6.9±2.3 mm2; P=.005). The control population (n=16) was otherwise comparable to the atherosclerotic study population (n=65) (Table 1⇑). All 65 patients in the atherosclerotic study population had a significant stenosis and ≥1 segment that could be evaluated according to the protocol. We found a positive correlation between wall area and EEL area (r=.82; P<.001; n=104).
The mean lumen diameter ratio calculated from the perpendicular lumen diameters as illustrated in Fig 1⇑ was 1.0±0.03 in the control segments. With application of the IVUS definitions for symmetry described in the methods, lumen diameter ratios between 0.9 and 1.1 were considered to represent circular lumens.
The degree of atherosclerosis was evenly distributed around the full circumference of the lumen (concentric plaque) in 22% (19 segments from 19 patients) of the 88 atherosclerotic artery segments. All 19 had circular lumens. Eccentric plaques were found in 78% (69 artery segments from 52 of the 61 patients); 57% (39 segments) of these arterial segments with an eccentric plaque exhibited circular lumens, and 43% (30 segments) had an oval lumen shape (Fig 2⇑).
Average dimensions in the three shape categories described by plaque formation and lumen shape are given in Table 2⇓. Because clear delineation of the EEL requires visualization of ≥300° of the total arterial circumference, the most stenotic atherosclerotic segments could not be classified; this was due to acoustic shadowing and/or thinning of the EEL. As a consequence, the mean area stenosis studied ranged from 51.1±7.7% to 56.7±10.7% (Table 2⇓). EEL area, lumen area, arterial wall area, and degree of area stenosis based on the plaque and lumen shape analyses of the atherosclerotic segments are illustrated in Fig 3⇓. Mean lumen diameter ratios were almost identical in segments with concentric atherosclerosis (lumen diameter ratio, 1.0±0.05) and arterial segments with eccentric plaque and a circular lumen (lumen diameter ratio, 1.0±0.03), whereas consistent with the definition, the group with an oval lumen had an altered lumen diameter ratio (0.8±0.08; P values <.001) compared with the other groups. Arterial segments with an eccentric plaque and a circular lumen were further characterized by an oval artery compared with the group with concentric atherosclerosis (artery diameter ratio, 1.2±0.09 versus 1.0±0.04; P<.001) and the arterial segments with an oval lumen (vessel ratio, 1.0±0.06; P<.001). The wall thickness ratio did not differ between segments with eccentric plaques regardless of whether the lumen was circular (mean wall thickness ratios, 2.7±1.02 and 2.8±1.15, respectively). As expected, both groups differed from those with segments with concentric plaques (ratio, 1.2±0.23; P<.001). The coronary arterial segments with an oval lumen also had a higher proportion of fibrocalcific plaques (19 of 30) than did segments with concentric atherosclerosis (2 of 19) and segments with eccentric plaques and circular lumen (5 of 39) (P.001).
In 25 patients, more than one atherosclerosis-lumen pattern was seen in the examined coronary artery. Eight patients with concentric atherosclerosis also had a segment in the imaged artery with eccentric atherosclerosis and a circular lumen. Fifteen patients exhibited eccentric plaques with segments with both circular and oval lumens, and 2 patients had the three different patterns of atherosclerosis lumen shape in the same coronary artery (Fig 4⇓).
The proportion of left anterior descending, left circumflex, and right coronary arteries did not differ among different plaque-lumen combinations. The different groups were not distinguished by differences in age, gender, localization of the segment in the respective artery, relative localization of the segment to the maximum stenosis in the artery, or maximum area stenosis in the imaged artery (Table 1⇑).
