T cell receptor repertoire dynamics and newly identified functional regulators of autologous tumor-infiltrating lymphocyte therapy in advanced solid tumors

Abstract

Background

Tumor-infiltrating lymphocyte (TIL) therapy has demonstrated clinical potential in melanoma; however, its applicability across diverse solid tumors in Asian patients remains unclear. Identification of molecular targets to enhance the antitumor activity of TIL is urgently needed.

Materials and methods

In a phase I investigator-initiated trial, we evaluated the safety, feasibility, and preliminary efficacy of autologous TIL therapy in twelve patients with advanced melanoma, cervical, lung, or head and neck cancers. Twelve patients who had progressed after standard therapies were enrolled between August 2022 and December 2024. Patients received lymphodepletion with cyclophosphamide (30 mg/kg) for 2 days, followed by fludarabine (25 mg/m²) for 5 days, approximately 24 h before intravenous infusion of autologous TIL. High-dose interleukin-2 (IL-2) was administered for 6–12 days to support T-cell survival and expansion. After a 28-day safety observation period, tumor assessments were performed every 6 weeks for the first 6 months and thereafter every 12 weeks during long-term follow-up, according to RECIST 1.1 criteria. T cell receptor (TCR) sequencing was conducted to evaluate in vivo persistence of infused TIL. Differential gene expression analysis was performed using bulk RNA sequencing on TIL from responders and non-responders. Target gene knockout in TIL was achieved using CRISPR/Cas9 technology to assess enhancement of antitumor function.

Results

The most common adverse events were fever, anemia, nausea, hypertension, and hyponatremia. The objective response rate (ORR) was 33.3% (4/12), including one complete response (CR) and three partial responses (PR), with a disease control rate (DCR) of 75% (9/12). Infused TIL persisted in peripheral blood and induced a reversal in the peripheral CD4 + T to CD8 + T cell ratio. TCR sequencing revealed dynamic clonal remodeling in responders. Transcriptomic profiling identified ASS1 and CEP20 as genes negatively associated with therapeutic efficacy. CRISPR/Cas9-mediated knockout of ASS1 or CEP20 enhanced TIL memory phenotypes, cytokine production, and tumor cytotoxicity both in vitro and in vivo.

Conclusions

This study demonstrates the efficacy of TIL therapy in diverse solid tumors among Asian patients, reveals dynamic TCR clonal remodeling in responders, and identifies ASS1 and CEP20 knockout as strategies to potentiate antitumor activity. These findings provide mechanistic insights that may guide improvement of TIL therapy for broader clinical application. Trial registration: NCT, NCT05366478. Registered 01 August 2022, http://www.clinicalTrial.gov/NCT05366478.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-08781-z.

Highlights

TIL therapy shows clinical activity in diverse solid tumors of Asian patients with a 33.3% objective response rate (ORR) and 75% disease control rate (DCR).

T cell receptor (TCR) sequencing revealed dynamic clonal remodeling correlates with efficacy of TIL therapy.

Transcriptomic profiling identifies ASS1 and CEP20 as negative regulators of TIL function.

CRISPR/Cas9-edited TIL exhibit enhanced persistence, cytokine secretion, and tumor control in vivo and in vitro.

This study links clinical and molecular findings to enable reverse translational improvement of TIL therapy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-08781-z.

Introduction

Adoptive cell therapy (ACT) [1] utilizing tumor-infiltrating lymphocyte (TIL) [2] has emerged as a promising immunotherapeutic strategy, particularly demonstrating clinical efficacy in metastatic melanoma. A landmark in the development of TIL therapy was the recent FDA approval of lifileucel (Amtagvi), the first TIL-based product, for the treatment of patients with metastatic melanoma who have progressed on PD-1 blockade therapy. TIL therapy involves the isolation and ex vivo expansion of autologous T cells derived from resected tumor tissue, followed by reinfusion into the patient, with the goal of inducing tumor regression [3]. Although TIL therapy has achieved notable success in melanoma, its application in a broader range of solid tumors remains under active investigation.

Recent clinical trials have begun to evaluate the feasibility and efficacy of TIL therapy across various solid malignancies [4]. Despite these advances, several challenges limit the widespread implementation of TIL therapy, including the labor-intensive and time-consuming process of TIL expansion [5], variability in TIL quality and functional potency [6–8], and the immunosuppressive tumor microenvironment [9–12], which may impair TIL activity and persistence. To address these limitations, novel strategies are being explored to enhance TIL function and durability. One such approach involves the application of CRISPR/Cas9 gene-editing technology to genetically modify TIL, with the aim of augmenting their antitumor activity [13–16]. Preclinical evidence indicates that disruption of genes associated with T cell exhaustion or inhibitory signaling pathways can enhance T cell cytotoxicity and persistence. For instance, knockout of the adenosine A2A receptor [14] in CAR-T cells has been shown to improve effector function and tumor-killing capacity. These findings suggest that CRISPR/Cas9-mediated gene editing may offer a viable strategy for optimizing T cell-based therapies.

