Author + information
- Received February 22, 2002
- Revision received June 4, 2002
- Accepted June 17, 2002
- Published online November 6, 2002.
- Ivan P Casserly, MB BCh*,
- Herbert D Aronow, MDMPH*,
- Paul Schoenhagen, MD*,
- Hiroshi Tsutsui, MD*,
- Jennifer Popovich, BA*,
- Marlene Goormastic, MPH*,
- Jeffrey J Popma, MD, FACC†,
- Steven E Nissen, MD, FACC* and
- E.Murat Tuzcu, MD, FACC*,* ()
- ↵*Reprint requests and correspondence:
Dr. E. Murat Tuzcu, Department of Cardiology, Desk F25, 9500 Euclid Ave, Cleveland, Ohio, USA 44195.
Objectives The objective of this study was to examine the relationship between quantitative volumetric and cross-sectional measures of residual atheroma burden and neointimal growth after coronary stenting.
Background Previous intravascular ultrasound (IVUS) studies have demonstrated a correlation between residual atheroma burden and neointimal growth after coronary stenting. However, postmortem studies contradict this finding.
Methods The study population included 34 patients who underwent IVUS six to eight months after stent placement, including 26 patients who underwent IVUS immediately after stent placement and at six to eight months follow-up. Using manual planimetry, the lumen cross-sectional area (LA), stent cross-sectional area (SA) and external elastic membrane cross-sectional area (EEM) were measured at 1-mm intervals after the procedure and at follow-up. Percent neointimal area (NA) and atheroma area (AA) were calculated as: percent neointimal area = ([SA − LA]/SA) × 100; percent AA = ([EEM − SA]/EEM) × 100 in the entire cross section and in individual quadrants. Postinterventional atheroma volume and neointimal volume at follow-up were calculated using Simpsons’s rule.
Results In pooled analyses using all cross sections and cross-sectional quadrants, there was a weak correlation between percent AA and NA (r = 0.11 and 0.12, respectively). Analysis in individual patients demonstrated no significant relationship between total or quadrant measurements of percent AA and NA (p = 0.47 and 0.4, respectively). No relationship between atheroma volume postintervention and neointimal volume at follow-up was observed (r = 0.1, p = 0.62).
Conclusions This study failed to demonstrate a clinically significant relationship between quantitative volumetric and cross-sectional measures of residual atheroma burden and subsequent neointimal growth.
Coronary stents are used in approximately 80% of percutaneous coronary intervention procedures in the U.S. (1). By preventing negative remodeling (2,3), stents reduce the overall rate of restenosis compared with angioplasty alone (4,5). However, stent placement is associated with an increased amount of neointimal tissue formation compared with balloon angioplasty and atherectomy (6,7), resulting in in-stent restenosis rates of 15% to 50% depending on a variety of patient-, lesion- and procedure-related factors (7).
A direct correlation between various measures of atheroma burden and the degree of neointimal growth after stent placement has been reported in nonrandomized intravascular ultrasound (IVUS) studies (8–12). Methods used to describe this relationship include volumetric analyses relating atheroma volume immediately after intervention to neointimal volume at the time of follow-up (usually six months) (11). Other two-dimensional IVUS studies have examined the axial and circumferential relationship between atheroma and neointima (8–11). In contrast, a large postmortem histologic analysis of in-stent restenosis obtained from 55 stents in 32 human patients demonstrated that neointimal thickness was greatest at sites of medial damage independent of whether stent struts were in contact with atheroma versus intact media (13).
The IVUS substudy of the Atherectomy before MULTI-LINK Improves Lumen Gain and Clinical Outcomes (AMIGO) trial offered the opportunity to re-examine the relationship between the quantitative measures of atheroma burden and neointimal growth. An accurate determination of these relationships has important implications for the understanding of the pathogenesis of neointimal growth and also on the clinical practice of lesion “debulking” (i.e., atherectomy) before stenting as a strategy to reduce in-stent restenosis.
The study population consisted of patients from the IVUS substudy of the AMIGO trial. The AMIGO trial was a randomized trial that compared the late angiographic and clinical outcomes of patients undergoing stent placement alone, or stent placement after debulking using directional atherectomy of a single de novo or first-time restenotic lesion in native coronary arteries. In a subset of these patients, IVUS examinations were performed immediately after intervention and at six to eight months follow-up.
