Abstract

Backgrounds:

Triglycerides (TG) and triglyceride-rich lipoproteins contribute to the development and progression of atherosclerosis. However, the prognostic value of TG levels in patients with left main coronary artery disease (LMCAD) remains unexplored. This study aimed to examine the association between TG levels and long-term mortality within this population.

Methods:

We conducted a single-center retrospective study of 2,778 patients with LMCAD undergoing percutaneous coronary intervention (PCI). We modeled the association between TG levels and the hazard ratio (HR) for mortality using restricted cubic splines (RCS). An optimal TG cutoff for stratification was identified using the maximum selected rank statistic, and patients were then divided into two groups. We assessed the proportional hazards assumption with plots of Schoenfeld residuals. The primary endpoint was all-cause death. Secondary endpoints included cardiovascular death, myocardial infarction, stroke, stent thrombosis, and target vessel revascularization.

Results:

Over a mean follow-up of 47.4 ± 30.3 months, 351 (12.6%) patients died, including 207 cardiovascular deaths. Restricted cubic spline analysis showed a linear inverse relationship between TG levels and all-cause mortality. An optimal TG cutoff of 0.93 mmol/L was identified, dividing patients into high-TG (n = 2214) and low-TG (n = 564) groups. The low-TG group had significantly higher all-cause mortality (17.2% vs. 11.5%, log-rank P<0.001). After multivariable adjustment, low TG remained independently associated with a higher risk of all-cause mortality [adjusted HR: 1.379; 95% confidence interval (CI): 1.069–1.778; P = 0.013]. The low-TG group was also associated with a higher risk of cardiovascular mortality (9.4% vs. 7.0%, log-rank P = 0.011). No significant associations were observed between TG levels and other secondary endpoints. Subgroup analyses confirmed the consistent prognostic value of TG across clinical subgroups.

Conclusion:

Our findings indicate that low TG levels are an independent prognostic factor for all-cause mortality among LMCAD patients undergoing PCI.

1 Introduction

The left main (LM) coronary artery supplies 60%–75% of the left ventricular myocardium. Consequently, LM lesions pose a high risk and are associated with poor outcomes. Although coronary artery bypass grafting (CABG) remains the gold standard, advances in next-generation drug-eluting stents and intravascular imaging have established percutaneous coronary intervention (PCI) as a robust alternative (1). Hypertriglyceridemia is a known risk factor for pancreatitis and atherosclerotic cardiovascular disease (ASCVD) (2, 3). Current evidence suggests that while triglycerides (TG) themselves may not be directly atherogenic, the cholesterol content within TG-rich lipoproteins (termed remnant cholesterol) contributes significantly to residual cardiovascular risk (4–6). However, the prognostic value of serum TG levels in LM lesions remains poorly defined. Therefore, this study aimed to investigate the association between baseline serum TG levels and long-term mortality in patients undergoing PCI for significant left main coronary artery disease (LMCAD).

2 Method2.1 Study population

We identified 3,347 consecutive patients with LMCAD undergoing PCI at Nanjing First Hospital from January 2011 to August 2022. After excluding patients with prior CABG (62), procedural failure or death (82), in-stent restenosis (197), non-significant LM disease (78), lost follow-up (130), other reasons (20), including concurrent transcatheter aortic valve replacement (TAVR) during the PCI procedure (10), iatrogenic left main coronary artery dissection (7), and spontaneous left main coronary artery dissection (3), 2,778 patients remained for final analysis (Figure 1). PCI was performed according to current clinical guidelines. For the purpose of this study, PCI was defined as successful stent implantation in the left main coronary artery. Patients treated with percutaneous transluminal coronary angioplasty (PTCA) alone or those with failed stent delivery were excluded from the final analysis. Participants were then stratified by TG level. This study adhered to the Declaration of Helsinki and was approved by the institutional ethics committee (KY20170904-06); written consent was waived due to the retrospective design.

Flowchart of patient selection.

2.2 Data collection

All key procedural decisions were made by interventional cardiologists. Procedural success was defined as achieving Thrombolysis in Myocardial Infarction (TIMI) flow grade 3 with residual stenosis <30% (7). We collected data on clinical comorbidities (hypertension, dyslipidemia, and diabetes mellitus), medications, and laboratory values, including hemoglobin (Hb), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein (HDL-C) cholesterol, triglycerides (TG), fast blood glucose (FBG), creatine kinase MB (CK-MB), albumin, uric acid, and creatinine (Scr). We also recorded specific procedural data, including the number of LM stents, stent diameter and length, lesion location, intravascular ultrasound (IVUS), dual antiplatelet therapy (DAPT), and the use of intra-aortic balloon pump (IABP).

