Convergent Multimodal Evidence of Cortical Excitation-Inhibition Imbalance in Psychosis

This study provides convergent multimodal evidence from resting-state fMRI and EEG that psychosis is characterized by a shift toward cortical hyperexcitability, evidenced by reduced Hurst exponents and aperiodic spectral exponents that align with specific molecular pathways involving GABA, potassium channels, and various neurotransmitter receptors.

The Gist

The Big Picture: A Brain That's "Too Loud"

Imagine your brain is a massive, bustling city. For the city to run smoothly, there needs to be a perfect balance between construction crews (excitatory neurons that build activity) and traffic police (inhibitory neurons that stop things from getting chaotic).

In people with psychosis (a condition that includes schizophrenia), scientists believe this balance is broken. The "construction crews" are working overtime, and the "traffic police" are taking a coffee break. The result? The city is in a state of hyper-excitability—too much noise, too much activity, and not enough control.

This study set out to find a way to "hear" this noise without needing to open up the skull. They used two different types of "microphones" to listen to the brain's rhythm in two separate groups of people: those with early psychosis and those with long-term psychosis.

The Two Microphones: fMRI and EEG

The researchers used two different tools to measure this brain activity:

1. The fMRI Microphone (The Long-Range Echo):

   • What it does: This measures how the brain's activity today relates to its activity in the past. Think of it like listening to a song. If the song has a very steady, predictable rhythm where every beat depends on the one before it, it has high "persistence."
   • The Finding: In healthy brains, the rhythm is steady and predictable. In the brains of people with psychosis, the rhythm was less persistent. It was more chaotic and less connected to the past.
   • The Analogy: Imagine a drummer. A healthy drummer keeps a steady beat that you can predict. A brain with psychosis is like a drummer who is rushing, skipping beats, and losing the groove. This "loss of groove" suggests the brain is too excited and lacks the braking power of inhibition.

2. The EEG Microphone (The Static Noise):

   • What it does: This measures the "background static" or the slope of the brain's electrical noise. In a healthy brain, the noise drops off smoothly as you look at higher frequencies (like a gentle hill).
   • The Finding: In people with psychosis, this "hill" was flatter. The brain was buzzing with too much high-frequency noise.
   • The Analogy: Think of a radio. A healthy brain is tuned to a clear station with a little bit of background static. A brain with psychosis is like a radio that has been turned up too high; the static is so loud it drowns out the signal. This "flatter" noise confirms the brain is in a state of over-excitement.

The "Aha!" Moment: It's the Same Problem, Different Tools

The most exciting part of this study is that both microphones heard the same thing.

• The fMRI showed the rhythm was chaotic.
• The EEG showed the noise was too loud.

Even though they looked at different groups of people (some with early illness, some with long-term illness) and used different machines, the result was the same: The brain's "brakes" are failing. This proves that the imbalance isn't just a fluke; it's a core feature of the disorder that persists over time.

Where is the Problem?

The researchers mapped exactly where this "chaos" was happening in the brain. It wasn't everywhere; it was concentrated in specific neighborhoods:

• The Sensory District: Areas that process what we see, hear, and feel.
• The Control Tower: Areas that help us focus and make decisions.
• The Relay Station: The thalamus, which acts as a switchboard for information.

The Analogy: It's like the traffic lights in the city center and the main highway are broken, causing gridlock and accidents, while the quiet suburbs (other parts of the brain) are still functioning normally.

The Molecular Clues: Why are the brakes failing?

The researchers then asked: What is the chemical reason for this? They compared the brain maps to a library of genetic blueprints and chemical maps.

They found that the chaotic areas were rich in specific genes and chemicals:

• Potassium Channels: Think of these as the "shock absorbers" for neurons. They help calm a neuron down after it fires. The study found that in the chaotic brain areas, these shock absorbers were missing or broken.
• GABA Receptors: These are the actual "brakes" (the traffic police). The study found a strong link between the chaos and the genes that build these brakes.

The Takeaway: The study suggests that the "traffic police" (GABA) and the "shock absorbers" (Potassium channels) are the specific parts of the brain's machinery that are malfunctioning in psychosis.