The mean EEL diameter ratio in controls was 1.0±0.35; hence, EEL diameter ratios between 0.9 and 1.1 were considered to define circular arteries. The mean geometric form index was 1.0±0.02 for lumen and 1.0±0.02 for the EEL. Accordingly, form indexes of <0.95 defined an oval lumen, and indexes of <0.95 defined an oval artery. The respective criteria for eccentricity or concentricity of plaque and symmetry of lumen and arterial circumference within the external elastic lamina were subsequently applied for each atherosclerotic arterial segment. The frequency distributions of the resulting plaque-lumen-artery patterns are presented in Fig 5⇓. These data demonstrate that with the use of IVUS-derived diameter ratios to describe shape, (1) 40% (n=35) had an eccentric plaque with a circular lumen and an oval arterial circumference within the EEL, (2) 30% (n=26) had an eccentric plaque with an oval lumen and a circular EEL, (3) 19% (n=17) had a concentric plaque with a circular lumen and a circular EEL, (4) 4.5% (n=4) had an eccentric plaque with a circular lumen and EEL, (5) 4.5% (n=4) had an eccentric plaque with an oval lumen and a circular EEL, and (6) 2% (n=2) had a concentric plaque with a circular lumen and EEL. There were no segments with concentric plaque with an oval lumen and a circular or oval EEL. With application of the geometric form index criteria for shape, the patterns found were restricted to the first four patterns: eccentric plaque/circular lumen/oval EEL in 33% (n=29), eccentric plaque/oval lumen/circular EEL in 33% (n=29), concentric plaque/circular lumen/circular EEL in 22% (n=19), and eccentric plaque/oval lumen/oval EEL in 12% (n=11).
We demonstrated (1) a novel way to evaluate coronary arterial geometry at the time of IVUS imaging, (2) that residual luminal geometry tends to be circular or oval with respect to expanding atherosclerosis, and (3) that the presence of eccentrically placed atheroma does not necessarily determine the shape of the residual lumen. Our discussion focuses on the importance of these findings, the importance of our classification method, factors influencing our results, and implications for the evaluation and treatment of coronary disease.
Description and Classification
In vivo, vessel segments with eccentric atherosclerosis, selected according to our criteria, in a majority of cases demonstrated a crescent-shaped plaque on the cross-sectional images with a gradual thinning toward the normal portion of the artery. There were no plaques bulging into the lumen. Our findings are in accordance with histopathological findings when methods using physiological pressure distention during fixation of the artery are used.3
We developed a descriptive classification system defining the cross section of the artery in terms of concentricity or eccentricity of the atherosclerotic plaque, symmetry of the lumen, and symmetry of the arterial outer circumference of the EEL. Such a three-level classification system that theoretically describes eight classes of arteries has not been previously proposed. The definitions of lumen and EEL asymmetry were based on the assumption that normal arteries are circular. The definition of plaque eccentricity was designed to allow detection of local arterial wall remodeling in response to atherosclerotic plaque formation. This IVUS-based classification was found to be a simple tool to determine whether effective remodeling mechanisms were operative in the arterial wall. The vast majority of arterial segments with concentric plaques had circular lumens and circular EELs. Eccentric plaques were usually associated with either a circular lumen in combination with an oval EEL or with an oval lumen in combination with a circular artery. These three patterns accounted for 89% of all arterial segments. A circular lumen was preserved in 66% of atherosclerotic artery segments. Describing the cross section of the atherosclerotic artery on three levels may contribute to the understanding of the evolving distinctions between atherosclerotic burden and lumen stenosis derived from IVUS data. Our approach provides a simple method to distinguish among lumen changes, adaptive responses in the artery wall, and alterations in the plaque itself, which may prove to be fruitful in longitudinal studies on progression or regression of atherosclerosis. This system may also help investigators evaluate the results of an intervention in a given segment to identify atherosclerotic patterns optimal for a given intervention. The potential use of our system to improve the description of the atherosclerotic artery and its dynamic changes should therefore be considered by investigators in this field. For example, an analysis of the localized remodeling according to our model may help predict the acute recoil after balloon angioplasty as well as the predilection of an arterial segment for restenosis.
