Clinical correlates of early post-PCI left ventricular ejection fraction in patients with STEMI: a single-center retrospective study

Abstract

Introduction

Early left ventricular systolic dysfunction after primary percutaneous coronary intervention (PCI) remains common in ST-segment elevation myocardial infarction (STEMI) and is associated with adverse remodeling and heart failure. The clinical relevance of lipid markers and lipid ratios for characterizing early post-PCI ventricular dysfunction beyond angiographic and hemodynamic features remains uncertain.

Methods

We conducted a single-center retrospective study of consecutive patients with STEMI undergoing emergency PCI at Nanjing BenQ Hospital (2014–2023). LVEF was assessed by transthoracic echocardiography 24–72 h post-PCI, and reduced LVEF was defined as < 50%. Lipids were measured from the first blood sample obtained on admission, and LDL-C/HDL-C and TG/HDL-C were calculated. Multivariable linear and logistic regression models with hierarchical adjustment (Model 2 additionally including log1p-transformed troponin I measured from the first blood sample obtained on admission) were used to identify factors associated with LVEF and reduced LVEF. Discrimination of lipid markers was evaluated using ROC analysis.

Results

Among 254 patients, 45 (17.7%) had reduced LVEF. LAD as the infarct-related artery (IRA) was more frequent in the reduced-LVEF group (73.3% vs. 48.3%, P = 0.002). LAD-IRA was independently associated with lower LVEF (β = −5.240; P < 0.001) and higher odds of reduced LVEF (OR = 4.005; P < 0.001). Higher systolic blood pressure was associated with lower odds of reduced LVEF (P = 0.026). TG/HDL-C showed modest discrimination (AUC = 0.605), whereas LDL-C showed no meaningful discrimination (AUC = 0.492).

Conclusion

In PCI-treated STEMI, LAD culprit territory and admission systolic blood pressure were the main clinical correlates of early post-PCI LV dysfunction, whereas TG/HDL-C showed only modest discrimination and limited standalone clinical utility. These findings should be considered exploratory and hypothesis-generating.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12872-026-06309-3.

Introduction

Acute ST-segment elevation myocardial infarction (STEMI) remains a leading cause of cardiovascular mortality and morbidity worldwide [1]. Primary percutaneous coronary intervention (PCI) is well established as the standard of care for timely restoration of coronary blood flow and limitation of infarct size [2]. However, despite angiographically successful epicardial reperfusion, a substantial proportion of patients develop early left ventricular (LV) systolic dysfunction, which constitutes a critical determinant of subsequent heart failure, adverse remodeling, and long-term mortality [3]. Early stratification of patients at high risk for reduced left ventricular ejection fraction (LVEF) is therefore essential for informing adjuvant therapies, optimizing in-hospital monitoring intensity, and tailoring surveillance strategies during the index hospitalization [4]. In addition to imaging- and biomarker-based approaches, electrocardiographic markers such as the frontal QRS-T angle have also been proposed as noninvasive indicators of ventricular depolarization–repolarization heterogeneity and cardiovascular risk, including myocardial infarction, heart failure, and sudden cardiac death [5].

The extent of myocardial injury and the trajectory of LV recovery are shaped by a complex interplay of anatomical, hemodynamic, and metabolic factors [6]. Historically, the location of the culprit vessel—particularly the left anterior descending (LAD) artery—and the duration of ischemia have been recognized as fundamental determinants of infarct size and early functional impairment [7]. Nevertheless, the association between symptom onset timing and functional outcomes may be confounded by heterogeneous pre-hospital delays, treatment-seeking behaviors, and survivor bias in real-world cohorts. Furthermore, admission hemodynamic parameters, such as systolic blood pressure (SBP), may capture the degree of immediate physiological compromise and infarct-related circulatory stress, yet they are frequently underweighted in risk models compared with angiographic variables.

In addition to anatomic and hemodynamic markers, metabolic dysregulation is increasingly recognized as an important contributor to the pathophysiology and clinical heterogeneity of acute coronary syndromes [8]. Although low-density lipoprotein cholesterol (LDL-C) remains the primary therapeutic target for secondary prevention, its prognostic value during the acute phase of acute coronary syndrome may be attenuated by the “lipid paradox,” in which lower cholesterol levels are paradoxically associated with worse outcomes, potentially reflecting acute-phase responses, systemic inflammation, or underlying frailty [9]. Consequently, composite lipid indices, such as the triglyceride-to-high-density lipoprotein cholesterol (TG/HDL-C) ratio, have emerged as clinically accessible surrogate markers of insulin resistance and atherogenic dyslipidemia [10]. However, whether such lipid ratios are associated with early post-PCI ventricular dysfunction and whether they provide useful standalone discrimination beyond conventional lipid measures remain uncertain. Accordingly, the present study aimed to identify factors independently associated with early post-PCI LVEF in a cohort of STEMI patients treated with contemporary emergency PCI. Specifically, we evaluated the associations of infarct-related artery location, admission hemodynamics, symptom-to-admission time, and lipid profiles with reduced LVEF (< 50%), thereby providing exploratory real-world evidence regarding clinical correlates of early post-PCI LVEF in routine practice.

