If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Semmelweis University Heart and Vascular Center, 1122 Budapest, Városmajor street 68., HungarySemmelweis University Medical Imaging Center, 1082 Budapest, Korányi Sándor street 2., Hungary
Cardiovascular Imaging Research Center, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USASemmelweis University Heart and Vascular Center, 1122 Budapest, Városmajor street 68., Hungary
Coronary CT angiography (CCTA) pericoronary adipose tissue (PCAT) markers are promising indicators of inflammation.
Objective
To determine the effect of patient and imaging parameters on the associations between non-calcified plaque (NCP) and PCAT attenuation and gradient.
Methods
This was a single-center, retrospective analysis of consecutive patients with stable chest pain who underwent CCTA and had zero calcium scores. CCTA images were evaluated for the presence of NCP, obstructive stenosis, segment stenosis and involvement score (SSS, SIS), and high-risk plaque (HRP). PCAT markers were assessed using semi-automated software. Uni- and multivariable regression models correcting for patient and imaging characteristics between plaque and PCAT markers were evaluated.
Results
Overall, 1652 patients had zero calcium score (mean age: 51 years ± 11 [SD], 871 women); PCAT attenuation values ranged between −123 HU and −51 HU, and 649 patients had plaque. In univariable analysis, the presence of NCP, SSS, SIS, and HRP were associated with PCAT attenuation (2, 1, 1, 6 HU; respectively; p < .001 all); while obstructive stenosis was not (1 HU, p = .58). In multivariable analysis, none of the plaque markers were associated with PCAT attenuation (0 HU p = .93, 0 HU p = .39, 1 HU p = .18, 2 HU p = .10, 1 HU p = .71, respectively), while patient and imaging characteristics showed significant associations, such as: male sex (1 HU, p = .003), heart rate [1/min] (−0.2 HU, p < .001), 120 kVp (8 HU, p < .001) and pixel spacing [mm3] (32 HU, p < .001). Similar results were observed for PCAT gradient.
Conclusion
PCAT markers were significantly associated with NCP, however the associations did not persist following correction for patient and imaging characteristics.
This residual cardiovascular risk stimulates the search for markers capable of identifying patients susceptible to major adverse cardiac events. Imaging is a noninvasive alternative to blood-based markers for detecting pathological changes indicative of diseases.
Validated imaging biomarkers as decision-making tools in clinical trials and routine practice: current status and recommendations from the EIBALL∗ subcommittee of the European Society of Radiology (ESR).
Inflammation is increasingly being recognized as an equally important pathological mechanism next to lipoprotein accumulation and endothelial dysfunction to promote coronary artery disease (CAD).
Relationship between changes in pericoronary adipose tissue attenuation and coronary plaque burden quantified from coronary computed tomography angiography.
Inflammation inhibits local adipogenesis, leading to smaller adipocytes with lower intracellular lipid content. This in turn results in higher Hounsfield unit (HU) values of PCAT on CCTA,
which can be captured by PCAT markers (PCAT attenuation and PCAT gradient). Mean PCAT attenuation is the average HU within radial distance of the vessel within the range of −190- and −30 HU, while PCAT gradient is defined as the relative change in PCAT HU values perpendicular to the vessel.
PCAT attenuation was developed to describe adipocyte and lipid content while PCAT gradient provides additional information regarding local stimuli and is superior in identifying non-calcified plaque (NCP) as compared to PCAT attenuation.
Pericoronary adipose tissue and quantitative global non-calcified plaque characteristics from CT angiography do not differ in matched South Asian, East Asian and European-origin Caucasian patients with stable chest pain.
Based on these findings, PCAT markers may be used to differentiate between stages and characteristics of CAD. However, reported changes in PCAT metrics between patient groups are subtle (Supplementary table 1), and may be affected by patient and image acquisition characteristics.
Coronary calcium score is a well-established independent risk marker for major adverse cardiac events.
A zero calcium score is an indicator of low cardiovascular risk. However, even in these patients, there is a considerable prevalence and progression of NCP.
Our aim was 1) to evaluate the associations between CAD markers (presence of plaque, presence of obstructive stenosis, segment stenosis score [SSS], segment involvement score [SIS], presence of high-risk plaque) and PCAT attenuation and gradient in patients with a low cardiovascular risk (calcium score = 0); 2) to evaluate to what degree patient and imaging characteristics influence PCAT markers; 3) to assess whether potential associations between NCP and PCAT markers persisted after correcting for patient and imaging characteristics; 4) to validate our findings in a cohort of individuals with zero calcium scores but imaged using a different CT scanner, and on a group of patients with moderate to severe stenosis on CCTA.
2. Materials and methods
2.1 Patients
In our single-center, retrospective, observational study done at a university teaching hospital, 4120 consecutive patients underwent CCTA between April 2016 and August 2019 for the assessment of stable chest pain. In a previous analysis of these patients, we examined the effect of clinical characteristics, image acquisition, and calcium score values on additional testing following CCTA.
In the current sub-analysis, only individuals who had a calcium score of zero were included (Zero calcium score group; n = 1652), which is an established indicator of low cardiovascular risk.
