Immunohistochemical and gene expression profiles as predictors of survival in recurrent high-grade glioma treated with intracranial nivolumab, ipilimumab, and autologous CD1c(BDCA-1) ⁺ /CD141(BDCA-3) ⁺ myeloid dendritic cells (m

2071 Background: Innovative treatments are needed for recurrent high-grade glioma (rHGG) patients (pts) as current salvage therapies fail to improve overall survival (OS). Immune checkpoint inhibitors lack efficacy in rHGG when administered IV. This single center, multicohort phase I trial (Glitipni, NCT03233152) investigated intracerebral (iCer) administration of ipilimumab (IPI) +/- nivolumab (NIVO) +/- myDC after maximal safe resection (MSR), followed by adjuvant intracavitary (iCav) IPI/NIVO through an Ommaya reservoir. Methods: Eligible pts (ECOG ≤ 2, ≤ 8mg/day methylprednisolone) with rHGG (WHO 2021 grade 3/4, IDH-1/2 wild type (wt) or mutant) after standard postoperative radiotherapy (RT) and temozolomide (TMZ) were included. Pts underwent MSR or stereotactic biopsy (if unresectable) < 24h after receiving NIVO IV (10mg), followed by iCer injection of varying doses of IPI +/- NIVO +/- myDC and Ommaya catheter placement depending on the cohort (C). NIVO was administered IV (all cohorts) +/- iCav (C3-7) bi-weekly up to 12 cycles. Baseline tumor microenvironment characteristics were assessed by immunohistochemical (IHC) analysis and gene expression profiling (GEP). Results: Between 2016 and 2023, 110 pts (68% male, median age 57, 92% ECOG 0/1) were enrolled. At primary diagnosis, the majority (85%) were glioblastoma pts (WHO grade 4, IDH-wt), treated with the standard of care (MSR + RT + TMZ) (71%). All pts received 10mg NIVO IV preoperatively. Ninety percent of the pts who underwent the neurosurgical procedure started the postoperative treatment. Early discontinuation of study treatment occurred in 76% of pts, mainly due to tumor progression (86%). Treatment-related adverse events (TRAE) were mild (CTCAE grade 1/2), no grade 5 TRAE occurred. Most frequent TRAE were fatigue, headache and fever. At database lock (Jan 1st, ‘25), 9 pts remained progression-free. When including durable benefit from bevacizumab at first progression (13 pts), PFS and OS were significantly higher in C5/6 (+myDC) compared to other cohorts (-myDC) of our trial with resectable rHGG, and to a historical control group treated with VEGF(R)-inhibitors (descriptive p < 0.05 for each pairwise comparison). Absence of B7H3 on resected tumor tissues as demonstrated by IHC (C4, 5, 7) showed longer median OS, which was consistent with GEP. PD-L1 expression and density of CD8, Granzyme B or FOXP3 positive cells/mm 2 did not correlate with survival. A proliferative gene signature on GEP was significantly correlated with shorter PFS and OS. Conclusions: Intracranial administration of IPI/NIVO co-administered with myDC was feasible and safe, resulting in encouraging survival in pts with resectable rHGG. Baseline B7H3 levels and a proliferative gene signature correlated with survival. Clinical trial information: NCT03233152 .