Triglyceride-glucose index, low-density lipoprotein levels, and cardiovascular outcomes in chronic stable cardiovascular disease: results from the ONTARGET and TRANSCEND trials

Abstract

Aims

The triglyceride-glucose index (TyG) has been proposed as an alternative to insulin resistance and as a predictor of cardiovascular outcomes. Little is known on its role in chronic stable cardiovascular disease and its predictive power at controlled low density lipoprotein (LDL) levels.

Methods and results

Our study population consisted of 29 960 participants in the ONTARGET and TRANSCEND trials that enrolled patients with known atherosclerotic disease. Triglycerides and glucose were measured at baseline. TyG was calculated as the logarithmized product of fasting triglycerides and glucose divided by 2. The primary endpoint of both trials was a composite of cardiovascular death, myocardial infarction, stroke, or hospitalization for heart failure. The secondary endpoint was all-cause death and the components of the primary endpoint. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CI) with extensive covariate adjustment for demographic, medical history, and lifestyle factors. During a mean follow-up of 4.3 years, 4895 primary endpoints and 3571 all-cause deaths occurred. In fully adjusted models, individuals in the highest compared to the lowest quartile of the TyG index were at higher risk for the primary endpoint (HR 1.14; 95% CI 1.05–1.25) and for myocardial infarction (HR 1.30; 95% CI 1.11–1.53). A higher TyG index did not associate with the primary endpoint in individuals with LDL levels < 100 mg/dL.

Conclusion

A higher TyG index is associated with a modestly increased cardiovascular risk in chronic stable cardiovascular disease. This association is largely attenuated when LDL levels are controlled.

Registration

www.clinicaltrials.gov: NCT00153101

Graphical abstract

Lay Summary

• The association of triglyceride-glucose index (TyG) with cardiovascular disease in chronic stable cardiovascular disease and its predictive power at controlled low density lipoprotein (LDL) levels is unclear. Using a study population of 29 960 participants with chronic stable cardiovascular disease, we found that higher TyG levels were associated with a modestly increased risk for incident cardiovascular events and low LDL levels largely attenuated the association of TyG with cardiovascular risk.

Triglyceride-glucose index, Cardiovascular disease, Mortality, Myocardial infarction

Introduction

Triglyceride-glucose index (TyG) has been proposed as an alternative low-cost and efficient biomarker of insulin resistance and metabolic syndrome.^1–3 Many studies have demonstrated that TyG can outperform the homoeostasis model assessment of insulin resistance (HOMA-IR).^4–6 TyG is calculated from routinely available laboratory variables and combines fasting plasma glucose and triglyceride levels and has been shown to be related to cardiometabolic outcomes, such as diabetes, coronary heart disease, and stroke, and was suggested for early identification of people at high cardiovascular risk.^7–13 Limited data are available on the role of TyG in patients already treated for stable cardiovascular disease (CVD).¹⁴ Importantly, it remains unknown whether TyG remains predictive for CVD and death risk at controlled low density lipoprotein (LDL) levels.² This is of significant interest as elevated triglycerides are associated with small-dense LDL particles that are pro-atherogenic and pro-inflammatory.^15,16

This analysis of the ONTARGET and TRANSCEND trials aimed to investigate the relationship of TyG levels with future cardiovascular outcomes and all-cause death in individuals with proven CVD and control of risk factors in a vast majority of participants. Further, we examined whether TyG still associates with the above risks when serum LDL is controlled to low levels.