Highly Localized Remodeling of the Coronary Artery
Focusing on eccentric wall thickening (considered to reflect eccentric plaque deposition) allowed an analysis of compensatory mechanisms in a portion of the artery wall. We observed preservation of a circular lumen in 57% of these arterial segments. The maintenance of a circular lumen shape was associated with an increase of the artery diameter along the line through the maximum thickness of the eccentric plaque and the center of the lumen. Notably, luminal area was significantly larger in arterial segments with a preserved circular lumen despite a strong overall correlation between EEL area and artery wall area. Also, this group with a circular lumen had a slightly larger mean EEL area. Consequently, the calculated area stenosis was less pronounced in arterial segments with circular lumens than in arterial segments with concentric atherosclerosis as well as segments with eccentric plaques and an oval lumen. The arterial segments were otherwise comparable between groups. Our observations suggest that very precise mechanisms capable of maintaining a virtually circular lumen are operant in the majority of coronary artery segments with eccentric atherosclerotic plaque deposition.
In contrast to the arterial segments with a circular lumen, an oval lumen was strongly associated with a round artery, suggesting inadequate remodeling. The failure to expand the artery wall at sites of these eccentric lesions may thus be an important determinant for development of plaques causing stenoses. In this instance, a given degree of atherosclerosis will result in a more narrowed lumen compared with an artery segment showing local expansion of the affected portion of the artery wall.
Factors Influencing the IVUS Assessment of Atherosclerosis
From histopathological descriptions of atherosclerosis, it is known that the plaques usually have irregular morphologies within the artery that involve varying proportions of the circumference along the artery.2 Consequently, the cross-sectional description of the plaque is dependent on the sampling site. However, in our study, there was no systematic influence of sampling site, but the described highly localized remodeling occurs in any location of any of the epicardial coronary arteries.
There is no generally accepted definition of eccentric atherosclerosis or plaque formation based on IVUS data, and proportions of eccentric plaque in studied coronary arteries are dependent on the definitions.13 In pathological studies, distinctions have been drawn between lesions that are “eccentric” (ie, associated with an arc of normal arterial wall within the lesion) and those that are “concentric” (ie, the atherosclerosis completely encircles the coronary artery).14 Our definition of eccentric plaques was designed to facilitate detection of local compensatory mechanisms in the artery wall. The presence of a normal arc of arterial wall was not necessary, but the bulk of the atherosclerosis had to be contained within one hemisphere of the artery. In our study, the wall thickness ratio was set at ≥1.5 to include mild and moderate atherosclerosis in which the postulated compensatory mechanisms are expected to be fully operative.4 Moreover, the strict imaging criteria for inclusion requiring clear delineation of the EEL for ≥300° of the 360° of the arterial circumference resulted in our evaluating early and intermediate stages of atheroma formation.
Every classification is dependent on its definitions and methods. In our study, the higher proportion of a circular shape of the arterial circumference within the EEL with the form index compared with the IVUS-based diameter ratio probably reflects the fact that the perpendicular artery diameter used in the IVUS-based diameter ratio often does not represent the largest diameter of the artery in this direction. There is a possibility that an oblique imaging plane may have influenced our results. This problem has been minimized by (1) ensuring fluoroscopically that the imaging catheter was coaxial to the artery during the pull-back procedure and (2) using measurement ratios for our definitions of eccentric or concentric plaques that consequently are not influenced by the imaging plane. The diameter ratios for the lumen and EEL are equally influenced by the imaging plane. This ensures that a measured difference represents a true difference in shape between lumen and EEL circumferences. (3) We defined normality from the same patient cohort and with the same imaging and measurement procedure.
The major limitation of our study relates to the potential limitation associated with IVUS technology; imaging of the most severe stenosis is limited by two factors: (1) the IVUS catheter is 3.5F and thus may not cross severe stenoses, and (2) determination of the entire EEL at the site of the most severe stenosis was not possible due to acoustic shadowing from plaque calcification and/or disruption of the EEL. Thus, our findings are currently applicable to mild and moderately severe atherosclerotic stenoses.
