Proteomic biomarkers of cognitive function, APOE ε4 status, and dementia in Generation Scotland

This study utilizes untargeted mass spectrometry in the Generation Scotland cohort to identify specific blood-based proteomic biomarkers associated with cognitive function, APOE ε4 status, and reduced risk of incident dementia, highlighting selenoprotein P, afamin, and complement proteins as promising candidates for future replication and early detection efforts.

The Gist

Imagine your blood is a bustling city, and the proteins floating inside it are the workers, messengers, and construction crews keeping everything running. Scientists have long tried to find specific "workers" in this city that might signal if the brain's "library" (memory and thinking skills) is in trouble. Usually, researchers only look at a short, pre-approved list of workers (like a fixed menu at a restaurant). This new study, however, decided to open the kitchen and take a snapshot of 439 different workers at once using a high-tech camera called "mass spectrometry."

Here is what the researchers found in the "Generation Scotland" group of over 14,000 people, explained simply:

1. The "Brain Fuel" Worker (Selenoprotein P)

The researchers looked for workers linked to how well people performed on four different brain tests (like a word puzzle, a memory game, and a speed test).

• The Finding: They found one specific worker, Selenoprotein P, that showed up more often in people who did well on three of the tests.
• The Analogy: Think of Selenoprotein P as a delivery truck for a special vitamin (selenium) that the brain needs to function. The study suggests that people with more of these delivery trucks in their blood tended to have sharper minds. Interestingly, this truck wasn't linked to the memory game (logical memory), only the other speed and language tests.

2. The "Genetic Risk" Signal (APOE ε4)

You may have heard of the APOE ε4 gene; it's like a genetic "warning label" that increases the risk of Alzheimer's disease. The researchers wanted to see if people with this warning label had different "workers" in their blood compared to those without it.

• The Finding: People with one or two copies of this risk gene had significantly fewer of a worker called Afamin in their blood.
• The Analogy: Imagine Afamin as a specialized cleaner that helps manage fats and vitamins in the body. The study found that if you carry the "risk gene," your body seems to have fewer of these cleaners. This happened even after the researchers accounted for other things like diet, weight, and smoking, suggesting this is a direct link to the gene itself.

3. The "Future Warning" Signal (Complement System)

Finally, the team looked at who developed dementia over the next 17 years. They wanted to see if the "worker" levels in the blood could predict who would get sick later.

• The Finding: They found that people with higher levels of a specific group of workers (called Protein Group 21, which includes Complement C2 and Complement Factor B) were less likely to develop dementia later in life.
• The Analogy: Think of these workers as the immune system's security guards. The study suggests that having a stronger, more active security team in your blood might act as a shield, lowering the risk of the brain's "library" getting damaged by dementia. The risk was cut by about 25% for those with higher levels of these guards.

What the Study Doesn't Say

It is important to stick to what the paper actually claims:

• It's a Discovery, Not a Diagnosis: The authors call this an "exploratory" study. They found clues, but they haven't proven these proteins cause the changes or that doctors can use them to diagnose patients today.
• No Guarantees: Just because these proteins are linked to brain health in this specific group of Scottish people doesn't mean the same rules apply to everyone else yet. The authors explicitly state that these findings need to be tested in other populations to see if they hold up.
• Correlation vs. Causation: The study shows these proteins move together with brain health, but it doesn't prove that changing the protein levels will fix the brain.

The Bottom Line

This study is like a detective taking a wide-angle photo of a crime scene (the blood) instead of just looking at one specific clue. They found three interesting suspects:

1. Selenoprotein P: More of this = better current brain performance.
2. Afamin: Less of this = having the APOE ε4 risk gene.
3. Complement Guards: More of these = lower risk of developing dementia in the future.

These findings offer new leads for scientists to investigate, but they are not yet ready to be used as medical tests for patients.