Methods

Study population

The design, treatment algorithm, and results of ONTARGET (Ongoing Treatment Alone and in Combination With Ramipril Global End Point Trials) and TRANSCEND (Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease) have been reported in detail previously.¹⁷ Briefly, the studies were conducted in 733 centres in 40 countries and coordinated by the Population Health Research Institute at McMaster University (Hamilton, Ontario, Canada), where data analysis was undertaken. A central committee blinded to study treatments adjudicated the endpoints. An independent Data and Safety Monitoring Board monitored the progress of all aspects of the study. The Steering Committee had overall responsibility for interpretation of the results. Both ONTARGET and TRANSCEND were multicentre trials that enrolled patients with known atherosclerotic disease (myocardial infarction, stroke, and proven peripheral artery disease) or diabetes mellitus with end-organ damage. Patients with a history of heart failure and a systolic blood pressure > 160 mmHg or a diastolic blood pressure > 100 mmHg were excluded. The main objectives of the ONTARGET trial were to determine whether the cardiovascular protection offered by telmisartan, and the combination of telmisartan and ramipril was, respectively, non-inferior or superior to that offered by ramipril alone. The main objective of the TRANSCEND trial was to determine whether the cardiovascular protection provided by telmisartan was superior to placebo in people intolerant to ACE inhibitors. All patients provided written informed consent, and protocols were approved by the ethics committees at the participating centres. For this analysis, we included all randomized participants of both trials except individuals with missing data on glucose or triglycerides (n = 345) as well as on important covariates (n = 1241). Our final study population consisted of 29 960 patients (see Supplementary material online, Figure S1).

Biochemical analyses and triglyceride-glucose index

Blood samples were obtained from each patient at baseline at regional study centres. Concentrations of total cholesterol, triglycerides, low density lipoprotein cholesterol, high density lipoprotein cholesterol, and glucose levels were measured using enzymatic assays.^17,18 TyG index was calculated using the following formula: ln [fasting triglycerides (mg/dL) × fasting plasma glucose (mg/dL)]/2.²

Cardiovascular outcomes

In the ONTARGET and TRANSCEND trials, the primary endpoint was a composite of death from cardiovascular causes, non-fatal myocardial infarction, non-fatal stroke, and hospitalization for heart failure.¹⁷ The secondary endpoint of this analysis was all-cause death and the individual components of the primary outcome.

Statistical analysis

Data from the treatment and placebo groups of ONTARGET and TRANSCEND trials were pooled. TyG index quartiles were tested for differences using analysis of variance for continuous data and the χ² test for categorical data. Event rates and cumulative incidence curves of endpoints (considering competing risk of death, where appropriate) are presented by quartiles of TyG at baseline. Cox proportional hazards regression models were constructed, and hazard ratios (HRs) and 95% confidence intervals (CI) for incident primary or secondary events were estimated considering the competing risk of death where appropriate. Beside the raw calculation without any adjustment, two models were used to adjust for confounding. Model 1 adjusted for age, sex, ethnicity, region, education, body mass index (BMI), waist–hip ratio, physical activity, LDL, high density lipoprotein (HDL), systolic blood pressure, diastolic blood pressure, heart rate, tobacco use, estimated glomerular filtration rate (eGFR), alcohol, and trial and randomized treatment group. Adjustment for numeric variables (waist–hip ratio, vital signs, and laboratory values) was made using restricted cubic splines, to account for a potentially non-linear relationship. Model 2 included all variables of Model 1 and additionally adjusted for history of hypertension, myocardial infarction, stroke/transient ischemic attack, and heart rhythm. A subgroup analysis stratifying by diabetes status and TyG quartile including all respective interactions was done. Finally, we conducted an analysis with TyG index and LDL levels as continuous values using restricted cubic splines. All subgroup analyses as well as the analysis with both, TyG and LDL, as continuous variables were fully adjusted according to Model 2. Analyses were performed by SAS statistical software version 9.4 (SAS Institute Inc., Cary, NC). All of the statistical tests were two-sided.

Results

Baseline characteristics by quartiles of TyG index are presented in Table 1. Individuals with higher TyG index were more likely to be younger, female, less physically active, had lower education as well as higher BMI, waist–hip ratio, total cholesterol, LDL, systolic and diastolic blood pressure, and heart rate levels while lower eGFR and HDL levels. These individuals were also more likely to consume alcohol and smoke as well as to have a history of diabetes, hypertension, and CVD.