Methods

Study design and treatments

This single-center, retrospective observational study consecutively identified patients admitted to Nanjing BenQ Hospital between January 1, 2014 and December 31, 2023. Using the electronic medical record system, we screened 1,145 patients from the emergency PCI registry at Nanjing BenQ Hospital. After exclusion of patients with NSTEMI, unstable angina, or stable angina (n = 427), 718 patients with STEMI were screened for study eligibility.

STEMI was diagnosed according to the Fourth Universal Definition of Myocardial Infarction (2018) and required typical ischemic chest pain lasting ≥ 20 min together with new ST-segment elevation in at least two contiguous leads, defined as ≥ 0.1 mV in all leads other than V2–V3, where the following cutoffs apply: ≥0.25 mV in men < 40 years, ≥ 0.20 mV in men ≥ 40 years, and ≥ 0.15 mV in women; or new-onset left bundle branch block [11]. For study eligibility, we further restricted inclusion to patients presenting within 12 h of symptom onset who underwent emergency PCI. Only patients presenting with their first documented AMI and undergoing their first PCI procedure during the index hospitalization were included. Patients with a documented history of previous AMI, prior PCI, or coronary artery bypass grafting were not included in the final analytic cohort. Accordingly, we excluded patients with symptom onset to admission > 12 h and resolved chest pain (n = 139), contraindications to PCI, refusal of PCI, or those deemed unsuitable for PCI (n = 37), severe comorbid conditions likely to affect survival (including advanced pulmonary disease, severe hepatic/renal dysfunction, or malignancy; n = 65), significant structural heart disease (congenital heart disease, rheumatic heart disease, or severe valvular disease; n = 3). After applying these clinical eligibility criteria, 474 patients with STEMI undergoing emergency PCI remained. Among them, 220 patients were further excluded because of incomplete clinical records and/or missing laboratory data, leaving a final analytic cohort of 254 patients. The patient inclusion and exclusion process is summarized in Fig. 1.

Fig. 1.

Flowchart of patient selection and group allocation according to post-PCI LVEF

Patients were categorized into a reduced LVEF group (LVEF < 50%; n = 45) and a preserved LVEF group (LVEF ≥ 50%; n = 209) according to a prespecified EF threshold. All included patients received guideline-directed emergency PCI with standard peri-procedural antithrombotic therapy and secondary prevention medications at the discretion of the treating cardiologist, consistent with contemporary practice during the study period.

Echocardiography and LVEF assessment

Left ventricular ejection fraction (LVEF) was assessed by transthoracic echocardiography within 24–72 h after PCI during the index hospitalization. LVEF was calculated using the Teichholz method in all included patients, consistent with the routine echocardiographic reporting practice during the study period. Echocardiographic measurements were reviewed by two physicians, including an experienced echocardiographer and a senior cardiologist. If multiple echocardiographic examinations were available, the earliest post-PCI LVEF measurement was used for analysis. Because this was a retrospective study based on routine clinical echocardiography reports, formal repeated measurements for calculating interobserver variability metrics, such as intraclass correlation coefficients, were not available.

Laboratory measurements

Laboratory variables were extracted from the hospital laboratory information system. Lipid profiles, inflammatory markers, and cardiac biomarkers were obtained from the first available blood sample collected on admission when available. Lipid ratios were calculated as LDL-C/HDL-C and TG/HDL-C using values expressed in mmol/L. In addition to lipid parameters, white blood cell count, C-reactive protein (CRP), pro-B-type natriuretic peptide (pro-BNP), and troponin I were evaluated. Troponin I, CRP, and pro-BNP were log-transformed using the log1p function in regression analyses because of their skewed distributions.

Ethics & consent

Ethics approval was obtained from the Ethics Committee of Nanjing BenQ Hospital (Approval No. 2026-KL020). The study was conducted in accordance with the Declaration of Helsinki. Given the retrospective design and use of de-identified data, the requirement for written informed consent was waived.

