🧠 neuroscience

Quantitative Modelling of Amyloid-β Dynamics in Brain, CSF, and Plasma During Sleep and Wakefulness

This study presents a quantitative model of Amyloid-beta dynamics across the brain, CSF, and plasma during sleep-wake cycles that reproduces observed oscillations, explains elevated CSF levels from repeated sampling as a result of pressure changes or impaired clearance, and underscores the protective role of sleep in A-beta regulation.

Original authors: Sangeet, S., Phillips, C. L., Hoyos, C. M., D'Rozario, A. L., Lucey, B. P., Postnova, S.

Published 2026-09-14

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Sangeet, S., Phillips, C. L., Hoyos, C. M., D'Rozario, A. L., Lucey, B. P., Postnova, S.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The brain is a busy organ that never truly shuts down, even when we are asleep. It constantly produces tiny protein fragments called amyloid-beta as a byproduct of its daily work. In a healthy brain, these fragments are cleared away efficiently, maintaining a delicate balance. However, when this cleaning process falters, the proteins can accumulate and clump together, forming the sticky plaques that are a hallmark of Alzheimer's disease. For years, scientists have known that sleep plays a critical role in this cleanup. During wakefulness, the brain's activity generates more of these proteins, while sleep seems to trigger a specialized flushing system that washes them out. Yet, the exact mechanics of how this cycle works across the brain, the fluid surrounding it, and the bloodstream remain difficult to pin down, largely because measuring these processes directly in living humans is invasive and technically challenging.

To bridge this gap, a team of researchers developed a computer model that simulates the movement of amyloid-beta through three connected spaces: the brain tissue itself, the cerebrospinal fluid that cushions the brain, and the blood plasma. This mathematical framework allowed them to recreate the natural 24-hour rhythm of protein levels in healthy individuals without needing to perform invasive procedures on every subject. The model successfully reproduced the observed patterns where protein levels in the fluid surrounding the brain and in the blood rise during the day and fall at night. Specifically, it showed that during a typical night of sleep, the concentration of these proteins in the fluid around the brain drops by about 13 percent, while levels in the blood decrease by roughly 16 percent. The simulation also provided a window into the brain tissue itself, predicting that protein levels there fluctuate between 240 and 280 units per milliliter, rising during the day and falling at night, a dynamic that is nearly impossible to measure directly in humans.

A significant portion of the study focused on a puzzling artifact found in previous human experiments. When researchers collect cerebrospinal fluid using a needle inserted into the lower back, a procedure known as a lumbar puncture, they often see protein levels rise unexpectedly on the second day of sampling, rather than following the normal nightly decline. The researchers used their model to test three different explanations for this anomaly. They considered whether the physical removal of fluid lowered pressure in the head, altering how proteins moved; whether the discomfort of the procedure disrupted sleep and thus the brain's natural cleaning cycle; or if both factors worked together. The simulations suggested that the most likely cause was a combination of both. The repeated removal of fluid appeared to change the pressure gradients, pushing more proteins from the brain into the fluid, while the disruption of sleep likely slowed down the system's ability to clear those proteins away. This finding implies that the rising protein levels seen in some studies are not a sign of the body failing to clear waste naturally, but rather a side effect of the measurement method itself.

The researchers also tested whether their model could explain the behavior of a slightly different version of the protein, known as amyloid-beta 40, which exists alongside the more commonly studied 42 form. Without needing to change the core rules of their simulation, the model accurately predicted the movements of this second protein type as well, suggesting that the brain uses the same general flushing mechanisms for both. The study concludes that while sleep is essential for keeping the brain clean, the tools we use to measure this process can sometimes interfere with the very thing we are trying to observe. By creating a virtual representation of these dynamics, the team has provided a way to distinguish between the brain's natural rhythms and the distortions introduced by medical procedures, offering a clearer path for understanding how sleep protects the brain from the buildup of toxic proteins.

Technical Summary: Quantitative Modelling of Amyloid-β Dynamics in Brain, CSF, and Plasma During Sleep and Wakefulness

Problem Statement
Amyloid-beta (Aβ) accumulation in the brain is a hallmark of Alzheimer's disease, yet the precise mechanisms governing its regulation during sleep-wake cycles in healthy individuals remain incompletely understood. While experimental evidence indicates that Aβ levels rise during wakefulness and decline during sleep, direct measurement of brain interstitial fluid (ISF) Aβ in humans is invasive and technically challenging. Furthermore, existing experimental data often conflates brain, cerebrospinal fluid (CSF), and plasma compartments due to the interconnected nature of these systems and artifacts introduced by invasive sampling procedures, such as repeated lumbar punctures. Previous mathematical models have explored Aβ kinetics but often lack a quantitative framework that explicitly simulates the dynamic, rather than static, effects of sleep-wake transitions on Aβ production and clearance across multiple compartments. Additionally, the day-to-day accumulation of Aβ observed in CSF during lumbar catheter studies contradicts the expected homeostatic stability of healthy individuals, necessitating a mechanistic explanation for these experimental artifacts.