Our study inclusion criteria were 1) patients with suitable image quality for PCAT analysis and 2) zero calcium score on non-contrast CT images. Exclusion criteria were 1) images not accessible from the clinical PACS system, 2) poor image quality evaluated by a consensus read of BM and VB, both with 3 years of experience in CCTA, 3) CCTA acquired at a tube voltage other than 100 or 120 kVp (only a minority of CCTA were performed at 80 or 140 kVp), 4) prior known CAD.
2.2 Validation groups
To validate our findings, we retrospectively identified two additional validation groups consisting of 330 individuals each (20% of the Zero calcium score group, respectively). The first validation group (Zero calcium score group – different scanner; n = 330) consisted of individuals with identical inclusion and exclusion criteria but scanned using another scanner (CardioGraphe, GE Healthcare). The second validation (Moderate to severe CAD group; n = 330) group consisted of patients scanned on the same scanner as the main study group (Brilliance iCT 256, Philips Healthcare) but with moderate to severe (50–99%) stenosis.
2.3 Sensitivity analysis
To assess whether our findings regarding the association between patient and imaging characteristics and PCAT markers are true in individuals without any CAD, we conducted a sensitivity analysis in individuals from our main study group with zero calcium scores and no NCP.
2.4 Coronary CTA acquisition, plaque characterization, and image quality assessment
Details are provided in supplemental material. In brief, for CCTA data sets acquired using the Philips scanner, 0.8-mm section thickness and 0.4-mm spacing between the sections was used, while for the CardioGraphe 0.5-mm section thickness with continuous sections was reconstructed to a 512x512 matrix with varying field of views optimized cardiac evaluation. High-risk plaque features were defined as low-attenuation plaque, positive remodeling, spotty calcification, and napkin-ring sign using standard definitions.
Society of cardiovascular computed tomography/north American society of cardiovascular imaging - expert consensus document on coronary CT imaging of atherosclerotic plaque.
PCAT analysis was performed around the proximal segment of the right coronary artery using AutoPlaque (version 2.0, Cedars-Sinai Medical Center), as previously described.
PCAT analysis was performed in the 40 mm proximal RCA segment following the first 10 mm from the RCA ostium as a standardized model for the analysis of PCAT.
Relationship between changes in pericoronary adipose tissue attenuation and coronary plaque burden quantified from coronary computed tomography angiography.
Non-invasive detection of coronary inflammation using computed tomography and prediction of residual cardiovascular risk (the CRISP CT study): a post-hoc analysis of prospective outcome data.
Adipose tissue is defined as voxels with HU values between −190 HU and −30 HU, therefore all PCAT markers were calculated after removing all voxels with values below or above these thresholds.
PCAT attenuation was defined as the average HU attenuation within a radial distance from the outer coronary artery wall equal to the average diameter of the vessel.
Given the average radius of the right coronary artery was 3.4 ± 0.3 mm in the zero calcium score group, PCAT attenuation was uniformly calculated using a 3-mm radial distance from the vessel wall for all individuals.
PCAT gradient was calculated as the percentage change in PCAT attenuation when comparing HU values within the 3 mm radius as described above to non-PCAT HU values at a 20 mm distance from the outer vessel wall, defined as: = 100%∗(PCAT attenuation - non-PCAT attenuation)/|PCAT attenuation|.
All measurements were done blinded to all clinical data by BM and VB, both with 3 years of experience in coronary CTA.
2.6 Article search for PCAT manuscripts
We performed systematic search for original research article published until December 2021 to provide an overview of the effect sizes reported in the literature for PCAT markers. The search was performed in several electronic databases (PubMed, Scopus, Medline, Embase, Web of Science, The Cochrane Library and ProQuest). The following MeSH terms and free text words were used: pericoronary adipose tissue, fat attenuation index, volumetric perivascular characterization index, coronary CT angiography, to identify articles regarding PCAT. The data is reported in Supplementary table 1.
2.7 Cardiovascular outcome analysis
All patients were followed up using the Hungarian Myocardial Infarction registry to ascertain whether they experienced an acute myocardial infarction following the CT examination.
Quality assurance of national internet-based patient register data: experiences during the operation of the Hungarian Myocardial Infarction Registry, 2010-2020.
The registry is continuously checked and validated by dedicated personnel resulting in more than 92% of events being registered as compared to the national healthcare provider reimbursement dataset.
Comparison of management and outcomes of ST-segment elevation myocardial infarction patients in Estonia, Hungary, Norway, and Sweden according to national ongoing registries.
Eur Heart J Qual Care Clin Outcomes.2022; 8: 307-314
Continuous variables are presented as mean and standard deviation, while categorical parameters are presented as frequencies with percentages. We used linear regression models to identify clinical, imaging and CAD predictors of PCAT attenuation and PCAT gradient. We built multivariable models including all predictors into the regression model. Due to the limited number of events, survival analyses were not done. All analyses were conducted in R environment (v4.0.2). A p < .05 was considered significant.
The institutional review board approved our study protocol. Due to the retrospective nature of our analysis, written informed consent was waived. All procedures used in this study were in accordance with the Health Insurance Portability and Accountability Act, local and federal regulations, and the Declaration of Helsinki.