In this context, we conducted an investigator-initiated, phase I clinical trial to assess the safety, feasibility, and preliminary efficacy of autologous TIL therapy in Asian patients with advanced solid tumors, including melanoma, cervical, lung, and head and neck cancers. We included different tumor types for the study in the Asian population that aim to gain a more comprehensive understanding of how TIL behave across a wide range of malignancies in the Asian context. It allows for the exploration of potential commonalities and differences in TIL biology among various tumors. This can help in the development of more generalized immunotherapeutic strategies that may be applicable to multiple cancer types. Additionally, we performed T cell receptor (TCR) sequencing to monitor clonal dynamics and conducted transcriptomic profiling of the infused TIL products to identify molecular signatures associated with therapeutic response. Subsequent CRISPR/Cas9-mediated knockout of candidate genes, including ASS1 and CEP20, was carried out to evaluate their functional roles in modulating TIL activity. This reverse translational approach aims to guide the development of genetically engineered TIL products, with the potential to enhance the efficacy of adoptive cell therapy across diverse solid tumors [17].

Methods and materials

Data reporting

The present study was non-randomized, no statistical methods were used to predetermine sample size, and blinding was not performed by all investigators during experimental procedures and outcome assessments.

Clinical trial design and treatment

The primary objective of the trial was to evaluate the safety and tolerability of functionally enhanced autologous tumor-infiltrating lymphocyte (TIL) in the treatment of solid tumors, including melanoma, non-small cell lung cancer, and cervical cancer. The secondary objectives were to assess the pharmacokinetic and pharmacodynamic characteristics of TIL in the treatment of solid tumors and to observe the preliminary clinical efficacy of functionally enhanced autologous TIL in vitro and in vivo. Preliminary antitumor activity was described using objective response rate (ORR), disease control rate (DCR), and progression-free survival (PFS), according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 and the immune-based Response Evaluation Criteria in Solid Tumors (iRECIST). Patients enrolled in the study received lymphodepleting chemotherapy consisting of cyclophosphamide (30 mg/kg) for 2 consecutive days, followed by fludarabine (25 mg/m²) for 5 consecutive days, approximately 24 h prior to intravenous infusion of autologous TIL. Subsequently, patients received high-dose interleukin-2 (IL-2) for 6 to 12 days to support T-cell persistence and expansion. Tumor assessments were performed at 6 weeks post-TIL infusion and every 6 weeks thereafter.

Patient eligibility

The inclusion criteria for patient selection were as follows: (1) Age ≥ 18 years, regardless of gender; (2) Diagnosis of histologically confirmed solid tumors, including American Joint Committee on Cancer (AJCC) stage III or IV melanoma, non-small cell lung cancer, cervical cancer, or other malignancies for which no standard or effective therapy is available; (3) Patients who had previously collected tumor tissue and met manufacturing requirements must have measurable lesions for efficacy evaluation; patients without prior tumor collection must have residual lesions amenable to surgical resection (> 0.5 cm³), with measurable disease post-resection for both TIL generation and efficacy assessment; (4) At least 2 weeks must have elapsed between the last dose of prior chemotherapy, radiotherapy, biotherapy, or immunotherapy and enrollment; for oral tyrosine kinase inhibitor (TKI) targeted therapy, a minimum of 7 days must have passed; 5) Laboratory parameters must meet the following requirements:- Hematology: lymphocyte count ≥ 0.8 × 10⁹/L; neutrophil count > 1.0 × 10⁹/L; white blood cell count > 3.5 × 10⁹/L and < × 10⁹/L; platelet count > 80 × 10⁹/L; hemoglobin > 100 g/L; - Liver function: alanine aminotransferase (ALT) and aspartate aminotransferase (AST) ≤ 2.5 × upper limit of normal (ULN), or ≤ 5 × ULN in the presence of liver metastasis; alkaline phosphatase (ALP) ≤ 2.5 × ULN; total bilirubin (TBIL) ≤ 1.5 × ULN; - Renal function: urea ≤ 1.5 × ULN; serum creatinine (Cr) ≤ 1.5 × ULN; 6) Echocardiography demonstrating left ventricular ejection fraction (LVEF) ≥ 50%; 7) Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1; 8) Expected survival longer than 3 months; 9) Ability to comply with study procedures and follow-up requirements; 10) Provision of written informed consent prior to any study-specific procedures, indicating understanding of and voluntary agreement to participate in the clinical trial.

