Global myocardial work efficiency provides incremental prognostic value in STEMI patients undergoing PCI

Over two years, 225 consecutive STEMI patients admitted to our tertiary center in Timişoara, Romania, were prospectively enrolled in the study. All underwent primary PCI within 12 hours of symptom onset, followed by comprehensive transthoracic echocardiography (TTE). Exclusion criteria included significant valvular disease; fixed or dynamic left ventricular outflow tract obstruction; atrial fibrillation or marked R–R variability; permanent pacemaker; life expectancy < 1 year due to comorbidities; or poor echocardiographic quality. The final cohort comprised 215 patients.

Patients received guideline-directed therapy [20] and were followed through a structured outpatient program. MAE was defined as a composite of unplanned cardiovascular hospitalization, malignant ventricular arrhythmias, stroke, and/or all-cause mortality. Demographics, clinical parameters, cardiovascular risk factors, and laboratory data — including lipid profile, creatinine, glucose, hemogram, erythrocyte sedimentation rate, and peak cardiac biomarkers — were systematically recorded. The study was approved by institutional and university ethics committees (approval no. 10163/05.12.2023 and 07/31.01.2024), with informed consent obtained from all participants.

TTE was performed according to guideline recommendations [21] using a Vivid 9 system (GE Healthcare) within 4 days of admission (interquartile range [IQR] 3–5). Imaging was obtained at rest in the left lateral decubitus position with electrocardiogram gating, and brachial systolic blood pressure was measured immediately before echocardiographic acquisition with the patient at rest; these values were then used to estimate LV pressure for MW calculations. Pulsed-wave tissue Doppler imaging measured septal and lateral mitral annular velocities (S’, E’, and A’) over five cardiac cycles, averaged for analysis. Speckle-tracking echocardiography in standard apical views was used to derive GLS, excluding poorly tracked segments. Mechanical dispersion (MD) was calculated as the standard deviation of time to peak longitudinal strain across 17 LV segments, providing a measure of regional mechanical heterogeneity. MD was analyzed as an exploratory secondary parameter.

Pressure-strain loops integrated GLS with brachial systolic pressure. Valve timings were determined from continuous-wave Doppler and visually confirmed. Derived indices included global work index (GWI), global constructive work (GCW), global wasted work (GWW), and global work efficiency (GWE = GCW / [GCW + GWW]) [16,22,23]. Analyses were performed independently by two blinded observers.

Data distribution was assessed via Shapiro-Wilk test. Categorical variables were reported as counts and percentages; continuous variables as mean ± SD or median (IQR). Comparisons used Chi-square/Fisher’s exact or t-test/Mann-Whitney U, as appropriate. Univariable and multivariable Cox regression identified MAE predictors, with the covariate number limited to ≥ 10 events per variable to avoid overfitting. Multicollinearity was evaluated via variance inflation factors (VIF). Sequential models assessed the incremental prognostic value of GWE, comparing baseline clinical data (age, Killip class, estimated glomerular filtration rate [eGFR]), LVEF, and GWE using –2 log-likelihood, likelihood ratio X², and Akaike Information Criterion (AIC). Intra-observer variability (assessed by repeat measurements performed by the same observer at two different time points and interobserver reproducibility (evaluated by comparing measurements obtained by two independent observers in the same patients) were quantified using intraclass correlation coefficients (ICCs) with 95% confidence intervals. A two-sided p < 0.05 indicated significance.

The study population consisted of 215 patients (78.1% male) with a mean age of 61 ± 10 years. At the time of hospital admission, mean heart rate was 80 ± 16 bpm, systolic blood pressure 139 ± 22 mmHg, and diastolic blood pressure 86 ± 14 mmHg. The median BMI was 27.1 kg/m² (IQR 24.2–30.9).

Laboratory analyses revealed a mean peak creatine-kinase isoform MB (CK-MB) level of 289.6 ± 211 U/L and a median peak troponin I concentration of 3,351 ng/L (IQR 314–21,506). Mean estimated glomerular filtration rate (eGFR) rate was 74.1 ± 21.2 mL/min/1.73 m².

