PRELIVE: A Framework for Predicting Lipid Nanoparticles In Vivo Efficacy and Reducing Reliance on Animal Testing

Abstract Lipid nanoparticles (LNPs) have emerged as transformative delivery vehicles for nucleic acid therapeutics, yet their development remains heavily dependent on extensive and iterative animal testing to assess their in vivo performance. Here, PREdicting LNP In Vivo Efficacy (PRELIVE) is presented, a predictive framework that reduces reliance on animal studies by enabling rational design of organ‐targeted LNPs. Using a diverse set of LNPs with varying compositions and properties, LNP composition‐based models are developed that predict organ‐specific functional delivery with high accuracy. The models reveal distinct design spaces for each organ, revealing the optimal LNP compositions for targeting liver, spleen, kidney, bone marrow, lung, heart, brain, and blood. Protein corona (PC)‐based models are additionally developed using 220 identified proteins, achieving comparable predictive performance and revealing unique in vitro corona fingerprints that correspond to specific in vivo functional delivery. To accelerate translation and adoption across the scientific community, interactive in vivo design spaces are shared allowing researchers to explore LNP composition‐activity relationships in real‐time and optimize formulations for the desired organ tropism. This proof‐of‐concept study demonstrates that careful in vivo study design combined with predictive modeling can effectively reduce animal testing while enabling more rational design of organ‐specific nanomedicines.