The exclusion criteria for patient selection in the trial were as follows: (1) Active or pre-existing autoimmune diseases with potential for recurrence (e.g., systemic lupus erythematosus, rheumatoid arthritis, inflammatory bowel disease, autoimmune thyroid disease, multiple sclerosis, vasculitis, glomerulonephritis, psoriasis, uncontrolled asthma); (2) Severe hepatic or renal dysfunction, severe cardiac disease (such as NYHA class III or IV heart failure), coagulation disorders, or hematopoietic dysfunction; (3) Known psychiatric or neurological disorders that may impair the ability to comply with trial procedures; (4) Concurrent systemic active infections that cannot be effectively controlled; (5) Active central nervous system metastases and/or malignant meningitis; 6) Uncontrolled pleural effusion, pericardial effusion, or ascites requiring repeated drainage; 7) Requirement for systemic corticosteroid therapy; 8) Positive hepatitis B surface antigen (HBsAg), positive hepatitis C virus (HCV) antibody, positive HIV antibody, or positive syphilis serology; 9) Persistent adverse reactions from prior anti-tumor therapies not recovered to ≤ Grade 1 (excluding toxicities deemed without safety risk by investigators, such as alopecia, stable hypothyroidism on hormone replacement therapy); 10) Primary or secondary immunodeficiency not yet resolved; 11) Patients with familial genetic disorders or chromosomal abnormalities; 12) Participation in other clinical trials within 8 weeks prior to the first TIL infusion; 13) Receipt of any vaccine within 4 weeks prior to the first TIL infusion; 14) Known history of alcohol or substance dependence; 15) History of severe hypersensitivity to biologic products; 16) Planned pregnancy (in male or female patients), pregnant or breastfeeding women; 17) Any condition deemed by the investigators to potentially increase subject risk or interfere with study outcomes.

TIL manufacturing

Tumor-infiltrating lymphocyte (TIL) used for patient treatment or in vivo/in vitro studies were manufactured by Suzhou Lanma Biotechnology Co., Ltd., under standard Good Manufacturing Practice (GMP) conditions as previously described [18]. Briefly, freshly resected tumor specimens were placed in sterile RPMI 1640 medium and transported to the GMP facility. Tumors were dissected into fragments of 2–3 mm³ and seeded into culture vessels (5–10 fragments per vessel) in medium supplemented with 6,000 IU/mL interleukin-2 (IL-2) and 10% human AB serum. When TIL reached a density of ≥ 2 × 10⁶ cells/mL, they were split into new culture vessels. Medium containing 6,000 IU/mL IL-2 was replenished every 2–5 days to support TIL expansion over 10–14 days using a pre-rapid expansion protocol (pre-REP). TIL from pre-REP cultures were cryopreserved and later thawed for rapid expansion (REP) upon patient readiness for treatment. For REP, TIL were cultured in medium containing anti-CD3 antibody (ACRO Biosystems), irradiated feeder cells, and 6,000 IU/mL IL-2 over a 14-day period. During REP, TIL were split into new vessels every 2–3 days upon reaching ≥ 2 × 10⁶ cells/mL. Pathogen testing was performed on day 0, day − 7, and day − 14 relative to the initiation of REP. After 14 days of REP, TIL were harvested, resuspended in infusion medium, and formulated as the final product for infusion.

TIL gene-editing

CRISPR/Cas9 ribonucleoprotein (RNP) complexes were formed by incubating a mixture of 40 pmol Cas9 (Kactus Biosystems) and 80 pmol synthetic gRNA targeting ASS1 or CEP20 (GenScript) at room temperature for 15 min. TIL were collected after the pre-pre-REP step, washed twice with phosphate-buffered saline (PBS), and resuspended in P3 Buffer (Lonza) at a concentration of 5 × 10⁶ cells in 20 µL. The cell suspension was then mixed with RNP complexes. Electroporation was performed using a 4D-Nucleofector (Lonza) with pulse code EO-110 to deliver RNP into TIL. Electroporated TIL were allowed to recover for 1 h before initiation of the rapid REP as described above. ASS1 knockout TIL (ASS1 KO-TIL) and CEP20 knockout TIL (CEP20 KO-TIL) were harvested and resuspended in infusion medium for subsequent in vitro or in vivo patient-derived xenograft (PDX) mouse model studies.

