Abstract
Objectives
This study aimed to investigate the impact of calcific (Ca) on the efficacy of coronary computed coronary angiography (CTA) in evaluating plaque burden (PB) and composition with near-infrared spectroscopy-intravascular ultrasound (NIRS-IVUS) serving as the reference standard.
Materials and methods
Sixty-four patients (186 vessels) were recruited and underwent CTA and 3-vessel NIRS-IVUS imaging (NCT03556644). Expert analysts matched and annotated NIRS-IVUS and CTA frames, identifying lumen and vessel wall borders. Tissue distribution was estimated using NIRS chemograms and the arc of Ca on IVUS, while in CTA Hounsfield unit cut-offs were utilized to establish plaque composition. Plaque distribution plots were compared at segment-, lesion-, and cross-sectional-levels.
Results
Segment- and lesion-level analysis showed no effect of Ca on the correlation of NIRS-IVUS and CTA estimations. However, at the cross-sectional level, Ca influenced the agreement between NIRS-IVUS and CTA for the lipid and Ca components (p-heterogeneity < 0.001). Proportional odds model analysis revealed that Ca had an impact on the per cent atheroma volume quantification on CTA compared to NIRS-IVUS at the segment level (p-interaction < 0.001). At lesion level, Ca affected differences between the modalities for maximum PB, remodelling index, and Ca burden (p-interaction < 0.001, 0.029, and 0.002, respectively). Cross-sectional-level modelling demonstrated Ca’s effect on differences between modalities for all studied variables (p-interaction ≤ 0.002).
Conclusion
Ca burden influences agreement between NIRS-IVUS and CTA at the cross-sectional level and causes discrepancies between the predictions for per cent atheroma volume at the segment level and maximum PB, remodelling index, and Ca burden at lesion-level analysis.
Clinical relevance statement
Coronary calcification affects the quantification of lumen and plaque dimensions and the characterization of plaque composition coronary CTA. This should be considered in the analysis and interpretation of CTAs performed in patients with extensive Ca burden.
Key Points
Coronary CT Angiography is limited in assessing coronary plaques by resolution and blooming artefacts.
Agreement between dual-source CT angiography and NIRS-IVUS is affected by a Ca burden for the per cent atheroma volume.
Advanced CT imaging systems that eliminate blooming artefacts enable more accurate quantification of coronary artery disease and characterisation of plaque morphology.
Keywords: Computed tomography angiography, Vascular calcification, Atherosclerosis, Near-infrared spectroscopy intravascular ultrasound
Introduction
Computed tomography coronary angiography (CTA) is regarded as the first-line diagnostic modality for assessing the extent and severity of coronary artery disease (CAD). Cumulative data has shown that CTA has a high negative predictive value in excluding obstructive CAD, but it has limited efficacy in assessing plaque pathophysiology and vulnerability, especially in patients with CAD [1–4]. This has been attributed to the limited resolution of CTA compared to intravascular imaging, as well as to the blooming artefacts that are seen in calcified lesions that often preclude the accurate delineation of the lumen and vessel wall borders on CTA.
Several studies have attempted to examine the implications of coronary calcification on the performance of CTA in quantifying the extent and severity of CAD using intravascular imaging as a reference standard [1–3]. However, most of these reports relied on retrospective analysis of acquired data and focused only on the quantification of the lumen dimensions [1] or the performance of the modality in assessing lesion severity [4, 5] and calcific (Ca) burden [2], and did not provide a holistic evaluation of the implications of coronary calcification in the performance of CTA in measuring plaque burden (PB) and characterising plaque composition.
Methods
Study population
All the patients included in the study signed a consent form prior to recruitment; the study was approved by the local ethics committee (REC reference: 17/SC/0566) and was conducted in accordance with the Declaration of Helsinki. The present study is a post-hoc analysis of a recently completed prospective study that aimed to evaluate the performance of CTA in assessing coronary artery morphology and physiology using multimodality near-infrared spectroscopy-intravascular imaging (NIRS-IVUS) as reference standard (Clinicaltrials.gov NCT03556644). The study design and objectives have been described previously [6]. In brief, 70 patients with a chronic coronary syndrome that had a coronary angiogram showing significant CAD, requiring treatment with percutaneous coronary intervention (PCI) or further evaluation with invasive imaging of functional assessment were recruited and underwent CTA before being listed for repeat invasive angiogram + /− PCI and 3 vessel NIRS-IVUS imaging. The CTA and NIRS-IVUS data acquisition protocol is described in the supplementary file. Patients were recruited for the study between March 2018–July 2019
CTA and NIRS-IVUS data analysis
An interventional cardiologist with expertise in CTA analysis reviewed the NIRS-IVUS and CTA data and identified the most proximal and distal side branch in NIRS-IVUS that was visible in CTA and used this to define a segment of interest, i.e. a segment that was assessed by both modalities. Segments with poor image quality, significant artefacts, and those portraying a stented vessel were excluded from the analysis.
CTA analysis was performed in the segment of interest by an expert analyst with an established reproducibility blinded to NIRS-IVUS analysis using commercially available software (QAngioCT Research Edition 3.1, Medis Medical Imaging Systems) [7]. In each CTA cross-section within the segment of interest, the analyst manually drew the lumen and vessel wall borders. The established Hounsfield unit cut-off proposed by de Graaf et al, [8] incorporated in the software was used to characterise plaque composition in each CTA cross section and to define four tissue types, fibrotic tissue (FT), fibrofatty (FF), necrotic core (NC) and Ca tissue. These cutoffs are based on a histology study that included CTA data acquired with a tube voltage of 120 kV but phantom studies indicate that they also apply to CTA images obtained at 100 kV [9, 10]. The detected tissue types on CTA were used to generate spread-out plots of NC and Ca distribution with the X-axis indicating the longitudinal and the Y-axis the circumferential position of these plaque components, which were portrayed in yellow and semi-transparent white color, while the FT and the FF were shown in red color [11]. In these spread-out plots, the Ca and lipid core burden index (CaBI, LCBI) were estimated and defined as the number of semi-transparent white or yellow pixels indicating the presence of Ca and NC divided by the total number of pixels multiplied by 1000. Moreover, the maxLCBI4mm was defined as the maximum LCBI value in a 4 mm segment.
NIRS-IVUS segmentation was performed for the segment of interest by another expert analyst with established reproducibility blinded to CTA analysis [12]. NIRS-IVUS analysis included the extraction of the end-diastolic frames using a validated deep learning methodology and, in these frames, the lumen and external elastic membrane (EEM) borders were manually annotated [13]. In addition, in these frames, the circumferential distribution of the Ca was manually annotated with an arc and these annotations, together with the lipid distribution derived from the NIRS chemogram, were used to generate spread-out plots of Ca and lipid core distribution with the X-axis indicating the longitudinal and the Y-axis the circumferential location of these components. Similarly, in the spread-out plots generated from the CTA data, FT was marked as red, semi-transparent white for the Ca and yellow color for the lipid core tissue; these plots were used to calculate the LCBI, maxLCBI4mm, and the CaBI (Fig. 1).
Fig. 1.
Typical example of tissue distribution in NIRS-IVUS and CTA imaging data. A Displays a longitudinal section of the IVUS sequence that was matched with the CTA images obtained from a left anterior descending coronary artery. B Shows a spread-out plot of the circumferential and longitudinal distribution of the lipid (indicated with yellow color) and Ca (displayed in a semi-transparent white color) tissue in the studied vessel while C, C” corresponding NIRS-IVUS and CTA cross sections with the estimations of the two modalities about plaque composition. D Illustrates tissue components distribution in a spread-out plot on the CTA imaging data and E is a longitudinal section of the studied vessel with the annotated lumen and vessel borders in CTA
The CTA and NIRS-IVUS images were co-registered using a methodology described in the supplementary file and the estimations of the two modalities were compared.