3. Results
3.1 Patient characteristics
3.1.1 Zero calcium score group
Between 2016 and 2019, 4120 consecutive patients underwent CCTA for the assessment of stable chest pain. 1839 patients had a calcium score of zero. From these individuals 187 patients were excluded from the analyses. Overall, we analyzed 1652 patients in the zero calcium score group who met all inclusion and exclusion criteria (Fig. 1). The mean age of patients was 51 ± 11 years, and 871 patients (53%) were female. Overall, 649 (39%) individuals had plaque. Patient characteristics and CCTA findings are summarized in Table 1.
Fig. 1Patient flow charts of the different study groups. Abbreviations: AMI: acute myocardial infarction, PCAT: pericoronary adipose tissue.
3.1.2 Zero calcium score group – different scanner
Overall, 398 images were screened to have 330 scans suitable for PCAT measurements (Fig. 1). The average age was 52 ± 12 years and 53% (174/330) were males. Altogether, 137 individuals had coronary plaque (Table 1).
3.1.3 Moderate to severe CAD group
Overall, 402 scans were screened to achieve 330 scans suitable for PCAT measurements (Fig. 1). The average age was 63 ± 10 years and 61% (200/330) were males. Per inclusion criteria, everyone had at least one obstructive plaque. In 43 individuals (13%) there was at least one high risk plaque (Table 1).
3.2 Predictors of PCAT attenuation in the zero calcium score group
The PCAT attenuation values ranged between −123 HU and −51 HU in individuals with a calcium score of zero. Average PCAT attenuation was −92 ± 9 HU, while the median value was: −93 HU (25th percentile: −99 HU; 75th percentile: −87 HU). Among the 1003 individuals who had zero calcium score and no plaque, PCAT attenuation ranged between: −123 HU and −51 HU; the average was: −93 ± 9 HU, and the median was: −94 HU (25th percentile: −100 HU; 75th percentile: −88 HU).
3.2.1 Univariable analysis
Based on the univariable analyses, among CAD characteristics, the presence of NCP, SSS, SIS, and high-risk plaque were associated with an increasing PCAT attenuation (2, 1, 1, 6 HU respectively; p < .001 for all), while presence of obstructive stenosis showed no evidence of an association (1 HU, p = .58). Detailed results are shown in Table 2.
Table 2The relationship between clinical characteristics, CCTA acquisition parameters, coronary artery disease characteristics and PCAT attenuation in the different study groups.
Predictors
Zero calcium score group (n = 1652)
Zero calcium score group – different scanner (n = 330)
Moderate to severe CAD group (n = 330)
Univariable model
Multivariable model
Univariable model
Multivariable model
Univariable model
Multivariable model
HU
95% CI
p
HU
95% CI
p
HU
95% CI
p
HU
95% CI
p
HU
95% CI
p
HU
95% CI
p
Clinical characteristics
Age [y]
0.1
[0.0; 0.1]
.001
0.0
[-0.1; 0.0]
.23
0.1
[-0.04; 0.2]
.27
0.1
[-0.02; 0.2]
.09
0.01
[-0.1; 0.2]
.84
−0.1
[-0.2; 0.1]
.44
Male sex
2.6
[1.7; 3.5]
<.001
1.4
[0.5; 2.3]
.003
0.9
[-1.6; 3.5]
.47
2.6
[-0.1; 5.3]
.06
3.0
[0.001; 6.0]
.05
1.1
[-1.6; 3.9]
.41
BMI [kg/m2]
0.1
[0.0; 0.2]
.14
−0.4
[-0.5; −0.3]
<.001
−0.4
[-0.6; −0.1]
.01
−0.4
[-0.7; −0.1]
.01
−0.3
[-0.6; 0.05]
.09
−0.4
[-0.7; −0.1]
.02
Hypertension [mmHg]
1.1
[0.2; 2.0]
.01
0.3
[-0.6; 1.1]
.53
0.1
[-2.4; 2.7]
.93
0.6
[-2.1; 3.3]
.67
−0.1
[-3.4; 3.2]
.94
−0.9
[-3.7; 1.9]
.53
Diabetes
0.7
[-1.1; 2.4]
.46
0.7
[-1.0; 2.3]
.42
−1.3
[-13.0; 10.3]
.82
2.4
[-8.8; 13.7]
.67
0.3
[-4.0; 4.5]
.90
−1.4
[-4.9; 2.1]
.42
Dyslipidemia
0.5
[-0.5; 1.5]
.30
0.1
[-0.8; 1.0]
.88
−1.0
[-3.8; 1.8]
.48
−0.7