Statistical analysis

All statistical analyses were performed using R (version 4.3.1) and SPSS (version 27.0). Continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range [IQR]), as appropriate, and categorical variables as number (percentage). Between-group comparisons (reduced-LVEF vs. preserved-LVEF) were conducted using the independent-samples t test for normally distributed continuous variables or the Mann–Whitney U test for non-normally distributed variables, and the χ² test or Fisher’s exact test for categorical variables, as appropriate. To evaluate factors associated with post-PCI left ventricular ejection fraction (LVEF), multivariable linear regression was performed with LVEF (%) as a continuous outcome. Two hierarchical models were fitted: Model 1 adjusted for age, sex, diabetes mellitus, symptom-to-admission time (per 6 h), systolic blood pressure (SBP), estimated glomerular filtration rate (eGFR), and infarct-related artery (LAD vs. non-LAD); Model 2 additionally adjusted for troponin I transformed using log1p. Results are reported as β coefficients with 95% confidence intervals (CIs). To identify factors associated with reduced LVEF (< 50%), multivariable logistic regression was performed using the same covariate structure (Model 1 and Model 2), and results are presented as odds ratios (ORs) with 95% CIs. Model diagnostics included assessment of multicollinearity, influential observations, and model calibration. Calibration of the fully adjusted logistic regression model was evaluated using the Hosmer–Lemeshow goodness-of-fit test. Given the limited number of reduced-LVEF events, the events-per-variable ratio was considered when interpreting the logistic regression models. Covariates were selected a priori based on clinical relevance, and no data-driven variable selection procedure was used. The primary analyses were performed using a complete-case approach. Patients with missing key clinical, laboratory, angiographic, or echocardiographic variables required for the main analyses were excluded. Because detailed baseline data were not consistently available for all excluded patients, a reliable comparison between excluded patients and the final analytic cohort could not be performed. The discriminative performance of lipid parameters and lipid ratios for reduced LVEF (< 50%) was evaluated using receiver operating characteristic (ROC) curve analysis, and the area under the curve (AUC) with 95% CIs was reported. Additional exploratory analyses were performed to evaluate inflammatory and cardiac biomarkers, including CRP, pro-BNP, WBC count, and troponin I. Between-group comparisons were summarized as median (IQR) and compared using the Mann–Whitney U test. Exploratory logistic and linear regression analyses were performed when biomarker data were available, using reduced LVEF (< 50%) and LVEF (%) as outcomes, respectively. These models were adjusted for age, sex, diabetes mellitus, symptom-to-admission time, SBP, eGFR, LAD as the infarct-related artery, and log1p-transformed troponin I. Because CRP and pro-BNP were not available for all patients, biomarker analyses were conducted among patients with available biomarker data, and no imputation was performed. All tests were two-sided, and P < 0.05 was considered statistically significant.

Results

Baseline characteristics of the study population

A total of 254 patients were included and stratified into the reduced-LVEF group (n = 45) and the preserved-LVEF group (n = 209). Baseline clinical characteristics are summarized in Table 1. The two groups were comparable in demographic characteristics, including age, sex, and marital status, as well as in the prevalence of hypertension, diabetes, and hyperlipidemia (all P > 0.05). Regarding angiographic features, CAD complexity was comparable between the reduced- and preserved-LVEF groups. Multivessel disease was present in 202 patients (79.53%) overall, including 36 patients (80.00%) in the reduced-LVEF group and 166 patients (79.43%) in the preserved-LVEF group (P = 1.000). Left main involvement was observed in 6 patients (2.36%) and did not differ between groups (P = 1.000). The distribution of culprit vessels differed between the two groups (P = 0.003), with LAD involvement as the infarct-related artery being more frequent in the reduced-LVEF group than in the preserved-LVEF group (73.33% vs. 48.33%, P = 0.002). In the overall cohort, LAD was the most common culprit vessel category (50.00%), followed by RCA (36.61%), LCX (9.84%), multiple culprit vessels recorded (3.15%), and left main coronary artery (0.39%). Post-PCI TIMI flow data were available for 253 patients. TIMI 3 flow was achieved in nearly all patients after PCI, including 45/45 patients (100.00%) in the reduced-LVEF group and 207/208 patients (99.52%) in the preserved-LVEF group. TIMI 0–2 flow was observed in one patient in the preserved-LVEF group and in none of the reduced-LVEF patients (P = 1.000). In terms of vital signs and laboratory measurements, patients in the reduced-LVEF group had lower systolic blood pressure (122.80 ± 20.15 vs. 129.54 ± 19.87 mmHg, P = 0.016) and higher WBC count (12.00 ± 3.37 vs. 10.69 ± 3.23 × 10^9/L, P = 0.019). Aspartate aminotransferase (AST) levels were also higher in the reduced-LVEF group (62.32 ± 83.82 vs. 39.47 ± 46.44 U/L, P = 0.021). Troponin I differed significantly between groups (P = 0.026), with higher values observed in the reduced-LVEF group. Triglyceride levels were lower in the reduced-LVEF group (P = 0.043), whereas other lipid parameters (total cholesterol, LDL-C, HDL-C, ApoA1, and ApoB) did not differ significantly between groups.

Table 1.