Before the IVUS study, patients received 100–200 μg of intracoronary nitroglycerin. Commercially available catheter systems with automated (motorized) pullback were used. These included the 3.5F Endosonics VISIONS Five-64 F/X catheter (20 MHz), the 2.9F and 3.2F BSC Ultracross imaging catheters (30 MHz) or the 2.6Fr Discovery imaging catheter (40 MHz). The same type of catheter and equipment were used at baseline and follow-up examinations in each patient. Motorized pullbacks were performed at 0.5 mm/s and were recorded on super-VHS videotape for offline analysis at the Cleveland Clinic Foundation IVUS core laboratory.
The entire postintervention and follow-up pullbacks were digitized. Cross sections at 1-mm intervals were selected for analysis and single-slice two-dimensional cross sections were selected at 1-mm intervals for quantitative analysis. Manual planimetry at each cross section was performed and measurements were obtained at each cross-section (Fig. 1), including lumen area (LA), stent area (SA) and external elastic membrane area (EEM). In cross sections acquired immediately postintervention, LA and SA were identical. Atheroma area (AA) and neointimal area (NA) were calculated as follows: AA = EEM − SA; NA = SA − LA.
Axial relationship between atheroma and neointima
The axial relationship between atheroma and neointima was determined by examining the relationship between a measure of total atheroma and total neointima in the samecross section from follow-up IVUS images. Using the data from all cross sections at six to eight months follow-up with complete measurements of LA, SA, and EEM, we calculated percent NA ([SA − LA)/SA] × 100) and percent AA ([EEM − SA)/EEM] × 100) in each cross section.
Circumferential relationship between atheroma and neointima
The circumferential relationship between atheroma and neointima was determined by examining the relationship between a quantitative measure of atheroma and neointima distribution in the samecross section from follow-up IVUS images. Using the IVUS pullbacks obtained at six to eight months follow-up, five representative cross sections along the length of the stent were identified in each patient using a set of predefined rules: for stents ≤20 mm in length, cross sections at 3 and 6 mm from the stent edges and at the center of the stent were analyzed; for stents >20 mm in length, cross sections at 4 and 8 mm from the stent edges, and at the center of the stent were analyzed. Where nonmeasurable borders (because of calcification or artifacts) existed at these predefined sites, the nearest proximal cross section was included in the analysis.
Each cross section was divided into four quadrants such that one of the quadrants contained the maximal AA. This was achieved using two intersecting perpendicular lines placed over the image with the intersection point of the lines being placed at the center of the lumen (Fig. 2). Within each quadrant, the EEM, AA, SA, and NA were determined by planimetry. Percent AA (AA/EEM) and percent NA (NA/SA) for each quadrant was calculated as previously described. Where there was a uniform distribution of atheroma in all four quadrants, perpendicular lines extending from 0° to 180° and 90° to 270° were constructed and the relevant measurements made in each quadrant.
Volumetric analysis relating atheroma volume postintervention to neointimal volume at follow-up was performed in those patients with IVUS pullback data immediately after intervention and at follow-up and in which complete planimetry of the appropriate borders in ≥70% of cross sections in a given pullback was possible. Volumes were calculated using Simpson’s rule:
Atheroma volume post-intervention:Neointimal volume at follow-up:Percent atheroma and neointimal volumes were derived as follows: percent atheroma volume = atheroma volume/vessel volume; percent neointimal volume = neointimal volume/stent volume.
To assess the reproducibility of planimetered measurements, 20 cross sections from 20 different patients were randomly selected. Using manual planimetry, two independent observers measured total EEM, SA and LA in each cross section. To assess the reproducibility of the method we used to quantify the circumferential relationship between atheroma and neointima, two independent observers performed manual planimetry of the EEM, SA, NA and AA in the quadrant of these cross sections that were deemed to have the maximal AA.
Continuous variables are reported as means with standard deviations and categorical variables as frequencies with percentages. Simple linear regression was used to relate the percent AA (or volume) to percent NA (or volume). A one-sided ttest was used to assess whether the mean linear regression slope (i.e., β parameter estimate) across all patients differed from zero.
The entire cohort of 34 patients was included in the cross-sectional analysis. Of these, 26 patients had IVUS pullbacks in which ≥70% of the cross sections had complete planimetry of EEM, SA, and LA borders and were included in the volumetric analysis. The clinical and procedural characteristics of the entire cohort are shown in Table 1. The MULTI-LINK ACS was the only stent used in the study with an average stent length of ∼23 mm. All but two patients had a single stent inserted. Follow-up IVUS examination was performed as part of the study protocol in 31 patients and was performed for a clinical indication in the remaining three patients. The angiographic restenosis rate (i.e., angiographic stenosis >50%) for the 34 patient cohort was 12.5%.
