In a recent study published in Nature Aging, researchers from China analyzed the proteomic data of adults without dementia. They found that GFAP (short for glial fibrillary acidic protein) is an optimal biomarker for predicting dementia, even more than ten years before diagnosis, highlighting its potential role in screening high-risk individuals and enabling early intervention.
Study: Plasma proteomic profiles predict future dementia in healthy adults. Image Credit: mapush / Shutterstock
Background
Detecting the progression of dementia from asymptomatic stages to clinical manifestation is crucial, given the current lack of effective therapy. The emergence of blood-based biomarkers offers a potential breakthrough, presenting a promising tool for early risk screening and intervention in the preclinical phase of dementia. Previous efforts in biomarker discovery for dementia risk have focused on a limited number of proteins due to technical constraints and a lack of systematic comparison. Prior investigations employing proteomics strategies revealed blood protein differences but were often cross-sectional and did not address temporal aspects or specific dementia subtypes. Large-scale prospective studies with data on blood proteomics and various dementia types are essential for improving our understanding.
The predictive performance of proteins (individual or combined) in various incidence time groups (e.g., 10 years, >10 years) has been overlooked despite being a crucial aspect of ultra-early detection and prevention. Despite previous efforts, there remains a dearth of blood proteomic biomarkers with the necessary sensitivity and specificity for predicting future dementia. Therefore, researchers in the present study employed a large-scale proteomic approach to identify plasma biomarkers associated with dementia, exploring their predictive performance and trajectories related to clinical progression.
About the study
In this prospective cohort study, data from the United Kingdom Biobank (UKB) was used. The study included 52,645 dementia-free adults of median age 58 years; 93.7% were White and 53.9% were female. Although the blood samples were collected and preserved between 2007 and 2010, their proteomic profiling was conducted between 2021 and 2022. Normalized Protein eXpression (NPX) values were generated to account for variations.
The primary outcomes were incident events related to AD, vascular dementia (VaD), and all-cause dementia (ACD). The outcomes were identified through comprehensive data sources, such as reports from primary care records, hospital admissions, and death registry records. The earliest recorded date of diagnosis from these sources determined the dementia diagnosis date. Follow-up (median 14.1 years) for participants commenced from their attendance at the assessment center. It continued until the earliest of the recorded diagnosis date, mortality date, or the last available date provided by healthcare providers. The study ensured the reliability of dementia diagnoses by linking data to UK electronic health records, where professional clinicians reported and classified cases based on the International Classification of Diseases (ICD)-9 and ICD-10 codes. Self-reported disease cases were excluded to enhance diagnostic accuracy. Statistical analysis involved the use of Cox proportional hazard models, chi-square tests, Student's t-tests, enrichment analysis, receiver operating characteristic (ROC) analysis, Kaplan–Meier curves, and Mann-Kendall trend tests.
Results and discussion
Over the follow-up, the incidence of dementia was found to be 2.7%. The median age of incident AD participants was 66 years, with 48.5% females. The incidence rate of AD and other dementias increased with age, peaking at 6.26 per 1,000 person-years in the 65–69-year age group.
After adjusting for relevant factors, GFAP and NEFL (short for neurofilament light) demonstrated the most significant associations with incident all-cause dementia, AD, and VaD. Additionally, GDF15 (short for growth/differentiation factor 15) and LTBP2 (short for latent transforming growth factor beta binding protein 2) were implicated in increased dementia risk, with enrichments in pathways related to extracellular matrix organization, immune system, and infectious diseases. These proteins also ranked highest in predicting all-cause dementia and its subtypes.
NEFL, GFAP, and GDF15 exhibited modest individual predictive accuracy for all-cause dementia, with potential improvement when combined with demographic indicators and cognitive tests. Notably, combining NEFL or GFAP with demographic features and brief cognitive tests showed significant enhancement in accuracy.
Further, individuals with higher baseline levels of GFAP had a 2.91 times greater likelihood of developing AD, while those with higher GDF15 levels had a 2.45 times greater likelihood of developing VaD. Interestingly, although GFAP levels were found to be associated with an increased risk of other dementias, they did not appear to be associated with other neurological ailments. The study is strengthened by an extended follow-up and the high-throughput proteomic analysis of a large community-based sample. However, it is limited by potential biases, lower dementia incidence in participants, lack of external validation datasets, variability in detection techniques, and diagnostic uncertainty.
Conclusion
In conclusion, the study suggests that GFAP levels hold high predictive value in identifying the risk of dementia in the future, even ten years before onset. These findings may have significant implications for screening individuals at an increased risk of dementia and for implementing early intervention strategies for improved prognosis.