Baseline clinical characteristics of the study population according to post-PCI LVEF

Variable	Total (N = 254)	Reduced LVEF group (LVEF < 50%) (N = 45)	Preserved LVEF group (LVEF ≥ 50%) (N = 209)	P-value
Age, years	58.67 ± 13.45	61.91 ± 15.19	57.97 ± 12.98	0.123
Sex, n (%)				0.115
Male	212 (83.46)	34 (75.56)	178 (85.17)
Female	42 (16.54)	11 (24.44)	31 (14.83)
Marital status (%)				0.787
Married	235 (92.52)	43 (95.56)	192 (91.87)
Divorced	6 (2.36)	0 (0.00)	6 (2.87)
Never married	13 (5.12)	2 (4.44)	11 (5.26)
Hypertension, n (%)				0.965
yes	146 (57.48)	26 (57.78)	120 (57.42)
no	108 (42.52)	19 (42.22)	89 (42.58)
Diabetes, n (%)				0.089
yes	55 (21.65)	14 (31.11)	41 (19.62)
no	199 (78.35)	31 (68.89)	168 (80.38)
Hyperlipidemia, n (%)				0.671
yes	89 (35.04)	17 (37.78)	72 (34.45)
no	165 (64.96)	28 (62.22)	137 (65.55)
Smoking, n (%)				0.059
yes	145 (57.09)	20 (44.44)	125 (59.81)
no	109 (42.91)	25 (55.56)	84 (40.19)
Drinking, n (%)				0.105
yes	21 (8.27)	1 (2.22)	20 (9.57)
no	233 (91.73)	44 (97.78)	189 (90.43)
Infarct-related artery (LAD), n (%)				0.002
yes	134 (52.76)	33 (73.33)	101 (48.33)
no	120 (47.24)	12 (26.67)	108 (51.68)
CAD complexity, n (%)				1.000
Single-vessel disease	52 (20.47)	9 (20.00)	43 (20.57)
Multivessel disease	202 (79.53)	36 (80.00)	166 (79.43)
Left main involvement, n (%)				1.000
yes	6 (2.36)	1 (2.22)	5 (2.39)
no	248 (97.64)	44 (97.78)	204 (97.61)
Culprit vessel distribution, n (%)				0.003
LAD	127 (50.00)	30 (66.67)	97 (46.41)
RCA	93 (36.61)	8 (17.78)	85 (40.67)
LCX	25 (9.84)	3 (6.67)	22 (10.53)
Left main coronary artery	1 (0.39)	1 (2.22)	0 (0.00)
Multiple culprit vessels recorded	8 (3.15)	3 (6.67)	5 (2.39)
Post-PCI TIMI flow, n (%)				1.000
TIMI 3	252 (99.60)	45 (100.00)	207 (99.52)
TIMI 0–2	1 (0.40)	0 (0.00)	1 (0.48)
SBP, mmHg	128.35 ± 20.05	122.80 ± 20.15	129.54 ± 19.87	0.016
DBP, mmHg	78.66 ± 14.19	77.84 ± 12.24	78.84 ± 14.60	0.685
WBC count, 109/L	10.92 ± 3.29	12.00 ± 3.37	10.69 ± 3.23	0.019
Troponin I, ng/mL	0.50 (0.07, 0.55)	0.50 (0.10, 5.10)	0.50 (0.06, 0.50)	0.026
Scr, µmol/L	85.07 ± 32.02	89.76 ± 52.91	84.06 ± 25.48	0.506
TC, mmol/L	4.66 ± 1.26	4.59 ± 1.26	4.67 ± 1.26	0.533
TG, mmol/L	2.30 ± 2.62	1.76 ± 0.96	2.42 ± 2.84	0.043
LDL-C, mmol/L	3.13 ± 0.98	3.25 ± 1.23	3.11 ± 0.92	0.873
HDL-C, mmol/L	1.05 ± 0.27	1.12 ± 0.24	1.04 ± 0.28	0.051
ApoA1, g/L	1.24 ± 0.65	1.20 ± 0.24	1.25 ± 0.71	0.979
ApoB, g/L	1.39 ± 6.15	1.00 ± 0.34	1.48 ± 6.78	0.282
ALT, U/L	31.77 ± 28.72	30.31 ± 17.09	32.09 ± 30.68	0.654
AST, U/L	43.52 ± 55.42	62.32 ± 83.82	39.47 ± 46.44	0.021

Annotation: Data are presented as mean ± SD, median (IQR), or n (%)

Multivariable analysis of factors associated with post-PCI LVEF

Table 2 summarizes the multivariable linear regression results for factors associated with post-PCI LVEF. Two models were fitted to evaluate variables independently associated with LVEF. In Model 1, LAD as the infarct-related artery (IRA) was significantly associated with lower LVEF (β = −5.500; 95% CI, − 7.666 to − 3.334; P < 0.001). Other clinical variables, including age, sex, diabetes, systolic blood pressure (SBP), and eGFR, were not significantly associated with LVEF. In Model 2, after additional adjustment for troponin I, LAD-IRA remained a strong negative correlate of LVEF (β = −5.240; 95% CI, − 7.414 to − 3.066; P < 0.001). Notably, symptom-to-admission time (per 6-hour increment) was positively associated with LVEF in the fully adjusted model (β = 1.794; 95% CI, 0.203 to 3.384; P = 0.027), whereas it showed only a borderline association in Model 1 (P = 0.056). Troponin I was not statistically significant (P = 0.068).