Cardiovascular risk factors were highly prevalent: 73% of patients were current or former smokers, 97.2% had dyslipidemia, 21.9% had diabetes mellitus, and 71.6% had hypertension. A history of coronary artery disease and a positive cardiovascular family history were each present in 6.5% of patients, while chronic kidney disease was documented in 10.2%. At presentation, 82.3% of patients were classified as Killip Class I, 14% as Class II, 1.4% as Class III, and 2.3% as Class IV.

Pre-hospital thrombolytic therapy was administered in 26% of cases. Coronary angiography showed single-, two-, and three-vessel disease in 41.9%, 34.4%, and 23.7% of the patients, respectively.

At discharge, nearly all patients received dual antiplatelet therapy and statins. Additional treatments included beta-blockers (78.6%), mineralocorticoid receptor antagonists (79.1%), angiotensin-converting enzyme inhibitors (58.1%), nitrates (42.8%), loop diuretics (74%), sodium-glucose cotransporter-2 inhibitors (20.5%), and angiotensin receptor–neprilysin inhibitors (3.3%). Only 0.5% of patients received angiotensin receptor blockers.

During a median follow-up of 7 months (IQR 5–15), 40 patients experienced MAE, including all-cause mortality (n = 5), unplanned cardiovascular hospitalization for heart failure or acute coronary syndrome (n = 30), ventricular arrhythmias (n = 3), and stroke (n = 2). Baseline characteristics of the study population are summarized in Table 1.

When stratified by the occurrence of MAE, significant differences were observed in selected clinical and laboratory parameters. Patients who experienced MAE had significantly higher peak CK-MB levels (353.3 ± 291.5 vs. 244.1 ± 189.7 U/L, p = 0.004) and a trend toward higher peak troponin I concentrations (median 13,404 [IQR 1,434–38,967] vs 1,866 ng/L [IQR 218–18,773], p = 0.057).

Regarding clinical presentation and coronary anatomy, fewer patients with MAE were classified as Killip Class I at admission (70% vs 85.1%, p = 0.024). Single-vessel coronary artery disease was less frequent in patients who developed MAE (22.5% vs 46.3%, p = 0.006), whereas three-vessel disease was more prevalent (40% vs 20.0%, p = 0.007). Anterior STEMI was present in 16 patients (40.0%) in the MAE group and 75 patients (42.9%) in the event-free group, with no statistically significant difference between groups (p = 0.741), indicating a balanced distribution of infarct location. These findings suggest that, in this cohort, infarct location alone did not significantly influence the risk of MAE, and that adverse outcomes were more closely related to infarct severity, extent of coronary artery disease, and early hemodynamic status, rather than anatomical localization.

Comprehensive echocardiographic evaluation was performed in all patients (Table 2). Compared with event-free patients, those who developed MAE exhibited a trend toward higher LV end-diastolic volume and lower LVEF, GLS, GWI, and GCW; however, these differences did not reach statistical significance. In contrast, GWE was significantly reduced in patients with MAE (83 ± 8% vs. 86 ± 7%, p = 0.022). Figure 1 illustrates MW analysis in a representative patient with anterior STEMI, demonstrating reduced GWI, GCW, and GWE, increased GWW, and impaired GLS (Figure 1).

Figure 1

Myocardial work analysis in a patient with acute anterior myocardial infarction, showing reduced myocardial work indices (GWI, GCW, GWE), elevated GWW and impairment of GLS. Abbreviations: GLS, global longitudinal strain; GWI, global work index; GCW, global constructive work; GWE, global work efficiency; BP, blood pressure.

In univariable Cox regression analysis, factors significantly associated with MAE included peak CK–MB levels (hazard ratio [HR] 1.001, 95% confidence interval [CI] 1.000–1.003; p = 0.008), hypertension (HR 0.499, 95% CI 0.265–0.941; p = 0.03), global work index (GWI; HR 0.999, 95% CI 0.998–1.000; p = 0.04), global constructive work (GCW; HR 0.999, 95% CI 0.998–1.000; p = 0.03), and global work efficiency (GWE; HR 0.811, 95% CI 0.688–0.955; p = 0.01). GLS and MD showed borderline associations with outcomes (GLS, p = 0.06; MD, p = 0.05).