[-3.5; 2.0]
.61
1.3
[-1.7; 4.2]
.41
2.1
[-0.5; 4.6]
.11
Smoking
0.3
[-0.9; 1.6]
.62
0.1
[-1.0; 1.2]
.92
0.8
[-2.8; 4.4]
.66
1.5
[-1.9; 4.9]
.39
0.1
[-3.9; 4.2]
.95
−0.03
[-3.4; 3.3]
.99
CCTA acquisition parameters
Non-sinus rhythm
1.3
[-1.9; 4.4]
.44
2.6
[-0.3; 5.5]
.08
−4.5
[-13.4; 4.3]
.31
4.9
[-4.7; 14.4]
.32
−8.9
[-18.4; 0.6]
.07
0.8
[-8.5; 10.0]
.87
Heart rate [beats/minute]
−0.2
[-0.3; −0.2]
<.001
−0.2
[-0.2; −0.1]
<.001
−0.2
[-0.3; −0.1]
<.001
−0.2
[-0.3; −0.1]
<.001
−0.1
[-0.2; 0.03]
.12
−0.1
[-0.2; 0.1]
.32
Poor image quality
−3.6
[-5.4; −1.8]
<.001
−1.3
[-3.0; 0.5]
.15
−7.4
[-11.7; −3.1]
<.001
−5.2
[-9.5; −0.9]
.02
−5.7
[-11.6; 0.1]
.05
−0.6
[-5.4; 4.2]
.80
Tube voltage 120 kVp
6.2
[5.4; 7.1]
<.001
7.7
[6.7; 8.7]
<.001
0.2
[-0.1; 0.4]
.14
0.2
[-0.1; 0.4]
.17
0.5
[0.3; 0.7]
<.001
0.5
[0.3; 0.7]
<.001
Tube current [mAs]
−0.02
[-0.03; −0.004]
.01
−0.02
[-0.03; −0.01]
<.001
−0.01
[-0.02; −0.0002]
.04
−0.01
[-0.02; 0.002]
.12
−0.1
[-0.1; −0.05]
<.001
−0.04
[-0.1; −0.02]
<.001
CNR
0.0
[-0.1; 0.1]
.94
−1.3
[-1.7; −1.0]
<.001
0.4
[0.2; 0.6]
<.001
0.3
[-0.8; 1.4]
.59
−0.1
[-0.3; 0.1]
.23
−1.5
[-2.5; −0.5]
.002
SNR
0.0
[0.0; 0.1]
.48
1.4
[1.0; 1.7]
<.001
0.4
[0.2; 0.6]
<.001
0.1
[-1.1; 1.3]
.90
−0.1
[-0.3; 0.2]
.56
1.6
[0.5; 2.7]
.004
Pixel Spacing [mm3]
30.5
[22.5; 38.6]
<.001
32.4
[24.9; 39.9]
<.001
49.3
[-20.4; 119.1]
.16
80.9
[13.2; 148.5]
.02
129.6
[103.5; 155.7]
<.001
100.0
[73.9; 126.1]
<.001
Coronary artery disease characteristics
Presence of plaque
2.3
[1.4; 3.3]
<.001
0.1
[-1.2; 1.4]
.93
2
[-0.6; 4.5]
.13
3.6
[-0.2; 7.3]
.06
–
–
–
–
–
–
Presence of obstructive stenosis
1.3
[-3.4; 6.1]
.58
1.1
[-4.5; 6.6]
.71
9.9
[-3.5; 23.3]
.15
13.1
[-2.0; 28.2]
.09
3.5
[-0.2; 7.2]
.06
2.3
[-1.5; 6.2]
.24
SSS [n]
0.6
[0.3; 0.8]
<.001
−0.4
[-1.1; 0.4]
.39
−0.1
[-0.7; 0.6]
.88
−0.5
[-2.8; 1.8]
.66
0.5
[0.2; 0.7]
<.001
−0.2
[-0.8; 0.3]
.46
SIS [n]
1.0
[0.6; 1.4]
<.001
0.8
[-0.4; 2.0]
.18
−0.1
[-0.9; 0.8]
.84
−0.4
[-3.3; 2.4]
.77
1.1
[0.6; 1.6]
<.001
0.9
[-0.2; 1.9]
.09
Presence of HRP
5.6
[3.2; 8.1]
<.001
1.9
[-0.3; 4.2]
.10
−3.3
[-16.7; 10.1]
.62
−0.2
[-14.0; 13,7]
.98
2.9
[-1.5; 7.2]
.19
3.7
[0.1; 7.2]
.04
Univariable and multivariable linear regression models demonstrating the effects of clinical characteristics, CTA acquisition setting and CAD characteristics on PCAT attenuation. Significant predictors are marked in bold. All variables were entered into the multivariable models.
Abbreviations: BMI: body mass index, CNR: contrast to noise ratio, CTA: coronary CT angiography, HRP: High-risk plaque, kVp: kilovoltage peak, mAs: milliampere-second, PCAT: pericoronary adipose tissue, SIS: Segment involvement score, SNR: Signal to noise ratio, SSS: Segment Stenosis Score.
After correcting for all factors, among clinical characteristics and image acquisition parameters, independently male sex, 120 kVp instead of 100 kVp, signal to noise ratio and pixel spacing were associated with higher PCAT attenuation values (1 HU p = .003, 8 HU p < .001, 1 HU p < .001, 32 HU p < .001; respectively). While each kg/m2 increase in body mass index, beat/minute in heart rate, mAs in tube current, unit in contrast to noise ratio independently decreased PCAT attenuation (−0.4 HU, −0.2 HU, −0.02 HU, −1.3 HU; p < .001 for all; respectively). Furthermore, none of the CAD characteristics was associated with PCAT attenuation (presence of NCP (0 HU, p = .93) presence of obstructive stenosis (1 HU, p = .71), SSS (0 HU, p = .39), SIS (1 HU, p = .18), high-risk plaque (2 HU, p = .10)). Detailed results are presented in Table 2.
