Technical performance and biotemporal stability evaluation of Olink proximity extension assay for blood‐based biomarker discovery

• Published in: Alzheimer's & dementia Vol. 17; no. S5
• Main Authors: Mattingly, Zoe A., Celia, Amanda M., Kuo, Yi‐kai, Trombetta, Bianca A., Ramirez, Christopher, LaCasse, Kathryn, Kivisäkk, Pia, Carlyle, Becky C., Arnold, Steven E.
• Format: Journal Article
• Language: English
• Published: 01.12.2021
• ISSN: 1552-5260; 1552-5279
• DOI: 10.1002/alz.056318

Background Novel biomarkers are necessary for improving differential diagnosis of Alzheimer’s Disease (AD), disease monitoring, and treatment personalization. Core cerebrospinal fluid (CSF) AD biomarkers: amyloid‐ß (Aß), total‐tau (T‐tau), and phosphorylated‐tau (P‐tau181) are strong markers of the presence of amyloid pathology, but are not suitable for measuring disease progression or drug response in a clinical trial . In this context, blood‐based biomarkers are increasingly desirable, as plasma is less invasive than CSF, enabling short‐term repeated sampling. Method Technical performance of Olink Proteomic’s multiplex proximity extension assay was evaluated using plasma samples from the MADRC longitudinal cohort. Over 400 analytes were measured on five off‐the‐shelf panels. Inter‐plate and intra‐plate coefficient of variations (CVs) were calculated from 3 samples run in duplicate on each plate. ANOVA was used to assess proportion of technical versus biological sources of variance. Multi‐protein investigations require consideration of multiple testing. Power calculations were performed using baseline samples from n=34 Controls and n=20 Dementia‐AD subjects to demonstrate optimal sample size for Olink studies. Result The majority of analyte mean CVs fell within the acceptable range for inter and intra‐plate measurements (<15%). Higher CVs were generally related to lower analyte abundance. Most analytes were relatively stable (Biotemporal CV < 15%) in control individuals year to year. ANOVA determined that the greatest source of variation in plasma was due to biological inter‐individual variability, as opposed to technical variation for all but 73 proteins. With regards to experimental power in a 450 protein experiment, a high effect size protein such as NfL required fewer than 50 samples per group to achieve confidence in observed significant differences. In contrast, the moderate effect size MCP‐1 required just under 500 samples per group when 450 proteins are measured. This decreased to n=400 if only 100 proteins are measured, and n=170 if MCP‐1 is measured alone. Conclusion Olink technology is technically robust and reliable, with biological factors being the primary source of variation for most proteins in plasma. Particularly for medium effect size proteins, sample size is an important consideration when planning experiments using this technology.