The study recruited patients presenting with STEMI to the emergency departments of The First Affiliated Hospital of Bengbu Medical University and Suzhou First People's Hospital between January 1, 2021, and July 1, 2023, who underwent PCI. The inclusion criteria were as follows: (1) age 18 years or older, (2) a definitive diagnosis of STEMI, (3) successful PCI performed during the acute phase. Successful PCI was defined by a significant reduction in coronary stenosis and the attainment of thrombolysis in myocardial infarction (TIMI) grade 3 flow, indicating complete reperfusion, as confirmed by angiography. (4) Patients with single-vessel infarct-related STEMI only. The exclusion criteria comprised: (1) patients with a diagnosis of STEMI who declined further treatment or were unable to maintain follow-up due to individual circumstances; (2) CAG images that did not meet the required analytical standards, such as being blurry, having irregular data formats, incomplete imaging, significant vascular overlap or artifacts, or being single-view post-PCI; (3) patients with STEMI involving multiple infarct-related arteries; (4) patients who had experienced recent major bleeding events or had a predisposition to bleeding. The study protocol was approved by the ethics committees of The First Affiliated Hospital of Bengbu Medical University (Approval No.: KY046) and Suzhou First People's Hospital (Approval No.: SZYYLLky2024016), and all participants provided informed consent. This study complied with the Declaration of Helsinki and pertinent medical ethical guidelines, guaranteeing the voluntary participation of patients.

After reperfusion of the culprit vessel, single-view QFR and AMR analyses, guided by Murray's law, were performed using the QFR software (AngioPlus Gallery, Pulse Medical Technology Inc., Shanghai, China). The process is fully automated, allowing the software to accurately identify optimal arterial lumen contours. Manual adjustments were made only in cases where the automatic identification was suboptimal, ensuring minimal vessel overlap and optimal image clarity. The hyperemic flow velocity was calculated by dividing the vessel centerline length by the time required for the contrast agent to completely fill the vessel. The analytical framework employs comprehensive contrast enhancement and full-lumen exposure techniques to automatically delineate the boundaries of the vessel and its major branches. As previously described, the pressure gradient in bifurcated vessels was evaluated using Murray's bifurcation fractal law. The AMR was determined by the ratio of distal coronary pressure to hyperemic flow rate (18, 19). The calculation procedure and schematic diagram of AMR are detailed in Supplementary Material 1.

The main outcome measure was established as a combination of cardiac death or rehospitalization for heart failure within one year after undergoing PCI. Secondary endpoints encompassed the components of the primary endpoint and patient-centered adverse events, including all-cause mortality, recurrent myocardial infarction, and revascularization procedures. All clinical outcomes were evaluated according to standardized definitions established by the Academic Research Consortium. Cardiac death encompassed fatalities due to cardiac causes, unknown etiology, or indeterminate origin. Heart failure readmission was defined as a recent exacerbation of symptoms or a significant decline in cardiac function, characterized by a left ventricular ejection fraction of less than 50%, elevated B-type natriuretic peptide levels, or a diagnosis of heart failure at discharge (20). All incidents were independently evaluated by seasoned cardiologists in a blinded manner, and any disagreements were resolved through consensus among the experts. Follow-up data were collected through telephone interviews, outpatient records, and hospitalization documents.

All statistical analyses were performed using SPSS 27.0 (IBM Corporation, Armonk, NY) and R 4.4.1 (The R Foundation for Statistical Computing). Continuous variables are reported as mean ± standard deviation (SD) or median (interquartile range, IQR), with normal distribution evaluated using the ShapiroWilk test. For variables that follow a normal distribution, the independent samples t-test was employed for between-group comparisons. In contrast, the Mann–Whitney U test was utilized for variables that do not follow a normal distribution. Categorical variables are reported as frequencies and proportions, and comparisons were conducted using either the chi-square test or Fisher's exact test. The independent association between AMR and MACE was assessed using a multivariate logistic regression model, with results presented as odds ratios (ORs) and 95% confidence intervals (CIs). All variables in the model underwent multicollinearity testing using the Variance Inflation Factor (VIF), confirming that multicollinearity levels were maintained within an acceptable range (VIF < 10). To ensure statistical robustness, we performed two-sided tests at a significance level of P < 0.05. The Kaplan–Meier method was used to assess the cumulative MACE incidence across varying AMR levels, with intergroup differences evaluated using the log-rank test. Receiver operating characteristic (ROC) curve analysis was performed to determine the optimal antimicrobial resistance (AMR) cutoff, and its discriminatory performance was assessed using the area under the curve (AUC), sensitivity, and specificity. The threshold for statistical significance was established at P < 0.05.