Implications of coronary calcification on the performance of CTA in assessing plaque pathology
The effect of coronary calcification on the performance of CTA in assessing plaque pathology was examined at segment-, lesion- and cross-sectional-level.
Segment-level analysis: the mean Ca area was estimated for each segment of interest defined as: Ca volume/segment length and this was used to split the studied segments in terciles. In each tercile, we compared the estimations of NIRS-IVUS and CTA for the lumen, vessel, total atheroma volume (TAV, defined as vessel—lumen volume) and percentage atheroma volume (PAV, defined as: 100 × TAV/vessel volume), as well as for the CaBI, LCBI, and maxLCBI4mm. Moreover, we examined the effect of mean Ca area, and arc (defined as 360 × CaBI/1000) on the association between NIRS-IVUS and CTA measurements for the entire segment dataset.
Lesion-level analysis: lesions in NIRS-IVUS as three consecutive end-diastolic frames with a PB ≥ 40%. Two lesions were considered separate if there was a segment between these two lesions with no significant atheroma (PB < 40%) with a length ≥ 5 mm [14, 15]. The correspondence between NIRS-IVUS and CTA cross-sections—defined using dedicated software—was used to identify the location of the lesions detected by NIRS-IVUS on CTA and these segments were classified as lesions in CTA. In each lesion, the mean Ca area was computed, and this was used to split the lesions in terciles. In each tercile, we compared the estimations of CTA and NIRS-IVUS for the minimum lumen area (MLA), maximum PB, reference lumen and vessel area and the remodelling index, as well as the CaBI, LCBI, maxLCBI4mm. Moreover, we examined the effect of mean Ca area and arc on the agreement between CTA and NIRS-IVUS for the entire lesion dataset.
Cross-sectional-level analysis: the Ca area of each CTA cross-section was used to split the matched NIRS-IVUS and CTA cross-section terciles. In each tercile, we compared the estimations of the two modalities for the lumen, vessel, plaque area, PB, CaBI, and LCBI (indicated by the Ca and lipid arc). In addition, we assessed the effect of the Ca area and arc on the association between NIRS-IVUS and CTA estimations for the above metrics.
Statistical analysis
Continuous variables are presented as median (interquartile range), while categorical as absolute values and percentages. The estimations of NIRS-IVUS and CTA were compared using Wilcoxon signed-rank tests. Bland–Altman (BA) analysis was employed to assess bias and the limits of agreements (LOA) between CTA and NIRS-IVUS measurements. Non-parametric BA analysis was chosen for the segment- and lesion-level analysis due to the small sample size and the non-normal distribution of differences, while parametric analysis was used for cross-sectional level analysis due to the larger sample size. In the parametric BA analysis, bias was calculated as the mean difference between the two methods, while the LOA was defined as a range centred around the mean bias ± 1.96 × standard deviation. For the non-parametric BA analysis, the median bias and the 2.5th–97.5th percentiles were computed using quantile regression.
The agreement between the two modalities was tested using intra-class correlation coefficients (ICC) under a two-way random effects model, considering absolute agreement and average measurements using a linear mixed model. First, the unadjusted ICC was estimated: subjects and modalities (NIRS-IVUS/CT) were included as random effects in the model. Subsequently, to obtain the adjusted ICC, the mean Ca area/mean Ca arc were included as fixed covariates in the model (using three knots-restricted cubic spline transformations for lesion- and segment-analysis and using 5 knots-restricted cubic spline transformations for frame-level data). When the confidence intervals of the unadjusted and adjusted ICCs overlapped, the comparative p-value was considered non-significant. For sensitivity analysis, the mean Ca area was divided into terciles and deciles and then ICC was calculated for each group. To detect the presence of heterogeneity among the terciles for ICCs, a meta-analysis was conducted using a tau-based heterogeneity p-value.
Finally, proportional odds models (for both segment- and lesion-level data) and ordinary least square models (for cross-sectional-level data) were used to estimate the differences between the predictions of NIRS-IVUS and CTA for lumen and vessel wall dimensions and PB and composition. The models were constructed both with (adjusted differences) and without the inclusion of the Ca area (unadjusted differences). The interaction between the Ca area on CTA and NIRS-IVUS measurements was assessed to examine whether the estimated differences between CTA and NIRS-IVUS measurements are influenced by the Ca area on CTA and explore whether the performance of the CTA to predict the NIRS-IVUS measurements varies with the changes in Ca.
All statistical tests were two-tailed, and statistical significance was set at p < 0.05; analyses were performed in the R software v. 4.2.2 (R statistical software) using “lme4”, “ppcor”, “dplyr”, “rms”, “meta” and “ggplot2” packages.
Results
Studied patients
Of the 70 patients recruited in the study, 65 patients had NIRS-IVUS and CTA imaging. In one case, a malignancy was found on CTA and the patient did not have a cardiac catheterization, one patient developed acute kidney injury post CTA and was excluded from the study, and in three cases there was NIRS-IVUS catheter malfunction and intravascular imaging was not performed. Moreover, 1 patient was excluded because of poor CTA image quality caused by an increased Ca burden that did not allow CTA segmentation. From the remaining patients, 9 additional vessels were excluded because of motion artefacts that prohibited frame-by-frame NIRS-IVUS and CTA matching. Therefore, in total, 64 patients, and 186 vessels with interpretable CTA and NIRS-IVUS imaging data (23,605 matched cross-sections) were included in the final analysis. The baseline characteristics of the patients included in the study are shown in Supplementary Table S1.
Segment-level analysis
Segment-level analysis results are shown in Table 1. CTA underestimated the lumen and vessel volume, TAV, PAV, and CaBI compared to NIRS-IVUS, but there was no difference between the two modalities for LCBI and maxLCBI4mm although the LOA was large. Results were consistent in the 3 terciles with the only difference in the CaBI in the high Ca area tercile where there was no difference between the two groups (Table 1).
Table 1.