Enzyme-linked immunosorbent assay (ELISA)

Peripheral blood was collected from patients at specified time points according to the study protocol. Plasma was isolated by centrifugation and used for cytokine analysis. The concentrations of multiple cytokines were measured using pre-coated ELISA kits according to the manufacturers’ instructions (IFN-γ: Biolegend 430107; TNF-α: Dakewe 1117202; IL-2: Biolegend 431802; IL-6: Dakewe 1110602; IL-7: Multi Sciences 70-EK107-96; IL-10: Dakewe 1111002; IL-13: Multi Sciences 70-EK113-96; IL-21: Biolegend 433807; TGF-β1: Dakewe 1117102). Absorbance was measured at 450 nm using a SpectraMax 190 Microplate Reader (Molecular Devices, LLC), and cytokine concentrations were calculated using SoftMax Pro 7.1.2 software (Molecular Devices, LLC).

Flow cytometry analysis

At the end of the REP, ≥ 1 × 10⁶ TIL per staining sample were collected and washed twice with 1× PBS. Cells were resuspended in 100 µL staining buffer (PBS containing 0.5% fetal bovine serum) supplemented with flow cytometry antibodies diluted at 1:50 and incubated for 30 min at 4 °C in the dark. Following staining, cells were washed twice with 1× PBS and analyzed using a BD FACSCanto II flow cytometer (BD Biosciences). Data were analyzed with FlowJo software (BD Biosciences). The following antibodies were used: FITC anti-human CD3 (Biolegend 300406), PE anti-human CD4 (Biolegend 317410), PerCP anti-human CD8 (Biolegend 344708), APC anti-human CD326 (EpCAM) (Biolegend 369810), and PE/Cyanine7 anti-human TCRα/β (Biolegend 306720).

Peripheral blood samples were collected from patients at designated time points, and red blood cells were lysed using RBC Lysis Buffer (Solarbio R1010). After lysis, samples were washed, stained, and analyzed following the same procedure as described for TIL. The following antibodies were used at a 1:50 dilution: FITC anti-human CD3 (Biolegend 300406), PE anti-human CD4 (Biolegend 317410), PerCP anti-human CD8 (Biolegend 344708), APC anti-human CD25 (Biolegend 356110), APC anti-human CD69 (Biolegend 310910), APC anti-human CD279 (PD-1) (Biolegend 329908), PE anti-human CD39 (Biolegend 328208), PE anti-human CD103 (Biolegend 350206), APC anti-human CD197 (CCR7) (Biolegend 353214), and PE anti-human CD352 (NTB-A) (Biolegend 332304).

Laboratory testing for tumor biomarkers

Peripheral blood samples were collected from patients and serums were collected Serum levels of cancer antigen 125 (CA125) were determined using a specific CA125 antigen detection kit (Product No. 386357, Beckman, United States). Briefly, add 100ul of the serum sample into the designated wells of the kit’s microplate. Then add the CA125 antibody (1:100) and ensure proper mixing. Incubate the plate at 2–8 °C for 30 min. After incubation, wash the plate thoroughly to remove unbound substances. Add the substrate solution and wait for the color to develop and then measure the absorbance using a microplate reader (Molecular Devices, LLC) to determine the serum CA125 levels. Lactate dehydrogenase (LDH) activity was measured using the LDH Activity Assay Kit (E-BC-K046-M, Elabscience) according to the manufacturer’s protocol. First, prepare the working solution as per the manufacturer’s instructions. Add 50ul of the serum sample into a cuvette followed by the working solution. Mix well and incubate the mixture at the room temperature for 45 min. After the reaction time, measure the change in absorbance at the 435 nm wavelength using a spectrophotometer (DS-39D, Shanghai).

HLA typing

Peripheral blood was collected at enrollment for high-resolution human leukocyte antigen (HLA) typing. HLA loci were typed using polymerase chain reaction (PCR) sequence-based typing with a DNA amplification step (CapitalBio, China). Briefly, genomic DNA was extracted from peripheral blood samples using the Magic Beads DNA Extraction Kit (TANBead, China) according to the manufacturer’s instructions. Exons 2 and 3 of HLA class I genes (HLA-A, HLA-B, and HLA-C) and exon 2 of HLA class II α genes were amplified and sequenced for allele determination. β genes (HLA-DQ and DR) were amplified and purified, and PCR products were sequenced on an ABI 3730XL DNA Analyzer (Applied Biosystems, USA). Sequence chromatograms were analyzed using ATF1.5 software (Conexio Genomics, Australia).

