Introduction

FFRCT is a reliable method for detecting functional ischemia, but it can be difficult to interpret because it is affected by a variety of factors such as vessel length [1], plaque characteristics [2, 3], left ventricular mass [4, 5], ramus artery [6], or collateral circulation [7]. The left coronary artery bifurcation angle has been shown to affect wall shear stress and cause modifications to the bloodstream [8, 9]. Significant modifications of laminar flow can contribute to FFR values when taking into account well-established factors such as lumen narrowing and lesion length [10]. We previously reported that diverse FFRCT changes at distal segments occur due to the differences in the left coronary bifurcation angle even with the same vessel length and plaque characteristics [11]. The left coronary artery bifurcation angle contributes to hemodynamic changes in the left coronary arteries and may affect the FFRCT. However, the effect of the left coronary artery bifurcation angle on FFR (FFRCT and invasive FFR) has not been clarified. The present study aimed to identify the effect of atypical left coronary artery bifurcation angle on an FFRCT decline below the clinically relevant value of 0.80 in vessels with no apparent CAD.

Methods

Patient population

A total of 1278 outpatients with suspected coronary artery disease (CAD) and who had a CT angiography (CTA) with FFRCT analysis examined at the Universitair Ziekenhuis Brussel between January 2017 and March 2021 were evaluated. A retrospective case-control cohort analysis was conducted with approval from the research ethics board at the Universitair Ziekenhuis Brussel under protocol number B.U.N. 14320. The main inclusion criteria were normal vessels or vessels with no apparent CAD on invasive coronary angiography (ICA) and CTA in the left main trunk, left anterior descending artery (LAD) and left circumflex artery (LCX). FFRCT of other vessels, such as the right coronary artery or side branches (diagonal and obtuse side branches) were not considered. No apparent CAD was defined as vessels with < 20% coronary stenosis or luminal irregularities [12]. The severity of coronary stenosis was assessed by experienced interventional cardiologists (B.C., JF.A.) using ICA and by experts in cardiac radiology imaging (T.T., K.T.) using CTA. A total of 631 patients who had not undergone ICA and 463 patients who had coronary stenosis with ICA and CTA were excluded from the study. The following categories of patients were also excluded from the study: the large ramus artery that was large enough to be measured by FFRCT (34 patients), the interval between FFRCT and ICA > 90 days (25 patients), inappropriate image quality (motion artifact, blooming artifact, or noise artifact) (22 patients), post-trans-catheter aortic valve implantation (6 patients), post coronary artery bypass graft (6 patients), congenital heart disease (5 patients), post coronary artery stenting (3 patients), post aortic valve replacement (2 patients). Myocardial bridging leads to compression of vessels during the systolic phase, which is prolonged to the diastolic phase, resulting in hemodynamic changes and affecting FFR [13, 14]. Myocardial bridging was absent in our study. In total, 166 vessels including 83 LAD and 83 LCX from 83 patients were enrolled. The patient selection flowchart is presented in Figure 1.

Fig. 1
figure 1

Patient selection flow diagram showing patient flow in the study. CTA, CT angiography; FFRCT, CT derived fractional flow reserve; ICA, invasive coronary angiography; LAD, left anterior descending artery; LCX, left circumflex artery; LMT, left main trunk

Full size image

Coronary computed tomography angiography acquisition

All coronary CT angiography (CTA) scans were acquired with the GE Revolution scanner (GE Healthcare) that had a spatial resolution of 230 μm, a rotational speed of 0.28 s, and a Z-axis coverage of 16 cm enabling to image the heart in one heartbeat. Beta-blockers were administered when necessary for targeting a heart rate of < 60 beats/min. Sublingual nitrates were administered before scanning in all patients. The images were reconstructed at 75 ± 10% of the R-R interval. Coronary arteries were classified based on the American Heart Association classification [15]. Only the major left coronary vessels were considered for analysis and each segment was classified into three categories (proximal, #6 and #11; middle, #7 and #13; distal, #8 and #15).

FFRCT

FFRCT was analyzed by HeartFlow Inc. Computational fluid dynamics and blood flow simulations were performed to calculate FFRCT available at any arbitrary point in the coronary artery. Coronary arteries were classified based on the American Heart Association classification [15] and each segment was divided into three equal segments, proximal, middle, and distal. FFRCT was measured at each point (Supplementary Figure 1A). The magnitude of change in FFRCT (ΔFFRCT) was measured from the proximal to the distal at each left coronary artery. A positive FFRCT was defined as a value ≤ 0.80 in accordance with previously published invasive and non-invasive literature [2, 4, 16, 17].

