Computed Tomography Angiography Characteristics of Thin‐Cap Fibroatheroma in Patients With Diabetes

Abstract

Background

It was recently reported that thin‐cap fibroatheroma (TCFA) detected by optical coherence tomography was an independent predictor of future cardiac events in patients with diabetes. However, the clinical usefulness of this finding is limited by the invasive nature of optical coherence tomography. Computed tomography angiography (CTA) characteristics of TCFA have not been systematically studied. The aim of this study was to investigate CTA characteristics of TCFA in patients with diabetes.

Methods and Results

Patients with diabetes who underwent preintervention CTA and optical coherence tomography were included. Qualitative and quantitative analyses were performed for plaques on CTA. TCFA was assessed by optical coherence tomography. Among 366 plaques in 145 patients with diabetes, 111 plaques had TCFA. The prevalence of positive remodeling (74.8% versus 50.6%, P<0.001), low attenuation plaque (63.1% versus 33.7%, P<0.001), napkin‐ring sign (32.4% versus 11.0%, P<0.001), and spotty calcification (55.0% versus 34.9%, P<0.001) was significantly higher in TCFA than in non‐TCFA. Low‐density noncalcified plaque volume (25.4 versus 15.7 mm³, P<0.001) and remodeling index (1.30 versus 1.20, P=0.002) were higher in TCFA than in non‐TCFA. The presence of napkin‐ring sign, spotty calcification, high low‐density noncalcified plaque volume, and high remodeling index were independent predictors of TCFA. When all 4 predictors were present, the probability of TCFA increased to 82.4%.

Conclusions

The combined qualitative and quantitative plaque analysis of CTA may be helpful in identifying TCFA in patients with diabetes.

Registration Information

URL: https://www.clinicaltrials.gov; Unique identifier: NCT04523194.

Nonstandard Abbreviations and Acronyms

CTA

computed tomography angiography

HRP

high‐risk plaque

LAP

low‐attenuation plaque

LDNCP

low‐density noncalcified plaque

NCP

noncalcified plaque

NRS

napkin‐ring sign

PR

positive remodeling

RI

remodeling index

SC

spotty calcification

TCFA

thin‐cap fibroatheroma

Clinical Perspective.

What Is New?

• Detailed computed tomography angiography characteristics of thin‐cap fibroatheroma in patients with diabetes have not been systematically studied.

• The presence of napkin‐ring sign, spotty calcification, high low‐density noncalcified plaque volume, and high remodeling index were independent predictors of thin‐cap fibroatheroma; when all 4 parameters were present, the probability of thin‐cap fibroatheroma increased to 82.4%.

What Are the Clinical Implications?

• Although diagnosis of thin‐cap fibroatheroma currently requires optical coherence tomography, our results may suggest that the integration of qualitative and quantitative computed tomography angiography analysis could noninvasively detect thin‐cap fibroatheroma in patients with diabetes.

Thin‐cap fibroatheroma (TCFA) is characterized as a plaque with large lipid pool with an overlying thin fibrous cap. ¹ It was recently reported that TCFA detected by optical coherence tomography (OCT) was an independent predictor of future cardiac events in patients with diabetes. ² However, the clinical usefulness of this finding is limited by the invasive nature of OCT. Computed tomography angiography (CTA) is a sensitive noninvasive technique for the detection of coronary atherosclerotic plaque and allows accurate measurement of plaque volume as well as identification of plaque features of vulnerability. ³ , ⁴ A previous study showed that higher low‐density noncalcified plaque (LDNCP) volume on CTA was associated with intravascular ultrasound‐detected TCFA. ⁵ CTA features of high‐risk plaque (HRP) include positive remodeling (PR), low‐attenuation plaque (LAP), napkin‐ring sign (NRS), and spotty calcification (SC). Several studies have reported that HRP was associated with TCFA. ⁶ , ⁷ , ⁸ , ⁹ However, detailed CTA characteristics of TCFA in patients with diabetes have not been systematically studied. The aim of the current study was to identify CTA characteristics of TCFA in patients with diabetes.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Population

