Focal pericoronary adipose tissue attenuation is related to plaque presence, plaque type, and stenosis severity in coronary CTA

Objectives To investigate the association of pericoronary adipose tissue mean attenuation (PCATMA) with coronary artery disease (CAD) characteristics on coronary computed tomography angiography (CCTA). Methods We retrospectively investigated 165 symptomatic patients who underwent third-generation dual-source CCTA at 70kVp: 93 with and 72 without CAD (204 arteries with plaque, 291 without plaque). CCTA was evaluated for presence and characteristics of CAD per artery. PCATMA was measured proximally and across the most severe stenosis. Patient-level, proximal PCATMA was defined as the mean of the proximal PCATMA of the three main coronary arteries. Analyses were performed on patient and vessel level. Results Mean proximal PCATMA was −96.2 ± 7.1 HU and −95.6 ± 7.8HU for patients with and without CAD (p = 0.644). In arteries with plaque, proximal and lesion-specific PCATMA was similar (−96.1 ± 9.6 HU, −95.9 ± 11.2 HU, p = 0.608). Lesion-specific PCATMA of arteries with plaque (−94.7 HU) differed from proximal PCATMA of arteries without plaque (−97.2 HU, p = 0.015). Minimal stenosis showed higher lesion-specific PCATMA (−94.0 HU) than severe stenosis (−98.5 HU, p = 0.030). Lesion-specific PCATMA of non-calcified, mixed, and calcified plaque was −96.5 HU, −94.6 HU, and −89.9 HU (p = 0.004). Vessel-based total plaque, lipid-rich necrotic core, and calcified plaque burden showed a very weak to moderate correlation with proximal PCATMA. Conclusions Lesion-specific PCATMA was higher in arteries with plaque than proximal PCATMA in arteries without plaque. Lesion-specific PCATMA was higher in non-calcified and mixed plaques compared to calcified plaques, and in minimal stenosis compared to severe; proximal PCATMA did not show these relationships. This suggests that lesion-specific PCATMA is related to plaque development and vulnerability. Key Points • In symptomatic patients undergoing CCTA at 70 kVp, PCATMA was higher in coronary arteries with plaque than those without plaque. • PCATMA was higher for non-calcified and mixed plaques compared to calcified plaques, and for minimal stenosis compared to severe stenosis. • In contrast to PCATMA measurement of the proximal vessels, lesion-specific PCATMA showed clear relationships with plaque presence and stenosis degree. Supplementary Information The online version contains supplementary material available at 10.1007/s00330-021-07882-1.


Introduction
Coronary inflammation plays an important role in atherosclerosis development [1][2][3]. Detection and quantification of coronary inflammation could assist in early risk stratification of coronary artery disease (CAD) patients, possibly even before the development of coronary plaque [4]. Recently, a non-invasive biomarker for coronary inflammation was proposed: computed tomography angiography (CCTA) derived pericoronary adipose tissue mean attenuation (PCAT MA ) [5]. PCAT MA has shown value as a predictor for cardiac mortality [6]. Few studies, predominantly using the proximal right coronary artery (RCA) as a representative location for patient-level analysis, have shown a relationship of PCAT MA with CAD and atherosclerosis progression [5,[7][8][9].
CCTA-based plaque composition and stenosis severity give information about plaque vulnerability and hemodynamic significance, and can be used for prognostication [10][11][12][13]. A previous study showed a PCAT MA difference of 3-4HU in the proximal RCA between CAD and non-CAD patients [5]. However, they found no significant difference of RCA-based PCAT M A between noncalcified plaques (NCP) and mixed or calcified plaques (CP) in patients with high plaque burden. Another study demonstrated that increased NCP and total plaque burden were associated with higher PCAT MA [8].
Most studies measured PCAT MA at one proximal coronary location [5,6,8,14]. Compared to proximal PCAT MA , there may be a stronger relation of lesionspecific PCAT MA with plaque considering a hypothesized local effect of coronary inflammation. Three PCAT MA s t u d i e s ( 3 5 -1 9 9 p a t i e n t s ) u s e d a l e s i o n -b a s e d measurement method considering all three main coronary arteries [9,15,16]. One study showed that lesion-specific PCAT MA was higher around culprit lesions in acute coronary syndrome (ACS) patients compared to non-culprit lesions in ACS and CAD patients [15]. Another study revealed lesion-specific PCAT MA was significantly increased in patients with abnormal FFR [9]. However, lesion-specific PCAT MA failed to show a significant difference between patients with and without elevated highsensitivity C-reactive protein [16]. Currently, there is a lack of knowledge on the relationship between PCAT MA and plaque presence, plaque type, and stenosis severity. In addition, the majority of studies only investigated a single, proximally measured PCAT MA value (mostly RCA) to represent overall pericoronary attenuation but did not investigate a potentially more relevant, focal PCAT MA value across coronary plaque.
The aim of this study was to evaluate the relationship of proximal and lesion-specific PCAT MA with coronary plaque presence, type, and severity.