Left coronary artery bifurcation angle

As in previous studies [18, 19], each bifurcation angle was measured as the crossing angle between the center lines of the left main trunk and each coronary artery (LAD or LCX) on the three-dimensional volume-rendered image of the coronary artery tree. For the left coronary artery bifurcation angle both LAD angle and LCX were determined by assessing the crossing line with the left main trunk (Supplementary Figure 2: center panel).

Vessel morphology and plaque characteristics

Vessel length, lumen volume, and composition of each vessel were measured using GE AW server 3.2 software (GE Healthcare). Aligning the vessel to the region of interest on FFRCT and vessel length and lumen volume were obtained semi-automatically with Color Code Plaque (GE Healthcare) (Supplementary Figure 1B). Vessel constituents were characterized based on Hounsfield units (HU) into low-attenuation plaque (< 30 HU), intermediate-attenuation plaque (30–150 HU), and calcified plaque (> 150 HU) [20, 21].

Statistical analysis

Continuous variables were expressed as mean ± standard deviation (SD). The 95% confidence interval (CI) was calculated as ± 1.96 SDs from the mean. Two-group comparisons were performed with unpaired Student’s t tests for means if the data were normally distributed or Mann–Whitney U-test if the data were not normally distributed. Correlation between continuous variables was performed using the Spearman correlation test. Multivariable linear regression analysis was performed to examine the independent correlations between FFRCT and baseline parameters. A hierarchical cluster analysis (Ward’s method) was performed to classify parameters related to bifurcation angles. Receiver operating characteristics curves were generated to determine the cut-off value that the highest diagnostic performance of an FFRCT ≤ 0.80. All statistical analyses were performed using JMP 11.0 statistical software (SAS Institute).

Results

Demographic data

Table 1 summarizes the demographic and CT image acquisition condition data of the study population. The time phase in the cardiac cycle was 75.2 ± 2.8%, and all CT images could be acquired at the mid-diastolic phase.

Table 1 Characteristics of the patient population
Full size table

Distribution of distal FFRCT

The distribution of distal FFRCT for each vessel is presented in Supplementary Figure 3. Compared to LCX, a lower proportion of FFRCT > 0.80 was observed in LAD (54.2% vs. 89.2).

Changes in FFRCT

FFRCT showed a continuous gradual decline from the proximal to the distal in LAD and LCX vessels in both FFRCT > 0.80 and ≤ 0.80. LAD presented a more stepwise variation than LCX, and distal end of FFRCT was significantly lower in LAD (0.81 ± 0.08 vs. 0.88 ± 0.06, p < 0.01). In LAD, proximal FFRCT did not differ between FFRCT > 0.80 and ≤ 0.80 (0.96 ± 0.02 vs. 0.95 ± 0.04, p > 0.05), but FFRCT ≤ 0.80 showed a significant decline from the level of the middle segment 6 (0.94 ± 0.03 vs. 0.91 ± 0.07, p < 0.01), resulting in the distal end of FFRCT to be significantly lower in FFRCT ≤ 0.80 (0.86 ± 0.04 vs. 0.74 ± 0.07, p < 0.01). In LCX, FFRCT ≤ 0.80 was significantly decreased at the proximal and distal end of FFRCT. ΔFFRCT was significantly higher in LAD (0.14 ± 0.07) than in LCX (0.09 ± 0.05) (Table 2 and Figure 2).

Table 2 FFRCT changes in LAD and LCX
Full size table
Fig. 2
figure 2

Changes in FFRCT of LAD and LCX. Each coronary artery segment number is based on the American Heart Association classification. Dist., distal; Mid., middle; Prox., proximal. *p < 0.01 vs. FFRCT ≤ 0.80

Full size image

Changes in FFRCT according to vascular characteristics

Vessel length, lumen volume, and plaque characteristics including, low-attenuation plaque, intermediate-attenuation plaque, and calcified plaque volume were significantly higher in LAD. The bifurcation angle was significantly lower in the LAD angle (31.1 ± 6.6° vs. 42.8 ± 14.9°, p < 0.01). Both in LAD and LCX, bifurcation angle and vessel length were significantly higher in FFRCT ≤ 0.80 (Table 3).

Table 3 Vessel morphology and composition of each vessel in LAD and LCX
Full size table

Univariate and multivariate analysis of the relationship between FFRCT and vessel characteristics

In LAD, ΔFFRCT significantly correlated with the bifurcation angle (r = 0.35, p = 0.001) and vessel length (r = 0.30, p = 0.005). With each 10° increase in bifurcation angle, FFRCT changed by 0.063. In LCX, ΔFFRCT also significantly correlated with the bifurcation angle (r = 0.26, p = 0.02), vessel length (r = 0.49, p < 0.0001). With each 10° increase in bifurcation angle, FFRCT changed by 0.047 (Figure 3).