The same individual may express compensatory remodeling or lack of remodeling in different portions of the same coronary artery (Fig 4⇑), as indicated by the 25 patients who expressed more than one pattern of plaque-lumen interaction in the same coronary artery. This particular finding favors the concept of localized determinants for regional adaptation in an atherosclerotic artery, such as responses to localized alterations in wall stress or degradation of underlying media and adventitia, rather than systemic factors.15 The outward bulging of the plaque and underlying wall in a majority of our segments with an eccentric plaque may be related to the focal nature of the destruction or degradation of the media underlying the plaque as well as the mechanical effect of intraarterial distention pressure at the site of a thinned or absent (weakened) arterial media. This observation is in agreement with previous pathological observations.2 16 Lack of localized compensatory enlargement was associated with a higher proportion of fibrocalcific plaques. This observation may mirror a less compliant plaque, which reduces a possible, generalized passive pressure–mediated wall expansion. This interpretation is compatible with the concept of McPherson et al.17 However, if calcification is a factor influencing the efficiency of artery wall remodeling, it is likely to be only one of several possible mechanisms for modulation of localized compensatory enlargement, as indicated by the 11 noncalcified atherosclerotic segments presenting without localized enlargement. In their study on 100 human and 328 monkey arteries, Clarkson et al5 failed to detect any interference by the degree of calcification on arterial remodeling. In the portion of their study concerning human arteries, eccentricity of the plaque did not influence medial status. An endothelium-mediated flow-velocity–related enlargement of the uninvolved segment of the artery has been proposed as an alternative explanation.4 A prominent role for locally mediated endothelial effects has also been advocated.15 Nevertheless, the exact mechanisms for compensatory dilatation remain unknown, and tentative explanations are purely speculative.
In conclusion, our IVUS-based classification system provides information regarding highly localized remodeling responses to plaque formation in vivo in human coronary arteries. A circular lumen was maintained in the majority of cases. Arterial segments with a failure of this mechanism to be operant may be prone to develop stenotic lesions. These data in themselves may have implications for ongoing follow-up and evaluation of the growth and development of atherosclerosis and their effect on vascular flow and reactivity.
This study was supported by the Wenner-Gren Foundation, Stockholm; Einar Belvén Foundation; the Swedish Institute; Lee E. Siegel, MD Memorial Fund; Rexford Kenamer, MD; and the Western Cardiac Foundation.
- Received February 25, 1997.
- Revision received April 8, 1997.
- Accepted April 13, 1997.
- Copyright © 1997 by American Heart Association
Glagov S, Zarins CK. Quantitating atherosclerosis: problems of definition. In: Bond MG, Insull W Jr, Glakov S, Chandeler AB, Cornhill JF, eds. Clinical Diagnosis of Atherosclerosis: Quantitative Methods of Evaluation. New York, NY: Springer-Verlag; 1983:11-35.
Tobis JM, Mallery J, Mahon D, Lehman K, Zalesky P, Griffith J, Gessert J, Moriuchi M, McRae M, Dwyer ML, Greep N, Henry WL. Intravascular ultrasound imaging of human coronary arteries: in vivo analyses of tissue characterizations with comparison to in vitro histological specimens. Circulation. 1991;83:913-926.
Nissen SE, Gurley JC, Grines CL, Booth DC, McClure R, Berk M, Fischer C, DeMaria AN. Intravascular ultrasound assessment of lumen size and wall morphology in normal subjects and patients with coronary artery disease. Circulation. 1991;84:1087-1099.
Mintz GS, Popma JJ, Pichard AD, Kent KM, Satler LF, Chuang YC, DeFalco RA, Leon MB. Limitations of angiography in the assessment of plaque distribution in coronary artery disease: a systematic study of target lesion eccentricity in 1446 lesions. Circulation. 1996;93:924-931.
Roberts WC. Coronary heart disease: a review of abnormalities observed in the coronary arteries. Cardiovasc Med. 1977;2:29-49.
Crawford T, Levine CI. Medial thinning in atheroma. J Pathol. 1953;66:19-23.
McPherson DD, Sirna SJ, Hiratzka LF, Thorpe L, Armstrong ML, Marcus ML, Kerber RE. Coronary arterial remodeling studied by high-frequency epicardial echocardiography: an early compensatory mechanism in patients with obstructive coronary atherosclerosis. J Am Coll Cardiol. 1991;17:79-86.