Coronary artery disease (CAD) is a leading cause of morbidity and mortality worldwide [1]. It is important to understand the coronary atherosclerotic progression for the prevention of adverse cardiovascular (CV) events. Numerous previous studies have suggested the significant role of insulin resistance (IR) in the development of CAD [2,3,4]. Recently, the triglyceride glucose (TyG) index has been suggested to be a reliable surrogate marker of IR [5,6,7]. Several cross-sectional studies have reported that TyG index is associated with CAD, especially with coronary artery calcification (CAC) [8, 9]. However, longitudinal data on the association between TyG index and coronary plaque progression (PP) is scarce. Coronary computed tomography angiography (CCTA) is a well-established non-invasive imaging tool with high diagnostic performance for coronary atherosclerosis and predictive value for adverse CV events [10,11,12,13]. Therefore, we aimed to examine the association between baseline TyG index and coronary PP using serial CCTA.


Study design and populations

The Progression of AtheRosclerotic PlAque DetermIned by Computed TomoGraphic Angiography IMaging (PARADIGM) is a prospective, international, and multicenter observational registry designed to evaluate associations between clinical variables and coronary atherosclerotic changes using serial CCTA [14]. Between 2003 and 2015, 2252 consecutive subjects underwent serial CCTA at 13 centers in 7 countries. Among these subjects, 1143 subjects with available information on TyG index and diabetic status were included in the present study. The characteristics of coronary plaques in all participants were categorized based on the TyG index tertile. TyG index was calculated as ln [fasting triglycerides (mg/dL) × fasting glucose (mg/dL)/2]. Diabetes was defined as treatment with oral hypoglycemic agent or insulin or fasting blood glucose (FBG) ≥ 126 mg/dL. The institutional review boards approved this study at each site.

Acquisition and interpretation of CCTA

All data acquisition and post-processing of CCTA were in accordance with the Society of Cardiovascular Computed Tomography guidelines [15, 16]. CCTA was performed with a ≥ 64-detector row scanner at all centers. All datasets from each center were transferred to an offline workstation for analysis with a semi-automated plaque analysis software (QAngioCT Research Edition v2.1.9.1; Medis Medical Imaging Systems, Leiden, the Netherlands) using manual correction. Segments with diameter ≥ 2 mm were evaluated using a modified 17-segment American Heart Association model [16]. Regardless of the presence of atherosclerotic plaques, plaque volume (PV) (mm3) of every coronary segment was obtained and summated to generate total PV per patient. Coronary plaques were further classified by composition according to the pre-defined intensity cut-offs in Hounsfield units (HU) for calcified plaques (≥ 351 HU), fibrous plaques (131–350 HU), fibro-fatty plaques (31–130 HU), and necrotic cores (-30 to 30 HU) [17, 18]. For comparing longitudinal CCTA images, all baseline and follow-up coronary segments were registered together with fiduciary landmarks, including the distance from the ostia or branch vessel take-offs. PV change was defined as plaque volume at follow-up CCTA minus plaque volume at baseline CCTA. Annual change of PV (mm3/year) was defined as total PV change divided by inter-scan period. Moreover, normalized total atheroma volume (TAVnorm) (mm3) was defined as total PV divided by vessel length, multiplied by the mean participants’ vessel length. Annual change of TAVnorm (mm3/year) was defined as TAVnorm divided by the inter-scan period. While total percent atheroma volume (PAVtotal) (%) was defined as PV divided by vessel volume, multiplied by 100, annual change of PAVtotal (%/year) was defined as total PAV divided by inter-scan period, and plaque progression (PP) was defined as the difference in plaque volume between follow-up and baseline CCTA > 0. Further, rapid PP (%/year) was defined as an annual progression of PAV ≥ 1.0% [19, 20]. Representative CCTA images are presented in Fig. 1.

Fig. 1
figure 1

Representative CCTA images. CCTA coronary computed tomography angiography, TyG triglyceride glucose

Statistical analysis

Continuous variables are expressed as mean ± SD or medians and interquartile range, while categorical variables are presented as absolute values and proportions. Continuous variables were compared using an independent t test or the Mann–Whitney U-test, as appropriate and categorical variables were compared using the χ2-test or Fisher’s exact test, as appropriate. Coronary characteristics across TyG index tertiles were compared using one-way analysis of variance or the Kruskal–Wallis test for continuous variables, as appropriate. Univariate logistic regression analysis was performed to evaluate the association between clinical variables and coronary PP. Further, multivariate logistic regression analyses were performed to identify the independent impact of TyG index on coronary PP. Variables with P < 0.05 in the univariate logistic regression analysis were considered confounding variables and entered into the multivariate logistic regression models, except the individual component of TyG index. All statistical analyses were performed using the Statistical Package for the Social Sciences version 19 (SPSS, Chicago, Illinois). A P value < 0.05 was considered statistically significant for all analyses.