Did it predict symptoms?

Interestingly, the "chaos meter" (the brain measurements) did not predict how severe a person's hallucinations or negative thoughts were.

• The Analogy: Imagine a car engine that is revving way too high (the brain imbalance). You can measure the engine noise perfectly, but that noise doesn't tell you if the driver is angry, sad, or happy. The engine is broken in everyone, but the experience of the illness depends on other factors (like other chemicals or how the brain tries to compensate).

Why Does This Matter?

1. New Tools for Diagnosis: We now have two reliable, non-invasive ways (fMRI and EEG) to detect this "brain noise" without needing to guess based on symptoms alone.
2. Better Treatments: Since we know the problem is likely related to broken "shock absorbers" (potassium channels) or missing "brakes" (GABA), drug companies can focus on making medicines that fix these specific parts. For example, new drugs targeting these channels are already being tested.
3. A Universal Language: Because these brain rhythms are similar across humans and animals, scientists can test new drugs in mice and be more confident they will work in humans.

Summary

This study is like finding a universal "check engine light" for psychosis. By listening to the brain's rhythm and noise, the researchers confirmed that the brain is stuck in a state of over-drive because its braking system is broken. While this doesn't explain every symptom a patient feels, it gives scientists a clear target for fixing the engine and developing better treatments.

Technical Summary

1. Problem Statement

Psychosis is increasingly theorized to stem from a disruption in the cortical Excitation-Inhibition (E/I) balance, specifically a shift toward hyperexcitability. However, robust, non-invasive translational biomarkers to quantify this imbalance in humans are lacking.

• Current Limitations: Traditional neuroimaging metrics (e.g., fMRI functional connectivity, EEG power spectra, MRS metabolites) provide indirect or non-specific proxies for microcircuit dysfunction.
• Proposed Solution: Two emerging metrics offer complementary perspectives on E/I balance:
  1. Resting-state fMRI Hurst Exponent (HE): Quantifies the persistence of temporal dependencies in brain activity. Lower HE values indicate reduced inhibition and increased cortical excitability.
  2. EEG Aperiodic Spectral Exponent (1/f slope): Captures the broadband spectral structure of neural activity. A flatter slope (lower exponent) similarly indicates a shift toward reduced inhibitory drive.
• Knowledge Gap: These metrics have not been jointly examined in psychosis, and the spatial, molecular, and receptor-level architecture of HE alterations remains poorly defined.

2. Methodology

The study utilized a multimodal, cross-cohort design involving two independent datasets to test for convergent signatures of E/I imbalance.

Datasets & Participants

• fMRI Cohort (HCP-EP): 107 patients with early psychosis (within 3 years of onset) and 53 healthy controls.
  • Acquisition: 3T Siemens scanner, multiband EPI (TR=0.8s), 4 runs of 5 minutes (eyes open).
  • Assessments: PANSS (symptoms), NIH Toolbox (cognition).
• EEG Cohort (BSNIP2): 547 patients with chronic schizophrenia/schizoaffective disorder (>5 years duration) and 363 healthy controls.
  • Acquisition: 64-channel Quik-Cap, 1000 Hz sampling.
  • Assessments: PANSS, BACS (cognition).

Data Processing & Analysis

• fMRI Preprocessing: fMRIPrep v25.0, wavelet despiking, nuisance regression (motion, CSF), parcellation into 360 cortical + 66 subcortical regions (HCPex atlas).
• EEG Preprocessing: MNE pipeline, ICA for artifact removal, band-pass filtering (0.5–40 Hz), spectral parameterization using FOOOF to isolate the aperiodic 1/f component.
• Metric Estimation:
  • HE: Calculated via wavelet-based maximum likelihood estimator. Whole-brain HE was the average of regional HEs.
  • Harmonization: ComBat was used to minimize site effects (covariates: age, sex, diagnosis).
• Statistical Modeling:
  • Group differences assessed via Ordinary Least Squares (OLS) regression.
  • Biological Correlates: Regional HE differences were correlated with:
    • Cortical thickness (CT).
    • Gene Expression: Allen Human Brain Atlas (using PLSR and GOrilla for enrichment).
    • Receptor Density: 19 receptor maps (Hansen Atlas) using PLSR.
  • Clinical Correlates: Partial Least Squares Regression (PLSR) with 10-fold cross-validation to link HE/slope to symptoms and cognition.