Segment-level comparison of the estimations of NIRS-IVUS and CTA in the entire dataset and in the low, intermediate and high Ca area groups
| NIRS-IVUS | CTA | p-value * | Median bias (95% CI) | |
|---|---|---|---|---|
| Entire set of studied segments | ||||
| Lumen volume, mm3 | 420.39 (223.82, 612.78) | 309.20 (176.73, 463.81) | < 0.001 | 89.43 (1.26, 251.68) |
| Vessel volume, mm3 | 735.01 (385.20, 1059.41) | 454.59 (247.49, 672.14) | < 0.001 | 251.32 (42.81, 573.23) |
| TAV, mm3 | 292.80 (149.37, 465.17) | 116.09 (64.87, 229.72) | < 0.001 | 139.93 (9.57, 368.73) |
| Percent atheroma volume, % | 41.06 (33.75, 47.85) | 27.67 (20.61, 37.17) | < 0.001 | 10.60 (− 1.34, 23.91) |
| LCBI | 33.87 (4.77, 91.15) | 28.36 (4.11, 78.62) | 0.758 | 0 (− 121.09, 96.91) |
| CaBI | 59.35 (16.51, 131.40) | 50.81 (9.95, 111.66) | < 0.001 | 11.27 (− 43.62, 78.94) |
| MaxLCBI4mm | 228.12 (60.00, 407.19) | 230.00 (33.44, 538.75) | 0.242 | − 10 (− 537.50, 343.75) |
| Low Ca area tercile | ||||
| Lumen volume, mm3 | 287.39 (153.62, 502.66) | 216.02 (132.14, 375.12) | < 0.001 | 87.94 (− 0.12, 239.94) |
| Vessel volume, mm3 | 468.81 (246.43, 845.13) | 319.13 (166.07, 521.45) | < 0.001 | 147.16 (20.36, 573.23) |
| TAV, mm3 | 151.91 (92.51, 343.44) | 63.33 (37.08, 129.92) | < 0.001 | 80.59 (22.17, 361.04) |
| Percent atheroma volume, % | 37.99 (30.82, 42.70) | 21.65 (18.54, 25.64) | < 0.001 | 13.77 (2.34, 26.00) |
| LCBI | 16.13 (2.01, 52.58) | 19.39 (1.01, 64.83) | 0.085 | − 4.77 (− 137.79, 62.24) |
| CaBI | 13.53 (1.02, 42.46) | 3.19 (0.00, 12.50) | < 0.001 | 5.99 (− 3.99, 53.18) |
| MaxLCBI4mm | 173.75 (44.38, 310.94) | 141.88 (11.01, 530.00) | 0.078 | − 41.25 (− 530.00, 248.75) |
| Intermediate Ca area tercile | ||||
| Lumen volume, mm3 | 461.85 (275.35, 664.55) | 355.44 (219.38, 496.04) | < 0.001 | 105.72 (9.23, 261.63) |
| Vessel volume, mm3 | 792.43 (544.06, 1106.64) | 523.01 (317.21, 661.53) | < 0.001 | 319.97 (76.856, 586.07) |
| TAV, mm3 | 315.25 (184.06, 486.80) | 111.83 (76.01, 188.82) | < 0.001 | 174.06 (42.53, 386.64) |
| Percent atheroma volume, % | 41.32 (33.50, 48.99) | 26.04 (20.32, 31.04) | < 0.001 | 13.10 (3.23, 25.87) |
| LCBI | 22.49 (1.98, 63.93) | 9.96 (1.73, 51.43) | 0.112 | 1.68 (− 65.23, 102.44) |
| CaBI | 58.52 (34.11, 102.46) | 53.07 (30.75, 76.07) | 0.001 | 17.64 (− 27.45, 77.30) |
| MaxLCBI4mm | 184.38 (45.00, 340.31) | 140.62 (20.00, 385.00) | 0.754 | 0 (− 427.50, 343.75) |
| High Ca area tercile | ||||
| Lumen volume, mm3 | 441.49 (267.54, 572.41) | 336.96 (216.94, 466.19) | < 0.001 | 86.53 (14.45, 242.89) |
| Vessel volume, mm3 | 776.69 (537.17, 1073.69) | 573.94 (378.55, 772.33) | < 0.001 | 252.891 (48.33, 509.59) |
| TAv, mm3 | 344.86 (241.61, 522.55) | 227.51 (128.34, 321.65) | < 0.001 | 157.546 (− 10.19, 327.07) |
| Percent atheroma volume, % | 45.52 (39.07, 49.18) | 37.97 (32.82, 45.70) | < 0.001 | 6.75 (− 8.99, 17.76) |
| LCBI | 55.77 (25.81, 114.65) | 65.13 (21.02, 138.40) | 0.669 | 3.18 (− 175.72, 110.66) |
| CaBI | 146.72 (93.53, 245.42) | 139.34 (105.04, 200.22) | 0.220 | 19.30 (− 111.68, 139.77) |
| MaxLCBI4mm | 332.50 (188.75, 466.88) | 323.12 (158.75, 612.19) | 0.488 | − 2.5 (− 560.00, 375.00) |
CaBI calcific burden index, CTA computed tomography coronary angiography, LCBI lipid core burden index, NIRS-IVUS near-infrared spectroscopy intravascular ultrasound, MaxLCBI4mm maximum LCBI in 4 mm segment
* p-values were obtained by the Wilcoxon signed rank test
The unadjusted ICCs between NIRS-IVUS and CTA are shown in Fig. 2; the ICC was high for the lumen and vessel volume, moderate for the TAV and CaBI, and weak for the PAV, LCBI, and maxLCBI4mm. After adjusting for the mean Ca area or mean Ca arc, the ICC for the CaBI was significantly lower but there was no difference for other variables (Fig. 2). The effect of an increasing mean Ca area on the association of NIRS-IVUS and CTA at segment-level is illustrated in Fig. 3A. Overall, there was no statistically significant difference between the ICC values for the estimations of NIRS-IVUS and CTA in the 3 Ca area groups (Table 2 and Supplementary Fig. 1).
Fig. 2.
Unadjusted and adjusted (by mean Ca area and mean ca arc) ICC and 95% CI for measurements between NIRS-IVUS and CTA estimated by linear mixed model at segment level
Fig. 3.
ICC for measurements between NIRS-IVUS and CTA according to Ca area deciles at segment (A), lesion- (B), and frame-level analysis (C). Horizontal black dashed lines have been added to facilitate the interpretation of ICC, and they represent 0.50, 0.75, and 0.90 (< 0.50 indicating poor, 0.50–0.75 indicating moderate, 0.75–0.90 indicating good, and > 0.90 indicating excellent agreement). Only ICC points according to Ca area tertiles are shown on the plot, 95% CI is not displayed
Table 2.
ICC values between NIRS-IVUS and CTA estimations at segment-, lesion- and cross-sectional-level in the different Ca area terciles
| Low Ca tercile ICC | p-value | Intermediate Ca tercile ICC | p-value | High Ca tercile ICC | p-value | Heterogeneity p-value | |
|---|---|---|---|---|---|---|---|
| Segment-level analysis | |||||||
| Lumen volume | 0.95 (0.72, 0.98) | < 0.001 | 0.94 (0.55, 0.98) | < 0.001 | 0.94 (0.65, 0.98) | < 0.001 | 0.993 |
| Vessel volume | 0.87 (0.27, 0.96) | < 0.001 | 0.88 (−0.05, 0.97) | < 0.001 | 0.90 (0.28, 0.97) | < 0.001 | 0.994 |
| TAV | 0.53 (− 0.08, 0.77) | < 0.001 | 0.70 (− 0.16, 0.89) | < 0.001 | 0.79 (0.13, 0.92) | < 0.001 | 0.669 |
| PAV | 0.19 (− 0.16, 0.48) | 0.021 | 0.37 (− 0.21, 0.69) | < 0.001 | 0.50 (0.15, 0.70) | 0.001 | 0.363 |
| LCBI | 0.27 (− 0.19, 0.55) | 0.105 | 0.55 (0.26, 0.73) | 0.001 | 0.25 (− 0.24, 0.55) | 0.129 | 0.281 |
| CaBI | 0.36 (− 0.03, 0.61) | 0.015 | 0.65 (0.39, 0.80) | < 0.001 | 0.25 (− 0.24, 0.55) | 0.130 | 0.111 |
| MaxLCBI4mm | 0.36 (− 0.04, 0.61) | 0.036 | 0.55 (0.26, 0.73) | 0.001 | 0.18 (− 0.35, 0.51) | 0.217 | 0.290 |
| Lesion-level analysis | |||||||
| MLA | 0.93 (0.75, 0.97) | < 0.001 | 0.90 (0.65, 0.96) | < 0.001 | 0.79 (0.31, 0.91) | < 0.001 | 0.692 |
| Maximum PB | 0.45 (− 0.22, 0.75) | < 0.001 | 0.74 (0.46, 0.86) | < 0.001 | 0.74 (0.56, 0.83) | < 0.001 | 0.529 |
| Reference lumen area | 0.82 (0.04, 0.94) | < 0.001 | 0.78 (0.33, 0.90) | < 0.001 | 0.65 (− 0.09, 0.85) | < 0.001 | 0.858 |