This is an observational, single‐center cohort study. Patients who underwent CTA for suspected coronary artery disease were enrolled. Of those, patients who underwent OCT before coronary intervention due to acute coronary syndrome (unstable angina pectoris or non–ST‐segment–elevation myocardial infarction) or chronic coronary syndrome between January 2011 and July 2020 were included from the established database, Massachusetts General Hospital and Tsuchiura Kyodo General Hospital Coronary Imaging Collaboration (NCT04523194). Patients with ST‐segment–elevation myocardial infarction were not included in this study. The definitions of unstable angina pectoris, non–ST‐segment–elevation myocardial infarction, and chronic coronary syndrome are described in Data S1. The study period was chosen so that CTA image acquisition was performed using a single computed tomography (CT) system. All CTA and OCT images, as well as clinical data, were acquired at Tsuchiura Kyodo General Hospital (Ibaraki, Japan) and submitted to Massachusetts General Hospital (Boston, MA) for analysis. Exclusion criteria included a span of >180 days between the time of CTA and OCT imaging, ¹⁰ poor image quality, calcified culprit plaque that caused acute coronary syndrome, culprit lesions of the left main, culprit lesions at branches (eg, diagonal branch), and indeterminate culprit lesions. Diabetes was defined by baseline medical history, glycohemoglobin ≥6.5%, or reported use of glucose‐lowering medication. In the final analysis, 145 patients with diabetes were included (Figure S1). All coronary segments >2 mm in diameter in the target vessel for percutaneous coronary intervention were evaluated for the presence of plaque (calcified, noncalcified, or partially calcified) on CTA. ¹¹ Of those, 366 lesions imaged by OCT were included in the final analysis. The study protocol was approved by the respective Institutional Review Boards at Tsuchiura Kyodo General Hospital and Massachusetts General Hospital. All participants provided written informed consent for the Tsuchiura Kyodo General Hospital's institutional database for potential future investigations. Thus, a waiver of consent for this project was granted by the Tsuchiura Kyodo General Hospital ethics committee.

CTA Acquisition

CT image acquisition was performed using a 320‐slice CT scanner (Aquilion ONE; Canon Medical Systems) in accordance with the Society of Cardiovascular Computed Tomography guidelines. ¹² Oral or intravenous β‐blockers were administered if a patient's resting heart rate was >65 bpm. Sublingual nitroglycerin (0.3 or 0.6 mg) was administered immediately before CTA scanning. CTA images were acquired with the following scan protocol: tube voltage of 120 kVp, tube current of 50 to 750 mA, gantry rotation speed of 350 ms per rotation, field matrix of 512×512, and scan slice thickness of 0.5 mm. Acquisition of CT data and the electrocardiography trace were automatically started as soon as the signal density level in the ascending aorta reached a predefined threshold of 150 Hounsfield units. Images were acquired after a bolus injection of 30 to 60 mL of contrast media (iopamidol, 370 mg iodine/mL; Bayer Yakuhin, Osaka, Japan) at a rate of 3 to 6 mL/s, using prospective ECG‐triggering or retrospective ECG‐gating with automatic tube current modulation. All scans were performed during a single breath‐hold. Images were reconstructed at a window centered at 75% of the R–R interval to coincide with left ventricular diastasis.

Qualitative Plaque Analysis in CTA

HRP features were defined as PR, LAP, NRS, and SC. The definitions of HRP are described in Data S1.

Quantitative Plaque Analysis in CTA

Quantitative plaque analysis was performed using semiautomated software (Autoplaque version 2.5; Cedars‐Sinai Medical Center) ¹³ (Data S1). Low intra‐ and interobserver variability for plaque assessment using Autoplaque has been demonstrated in earlier work. ¹⁴ Plaque volume (cubic millimeters) was calculated on a per‐lesion level for the following plaque components: total plaque, calcified plaque, noncalcified plaque (NCP), and LDNCP (defined by an attenuation of <30 Hounsfield units). The software automatically quantified vessel remodeling index (RI) (Data S1). Quantitative CTA plaque analysis was performed by 2 independent investigators who were blinded to patients' data, using an offline review workstation at Massachusetts General Hospital.

OCT Image Acquisition and Analysis

OCT examination was performed using either a frequency‐domain (C7/C8; OCT Intravascular Imaging System, St. Jude Medical) or a time‐domain (M2/M3; Cardiology Imaging Systems, Light Lab Imaging) OCT system. All OCT images were submitted to the core laboratory at Massachusetts General Hospital and analyzed by 2 independent investigators who were blinded to patients' data, using an offline review workstation (St. Jude Medical). Lipid was defined as a low‐signal region with diffuse border. ¹⁵ Lipid‐rich plaque was defined as a plaque with a maximal lipid arc >90°. ¹⁵ TCFA was defined as a plaque with a maximal lipid arc >90° and thinnest fibrous cap thickness of ≤65 μm. ¹⁵