Table 1

Baseline characteristics by quartiles of triglyceride-glucose index

Characteristic	N (n = 29 960)	Q1 TyG 3.57–4.59 (n = 7490)	Q2 TyG 4.59–4.78 (n = 7489)	Q3 TyG 4.78–5.00 (n = 7491)	Q4 TyG 5.00–6.45 (n = 7490)	P-value
		Mean or %
Age	66.5	67.6	67.0	66.2	65.3	<0.0001
Age group						<0.0001
55–<65 years	42.3	36.8	40.0	43.9	48.5
≥65–<75 years	42.6	43.8	43.5	42.5	40.5
≥75 years	15.1	19.3	16.5	13.6	11.0
Male	70.3	73.3	71.9	68.8	67.1	<0.0001
Race						<0.0001
White	71.4	73.3	72.0	70.7	69.6
Black	2.3	3.2	2.1	2.2	1.9
Asian	15.7	14.5	16.0	16.2	15.9
Other	10.6	9.0	9.9	10.9	12.9
Region						<0.0001
Asia incl. Middle East	15.8	14.8	15.8	16.2	16.5
Australia/New Zealand	6.8	8.4	7.8	6.4	4.6
Europe incl. South Africa	46.9	48.9	47.5	44.9	46.1
Latin America	10.6	9.4	9.6	11.1	12.2
North America	19.9	18.5	19.1	21.5	20.6
Physical activity						<0.0001
Mainly sedentary	22.8	19.7	21.2	23.9	26.6
<Once/week	11.6	10.6	10.3	11.7	13.7
2–4 times/week	22.9	22.9	23.2	23.1	22.5
5–6 times/week	7.7	8.5	8.1	7.7	6.5
Everyday	34.9	38.3	37.1	33.5	30.7
Formal education						0.0001
≤8 years	33.9	32.3	33.0	34.3	35.8
9–12 years	29.4	28.3	28.8	30.1	30.1
Trade/technical School	17.7	18.7	19.1	16.9	16.4
College/university	19.0	20.7	19.1	16.9	16.4
Glucose [mg/dL]	117.5	93.3	103.1	115.7	157.9	<0.0001
Triglycerides [mg/dL]	153.8	80.0	119.4	161.4	254.2	<0.0001
Cholesterol [mg/dL]	190.6	177.1	185.9	193.8	205.8	<0.0001
HDL [mg/dL]	48.7	53.9	49.4	47.0	44.4	<0.0001
LDL [mg/dL]	113.3	106.9	113.3	116.6	116.2	<0.0001
SBP [mmHg]	141.7	140.9	141.1	141.8	142.9	<0.0001
DBP [mmHg]	82.1	81.4	81.9	82.3	82.7	<0.0001
Pulse rate [b.p.m.]	68.0	66.1	67.0	68.1	70.8	<0.0001
Waist/hip ratio	0.94	0.92	0.93	0.94	0.95	<0.0001
Body mass index [kg/m2]	28.1	26.7	27.7	28.7	29.5	<0.0001
eGFR CKD-EPI (mL/min/1.73 m2)	70.7	71.8	70.9	70.4	69.8	<0.0001
Alcohol consumption (%)	38.5	43.3	38.7	37.5	34.5	<0.0001
Tobacco use (%)						0.21
Current	12.0	11.5	11.7	12.9	12.8
Formerly	50.3	50.3	50.7	50.7	49.4
Never	37.7	38.3	37.6	37.3	37.8
Diabetes (%)	40.7	19.1	27.6	43.1	72.9	<0.0001
History of hypertension (%)	70.1	64.1	67.1	71.7	77.3	<0.0001
History of myocardial infarction (%)	48.5	50.9	49.8	48.8	44.3	<0.0001
History of stroke/TIA (%)	21.1	23.3	22.2	20.2	18.7	<0.0001
Heart rhythm at baseline						<0.0001
Sinus	28 034	6953	6953	7082	7046
Atrial fibrillation	987	276	260	198	253
Other	939	261	276	211	191

Q, quartile; LDL, low density lipoprotein; HDL, high density lipoprotein; SBP, systolic blood pressure; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; TIA, transient ischemic attack.