| Reference vessel area | 0.73 (− 0.16, 0.91) | < 0.001 | 0.74 (− 0.15, 0.91) | < 0.001 | 0.52 (− 0.23, 0.81) | < 0.001 | 0.802 |
| Remodelling index | 0.47 (0.19, 0.65) | 0.002 | 0.02 (− 0.53, 0.37) | 0.470 | 0.14 (− 0.34, 0.45) | 0.253 | 0.128 |
| LCBI | 0.59 (0.38, 0.73) | < 0.001 | 0.41 (0.10, 0.61) | 0.008 | 0.24 (− 0.15, 0.50) | 0.098 | 0.147 |
| CaBI | 0.63 (0.39, 0.77) | < 0.001 | 0.31 (− 0.04, 0.54) | 0.037 | 0.60 (0.39, 0.74) | < 0.001 | 0.155 |
| MaxLCBI4mm | 0.61 (0.35, 0.76) | < 0.001 | 0.61 (0.39, 0.75) | < 0.001 | 0.54 (0.29, 0.70) | < 0.001 | 0.863 |
| Cross-sectional-level analysis | |||||||
| Lumen area | 0.89 (0.63, 0.95) | < 0.001 | 0.82 (0.12, 0.93) | < 0.001 | 0.82 (0.35, 0.92) | < 0.001 | 0.883 |
| Vessel area | 0.75 (− 0.10, 0.91) | < 0.001 | 0.66 (− 0.21, 0.88) | < 0.001 | 0.76 (0.05, 0.90) | < 0.001 | 0.961 |
| Plaque area | 0.43 (− 0.09, 0.68) | < 0.001 | 0.42 (− 0.16, 0.69) | < 0.001 | 0.68 (0.42, 0.80) | < 0.001 | 0.345 |
| PB | 0.35 (− 0.07, 0.58) | < 0.001 | 0.47 (− 0.05, 0.70) | < 0.001 | 0.67 (0.65, 0.69) | < 0.001 | 0.999 |
| LCBI | 0.27 (0.24, 0.29) | < 0.001 | 0.36 (0.32, 0.39) | < 0.001 | 0.20 (0.15, 0.25) | < 0.001 | < 0.001 |
| CaBI | 0.00 (− 0.03, 0.04) | 0.416 | 0.20 (0.15, 0.24) | < 0.001 | 0.46 (0.44, 0.51) | < 0.001 | < 0.001 |
CaBI calcific burden index, CTA computed tomography coronary angiography, LCBI lipid core burden index, MaxLCBI4mm maximum LCBI in 4 mm segment, MLA minimum lumen area, PAV percentage atheroma volume, PB plaque burden, TAV total atheroma volume
Both unadjusted and adjusted by mean Ca area proportional odds model showed that the estimated median lumen volume, vessel volume, TAV and PAV were larger on NIRS-IVUS than the CTA. There was no significant interaction between the mean Ca area and lumen and vessel volume but there was a borderline interaction (p = 0.070) for the TAV and a significant interaction for the PAV (p-interaction < 0.001). No significant differences were noted between NIRS-IVUS and CTA for the estimated CaBI, LCBI, and maxLCBI4mm measurements in both unadjusted and adjusted models. For these variables, there was no interaction between the two modalities and the mean Ca area (Supplementary Fig. 2).
Lesion-level analysis
In total, 304 lesions were detected by NIRS-IVUS and were assessed by CTA and included in the analysis. The estimations of NIRS-IVUS were higher for the MLA, maximum PB and reference lumen and vessel areas; this trend was noted in all the Ca terciles although there was a variation in the median bias and its 95% confident interval for the differences for the above variables in the three groups (Table 3). NIRS-IVUS overestimated the remodelling index in the low and intermediate Ca terciles but there was no difference for the high Ca tercile. The lesion composition analysis results were different in Ca terciles. In the low Ca area tercile, there were differences between CTA and NIRS-IVUS for the CaBI, LCBI, and maxLCBI4mm while, in the intermediate Ca area tercile, CTA underestimated both CaBI and maxLCBI4mm and there was no difference between the two modalities for the LCBI. Finally, in the high Ca area tercile, the CTA estimations were different from NIRS-IVUS for the maxLCBI4mm but similar for the CaBI and LCBI (Table 3). The unadjusted ICC was high between NIRS-IVUS and CTA estimations for the MLA and reference lumen area, and CaBI, moderate for the maximum PB, reference vessel area and maxLCBI4mm, and poor for the LCBI and remodelling index. The adjusted mean Ca area or thickness ICC values were not different to the unadjusted for all the studied variables apart from the CaBI whereby the ICC markedly decreased (Fig. 4). The ICC between NIRS-IVUS and CTA estimations in different Ca terciles are presented in Table 2 and Supplementary Fig. 3 while the effect of an increasing mean Ca on the association of NIRS-IVUS and CTA for the lesion level analysis is shown in Fig. 3B. Overall, there was no statistically significant difference between the ICC values reported in different Ca area terciles.
Table 3.
Lesion-level comparison of the estimations of NIRS-IVUS and CTA in the entire dataset in the low, intermediate, and high Ca area groups
| NIRS-IVUS | CTA | p-value * | Median bias (95% CI) | |
|---|---|---|---|---|
| Entire set of studied lesions | ||||
| MLA, mm2 | 4.20 (2.46, 6.35) | 3.17 (1.96, 5.07) | < 0.001 | 1.23 (− 0.66, 3.06) |
| Maximum PB, % | 61.56 (51.75, 74.44) | 54.61 (35.29, 74.54) | < 0.001 | 7.00 (− 14.33, 32.36) |
| Reference lumen area, mm2 | 9.84 (7.44, 12.85) | 7.27 (5.44, 10.09) | < 0.001 | 2.65 (0.34, 5.57) |
| Reference vessel area, mm2 | 13.80 (10.44, 17.98) | 8.73 (6.72, 12.03) | < 0.001 | 5.16 (1.60, 9.83) |
| Remodelling index | 0.92 (0.78, 1.08) | 0.85 (0.59, 1.05) | 0.005 | 0.08 (− 0.58, 0.56) |
| LCBI | 29.50 (0.00, 110.00) | 26.66 (0.00, 110.42) | 0.279 | 0.00 (− 168.49, 113.33) |
| CaBI | 104.00 (13.75, 203.25) | 60.59 (0.00, 180.78) | < 0.001 | 15.11 (− 87.87, 125.00) |
| MaxLCBI4mm | 173.10 (0.00, 343.81) | 88.12 (0.00, 375.00) | 0.121 | − 0.25 (− 427.50, 290.00) |
| Low Ca area tercile | ||||
| MLA, mm2 | 4.42 (3.09, 6.25) | 3.32 (2.37, 5.49) | < 0.001 | 0.77 (− 0.69, 2.54) |
| Maximum PB, % | 54.22 (49.15, 66.87) | 28.13 (19.56, 49.66) | < 0.001 | 23.46 (1.10, 41.97) |
| Reference lumen area, mm2 | 8.15 (6.07, 10.36) | 6.48 (4.62, 8.14) | < 0.001 | 2.08 (− 0.06, 4.61) |
| Reference vessel area, mm2 | 11.25 (8.49, 14.58) | 7.52 (5.43, 9.57) | < 0.001 | 4.52 (1.28, 8.54) |
| Remodelling index | 0.96 (0.84, 1.10) | 0.83 (0.65, 1.02) | 0.002 | 0.12 (− 0.47, 0.51) |
| LCBI | 0.00 (0.00, 45.88) | 6.67 (0.00, 71.30) | 0.045 | 0 (− 135.00, 54.00) |
| CaBI | 0.00 (0.00, 46.25) | 0.00 (0.00, 2.07) | < 0.001 | 0 (− 3.48, 83.00) |
| MaxLCBI4mm | 22.50 (0.00, 181.94) | 25.00 (0.00, 187.19) | < 0.001 | − 10 (− 437.75, 115.00) |
| Intermediate Ca area tercile | ||||
| MLA, mm2 | 3.71 (2.29, 6.29) | 2.98 (1.91, 4.97) | < 0.001 | 1.11 (− 0.70, 2.94) |
| Maximum PB, % | 62.77 (53.95, 76.91) | 53.95 (41.13, 74.21) | < 0.001 | 6.81 (− 9.40, 30.00) |
| Reference lumen area, mm2 | 9.84 (7.68, 13.30) | 6.97 (5.51, 10.40) | < 0.001 | 2.76 (0.27, 5.63) |
| Reference vessel area, mm2 | 13.67 (10.77, 18.17) | 8.85 (7.03, 12.21) | < 0.001 | 5.06 (1.60, 9.10) |
| Remodelling index | 0.90 (0.74, 1.06) | 0.76 (0.51, 0.99) | 0.015 | 0.08 (− 0.45, 0.66) |
