Presymptomatic plasma biomarkers in autosomal dominant Alzheimer's disease: sequence and timing

This study utilized a longitudinal cohort of autosomal dominant Alzheimer's disease mutation carriers to identify a sequential cascade of nine plasma protein biomarkers that diverge from non-carriers between 26 and 6 years prior to expected symptom onset, offering a refined framework for staging presymptomatic disease and optimizing clinical trial design.

The Gist

The Big Picture: Predicting a Storm Before the Rain Starts

Imagine Alzheimer's disease not as a sudden storm that hits your house, but as a slow-building weather system. For decades, the clouds gather, the wind picks up, and the pressure drops long before the first drop of rain (symptoms) actually falls.

This study is like having a team of meteorologists who have found nine different weather instruments (blood tests) that can tell us exactly when that storm is coming, and in what order the weather changes will happen.

The researchers focused on a special group of people: families with a "genetic switch" (a mutation) that guarantees they will develop Alzheimer's at a young age. Because the timing of their "storm" is predictable (based on when their parents got sick), the scientists could look at their blood years in advance and see exactly when the warning signs appeared.

The "Weather Instruments" (The Biomarkers)

The scientists used a high-tech blood test (called NULISA) that can measure 124 different proteins at once. Think of these proteins as tiny messengers in your blood that shout out when something is wrong in the brain.

They found nine specific messengers that start shouting at different times. Here is the sequence of events, from the very first warning sign to the final storm:

1. The First Whisper: Aβ42 (Amyloid Beta)

• When it happens: About 26 years before symptoms start.
• The Analogy: Imagine a factory that makes bricks. In this disease, the factory starts making the wrong kind of bricks (sticky, clumpy ones) decades before the house starts to collapse.
• What the study found: This was the very first thing to change. The blood showed an increase in these "wrong bricks" (Aβ42) more than two decades before the person felt sick. Interestingly, in these genetic families, the level of these bricks goes up in the blood, which is different from the usual sporadic Alzheimer's where they go down.

2. The Construction Crew Goes Wild: Tau Proteins (pTau)

• When it happens: About 21 to 24 years before symptoms.
• The Analogy: Inside the brain, there are railroad tracks (called microtubules) that trains (nutrients) use to move around. The "Tau" protein is the glue that holds these tracks together. When the disease starts, the glue gets sticky and clumps up, derailing the trains.
• What the study found: The blood showed that this "glue" was getting messy and clumping up (phosphorylated tau) roughly 20 years before the person forgot their name or got lost. This was the second major wave of change.

3. The Firefighters Arrive: GFAP and BACE1

• When it happens: About 14 years before symptoms.
• The Analogy: Now that the tracks are derailing and the bricks are piling up, the brain's "firefighters" (immune cells called astrocytes) and "construction managers" (enzymes) are rushing to the scene to put out the fire and fix the mess.
• What the study found: The blood showed these emergency responders were active about 14 years before symptoms. This tells us the brain is now actively trying to fight the damage.

4. The Structural Damage: NfL (Neurofilament Light)

• When it happens: About 6 years before symptoms.
• The Analogy: The fire has been burning for a while, and now the actual walls of the house are starting to crack. NfL is a piece of the wall that falls into the blood when the brain cells are dying.
• What the study found: This marker appeared much later, just a few years before the person started showing symptoms. It signals that the "house" (the brain) is finally starting to lose its structure.

Why Does This Matter?

1. We Can Now "Stage" the Disease
Before this, we knew the disease happened early, but we didn't know the exact order. Now, we have a timeline. If you test a person's blood and see the "bricks" (Aβ42) are up but the "firefighters" (GFAP) aren't active yet, we know they are in the very early stages. If the "wall pieces" (NfL) are up, they are closer to needing help.

2. Better Clinical Trials
Imagine trying to test a new medicine to stop a storm. If you only test people who are already getting wet (symptomatic), it's too late; the roof is already gone.

• This study tells researchers: "Hey, you need to start testing your medicine 20 years before the symptoms, when the 'glue' (Tau) starts getting sticky."
• This helps scientists pick the right people for trials at the right time, making it much more likely that new drugs will work.

3. A Warning About Treatment
The study also found that one marker (AChE) only changed in people who were already taking medication for memory loss. This is a reminder that if you are testing blood for research, you have to be careful not to confuse the disease with the medicine used to treat it!

The Bottom Line

This research is like finding the perfect timeline for a disease. It shows us that Alzheimer's is a long, slow process that starts with a change in how the brain makes proteins, moves to a change in how it holds itself together, and finally leads to cell death.

By using a simple blood test to track these nine messengers, we can one day predict exactly when the "storm" is coming and intervene with treatments long before the first drop of rain falls.

Technical Summary

1. Problem Statement

Autosomal Dominant Alzheimer's Disease (ADAD) serves as a critical model for understanding the pathogenesis of Alzheimer's disease (AD) because it allows for the study of individuals carrying pathogenic mutations decades before symptom onset. While blood-based biomarkers for AD have been identified, the temporal sequence and precise timing of their divergence from non-carriers in the presymptomatic phase remain incompletely characterized. Previous studies have often focused on single biomarkers or limited panels with variable results. There is a need for a high-throughput, simultaneous analysis of multiple plasma proteins to establish a robust biomarker timeline, which is essential for staging disease progression and optimizing the design of prevention trials.