Table 2.

Multivariable linear regression of post-PCI LVEF

Variables	Model 1		Model 2
	β (95% CI)	P-value	β (95% CI)	P-value
Age, years	-0.038 (-0.148, 0.073)	0.500	-0.033 (-0.144, 0.077)	0.550
Sex, Female	-0.879 (-3.986, 2.227)	0.578	-0.860 (-3.952, 2.231)	0.584
Diabetes	-1.343 (-3.753, 1.068)	0.274	-1.278 (-3.678, 1.122)	0.295
Symptom-to-admission time, per 6 h	1.531 (-0.041, 3.103)	0.056	1.794 (0.203, 3.384)	0.027
SBP, mmHg	0.027 (-0.027, 0.081)	0.322	0.027 (-0.026, 0.081)	0.317
eGFR, mL/min/1.73 m2	0.029 (-0.039, 0.096)	0.401	0.025 (-0.042, 0.093)	0.460
Infarct related artery (LAD)	-5.500 (-7.666, -3.334)	< 0.001	-5.240 (-7.414, -3.066)	< 0.001
Troponin I (log1p)	—	—	-1.004 (-2.083, 0.076)	0.068

Abbreviations: CI Confidence interval, DM Diabetes mellitus, eGFR estimated glomerular filtration rate, LAD Left anterior descending artery, LVEF Left ventricular ejection fraction, SBP Systolic blood pressure

Models: Model 1 adjusted for age, sex, DM, symptom-to-admission time (per 6 h), SBP, eGFR, and infarct-related artery (LAD vs. non-LAD); Model 2 additionally adjusted for troponin I (log1p)

 “—” indicates not included. β indicates the estimated change in LVEF (%) per unit increase in the corresponding variable. P < 0.05 was considered statistically significant

Factors associated with reduced post-PCI LVEF (< 50%)

Table 3 summarizes the multivariable logistic regression results for factors independently associated with reduced LVEF (< 50%). In Model 1, LAD as the infarct-related artery (IRA) was strongly associated with higher odds of reduced LVEF (OR, 4.243; 95% CI, 2.025 to 9.481; P < 0.001). In contrast, higher systolic blood pressure (SBP) was associated with lower odds of reduced LVEF (OR, 0.980; 95% CI, 0.961 to 0.998; P = 0.031). In the fully adjusted Model 2, which additionally included troponin I, LAD-IRA remained significantly associated with reduced LVEF (OR, 4.005; 95% CI, 1.901 to 8.977; P < 0.001). Symptom-to-admission time (per 6-hour increment) was also associated with lower odds of reduced LVEF (OR, 0.357; 95% CI, 0.118 to 0.791; P = 0.036), whereas it showed only a borderline association in Model 1 (P = 0.055). SBP remained statistically significant in Model 2 (P = 0.026), while age, sex, diabetes, eGFR, and troponin I were not significantly associated with reduced LVEF.

Table 3.

Factors associated with post-PCI reduced LVEF (< 50%): multivariable logistic regression

Variables	Model 1		Model 2
	OR (95% CI)	P-value	OR (95% CI)	P-value
Age, years	1.013 (0.978, 1.050)	0.472	1.012 (0.977, 1.049)	0.493
Sex, Female	1.636 (0.643, 4.024)	0.289	1.573 (0.606, 3.927)	0.338
Diabetes	1.520 (0.699, 3.228)	0.280	1.471 (0.669, 3.146)	0.326
Symptom-to-admission time, per 6 h	0.429 (0.157, 0.878)	0.055	0.357 (0.118, 0.791)	0.036
SBP, mmHg	0.980 (0.961, 0.998)	0.031	0.979 (0.960, 0.997)	0.026
eGFR, mL/min/1.73 m2	0.995 (0.975, 1.017)	0.662	0.997 (0.976, 1.019)	0.807
Infarct related artery (LAD)	4.243 (2.025, 9.481)	< 0.001	4.005 (1.901, 8.977)	< 0.001
Troponin I (log1p)	—	—	1.319 (0.942, 1.814)	0.094

Abbreviations: CI Confidence interval, DM Diabetes mellitus, eGFR estimated glomerular filtration rate, LAD Left anterior descending artery, LVEF Left ventricular ejection fraction, OR Odds ratio, SBP systolic blood pressure

Models: Model 1 adjusted for age, sex, DM, symptom-to-admission time (per 6 h), SBP, eGFR, and culprit vessel (LAD vs. non-LAD); Model 2 additionally adjusted for troponin I (log1p)

“—” indicates not included. P < 0.05 was considered statistically significant

Model calibration

For the fully adjusted logistic regression model, the Hosmer–Lemeshow goodness-of-fit test showed no evidence of poor calibration (χ² = 5.335, df = 8, P = 0.721).