3.2.3 Sensitivity analysis
In the 1003 individuals with calcium score of zero and no CAD, in univariable results we found the same patient and image acquisition parameters to be associated with PCAT attenuation as the whole group, except for hypertension and tube current which were non-significant. In multivariable analysis, we found the same patient and imaging characteristics to be significant. Results are presented in Supplementary table 2.
3.3 Predictors of PCAT gradient in the zero calcium score group
The PCAT gradient values ranged between −31% and 75% in individuals with a calcium score of zero. The average PCAT gradient was: −1±12%, while the median value was: −2% (25th percentile: −9%; 75th percentile: 6%).
3.3.1 Univariable analysis
All CAD characteristics were significantly associated with PCAT gradient: presence of NCP (−2% p < .001), presence of obstructive stenosis (8% p = .01), SSS (1% p < .001), SIS (1% p < .001), high-risk plaque (5% p < .001). Detailed results are presented in Table 3.
Table 3The relationship between clinical characteristics, CCTA acquisition parameters, coronary artery disease characteristics and PCAT gradient in the different study groups.
Predictors
Zero calcium score group (n = 1652)
Zero calcium score group – different scanner (n = 330)
Moderate to severe CAD group (n = 330)
Univariable model
Multivariable model
Univariable model
Multivariable model
Univariable model
Multivariable model
%
95% CI
p
%
95% CI
p
%
95% CI
p
%
95% CI
p
%
95% CI
p
%
95% CI
p
Clinical characteristics
Age [y]
0.1
[0.1; 0.2]
<.001
0.1
[0.1; 0.2]
<.001
0.1
[-0.04; 0.3]
.12
0.2
[0.03; 0.4]
.03
0.2
[-0.1; 0.4]
.14
0.1
[-0.2; 0.3]
.47
Male sex
3.1
[2.0; 4.2]
<.001
3.1
[1.8; 4.3]
<.001
6.8
[2.6; 11.0]
.002
9.1
[4.5; 13.7]
<.001
5.6
[0.3; 10.9]
.04
4.0
[-1.2; 9.1]
.13
BMI [kg/m2]
0.4
[0.3; 0.5]
<.001
−0.1
[-0.2; 0.1]
.38
−0.2
[-0.7; 0.2]
.32
−0.4
[-0.9; 0.03]
.06
−0.1
[-0.7; 0.4]
.63
−0.2
[-0.7; 0.4]
.55
Hypertension [mmHg]
2.2
[1.1; 3.4]
<.001
0.2
[-1.0; 1.3]
.78
1.5
[-2.8; 5.7]
.49
−0.1
[-4.7; 4.4]
.96
3.2
[-2.6; 9.1]
.28
0.5
[-4.8; 5.8]
.84
Diabetes
1.6
[-0.6; 3.9]
.16
0.1
[-2.1; 2.2]
.97
−7.6
[-27.0; 11.8]
.44
−4.4
[-23.5; 14.7]
.65
1.2
[-6.3; 8.7]
.76
−3.4
[-9.9; 3.2]
.31
Dyslipidemia
1.8
[0.6; 3.1]
.004
0.8
[-0.5; 2.0]
.22
−1.2
[-5.9; 3.5]
.63
−0.4
[-5.1; 4.3]
.85
3.5
[-1.7; 8.8]
.19
3.8
[-0.9; 8.5]
.12
Smoking
0.8
[-0.8; 2.3]
.35
0.9
[-0.6; 2.4]
.25
−1.8
[-7.8; 4.2]
.55
−0.2
[-6.1; 5.6]
.93
−0.1
[-7.3; 7.0]
.97
0.03
[-6.2; 6.2]
.99
CCTA acquisition parameters
Non-sinus rhythm
4.8
[0.7; 8.8]
.02
5.6
[1.7; 9.5]
.01
−5.2
[-19.9; 9.6]
.49
7.7
[-8.5; 23.9]
.35
−17.6
[-34.4; −0.9]
.04
−0.8
[-18.1; 16.5]
.93
Heart rate [beats/minute]
−0.1
[-0.2; −0.1]
<.001
−0.1
[-0.15; −0.01]
.02
−0.3
[-0.4; −0.1]
.002
−0.3
[-0.5; −0.1]
.005
−0.2
[-0.4; −0.012]
.04
−0.1
[-0.3; 0.1]
.21
Poor image quality
−3.2
[-5.5; −0.8]
.008
−2.2
[-4.5; 0.1]
.07
−10.1
[-17.3; −2.9]
.01
−7.8
[-15.1; −0.5]
.04
−11.0
[-21.4; −0.7]
.04
−2.8
[-11.9; 6.2]
.54
Tube voltage 120 kVp
5.4
[4.3; 6.6]
<.001
4.0
[2.7; 5.3]
<.001
0.4
[0.02; 0.7]
.04
0.3
[-0.1; 0.6]
.19
0.7
[0.2; 1.1]
.002
0.4
[0.1; 0.8]
.02
Tube current [mAs]
0.0
[0.0; 0.0]
.82
−0.02
[-0.04; −0.01]
.003
−0.01
[-0.03; 0.01]
.36
−0.01
[-0.03; 0.01]
.38
−0.1
[-0.1; −0.1]
<.001
−0.1
[-0.1; −0.03]
<.001
CNR
−0.2
[-0.3; −0.1]
<.001
−1.4
[-1.8; −0.9]
<.001
0.5
[0.1; 0.8]
.004
0.4
[-1.4; 2.2]
.64
−0.2
[-0.5; 0.1]
.26
−2.3
[-4.2; −0.5]
.01
SNR
−0.2
[-0.3; −0.1]
<.001
1.3
[0.8; 1.8]
<.001
0.5
[0.2; 0.9]
.003
0.03
[-2.0; 2.0]
.98
−0.1
[-0.5; 0.3]
.57
2.6
[0.5; 4.6]
.02
Pixel Spacing [mm3]
41.1
[30.9; 51.4]
<.001
36.5
[26.4; 46.6]
<.001
16.1
[-100.6; 132.7]
.79
63.3
[-51.7; 178.2]
.28
211.9
[164.8; 259.0]
<.001
157.4
[108.6; 206.1]
<.001
Coronary artery disease characteristics
