Atherosclerosis has been shown to develop preferentially at sites of coronary bifurcation, yet culprit lesions resulting in ST-elevation myocardial infarction do not occur more frequently at these sites. We hypothesized that these findings can be explained by similarities in intracoronary lipid and that lipid and lipid core plaque would be found with similar frequency in coronary bifurcation and nonbifurcation segments. One hundred seventy bifurcations were identified, 156 of which had comparative nonbifurcation segments proximal and/or distal to the bifurcation. We compared lipid deposition at bifurcation and nonbifurcation segments in coronary arteries using near-infrared spectroscopy (NIRS), a novel method for the in vivo detection of coronary lipid. Any NIRS signal for the presence of lipid was found with similar frequency in bifurcation and nonbifurcation segments (79% vs 74%, p = NS). Lipid core burden index, a measure of total lipid quantity indexed to segment length, was similar across bifurcation segments as well as their proximal and distal controls (lipid core burden index 66.3 ± 106, 67.1 ± 116, and 66.6 ± 104, p = NS). Lipid core plaque, identified as a high-intensity focal NIRS signal, was found in 21% of bifurcation segments, and 20% of distal nonbifurcation segments (p = NS). In conclusion, coronary bifurcations do not appear to have higher levels of intracoronary lipid or lipid core plaque than their comparative nonbifurcation regions.
If coronary bifurcations preferentially contained lipid core plaque (LCP), one would expect them to be the dominant site of plaque rupture and acute coronary syndromes. One study of 385 patients presenting with ST-elevation myocardial infarctions arrived at this conclusion, reporting culprit lesions within 20 mm of a bifurcation in 79% of cases. However, the effect of bifurcation induced shear stress, the presumed mechanism for increased plaque deposition, has been shown to dissipate within 3 mm of a side branch, suggesting that a number of these cases involved lesions outside of the influence of a bifurcation. In fact, culprit lesions were found within 10 mm of a bifurcation only 57% of the time. Fujii et al reported a similar finding using optical coherence tomography; these authors found lipid rich, thin-capped fibroatheromas in close proximity to a side branches in only 51% of cases. Thus, we hypothesized that despite the propensity to develop atherosclerosis at bifurcation segments, plaque lipid would not be found preferentially at bifurcations. Using near-infrared spectroscopy (NIRS), we compared the lipid composition of bifurcation versus nonbifurcation atherosclerosis. We hypothesized that intracoronary lipid and lipid-rich plaques would be found with similar distribution at coronary artery bifurcation and nonbifurcation coronary segments.
Methods
Lipid maps of ≥1 coronary arteries were obtained in 100 patients. Written informed consent as approved by the institutional review board was obtained before angiography. NIRS data were obtained using the Lipiscan catheter (InfraReDx, Burlington, Massachusetts) from patients undergoing coronary angiography. All data were submitted to the Chemometric Observation of LCP of Interest in Native Coronary Arteries Registry (COLOR), a repository of NIRS data maintained by the manufacturer of the catheter used for lipid assessment. Enrollment in COLOR was open to any nonpregnant patient aged over 18 years in whom a native coronary artery was imaged with the device.
In each patient, the NIRS catheter was advanced over a coronary guidewire into at least 1 epicardial artery and then withdrawn at 0.5 mm/s by automated pullback device. The spectroscopic signal generates a chemogram image, a block chemogram, and a lipid core burden index (LCBI). The LCBI is a measure of total lipid signal indexed to the length of an arterial segment. The block chemogram is an assessment for focal lipid-rich plaque within 2-mm blocks (see Figure 1 ). Each block is assessed for the presence of lipid, with a LCP being defined as a yellow chemogram signal, a signal characteristic previously shown to correlate closely with the presence of a lipid-rich plaque.
All scans were reviewed for the presence of major bifurcations marked by the operator at the time of image acquisition. For those with an identified side branch, we determined the LCBI and assessed for LCP in the 10-mm region surrounding the side branch. The LCBI and presence of LCP were also determined for a 10-mm segment proximal and distal to the bifurcation region. If a proximal or distal nonbifurcation segment overlapped with another bifurcation region, it was excluded.
Mean LCBI at bifurcation and nonbifurcation segments were compared, as were the percentage of segments containing LCP. In determining the effects of coronary side branches on LCBI, we had to account for significant zero-inflation (i.e., >20% of the LCBI measures had a value of 0). The nonzero LCBI measures were distributed log-normally.
A 2-part mixed-effects model was designed to handle zero-inflated data with repeated measures to model LCBI (SAS macro MIXCORR, SAS Institute Inc., Cary, North Carolina ). Using this method, we simultaneously fit a logistic regression model for proportion of non-zero LCBI measures for all participants and a linear regression for the log of the mean LCBI level among observations having LCBI values >0. Models were fit with and without correlated random effects using a backward selection procedure. All statistical analyses were conducting using SAS v9.2 (SAS Institute Inc.). Odds ratios were calculated to determine whether any demographic variables correlated with LCP deposition at bifurcation segments. Two-sample t tests were used to compare age and body mass index. We used chi-square tests and, when appropriate, Fisher’s exact test to compare all categorical variables.
Results
The study population included 100 patients, 79 men and 21 women. Demographic and clinical information for all participants is shown in Table 1 . The majority of patients were referred for catheterization due to angina (n = 62) or an acute myocardial infarction (n = 21), with the remainder for atypical chest pain, heart failure, posttransplant surveillance, preoperative evaluation, or silent ischemia (n = 17). Obstructive coronary artery disease was identified in 80% of patients at the time of heart catheterization.
| Characteristic | % ∗ |
|---|---|
| Age (yrs) | 62 ± 11 |
| Body mass index (kg/m 2 ) | 30 ± 6 |
| Caucasian | 83 |
| African-American | 15 |
| Other race | 2 |
| Never smoker | 32 |
| Former smoker (quit >6 mos ago) | 35 |
| Current smoker | 33 |
| Diabetes mellitus | 38 |
| Hypertension | 88 |
| Hyperlipidemia | 84 |
| History of coronary heart disease | 71 |
Of the 170 bifurcations identified, 156 had adequate comparative regions proximal to (n = 89) and/or distal to (n = 135) the bifurcation. Any NIRS signal for the presence of lipid was found with similar frequency at coronary bifurcation (79%), proximal (78%), and distal nonbifurcation segments (69%; p = NS). There were no significant differences in total lipid signal as assessed by LCBI and no increase prevalence of LCP when comparing bifurcation, proximal, and distal segments ( Figure 2 ).
When all patients were compared, no clinical predictors of lipid deposition as assessed by LCBI were found. Even when using our 2-part model to account for the large number of zeros (in which only patients found to have any lipid were further analyzed), there was no significant association with the probability of a nonzero LCBI measurement with bifurcation (point of interest vs proximal/distal), age, body mass index, gender, hypertension, hyperlipidemia, diabetes, coronary artery disease, or taking lipid-lowering medication.
The impact of demographic and historical factors on the presence of LCP is shown in Figure 3 . In this multivariate model, neither the presence of a bifurcation nor any coronary artery disease risk factor correlated with an increased presence of LCP. Younger age was the only statistically significant association with LCP when age was analyzed as a continuous variable and each year younger in age was associated with an increased risk of finding plaque (odds ratio 0.97, confidence interval 0.94–0.99).
