Identification of blood plasma protein ratios for distinguishing Alzheimer's disease from healthy controls using machine learning.

Abstract

Early detection of Alzheimer's disease is essential for effective treatment and the development of therapies that modify disease progression. Developing sensitive and specific noninvasive diagnostic tools is crucial for improving clinical outcomes and advancing our understanding of this condition. Liquid biopsy techniques, especially those involving plasma biomarkers, provide a promising noninvasive method for early diagnosis and disease monitoring. In this study, we analyzed the plasma proteomic profiles of 38 healthy individuals, with an average age of 66.5 years, and 22 patients with Alzheimer's disease, with an average age of 79.7 years. Proteins in the plasma were quantified using specialized panels designed for proteomic extension assays. Through computational analysis using a linear support vector machine algorithm, we identified 82 differentially expressed proteins between the two groups. From these, we calculated 6642 possible protein ratios and identified specific combinations of these ratios as significant features for distinguishing between individuals with Alzheimer's disease and healthy individuals. Notably, the protein ratios kynureninase to macrophage scavenger receptor type 1, Neurocan to protogenin, and interleukin-5 receptor alpha to glial cell line-derived neurotrophic factor receptor alpha 1 achieving accuracy up to 98 % in differentiating between the two groups. This study underscores the potential of leveraging protein relationships, expressed as ratios, in advancing Alzheimer's disease diagnostics. Furthermore, our findings highlight the promise of liquid biopsy techniques as a noninvasive and accurate approach for early detection and monitoring of Alzheimer's disease using blood plasma.