Intralesional PD-1 blockade for oral cancer prevention: First-in-class phase 1 trial

Oral premalignant lesions affect 5% of the global population with transformation rates of 1-8% in mild-moderate dysplasia and up to 36% in severe dysplasia. Current management via surgical excision yields 30-40% recurrence despite negative margins, with cumulative functional morbidity. While systemic PD-1 blockade demonstrated biological activity in oral dysplasia, immune-related toxicity precludes use in non-cancer populations. We hypothesized intralesional delivery would achieve immune remodeling while eliminating systemic exposure. We conducted a phase 1, open-label, dose-escalation trial of intralesional nivolumab in patients with histologically confirmed oral epithelial dysplasia (NCT05327270). Twenty-nine patients received 10 mg or 20 mg intralesional nivolumab every three weeks for four cycles. Primary endpoints were safety and tolerability; secondary endpoints included lesion response, progression to carcinoma, pharmacokinetics, spatial immune profiling, and, uniquely for this population, prospective patient-reported outcomes (PROs). No dose-limiting toxicities occurred; 94% of adverse events were grade 1-2 with no systemic immune-related toxicities. Plasma nivolumab concentrations remained 10-fold below IV dosing without accumulation. Lesion area decreased 60% across cohorts with 41% histologic downgrading. Twelve-month cancer-free survival was 75.8%; all progression events were detected early and surgically salvageable. We performed the first prospective longitudinal analysis of PROs in oral premalignancy. High study completion (86%) and adherence, despite significant travel burden, coupled with stable or improved quality-of-life scores (specifically pain and swallowing), indicate that repeated intralesional injections are a feasible, non-toxic approach associated with no functional adversity. Unlike surgical standards that degrade function, this strategy preserved patient quality-of-life. Mechanistic analyses using spatial transcriptomics (Xenium) and multiplexed imaging (CODEX) revealed immune activation exclusively in treated lesions: increased CD4+ and CD8+ T cell infiltration, enriched CCR7+ activated dendritic cells, elevated CD103+ tissue-resident CD8+ T cells, and formation of higher-order immune assemblies. Untreated non-index lesions from the same patients showed no immune changes, definitively demonstrating anatomically restricted immune activation. PBMC profiling confirmed absence of systemic immune perturbation. This first-in-class trial demonstrates intralesional PD-1 blockade safely reprograms premalignant tissue immunity without systemic perturbation, establishing lesion-directed checkpoint inhibition as a viable cancer interception strategy. These findings have established the foundation for a randomized, placebo-controlled Phase 2 trial currently enrolling at MD Anderson Cancer Center (NCT06561087) and support broader applicability to accessible epithelial precancers including cervical, anal, and cutaneous dysplasia.