Abstract
BACKGROUND: Heart failure with preserved ejection fraction (HFpEF) remains a formidable clinical challenge due to its intricate pathophysiology and the difficulty in early diagnosis. Traditional biomarkers for HFpEF are often insufficient in clinical practice, driving the need for more novel, sensitive diagnostic markers. METHODS: Serum samples from 58 HFpEF patients and 30 healthy controls were analyzed using Olink proteomics technology. Advanced machine-learning algorithms were employed to comprehensively compare the diagnostic performance of individual biomarkers and their combinations. RESULTS: The serum levels of DLK-1, PSP-D, and PCSK-9 in HFpEF patients were higher than those in the control group. NT-proBNP, UA and PCSK-9 were identified as risk factors for HFpEF, with regression coefficients of 0.009 for NT-proBNP, 0.006 for UA, and 1.061 for PCSK-9 respectively. The area under the receiver operating characteristic curve (AUC) of the combination of NT-proBNP, DLK-1, PSP-D and PCSK-9 for the diagnosis of HFpEF reached 0.794. This value outperformed the AUC of the combination of PCSK-9 and NT-proBNP, 0.788, the combination of three markers (DLK-1, PSP-D, and PCSK-9, 0.622), as well as the AUCs of each of the four markers alone (NT-proBNP: 0.778, DLK-1: 0.578, PSP-D: 0.523, PCSK-9: 0.628). CONCLUSION: The combined detection of NT-proBNP, DLK-1, PSP-D and PCSK-9 can significantly enhance the specificity and sensitivity of the clinical diagnosis of HFpEF patients, holding great potential for improving the diagnostic accuracy of HFpEF in clinical settings.