Study population
This single-center, cross-sectional study was performed at the University Medical Center Groningen. The study was compliant with the Declaration of Helsinki and approved by the institutional ethical review board, who waived the need for informed consent.
In total, 2621 patients underwent cardiac CTA for routine indications between January 2015 and November 2017. Of these patients, a random sample of 1280 patients was further characterized by gathering hospital record information on CT indication, demographics, and clinical risk factors, to be used in various CT analyses. In a previous analysis (Ma et al) [17], we studied a cohort of patients with a zero calcium score and no coronary plaque on CCTA ("normal patients"); from this population, we selected patients with CCTA at 70 kilovoltage peak (kVp) as a reference category for the current study (n = 72). From the 697 patients (out of 1280) who underwent CCTA because of angina, we randomly selected patients with CAD, defined as patients with plaque on their CCTA images, for the current analysis based on the following inclusion criteria: 1, age > 18 years; 2, CCTA performed at 70 kVp; 3, no coronary stents or coronary artery bypass grafts. Tube voltage was restricted to 70 kVp in view of known influence of kVp on PCAT MA [17]. In total, 171 patients (72 + 99) were included. Six CAD patients were excluded for the following reasons: anomalous origin of coronary artery (n = 2), insufficient image quality (n = 1), incomplete coronary image coverage (n = 3) (Fig. 1). A radiologist with 10-year experience in cardiac radiology performed the CCTA evaluation (R.M.). In case of doubt, a radiologist with 14 years of experience was consulted and consensus was obtained (R.V.).

CCTA scan protocol
CCTA imaging was performed according to the routine clinical protocol using third-generation dual-source CT (SOMATOM Force, Siemens Healthineers). First, a nonenhanced ECG-gated CT at a high pitch (tube voltage 120 kVp, reference tube current 64 mAs, reconstructed slice thickness 3.0mm) was performed for coronary calcium score (CACS) analysis. Subsequently, CCTA was performed using CarekV (kVp optimization assistance), depending on patient size; patients scanned at 70 kVp were included. ECG-gated high-pitch spiral scanning was performed in low, regular heart rate, otherwise ECG-triggered sequential scanning. Patients received sublingual nitroglycerin, unless contraindicated. If the heart rate was > 70-73 beats/min, the patient received intravenous beta-blocker, unless contraindicated. Contrast timing was determined using a test bolus. Iomeprol (Iomeron 350) was injected with dose-and flow-rate depending on patient characteristics and scan mode. A dual-injection technique was used followed by a saline flush. CCTA images were reconstructed at 0.6 mm thickness.

Patient characteristics
Baseline patient characteristics were collected from clinical records. Age, sex, and CAD risk factors were collected. The classification criteria of risk factors were as follows: (a) hypertension-systolic blood pressure > 140 mmHg or diastolic blood pressure > 90 mmHg according to guidelines [18] and/ or anti-hypertension medication use; (b) hyperlipidemiapatients with a low-density lipoprotein > 4.5 mmol/L or total cholesterol > 6.5 mmol/L based on guidelines [19] were considered as hyperlipidemic; lipid-lowering medications used at the time of CT scanning was considered as a separate factor indicating treated hyperlipidemia; (c) diabetes mellitus-antidiabetic medication use; (d) smoking status was classified as non-smoker, current smoker, or former smoker. Depending on the risk factors, information was missing in 26 to 51 patients. If there was no mention of a risk factor, the risk factor was considered absent. Body mass index (BMI) information was collected as well.