3. Key Results

A. Primary Findings: E/I Imbalance Signatures

• fMRI (Early Psychosis): Patients showed significantly lower whole-brain HE compared to controls ($\beta = -0.031$, $d = -0.68$).
  • Regional Pattern: Widespread reductions (62% of cortical regions nominally significant). Strongest effects in somatomotor, insular-opercular, midcingulate, dorsolateral prefrontal, and superior parietal regions, as well as bilateral thalamus.
  • Medication: No significant difference between medicated and unmedicated patients; no correlation with chlorpromazine equivalent dose.
• EEG (Chronic Psychosis): Patients showed a significantly lower aperiodic 1/f exponent (flatter slope) compared to controls ($\beta = -0.101$, $d = -0.22$).
  • Convergence: Both modalities indicate a shift toward cortical hyperexcitability across different illness stages (early vs. chronic) and modalities (fMRI vs. EEG).

B. Biological Substrates

• Receptor Density: Regional HE differences were significantly predicted by receptor density maps ($r_\rho = 0.69$). The strongest contributors were Noradrenergic (NET), Cholinergic (VAChT), Serotonergic (5-HT1A, 5-HT2A), Glutamatergic (NMDAR), and Dopaminergic (D2) receptors.
• Gene Expression: Cortical gene expression patterns significantly predicted HE spatial distribution ($r_\rho = 0.55$).
  • Enrichment: Synaptic signaling, ion transport, and CNS development.
  • Key Pathways: Potassium channels (specifically KCNB2) and GABA receptors. Regions with lower KCNB2 expression showed larger illness-related HE differences.
• Structural Correlates: No significant correlation was found between HE reductions and cortical thickness reductions, suggesting HE captures functional/microcircuit dynamics distinct from gross structural atrophy.

C. Clinical Correlates

• Symptoms & Cognition: Neither HE (fMRI) nor the aperiodic slope (EEG) showed robust associations with PANSS symptom scores or cognitive performance in cross-validated models.
• Interpretation: E/I imbalance may represent a stable trait-like vulnerability rather than a state-dependent marker of current symptom severity.

4. Key Contributions

1. Multimodal Convergence: First study to demonstrate that reduced HE (fMRI) and flattened 1/f slope (EEG) converge as evidence of cortical disinhibition in psychosis across independent cohorts.
2. Spatial & Molecular Mapping: Mapped the spatial distribution of E/I imbalance to specific functional networks (Somatomotor, Salience) and identified specific molecular drivers (Potassium channels, GABA receptors, and monoaminergic systems).
3. Trait vs. State: Provided evidence that these metrics reflect a stable biological vulnerability (unaffected by medication status or symptom severity) rather than a transient state.
4. Translational Biomarkers: Validated HE and aperiodic slope as scalable, non-invasive biomarkers suitable for cross-species translation and precision psychiatry.

5. Significance and Implications

• Mechanistic Insight: The findings support the hypothesis that psychosis involves a fundamental failure of inhibitory control, likely driven by deficits in GABAergic interneurons and potassium channel function (e.g., KCNB2).
• Therapeutic Targets: The strong association with cholinergic and noradrenergic receptors supports the potential efficacy of muscarinic agonists (e.g., xanomeline-trospium/KarXT) and Kv3 channel agonists in restoring E/I balance.
• Clinical Utility: These metrics could serve as objective biomarkers for:
  • Early Detection: Identifying individuals at risk before full symptom onset.
  • Patient Stratification: Selecting patients likely to respond to specific E/I-modulating treatments.
  • Drug Development: Providing a quantitative endpoint for clinical trials targeting microcircuit dysfunction.

Limitations: The study is cross-sectional (limiting causal inference), relies on indirect measures of E/I balance, and uses postmortem transcriptomic data that may not fully generalize to living populations. Future longitudinal and pharmacological challenge studies are recommended.