Phase 2 trial of TIL therapy for metastatic uveal melanoma: Evaluation of T cell potency and an in situ precision biomarker.

2512 Background: Uveal melanoma (UM) is a rare eye cancer that frequently metastasizes to the liver and responds poorly to systemic therapies. We previously reported that adoptive transfer of autologous tumor-infiltrating lymphocytes (TIL) produced a 35% objective response rate (ORR) in metastatic UM, with infusion product tumor reactivity being the strongest predictor of response. To guide surgical harvest of metastases enriched for tumor-reactive TIL, we developed Uveal Melanoma Immunogenomic Score (UMIS), an in situ tumor transcriptomic biomarker. Here we present results of a phase 2 trial designed to validate the efficacy of TIL therapy in metastatic UM and prospectively evaluate predictors of response. Methods: In this single-center phase 2 trial (NCT03467516) metastatic UM patients underwent metastatectomy to generate early TIL cultures. Cultures demonstrating growth potential and autologous anti-tumor reactivity were expanded for infusion. Patients received non-myeloablative lymphodepletion with cyclophosphamide and fludarabine, followed by infusion of TIL and high-dose interleukin-2. The primary endpoint was ORR per RECIST v1.1. Exploratory analyses to identify response predictors were performed in the current cohort and a pooled dataset including the NCT01814046 cohort. Results: Thirty-three UM patients received TIL therapy. Most had high disease burden (82% M1b/M1c; 76% elevated LDH; 70% hepatic and extrahepatic metastases; 24% CNS), received a median of 2 prior metastatic treatments (64% checkpoint blockade, 33% tebentafusp), and had liver metastases as their TIL source (52%). Infusion products had a median of 5.24E10 TIL, were CD8-enriched (median 68%), and demonstrated autologous tumor reactivity in 61% (19/31). Among 32 evaluable patients, the ORR was 22% (7/32) with a median response duration of 10.8 months (maximum ongoing at 58 months). Tumor-reactive TIL conferred higher response rates (37%, 7/19) than nonreactive products (0%, 0/12; P = 0.026). Responders demonstrated higher persistence of infused TIL clones in peripheral blood (R/NR ratio = 10.69, P = 0.020). In the pooled cohort (n = 54, 52 evaluable), ORR was higher with reactive products (43%, 12/28) versus nonreactive products (0%, 0/20; P = 5.06E-4). UMIS of the source metastases predicted growth of tumor-reactive TIL cultures (Rho = 0.55, P = 2.72E-5), final infusion product reactivity (P = 0.040), and clinical response (P = 0.004). Conclusions: TIL therapy demonstrates clinically meaningful activity in advanced, pretreated metastatic UM. Infusion product tumor reactivity was the strongest predictor of response. UMIS identified metastases enriched with tumor-reactive TIL and predicted final product potency and clinical outcome. These findings support implementation of UMIS as a minimally invasive precision biomarker to guide patient and tumor selection for optimal therapeutic benefit. Clinical trial information: NCT03467516 .