Fig. 3
figure 3

Relationship between ΔFFRCT, bifurcation angle, and vessel length. LAD, left anterior descending artery; LCX, left circumflex artery

Full size image

Both angles of LAD- and LCX-related parameters were classified into three groups. The angles of LAD parameters were classified into the following groups: (vessel length and IAP volume) (LAP volume, ΔFFRCT, LAD angle, and CP volume) (lumen volume). The angles of LCX parameters were classified into the following groups: (vessel length, IAP volume, and LCX angle) (LAP volume, CP volume, and ΔFFRCT) (lumen volume) (Supplementary Figure 4). For LAD, multivariable analysis showed that angle (β-coefficient = 0.24, p = 0.02) and vessel length (β-coefficient = 0.55, p = 0.0003) had a predictive value for ΔFFRCT. For LCX, vessel length (β-coefficient = 0.46, p = 0.02) and calcified plaque volume (β-coefficient = 0.34, p = 0.001) were predictive for ΔFFRCT (Table 4). Receiver operating characteristic curve revealed that both LAD and LCX angle had a predictive value for FFRCT ≤ 0.80 at the distal aspect of the vessel (LAD; cut-off 31.0°, AUC 0.70, 95% CI 0.58–0.83, sensitivity 74%, specificity 68%, p = 0.007; LCX, cut-off 52.6°, AUC 0.86, 95% CI 0.76–0.96, sensitivity 88%, specificity 85%, p = 0.0006) (Figure 4).

Table 4 Univariable and multivariable analysis for ΔFFRCT
Full size table
Fig. 4
figure 4

Receiver operating curve of bifurcation angle for predicting the distal vessel FFRCT of ≤ 0.80. LAD, left anterior descending artery; LCX, left circumflex artery

Full size image

Discussion

In patients with an ICA and CTA showing essentially vessels with no apparent CAD, our study highlighted the following: (1) Both LAD and LCX angles correlated with ΔFFRCT; (2) Bifurcation angle was one of the predictors of a distal FFRCT ≤ 0.80 and an optimal cut-off value of 31.0° for the LAD and 52.6° for the LCX were demonstrated in our investigation.

To the best of our knowledge, few investigations have studied the role of the bifurcation angle on FFRCT despite the fact that the bifurcation angle may even be one of the most important factors contributing to an unexpected FFRCT decline in vessels with no apparent CAD. This has significant clinical implications.

There was some accordance between our study of bifurcation angles and previous articles. In previous studies that enrolled patients with suspected CAD, bifurcation angles of LAD and LCX assessed by CTA were 34.2 ± 13.4° and 44.3 ± 13.4° [8] and 37 ± 13° and 59 ± 21° [22], and assessed by ICA were 34.1 ± 18.5° and 42.8 ± 20.1° [19]. These bifurcation angles were larger than in our study where they were 31.1 ± 6.6° and 42.8 ± 14.9°, respectively. This difference might be due to selection bias. LAD angle was greater in patients with advanced calcification (33.8 ± 11.6°) than in those without (40.3 ± 10.0°) [14] LCX angle was also higher in patients with coronary stenosis ≥ 50% (39.5 ± 27.0°) than < 50% (46.9 ± 24.6°) [8].

An advantage of our investigation was that our results of the LAD and LCX angles were studied in patients with no apparent CAD. This is valuable as it shows that even with no apparent CAD the bifurcation angle has an influence on FFRCT. In the previous studies, many confounding factors could potentially have influenced the results.

The finding in our work that vessel morphology was the most important factor influencing FFRCT decline was in line with previous studies [3]. However, we showed that the bifurcation angle also had a considerable influence on FFRCT decline in vessels with no apparent CAD. We believe our findings are of clinical interest. As shown by different studies the important FFRCT value should be placed in perspective and many factors should be taken into account when interpreting it. Both in LAD and LCX, vessel length was the strongest factor influencing FFRCT, followed by bifurcation angle in LAD, and CP volume in LCX. Experimental studies indeed showed that a wide bifurcation angle was closely associated with modifications of the bloodstream in vivo [23] as well as in vitro [22,23,24]. Fluid dynamically, a disturbance of laminar flow generates thermal energy, which is consumed, resulting in FFRCT decline in not only atherosclerotic vessels but also normal vessels [10]. Turbulence may be an important factor in this setting. Previous investigations in patients with CAD with a wider bifurcation angle assessed by CTA [25], intravascular ultrasound [26], or angiography [19] showed a higher prevalence of CAD, high-risk plaque, or restenosis after stent implantation.

