Association between perioperative renal function trajectory and post-discharge prognostic outcomes in STEMI patients undergoing PCI

The association between peri-procedural renal function trajectories and long-term prognosis in STEMI patients after percutaneous coronary intervention (PCI) remains unclear. This retrospective cohort study enrolled patients from two tertiary centers in Zhejiang, China (Taizhou Hospital of Zhejiang Province [January 2015–December 2024] and the Fourth Affiliated Hospital of Zhejiang University School of Medicine [January 2015–December 2020]). Latent class growth modeling (LCGM) was applied to renal function indices (serum creatinine and eGFR) measured at three peri‑procedural time points (before PCI [T1], within 72 h after PCI [T2], and before discharge [T3]) to identify distinct trajectories of renal function change. Multivariable Cox regression and competing‑risk analyses were used to assess associations between the identified trajectories and post‑discharge outcomes, including major adverse cardiovascular events (MACEs) and worsening renal function (WRF). A total of 2,413 patients were included (mean follow‑up 32.50 ± 25.28 months). During follow‑up, 593 (24.58%) experienced MACEs and 103 (4.27%) developed WRF. Two serum creatinine trajectories (low‑stable and high‑increase) and three eGFR trajectories (low‑stable, moderate‑stable, and high‑stable) were identified. Compared with the low‑stable creatinine group, the high‑increase group had a 31.0% higher risk of MACEs (HR = 1.310, 95% CI: 1.064–1.613, P = 0.011) and a 2.312‑fold increased risk of WRF (HR = 2.312, 95% CI: 1.468–3.640, P < 0.001). For eGFR trajectories, the moderate‑stable group showed a 51.5% lower risk of WRF versus the low‑stable group (HR = 0.485, 95% CI: 0.283–0.831, P = 0.008). Peri-procedural renal function trajectories in STEMI patients post-PCI are associated with long-term prognosis, but they do not demonstrate incremental prognostic value beyond conventional baseline and discharge renal function measurements in fully adjusted models, and their clinical utility requires further investigation.