Combined Biomarker Panel of Presepsin, HE4, and Oxygenation Index for Sepsis Diagnosis and Prognosis in Intensive Care Unit Patients

Jinmei Luo,1,* Shu An,2,* Xuanren Liao,2,* Juehui Wu,2 Laisheng Li2 1Department of Medical Intensive Care Unit, the Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, 510630, People’s Republic of China; 2Department of Laboratory Medicine, the First Affiliated Hospital of Sun Yat-Sen University, Guangzhou, 510080, People’s Republic of China*These authors contributed equally to this workCorrespondence: Juehui Wu, Department of Laboratory Medicine, the First Affiliated Hospital of Sun Yat-sen University, 58 Zhongshan Road, Guangzhou, 510080, People’s Republic of China, Tel/Fax +86-20-28823350-8464, Email [email protected] Laisheng Li, Department of Laboratory Medicine, the First Affiliated Hospital of Sun Yat-sen University, 58 Zhongshan Road, Guangzhou, 510080, People’s Republic of China, Tel/Fax +86-20-28823350-8464, Email [email protected]: Sepsis remains a lethal global health crisis with persistently high mortality, exacerbated by diagnostic challenges stemming from its heterogeneous clinical presentation and limitations of current diagnostic tools like SOFA score. While biomarkers such as Presepsin or human epididymis protein 4 (HE4) show promise, single-marker approaches exhibit insufficient accuracy for reliable early detection and prognosis.Objective: We evaluated the diagnostic and prognostic utility of Presepsin, HE4, Oxygenation Index (OI), and their combined panel (PHO) in critically ill patients with or without sepsis.Methods: This single-center study analyzed 411 ICU patients (165 non-sepsis, 246 sepsis or septic shock). Clinical parameters—including SOFA, and OI—alongside laboratory parameters (hematological indices, hepatic/renal function markers, inflammatory biomarkers, blood culture results) were extracted from the hospital’s Laboratory Information System. Residual admission samples were used to quantify Presepsin and HE4. Receiver operating characteristic (ROC) curve analysis, Spearman’s rank correlation, and gradient boosting machine learning models were employed to evaluate the diagnostic and prognostic performance.Results: Presepsin, HE4, and OI were significantly elevated in septic patients compared to controls (all P < 0.05) and correlated strongly with 30-day mortality. The gradient boosting algorithm identified these three markers as the most significant predictors. Importantly, the combined biomarker panel PHO demonstrated superior performance in both diagnosis and prognosis. For sepsis diagnosis, PHO achieved an outstanding AUC of 0.892 (95% CI: 0.860– 0.924), significantly outperforming individual biomarkers (Presepsin: 0.821; HE4: 0.803; OI: 0.752) and conventional inflammatory markers (all P< 0.05). For mortality prediction, PHO maintained the highest prognostic accuracy (AUC 0.706, 95% CI: 0.641– 0.772), with improved sensitivity and specificity compared to single biomarkers.Conclusion: The combined Presepsin, HE4 and OI biomarker panel significantly outperforms individual markers in sepsis diagnosis and prognosis. This machine learning-validated composite indicator enables early risk stratification and may guide timely interventions to improve outcomes.Keywords: presepsin, HE4, oxygenation index, sepsis, diagnosis, prognosis