Plaque analysis
Visual, qualitative analysis For visual plaque evaluation only, the main coronary arteries, left anterior descending (LAD), left circumflex (LCx), and right coronary artery (RCA) were taken into account to optimize patient comparability. Plaque composition and diameter stenosis (DS) were assessed for the most severe plaque per coronary artery. Plaque components were classified into noncalcified plaque (NCP), mixed plaque, and calcified plaque (CP). Using visual analysis, CP was defined as plaque when it had > 75% volume with density higher than the luminal contrast; NCP was defined as plaque when it had > 75% volume with a density lower than the lumen contrast and higher than soft tissues around. Mixed plaque was defined as plaque comprising 25 to 75% volume with density higher than the luminal contrast [20,21]. DS was classified into 4 stenosis categories: minimal, DS 1-24%; mild, DS 25-49%; moderate, DS 50-69%; and severe, DS 70-100% [22].
Quantification of the plaque composition was semiautomatically performed by the software (vascuCAP, Research Edition, Elucid Bioimaging) [23]. Automatic segmentation of the entire coronary lumen and wall was performed, allowing manual corrections if needed. Subsequently, the matrix burden, CP burden, and lipidrich necrotic core (LRNC) burden were automatically calculated by the software on a per-vessel level [24]. The classification of the different plaque components, which was validated with plaque histology, was based on an adaptive threshold. The LRNC lower limit was defined as −300HU; LRNC-IPH boundary was defined as 25HU. The lower limit and upper limit of the CP were 250 and 3000HU. Matrix burden was calculated by dividing the total wall volume by the matrix volume, where the matrix is defined as normal organization tissues in the vessel wall [23]. Plaque burden was defined as 1-matrix burden [24].

PCAT MA measurements
PCAT MA was measured proximally in the RCA, LAD, and LCx, using dedicated software (Aquarius iNtuition, TeraRecon, Version 4.4.13). The starting point of the proximal PCAT MA measurement was 10mm after the left main bifurcation for LAD, at the bifurcation point for LCx, and 10mm after the ostium for RCA [17]. In vessels with plaque, a lesion-specific PCAT MA measurement was performed centered around the most severely stenotic plaque. The proximal and distal ends of the measurement were 5mm away from the lesion center. The measurement length and width for all measurements were 10mm and 1mm. A 1mm gap was left between the outer vessel wall, taking into account eccentric plaques, and the measured cylindrical volume to avoid artifacts. PCAT MA was defined as the mean CT value in the measured area within the range of −190 to −30 HU (Fig. 2).

Data analysis
First, PCAT MA was studied on per-patient level (Fig. 1). Patients with any coronary plaque were considered as CAD patients; patients without plaque were considered non-CAD patients. For the per-patient PCAT MA , the mean of the proximal PCAT MA values based on the three main coronary arteries was calculated to represent an overall, patient-based PCAT MA value. Patient-based CACS and DS were analyzed in conjunction with the per-patient PCAT MA . Patient-level categorization of DS degree was based on the most severe DS in all three coronary arteries. To allow comparison with prior studies that used only the proximal measurement of PCAT MA of the RCA, we additionally performed analyses for RCA-based PCAT MA . Additionally, a comparison of patients with and without at least 50% stenosis was performed. The total plaque burden of the main coronary arteries was considered as the patient-based plaque burden.
Second, vessel-based analysis was performed (Fig. 2). We discriminated arteries with any plaque, and arteries without plaque. CAD patients could contribute arteries without plaque. For arteries with multiple plaques, the lesion with the highest DS was used. The proximal PCAT MA was used in arteries without plaque to compare with lesion-specific PCAT MA in arteries with plaque. Lesion-specific PCAT MA was analyzed based on plaque type and DS severity.

Statistical methods
Normality testing for continuous variables was performed with the Shapiro-Wilk test. Continuous variables are represented as mean± standard deviation (SD) or median (interquartile range [IQR]), according to distribution. The model estimated values are given in mean with 95% confidence interval (CI). Categorical variables were recorded as numbers (n) and percentages (%). Paired t-tests were used to evaluate differences between proximal and lesion-specific PCAT MA . Independent t-tests were used to compare PCAT MA measurements between patients. One-way analysis of variance (ANOVA) testing was used to compare PCAT MA between categories of plaque type and DS severity. Spearman correlation testing was used to assess the correlation of PCAT MA with plaque burden and plaque component burden. A generalized linear model was used to evaluate the influencing factors for patient-based PCAT MA . Using mixed models with random intercepts, the model estimated marginal means and 95% CI of the corrected PCAT MA were calculated. The basic model included age, sex, and vessel, while the advanced models included CAD risk factors. The models did not include BMI because of 43 missing values. PCAT MA was taken as a dependent variable in order to study the relationship between PCAT MA and plaque features. A p value < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 25 (IBM).

