🧠 neuroscience

Brain network modeling with The Virtual Brain derives pharmacodynamics of ketamine

This study utilizes The Virtual Brain framework to implement a dose-dependent NMDAR antagonism model, revealing that low-dose ketamine primarily impairs excito-inhibitory transmission via disinhibition while high doses additionally affect excito-excitatory connections, thereby elucidating the distinct neural mechanisms underlying its varying clinical effects.

Original authors: Them, J., Deger, L., Taher, H., Stasinski, J., Martin, L. K., Meier, J. M., Stefanovski, L., Ritter, P.

Published 2026-02-25

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Them, J., Deger, L., Taher, H., Stasinski, J., Martin, L. K., Meier, J. M., Stefanovski, L., Ritter, P.

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 ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed.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

Imagine your brain as a massive, bustling city with millions of workers (neurons) constantly sending messages to keep everything running. Some workers are "Exciters" who shout, "Let's go!" and others are "Inhibitors" who whisper, "Calm down!"

Ketamine is a special chemical that acts like a master switch for a specific type of communication line in this city called the NMDAR. Usually, this line is blocked by a heavy gate (Magnesium ions) that only opens when the workers are already excited.

The big mystery scientists have is: How does the same drug cause such different effects depending on the dose?

Low Dose: It makes people feel happy, less anxious, and sometimes a bit "out of it" (dissociation).
High Dose: It knocks people out completely (anesthesia).

This paper uses a super-powerful computer simulation (called The Virtual Brain) to solve this mystery. Here is how they did it, explained simply:

1. The City Model

The researchers built a digital twin of the human brain. Instead of simulating every single neuron (which would take forever), they grouped them into "neighborhoods" (brain regions). Each neighborhood has three types of workers:

Pyramidal Cells (PCs): The main workers who do the heavy lifting.
Excitatory Interneurons (EINs): The cheerleaders who encourage the main workers.
Inhibitory Interneurons (IINs): The security guards who tell the main workers to stop.

2. The "One-Size-Fits-All" Mistake

First, the researchers tried a simple idea: What if Ketamine just weakens all communication lines equally, no matter the dose?

The Result: The simulation showed that the brain's rhythm slowed down (which is true for anesthesia), but the main workers' activity didn't behave like real humans do. In real life, low doses make the brain more active, but this simple model didn't show that. It was like trying to fix a car engine by turning down the volume on the radio; it didn't fix the actual problem.

3. The "Disinhibition" Breakthrough

Then, they tried a smarter approach based on a theory called Disinhibition. They realized that Ketamine doesn't treat all workers the same way.

The Low Dose Scenario (The "Security Guard" Theory):
Imagine the security guards (IINs) are very sensitive to Ketamine. When you take a low dose, the drug mostly targets these guards. It knocks them out or makes them sleepy.
- The Analogy: If you take away the security guards, the main workers (PCs) stop getting told to "calm down." They start shouting and working harder!
- The Result: The brain becomes more active and excited. This explains the antidepressant and "buzz" effects.
The High Dose Scenario (The "Total Shutdown" Theory):
As you increase the dose, the drug eventually overpowers the main workers (PCs) too.
- The Analogy: Now the drug is so strong it knocks out the main workers and the cheerleaders. Even though the guards are still asleep, there's no one left to do the work.
- The Result: The whole city shuts down. The brain goes quiet. This explains the anesthesia (knockout) effect.

4. The Computer's "Light Show"

The researchers watched how the brain's "light show" (brain waves) changed in their simulation:

Normal State: The brain hums with a steady, rhythmic beat (Alpha waves).
Low Dose: The beat gets faster and more chaotic (Gamma waves), and the rhythm shifts. The city is buzzing with energy.
High Dose: The fast rhythms die out, and the city slows down to a deep, slow slumber (Theta and Delta waves).

Why This Matters

This study is like having a flight simulator for drugs.

Before, doctors had to guess how Ketamine worked in the human brain by looking at rats or taking EEGs (brain scans) on humans. It was like trying to understand a plane crash by looking at a few pieces of wreckage.
Now, they have a virtual lab. They can test exactly how the drug interacts with different parts of the brain without hurting anyone.

The Takeaway:
Ketamine is a "smart" drug that changes its behavior based on how much you take. At low doses, it acts like a remover of brakes (taking away the inhibitors), making the brain rev up. At high doses, it acts like a brake pedal for the whole engine, shutting everything down.

This computer model helps scientists understand why this happens, which could lead to better treatments for depression, anxiety, and PTSD, and help doctors give the perfect dose to the right patient.

1. Problem Statement

Ketamine is a dissociative anesthetic and a rapid-acting antidepressant that functions as an N-Methyl-D-aspartate receptor (NMDAR) antagonist. Its effects are highly dose-dependent:

Low (sub-anesthetic) doses: Induce dissociation and have antidepressant/anxiolytic effects, often associated with increased cortical excitability and gamma-band activity.
High (anesthetic) doses: Induce anesthesia, characterized by reduced firing rates, decreased alpha-band activity, and increased slow-wave (theta/delta) activity.

The Core Challenge: While the "disinhibition theory" suggests that low doses preferentially block NMDARs on inhibitory interneurons (IINs), leading to increased pyramidal cell (PC) firing, it remains unclear how these microscopic molecular changes translate into large-scale brain dynamics across the entire dose spectrum. Existing computational models either focus on single neurons, specific circuits (cortico-thalamic), or fail to integrate the full dose spectrum (from low to anesthetic) within a single whole-brain framework.