2.3 Clinical outcomes

All patients underwent regular follow-up via telephone interviews or clinical visits, with an initial minimum follow-up duration of 1 year. The primary endpoint was all-cause death. Secondary efficacy endpoints, defined according to the ULTIMATE-DAPT trial criteria (8), comprised cardiovascular death, myocardial infarction (MI), ischemic stroke, stent thrombosis (ST), and target vessel revascularization (TVR). The primary safety endpoint was major bleeding, defined as Bleeding Academic Research Consortium (BARC) types 3–5 (9).

2.4 Statistical analysis

Continuous variables are expressed as mean ± standard deviation (SD), and categorical variables are expressed as frequencies and percentages. Comparisons between groups were performed using Student's t-test or the Mann–Whitney U test for continuous data, and the Chi-square or Fisher's exact test for categorical data. The association between TG levels and mortality was modeled using restricted cubic splines (RCS) within a Cox proportional hazards framework. Nonlinearity was evaluated using likelihood ratio tests, and dose-response relationships were visualized using RCS curves with hazard ratios (HRs) and 95% confidence intervals (CIs) (10). The optimal TG cutoff for risk stratification was determined using maximally selected rank statistics (11). The proportional hazards assumption was verified using Schoenfeld residual plots, confirming its validity throughout the follow-up period. Survival outcomes across TG groups were compared using Kaplan–Meier curves and the log-rank test. Multivariable Cox proportional hazards analysis was performed using a block-wise entry approach. Established cardiovascular risk factors, including age, gender, BMI, LDL-C, and FBG, were force-entered (Block 1) into the model to ensure adequate adjustment. Other candidate laboratory and clinical variables were entered in the second block and subjected to a backward stepwise elimination process (Likelihood Ratio test) with a removal criterion of P > 0.05 (Block 2). This strategy ensured that the independent association between TG levels and mortality was adjusted for both statistically significant predictors and clinically essential confounders. Consistency of the TG-mortality association was evaluated through pre-specified subgroup analyses; interaction terms were included in the models, with P > 0.05 indicating a consistent effect across subgroups.

Statistical analyses were performed using IBM SPSS Statistics version 27.0 (IBM Corp., Armonk, NY, USA) and R software version 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria). Specifically, baseline characteristics and Cox proportional hazards regression analyses were conducted in SPSS. R software, incorporating the rms and survival packages, was utilized for data visualization and advanced modeling, including the construction of the nomogram and Kaplan–Meier survival curves.

3 Results3.1 Baseline characteristics

A total of 2,778 patients with LMCAD undergoing PCI were included in this study. The cohort's mean age was 67.34 ± 10.33 years, and 77.1% were male (Table 1). During a mean follow-up of 47.4 ± 30.3 months, 351 (12.6%) patients died, including 207 cardiovascular deaths.

CharacteristicsOverall (n = 2778)Low TG (n = 564)High TG (n = 2214)P-valueAge (y)67.34 ± 10.3370.10 ± 10.0266.64 ± 10.29<0.001Male (n)2143 (77.1)490 (86.9)1653 (74.7)<0.001BMI (kg/m2)24.48 ± 3.1623.40 ± 3.2324.75 ± 3.08<0.001Heart rate (bpm)73.67 ± 11.7274.02 ± 12.3673.58 ± 11.550.421Mean blood pressure (mmHg)96.31 ± 11.1695.84 ± 11.1596.43 ± 11.160.261Hypertension (n)1983 (71.4)379 (67.2)1604 (72.4)0.014DM (n)931 (33.5)175 (31.0)756 (34.1)0.161Dyslipidemia (n)2058 (74.1)307 (54.4)1751 (79.1)<0.001Prior myocardial infarction(n)293 (10.5)73 (12.9)220 (9.9)0.038Prior PCI (n)539 (19.4)127 (22.5)412 (18.6)0.036AMI (n)651 (23.4)128 (22.7)523 (23.6)0.643Hb (g/L)130.44 ± 17.28126.45 ± 17.13131.46 ± 17.18<0.001FBG (mmol/L)6.23 ± 2.305.71 ± 1.956.36 ± 2.37<0.001Albumin (g/L)38.25 ± 3.5437.18 ± 3.8638.52 ± 3.40<0.001Uric acid (umol/L)344.33 ± 107.61321.55 ± 97.11350.13 ± 109.39<0.001LDL-C (mmol/L)2.36 ± 1.471.95 ± 0.792.47 ± 1.58<0.001HDL-C (mmol/L)0.98 ± 0.231.07 ± 0.260.96 ± 0.22<0.001Scr (umol/L)86.21 ± 64.3986.61 ± 55.6186.11 ± 66.450.869CK-MB (ng/mL)18.05 ± 33.3219.39 ± 40.3717.71 ± 31.280.285Ticagrelor (n)1338 (48.2)266 (47.2)1072 (48.4)0.594ACE inhibitor or ARB (n)1637 (58.9)311 (55.1)1326 (59.9)0.041Beta-blockers (n)1780 (64.1)325 (57.6)1455 (65.7)<0.001Statins (n)2716 (97.8)548 (97.2)2168 (97.9)0.276Oral Antidiabetic Drugs (n)688 (24.8)131 (23.2)557 (25.2)0.343Insulin (n)344 (12.4)66 (11.7%)278 (12.6)0.582DAPT (m)15.33 ± 11.3814.80 ± 10.0815.46 ± 11.690.219LM lesion location0.856 Ostium (n)246 (8.9)47 (8.3)199 (9.0) Shaft (n)68 (2.4)13 (2.3)55 (2.5) Distal bifurcation (n)2464 (88.7)504 (89.4)1960 (88.5)LM True-bifurcation (n)1323 (47.6)265 (47.0)1058 (47.8)0.734triple-vessel disease (n)1441 (51.9)297 (52.7)1144 (51.7)0.675Length of LM stent (mm)25.29 ± 7.6225.17 ± 7.1825.32 ± 7.730.675Diameter of LM stent (mm)3.43 ± 0.353.42 ± 0.353.43 ± 0.350.5962-stent strategy (n)679 (24.4)139 (24.6)540 (24.4)0.900IVUS (n)1046 (37.7)212 (37.6)834 (37.7)0.972IABP (n)175 (6.3)35 (6.2)140 (6.3)0.918