Statistical Analysis

Continuous variables were expressed as mean±SD. The calcified plaque volume that was skewed with high extreme values was expressed as median (interquartile range). The median time interval between CTA and OCT imaging was also expressed as median (interquartile range). Per‐plaque data were analyzed using the generalized estimating equations with an identity link for continuous and with a logit link for the binary variables to account for potential clustering of multiple plaques in a single patient. For calcified plaque volume, logarithmic transformation was made before fitting the generalized estimating equations model. The multivariable logistic regression model with a generalized estimating equation was used to identify independent predictors of TCFA. For CTA characteristics, NRS and SC, which can only be evaluated by qualitative analysis, and LDNCP volume and RI, which allow quantitative evaluation of HRP, were selected. Receiver operating curves were analyzed to assess the best cutoff values of the LDNCP volume, RI, and total plaque volume for predicting TCFA. The optimal cutoff value was calculated using the Youden index. Variables with a P value <0.10 in the univariable model were entered into the multivariable model. A P value <0.05 was considered statistically significant. All analyses were performed with SPSS (version 28 for Windows; IBM).

Results

Baseline Characteristics and Prevalence of TCFA

Baseline characteristics are shown in Table 1. The mean age was 66.1 years, and 80.7% (117 of 145 patients) were men. The median time interval between CTA and OCT imaging was 14 (1–43) days. Among 366 plaques in 145 patients with diabetes, 111 plaques (30.3%) had TCFA.

Table 1.

Baseline Characteristics (n=145)

Characteristic	Value
Age, y	66.1±11.2
Men, n (%)	117 (80.7)
Clinical presentation
ACS
NSTEMI, n (%)	52 (35.9)
Unstable angina pectoris, n (%)	10 (6.9)
CCS, n (%)	83 (57.2)
Hypertension, n (%)	104 (71.7)
Dyslipidemia, n (%)	91 (62.8)
Current smoking, n (%)	40 (27.6)
Laboratory data
eGFR, mL/min per 1.73 m2	71.5±20.6
Low‐density lipoprotein cholesterol, mg/dL	105.7±35.2
High‐density lipoprotein cholesterol, mg/dL	46.8±12.3
Triglyceride, mg/dL	170.1±126.6
HbA1c, %	7.2±1.3
WBC, count/μL	6988±2483
hsCRP, mg/dL	0.61±1.95
Statin use at admission, n (%)	82 (56.6)

Values are mean±SD or n (%). ACS indicates acute coronary syndrome; CCS, chronic coronary syndrome; eGFR, estimated glomerular filtration rate; HbA1c, glycosylated hemoglobin; hsCRP, high‐sensitivity C‐reactive protein; NSTEMI, non–ST‐segment–elevation myocardial infarction; and WBC, white blood cell.

Qualitative Plaque Parameters

Among 366 plaques, PR was the most prevalent feature, followed by LAP, SC, and NRS (57.9%, 42.6%, 41.0%, and 17.5%, respectively). The prevalence of PR, LAP, NRS, and SC was significantly higher in TCFA than in non‐TCFA (Table 2, Figure 1A). When all 4 HRP features were present, the probability of TCFA increased to 66.7% (Figure S2).

Table 2.

Computed Tomography Angiography Characteristics

Characteristic	Plaques with TCFA (n=111)	Plaques without TCFA (n=255)	P value
Qualitative parameters
Positive remodeling	83 (74.8)	129 (50.6)	<0.001
Low attenuation plaque	70 (63.1)	86 (33.7)	<0.001
Spotty calcification	36 (32.4)	28 (11.0)	<0.001
Napkin‐ring sign	61 (55.0)	89 (34.9)	<0.001
Quantitative parameters
Plaque volume, mm3
Total plaque	186.0±113.3	136.9±86.0	<0.001
Noncalcified plaque	175.0±111.2	119.1±73.2	<0.001
Low‐density noncalcified plaque	25.4±22.0	15.7±16.1	<0.001
Calcified plaque	2.2 (0–14.2)	4.3 (0–17.4)	0.172
Remodeling index	1.30±0.28	1.20±0.19	0.001

Values shown are n (%), mean±SD, or median (interquartile range). TCFA indicates thin‐cap fibroatheroma.

Figure 1. CTA characteristics.

