USC Launches Study Targeting APOE4 Brain Inflammation

The Norman and Mary Pattiz Foundation gave USC's Keck School of Medicine a $3 million gift, announced June 2, 2026, to fund Alzheimer's prevention research at the USC Center for Personalized Brain Health (CPBH), directed by Hussein Yassine, MD. The gift backs AI-driven drug discovery targeting calcium-dependent phospholipase A2 (cPLA2), an enzyme Yassine's team found is elevated in APOE4 carriers who progress to dementia; it also funds a high-risk early-detection registry combining APOE4 genetic data with cardiovascular risk factors, and gives neuropathologist Anne Hiniker, MD, PhD, protected time to screen more than 1,100 human brain tissue samples in the USC Alzheimer's Disease Research Center's Neuropathology Core.

The AI-screening work the gift supports builds on V-SYNTHES, a computational platform developed by Vsevolod Katritch at the USC Michelson Center for Convergent Bioscience that combines AI with physics-based molecular simulation. Per USC Today, Katritch's team can computationally narrow trillions of possible small molecules down to 100 to 200 synthesizable candidates likely to bind a given target; applied to cPLA2, the Yassine-Katritch collaboration screened billions of compounds for blood-brain-barrier penetration and enzyme binding and identified leading drug candidates in under six months. The cPLA2 program already carries a separate NIH grant supporting translation into a drug therapy with Stan Louie, professor of clinical pharmacy at USC's Mann School of Pharmacy.

Teams building AI screening pipelines for CNS targets should note what made this workable: a biologically specific, mechanistically validated target (cPLA2's role in APOE4-linked inflammation) narrowed the search space before AI screening began, and a hybrid AI-plus-physics approach, rather than AI alone, was used to filter for blood-brain-barrier permeability, a property still hard to predict from structure alone. The harder, unresolved work is orthogonal validation: confirming target engagement in human tissue and moving from computationally ranked candidates to experimentally confirmed hits.

Watch for peer-reviewed preclinical data on the cPLA2 inhibitors, since the current record consists of USC's own announcements rather than published, externally reviewed results; and watch how the new early-detection registry (drawing on USC's GeneScreen and CPBH SPARK cohorts) and the expanded 1,100-sample tissue library get used to generate training labels for future inflammation-focused AI models.