Baseline demographic, angiographic, and procedural characteristics between low and high TG groups.

Values are n (%) or mean ± SD. BMI, body mass index; DM, diabetes; PCI, percutaneous coronary intervention; AMI, acute myocardial infarction; Hb, hemoglobin; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride; FBG, fasting blood glucose; Scr, serum creatinine; CK-MB, creatine kinase MB; DAPT, dual antiplatelet therapy; ARB, angiotensin receptor blocker; ACEI, angiotensin-converting enzyme inhibitor. IVUS, intravascular ultrasound; LM, left main; IABP, intra-aortic balloon pump. True bifurcation was defined as a Medina classification type of 1,1,1, 1,0,1, or 0,1,1.

3.2 Univariate and multivariable predictors of all-cause mortality

Univariate analysis screened 23 potential variables associated with all-cause mortality (Table 2). Subsequent multivariable Cox regression analysis confirmed that age, diabetes mellitus, prior MI, hemoglobin, albumin, uric acid, FBG, TG, creatinine, CK-MB, DAPT duration, triple-vessel disease, LM stent diameter, and IABP use remained independent predictors of all-cause mortality.

CovariatesHR (95%CI)P-valueHR (95%CI)P-valueAge, per 5 years increase1.36 (1.28–1.445)<0.0011.259 (1.179–1.344)<0.001Male0.958 (0.748–1.226)0.7320.909 (0.691–1.195)0.492BMI (kg/m2)0.942 (0.91–0.975)0.0011.000 (0.966–1.036)0.983Heart rate, per 5 bpm increase1.095 (1.053–1.138)<0.001Mean blood pressure, per 5 mmHg increase0.998 (0.952–1.047)0.937Hypertension1.411 (1.093–1.822)0.008DM1.577 (1.276–1.948)<0.0011.338 (1.045–1.715)0.021Dyslipidemia1.119 (0.871–1.439)0.378Prior MI1.802 (1.357–2.393)<0.0011.740 (1.298–2.332)<0.001Prior PCI1.302 (1.019–1.664)0.035AMI1.576 (1.256–1.977)<0.001Hb, per 5 g/L increase0.860 (0.835–0.886)<0.0010.927 (0.895–0.960)<0.001Albumin (g/L)0.873 (0.848–0.899)<0.0010.941 (0.912–0.972)<0.001Uric acid, per 50 umol/L increase1.152 (1.104–1.203)<0.0011.108 (1.060–1.157)<0.001LDL-C (mmol/L)0.953 (0.854–1.064)0.3911.026 (0.966–1.090)0.406FBG (mmol/L)1.077 (1.039–1.117)<0.0011.068 (1.023–1.114)0.003TG (mmol/L)0.730 (0.624–0.855)<0.0010.815 (0.688–0.966)0.018HDL-C (mmol/L)0.578 (0.355–0.942)0.028Scr, per 5 umol/L increase1.013 (1.010–1.016)<0.0011.008 (1.004–1.012)<0.001CK-MB, per 20 ng/mL1.046 (1.009–1.084)0.0141.073 (1.029–1.119)0.001Ticagrelor0.730 (0.630–1.007)0.057ACE inhibitor or ARB0.873 (0.706–1.081)0.214Betablockers0.930 (0.748–1.156)0.514Statins1.552 (0.642–3.752)0.330Oral Antidiabetic Drugs1.223 (0.967–1.547)0.092Insulin2.049 (1.574–2.668)<0.001DAPT (m)0.979 (0.967–0.991)0.0010.967 (0.955–0.980)<0.001LM distal bifurcation1.329 (0.937–1.885)0.111LM True-bifurcation1.495 (1.211–1.846)<0.001Triple-vessel disease1.892 (1.519–2.356)<0.0011.405 (1.119–1.765)0.003Length of LM stent1.000 (0.986–1.014)0.967Diameter of LM stent0.442 (0.327–0.597)<0.0010.597 (0.434–0.820)0.0012-stent strategy1.327 (1.060–1.660)0.013IVUS0.824 (0.661–1.027)0.085IABP2.634 (1.952–3.553)<0.0011.455 (1.059–1.999)0.021