The prevalence of PR, LAP, NRS, and SC was significantly higher in TCFA than in non‐TCFA (A), Total plaque volume, noncalcified plaque volume, low‐density noncalcified plaque volume, and remodeling index were higher in TCFA than in non‐TCFA (B). CTA indicates computed tomography angiography; LAP, low‐attenuation plaque; NRS, napkin‐ring sign; PR, positive remodeling; SC, spotty calcification; and TCFA, thin‐cap fibroatheroma.

Quantitative Plaque Parameters

The results of quantitative analysis are shown in Table 2 and Figure 1B. Total plaque volume, NCP volume, LDNCP volume, and RI were higher in TCFA than in non‐TCFA.

Predictors of TCFA

Table 3 shows the univariable and multivariable analyses of TCFA, clinical characteristics, and CTA characteristics. The best cutoff values of the LDNCP volume and RI to predict the presence of TCFA were 16.7 mm³ and 1.42, respectively. In multivariable analyses, the presence of NRS, SC, LDNCP volume ≥16.7 mm³, and RI ≥1.42 were associated with TCFA. In the model, with the addition of the cutoff value for total plaque volume, these 4 parameters continued to be associated with TCFA (Table S1). When all 4 parameters were present, the probability of TCFA increased to 82.4% (Figure 2).

Table 3.

Univariable and Multivariable Analysis for Thin‐Cap Fibroatheroma

Variable	Univariable		Multivariable
	OR [95% CI]	P value	OR [95% CI]	P value
Age	0.994 [0.966–1.022]	0.654
Men	0.798 [0.417–1.525]	0.494
ACS	1.783 [1.039–3.059]	0.036	1.377 [0.771–2.460]	0.280
Hypertension	1.510 [0.867–2.632]	0.145
Dyslipidemia	1.108 [0.635–1.932]	0.718
Napkin‐ring sign	3.891 [2.242–6.755]	<0.001	1.993 [1.051–3.780]	0.035
Spotty calcification	2.276 [1.444–3.586]	<0.001	1.806 [1.107–2.948]	0.018
LDNCP volume ≥16.7 mm3	3.130 [1.994–4.912]	<0.001	2.197 [1.324–3.646]	0.002
Remodeling index ≥1.42	3.468 [1.963–6.129]	<0.001	2.036 [1.138–3.641]	0.017

ACS indicates acute coronary syndrome; LDNCP, low‐density noncalcified plaque; and OR, odds ratio.

Figure 2. Frequency of TCFA according to number of relevant CTA features.

The presence of napkin‐ring sign, spotty calcification, high low‐density noncalcified plaque volume (≥16.7 mm³), and high remodeling index (≥1.42) were independent predictors of TCFA. As the number of these CTA features increased, the prevalence of TCFA also increased significantly. When all 4 predictors were present, the probability of TCFA increased to 82.4%. CTA indicates computed tomography angiography; and TCFA, thin‐cap fibroatheroma.

Discussion

The current study demonstrated that, in patients with diabetes, (1) TCFA, compared with non‐TCFA, had a significantly higher prevalence of PR, LAP, NRS, and SC as well as significantly higher LDNCP volume and RI on CTA, (2) the presence of NRS, SC, LDNCP volume ≥16.7 mm³, and RI ≥1.42 were independent predictors of TCFA, and (3) when all 4 parameters were present, the probability of TCFA increased to 82.4%. To the best of our knowledge, this is the first study that describes the CTA features of TCFA in patients with diabetes.

CTA Features for TCFA

An autopsy study showed that PR is associated with large lipid core size. ¹⁶ Terashima et al reported that PR and plaque vulnerability were associated with increased oxidative stress using intravascular ultrasound and immunohistochemistry analyses. ¹⁷ A previous intravascular ultrasound study reported that mean CT density was negatively correlated with a necrotic core. ¹⁸ Previous studies revealed that PR, LAP, and NRS were associated with an OCT‐detected TCFA. ⁶ , ⁹ , ¹⁹ In these studies, NRS was consistently an independent predictor of TCFA. This finding was supported by a histology validation study. ²⁰ A lipid‐rich necrotic core corresponds to the central low‐attenuation region on CTA, whereas the fibrous plaque tissue corresponds to the rim of CTA high attenuation. ²⁰ The NRS was thought to be caused by the difference between these attenuations. ²⁰ In a virtual histology intravascular ultrasound study, SC was associated with the lesion with a large lipidic‐necrotic core. ²¹ Persistent hyperglycemia and insulin resistance in patients with diabetes induce oxidative stress and metabolic disarrangements through diverse molecular processes that stimulate the accelerated development, progression, and instability of plaques. ²² , ²³ Several CTA studies have reported that patients with diabetes had a higher prevalence of HRP than those without diabetes. ²⁴ , ²⁵ The current study demonstrated, for the first time, that the prevalence of PR, LAP, NRS, and SC was significantly higher in TCFA than in non‐TCFA in patients with diabetes.