Presence of plaque
−2.2
[2.0; 4.4]
<.001
0.2
[-1.5; 2.0]
.79
5.3
[1.0; 9.6]
.02
5.1
[-1.2; 11.5]
.11
–
–
–
–
–
–
Presence of obstructive stenosis
7.5
[1.5; 13.5]
.01
4.0
[-3.4; 11.5]
.28
14.0
[-8.3; 36.4]
.22
8.2
[-17.5; 33.9]
.53
6.4
[-0.1; 13.0]
.05
2.0
[-5.2; 9.2]
.58
SSS [n]
0.9
[0.6; 1.2]
<.001
0.0
[-1.0; 1.1]
.94
0.7
[-0.4; 1.8]
.21
0.3
[-3.6; 4.1]
.88
1.0
[0.6; 1.4]
<.001
−0.1
[-1.1; 1.0]
.92
SIS [n]
1.4
[0.9; 1.8]
<.001
0.4
[-1.2; 2.0]
.65
0.8
[-0.6; 2.2]
.25
−1.5
[-6.3; 3.4]
.56
2.3
[1.5; 3.2]
<.001
1.4
[-0.6; 3.3]
.17
Presence of HRP
5.3
[2.2; 8.4]
<.001
1.8
[-1.3; 4.9]
.25
10.2
[-12.1; 32.6]
.37
10.0
[-13.5; 33.5]
.40
2.7
[-5.0; 10.4]
.49
3.5
[-3.1; 10.2]
.30
Univariable and multivariable linear regression models demonstrating the effects of clinical characteristics, CTA acquisition setting and CAD characteristics on PCAT gradient. Significant predictors are marked in bold. All variables were entered into the multivariable models.
Abbreviations: BMI: body mass index, CNR: contrast to noise ratio, CTA: coronary CT angiography, HRP: High-risk plaque, kVp: kilovoltage peak, mAs: milliampere-second, PCAT: pericoronary adipose tissue, SIS: Segment involvement score, SNR: signal to noise ratio, SSS: segment stenosis score.
In multivariable analysis, age (0.1% p < .001), male sex (3% p < .001), non-sinus rhythm (6% p = .01), 120 kVp (4% p < .001), signal to noise ratio (1% p < .001) and pixel spacing (37% p < .001) were all independently associated with higher PCAT gradient values. Also, each beat per minute increase in heart rate (−0.1% p = .02), each increase in mAs (−0.02% p = .003) and contrast to noise ratio (−1% p < .001) was independently associated with lower PCAT gradient values. Furthermore, none of the CAD characteristics showed evidence of an association with PCAT gradient: presence of NCP (0% p = .79), presence of obstructive stenosis (4% p = .28), SSS (0% p = .94), SIS (0% p = .65) and presence of high-risk plaque (2% p = .25). Detailed regression results are presented in Table 3.
3.3.3 Sensitivity analysis
In the individuals with calcium score of zero and no CAD (n = 1003), in univariable analyses we found the same parameters to have a significant association with PCAT gradient as in the whole patient group with zero calcium score, except for poor image quality and heart rate which were non-significant. In multivariable analysis, the same patient characteristics and imaging characteristics were significant, except for heart rate. Results are presented in Supplementary table 3.
3.4 Validation group: zero calcium score group – different scanner
Using a different scanner, PCAT attenuation values were higher compared to the main zero calcium score group (−77 ± 12 vs −92 ± 9, p < .001). The attenuation values ranged between −105 HU and −49 HU, while gradient values ranged between −21% and 90%.
3.4.1 Univariable analysis
In contrast to our main study group, none of the plaque markers were associated with PCAT attenuation (Table 2), while only the presence of plaque was associated with higher PCAT gradient values (5% p = .02; Table 3). However, several patient and imaging characteristics were associated with PCAT HU and gradient similar to our findings in the main zero calcium score group (Table 2, Table 3).
3.4.2 Multivariable analysis
Similar to our findings in the main study group with zero calcium score, none of the plaque markers were associated with PCAT attenuation or gradient in multivariable analyses. However similarly, several patient and imaging characteristics remained significantly associated with PCAT markers (Table 2, Table 3).
3.5 Validation group: moderate to severe CAD group
PCAT values ranged between −82 HU and −50 HU, while the gradient values ranged between −22% and 116%.