Participants in our analysis were followed for the primary endpoint for a mean (standard deviation) of 4.32 (±1.23) years and 4.56 (±0.94) years for all-cause death. During the follow-up, 4895 primary endpoints occurred (including 2134; 1431; 1310; and 1254 cases of cardiovascular death, myocardial infarction, stroke, and hospitalization for heart failure (HF), respectively) and 3571 all-cause deaths. The annualized crude rate of events was 3.78 per 100 person-years for the composite primary endpoint and 2.61 per 100 person-years for all-cause death. Cumulative incidence plots showing the cumulative incidence of the composite primary endpoint and all-cause death by quartiles of TyG are presented in Figure 1.

Figure 1

Cumulative incidence for the composite primary endpoint (A) and its components (B to F) by quartiles of TyG. CHF, congestive heart failure.

The results of the Cox proportional hazards analysis for risk of the primary endpoint including its individual components and of all-cause death by quartiles of TyG are presented in Figure 2 and Supplementary material online, Table S1. In our unadjusted analyses, we observed a highly significant association between TyG quartiles and the composite primary endpoint (P < 0.0001). Individuals in the highest quartile of TyG index were at a significantly higher risk compared to the lowest quartile (unadjusted Model: HR_{4th vs. 1st quartile} 1.24; 95% CI 1.15–1.34). The association to all-cause death was modest (unadjusted Model: HR_{4th vs. 1st quartile} 1.12; 95% CI 1.02–1.23; P = 0.01). These associations were attenuated after accounting for established cardiovascular risk factors (Model 2: HR _{4th vs. 1st quartile} 1.14; 95% CI 1.05–1.25; P = 0.01), rendering the association with all-cause death non-significant (Model 2: HR_{4th vs. 1st quartile} 1.10; 95% CI 0.99–1.21; P = 0.27), respectively. The associations between TyG index and stroke or heart failure hospitalization were less pronounced than for the combined endpoint (stroke, P = 0.074; heart failure hospitalization, P = 0.05) whereas a robust relationship with myocardial infarction was detected (HR_{4th vs. 1st quartile} 1.30; 95% CI 1.11–1.53; P = 0.01). In the adjusted analyses, no association with cardiovascular death was observed.

Figure 2

Hazard ratios (HRs, 95% CI) for primary (A) and secondary endpoints (B to F) by quartiles of TyG.

The results of the Cox proportional hazards analysis for risk of the primary endpoint and all-cause death by quartiles of TyG and diabetes status are presented in Figure 3 and Supplementary material online, Table S2. There was a modest interaction between diabetes and TyG (P = 0.08 for the combined endpoint; P = 0.11 for all-cause death). Diabetic individuals in all TyG quartiles were at significantly higher risk for the primary endpoint and all-cause death compared to non-diabetic individuals with similar TyG levels (P-value for main effect of diabetes: <0.0001). In patients without diabetes, no risk increase with increasing TyG index was observed.

Figure 3

Hazard ratios (HRs, 95% CI)^† for primary endpoint (A) and all-cause death (B) by quartiles of TyG and diabetes status.

The relationship of TyG index and LDL combined, when both parameters are considered as continuous variables, with the primary endpoint and all-cause death is displayed in Figure 4, and specifically for TyG quartiles and various LDL cut-off levels (i.e. 55, 70, and 100 mg/dL) in Figure 5 and Supplementary material online, Table S3A/B. For the primary endpoint a significant interaction between TyG quartiles and LDL groups was observed (P = 0.02), indicating that the differences between TyG quartiles were different between LDL levels. At higher LDL levels (≥100 mg/dL), there was a highly significant difference between TyG quartiles (P < 0.0001); vice versa in the 4th TyG quartile, there was a highly significant difference between LDL levels (P < 0.0001). In participants in the highest TyG quartile with LDL higher than 100 mg/dL, the risk for primary endpoint was increased by more than 50% compared to the reference (1st TyG quartile/LDL < 55 mg/dL) (HR 1.52; 1.06–2.17) while no increased risk for individuals in the highest TyG quartile with LDL lower than 55 mg/dL compared to the reference (1st TyG quartile/LDL < 55 mg/dL) was detected (HR 1.02; 0.65–1.59). No association with all-cause death was detected.