Tumor cell lines culture

The A375 cell line, bearing the HLA-A*02:01 haplotype, and the HMY-1 cell line with the HLA-A*24:02 haplotype were obtained from the American Type Culture Collection (ATCC). The selected tumor cell lines were placed in specialized cell culture flasks. RPMI 1640 medium, sourced from HyClone (Cat No.: 30809), was used as the base. To ensure optimal growth, it was supplemented with 10% fetal bovine serum from HyClone (Cat No.: SH30406.06), which is rich in nutrients and growth-promoting factors. Then, 1% penicillin-streptomycin from Gibco (Cat No.:15140012) was added to safeguard against microbial contamination. The flasks were then incubated at 37 °C in a 5% CO₂ atmosphere for cell growth. Cells were passage and harvested as needed for in vitro experiments.

Organoids culture protocol

Organoid culture was performed using the bioGenous™ Kit according to the manufacturer’s instructions. Briefly, approximately 0.5 cm³ surgical tumor specimens were washed twice with PBS and once with RPMI 1640 medium on ice. During washing, necrotic tissue, adipose tissue, and muscle were removed, and tumors were minced into fragments of 1–3 mm³. Tissue fragments were digested in Digestion Buffer (bioGenous, K601003) for 30–90 min at 37 °C. Digestion was terminated by addition of 4% FBS, and the suspension was filtered through a 100 μm strainer. Cells from digested tissues were washed twice with RPMI 1640 medium, centrifuged at 300 × g for 3 min, and resuspended in Matrigel (Corning, 356231). The cell-Matrigel mixture was seeded into 24-well plates at 20 µL per well and allowed to solidify at 37 °C for 15 min. Thereafter, 500 µL of complete medium (bioGenous, B213152) was added to each well, and organoids were maintained at 37 °C in a 5% CO₂ atmosphere. Organoids were passaged every 2–4 days by digestion with collagenase IV (200 U/mL) at 37 °C for 20 min, followed by splitting at a 1:2 ratio into new wells.

In vitro cytotoxicity assay

A flow cytometry-based assay was used to evaluate the in vitro cytotoxicity of melanoma-derived TIL against HLA-matched melanoma cell lines. Target cells (HMY-1) were labeled with 500 nM carboxyfluorescein succinimidyl ester (CFSE) and seeded into 96-well plates (50,000 tumor cells per well) with varying numbers of effector cells in AIM-V medium for 16 h (E: T ratios of 40:1 or 20:1). Immediately before analysis, cells were harvested and stained with 1 µg/mL 7-aminoactinomycin D (7-AAD; BioLegend, 420403) for 20 min at 4 °C in the dark, followed by flow cytometric analysis. The percentage of 7-AAD⁺ cells within the CFSE⁺ population was calculated and reported as specific cytolytic lysis (%) after subtraction of the non-effector control value.

For in vitro cytotoxicity assays against autologous tumor organoids, organoids were pretreated with 200 ng/mL IFN-γ for 24 h, washed twice with PBS, labeled with 500 nM CFSE, and co-cultured with TIL in AIM-V medium in 96-well plates for 48 h (E: T ratio of 5:1). Cells were then harvested and analyzed by flow cytometry.

Intracellular cytokine staining

Melanoma-derived TIL were co-cultured at 37 °C with HLA-matched HMY-1 or HLA-mismatched A375 cell lines at an E: T ratio of 5:1. GolgiPlug (BD, 555029; 1:1000 dilution) was added during the incubation. After 16 h, cells were stained with FITC-conjugated anti-human CD3 antibody (BioLegend, 300406; 1:50 dilution) to identify TIL, followed by fixation and permeabilization using the Cytofix/Cytoperm Kit (BD, 555028). For detection of IFN-γ and TNF-α, permeabilized cells were incubated with APC-conjugated anti-human IFN-γ (BioLegend, 502512; 1:50 dilution) and PE-conjugated anti-human TNF-α (BioLegend, 502908; 1:50 dilution) for 30 min at 4 °C in the dark. After two washes with BD Wash Buffer, cells were analyzed using a BD FACSCanto™ II flow cytometer (BD Biosciences).

Bulk RNA sequencing

Bulk RNA sequencing was performed on the infused TIL from all patients. mRNA was purified from total RNA of TIL using poly-T oligonucleotides and fragmented into 300–350 bp fragments. The first-strand cDNA was synthesized via reverse transcription of fragmented RNA using dNTPs (dATP, dTTP, dCTP, and dGTP), followed by second-strand cDNA synthesis. Remaining overhangs in the double-stranded cDNA were converted into blunt ends using exonuclease and polymerase activities. After adenylation of the 3′ ends of the DNA fragments, sequencing adaptors were ligated to the cDNA, and the library fragments were purified. The template was enriched by PCR amplification, and the PCR product was purified to generate the final library. Following library construction, library concentration was quantified using a Qubit^® fluorometer. The precise concentration of the cDNA library was further validated by qPCR. Library size distribution was assessed by agarose gel electrophoresis. After library preparation and pooling of samples, sequencing was carried out on an Illumina platform. Typically, lncRNA-seq employed PE150 (paired-end 150 bp) sequencing.