3.5.1 Univariable analysis
Among plaque markers, SSS and SIS were associated with PCAT attenuation (1 HU p < .001; 1 HU p < .001, respectively). In case of PCAT gradient, similarly only SSS and SIS were significantly associated (1% p < .001, 2% p < .001, respectively). However, several patient and image acquisition characteristics also showed a significant association with PCAT HU (Table 2) and gradient (Table 3).
3.5.2 Multivariable analysis
After correcting for all patient and image acquisition parameters, only the presence of high risk plaque showed borderline association with PCAT attenuation (4 HU p = .04; Table 2), while similar to the main cohort of patients with zero calcium score, none of the plaque markers were associated with PCAT gradient (Table 3). Nevertheless, similarly to the initial cohort, BMI and several image acquisition parameters were associated with PCAT HU (Table 2), while only image acquisition settings showed a significant association with PCAT gradient (Table 3).
Representative images from all three groups showing how different heart rates influence PCAT attenuation and PCAT gradient values are shown in Fig. 2.
Fig. 2Representative images showing PCAT attenuation and PCAT gradient differences in patients with different heart rates. Panel A, B show two representative images from the zero calcium score group. Patients were scanned on a Philips - Brilliance iCT 256 and had a calcium score of zero. Panel C, D show two representative cases from the zero calcium score group - different scanner group. Patients were scanned on a GE - CardioGraphe and had a calcium score of zero. Panel E, F show two representative images from the moderate to severe CAD group. Patients were scanned on a Philips - Brilliance iCT 256 and had obstructive CAD. Patients within all three cohorts had a wide range of PCAT attenuation and gradient values. Abbreviations: BMI: body mass index, CAD: coronary artery disease, HR: heart rate, HU: Hounsfield unit, PCAT: pericoronary adipose tissue.
During the average follow-up time of 3.0 years in the zero calcium score group, two individuals suffered acute myocardial infarction, their PCAT attenuation and gradient values were: −93 HU and −79 HU and −1% and 19% respectively. In the validation cohort with zero calcium score scanned on a different scanner, there were no events within the average follow-up time of 0.7 years, while there were 7 events in the moderate to severe CAD group over the average of 2.2 years. Their PCAT attenuation ranged between: −88 HU and −64 HU, while the gradient values were between 7% and 51%.
4. Discussion
PCAT attenuation and gradient are considered markers of perivascular inflammation and have been shown to be associated with several cardiovascular conditions. However, being based on HU values, they are potentially confounded by patient and image acquisition characteristics. In our primary study group of individuals with zero calcium score, we show in 1652 patients that PCAT attenuation and gradient can have a wide range of values. Nevertheless, in our univariable results CAD markers showed associations with PCAT markers. However, many of the patient and imaging parameters such as: male sex, heart rate, tube voltage and current and pixel spacing were also significantly associated with PCAT attenuation. While our univariable results confirm previous associations of PCAT attenuation and gradient with NCP characteristics, these associations did not persist after correcting for patient and image acquisitions parameters, of which many were independently associated with PCAT markers in multivariable analysis. Similar results were observed in individuals imaged on a different CT scanner and also in individuals with moderate to severe CAD.
In a study with in vivo and ex vivo models, Antonopoulos et al. demonstrated phenotypic changes in PCAT, hence in PCAT attenuation values, as a response to vascular inflammation and inflamed vulnerable plaques at the time of a major adverse cardiac event. Strengthening the association between PCAT and inflammation, Goeller et al. found a positive correlation between pro-inflammatory serum markers (MCP-1 and IL-7) and PCAT attenuation and a negative correlation between anti-inflammatory cytokines (IL-4, IL-10, IL-13) and PCAT attenuation,
Pericoronary adipose tissue CT attenuation and its association with serum levels of atherosclerosis-relevant inflammatory mediators, coronary calcification and major adverse cardiac events.
although these correlations were weak (Pearson correlation coefficient range: −0.12 to 0.23). Similarly, PCAT attenuation showed an association with 18F-sodium fluoride uptake on PET-CT imaging in stable patients with high-risk plaque features on CCTA, providing another link with established markers of inflammation.
Regarding the association of PCAT attenuation with coronary plaques, Goeller et al. found higher PCAT attenuation values around the culprit lesions as compared with non-culprit lesions in patients with acute coronary syndrome.
However, our results indicate that there are certain factors that must be considered before drawing conclusions using PCAT markers.
First, in our study, the patients with a low cardiovascular risk referred for a CCTA demonstrated a large variability of PCAT attenuation and gradient values. A wealth of evidence demonstrates that patients with zero calcium score are at low risk for adverse events.
Therefore, one would expect that the PCAT values would be universally low and exhibit low variability, especially in individuals with zero calcium scores and no plaque. However, PCAT attenuation and PCAT gradient had a wide range of values. This may be mainly due to our second finding that PCAT markers are significantly associated with imaging and patient characteristics. Most prior studies use the original definition of PCAT attenuation (originally called fat attenuation index) based on the study by Antonopoulos et al.,
Pericoronary adipose tissue and quantitative global non-calcified plaque characteristics from CT angiography do not differ in matched South Asian, East Asian and European-origin Caucasian patients with stable chest pain.