Between January 1, 2021, and July 1, 2023, 748 patients undergoing PCI were screened. Following stringent clinical evaluation, 192 patients who failed to meet the inclusion criteria were excluded, and an additional 73 patients were excluded due to suboptimal angiographic image quality. A total of 483 patients were incorporated into the final analyses. Patients were divided into two groups according to their QFR and AMR measurements: the non-MACE group (n = 431) and the MACE group (n = 52) (Figure 1).

There were significant differences in baseline characteristics between the non-MACE and MACE groups (Table 1). Patients with MACE were older (67.00 [54.75–75.50] vs. 64.00 [54.00–74.00], P = 0.025) and had a higher prevalence of hypertension (88.5% vs. 59.9%, P < 0.001), diabetes (63.5% vs. 33.0%, P < 0.001), and hyperlipidemia (84.6% vs. 64.5%, P = 0.004). N-terminal pro B-type natriuretic peptide (NT-proBNP) concentrations were markedly elevated in the MACE group compared to the non-MACE group (1,190.00 [355.50, 4,777.50] pg/ml vs. 243.10 [82.12, 911.33] pg/ml, P < 0.001). Additionally, monocyte and platelet counts were elevated in the MACE group (0.62 [0.45, 0.91] vs. 0.48 [0.37, 0.64], P = 0.003; 244.50 [215.25, 266.00] vs. 208.00 [167.50, 245.00], P < 0.001), while lymphocyte counts were reduced (1.36 [1.00, 1.77] vs. 1.89 [1.50, 2.55], P < 0.001). Aspirin use at discharge was significantly lower in the MACE group compared to the non-MACE group (94.2% vs. 99.1%, P = 0.006), while the numerical of value AMR was notably higher in the MACE group than in the non-MACE group (292.50 [265.00, 351.25] vs. 230.00 [199.00, 256.00], P < 0.001).

The analysis of baseline characteristics uncovered statistically significant disparities in clinical parameters between the MACE and non-MACE groups. A multivariate logistic regression analysis was utilized to evaluate the independent predictive significance of these variables, while accounting for potential confounders.

The multivariate logistic regression analysis identified several variables as significant predictors of MACE (Table 2). Advanced age (OR: 1.310, 95% CI: 1.102–1.564; P = 0.033) and an elevated neutrophil count (OR: 1.807, 95% CI: 1.228–2.658; P = 0.003) were identified as significant risk factors.

An elevated platelet count (OR: 1.032, 95% CI: 1.012–1.052; P = 0.002) was significantly associated with an increased risk of MACE, whereas a higher lymphocyte count (OR: 0.839, 95% CI: 0.762–0.964; P = 0.001) demonstrated a protective effect. Furthermore, an elevated Killip class (OR: 1.072, 95% CI: 1.028–1.153; P = 0.027) and increased NT-proBNP levels (OR: 1.231, 95% CI: 1.149–1.437; P < 0.001) were independently associated with a higher incidence of MACE. In contrast, the high-value QFR demonstrated a significant association with a reduced risk of MACE (OR, 0.674; 95% CI, 0.519–0.874; P = 0.014). Additionally, an elevated AMR (OR: 1.085, 95% CI: 1.037–1.248; P < 0.001) was significantly correlated with an increased incidence of MACE. All the variables listed above demonstrated statistical significance, with P-values <0.05.