| LCBI | 39.00 (0.00, 112.00) | 13.13 (0.00, 65.36) | 0.263 | 0 (− 153.71, 107. 33) |
| CaBI | 112.00 (60.00, 169.00) | 64.34 (35.00, 105.81) | < 0.001 | 48.51 (− 60.00, 135.00) |
| MaxLCBI4mm | 224.00 (0.00, 340.31) | 52.50 (10.00, 248.75) | < 0.001 | 0 (− 323.50, 322.50) |
| High Ca area tercile | ||||
| MLA, mm2 | 4.21 (2.69, 6.41) | 3.14 (1.34, 4.66) | < 0.001 | 1.58 (− 0.18, 3.78) |
| Maximum PB, % | 66.55 (57.60, 75.70) | 74.34 (62.65, 83.63) | 0.002 | − 6.34 (− 20.96, 17.48) |
| Reference lumen area, mm2 | 11.55 (9.74, 14.86) | 8.22 (6.92, 11.00) | < 0.001 | 3.13 (1.11, 6.60) |
| Reference vessel area, mm2 | 16.09 (14.02, 21.09) | 9.93 (8.31, 13.37) | < 0.001 | 5.88 (2.78, 11.54) |
| Remodelling index | 0.95 (0.76, 1.08) | 0.91 (0.67, 1.25) | 0.498 | − 0.05 (− 0.74, 0.50) |
| LCBI | 58.00 (8.00, 166.00) | 69.47 (17.08, 179.91) | 0.275 | − 0.613 (− 230.84, 185.00) |
| CaBI | 236.59 (151.00, 340.00) | 222.78 (157.43, 319.56) | 0.539 | 26.2 (− 240.43, 171.16) |
| MaxLCBI4mm | 223.75 (70.25, 435.00) | 220.00 (31.25, 470.00) | < 0.001 | − 10 (− 532.50, 326.25) |
CaBI calcific burden index, CTA computed tomography coronary angiography, LCBI lipid core burden index, MaxLCBI4mm maximum LCBI in 4 mm segment, MLA minimum lumen area, NIRS-IVUS near-infrared spectroscopy intravascular ultrasound, PB plaque burden
* p-values were obtained by Wilcoxon signed-rank test
Fig. 4.
Unadjusted and adjusted (by mean Ca area and mean ca arc) ICC and 95% CI for measurements between NIRS-IVUS and CTA estimated by linear mixed model at lesion level
The unadjusted proportional odds model showed that NIRS-IVUS had a higher estimated median MLA, maximum PB, reference vessel and lumen area, remodelling index, CaBI and maxLCBI4mm compared to CTA and there was no difference between the two modalities for the LCBI. The adjusted model (by mean Ca area) eliminated the difference between the predictions of the two modalities for the remodelling index and maxLCBI4mm but had no other effect on the other variables. Interaction analysis showed that CTA-derived Ca affected the differences in the estimations of NIRS-IVUS and CTA for the maximum PB, remodelling index, and CaBI (p-interaction < 0.001, 0.029 and 0.002, respectively) and there was a trend for the MLA (p-interaction = 0.052, Supplementary Fig. 4).
Cross-sectional-level analysis
In total 23,605 NIRS-IVUS frames were matched with CTA and included in the analysis. The estimations of the two modalities were different for all the studied variables. The results were consistent in all terciles (Table 4).
Table 4.
Cross-sectional-level comparison of the estimations of NIRS-IVUS and CTA in the entire dataset in the low, intermediate, and high Ca area groups
| NIRS-IVUS | CTA | p-value * | Mean bias (95% CI) | |
|---|---|---|---|---|
| Entire set of matched frames | ||||
| Lumen area, mm2 | 7.23 (4.90, 10.42) | 5.42 (3.66, 8.09) | < 0.001 | 1.98 (− 2.51, 6.47) |
| Vessel area, mm2 | 12.88 (8.61, 18.05) | 7.74 (5.11, 11.84) | < 0.001 | 4.95 (− 2.49, 12.39) |
| Plaque area, mm2 | 5.25 (2.87, 7.89) | 1.59 (0.95, 3.32) | < 0.001 | 2.97 (− 3.03, 8.97) |
| PB, % | 38.65 (27.30, 50.78) | 19.08 (16.68, 34.06) | < 0.001 | 12.39 (− 20.73, 45.50) |
| LCBI | 0.00 (0.00, 0.00) | 0.00 (0.00, 0.00) | < 0.001 | − 0.01 (− 0.41, 0.39) |
| CaBI | 0.00 (0.00, 0.08) | 0.00 (0.00, 0.00) | < 0.001 | 0.01 (− 0.35, 0.37) |
| Low Ca area tercile | ||||
| Lumen area, mm2 | 6.45 (4.53, 9.64) | 5.03 (3.49, 7.61) | < 0.001 | 1.64 (− 2.38, 5.65) |
| Vessel area, mm2 | 11.04 (7.62, 16.28) | 6.54 (4.49, 10.02) | < 0.001 | 4.65 (− 2.37, 11.67) |
| Plaque area, mm2 | 4.31 (2.24, 6.71) | 1.21 (0.80, 2.04) | < 0.001 | 3.01 (− 2.62, 8.64) |
| PB, % | 36.16 (25.29, 47.96) | 18.27 (16.08, 19.95) | < 0.001 | 15.79 (− 15.48, 47.06) |
| LCBI | 0.00 (0.00, 0.00) | 0.00 (0.00, 0.00) | < 0.001 | − 0.01 (− 0.40, 0.38) |
| CaBI | 0.00 (0.00, 0.00) | 0.00 (0.00, 0.00) | < 0.001 | 0.02 (− 0.22, 0.26) |
| Intermediate Ca area tercile | ||||
| Lumen area, mm2 | 8.67 (6.40, 11.37) | 6.20 (4.45, 8.65) | < 0.001 | 2.53 (− 1.93, 6.99) |
| Vessel area, mm2 | 13.99 (10.55, 18.52) | 8.19 (5.90, 11.43) | < 0.001 | 6.06 (− 1.04, 13.16) |
| Plaque area, mm2 | 5.29 (3.04, 7.62) | 1.58 (1.09, 2.50) | < 0.001 | 3.53 (− 2.10, 9.17) |
| PB, % | 35.20 (25.87, 45.71) | 18.60 (16.48, 24.65) | < 0.001 | 13.74 (− 13.49, 40.97) |
| LCBI | 0.00 (0.00, 0.00) | 0.00 (0.00, 0.00) | < 0.001 | 0.01 (− 0.28, 0.30) |
| CaBI | 0.00 (0.00, 0.08) | 0.00 (0.00, 0.00) | < 0.001 | 0.01 (− 0.35, 0.37) |
| High Ca area tercile | ||||
| Lumen area, mm2 | 8.04 (5.38, 11.70) | 5.87 (3.67, 8.80) | < 0.001 | 2.52 (− 2.92, 7.95) |
| Vessel area, mm2 | 16.97 (12.62, 21.62) | 11.95 (8.39, 16.25) | < 0.001 | 4.77 (− 3.79, 13.33) |
| Plaque area, mm2 | 8.33 (5.89, 10.68) | 5.16 (3.50, 7.86) | < 0.001 | 2.26 (− 4.85, 9.36) |
| PB, % | 50.15 (39.92, 60.42) | 48.27 (36.26, 63.19) | < 0.001 | 0.20 (− 33.24, 33.65) |
| LCBI | 0.00 (0.00, 0.17) | 0.00 (0.00, 0.17) | < 0.001 | − 0.02 (− 0.52, 0.48) |
| CaBI | 0.25 (0.08, 0.42) | 0.33 (0.17, 0.50) | < 0.001 | − 0.05 (− 0.63, 0.54) |
CaBI calcific burden index, CTA computed tomography coronary angiography, LCBI lipid core burden index, NIRS-IVUS near-infrared spectroscopy-intravascular ultrasound, PB plaque burden
* p-values were obtained by Wilcoxon signed-rank test
Figure 5 shows the ICC between the NIRS-IVUS and CTA estimations. The addition of the Ca area or arc to the ICC had an effect only on the ICC for the LCBI, CaBI suggesting that there was an effect of the Ca burden on the performance of CTA for these metrics. The ICC between the estimations of the two modalities in the three terciles groups are shown in Table 2. A significant heterogeneity was noted between the ICC reported in the three groups for the LCBI, and CaBI indicating an effect of the Ca burden in the agreement of the two modalities for these metrics. CTA estimations for the lipid component seem to correlate better with the NIRS-IVUS in the intermediate Ca tercile while for the CaBI component in the high Ca area tercile (Supplementary Fig. 5). The effect of an increasing mean Ca on the association between NIRS-IVUS and CTA estimations for the cross-sectional level analysis is shown in Fig. 3C.