Our study is original in that it investigates the bifurcation angle in patients without CAD. The present study included only patients with no apparent CAD. In contradistinction, other studies have investigated the relationship between bifurcation angle and plaque formation. In these reports, bifurcation angle was shown to affect the shear stress and consequently influence plaque formation [27]. A wider bifurcation angle was shown to be related to higher turbulence and low shear stress which might cause plaque formation in the areas of bifurcation [24, 28]. Our study showed that plaque volume did not affect FFRCT dynamics. We suggest that there was only limited total plaque volume to affect FFRCT because of the enrolled patients with no apparent CAD. However, our results offer important additional information even in vessels with no apparent CAD the bifurcation angle has a significant effect on FFRCT and erroneous clinical decisions could occur. Whether at a later phase of the disease, the bifurcation angle caused CAD was beyond the scope of our study.

In our study, the use of a cut-off value of 31.0° for LAD and 52.6° for LCX bifurcation after corrections for confounding factors, was still significant, suggesting that LAD and LCX angles were a reliable predictor of a distal vessel FFRCT of ≤ 0.80.

To the best of our knowledge, this is the first work to investigate the effects of the left coronary bifurcation angle on FFRCT in vessels with no apparent CAD. The present study included vessels with no apparent CAD to exclude confounding factors such as coronary stenosis. In common coronary artery disease, the effect of the bifurcation angle may be more important than the results of our study suggest. Our findings indicate that a large bifurcation angle can affect hemodynamics, resulting in the overestimation of FFRCT. FFRCT should be interpreted more carefully and a correction may be required depending on the left coronary artery bifurcation angle for assessment of myocardial ischemia. Since the bifurcation angle has an effect, the addition of an angulation measurement to conventional FFRCT evaluation may have clinical value and should be considered.

Limitations

Our study has several limitations. First, this is a single-center study. Still, the number of patients included was sufficient for statistical analysis. Second, according to the American Heart Association [29] and European Society of Cardiology guidelines [30], FFR is recommended to assess angiographically intermediate-grade coronary stenosis. The present study included patients with vessels with < 20% coronary stenosis. Therefore, invasive procedures (measurement of FFR or bifurcation angle) were not performed. Previous studies have shown that FFRCT can be an alternative test to invasive FFR due to the high concordance between FFRCT and invasive FFR [31]. FFR values [31] and bifurcation angle [9, 32] assessed by CTA correlate well with invasive measurements. However, it is indeed uncertain whether distal FFR changes related to the bifurcation angle occur in invasive FFR as well as FFRCT. Third, despite the fact that only vessels with no apparent CAD were selected, plaque burden was still observed. Plaque burden observed with no apparent CAD may be due to the following factors: (1) Plaque formation generates at the bifurcation angle due to higher turbulence and low shear stress. The vessel diameter immediately after the bifurcation angle is large, resulting in higher absolute plaque volume even with no apparent CAD; (2) Diffuse plaque deposits throughout the vessel. (3) Plaque deposits on the extravascular side. Our plaque analysis software could not differentiate between intravascular and extravascular plaque. However, it is controversial and difficult to investigate FFRCT in vessel models that completely eliminate the effects of plaque characteristics. Simulated vessel models are assumed to have a rigid wall rather than the elastic wall; therefore, such simulation does not reflect the physiological situation [23]. Plaque burden caused an impaired vasodilator capacity due to oxidative stress and inflammation [33, 34]. This vasodilator dysfunction could interfere with accurate assessment of FFRCT. Fourth, this study did not include ‘no obvious CAD’, thus minimal atherosclerotic lesions were present. Because vessels with advanced atherosclerosis also may show a complex morphology due to curvature or tortuosity in the actual clinical setting, measurement of the bifurcation would optimally be better performed by three-dimensional analysis. However, similar to our experience in previous studies [9, 18, 19, 25], it was quite difficult to assess the bifurcation angle by three-dimensional analysis. Moreover, complex vessel morphology may affect blood flow leading to paradoxical FFRCT changes. Blood flow velocity is accelerated at the stenotic lesion, resulting in potential energy being converted into kinetic energy. A small amount of turbulent eddies generated at the stenotic lesion leads to less thermal energy loss, thus kinetic energy is reconverted into potential energy (pressure recovery phenomenon), which can cause FFRCT increase In order to explore the influence of vessel morphology on FFRCT changes, further investigations of various types of lesions in a larger number of patients may be needed. Fifth, the relationship between bifurcation angle, turbulence, and FFRCT is not clear because turbulence is present not quantifiable [35].

Conclusions

FFRCT decline was dependent on each left coronary bifurcation angle. Vessel length most influences FFRCT, but also the bifurcation angle in vessels with no apparent CAD plays a significant role. As such, a correction may be required depending on the left coronary artery bifurcation angle when interpreting numerical values of FFRCT for assessment of myocardial ischemia. We provide cut-off values at which this effect occurs.