The Prune-Without-Repair Model for Schizophrenia Cognitive Impairment: Evidence from Convergent GWAS Re-Analyses

Background: Schizophrenia is driven by many common variants, and two biological themes—excessive synaptic pruning and reduced glutamatergic transmission—feature prominently in current models. Yet these mechanisms do not fully account for the early-emerging, severe cognitive difficulties seen in affected individuals. To examine how pruning and plasticity signals diverge or overlap with cognition, we contrasted their genetic footprints in schizophrenia and in general intelligence.Methods: Using identical analytic steps, we processed summary statistics from the Psychiatric Genomics Consortium Wave 3 schizophrenia genome-wide association study and a large-scale intelligence study. The pipeline combined three approaches: (1) MAGMA for competitive gene-set enrichment, (2) stratified LD-score regression to partition heritability, and (3) S-PrediXcan to infer transcriptome-wide associations. Seven predefined gene panels anchored the work: three capturing pruning biology, two plasticity and two controls.Results: All three methods converged on a robust enrichment of pruning genes in schizophrenia. For the shortened pruning panel, MAGMA yielded a Bonferroni-corrected p = 1.3 × 10⁻⁵ and LD-score regression indicated extreme enrichment (p ≈ 10⁻¹⁷⁹). Signal persisted after glutamatergic genes were removed, and S-PrediXcan suggested up-regulated expression of key complement components such as C4A. In contrast, glutamatergic pathways showed only modest schizophrenia involvement.Conclusions: A double dissociation emerges. Schizophrenia risk aligns mainly with overactive pruning, whereas successful cognitive performance depends more on balanced glutamatergic-driven plasticity. We outline a “prune-without-repair” model in which unchecked complement activity, combined with weak glutamatergic stabilization, progressively undermines cognitive circuits in schizophrenia.