2. Methodology

• Study Design & Cohort:
  • Population: A longitudinal cohort from the Dementia Research Centre, UCL, comprising 113 individuals (73 mutation carriers [MCs] and 40 non-carrier familial controls [NCs]).
  • Samples: 270 plasma samples collected longitudinally (median 2 visits per person).
  • Genetics: MCs carried pathogenic variants in PSEN1 (31 variants) or APP (7 variants).
  • Metric: Estimated Years from Onset (EYO) was calculated by subtracting the parent's age at onset (or the variant-specific average) from the participant's age at sampling. Negative EYO indicates presymptomatic status; positive EYO indicates symptomatic status.
• Biomarker Technology:
  • Platform: NULISA (NUcleic acid-Linked Immuno-Sandwich Assay) using the CNS 120 Panel on the ARGO analyser.
  • Scope: Quantified 124 proteins related to neurodegeneration from small plasma volumes (~30µl).
  • Key Analytes: Included phosphorylated tau species (pTau181, pTau217, pTau231), total tau (MAPT), Amyloid-beta (Aβ42, Aβ40, Aβ38), GFAP, NfL, BACE1, and AChE.
• Statistical Analysis:
  • Cross-sectional: Linear regression models comparing MCs vs. NCs, adjusted for sex and age. False Discovery Rate (FDR) correction was applied.
  • Longitudinal: Linear Mixed Effects (LME) models with restricted cubic splines to model non-linear relationships with EYO.
  • Divergence Point: The timing of biomarker divergence was defined as the EYO where the 95% confidence interval (CI) of the difference between MCs and NCs no longer included zero. Sensitivity analyses used 90% and 99% CIs.

3. Key Contributions

• High-Throughput Profiling: This is one of the first studies to simultaneously evaluate a broad panel of 124 plasma proteins in an ADAD cohort using the highly sensitive NULISA technology.
• Sequential Timeline: The study establishes a chronological sequence of plasma biomarker abnormalities relative to symptom onset, moving from amyloid metabolism to tau pathology, astrocytosis, and finally neurodegeneration.
• Gene-Specific Differences: It identifies distinct amyloid-beta profiles associated with APP versus PSEN1 mutations, highlighting that biomarker interpretation must account for the specific genetic etiology.
• Treatment Confounding: The study explicitly demonstrates how cholinesterase inhibitor (ChI) use can confound plasma AChE levels, a crucial consideration for future clinical trial biomarker selection.

4. Key Results

Cross-Sectional Findings:
Nine proteins were significantly elevated in mutation carriers compared to non-carriers (FDR < 0.05):

1. pTau217 (FC 2.63)
2. pTau231 (FC 2.54)
3. pTau181 (FC 2.17)
4. GFAP (FC 2.03)
5. Aβ42 (FC 1.70)
6. MAPT (Total Tau) (FC 1.69)
7. NfL (FC 1.60)
8. BACE1 (FC 1.17)
9. AChE (FC 1.30; significance lost when adjusting for age, likely due to ChI use in symptomatic patients).

Longitudinal Timing of Divergence (EYO):
The study mapped the approximate onset of significant biomarker elevation relative to expected symptom onset:

• ~26 Years Prior: Aβ42 levels were elevated in carriers. This is the earliest detectable change, suggesting constitutive alteration in APP processing.
• ~21–24 Years Prior: Phosphorylated Tau markers diverged.
  • pTau231: ~24 years prior.
  • pTau181: ~23 years prior.
  • pTau217: ~21 years prior.
• ~19 Years Prior: Total Tau (MAPT) divergence.
• ~14 Years Prior: GFAP (astrocytosis) and BACE1 (synaptic injury) divergence.
• ~6 Years Prior: NfL (neurodegeneration) divergence.
• Symptomatic Phase: AChE elevation was observed only in symptomatic individuals, correlating with cholinesterase inhibitor therapy rather than disease pathology.

Gene-Specific Nuances:

• APP Carriers: Showed elevated Aβ38 and reduced Aβ40 compared to PSEN1 carriers and controls.
• PSEN1 Carriers: Showed reduced Aβ38 compared to controls.
• These findings contrast with sporadic AD, where Aβ42/40 ratios typically decrease; in ADAD, Aβ42 is often elevated early.

5. Significance and Implications

• Disease Staging: The sequential nature of these biomarkers (Amyloid $\rightarrow$ Tau $\rightarrow$ Astrocytosis/Synaptic $\rightarrow$ Neurodegeneration) provides a refined framework for staging presymptomatic AD.
• Clinical Trial Design: The data suggests that prevention trials targeting amyloid pathology should enroll participants at least 20–25 years before expected onset, while trials targeting tau or neurodegeneration may require different enrollment windows.
• Biomarker Selection: The study validates plasma pTau217 and pTau231 as the most robust early markers for distinguishing carriers from non-carriers. It also cautions against using plasma Aβ42/Aβ40 ratios for ADAD diagnosis without considering the specific mutation, as Aβ42 may be elevated rather than reduced.
• Generalizability: While the findings are specific to monogenic ADAD, the temporal sequence supports the "amyloid cascade" hypothesis and offers a template for investigating biomarker timing in sporadic AD, though validation in sporadic cohorts is required.

Limitations:
The study is limited by a relatively small sample size (though large for ADAD), lack of concurrent CSF or PET imaging for validation, and potential confounders like non-fasting samples and BMI. However, the results align with and refine previous findings from larger cohorts using different assay technologies.