Philip Ma, United States of America
PrognomIQ Inc. -Author Of 1 Presentation
PROTEOGRAPH NANOPARTICLE-BASED PLASMA PROTEIN PROFILING OF ALZHEIMER’S AND MILD COGNITIVE IMPAIRMENT SUBJECTS HIGHLIGHTS NOVEL COMBINATIONS OF KNOWN/UNKNOWN CANDIDATE BIOMARKERS
Abstract
Aims
The identification of clinically useful biomarkers for Alzheimer’s disease (AD) from blood is a long-standing goal. Here we report the use of a platform for untargeted plasma protein profiling, Proteograph, to identify candidate protein biomarkers from plasma for AD and Mild Cognitive Impairment (MCI).
Methods
Plasma samples from 200 subjects comprising 50 AD, 50 MCI, and 100 controls were profiled using a recently reported plasma protein profiling platform, Proteograph1. Using a 5-nanoparticle panel and 85 µL of plasma per nanoparticle, proteins were quantified by data-independent acquisition (DIA) liquid-chromatography mass-spectrometry (LC-MS) in about 6 weeks. Normalized peptide intensities were used in ten rounds of 10-fold cross-validation to develop random forest models for class discrimination.
Results
Across the samples, 2,391 plasma proteins were detected, with 2,085 in at least 25%. 36 proteins with the highest AD OpenTargets2 score were detected, including Amyloid Beta (A4) Precursor Protein. 25,593 protein-comprising peptides were detected, with 15,661 in at least 25% of the samples. Univariate analysis identified 441 and 526 proteins that were significantly different in AD or MCI versus control, respectively. Random-forest classification for AD and MCI produced ROC AUCs that were at least 0.90. Top features by importance included known and unknown candidate biomarkers.
Conclusions
These analyses have identified novel combinations of candidate plasma protein markers, many without prior known relevance to AD. More broadly, the Proteograph platform confirmed its ability to generate profiling data in a deep, broad, and rapid fashion, enabling large-scale studies to detect novel insights with clinically useful potential.