A Four-Tier Computational Model of Microglial Pruning, Mitochondrial NAD⁺ Compensation, and Autophagy Collapse in ALS: Testing a Multi-Target Glutamatergic Regimen

Amyotrophic lateral sclerosis (ALS) remains a disease with few disease-modifying options and a short survival window. Recent genomic and pathological findings have suggested that excessive microglial synaptic pruning may represent a shared upstream vulnerability with major depressive disorder, with the ALS trajectory then intensified by mitochondrial bioenergetic stress and autophagy failure. A series of five 2026 preprints by Cheung proposed a four-tier mechanistic continuum linking these processes. The present study implemented that framework in a mesoscale neural-network model to examine whether the proposed cascade could reproduce clinical ALS trajectories and whether a multi-target Augmented Cheung Glutamatergic Regimen (aCGR) could modify progression relative to riluzole or ketamine-like interventions.The model incorporated eight circuit layers, including sub-layer branching for bulbar, respiratory, and fine-motor compartments. Tier 1 pruning was modeled using complement bias plus activity dependence. A per-layer NAD⁺ buffer represented mitochondrial compensation and depletion at Tier 2, while dynamic autophagy efficiency and weight corruption were used to represent aggregate spread and clearance failure at Tier 3. All downstream symptoms, together with the neurofilament light chain proxy, emerged directly from network state. Sixty scenarios were run across standard ALS, subtype, and comorbid profiles.In the untreated arm, sparsity accumulated rapidly after cycle 6, NAD levels declined, and the ALSFRS-R proxy fell below 12 by mid-stage. Riluzole and ketamine delayed deterioration modestly but did not prevent eventual NAD burnout or near-total sparsity. Both early-pulsed and maintenance aCGR schedules were associated with lower final sparsity, at 86%–87%, reduced NFL proxy values, and better preserved NAD levels than the other treatment arms. Subtype and comorbid simulations showed similar directional effects, with numerically lower depression scores in the comorbid aCGR arm.These computational findings suggest that multi-tier targeting may delay the critical sparsity threshold. The model also yields several testable predictions for future pilot work, including slower ALSFRS-R decline and biomarker shifts in genetically stratified patients. Validation in iPSC systems and careful early-phase trials would be necessary before any clinical use could be considered.