Raw data in FASTQ format were initially processed using in-house Perl scripts. During this step, clean data (clean reads) were generated by removing: (1) reads containing adapter sequences; (2) reads with more than three N bases; (3) reads with more than 20% of nucleotides having a Qphred score ≤ 5. Concurrently, Q20, Q30, and GC content of the clean data were calculated. Clean reads were then mapped to the SILVA database to remove rRNA sequences. All downstream analyses were conducted using rRNA-depleted clean data. Paired-end clean reads were aligned to the reference genome using Hisat2 [19]. FeatureCounts [20–21] was used to quantify reads mapped to each gene. Differential expression analysis was performed using EdgeR. P-values were adjusted using the Benjamini and Hochberg method to control the false discovery rate. Genes with |log2(fold change)| >1 and q-value < 0.05 were considered differentially expressed.

Bulk TCR sequencing

Bulk TCR sequencing was performed on peripheral blood mononuclear cells (PBMCs) collected from all patients on Day 15 and Week 6 post-TIL infusion, as well as on the infused TIL from Patients 3–12. RNA samples were subjected to deep high-throughput sequencing of the T-cell receptor beta (TRB) chain using a TCR profiling system. Briefly, a 5′ RACE-based unbiased amplification protocol was employed. This method incorporates unique molecular barcodes (UMBs) during cDNA synthesis to minimize amplification bias and correct for PCR and sequencing errors. Sequencing was conducted on an Illumina NovaSeq^® system in PE150 mode. A common adaptor containing a UMB was ligated to the 5′ end of cDNA during first-strand synthesis, and a reverse primer targeting the constant (C) regions of TRB was used to enable less biased PCR amplification. UMBs attached to raw sequence reads were utilized for error correction and removal of PCR duplicates. V, D, J, and C gene segments were mapped using NCBI databases, followed by CDR3 region extraction and clonotype assembly for all clones. The resulting nucleotide and amino acid sequences of the TRB CDR3 regions were determined, and sequences containing out-of-frame rearrangements or stop codons were excluded from the final TRB repertoire. Each TRB clonotype was quantified by summing the number of TRB clones sharing identical CDR3 nucleotide sequences.

Single-cell RNA/TCR sequencing

Single-cell RNA-seq and TCR-seq were performed on the infused TIL and PBMCs from Patients 1 and 11, collected at Week 24 post-TIL infusion. Single-cell TCR-seq was also conducted on the infused TIL from Patient 2. Single-cell RNA-seq and TCR-seq libraries were generated using the Chromium Single Cell 5’ Library and Gel Bead Kit (10x Genomics), following the manufacturer’s protocol. Briefly, cells were washed three times with PBS containing 0.04% BSA and resuspended to a concentration of 700–1,200 cells/µL (viability ≥ 85%), as assessed by Countess™ II. Cells were encapsulated in droplets with a targeted cell recovery. After reverse transcription, emulsions were broken, and barcoded cDNA was purified using Dynabeads, followed by PCR amplification. The amplified cDNA was used for both 5’ gene expression library construction and TCR enrichment. For gene expression library preparation, 50 ng of amplified cDNA was fragmented, end-repaired, and subjected to double size selection using SPRIselect beads, then sequenced on a NovaSeq platform (Illumina) to generate 150 bp paired-end reads. For TCR/BCR library construction, TCR transcripts were amplified via 10 cycles of PCR using primers provided in the kit. VDJ-enriched libraries were size-selected with SPRI beads and sequenced on an Illumina NovaSeq 6000 instrument using 150 bp paired-end reads. Data were analyzed using the Cell Ranger pipeline (10x Genomics) with default parameters and visualized using Loupe V(D)J Browser software.

Anti-tumor activity of TIL in patient-derived xenograft (PDX) mice

PDX models are generated by transplanting human tumor tissue into immunodeficient mice which allows them to retain many of the genetic and histological features of the original human tumors [22]. Thus, some aspects of TIL-tumor interactions, including the recognition of tumor-specific antigens by TIL can be maintained. To some extent, PDX models can preserve the tumor microenvironment. Tumor associated stromal cells, extracellular matrix components, and cytokines present in the original human tumor can be carried over into the PDX model [23]. Therefore, this provides a more realistic setting for studying TIL-tumor interactions compared to in vitro models. Immunodeficient NTG (NOD-scid Il2rg⁻/⁻) mice (6–8 weeks old) were obtained from SPF Biotech, Ltd. (Beijing, China). Two human cervical tumor samples were engrafted into NTG mice, and two PDX models—designated Model 001 and Model 006—were successfully established. A total of 2 × 10⁷ unedited control or edited TIL in saline were administered intravenously to NTG PDX mice bearing tumors of 50–100 mm³, while saline-treated mice served as controls. Recombinant human IL-2 (10,000 U) was administered daily to each mouse for the first 7 days post-infusion to support the survival and function of the infused TIL. Body weight and tumor volume were monitored twice weekly. Approximately 0.3 mL of blood was collected from each mouse at Week 2 and Week 6 post-TIL infusion. PBMCs were isolated from these samples, stained with anti-human CD45 antibody, and analyzed by flow cytometry to detect circulating TIL.