Axial relationship between atheroma and neointima
A total of 757 cross sections at 1-mm intervals were acquired from IVUS pullbacks of the stented region in 34 patients. Thirty-three cross sections (4.4%) were excluded because of calcification, 41 (5.4%) because of branch vessels, and 51(6.7%) because of shadowing from stent struts. The remaining 626 (83%) cross-sections with measurable EEM, LA and SA borders were included in the analysis.
In the pooled analysis of these 626 intrastent cross sections, the mean percent total AA and NA were 51.3 ± 9.3% and 28.1 ± 20.2%, respectively. A weak positive correlation between percent residual AA and percent NA was found (r = 0.11; Fig. 3). In 297 intrastent cross sections from the 15 patients who had percent NA ≥50% in any cross section, a greater but still weakly positive correlation between percent AA and NA was present (r = 0.24, r2= 0.06).
When linear regression analyses were performed for each individual patient, a statistically significant positive correlation between percent NA and AA was found in eight patients, a significant negative correlation was seen in another four patients, and no statistically significant relationship was seen in the remainder. The mean of all regression line slopes was 0.129 ± 1.033, which was not statistically different from zero (one-sided p value 0.47).
Circumferential relationship between distribution of atheroma and neointima in each cross section
In the pooled analysis of 556 quadrants from 139 representative cross sections identified in the 34 patients included in this analysis, a weak positive correlation was observed between percent atheroma and neointimal areas in each quadrant (r = 0.12). A statistically significant positive correlation was found in four patients and a statistically significant negative correlation was found in four patients. No correlation was found in quadrant measurements from the remaining 26 patients. The mean of all slopes was 0.074 ± 0.495 (one-sided ttest, p = 0.4).
No statistically significant relationship was observed between atheroma volume after intervention and neointimal volume at follow-up in 26 patients with an IVUS pullback performed at both time points (r = 0.1, p = 0.62) (Fig. 4).
The intraclass correlation coefficients for measurements of total EEM, SA, and LA were 0.93, 0.98, and 0.95, respectively. For measurements of the EEM, SA, NA and AA in the quadrant of the cross section with the maximal AA, the intraclass correlation coefficients were 0.89, 0.76, 0.73, and 0.92, respectively.
Despite using multiple cross-sectional and volumetric analytic methods, in a prospective randomized study we were unable to reproduce the findings of previous IVUS studies that demonstrated an association between the residual atheroma burden and subsequent neointimal growth. Our pooled cross-sectional analyses found that only 1.2% to 1.5% of the variation in percent NA could be attributed to percent AA. When cross-sectional analyses were performed for individual patients, we found no consistent association between these two variables. Finally, we found no relationship between residual atheroma burden and follow-up neointimal growth using volumetric analysis. These findings suggest that any apparent relationship between atheroma burden and neointimal growth is not clinically relevant.
The disparity between our findings and those of previous IVUS studies is not easily explained. Differences in methodology exist between our study and those of most previous investigators. We examined the axial and circumferential relationship between atheroma and neointima in the same cross section from IVUS pullbacks at follow-up. Although Prati et al. (11)used a similar methodology to examine the axial relationship between atheroma and neointima, the remaining IVUS studies have reported the axial and circumferential relationship between pre- or postintervention atheroma and follow-up neointima (8–11). Requiring the matching of pre- or postintervention images with follow-up images necessitates the use of cross sections with clear landmarks such as calcification or branch vessels to allow accurate matching of cross sections and cross-sectional orientation.
The advantage of our method is that all cross sections with complete planimetry are included in the analysis, eliminating selection bias. However, it is not clear how this difference in cross-sectional selection would result in the reported difference in the relationship between atheroma and neointima. The major disadvantage of our method is that inferences about the relationship between pre- and postintervention atheroma and neointima require the assumption that no significant change in atheroma occurs between the time of intervention and follow-up. This assumption is supported, however, by volumetric analyses showing no significant change in atheroma volume between the two time points (11). In addition, in a cross-sectional study by Mudra et al. (3), no significant change in AA was found in ∼80% of cross sections between intervention and six-month follow-up.
The method we used to examine the circumferential relationship between atheroma and neointima is novel. Previous studies have used semiquantitative methods to describe this relationship. The study by Shiran et al. (9)only studied those cross sections with both an eccentric distribution of atheroma (maximal/minimal atheroma thickness >2) and significant neointimal growth (neointimal thickness of >0.5 mm). Hibi et al. (8)used a method that required the measurement of an angle between the site of maximal atheroma and neointimal thickness. In this study, it is unclear whether cross sections with a uniform distribution of atheroma were included. Our method included all prespecified cross sections and reported the quantitative relationship between atheroma and neointima in each cross-sectional quadrant.