In multivariable analysis adjusted for clinically relevant covariates, GWE remained independently associated with MAE. Specifically, each 5% decrease in GWE was associated with a 16.5% increase in the risk of adverse events (HR 0.835, 95% CI 0.693–0.985; p = 0.033) (Table 3).

The incremental prognostic value of GWE was assessed using likelihood ratio testing of nested Cox proportional hazards models. Covariates were selected based on clinical relevance and prior evidence, with their number restricted to minimize overfitting relative to the number of observed events. Multicollinearity was evaluated using VIF, with all values < 5, indicating acceptable independence among predictors. LVEF was included as the conventional echocardiographic measure of systolic function. GLS was not added due to its strong correlation with GWE (r = 0.762, p < 0.001), which could introduce multicollinearity and overfitting.

The baseline clinical model — including age, Killip class, and eGFR — yielded a –2 log-likelihood (-2LL) value of 390.809. The addition of LVEF resulted in a negligible reduction in –2LL (390.771), corresponding to a non-significant improvement in model fit (likelihood ratio X²= 0.03, df = 1, p = 0.85). In contrast, inclusion of GWE led to a substantial decrease in –2LL to 385.870, reflecting a significant enhancement in model performance (likelihood ratio X² = 4.90, df = 1, p = 0.02). These findings demonstrate that GWE provides independent and incremental prognostic information beyond conventional clinical and echocardiographic parameters (Figure 2).

Figure 2

Incremental prognostic value of GWE for MAE. The bar chart shows –2 log likelihood for three sequential Cox models: the baseline clinical model (age, Killip class and eGFR), the model adding LVEF, and the model further including GWE. Data labels indicate the Akaike information criterion for each model. Addition of LVEF resulted in a minimal and non-significant improvement in model fit (likelihood ratio X²= 0.03, df =1, P = 0.85). Further inclusion of GWE produced a substantial reduction in –2LL, significantly improving model fit (likelihood ratio X² = 4.90, df = 1, p = 0.02). GWE, global work efficiency; –2LL, –2 log likelihood; LR, likelihood ratio; eGFR, estimated glomerullar filtration rate; LVEF, left ventricular ejection fraction; AIC, akaike information criterion.

Model comparison using the AIC supported these results. The baseline clinical model had an AIC of 396.8, which increased slightly after adding LVEF (398.7), indicating no improvement in model quality. In contrast, incorporation of GWE reduced the AIC to 395.8, confirming improved model performance while maintaining parsimony (Figure 2).

Intra-observer reproducibility was excellent for all MW parameters, as reflected by high ICCs: 0.988 (95% CI: 0.9690.995) for GWI, 0.987 (95% CI: 0.966–0.995) for GCW, 0.975 (95% CI: 0.936–0.990) for GWW, and 0.979 (95% CI: 0.946–0.992) for GWE. Likewise, interobserver agreement was consistently strong across myocardial work indices, with ICCs of 0.969 (95% CI: 0.923–0.987) for GWI, 0.963 (95% CI: 0.909–0.985) for GCW, 0.970 (95% CI: 0.928–0.988) for GWW, and 0.967 (95% CI: 0.921–0.987) for GWE.

Several limitations should be acknowledged. This was a single-center study with a moderate sample size, which may limit its generalizability. In addition, patients were enrolled only if they underwent PCI within 12 hours of symptom onset and had echocardiographic images of sufficient quality. This could have potentially led to excluding higher-risk patients, which may limit the applicability of our findings to the broader STEMI population. Furthermore, exclusion of individuals with severe valvular disease, LV outflow tract obstruction, pronounced R-R variability on electrocardiogram (including atrial fibrillation), and permanent pacemakers may reduce the applicability of our findings to such populations. Serial assessment of MW was not performed, precluding evaluation of temporal changes in myocardial efficiency. Additionally, the relatively short follow-up interval may have limited the detection of hard clinical endpoints such as mortality. Finally, speckle-tracking analysis remains operator-dependent, although standardized acquisition and blinded analysis were used to minimize variability.

Future multi-center studies with larger cohorts and longer followup periods are warranted to validate these findings and to explore the role of serial MW assessment in guiding therapy. Integration of MW indices into routine clinical workflows may ultimately enhance personalized risk stratification and improve outcomes in patients with STEMI.