Patient demographics
In total, 93 patients with CAD and 72 patients without CAD were included. Figure 2 shows an overview of the inclusion process. Patient characteristics are given in Table 1. Patients with CAD were significantly older (60.9 ± 8.7 vs. 51.2 ± 12.6 years, p < 0.001) and had significantly more hypertension

Patient-based PCAT MA analysis
An overview of PCAT MA values for CAD and non-CAD patients, CACS, and DS category is provided in Table 2 Table S1 and Table S2.

Vessel-based proximal PCAT MA analysis
There were 204 arteries with plaque and 291 without plaque (216 from patients without CAD and 75 from patients with CAD). The mean proximal PCAT MA of vessels without plaque was −95.6 ± 9.6 HU and −96.3 ± 8.3 HU for patients with and without CAD, respectively (p = 0.567). The different plaque components or degrees of stenosis groups did not show a difference in proximal PCAT MA .

Vessel-based lesion-specific PCAT MA analysis
Lesion-specific PCAT MA showed a significant difference (p = 0.002) for the coronary lesions with different plaque   Figure 3 gives an overview of proximal and lesionspecific PCAT MA measurements for different plaque components and degrees of stenosis.
After correction for CAD risk factors, LRNC burden and plaque burden had significant effects (estimate: −0.8 vs. −0.6) on proximal PCAT MA , while the CP burden had no significant effects on proximal PCAT MA (Table 3).