Additional exploratory analyses of inflammatory and cardiac biomarkers

Additional inflammatory and cardiac biomarkers are summarized in Supplementary Table 1. CRP was available in 188 patients and pro-BNP in 167 patients. Patients with reduced LVEF had higher CRP levels than those with preserved LVEF (5.00 [1.92, 9.60] vs. 2.30 [1.00, 6.00], P = 0.021) and higher pro-BNP levels (1586.00 [587.55, 4183.10] vs. 513.80 [235.00, 1265.10], P < 0.001). WBC count and troponin I were also higher in the reduced-LVEF group. In exploratory sensitivity analyses adjusted for age, sex, diabetes mellitus, symptom-to-admission time, SBP, eGFR, LAD as the infarct-related artery, and log1p-transformed troponin I, log1p(pro-BNP) and WBC count were associated with higher odds of reduced LVEF, whereas log1p(CRP) was not independently associated with reduced LVEF in the logistic model. In linear regression models using LVEF as a continuous outcome, log1p(CRP), log1p(pro-BNP), and WBC count were inversely associated with LVEF (Supplementary Table 2).

Sensitivity analysis for symptom-to-admission time

To further evaluate the association between symptom-to-admission time and early post-PCI LVEF, we performed a sensitivity analysis using categorized symptom-to-admission time (≤ 3 h, > 3–6 h, and > 6–12 h). The proportions of reduced LVEF were 18.44%, 20.41%, and 12.50%, respectively, with no significant difference across groups (P = 0.841). LVEF as a continuous outcome also did not differ significantly across these categories (P = 0.070). In adjusted logistic and linear regression models using the ≤ 3 h group as the reference, the associations for > 3–6 h and > 6–12 h were not statistically significant. These findings suggest that the association between symptom-to-admission time and early LVEF should be interpreted cautiously (Supplementary Table 3).

Receiver operating characteristic curves of lipid markers for identifying post-PCI reduced LVEF (< 50%)

Figure 2 shows the receiver operating characteristic (ROC) curves of lipid markers for identifying post-PCI reduced LVEF (< 50%). Among the evaluated indices, the TG/HDL-C ratio demonstrated the numerically highest, but still modest discriminative performance (AUC, 0.605; 95% CI, 0.515–0.694). The optimal cut-off value for the TG/HDL-C ratio was 1.665, yielding a sensitivity of 71.1% and a specificity of 50.7%. TG and HDL-C also exhibited modest discrimination, with AUCs of 0.596 (95% CI, 0.506–0.686) and 0.593 (95% CI, 0.505–0.681), respectively. In contrast, LDL-C showed no meaningful discriminative ability (AUC, 0.492; 95% CI, 0.394–0.591), consistent with performance close to chance. Given its AUC of only 0.605, TG/HDL-C should be interpreted as an exploratory lipid index rather than a standalone marker for clinical risk prediction.

Fig. 2.

Receiver operating characteristic (ROC) curves of lipid parameters and lipid ratios for identifying reduced LVEF (< 50%)

Discussion

In this single-center retrospective study of patients with STEMI undergoing primary PCI, we identified a pattern of anatomic and hemodynamic factors consistently associated with early left ventricular systolic dysfunction, alongside exploratory differences in the discriminative performance of lipid markers. Our principal findings are threefold. First, consistent with the concept of myocardium at risk, an infarct-related artery in the left anterior descending territory was the most consistently associated clinical factor for impaired left ventricular systolic function in our cohort, which is biologically plausible given its association with a larger area of jeopardized myocardium, greater infarct size, and more pronounced ventricular dysfunction and remodeling after STEMI [12, 13]. Second, lower admission systolic blood pressure (SBP) was also associated with reduced LVEF, providing clinically interpretable hemodynamic context beyond angiographic anatomy alone. Third, although LDL-C showed limited discriminative utility in our cohort—consistent with the acute-phase “lipid paradox” described in acute myocardial infarction and acute coronary syndrome—the TG/HDL-C ratio yielded the numerically highest AUC among the evaluated lipid indices, although its overall discriminative performance remained modest [14–16]. Notably, our data also revealed an inverse association between shorter symptom-to-admission times and lower LVEF, which likely reflects a specific high-risk clinical phenotype and selection processes in real-world cohorts rather than any protective effect of delayed presentation.