Baseline characteristics

The mean age of the 1143 participants (624 male, 54.6%) was 60.7 ± 9.3 years. Median inter-scan period was 3.2 (range, 2.6–4.4) years. Coronary PP was observed in 883 (77.3%) participants during follow-up. The clinical characteristics of participants according to PP are presented in Table 1. Age, systolic blood pressure (BP), body mass index (BMI), serum triglyceride and FBG levels, prevalence of male sex, hypertension, diabetes, hyperlipidemia, and the use of aspirin, angiotensin-converting enzyme inhibitor (ACEI)/angiotensin receptor blocker (ARB), and statin were significantly higher in subjects with PP than in those without it. Subjects with PP had significantly lower levels of high-density lipoprotein cholesterol (HDL-C) than those without PP. TyG index values were higher in subjects with PP than in those without it (8.85 ± 0.60 vs. 8.69 ± 0.55; P < 0.001).

Table 1 Baseline characteristics

Comparison of baseline PV and annual change of PV according to TyG index tertile

Baseline total PV (mm3) was as follows: group I [lowest]: 30.8 (0.0–117.7), group II: 47.2 (6.2–160.4), and group III [highest]: 57.5 (8.4–154.3), P < 0.001. Baseline TAVnorm values were as follows: group I, 33.0 (0.0–122.3); group II, 54.1 (7.4–192.3); and group III, 61.2 (9.5–165.6); P = 0.001. PAVtotal was as follows: group I, 1.6 (0.0–6.1); group II: 2.8 (0.4–8.8); and group III: 3.0 (0.5–8.1); P = 0.001. There were significant differences among the TyG index tertile groups at baseline. Regarding coronary plaque subtypes, there was a significant difference in the fibrous, fibrous-fatty, necrotic-core, and dense calcium PVs among all groups at baseline. During follow-up, the annual change of the total PV was as follows: group I, 5.7 (0.0–20.2); group II, 7.6 (0.5–23.5); and group III, 9.4 (1.4–27.7); P = 0.0101; 2) and of TAVnorm was as follows: group I, 6.2 (0.0–19.9); group II, 7.8 (0.5–25.4); and group III, 9.3 (1.7–31.2); P = 0.005. PAVtotal [group I: 0.3 (0.0–0.9), group II: 0.4 (0.0–1.3), and group III: 0.5 (0.1–1.4); P = 0.006] was different among all the groups. There was a significant difference in the annual change of fibrous and dense calcium PVs (Table 2).

Table 2 Baseline and changes in the coronary plaque characteristics

Association of clinical variables with coronary atherosclerotic change

Age (odds ratio [OR] 1.031; 95% confidence interval [CI] 1.016–1.047; P < 0.001), male sex (OR 1.490; 95% CI 1.129–1.967; P = 0.005), systolic BP (OR 1.012; 95% CI 1.003–1.022; P = 0.012), BMI (OR 1.077; 95% CI 1.026–1.130; P = 0.003), and HDL-C (OR 0.987; 95% CI 0.976–0.998; P = 0.017) were associated with coronary PP. Among the TyG tertile groups, PP risk was increased in group III compared with that in group I (OR 1.648; 95% CI 1.167–2.327; P = 0.005) (Table 3).

Table 3 Univariate logistic regression analysis for the association of clinical variables with the risk of coronary PP

Subgroup analysis for the relationship of TyG index with coronary PP

Figure 2 shows the subgroup analysis of the estimated OR of TyG index for coronary PP. The TyG index was significantly associated with an increased risk of PP in subgroups of aged < 65 years (OR 1.584; 95% CI 1.190–2.109; P = 0.002), females (OR 2.061; 95% CI 1.435–2.961; P < 0.001), as well as those without hypertension (OR 1.762; 95% CI 1.249–2.484; P = 0.001), and diabetes (OR 1.473; 95% CI 1.091–1.990; P = 0.012). The same association was observed with hyperlipidemia (OR 1.546; 95% CI 1.151–2.076; P = 0.004), BMI ≥ 25.0 kg/m2 (OR 1.564; 95% CI 1.134–2.157; P = 0.006), and current smoking status (OR 1.569; 95% CI 1.193–2.064; P = 0.001).

Fig. 2
figure 2

Subgroup analysis for the impact of TyG index on coronary PP. TyG triglyceride glucose, PP plaque progression

TyG index on the risk of coronary PP

The results of multiple logistic regression models for the association between TyG index and PP risk are presented in Table 4. Increased TyG index values were significantly related to an increased risk of PP after adjusting for other confounding variables. After adjusting for traditional CV risk factors, TyG index was associated with coronary PP (OR 1.308; 95% CI 1.004–1.703; P = 0.046) (Additional file 1: Table S1). TyG index was particularly associated with the calcified PP among coronary plaque sub-types (Additional file 2: Table S2). Regarding rapid coronary PP, multivariate logistic regression analysis showed that the risk of rapid PP was increased in group III (OR 1.557; 95% CI 1.109–2.185; P = 0.011) compared with group I (Table 5).