To overcome the aforementioned limitations, instead of adjusting for possible confounders in the statistical models of specific studies, Oikonomou et al. used a modified PCAT attenuation parameter to predict later outcomes in the CRISP CT-study.
Non-invasive detection of coronary inflammation using computed tomography and prediction of residual cardiovascular risk (the CRISP CT study): a post-hoc analysis of prospective outcome data.
This corrected differences in HU values between 100 and 120 kVp using a correction factor derived from 17 individuals scanned on a dual-source system by simply dividing the average attenuation of free-hand regions of interest drawn in the pericardial fat on 100 and 120 kVp images (−96.1/-86.2 = 1.11485).
However, several of the factors that our study found to independently influence PCAT attenuation values were not corrected for, such as heart rate. Furthermore, as such characteristics as heart rate significantly influence PCAT attenuation values beyond image quality parameters (eg. contrast to noise ratio, signal to noise ratio), it seems challenging to establish robust estimates of PCAT attenuation and to determine normal PCAT attenuation and gradient values. Chatterjee et al. proposed correcting PCAT attenuation with adjacent lumen attenuation values to account for variations in image and patient characteristics.
Following correction, they did not find any association between PCAT and major adverse cardiovascular events in the CORE320 trial participants. Also, simply using a different CT scanner the average PCAT attenuation was 15 HU higher, while both the main study group and the validation group consisted of individuals with zero calcium score. These results show the need for normalization of images prior to PCAT measurements. Further multi-center efforts are needed using a wide range of patients and imaging machinery to create a diverse database on which the optimal normalization protocols are developed. Conventional statistical and artificial intelligence techniques may help to correct for the possible confounding of these parameters to achieve standardized comparable PCAT measurements that are independent of patient and imaging characteristics. Until then, researchers should use as similar as possible scan characteristics and correct their statistical models for all possible confounders. In case of longitudinal analyses, same scanners and acquisition protocols should be used to minimize bias.
Our study has some limitations. First, our main study group consisted of patients with low cardiovascular risk with zero coronary calcium scores from a single center study scanned using the same CT and similar imaging protocols, which might have resulted in a selection bias. However, we found very similar results in our validation cohorts strengthening our results. Second, only two readers segmented PCAT volumes with a single software, which artificially reduced variation, therefore the results are only generalizable with caution. Third, only one CCTA reconstruction algorithm was used in each group. However, reconstruction algorithms are known to affect HU values thus representing an additional factor which may need to be considered when correcting for confounders when analyzing PCAT. Fourth, our outcome analyses may be underpowered. Nevertheless, in the individuals who experience myocardial infarction, we found a wide range of values. Moreover, other studies have found other factors also to be associated with PCAT markers, therefore our list of potential confounders may not be complete.
Based on our results, PCAT attenuation and gradient show a wide range of values in individuals with low cardiovascular risk (calcium score = 0). These markers are significantly influenced by different image acquisition and patient characteristics. After correcting for these metrics, associations between PCAT and CAD markers do not persist, highlighting the importance of correcting for all possible confounders before evaluating the additive value of PCAT markers. Future studies extensively correcting for possible confounders are needed to validate the additive value of PCAT measurements.
Funding
Project no. RRF-2.3.1-21-2022-00003 has been implemented with the support provided by the European Union. Project no. NVKP_16-1–2016-0017 (’National Heart Program’) has been implemented with the support provided from the National Research, Development and Innovation Fund of Hungary, financed under the NVKP_16 funding scheme. Melinda Boussoussou MD was supported by the ÚNKP-22-3-II-SE (ÚNKP-22-3-II-SE-51), New National Excellence Program of the Ministry for Innovation and Technology from the source of the National research, Development and Innovation fund and by the EFOP-3.6.3-VEKOP-16-2017-00009 project fund.
Conflicts of interest
None.
Appendix A. Supplementary data
The following is the supplementary data to this article:
Validated imaging biomarkers as decision-making tools in clinical trials and routine practice: current status and recommendations from the EIBALL∗ subcommittee of the European Society of Radiology (ESR).
Relationship between changes in pericoronary adipose tissue attenuation and coronary plaque burden quantified from coronary computed tomography angiography.
Pericoronary adipose tissue and quantitative global non-calcified plaque characteristics from CT angiography do not differ in matched South Asian, East Asian and European-origin Caucasian patients with stable chest pain.
Society of cardiovascular computed tomography/north American society of cardiovascular imaging - expert consensus document on coronary CT imaging of atherosclerotic plaque.
Non-invasive detection of coronary inflammation using computed tomography and prediction of residual cardiovascular risk (the CRISP CT study): a post-hoc analysis of prospective outcome data.
Quality assurance of national internet-based patient register data: experiences during the operation of the Hungarian Myocardial Infarction Registry, 2010-2020.
Comparison of management and outcomes of ST-segment elevation myocardial infarction patients in Estonia, Hungary, Norway, and Sweden according to national ongoing registries.
Eur Heart J Qual Care Clin Outcomes.2022; 8: 307-314
Pericoronary adipose tissue CT attenuation and its association with serum levels of atherosclerosis-relevant inflammatory mediators, coronary calcification and major adverse cardiac events.
00281-7&id=gr1.jpg){kind=link}
00281-7&id=gr2.jpg){kind=link}