Optimization of tumor-infiltrating lymphocyte (TIL) manufacturing in non-small cell lung cancer (NSCLC) using an IL-7/IL-15/IL-2 expansion protocol.

Abstract

2559 Background: Adoptive cell therapy with tumor-infiltrating lymphocytes (TILs) is a promising strategy in non-small cell lung cancer (NSCLC), although ex vivo expansion remains challenging due to variability in yield and product quality. Optimizing cytokine support during TIL expansion may improve manufacturing robustness while preserving a clinically relevant immunophenotype. Methods: Tumor tissue specimens were obtained from immediately resected NSCLC lesions and processed for TIL outgrowth and rapid expansion over 28 days using either standard IL-2 or an experimental IL-7/IL-15/IL-2 protocol. Manufacturing endpoints included expansion yield, viability and CD3⁺ proportion. Multiparametric flow cytometry assessed memory differentiation subsets and exhaustion-associated markers at baseline, pre-REP and post-REP. Clinical-pathological variables included disease stage, PD-L1 expression and prior systemic treatment. Results: 11 NSCLC TIL expansions were analyzed (experimental: n=7; IL-2: n=4). Median age was 73.1 years (range 48.5–92.3) and ECOG was 0–1 in 91% of patients; 64% had early-stage disease and 27% had advanced-stage (III–IV). PD-L1 was <1% in 73%, 9% showed PD-L1 1–49%, and none ≥50%. The cohort included pretreated patients, including prior anti–PD-1 exposure. Compared with IL-2, IL-7/IL-15/IL-2 resulted in a modest reduction in expansion yield (15–28%) and a lower CD4+/CD8+ ratio (6.62 vs 19.86), but showed improved manufacturing robustness, with all expansions meeting predefined quality criteria, whereas standard IL-2 expansions showed more variable product quality. Immunophenotypic profiling showed a more stable memory-oriented differentiation pattern with IL-7/IL-15/IL-2, including reduced naïve and TEMRA subsets and increased CD4+ TCM cells; CD8+ TCM remained predominant in both protocols. A polarisation towards the effector memory phenotype (CCR7⁻CD45RA⁻) was observed in the experimental protocol. CD28 remained highly expressed in both protocols, while CD134 tended to decrease with IL-7/IL-15. During pre-REP, PD-1 and TIM-3 were reduced in several subpopulations under the experimental protocol, suggesting a transiently less exhausted phenotype. After REP, these differences disappeared, with comparable levels between protocols. During REP, Th1 and Tfh increased in both protocols, while Th2 and Th17 remained low. Conclusions: Supplementation with IL-7 and IL-15 during NSCLC-derived TIL expansion improves consistency of product quality and supports a favorable memory-oriented and functional phenotype without increasing exhaustion markers at the end of manufacturing. This approach may enable more reliable generation of clinically applicable TIL products in heterogeneous and pretreated NSCLC populations, facilitating broader implementation of TIL-based therapies.