Figure 4

Hazard ratios for primary endpoint (A) and all-cause death (B) by TyG index and LDL levels.

Figure 5

Risk (hazard ratio, 95% CI) for primary endpoint (A) and all-cause death (B) by TyG index and LDL levels (55, 70, and 100 mg/dL).

Discussion

Meta-analyses of cohort studies reported a higher TyG index to be associated with an increased risk of coronary heart disease.^9,19,20 These analyses were restricted to individuals derived from the general population, who were free of known CVD.⁹ It remains uncertain whether in secondary prevention the residual cardiovascular risk associated with elevated TyG index is of relevance.² Our results from two randomized-controlled trials with ∼29 000 individuals and known atherosclerotic disease extensively treated with cardioprotective therapies close this knowledge gap.

First, we found that a higher TyG index was associated with an increased risk for the composite primary cardiovascular endpoint in a well-treated population. This risk was mainly driven by myocardial infarction events. TyG levels also tended to be related to stroke and heart failure hospitalization but did not reach statistical significance in our sample.^14,21 These results are in line with prior reports on the associations in populations suffering from coronary artery disease, however, our effect sizes were only modest compared to prior evidence. A recent meta-analysis of 12 studies (all conducted in China) comprising 28 795 patients with coronary artery disease reported a 2.14-fold higher risk of major adverse cardiac events for individuals in the highest TyG group compared with those in the lowest TyG group.²⁰ As a caveat, among these 12 studies, only one enrolled individuals with stable coronary artery disease, whereas the vast majority focused on individuals with acute coronary syndrome. In this one analysis, based on a nested case–control study among 3745 patients, TyG index was less strongly associated with cardiovascular risk (HR 1.36; 95% CI 1.10–1.69; P = 0.01).¹⁴ In accordance with prior data, we did not observe an association of TyG with all-cause death that may relate to the obesity paradox.²⁰ In subanalysis, we also found women to be particularly prone to the adverse effects of elevated TyG implying a potentially unmet need for addressing residual cardiovascular risk (data not shown). Sex has previously been identified as a significant effect modifier on the association between TyG and cardiovascular outcomes.²² The I-Lan Longitudinal Aging Study reported a sex difference in the correlation between high TyG and subclinical atherosclerosis, with women particularly affected.²³ Greater TyG was not found to be related to higher risk for cardiovascular outcomes or death in diabetics although a significant effect modification by diabetes was reported previously.^24–26 The underlying explanation for this discrepancy remains unclear but it is likely related to reverse causality as patients previously diagnosed with diabetes are generally intensively treated with cardioprotective agents.

The association of TyG with CVD has been predominantly investigated in Asian populations.^9,19,26 Interestingly, results from the PURE Study could show that the association of TyG with CVD was mainly observed in low- and middle-income countries but not in high-income countries.²⁷ It was hypothesized that this effect modification could be explained by an increased vulnerability of lower income populations.^28,29 Socioeconomic inequalities, mismatches between poverty and undernutrition in early life, and exposure to an obesogenic environment in adulthood are thought to lead to an increased susceptibility to insulin resistance, low-grade inflammation, and chronic diseases, such cardiovascular diseases.^30,31 The pathogenic consequences of TyG may thus affect inhabitants of lower income countries disproportionally compared with individuals with similar characteristics in high-income countries. In this context, another explanation for the varying effects of TyG levels across different countries might be the substantial regional, global as well as gender differences in residual risk factor control and secondary prevention, particularly control of lipids due to lack of access to adequate medication and treatment.^29,32,33