Univariate and multivariable Cox proportional hazards analysis of predictors for all-cause mortality in LMCAD patients undergoing percutaneous coronary intervention.

Values are n (%) or mean ± SD. BMI, body mass index; DM, diabetes; PCI, percutaneous coronary intervention; AMI, acute myocardial infarction; Hb, hemoglobin; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride; FBG, fasting blood glucose; Scr, serum creatinine; CK-MB, creatine kinase MB; DAPT, dual antiplatelet therapy; ARB, angiotensin receptor blocker; ACEI, angiotensin-converting enzyme inhibitor. IVUS, intravascular ultrasound; LM, left main; IABP, intra-aortic balloon pump.

3.3 Dose-response relationship between triglyceride levels and mortality risk

Both univariate and multivariable RCS analyses revealed a significant linear inverse association between TG levels and all-cause mortality (P-non-linear = 0.4326 and 0.2176, respectively). The multivariable model was rigorously adjusted for age, gender, BMI, LDL-C, FBG, DM, prior MI, hemoglobin, albumin, uric acid, creatinine, CK-MB, DAPT, triple-vessel disease, LM stent diameter, and IABP. In the crude model, the hazard ratio (HR) crossed the point of neutrality at a TG threshold of 1.36 mmol/L; notably, this threshold shifted to 0.76 mmol/L after adjusting for the aforementioned clinical and laboratory confounders (Figure 2).

Restricted cubic spline (RCS) curves for the association between triglyceride levels and all-cause mortality. (A) Unadjusted Cox proportional hazards model demonstrating the hazard ratio (HR) for TG levels. (B) Multivariable-adjusted model, accounting for Age, gender, BMI, LDL-C, FBG, DM, Prior MI, Hemoglobin, Albumin, Uric acid, creatinine, CK-MB, DAPT, Triple-vessel disease, LM stent Diameter, and IABP. The solid lines represent the HRs, and the shaded areas indicate the 95% confidence intervals. The likelihood ratio test confirmed a persistent, non-linear relationship that did not reach statistical significance after adjustment (P for nonlinearity = 0.2176).

3.4 Risk stratification using maximally selected rank statistics

We performed maximally selected rank statistics to identify the optimal TG cutoff for risk stratification of all-cause mortality. Using the maximally selected rank statistic approach, an optimal TG cutoff of 0.93 mmol/L was identified for risk stratification, corresponding to a maximum log-rank statistic of 19.29 (Figure 3). Patients were then categorized into a high-TG group (≥0.93 mmol/L, n = 2214) and a low-TG group (<0.93 mmol/L, n = 564). Evaluation of the proportional hazards assumption using Schoenfeld residuals confirmed the validity of the Cox model (global test P = 0.436), indicating that the effect of TG stratification remained consistent throughout follow-up (Figure 4).

Distribution of TG levels and the identified optimal cutoff value.

Schoenfeld residual plots for proportional hazards assumption.

3.5 Endpoint analysis

Compared with the high-TG group, patients in the low-TG cohort were older and exhibited significantly lower body mass index (BMI) and lower levels of hemoglobin, albumin, LDL-C, FBG, and uric acid (Table 1). Kaplan–Meier analysis (Figure 5A) showed a significantly lower cumulative survival rate in the low-TG group than in the high-TG group (P < 0.001, log-rank test). After multivariable adjustment, low TG levels remained independently associated with an increased risk of all-cause mortality (17.2% vs. 11.5%; adjusted HR: 1.379; 95% CI: 1.069–1.778; P = 0.013; Table 3). Similarly, the low-TG group experienced higher cardiovascular mortality (Log-rank P = 0.011; Figure 5B), which persisted as an independent predictor after adjusting for potential confounders (adjusted HR: 1.458; 95% CI: 1.041–2.043; P = 0.028). Notably, no significant differences were observed in other secondary endpoints, including myocardial infarction, stroke, target vessel revascularization, stent thrombosis, or major bleeding (BARC 3–5).

Kaplan-Meier curves for (A) all-cause death and (B) cardiovascular death between TG groups.