Discussion
This study investigated the relationship between PCAT MA and plaque presence, plaque type, and stenosis severity in the main coronary arteries in symptomatic patients undergoing CCTA at 70 kVp. PCAT MA was higher in vessels with plaque than in vessels without plaque, taking into account patients' risk factors. Lesion-specific PCAT MA was higher for non-calcified and mixed plaques compared to calcified plaques, and for minimal stenosis compared to severe stenosis. In contrast to proximal PCAT MA , lesion-specific PCAT MA showed clear relationships with plaque presence and stenosis degree.
The proof-of-concept paper by Antonopoulos et al [5] demonstrated that RCA-based PCAT MA differed by approximately 3HU between CAD and non-CAD patients, where CAD was defined as the presence of a stenosis of more than 50%. As PCAT MA values vary between coronary arteries and plaque distribution among the coronary arteries, with the LAD most often affected, taking only the RCA as a PCAT MA reference location may not accurately represent the patient's PCAT MA status. Oikonomou et al [6] reported that increased PCAT MA in the RCA and LAD rather than LCx was related to increased cardiac mortality risk. Gaibazzi et al [25] reported significant differences between the LAD/RCA and the LCX in vessels with a stenosis < 50%, with a HU difference of approximately 1.5 HU on 120kVp scans. In our previous study, comparing PCAT MA at different kVp levels in patients without plaque, there were significant differences between the PCAT MA of LAD, LCX, and RCA with a HU difference around 2~4 HU [17].
Besides the coronary artery, the measurement location may also have a significant effect on PCAT MA . Goeller et al [8] showed that, although there was a correlation between PCAT MA and epicardial adipose tissue (EAT), there was no correlation between changes in EAT and plaque burden progression. Dai et al [16] found no relationship between lesion-specific PCAT MA and high-sensitive C-reactive protein, suggesting that PCAT MA may be associated with local coronary inflammation rather than global inflammation. Previously mentioned studies used lesion-specific PCAT MA only; few investigated the relationship with coronary plaque. Kwiecinski et al [26] found that increased lesion-specific PCAT MA in patients with high-risk plaque was related to focal 18F-NaF PET uptake. Lin et al [27] reported on the relationship of PCAT radiomic features and PCAT MA in the proximal RCA and around (non-) culprit lesions at presentation and 6 months post-MI, in comparison to stable CAD and non-CAD cases. They report that the most significant radiomic parameters distinguishing patients with and without MI were based on texture and geometry, yielding information not  included in PCAT attenuation. They found that radiomic features were not different between culprit and non-culprit lesions, where the PCAT MA showed a significant difference. The authors mention that PCAT MA may have utility as a lesion-specific imaging biomarker, while radiomics features may have more value as a patient-specific biomarker of systemic inflammation. Our study, using both proximal and lesion-based PCAT MA , confirms that lesion-specific PCAT MA is a better representation of focal inflammation and plaque development. Only lesion-specific PCAT MA measurements showed a difference between vessels with and without plaque. Using an adjusted model, the PCAT MA of vessels with plaque was around 2HU higher than those without plaque. This result is similar to the HU difference in the study by Antonopoulos et al [5]. Lesion-specific PCAT MA differed by DS categories, taking into account age, sex, and coronary artery. Our results suggest that there may be more inflammation in mild and moderate DS than in severe DS. This fits with the hypothesis that as the plaque becomes more stabilized and more calcified in severe DS, inflammation could be relatively decreased [28]. Inflammatory cytokines play a critical role in the development and progression of coronary atherosclerosis [29,30]. The theory behind PCAT MA is that vessel wall atherosclerosis inhibits adipocyte maturation and lipid accumulation in the pericoronary fat tissue, increasing the attenuation. Additionally, corresponding increases in edema and amount of inflammatory cells possibly result in an additional increase in PCAT MA in patients at risk of or with CAD [31,32]. Results from previous studies suggest that the relationship between coronary inflammation and PCAT MA may be more evident in NCP than CP, since CPs are relatively stable and have only a minimal inflammatory component [31,32]. Goeller et al [8] investigated the relationship between PCAT MA and progression of plaque burden on CCTA. Measuring patient-based plaque burden/composition and RCA-based PCAT MA , they found that PCAT MA is related to progression of total plaque burden and NCP burden. PCAT MA > −75 HU of the proximal RCA was independently associated with increased NCP burden at 120kVp CCTA [8]. However, similar to our results, they found that there was no relationship with CP burden. In our study, the model-adjusted, lesion-specific PCAT MA values for NCP were 5-7 HU higher compared to CP and mixed plaques at 70kVp CCTA, measured in the three main coronary arteries. Our study showed only a weak correlation between vessel-based plaque burden and per-vessel PCAT MA , and no significant correlation between patient-based total plaque burden and patient-based PCAT MA . The per-vessel LRNC burden had a moderate correlation with PCAT MA whereas the CP burden showed a very poor correlation. Recent research revealed that LRNC burden is capable of predicting myocardial infarction better than CAC scoring, cardiovascular risk scores, and coronary artery stenosis [33].
There are reports that show that lipid-lowering medication could decrease the EAT attenuation independent of decreasing lipid values [34]. Our study also shows a significant effect of lipid-lowering medication on PCAT MA values, supporting the idea that statins have an effect on cardiac fat attenuation and, potentially, adipose tissue activity [35]. Additionally, we found that vessel, sex, and age had significant effects on PCAT MA . The relationship between age, sex, and CAD has been reported frequently [36][37][38]. Men showed generally higher PCAT MA values than women (−94.0 vs −97.3 HU). Gender-specific hormones may be the reason for the different effects on coronary inflammation.

Limitations
This is a single-center, cross-sectional study of patients with clinically indicated CCTA. No follow-up information is available; hence, CCTA results cannot be related to cardiovascular prognosis. Although our study demonstrates a relationship between plaque presence, type, and stenosis degree with PCAT MA , it was not designed to show direct causality between inflammatory status, plaque characterization, and PCAT MA . Plaque burden quantification was performed by automatic software, allowing manual corrections. In general, automatic analysis might be sensitive to errors due to image artifacts or decreased image quality and errors in segmentation. To avoid these errors in this study, scans were selected on image quality (2 scans were excluded), and at each segmentation step, the segmentation was visually assessed and manually corrected when necessary by an experienced radiologist to avoid errors. Window levels could be adjusted manually to reduce, for example, blooming effects from calcifications in order to optimize the segmentation and automated analysis.

Conclusion
PCAT MA was higher in coronary arteries with plaque, compared to vessels without plaque. Lesion-specific PCAT MA was higher in NCP and mixed plaque compared to CP, and in minimal stenosis compared to severe stenosis. Proximally measured PCAT MA only showed differences by plaque composition, and only when corrected for clinical parameters. This suggests that in particular lesion-specific PCAT MA is related to plaque development and vulnerability.