Table 4 Multiple logistic models for the impact of TyG index on coronary PP
Table 5 Association of TyG index and traditional risk factors with rapid PP


Main findings

To the best our knowledge, this is first study to evaluate the longitudinal quantitative changes of coronary plaques and their subtypes related to TyG index using serial CCTA. This study identified a significant association between TyG index and coronary atherosclerosis progression. Previous cross-sectional studies have reported a significant relationship between TyG index and CAC prevalence [8, 9]. A recent longitudinal study revealed that elevated TyG index is independently associated with CAC progression [21]. However, this study had a retrospective design and included only a Korean population, which were limitations. Additionally, considering that non-calcified plaques might be related to an increased risk of acute coronary syndrome events [22], it might be important to compare longitudinal changes of non-calcified plaques according to TyG index values. In the present PARADIGM study, which had a prospective, international, and observational design, we identified that the baseline total PV and all subtypes as well as annualized change in total, fibrous, and dense-calcium PV increased with increasing TyG index values. In addition, the TyG index had a positive association with the annual change of total PV, TAVnorm, and PAVtotal (Additional file 3: Table S3). Even after adjusting for confounding factors, TyG index was related to the increased risk of PP as well as rapid PP. Regarding coronary plaque sub-types, TyG index was found to be associated with calcified PP after adjusting for traditional CV risk factors in a previous cross-sectional cohort study [9].

Recent investigations on the longitudinal assessment of coronary atherosclerosis

To understand that the coronary atherosclerotic change is an important issue in clinical practice, it is well-known that diabetes has close association with the prevalence and severity of CCTA verified CAD progression [23]. Even asymptomatic diabetic patients experience plaque progression as well as evolution to overt or silent CAD, and an increase in the PV was reported to be associated with subsequent CV events [24]. In addition, the increased duration of diabetes combined with higher HbA1c levels deleteriously influences culprit-plaque characteristics among diabetic patients who suffered acute myocardial infarction [25]. A rapid plaque progression was specially observed in male patients and in patients with typical angina [26]. While helical flow in coronary arteries has a protective role against atherosclerotic wall thickness growth [27], an intrinsic calcification angle, defined as the angle externally projected by a vascular calcification, is a novel feature of coronary plaque vulnerability and its impact on fibrous cap stress is potentiated in more superficial calcifications, adding to the destabilizing role of smaller calcifications [28].

Focused issue for the significance of TyG index

It is well-established that IR is a main mechanism in the development of type 2 diabetes. A previous PARADIGM study identified that individuals with established diabetes experienced greater PP, particularly, significantly greater progression of adverse plaque formation than those without diabetes [29]. In addition, unlike diabetes, pre-diabetic condition was not independently associated with coronary PP in the sub-study of same registry [30]; however, although pre-diabetes was defined according to the criteria used in previous studies, glycemic status was assessed based on only the levels of FBG and HbA1c without considering IR status among non-diabetic participants. According to the results of a recent large cross-sectional cohort study [31], TyG index had an independent and positive association with the risk of CAD and obstructive CAD in non-diabetic individuals; however, glycemic control status reflected in HbA1c rather than IR parameters was significantly related to the risk of both CAD and obstructive CAD in individuals with established diabetes. These results might support the hypothesis for the different pathogenesis of CAD according to diabetic status. In clinical practice, atherosclerosis-related adverse events commonly occurred even in people with low CV risk burden [32,33,34]. Thus, early detection of the presence and progression of subclinical atherosclerosis in this population is important. Recent studies have focused on defining useful predictors for subclinical atherosclerosis in individuals with low CV risk [35, 36]. Interestingly, although the statistical significance could be influenced by the sample size of the individual subgroup, this study showed that TyG index had a significant predictive value for PP in individuals without the traditionally known CV risk factors, especially in female subgroup. This result suggests that TyG index is a potential surrogate marker for the early detection of subclinical atherosclerosis in the absence of CV risk factors as reported in a recent cross-sectional cohort study [37]. Considering the pivotal role of IR in atherosclerosis progression by promoting apoptosis of macrophages, endothelial cells, and vascular smooth muscle cells [38,39,40], further prospective studies with larger sample sizes will be necessary to address the predictive value of TyG index for subclinical atherosclerosis in individuals with low CV risk burden.


There are some limitations in the present study. First, we only evaluated the association between baseline TyG index and coronary atherosclerotic change; longitudinal consecutive changes of TyG index during follow-up could not be confirmed. Second, the effects of anti-hypertensive and anti-diabetic medications were not controlled for because of the observational nature of the study design. Third, the homeostatic model assessment of insulin resistance was not analyzed and compared with TyG index because insulin levels were not measured in the PARADIGM registry. Fourth, we could not confirm the TyG index of the small coronary arteries in the present study. Fifth, a selection bias might be present because of the retrospective inclusion of participants. In addition, the results of CCTA at baseline could affect the performance of follow-up CCTA. Finally, despite our application of strict and standardized criteria for assessing CCTA, atherosclerotic findings can be affected by HU density. Despite these limitations, this study used serial CCTA to estimate coronary PVC and PP according to TyG index values in a large multicultural cohort subjects.


The present study demonstrates the independent association between TyG index values and coronary PP based on serial quantitative assessment by CCTA during a relatively short-term period. Further large prospective and randomized studies with longer follow-up durations are necessary to confirm the results of the present study.