In a recent study published in Nature Ageing, a group of researchers explored the predictive power of plasma proteomic profiles for future dementia onset in adults without dementia, utilizing a large cohort from the United Kingdom (UK) Biobank (UKB), focusing on key biomarkers for effective early detection and intervention strategies.
Study: Plasma proteomic profiles predict future dementia in healthy adults. Image Credit: Atthapon Raksthaput/Shutterstock.com
Background
The gradual progression of dementia from an undetectable to a severe stage highlights the urgency of early detection. Blood-based biomarkers offer hope for identifying at-risk individuals before symptoms emerge.
However, the focus on limited proteins and lack of comprehensive analyses has stymied progress. Moreover, previous studies often overlooked the timing of biomarker changes.
Recent proteomic efforts fall short due to small sample sizes and broad outcomes. Further research is needed to identify precise, reliable biomarkers for early dementia detection, enabling timely intervention and advancing prevention strategies.
About the study
The present comprehensive prospective cohort study enrolled over 500,000 individuals aged 39 to 70 years between 2006 and 2010, aiming to monitor their health over an extended period.
Participants, registered with the UK National Health Service (NHS), were recruited from 22 assessment centers nationwide.
To facilitate this study, individuals diagnosed with dementia at baseline or who self-reported dementia, as well as those with incomplete proteomic data, were excluded, resulting in a cohort of 52,645 participants devoid of dementia.
Blood samples were collected across the UK between 2007 and 2010, with the majority taken during the baseline visit. These samples were stored in ethylenediaminetetraacetic acid (EDTA) tubes, centrifuged to separate plasma, and then aliquoted and frozen at -80°C until analysis.
The Olink Explore Proximity Extension Assay, a sophisticated antibody-based method, was employed for quantifying proteomic profiles in these samples at Olink Analysis Service in Sweden.
This process involved measuring 1,463 unique proteins, focusing on cardiometabolic health, inflammation, neurology, and oncology. Proteomic data were normalized and subjected to stringent quality control to ensure reliability.
Dementia outcomes were carefully determined using a combination of hospital admission records, primary care data, death registry entries, and specific algorithm definitions, ensuring a robust method for identifying incident dementia cases, including all-cause dementia (ACD), Alzheimer's disease (AD), and vascular dementia (VaD).
The study carefully tracked participants from their initial assessment until the earliest of either a dementia diagnosis, death, or the last follow-up in March 2023.
Statistical analyses did not precede with a predetermined sample size, but the cohort's scale was comparable or superior to previous studies.
The analytical approach utilized Cox proportional hazard regression models to explore the relationship between plasma protein levels and dementia incidence, adjusting for a range of demographic, genetic, and health-related variables.
Proteins significantly associated with dementia outcomes underwent further analysis, including enrichment analysis and a light gradient boosting machine classifier to rank proteins by their predictive importance.
Study results
The study included 52,645 adults free of dementia at the outset, predominantly white (93.7%) with a slight female majority (53.9%) and a median age of 58.
Over a follow-up of 14.1 years, 1,417 developed dementia, revealing incidences at various intervals post-baseline.
The median age at dementia onset was higher, with a slight shift towards male and white ethnicity predominance. AD and VaD were the most common diagnoses among these cases.
In exploring proteomic biomarkers, 1,463 proteins were analyzed, with significant associations found for various proteins with ACD, AD, and VaD after adjustments for demographic and genetic factors.
Notably, Glial Fibrillary Acidic Protein (GFAP) and Neurofilament Light Chain (NEFL) emerged as highly significant predictors across dementia types, alongside Growth Differentiation Factor 15 (GDF15) and Latent Transforming Growth Factor Beta Binding Protein 2 (LTBP2).
The study identified proteins with the greatest predictive power, with GFAP, NEFL, and GDF15 leading in importance for ACD, AD, and VaD prediction, respectively.
These proteins' predictive accuracy, individually and in combination with demographic and cognitive measures, was thoroughly evaluated.
Remarkably, combining GFAP or NEFL with demographic factors yielded high predictive accuracy for ACD and AD. These findings underscored the potential of specific plasma proteins, particularly GFAP, as promising biomarkers for early dementia detection.
Further analysis confirmed the prognostic value of high baseline levels of NEFL, GFAP, and GDF15 for increased dementia risk, with GFAP showing specific relevance to dementia.
This specificity was further supported by the lack of significant association between GFAP levels and non-dementia neurological or mental disorders, suggesting its potential as a dementia-specific marker.
These findings were validated through a split-sample approach, consistently highlighting the importance of GFAP, NEFL, GDF15, and LTBP2 in dementia prediction.
Additionally, predementia trajectories for these proteins indicated significant deviations from normal levels up to a decade before dementia onset, particularly for GFAP and NEFL in AD cases.
Furthermore, the specificity of GFAP and its early deviation in predementia stages particularly highlight its role as a valuable biomarker for future dementia risk assessment and intervention planning.
Frailty may predict dementia risk years before onset, study finds