We proceeded to analyse if the relationship between TyG and outcomes persisted when LDL levels were under control. Our data indicate that increasing TyG levels that are indicative of insulin resistance do not increase cardiovascular risk in case of low LDL levels. These findings coincide with most recent results from the PROMINENT trial showing that lowering of mild-to-moderate hypertriglyceridaemia among type 2 diabetes, who already had well controlled LDL levels, did not reduce the incidence of cardiovascular events.³⁴ Taken together, these results emphasize the importance of LDL lowering as the cornerstone of managing diabetic dyslipidaemia and question the prognostic effects of reducing TyG as a method of addressing residual cardiovascular risk among individuals with controlled LDL levels.³⁵ Additionally, these findings highlight the complexity of lipid modification by higher TyG index. Higher triglyceride and glucose levels regulate apoC-III metabolism and have been shown to enfold direct effects on atherogenesis by increasing the affinity of LDL for the artery wall.^36–38 This may partly explain our findings of a higher risk of myocardial infarction associated with higher TyG when LDL levels are not in target range.

This analysis has some strengths and limitations. The results are based on a large ethnically diversified study population with structured assessment of cardiovascular risk factors including a wide range of confounding variables to correct for, adjudicated outcomes, and long-time following with almost no lost-to-follow-up. Conversely, TyG was assessed at baseline, and time-varying data are not available. The findings are based on a post hoc exploratory observational analysis, and the allocation of individuals was not subject to randomization of the study trials.

In conclusion, higher TyG levels are associated with a modestly increased risk for cardiovascular outcomes in chronic stable CVD. However, low LDL levels modify and blunt the association of TyG with cardiovascular risk. These findings emphasize adherence to current guideline-directed treatment recommendations for the prevention of CVD.

Authors’ contributions

B.H., H.S., and M.B. contributed to the conception of design of the work. B.H., M.B., H.S., G.M., K.K.T., F.M., E.L., R.S., J.F.E.M., K.S., and S.Y. contributed to the acquisition, analysis, or interpretation of data for the work. B.H., H.S., and M.B. drafted the manuscript. F.M., G.M., K.K.T., E.L., R.S., J.F.E.M., K.S., and S.Y. critically revised the manuscript. All gave final approval and agreed to be accountable for all aspects of work, ensuring integrity and accuracy.

Supplementary material

Supplementary material is available at European Journal of Preventive Cardiology.

Acknowledgements

We thank Armin Schweitzer, Saarland University Hospital, for his help in preparing the figures of this manuscript.

Data availability

The data underlying this work can be obtained upon reasonable request to the ONTARGET/TRANSCEND Steering Committee.

Author notes

Conflict of interest: M.B. is supported by the Deutsche Forschungsgemeinschaft (German Research Foundation; TTR 219, project number 322900939) and reports personal fees from Abbott, Amgen, Astra Zeneca, Bayer, Boehringer Ingelheim, Cytokinetics, Medtronic, Novartis, Servier, and Vifor during the conduct of the study. B.H. reports no conflicts. F.M. reports grants and personal fees from Medtronic and personal fees from Recor, Boehringer Ingelheim, and Berlin Chemie. He is supported by Deutsche Gesellschaft für Kardiologie (DGK), Deutsche Forschungsgemeinschaft (SFB TRR219, project number 322900939), and Deutsche Herzstiftung, has received scientific support and/or speaker honoraria from Ablative Solutions, Medtronic, and ReCor Medical, and speaker honoraria/consulting fees from Ablative Solutions, Amgen, Astra-Zeneca, Bayer, Boehringer Ingelheim, Inari, Medtronic, Merck, ReCor Medical, Servier, and Terumo. G.M. reports no conflicts. J.F.E.M. reports no conflicts. R.S. reports no conflicts. H.S. reports personal fees from Boehringer Ingelheim. K.K.T. reports no conflicts. E.M.L. reports no conflicts. S.Y. reports institutional grants for the ONTARGET and TRANSCEND trials but not related to the current analysis.