Network-based disease fingerprinting with neuroinflammation PET imaging

Area of Science:

• Neuroscience
• Medical Imaging
• Biochemistry

Background:

• Neuroinflammation is a key feature in many neurological and psychiatric conditions.
• 18 kDa translocator protein (TSPO) positron emission tomography (PET) is used to image neuroinflammation in vivo.
• Current TSPO PET quantification methods lack spatial relationship analysis, limiting disease-specific pattern detection.

Purpose of the Study:

• To develop a novel network-based approach for analyzing TSPO PET data.
• To capture disease-specific neuroinflammatory patterns by examining inter-regional pharmacokinetic similarity.
• To assess the potential of this network approach for disease classification and understanding pathophysiology.

Main Methods:

• A data-driven, network-based method was developed to create individual brain-wide TSPO PET matrices.
• Euclidean distance was used to quantify inter-regional pharmacokinetic similarity.
• The approach was applied to a large multicenter dataset (528 scans) using three TSPO tracers across healthy controls and various patient groups (multiple sclerosis, traumatic brain injury, schizophrenia, depression, chronic low back pain).
• Statistical modeling and machine learning were employed for analysis and classification.

Main Results:

• TSPO similarity patterns exhibited high biological specificity and reproducibility, with strong test-retest correlations (mean Spearman's ρ = 0.84).
• Disease classification accuracy significantly exceeded chance performance (23-89%), driven by condition-specific regional hubs.
• Feature importance analysis showed minimal overlap across conditions, highlighting specificity.

Conclusions:

• Network-based analysis of human TSPO PET data effectively detects disease-specific neuroinflammatory signatures.
• This methodology enhances the biological significance and translational value of TSPO PET imaging.
• The findings support precision medicine strategies for neuroinflammatory disorders.