Technical Summary

Technical Summary: Proteomic Biomarkers of Cognitive Function, APOE ε4 Status, and Dementia in Generation Scotland

Problem Statement
Current large-scale biomarker discovery for dementia predominantly relies on targeted, affinity-based protein panels that measure a predefined subset of the proteome. This approach may limit the identification of novel biomarkers or biological pathways relevant to cognitive ageing. While blood-based biomarkers are highly desirable for their accessibility, there is a need to utilize untargeted mass spectrometry to explore the broader, understudied subset of the circulating proteome in relation to cognitive function, genetic predisposition (specifically APOE ε4 status), and incident dementia.

Methodology
The study utilized data from the Generation Scotland (GS) cohort, a family-based study of individuals aged 17–99. The analysis focused on 439 independent mass spectrometry signals (representing serum proteins and protein groups) measured via ultra-high-throughput scanning SWATH mass spectrometry.

The statistical approach was tailored to three distinct analytical goals:

1. Cognitive Function: A multivariate Bayesian joint sparse model (MAJA) was employed to simultaneously assess associations between the 439 protein signals and five outcomes: four cognitive tests (digit symbol, logical memory, verbal fluency, vocabulary) and a cognitive polygenic score (PGS). This method accounts for the covariance among both proteins and cognitive traits to improve power and control false positives. Models were adjusted for age, sex, family structure, and fully adjusted for BMI, alcohol intake, depression, education, hypertension, diabetes, smoking status (EpiSmokEr), and deprivation (SIMD).
2. APOE ε4 Status: Linear mixed-effects models tested associations between protein levels and APOE ε4 allele count (0, 1, or 2 copies). Significance was determined using a Bonferroni-adjusted threshold ($p < 3.1 \times 10^{-4}$) based on the number of principal components explaining 80% of proteomic variance.
3. Incident Dementia: Mixed-effects Cox proportional hazard models assessed associations between protein levels and time-to-dementia diagnosis (up to 17 years post-sampling). Models included adjustments for APOE ε4 status and other covariates.

Key Results

• Cognitive Function: In fully adjusted models, a single mass spectrometry signal mapping to selenoprotein P was significantly associated with three cognitive tests: digit symbol, vocabulary, and verbal fluency (absolute $\beta$ range = 0.02–0.03, Posterior Inclusion Probability $\ge$ 0.95). No associations were found for logical memory. Additionally, four protein groups (including Complement C4-B and groups mapping to Complement C4-A/B) were associated with the cognitive PGS.
• APOE ε4 Status: Carrying one or two copies of the APOE ε4 allele was associated with distinct proteomic profiles. Notably, a signal mapping to afamin showed a negative association with APOE ε4 carrier status ($\beta = -0.08$ for one copy, $\beta = -0.2$ for two copies). Several apolipoproteins also showed expected associations, such as increased apolipoprotein B-100 and decreased apolipoprotein E levels in carriers.
• Incident Dementia: In fully adjusted models, a higher mass spectrometry signal mapping to Protein Group 21 (comprising Complement Factor B and Complement C2) was associated with a lower hazard for incident dementia (Hazard Ratio = 0.75, $p < 3.1 \times 10^{-4}$).

Significance and Claims
The authors position this work as an exploratory, discovery-stage analysis demonstrating the utility of untargeted mass spectrometry in identifying candidate biomarkers for cognitive health and dementia risk. The study highlights specific independent signals—selenoprotein P for cognitive function, afamin for APOE ε4 status, and Complement Factor B/C2 for dementia risk—that warrant further investigation.

The paper explicitly states that these findings require replication in other populations and mechanistic follow-up to establish reliability. The authors acknowledge limitations, including the cross-sectional nature of the cognitive analysis (precluding causal inference), the restriction to European ancestry, and the fact that the study design cannot distinguish whether these proteins are causal drivers or secondary changes. Future work is suggested to include Mendelian randomization to explore causality. The study does not claim clinical utility at this stage, emphasizing that these are potential biological pathways and candidate biomarkers requiring validation.