2. Methodology

The authors utilized The Virtual Brain (TVB) framework to construct a whole-brain network model (BNM) that simulates the dose-dependent effects of ketamine.

Structural Connectome: The model is based on the Human Connectome Project (HCP) s900 release, comprising 82 cortical and subcortical regions of interest (ROIs) derived from the Desikan-Killiany and Fischl atlases. Structural connectivity (SC) was defined by fiber counts (weighted matrix $W$ ) and tract lengths ($TL$) to calculate conduction delays.
Neural Mass Model: Each brain region is modeled using the Jansen-Rit (JR) neural mass model. Each node consists of three subpopulations:
1. Pyramidal Cells (PCs)
2. Excitatory Interneurons (EINs)
3. Inhibitory Interneurons (IINs)
  These populations are coupled via local coupling parameters (LCPs): $c_{EI,PC}$ , $c_{PC,EI}$ , and $c_{II,PC}$ .
Modeling Ketamine Mechanisms:
- Ketamine's NMDAR antagonism was implemented as a reduction in the LCPs.
- Scenario A (Uniform Scaling): A linear trajectory where ketamine uniformly reduces all LCPs ( $c_{ij} = (1-Ket) \cdot c_{def}$ ).
- Scenario B (Selective Sigmoidal Scaling): Based on the disinhibition hypothesis, this scenario uses distinct sigmoidal transfer functions for different LCPs.
  - Low Doses: Preferentially reduces $c_{II,PC}$ (excitatory-to-inhibitory coupling) due to higher affinity of IIN NMDARs and their tonic activity removing the $Mg^{2+}$ block.
  - High Doses: As dose increases, $c_{EI,PC}$ and $c_{PC,EI}$ (excitatory-to-excitatory coupling) begin to decrease as PCs become more active and their NMDARs open.
Analysis Techniques:
- Bifurcation Analysis: Used auto-07p to analyze the dynamical landscape (fixed points vs. limit cycles) of single nodes under varying LCPs.
- Spectral Analysis: Computed Power Spectral Densities (PSD) for alpha (8–12 Hz) and theta (4–8 Hz) bands.
- Simulation: Ran 10 trials with stochastic noise across 1,000 equidistant ketamine dose values ( $Ket \in [0.0, 1.0]$ ).

3. Key Contributions

First Whole-Brain Dose-Spectrum Model: This is the first study to integrate ketamine's full dose spectrum (from sub-anesthetic to anesthetic) into a single whole-brain computational framework.
Mechanistic Validation of Disinhibition Theory: The study mathematically demonstrates that the "disinhibition hypothesis" (preferential IIN blockade at low doses) is necessary to reproduce the empirically observed increase in firing rates at low doses, which uniform antagonism fails to do.
Linking Micro to Macro: The model successfully bridges microscopic receptor-level interactions (NMDAR antagonism on specific cell types) to macroscopic electrophysiological signatures (alpha/theta power shifts and firing rate changes).
Bifurcation Insights: The authors provide a detailed bifurcation analysis showing how ketamine alters the stability of fixed points and limit cycles, explaining the transition from alpha oscillations to theta dominance.

4. Key Results

Failure of Uniform Antagonism: When ketamine was modeled as a uniform linear reduction of all LCPs, the model reproduced the slowing from alpha to theta bands but failed to reproduce the empirically observed increase in pyramidal cell firing rates at low doses. Instead, firing rates dropped monotonically or abruptly.
Success of Selective Sigmoidal Antagonism:
- Low Doses: The selective reduction of $c_{II,PC}$ (disinhibition) caused an initial increase in pyramidal cell firing rates (peaking at $Ket \approx 0.33$ ) and a decrease in alpha power, consistent with experimental data.
- High Doses: As the dose increased further, the reduction of excitatory-excitatory coupling ( $c_{EI,PC}, c_{PC,EI}$ ) took over, leading to a sharp decline in firing rates and a shift to dominant theta-band activity, mimicking anesthesia.
Spectral Dynamics:
- Alpha Band: Power decreased progressively with dose. Even at high doses, residual alpha power persisted due to noise-induced spiraling around stable fixed points, even after the limit cycles (oscillatory attractors) vanished.
- Theta Band: Power increased to a peak at intermediate-high doses ( $Ket \approx 0.36$ ) before declining, mirroring the transition to anesthesia.
Network Synchronization: Higher ketamine doses induced increased synchronization across the network, particularly in regions with high node strength, as the system moved toward stable fixed points.

5. Significance and Implications

Pharmacological Mechanism Decoding: The study confirms that the dose-dependent behavioral effects of ketamine (dissociation vs. anesthesia) arise from a specific sequence of NMDAR antagonism: first targeting inhibitory interneurons (disinhibition), then excitatory neurons.
Clinical Translation: By validating the disinhibition theory in a human-scale model, the work provides a framework for understanding how ketamine treats depression (via low-dose disinhibition) and how it induces anesthesia.
Precision Medicine Potential: The framework can be extended to use personalized connectomes, potentially predicting individual patient responses to ketamine therapy for treatment-resistant depression, bipolar disorder, or PTSD.
Future Directions: The authors note limitations, such as the model's inability to generate gamma-band oscillations (requiring a PING model extension) and the exclusion of ketamine's metabolites and non-NMDAR effects. However, the model establishes a robust foundation for testing other NMDAR-modulating therapeutics (e.g., memantine) and understanding NMDAR-related pathologies.

In conclusion, this paper demonstrates that computational whole-brain modeling, grounded in biophysical principles and bifurcation theory, is a powerful tool for disentangling complex, dose-dependent drug mechanisms and predicting their large-scale neural consequences.