Methodology
The authors developed a minimal, quantitative three-compartment ordinary differential equation (ODE) model representing the Brain, CSF, and Plasma. The model incorporates:

Production: Aβ production in the brain, modulated by a sleep-wake scaling factor ( $\sigma_A$ ) that reduces production during sleep based on decreased neuronal activity.
Clearance and Transport: First-order kinetics governing the transfer of Aβ from the brain to CSF (glymphatic pathway) and plasma (blood-brain barrier), and from CSF to plasma (blood-CSF barrier). These rates are modulated by sleep-dependent scaling factors ( $\sigma_{bc}, \sigma_{bp}, \sigma_{cp}$ ).
Systemic Removal: Clearance of Aβ from plasma via renal and hepatic pathways.

The model was calibrated against experimental time-series data from three studies involving cognitively healthy adults (Blattner et al., 2020; Lucey et al., 2018; Liu et al., 2023). To address the artifact of rising CSF Aβ levels caused by repeated lumbar sampling, the authors reconstructed "normal" sleep-wake dynamics using only the first 16 hours of wakefulness data, effectively detrending the upward drift.

Parameter constraints were derived from literature values, including plasma Aβ half-life (2–3 hours) and neuronal firing rate reductions during sleep. The model was fitted using a global optimization approach (MATLAB fmincon) minimizing the Normalized Root Mean Square Error (NRMSE). To investigate the mechanism behind CSF Aβ accumulation during lumbar puncture protocols, three hypotheses were tested by varying specific parameters:

Pressure Hypothesis: Altered transport rates due to intracranial pressure changes from CSF withdrawal.
Sleep Hypothesis: Disruption of sleep architecture impairing glymphatic and systemic clearance.
Combined Hypothesis: A synergistic effect of both pressure and sleep disruption.

Model selection was performed using the Corrected Akaike Information Criterion (AICc) to balance goodness-of-fit with model complexity. The framework was subsequently tested for generalizability to the Aβ40 isoform without additional parameter fitting.

Key Contributions and Results

Reproduction of 24-Hour Oscillations: The calibrated model successfully reproduces the experimentally observed 24-hour oscillations of Aβ42 in CSF (fluctuating between ~~660–760 pg/mL) and plasma (~~15–19 pg/mL). The model predicts brain Aβ dynamics, showing a ~15% oscillation between wake and sleep states, a compartment inaccessible to direct human measurement.
Mechanistic Explanation for Lumbar Puncture Artifacts: The study identifies that the day-to-day increase in CSF Aβ levels observed in lumbar catheter studies is likely an artifact of the procedure rather than a physiological phenomenon. The Combined Hypothesis (involving increased brain-to-CSF influx due to pressure gradients and decreased CSF-to-plasma/systemic clearance due to sleep disruption) provided the best statistical fit (lowest AICc) across datasets. Specifically, the model suggests that repeated sampling alters pressure-mediated transport (increasing brain-to-CSF flux) and impairs sleep-dependent clearance mechanisms.
Generalizability to Aβ40: The model successfully generalizes to Aβ40 dynamics by adjusting only the production rate (scaled by ~8.87) and systemic clearance rate (scaled by 1/1.44), maintaining the same transport pathways. This supports the view that bulk flow clearance mechanisms (glymphatic and CSF circulation) are non-selective between Aβ40 and Aβ42 in amyloid-negative individuals.
Physiological Insights: The model quantifies that while the brain shows modest Aβ fluctuations (~15%), CSF acts as a regulatory interface with competing fluxes balancing the system. It also estimates that the blood-brain barrier contributes approximately 26–37% of total brain efflux, with the majority occurring via the brain-to-CSF route.

Significance
The paper claims to provide a minimal yet mechanistically interpretable framework for understanding Aβ regulation under healthy physiological conditions. Its primary significance lies in:

Non-invasive Insight: Offering a method to infer brain Aβ dynamics, which are otherwise experimentally inaccessible in humans.
Artifact Identification: Providing a quantitative explanation for why lumbar puncture protocols induce apparent Aβ accumulation, highlighting the need for non-invasive measurement approaches or careful interpretation of invasive data.
Sleep's Protective Role: Reinforcing the quantitative relationship between sleep-wake cycles and Aβ clearance, suggesting that sleep disturbances could disrupt specific clearance pathways (glymphatic and systemic) leading to accumulation.
Foundation for Pathology: Establishing a baseline model of healthy dynamics that can serve as a reference for future investigations into pathological states, such as Alzheimer's disease, and the long-term effects of chronic sleep disruption.

The authors maintain a modest stance, acknowledging that the model represents effective rate constants rather than isolated biological mechanisms and that future extensions are required to model bidirectional transport in disease states or long-term plaque accumulation.

More like this