There is increasing evidence on the relationship between atherosclerotic plaque volume and plaque vulnerability. A recent study showed that NCP volume was the strongest independent prognostic predictor for the composite of death and nonfatal myocardial infarction, as compared with obstructive coronary artery disease, total plaque volume, and coronary artery calcium score. ²⁶ Hell et al reported that high NCP volume and LDNCP volume were associated with cardiac death. ²⁷ However, these previous studies used an analysis of the sum of plaque per patient. In our study, we applied a per‐plaque analysis to characterize plaques with TCFA. The current study demonstrated, for the first time, that NCP volume and LDNCP volume were higher in TCFA than in non‐TCFA in patients with diabetes. A previous study reported that LDNCP volume was associated with higher uptake of ¹⁸F‐sodium fluoride, a marker of plaque vulnerability. ²⁸ The current study also demonstrated that RI was higher in TCFA than in non‐TCFA, which is in line with previous studies reporting that high RI was associated with TCFA, ⁶ , ¹⁹ and RI was negatively correlated with fibrous cap thickness. ²⁹ Wen et al reported that high RI correlated with higher uptake of ¹⁸F‐sodium fluoride. ³⁰ RI in our study tended to be higher than that in the previous studies. This could be attributed to the fact that all patients in our study had diabetes. A previous pathology study reported that the internal elastic lamina area (a histological indicator of vessel size), when adjusted for the distance from the coronary ostium, was greater in patients with diabetes relative to those without diabetes. ³¹ In a previous CTA study, patients with diabetes had a higher prevalence of PR than patients without diabetes. ²⁵ Kim et al reported that patients with high insulin resistance had higher RI than patients with low insulin resistance. ³²

Clinical Significance of CTA Predictors of TCFA

Large population studies reported that patients with diabetes and no prior myocardial infarction had the similar cardiovascular risk as patients without diabetes with a previous myocardial infarction. ³³ , ³⁴ Patients with diabetes had a 30% higher mortality when developing a myocardial infarction compared with patients without diabetes. ³⁵ A recent study reported that OCT‐detected TCFA was an independent predictor of future cardiac events in patients with diabetes. ² Therefore, identification of TCFA and individualized aggressive treatment may improve the prognosis in patients with diabetes. The current study demonstrated the prevalence of HRP per plaque in patients with diabetes with coronary artery disease. When all 4 HRP features were present, the probability of TCFA increased to 66.7%. If high LDNCP volume and high remodeling index were used instead of LAP and PR, the probability of TCFA increased further to 82.4%. Although diagnosis of TCFA currently requires OCT, our results may suggest that the integration of qualitative and quantitative CTA analysis could noninvasively detect TCFA in patients with diabetes.

Limitations

This study has several limitations. First, this was a retrospective study that included only patients who underwent both CTA and OCT. In the current study, patients underwent CTA first. Thus, it is possible that those patients without significant stenosis on CTA were not brought to the catheterization laboratory for coronary angiography and OCT. This analysis did not include patients with ST‐segment–elevation myocardial infarction. Therefore, selection bias cannot be ruled out. Second, all patients were enrolled in a single center in Japan. Thus, the current results may not be generalizable to other populations with different ethnic backgrounds. Finally, we had limited information on diabetes‐related characteristics such as the type of diabetes, diabetic duration, and treatment.

Conclusions

The combined qualitative and quantitative plaque analysis of CTA may be helpful in identifying TCFA in patients with diabetes.

Sources of Funding

Dr Jang's research has been supported by Gillian Gray through the Allan Gray Fellowship Fund in Cardiology and by Mukesh and Priti Chatter through the Chatter Foundation. Dr Dey is supported by National Heart, Lung, and Blood Institute grants (1R01HL148787‐01A1 and 1R01HL151266). The funder had no role in the design or conduct of this research.

Disclosures

Dr Jang received educational grants from Abbott Vascular and consulting fees from Svelte Medical Systems. Dr Dey has received software royalties from Cedars‐Sinai Medical Center and has a patent. Dr Ferencik has received consulting fees from Siemens Healthineers, HeartFlow, and Elucid. The remaining authors have no disclosures to report.

Supporting information

Table S1

Figures S1–S2

References 36–39