The association between LAD culprit territory and lower early post-PCI LVEF is consistent with established evidence linking anterior infarction to a larger myocardium-at-risk and greater early systolic impairment [12, 17]. Occlusion of the LAD typically compromises the anterior wall and substantial portions of the interventricular septum, thereby involving a larger mass of myocardium than many non-LAD occlusions. Our results also corroborate prior registry and cohort data indicating that anterior infarction is associated with a higher risk of heart failure and adverse ventricular remodeling, even in the contemporary era of timely reperfusion [13, 18]. The persistence of this association in our multivariable models underscores that, despite successful epicardial flow restoration, the initial anatomic burden remains a primary determinant of early functional recovery. Collectively, these findings support the clinical relevance of culprit-vessel territory as a contextual factor when interpreting early post-PCI LVEF, rather than establishing a novel risk marker.

Hemodynamic status at presentation, particularly SBP, provided complementary risk information. Although admission SBP can be influenced by multiple acute-phase factors, lower SBP in our cohort remained independently associated with reduced LVEF [19]. This pattern likely reflects a continuum of the Killip classification, whereby lower systemic pressure signals early pump failure, impaired forward flow, and reduced stroke volume [20]. While composite shock indices are frequently used in emergency and critical care settings, admission SBP may provide simple contextual information regarding early hemodynamic compromise, although it should not be interpreted as a standalone predictor of post-PCI LVEF [21, 22]. Importantly, SBP remained significant even after adjustment for the culprit vessel territory, implying that it may capture downstream hemodynamic consequences—potentially related to microvascular obstruction, impaired reperfusion, or limited collateralization—that are not fully explained by infarct location alone.

A notable observation in our analysis was the “paradoxical” relationship between symptom-to-admission time and LVEF, whereby shorter times were associated with worse functional outcomes. This finding diverges from the conventional “time is muscle” paradigm, although the relationship between prehospital delay and clinical outcomes may be nonlinear and confounded by symptom severity and care-seeking behavior in retrospective analyses [23, 24]. Importantly, this finding should not be interpreted as evidence that delayed presentation is beneficial. In additional sensitivity analyses using categorized symptom-to-admission time, the proportions of reduced LVEF and LVEF as a continuous outcome did not differ significantly across time categories, suggesting that the observed association in the main model was not robust to alternative categorization. Therefore, this result should be regarded as exploratory and hypothesis generating [25]. Patients with more severe ischemic presentations may seek medical attention earlier and are more likely to activate emergency medical services, whereas complete coronary occlusion is associated with more extensive myocardial injury and a higher likelihood of early hemodynamic compromise [26]. Conversely, later presenters may represent a survivor cohort characterized by stuttering ischemia (potentially inducing ischemic preconditioning) or better-developed collateral circulation that preserves viability despite prolonged ischemic intervals. Prospective studies with standardized reperfusion timing variables, including symptom-to-wire time and door-to-balloon time, are needed to clarify this relationship.

Regarding metabolic profiling, our study highlights the limited discriminatory value of standard lipid measures in the acute setting. The near-chance performance of LDL-C in our cohort (AUC 0.492) is consistent with the “lipid paradox” described in AMI, in which lower LDL-C levels may paradoxically be associated with worse outcomes. This pattern may partly reflect acute-phase changes in lipid levels and inflammation-related alterations in lipoprotein metabolism, rather than lower baseline atherogenic risk alone [27, 28]. In contrast, the TG/HDL-C ratio showed numerically higher but still modest discriminative ability in our analysis. Elevated TG/HDL-C has been proposed as a clinically accessible surrogate marker of insulin resistance and an atherogenic lipoprotein phenotype characterized by smaller, denser LDL particles, while also reflecting an adverse metabolic–inflammatory milieu [29]. However, the absolute discriminative performance of TG/HDL-C was modest (AUC 0.605), and this level of discrimination is insufficient for standalone clinical risk prediction or decision-making. Therefore, our findings should be interpreted as exploratory and hypothesis-generating. TG/HDL-C may reflect an adverse metabolic background, but its incremental value beyond established clinical, angiographic, and hemodynamic factors remains uncertain and requires validation in larger prospective cohorts.