Biodistribution and pharmacokinetics of TIL in PDX mice

Immunodeficient NTG (NOD-scid Il2rg⁻/⁻) mice (6–8 weeks old) were obtained from SPF Biotech, Ltd. (Beijing, China). A human cervical tumor was engrafted into NTG mice to establish a patient-derived xenograft (PDX) model. DiR-labeled TIL (2 × 10⁷) in saline were administered intravenously to NTG PDX mice, with daily administration of 10,000 U human IL-2 for the first 7 days post-infusion; saline-treated mice served as controls. At 1 h, 2 h, 4 h, 6 h, 1d, 2d, 3d, 5d, 7d, 10d, and 14d post-infusion, mice were anesthetized and fluorescence signals were captured using the IVIS Lumina Series III imaging system (PerkinElmer). At 1d, 3d, 7d, 14d, and 28d post-infusion, 10 animals (5 male, 5 female) were euthanized at each time point. Organs—including xenograft tumor, heart, lungs, liver, kidneys, spleen, brain, stomach, intestine, ovaries, and testis—were collected for quantitative PCR (qPCR) analysis to assess TIL biodistribution. Approximately 0.3 mL of blood was collected at 1 d, 3 d, 7 d, 14 d, and 28 d post-infusion. PBMCs were isolated, stained with anti-human CD45 antibody, and analyzed by flow cytometry to evaluate TIL distribution and pharmacokinetics in peripheral blood.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 9.0 (GraphPad Software, La Jolla, California, USA). Survival curves were generated using the Kaplan-Meier method, and overall survival (OS) was measured from the date of enrollment to December 31, 2024, or the date of death, whichever occurred first. The two-tailed Student’s t test, Mann-Whitney U test, one-way ANOVA, or two-way analysis of variance with Dunnett’s post hoc test for multiple comparisons was used to assess statistical significance between groups. Correction for multiple testing was applied consistently in the clinical and omics analyses. For the transcriptomic work, we employed the Benjamini-Hochberg (BH) procedure to control the false discovery rate (FDR). This widely-used method is effective in adjusting p-values for multiple hypothesis testing in high-dimensional data. By setting an acceptable level of FDR ( 0.05), we were able to identify significant genes while minimizing the number of false positives. For repeated measurements, we used a mixed-effects model. This model takes into account both the fixed effects and the random effects. By including random effects, we can account for the correlation between repeated measurements within the same subject. This approach provides a more accurate estimate of the experiment effects and reduces the bias that could occur if we treated repeated measurements as independent observations. Furthermore, the read depth of the transcriptome data is 20 M with a data volume of 6G. We utilized the fastp (version 0.23.2) software for data quality control, with the specific details as follows: (1) remove the linker sequences from the reads and filter out reads with insert fragments shorter than 15 bp; (2) Filter low-quality reads with the proportion of bases with quality values below Q20 is greater than 0.20; (3) Filter and sequence reads containing N bases that the number of N bases in the reads exceeds 5. A p value of ≤ 0.05 was considered statistically significant.

Results

Clinical trial demonstrates safety and efficacy of TIL monotherapy in diverse solid tumors

We conducted a phase I, investigator-initiated trial to evaluate the safety, feasibility, and preliminary efficacy of autologous TIL monotherapy in patients with advanced solid tumors, with an emphasis on clinical activity in Asian patients (Figs. 1 and 2 and Supplemental Fig. 1). Patients underwent surgical tumor resection for TIL harvest, followed by ex vivo expansion using a pre-REP and REP protocol under good manufacturing practice (GMP) conditions. Prior to TIL infusion (range: 0.3–13 × 10¹⁰ cells), all patients received a lymphodepleting conditioning regimen consisting of cyclophosphamide (30 mg/kg) for 2 days and fludarabine (25 mg/m²) for 5 days. This was followed by a single intravenous infusion of expanded TIL and high-dose IL-2 administered for 6–12 days to support in vivo T-cell persistence and expansion (Fig. 1A). Twelve patients were enrolled and treated between August 2022 and December 2024. Tumor types included melanoma (n = 6), cervical cancer (n = 4), non-small cell lung cancer (n = 1), and head and neck cancer (n = 1) (Fig. 1B). All patients had relapsed or refractory disease with no remaining standard treatment options. Baseline demographic and clinical characteristics are summarized in Table 1.

