AUTHOR=Chen Zheng , Liu Xiaoli , Sun Yan , Zhang Dai , Cheng Yujing , Zhou Yujie 

TITLE=Prognostic value of the atherogenic index of plasma in patients with acute coronary syndrome without standard modifiable risk factors: a machine learning-based cohort study

JOURNAL=Frontiers in Cardiovascular Medicine

VOLUME=Volume 12 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2025.1681170

DOI=10.3389/fcvm.2025.1681170

ISSN=2297-055X

ABSTRACT=BackgroundPatients without any standard modifiable cardiovascular risk factors (SMuRF-less) who develop acute coronary syndrome (ACS), tend to have poor outcomes. However, the prognostic value of atherogenic index of plasma (AIP) in these patients is unclear. Therefore, we investigated the association between AIP and adverse outcomes in SMuRF-less patients with ACS.MethodsThis study retrospectively enrolled 722 SMuRF-less patients with ACS receiving percutaneous coronary intervention (PCI) at Beijing Anzhen Hospital from March 2017 to March 2018. Three patient-groups were formed using AIP tertiles. The primary outcome, major adverse cardiovascular and cerebrovascular events (MACCE), was a composite of all-cause mortality, non-fatal myocardial infarction (MI), unplanned revascularization, and non-fatal ischemic stroke. Association between AIP levels and MACCE risk was examined using restricted cubic spline (RCS) analysis. Prognostic value of AIP levels for MACCE was assessed using multivariable Cox regression models and machine learning approaches.ResultsDuring follow-up of the 722 patients (median age, 60 years [interquartile range, 53–67]; female, 29.8%; median follow-up duration, 59 months), 168 (23.3%) developed MACCE. The RCS results showed linear association of progressively increasing MACCE risk with increasing AIP levels. In multivariable Cox regression analysis, significantly higher MACCE risk occurred with the highest AIP tertile than with the lowest (hazard ratio [HR] 2.03, 95% confidence interval [CI]: 1.34–3.08; P < 0.001). Elevated AIP level was associated with higher risks of all-cause death (HR: 3.49, 95% CI: 1.09–11.23; P = 0.036); non-fatal MI (HR: 3.02, 95% CI: 1.08–8.48; P = 0.035); and unplanned revascularization (HR: 2.18, 95% CI: 1.34–3.52; P < 0.001). As a continuous variable, higher AIP levels were significantly associated with increased risks of MACCE (HR: 2.95, 95% CI: 1.74–4.98; P < 0.001), all-cause mortality (HR: 6.80, 95% CI: 1.85–24.96; P = 0.003), non-fatal myocardial infarction (HR: 3.58, 95% CI: 1.08–11.86; P = 0.037), and unplanned revascularization (HR: 2.84, 95% CI: 1.55–5.19; P < 0.001). Machine-learning models incorporating AIP levels improved outcome prediction. At 48 months, the gradient boosting machine model achieved the highest AUC (0.796; 95% CI: 0.703–0.889), while complementary assessments showed that the random survival forest model provided the greatest net clinical benefit and demonstrated excellent calibration.ConclusionAmong SMuRF-less patients with ACS undergoing PCI, AIP level was identified as an independent predictor of clinical prognosis.