Several limitations merit consideration. First, the retrospective, single-center design introduces potential selection bias and limits generalizability to other populations and practice settings. In addition, the complete-case approach may have introduced further selection bias because patients with incomplete clinical records or missing key data were excluded. Because detailed baseline information was not consistently available for all excluded patients, we could not reliably compare excluded patients with the final analytic cohort. Second, although we adjusted for key clinical variables, we lacked direct measures of infarct size such as cardiac magnetic resonance (CMR) quantification or comprehensive serial biomarker profiles, which would provide more granular correlates of myocardial necrosis and tissue-level injury. Third, LVEF was assessed using two-dimensional echocardiography obtained from routine clinical echocardiography reports. Although this approach is clinically standard, it is less sensitive than global longitudinal strain (GLS) for detecting subtle subclinical dysfunction and may be influenced by loading conditions and image quality. Although echocardiographic measurements were reviewed by two physicians, formal interobserver variability metrics, such as intraclass correlation coefficients, were not available because of the retrospective study design. In addition, LVEF was calculated using the Teichholz method in all patients, which may be less accurate in the presence of regional wall motion abnormalities; therefore, measurement variability could not be fully excluded. Fourth, only 45 patients had reduced LVEF, resulting in a relatively low events-per-variable ratio in the fully adjusted logistic regression model. Although covariates were prespecified and selected based on clinical relevance, the limited number of events may have affected model stability, increased the risk of overfitting, and reduced the precision of effect estimates. Therefore, the logistic regression findings should be interpreted cautiously and require validation in larger cohorts. Although model calibration was additionally assessed using the Hosmer–Lemeshow test and showed no evidence of poor calibration, the small number of reduced-LVEF events limited the robustness of model performance evaluation. In addition, CRP and pro-BNP were not routinely measured in all patients because of the retrospective design; therefore, analyses involving these biomarkers were exploratory and may be affected by missingness and reduced statistical power. Finally, although symptom-to-admission time was included in the multivariable models and post-PCI TIMI flow was additionally summarized, several important procedural and angiographic determinants of post-MI ventricular function were not consistently available in the electronic records. These included door-to-balloon time, symptom-to-wire time, myocardial blush grade, and complete versus incomplete revascularization. In particular, the interpretation of the association between symptom-to-admission time and early LVEF is limited by the lack of standardized symptom-to-wire and door-to-balloon times; moreover, this association was attenuated and no longer statistically significant when symptom-to-admission time was analyzed categorically. The absence of these variables may have introduced residual confounding. Given the limited number of reduced-LVEF events, we did not further expand the primary logistic regression model to avoid additional model instability and overfitting. Future prospective studies incorporating strain imaging, standardized reperfusion timing metrics, and more comprehensive metabolic phenotyping are warranted to determine whether TG/HDL-C provides any incremental value beyond established clinical, angiographic, and hemodynamic factors in STEMI.

Conclusion

In conclusion, in patients with STEMI treated with emergency PCI, LAD culprit territory and lower admission systolic blood pressure were the main clinical correlates of early post-PCI reduced LVEF. TG/HDL-C showed only modest standalone discriminative ability, whereas LDL-C had little utility in this setting. Given its limited AUC, TG/HDL-C should not be interpreted as a standalone clinical prediction tool. These findings should be considered exploratory and hypothesis-generating, and external validation with more comprehensive clinical, angiographic, and hemodynamic data is needed before any clinical application can be considered.

Abbreviations

ACS

Acute coronary syndromes

AMI

Acute myocardial infarction

ApoA1

Apolipoprotein A1

ApoB

Apolipoprotein B

AST

Aspartate aminotransferase

AUC

Area under the curve

CI

Confidence interval

CRP

C-reactive protein

CMR

Cardiac magnetic resonance

eGFR

Estimated glomerular filtration rate

EF

Ejection fraction

GDMT

Guideline-directed medical therapy

GLS

Global longitudinal strain

HDL-C

High-density lipoprotein cholesterol

IQR

Interquartile range

IRA

Infarct-related artery

LAD

Left anterior descending coronary artery

LBBB

Left bundle branch block

LDL-C

Low-density lipoprotein cholesterol

LV

Left ventricular

LVEF

Left ventricular ejection fraction

NSTEMI

Non–ST-segment elevation myocardial infarction

OR

Odds ratio

pro-BNP

pro-B-type natriuretic peptide

PCI

Percutaneous coronary intervention

ROC

Receiver operating characteristic

SBP

Systolic blood pressure

SD

Standard deviation

STEMI

ST-segment elevation myocardial infarction

TG

Triglycerides

TG/HDL-C

Triglyceride-to–high-density lipoprotein cholesterol ratio

TIMI

Thrombolysis In Myocardial Infarction

Authors’ contributions

H.T.: Writing − review & editing, Writing − original draft, Resources, Project administration, Formal analysis, Data curation, Conceptualization. W.L. , L.W. and X.H.: Writing – review & editing, Writing − original draft, Validation, Supervision, Project administration, Methodology. X.L. , Z.L. , L.W. and S.D. : Writing − review & editing, Validation, Investigation, Formal analysis. All authors read and approved the final paper.

Funding

Not applicable.

Data availability

The datasets generated and/or analysed during the current study are not publicly available because they contain information that could compromise the privacy of research participants, but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Nanjing BenQ Hospital (Approval No. 2026-KL020) and was conducted in accordance with the Declaration of Helsinki. Owing to the retrospective study design, the use of de-identified data, and the minimal risk to participants, the requirement for individual informed consent was waived by the ethics committee. All data were anonymized before analysis and handled in accordance with institutional and national data-protection regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.