IL-12/anti-PD-1 armored oncolytic HSV-1 reprograms CNS immunity: Integrated longitudinal immune, genomic, and metabolic CSF profiling in the MVR-C5252 PuMP Trial

Introduction/Rationale: Glioblastoma (GBM) remains refractory to immunotherapy and exhibits low basal IL-12 and IFN-gamma levels relative to other solid tumors, reflecting an immunologically quiescent microenvironment. MVR-C5252 is a replication-competent HSV-1 engineered to deliver IL-12 and an anti-PD-1 antibody fragment, coupling oncolysis with Th1 polarization and localized checkpoint blockade to overcome this resistance. Methods: Stage 1 of PuMP (NCT06126744) assessed safety in six adults with recurrent IDH-wildtype GBM treated with a single convection-enhanced intratumoral infusion (5×10⁶ or 1×10⁸ PFU). An Ommaya reservoir enabled CSF sampling at baseline, 1 hour, and 28 days. Serial CSF single-cell RNA sequencing (scRNA-seq) assessed immune cells, and whole-genome sequencing (WGS) quantified ctDNA and viral kinetics. Liquid chromatography-mass spectrometry (LC-MS)-based metabolomics interrogated IFN-gamma-linked tryptophan/kynurenine and arginine/NO pathways. Stage 2 (ongoing) incorporates repeat dosing via an implanted pump, and Stage 3 delivers 12 doses in a Bayesian Optimal Interval (BOIN) dose-escalation design, with the selected dose and schedule planned for expansion in Stage 4. Results: Six patients (3M/3F; 50-59 years; KPS 90 in 5, 80 in 1; MGMT unmethylated in 5) were treated in Stage 1. Median PFS was three months and median OS was 6.1 months, with two patients alive at last follow-up. MVR-C5252 was well tolerated (grade 1-2 only) with no PCR-detectable shedding in urine or saliva. scRNA-seq in four patients with CSF from all three timepoints yielded 14,049 high-quality cells and diverse T-cell and myeloid populations. CD8⁺ subsets included cytotoxic and exhausted states, while CD4⁺ subsets spanned naïve, memory, effector, and regulatory states. TAM-like microglia and macrophages were also identified. Rapid early IFN-driven activation and monocyte/dendritic-cell recruitment were observed, followed by sustained cytotoxic and myeloid remodeling at 28 days. CD8⁺ populations showed increased effector states and fewer exhausted cells, while CD4⁺ T-cell states spanned naïve and helper lineages. Tumor and viral genome kinetics were characterized by WGS, immune-metabolic pathways were quantified by LC-MS-based metabolomics, and integrated CSF-tumor-blood analyses will be presented. Conclusions: Stage 1 demonstrates safety and immunobiological activity of MVR-C5252, with a single infusion inducing immune remodeling that was not sustained. Meaningful antitumor activity is expected to require repeated intratumoral dosing, as planned in Stages 2-4. These findings establish a mechanistic human model in which localized IL-12/anti-PD-1 virotherapy shifts the tumor microenvironment from a quiescent to a Th1-polarized, cytotoxic state, providing a foundation for next-phase efficacy evaluation.