Fig. 5.
Unadjusted and adjusted (by mean Ca area and mean ca arc) ICC and 95% CI for measurements between NIRS-IVUS and CTA estimated by linear mixed model at the cross-sectional level
In both, the unadjusted and adjusted ordinary least square model, NIRS-IVUS was associated with higher estimated MLA, vessel area, plaque area PB and CaBI and lower estimated LCBI. Interaction analysis showed that the CTA-derived Ca area affected the differences in all the estimations of NIRS-IVUS and CTA for the lumen vessel and plaque dimensions and composition (Supplementary Fig. 6).
Discussion
This is the first study that examined the implications of coronary calcification quantified by CTA on the performance of the modality to assess lumen and plaque and its composition using NIRS-IVUS as the reference standard. We found that the Ca burden (1) did not affect the association of the two modalities for the lumen and plaque dimensions and its composition at segment-level analysis, (2) had no effect on the performance of CTA to quantify the MLA, maximum PB, and characterise its composition, and (3) but it influenced the correlation between CTA and NIRS-IVUS for the lipid and Ca estimations at cross-sectional level analysis. The prediction models built from the matched NIRS-IVUS and CTA data showed that Ca burden (1) had no effect on the agreement of the predicted estimations of NIRS-IVUS and CTA for the lumen and vessel dimensions and composition at the segment level but it affected the PAV estimations (2) influenced the agreement of the predictions of the two modalities at lesion-level for the maximum PB, remodelling index, and CaBI and there was a trend for the MLA, and (3) had a significant effect on the differences between the predictions of NIRS-IVUS and CTA for the lumen vessel and plaque dimensions and compositions.
Ca tissue generates blooming artifacts in CTA which affects data analysis and quantification of lesion severity [16, 17]. Studies examining the implications of coronary calcification on the performance of CTA in evaluating lesion hemodynamic severity using invasive fractional flow reserve (FFR) as a reference standard have shown a negligible effect of Ca burden on the computation of CTFFR [4, 5]. Conversely, there is inconsistency in the findings of studies that examined the effect of Ca on the efficacy of CTA to quantify lumen dimensions and PB against intravascular imaging; this can be at least partially attributed to the limitations of these analyses and more specifically to the small number of patients recruited, the use of early generation CT scanner, their retrospective design and the suboptimal CTA and IVUS co-registration that can cause bias and affect the reported results [1, 18–22].
Our study overcomes these limitations and provides unique insights into the effect of Ca burden on the performance of CTA in assessing lumen and vessel dimensions and quantifying PB and composition. In contrast to other studies, CTA imaging was performed using a 3rd generation CT scanner with a minimum tube voltage of 100 kV—to avoid bias in plaque quantification and effort was made to collect images of optimal quality and reconstruct the CTA using the ADMIRE five iterative reconstruction algorithm—which is not commonly used in clinical practice but appears to be superior to other approaches in the quantification of the PB [7]. Moreover, NIRS-IVUS imaging was performed in the entire coronary tree including the advanced disease and disease-free segments so as to have a representative evaluation of plaque distribution, and after administration of nitrates that were also given prior to CTA imaging, as indicated by the JCCT guidelines [23]. Additionally, we developed a module for accurate CTA and NIRS-IVUS co-registration and matched only the end-diastolic NIRS-IVUS frames—detected retrospectively by a dedicated method—with the CTA so as to avoid errors in the matching that occurred during the back-forward motion of the IVUS probe in the vessel during the cardiac cycle that can affect data co-registration and TAV quantification [13, 24]. Finally, we modified the output of the QAngioCT software and generated spread-out plots of plaque composition to perform a meaningful comparison of the estimations of the two approaches for tissue types.
We found that Ca burden—expressed by the mean plaque area or arc—had no effect on the agreement of the two approaches for the lumen, vessel, TAV, PAV and plaque components at the segment level. Similarly, for lesion-level analysis; Ca burden did not affect the difference between NIRS-IVUS and CTA for the MLA, maximum PB, reference lumen and vessel area, remodelling index and plaque composition. Conversely, the cross-sectional level analysis suggested that the Ca component influenced the ICC between NIRS-IVUS and CTA for the CaBI and LCBI. For the CaBI, the ICC was higher in the high Ca tercile—something that can be explained by the larger range of CaBI values in this group—while for the LCBI the highest value was reported in the moderate Ca tercile. An explanation of the latter finding is the fact that often NIRS detects lipids in disease-free segments with no large NC. Pathological intimal thickening plaque phenotypes can light up in NIRS as they can include lipid tissue and are associated with disease progression, but these lesions are unlikely to have attenuated plaques on CTA [25, 26]. The matched frames included in the intermediate Ca tercile have the atherosclerotic disease and are more likely to portray attenuated plaques in CTA—that are associated with the presence of lipids [27]—while in the high Ca tercile, the blooming artefacts noted in frames with large Ca deposits were likely to mask the surrounding tissues and affect the efficacy of CTA to portray lipid-rich plaques.
Results were different when proportional odds modelling was performed that took advantage of the NIRS-IVUS and CTA estimations to predict the agreement between the two modalities in different Ca patterns. This approach overcomes the limitation of the small sample size and allows a more comprehensive evaluation of the effect of Ca burden on the differences between NIRS-IVUS and CTA estimations. Segment-level analysis demonstrated an effect of the Ca burden on the difference between NIRS-IVUS and CTA for PAV with low Ca area in CTA was related to PAV underestimation that diminishes with the increase of Ca (Supplementary Fig. 3). The lesion-level analysis also showed a significant effect of Ca area on the difference between NIRS-IVUS and CTA for the PB (CTA underestimated PB in lesions with low calcification and overestimated PB in these with increased Ca burden), remodelling index (CTA underestimated remodelling index in lesions with low Ca but provided similar estimation to NIRS-IVUS in Ca lesions) and CaBI (CTA overestimated Ca burden in lesions with increased Ca) and there was a trend for the MLA with CTA tending to underestimate the MLA in lesions with increased calcification. Finally, the cross-sectional level analysis revealed a significant interaction between the Ca area and the NIRS-IVUS and CTA estimations; a large Ca area was associated with a larger underestimation of the lumen area and overestimation of the vessel and plaque area, PB, CaBI and LCBI. These findings should be attributed to the limited efficacy of CTA to assessing soft fibrotic-rich plaques as shown in histology studies and the blooming artefacts that are seen in large Ca deposits that mask the lumen, causing an overestimation of PB and CaBI which impedes the accurate quantification of the low attenuation component [28].