After conducting the multivariate logistic regression analysis, an ROC curve analysis was carried out to evaluate the model's ability to discriminate MACE, and the AUC was calculated.

The ROC curve analysis of the eight variables revealed varying degrees of predictive accuracy for MACE (Figure 2). AMR exhibited the highest predictive accuracy (AUC = 0.889), followed by lymphocyte count (AUC = 0.762), indicating a strong discriminatory capacity. NT-proBNP (AUC = 0.697) and platelet count (AUC = 0.658) demonstrated moderate predictive accuracy for major adverse cardiac events (MACE). Conversely, QFR (AUC = 0.583), neutrophil count (AUC = 0.573), age (AUC = 0.529), and Killip classification (AUC = 0.48) exhibited limited predictive value, with AUC values approaching 0.5, suggesting nearly random classification.

AMR demonstrated robust predictive accuracy for MACE, achieving an AUC of 0.889, which underscores its strong discriminative power. At an optimal threshold of 246.5, AMR attained a sensitivity of 0.98 and a specificity of 0.67, further emphasizing its effectiveness in predicting outcomes, especially in identifying high-risk patients.

In multivariate analysis, AMR achieved the highest AUC (0.889), indicating its potential role in predicting MACE (Figure 3). To establish AMR as an independent predictor, it was integrated into a multivariate logistic regression model to assess its independent predictive value for MACE.

In the multivariate logistic regression analysis of AMR (Table 3), several factors were found to be significantly associated with an increased risk of AMR: age (OR: 1.597, 95% CI: 1.228–1.967, P = 0.002), diabetes (OR: 2.727, 95% CI: 2.321–5.132; P = 0.017), hyperlipidemia (OR: 2.119, 95% CI: 2.051–5.489; P < 0.001), NT-proBNP (OR: 1.002, 95% CI: 1–1.258; P = 0.014), longer reperfusion time (OR: 1.018, 95% CI: 1.003–1.034; P = 0.023), CK/CKMB ratio (OR: 1.356, 95% CI: 0.524–2.188; P = 0.001), platelet count (OR: 1.095, 95% CI: 1.015–1.175; P = 0.020), smoking history (OR: 15.337, 95% CI: 5.083–25.591; P = 0.003), and postoperative QFR (OR: 1.340, 95% CI: 1.267–1.769; P < 0.001). In contrast, a higher lymphocyte count (OR: 0.876, 95% CI: 0.659–0.983; P = 0.030) was identified as a significant protective factor against AMR.

Participants were divided into two groups according to the optimal AMR cutoff value of 246.5 mmHg·s/m: the low AMR group (AMR < 246.5 mmHg·s/m, Group 1) and the high AMR group (AMR ≥ 246.5 mmHg·s/m, Group 2). A KaplanMeier survival analysis was conducted to assess the influence of AMR levels on the incidence of MACE and to compare the cumulative survival rates between the two groups.

Kaplan–Meier survival curve analysis revealed a statistically significant difference in the cumulative incidence of MACE between patients with low and high AMR (P < 0.0001) (Figure 4). The cumulative incidence of MACE was consistently higher in the high AMR group than that in the low AMR group during the follow-up period, with a marked divergence in the survival curves observed after approximately 200 days. This finding suggests an association between elevated AMR and increased MACE risk.

This study has several limitations that warrant careful interpretation of the findings. First, its single-center, retrospective design and relatively small sample size may introduce selection bias and unadjusted confounding, thereby limiting the generalizability and robustness of the results. Although multivariate analysis was performed, residual confounding cannot be entirely excluded. Second, the short follow-up duration may have restricted the assessment of AMR's predictive value for long-term MACE. Third, while AMR—derived noninvasively from coronary angiography—offers practical advantages such as avoiding adenosine and pressure wires, its accuracy depends on estimated parameters like flow velocity and vessel length, which may lead to discrepancies when compared to invasive measures such as the IMR. Furthermore, the lack of external validation against gold-standard methods, such as CMR imaging, limits the confidence in its physiological accuracy.