Differences in patient-, lesion- or procedure-related characteristics may also explain some of the disparity between this study and others. Given the small numbers of patients in this and other studies, it is difficult to correct for such differences. Previous studies have included patients who underwent routine follow-up examinations. It is unclear how accurately these studied patients included in previous analyses reflect the entire population undergoing coronary intervention. The current study, however, is also likely to be subject to patient selection bias, given the 50% follow-up rate in the IVUS substudy of the AMIGO trial, and the binary restenosis rate of only 12.5% in the patient cohort.
One of the peculiar differences between this study and that of Prati et al. (11)is that 11.6% of all cross sections in this study demonstrated no neointimal growth despite having a significant atheroma burden (50.2 ± 13.5%). Such cross sections were completely absent in the study by Prati et al. (11). The presence of such cross sections supports the contention that no consistent relationship exists between atheroma burden and neointima growth.
Despite the disparity between the findings in this study and other previous IVUS studies, there is supportive evidence for our conclusions. In the largest postmortem study of in-stent restenosis in humans, Farb et al. (13)demonstrated that neointimal thickness at stent strut sites was greatest at sites of medial injury. Neointimal thickness was not related to atheroma burden if the media was intact. Grewe et al. (14)have demonstrated in a postmortem model of stent deployment in coronary vessels with type B2 and C atherosclerotic lesions that the highest degree of vessel injury occurs in anatomically nondiseased or only slightly fibrosed portion of the vessel. Because vessel injury score has been shown to be highly predictive of neointimal proliferation in animal models of restenosis (15,16), this suggests that neointimal growth is unlikely to occur maximally at the sites of greatest atheroma burden.
Variation in the cellular or chemical composition of atheroma may explain differing relationships between atheroma burden and neointima and the overall absence of a relationship. This hypothesis is supported by a number of observations. Patients with type B versus type A atheroma have larger lumen loss because of neointimal growth (3). Farb et al. (13)have demonstrated that 44% of stent struts in contact with lipid core (as a result of focal penetration by stent strut) demonstrate +3 inflammation (i.e., >20 inflammatory cells) compared with 3% of stent struts in contact with fibrous plaque. A strong association between an aggressive inflammatory response and an exaggerated neointimal response is suggested from animal studies (17). Moreno et al. (18), using atherectomy specimens from patients with unstable angina, reported that macrophage-rich areas in the plaque tissue were larger in patients with in-stent restenosis versus those patients without restenosis.
The relationship between the various measures of atheroma burden and neointimal growth was a univariate analysis. In the analysis of the axial relationship between total plaque area and NA at follow-up, only those cross sections with planimetry of LA, SA, and EEM could be included in the analysis. Exclusion of cross sections with nonmeasurable borders, which is not a random process, may have altered the relationships that we observed. However, this limitation is common to all cross-sectional IVUS studies and cannot be remedied.
A remarkably consistent lack of a clinically significant relationship between various volumetric and cross-sectional measures of atheroma burden and neointimal growth was found in patients undergoing stenting alone in this study. This finding differs from previous studies and suggests that atheroma burden does not play an important role in the pathogenesis of neointimal growth. If atheroma plays a role in neointimal growth, atheroma composition rather than burden may be the more critical determinant.
- atheroma area
- Atherectomy Before MULTI-LINK Improves Lumen Gain and Clinical Outcomes trial
- external elastic membrane cross-sectional area
- intravascular ultrasound
- lumen cross-sectional area
- neointimal area
- stent cross-sectional area
- Received February 22, 2002.
- Revision received June 4, 2002.
- Accepted June 17, 2002.
- American College of Cardiology Foundation
- Hoffmann R,
- Mintz G.S,
- Dussaillant G.R,
- et al.
- Mudra H,
- Regar E,
- Klauss V,
- et al.
- Hoffmann R,
- Mintz G.S
- Hibi K,
- Honda Y,
- Hayase M,
- Suzuki T,
- Yeung A.C,
- Yock P.G
- Hoffmann R,
- Mintz G.S,
- Mehran R,
- et al.
- Prati F,
- Di Mario C,
- Moussa I,
- et al.
- Endo A,
- Hirayama H,
- Yoshida O,
- et al.
- Farb A,
- Sangiorgi G,
- Carter A.J,
- et al.
- Schulz C,
- Herrmann R.A,
- Beilharz C,
- Pasquantonio J,
- Alt E
- Schwartz R.S,
- Huber K.C,
- Murphy J.G,
- et al.
- Moreno P.R,
- Fallon J.T,
- Murcia A.M,
- et al.