From the above, it is apparent that a large Ca component may impose challenges in the accurate quantification of plaque pathology. Machine learning methodologies trained from intravascular imaging data may be useful in overcoming the limitations of current CT scanners with the first studies in the field showing promising results—however, prospective validation in large datasets is needed before advocating their use in routine practice [11]. In parallel, advances in CT scanners are expected to improve CTA image quality reduce the blooming effects of Ca and enhance its potential in clinical practice and research [29, 30]. Indeed, cumulative data indicate that photon counting CT enables a more accurate assessment of the lumen dimensions and quantification of the Ca burden in calcified lesions and outperforms conventional CTA in the evaluation of stented segments where the increased blooming artefacts limit lumen visibility [31–33].
Limitations
Despite the fact that this study is the largest in the field assessing in a comprehensive fashion the effect of Ca deposits on the performance of current CTA technologies to evaluate plaque pathology, it has limitations that should be acknowledged. Firstly, it is a single centre study conducted using a single CT scanner; it is, therefore, unclear whether the findings of this analysis apply to CTA data collected by other CT systems and reconstructed using different reconstruction approaches (that also appear to affect the quantification of the Ca volume) [34]. Secondly, despite that this is one of the largest studies in the field the number of the included patients and studied segments may have not provided sufficient power to thoroughly assess the effect of Ca on the agreement between the estimations of NIRS-IVUS and CTA. The use of proportional odd modelling may allow us to partially overcome this restraint, but it also has limitations as it is associated with a larger systematic bias than the assessment of the two methods’ agreement. Thirdly, the severity of lesion calcification was based on the estimations of CTA. It is well known that CTA is unable to detect small Ca deposits, and this is likely to have affected the reported results; however, it was preferred over NIRS-IVUS estimation as IVUS does not allow quantification of Ca volume because of the acoustic shadowing behind Ca tissue [35]. The latter did not allow volumetric comparison of intravascular and CTA and NIRS-IVUS estimations of the Ca tissue as it has been performed in recently reported OCT studies [2, 3]. Nevertheless, the use of NIRS-IVUS in this study has allowed us to evaluate, for the first time the implications of Ca burden on the CTA on the computation of lipid component. Finally, there was a significant difference in the length of the segment of interest between NIRS-IVUS and CTA (Δlength: − 3.49 ± 4.62 mm, p < 0.001); this may have affected the agreement between NIRS-IVUS and CTA for the volumetric analyses (i.e. lumen, vessel and TAV) at segment-level but not the lesion- and frame-level analyses. Moreover, this difference is unlikely to have affected the impact of the Ca burden on the agreement between NIRS-IVUS and CTA for the above variables as the difference in the length of the segment of interest was consistent in the three Ca terciles and it was not associated with the Ca burden (p = 0.215).
Conclusions
The efficacy of CTA imaging to assess the extent, severity, and compositional characteristics of coronary atheroma can be affected by Ca burden on CTA images. This should be taken into account in studies using CTA to assess plaque vulnerability and in revascularisation planning.
Supplementary information
Acknowledgements
Barts NIHR Biomedical Research Centre, London, United Kingdom: Anantharaman Ramasamy, Retesh Bajaj, Anthony Mathur, Andreas Baumbach, and Christos V Bourantas. Turkish Society of Cardiology, Istanbul, Turkey: Murat Cap.
Abbreviations
- BA
Bland–Altman
- Ca
Calcific
- CaBI
Calcific burden index
- CAD
Coronary artery disease
- CTA
Computed tomography coronary angiography
- FF
Fibrofatty
- FFR
Fractional flow reserve
- FT
Fibrotic tissue
- ICC
Intra-class correlation coefficients
- LCBI
Lipid core burden index
- LOA
Limits of agreements
- MaxLCBI4mm
Maximum LCBI in 4 mm segment
- MLA
Mean lumen area
- NC
Necrotic core
- NIRS-IVUS
Near-infrared spectroscopy-intravascular imaging
- PAV
Percentage atheroma volume
- PB
Plaque burden
- PCI
Percutaneous coronary intervention
- TAV
Total atheroma volume
Funding
This study has received joint funding by the British Heart Foundation (PG/17/18/32883), the University College London Biomedical Resource Centre (BRC492B), and Rosetrees Trust (A1773).
Compliance with ethical standards
Guarantor
This study is sponsored by the University College London, UK.
Conflict of interest
P.K., H.S., and J.H.C.R. are affiliated with Medis Medical Imaging Systems. The remaining authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.
Statistics and biometry
Professor Ibrahim Halil Tanboga MD, PhD. Department of Biostatistics and Cardiology, Nisantasi University Medical School, Istanbul, Turkey has significant statistical expertise. Professor Ibrahim Halil Tanboga is one of the listed authors.
Informed consent
Written informed consent was obtained from all subjects (patients) in this study.
Ethical approval
Institutional Review Board approval was obtained REC references: 17/SC/0566).
Study subjects or cohorts overlap
The data presented in this manuscript has not been published or presented elsewhere.
Methodology
Prospective
Single-centre observational study
Footnotes
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Nathan Angelo Lecaros Yap and Anantharaman Ramasamy contributed equally to this work.
Supplementary information
The online version contains supplementary material available at 10.1007/s00330-024-10996-x.
References
- 1.Kruk M, Noll D, Achenbach S et al (2014) Impact of coronary artery calcium characteristics on accuracy of CT angiography. JACC Cardiovasc Imaging. 10.1016/j.jcmg.2013.07.013 [DOI] [PubMed]
- 2.Monizzi G, Sonck J, Nagumo S et al (2020) Quantification of calcium burden by coronary CT angiography compared to optical coherence tomography. Int J Cardiovasc Imaging. 10.1007/s10554-020-01839-z [DOI] [PubMed]
- 3.Kurogi K, Ishii M, Ikebe S et al (2023) Calcium evaluation using coronary computed tomography in combination with optical coherence tomography. Int J Cardiovasc Imaging. 10.1007/s10554-023-02891-1 [DOI] [PubMed]
- 4.Belmonte M, Maeng M, Collet C et al (2023) Accuracy of a virtual PCI planner based on coronary CT angiography in calcific lesions. J Cardiovasc Comput Tomogr. 10.1016/j.jcct.2023.06.004 [DOI] [PubMed]
- 5.Norgaard BL, Gaur S, Leipsic J et al (2015) Influence of Coronary Calcification on the Diagnostic Performance of CT Angiography Derived FFR in Coronary Artery Disease: A Substudy of the NXT Trial. JACC Cardiovasc Imaging. 10.1016/j.jcmg.2015.06.003 [DOI] [PubMed]
- 6.Ramasamy A, Safi H, Moon JC et al (2020) Evaluation of the Efficacy of Computed Tomographic Coronary Angiography in Assessing Coronary Artery Morphology and Physiology: Rationale and Study Design. Cardiology. 10.1159/000506537 [DOI] [PubMed]
- 7.Ramasamy A, Hamid AKA, Cooper J et al (2023) Implications of computed tomography reconstruction algorithms on coronary atheroma quantification: Comparison with intravascular ultrasound. J Cardiovasc Comput Tomogr. 10.1016/j.jcct.2022.09.004 [DOI] [PubMed]
- 8.de Graaf MA, Broersen A, Kitslaar PH et al (2013) Automatic quantification and characterization of coronary atherosclerosis with computed tomography coronary angiography: cross-correlation with intravascular ultrasound virtual histology. Int J Cardiovasc Imaging. 10.1007/s10554-013-0194-x [DOI] [PubMed]
- 9.Brodoefel H, Reimann A, Heuschmid M et al (2008) Characterization of coronary atherosclerosis by dual-source computed tomography and HU-based color mapping: a pilot study. Eur Radiol. 10.1007/s00330-008-1019-5 [DOI] [PMC free article] [PubMed]
- 10.Horiguchi J, Fujioka C, Kiguchi M, Yamamoto H, Shen Y, Kihara Y (2011) In vitro measurement of CT density and estimation of stenosis related to coronary soft plaque at 100 kV and 120 kV on ECG-triggered scan. Eur J Radiol. 10.1016/j.ejrad.2009.08.002 [DOI] [PubMed]
- 11.Ramasamy A, Sokooti H, Zhang X et al (2023) Novel near-infrared spectroscopy-intravascular ultrasound-based deep-learning methodology for accurate coronary computed tomography plaque quantification and characterization. Eur Heart J Open. 10.1093/ehjopen/oead090 [DOI] [PMC free article] [PubMed]
- 12.Bajaj R, Huang X, Kilic Y et al (2021) Advanced deep learning methodology for accurate, real-time segmentation of high-resolution intravascular ultrasound images. Int J Cardiol. 10.1016/j.ijcard.2021.06.030 [DOI] [PubMed]
- 13.Bajaj R, Huang X, Kilic Y et al (2021) A deep learning methodology for the automated detection of end-diastolic frames in intravascular ultrasound images. Int J Cardiovasc Imaging. 10.1007/s10554-021-02162-x [DOI] [PMC free article] [PubMed]
- 14.Stone GW, Maehara A, Lansky AJ et al (2011) A prospective natural-history study of coronary atherosclerosis. N Engl J Med. 10.1056/NEJMoa1002358 [DOI] [PubMed]
- 15.Garcia-Garcia HM, Gomez-Lara J, Gonzalo N et al (2010) A comparison of the distribution of necrotic core in bifurcation and non-bifurcation coronary lesions: an in vivo assessment using intravascular ultrasound radiofrequency data analysis. EuroIntervention. 10.4244/EIJV6I3A54 [DOI] [PubMed]
- 16.Zhang S, Levin DC, Halpern EJ, Fischman D, Savage M, Walinsky P (2008) Accuracy of MDCT in assessing the degree of stenosis caused by calcified coronary artery plaques. AJR Am J Roentgenol. 10.2214/AJR.07.4026 [DOI] [PubMed]
- 17.Vavere AL, Arbab-Zadeh A, Rochitte CE et al (2011) Coronary artery stenoses: accuracy of 64-detector row CT angiography in segments with mild, moderate, or severe calcification--a subanalysis of the CORE-64 trial. Radiology. 10.1148/radiol.11110537 [DOI] [PMC free article] [PubMed]
- 18.Joshi SB, Okabe T, Roswell RO et al (2009) Accuracy of computed tomographic angiography for stenosis quantification using quantitative coronary angiography or intravascular ultrasound as the gold standard. Am J Cardiol. 10.1016/j.amjcard.2009.05.052 [DOI] [PubMed]
- 19.Park JK, Kim JY, Kwon HM et al (2014) Multidetector computed tomography for the evaluation of coronary artery disease; the diagnostic accuracy in calcified coronary arteries, comparing with IVUS imaging. Yonsei Med J. 10.3349/ymj.2014.55.3.599 [DOI] [PMC free article] [PubMed]
- 20.Li Y, Zhang J, Lu Z, Pan J (2021) Discrepant findings of computed tomography quantification of minimal lumen area of coronary artery stenosis: correlation with intravascular ultrasound. Eur J Radiol. 10.1016/j.ejrad.2012.03.010 [DOI] [PubMed]
- 21.Noll D, Kruk M, Demkow M et al (2018) Patterns of Coronary Calcification and Their Impact on the Diagnostic Accuracy of Computed Tomography Coronary Angiography. J Comput Assist Tomogr. 10.1097/RCT.0000000000000681 [DOI] [PubMed]
- 22.Matsumoto H, Watanabe S, Kyo E et al (2019) Standardized volumetric plaque quantification and characterization from coronary CT angiography: a head-to-head comparison with invasive intravascular ultrasound. Eur Radiol. 10.1007/s00330-019-06219-3 [DOI] [PMC free article] [PubMed]
- 23.Abbara S, Blanke P, Maroules CD et al (2016) SCCT guidelines for the performance and acquisition of coronary computed tomographic angiography: A report of the society of Cardiovascular Computed Tomography Guidelines Committee: Endorsed by the North American Society for Cardiovascular Imaging (NASCI). J Cardiovasc Comput Tomogr. 10.1016/j.jcct.2016.10.002 [DOI] [PubMed]
- 24.Erdogan E, Huang X, Cooper J et al (2022) End-diastolic segmentation of intravascular ultrasound images enables more reproducible volumetric analysis of atheroma burden. Catheter Cardiovasc Interv. 10.1002/ccd.29917 [DOI] [PubMed]
- 25.Pu J, Mintz GS, Biro S et al (2014) Insights into echo-attenuated plaques, echolucent plaques, and plaques with spotty calcification: novel findings from comparisons among intravascular ultrasound, near-infrared spectroscopy, and pathological histology in 2,294 human coronary artery segments. J Am Coll Cardiol. 10.1016/j.jacc.2014.02.576 [DOI] [PubMed]
- 26.Tomaniak M, Hartman EMJ, Tovar Forero MN et al (2022) Near-infrared spectroscopy to predict plaque progression in plaque-free artery regions. EuroIntervention. 10.4244/EIJ-D-21-00452 [DOI] [PMC free article] [PubMed]
- 27.Kitahara S, Kataoka Y, Miura H et al (2021) The feasibility and limitation of coronary computed tomographic angiography imaging to identify coronary lipid-rich atheroma in vivo: Findings from near-infrared spectroscopy analysis. Atherosclerosis. 10.1016/j.atherosclerosis.2021.02.019 [DOI] [PubMed]
- 28.Maurovich-Horvat P, Schlett CL, Alkadhi H et al (2012) Differentiation of early from advanced coronary atherosclerotic lesions: systematic comparison of CT, intravascular US, and optical frequency domain imaging with histopathologic examination in ex vivo human hearts. Radiology. 10.1148/radiol.12111891 [DOI] [PubMed]
- 29.Si-Mohamed SA, Boccalini S, Lacombe H et al (2022) Coronary CT Angiography with Photon-counting CT: First-In-Human Results. Radiology. 10.1148/radiol.211780 [DOI] [PubMed]
- 30.Marsh JF Jr. VanMeter PD, Rajendran K, Leng S, McCollough CH (2023) Ex vivo coronary calcium volume quantification using a high-spatial-resolution clinical photon-counting-detector computed tomography. J Med Imaging (Bellingham). 10.1117/1.JMI.10.4.043501 [DOI] [PMC free article] [PubMed]
- 31.Zsarnoczay E, Fink N, Schoepf UJ et al (2023) Ultra-high resolution photon-counting coronary CT angiography improves coronary stenosis quantification over a wide range of heart rates - A dynamic phantom study. Eur J Radiol. 10.1016/j.ejrad.2023.110746 [DOI] [PubMed]
- 32.Sandstedt M, Marsh J Jr Rajendran K et al (2021) Improved coronary calcification quantification using photoncounting- detector CT: an ex vivo study in cadaveric specimens. Eur Radiol. 10.1007/s00330-021-07780-6 [DOI] [PMC free article] [PubMed]
- 33.Petritsch B, Petri N, Weng AM et al (2021) Photon-Counting Computed Tomography for Coronary Stent Imaging: In Vitro Evaluation of 28 Coronary Stents. Invest Radiol. 10.1097/RLI.0000000000000787 [DOI] [PubMed]
- 34.Messerli M, Rengier F, Desbiolles L et al (2016) Impact of Advanced Modeled Iterative Reconstruction on Coronary Artery Calcium Quantification. Acad Radiol. 10.1016/j.acra.2016.08.008 [DOI] [PubMed]
- 35.van der Giessen AG, Gijsen FJ, Wentzel JJ et al (2011) Small coronary calcifications are not detectable by 64-slice contrast enhanced computed tomography. Int J Cardiovasc Imaging. 10.1007/s10554-010-9662-8 [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.