Fig. 1.

Clinical evaluation of tumor-infiltrating lymphocyte (TIL) therapy in patients with advanced malignancies demonstrates the therapeutic potential of adoptive cell immunotherapy. (A) Flow diagram of the clinical trial design and patient disposition. (B) Swimmer plot depicting clinical outcomes of 12 enrolled patients following TIL infusion, including 6 with melanoma, 4 with cervical cancer, 1 with non-small cell lung cancer, and 1 with head and neck squamous cell carcinoma. (C) Tumor size dynamics from baseline to post-treatment (n = 12), categorized according to RECIST 1.1 criteria: complete response/partial response (CR/PR, green), stable disease (SD, gray), and progressive disease (PD, red). (D) Target lesion measurements before and after TIL therapy, visualized using interval plots with response categorization: CR/PR (green lines), SD (black lines), and PD (red lines). (E) Objective response analysis revealed an overall response rate (ORR) of 33.3% (4/12) and a disease control rate (DCR) of 75% (9/12). CR: complete response; PR: partial response; SD: stable disease; PD: progressive disease. ICF, Informed Consent Form

Fig. 2.

Representative imaging findings and adverse event profile from the clinical trial. (A) Serial CT scans demonstrating complete response (CR) in patient 8 at baseline, 6 weeks, and 12 weeks post-TIL infusion. (B) Combined photographic documentation and CT imaging showing partial response (PR) in patient 11 at baseline, 15 days, 6 weeks, and 12 weeks following TIL administration. (C) Treatment-related adverse events profile of the top 15 most frequently reported events during the clinical investigation

Table 1.

Patient clinical and demographic characteristics at baseline

Characteristic	Mean ± SD (n = 12)	Range
Age(year)	50.33 ± 13.70	24–71
Gender(male/female)	4/8	-
Weight (Kg)	62.0 ± 10.84	44–80
Tumor size (cm)	8.66 ± 4.67	4.67–20.46
Disease stage (III/ IV)	6/6	-
PS score (≤ 2 / >2)	12/0	-
Cancer type
Melanoma (n)	6	-
Cervical cancer(n)	4
Lung cancer(n)	1	-
Head and neck(n)	1	-
Brain metastasis (yes/no)	2/10
Treatment history
Chemotherapy (yes/no)	11/1
Dabrafenib+Trametinib	3/9
PD-1 inhibitor (yes/no)	11/1	-
Radiotherapy (yes/no)	4/8
Number of target lesion	2.67 ± 1.61	1–5

PS Score, Performance Status Score

Tumor responses were assessed using both RECIST 1.1 and iRECIST criteria. Among the 12 treated patients, 1 achieved a complete response (CR), 3 achieved partial responses (PR), 5 had stable disease (SD), and 3 experienced progressive disease (PD), yielding an ORR of 33.3% and a DCR of 75% (Fig. 1B–E). Reductions in tumor size were observed as early as 6 weeks post-infusion, with durable tumor shrinkage maintained in CR and PR cases up to 12 weeks (Figs. 1C and 2A–B; Table 2, and Supplemental Fig. 1A–B). Representative radiographic responses include Patient 8 (cervical cancer, CR), who achieved complete resolution of measurable lesions by 12 weeks post-TIL therapy, and Patient 11 (melanoma, PR), who demonstrated consistent reduction in tumor burden starting on day 15 (Fig. 2A–B). These findings support the potential of TIL therapy to induce meaningful tumor regression in heavily pretreated patients with advanced disease.

Table 2.

Overall survival and progression-free survival of studied patients

Patient ID	Cancer type	Response	Response group	OS (month)	PFS (month)
Pt. 1	melanoma	SD	NR	28.0	7.0
Pt. 2	melanoma	PD	NR	16.0	1.5
Pt. 3	cervical	PR	R	10.0	5.0
Pt. 4	lung	SD	NR	6.0	3.0
Pt. 5	cervical	SD	NR	10.5	3.0
Pt. 6	cervical	PD	NR	4.0	1.5
Pt. 7	head and neck	SD	NR	3.0	1.5
Pt. 8	cervical	CR	R	12.0	6.0
Pt. 9	melanoma	PR	R	5.5	4.5
Pt. 10	melanoma	PD	NR	9.5	1.5
Pt. 11	melanoma	PR	R	7	4.5
Pt. 12	melanoma	SD	NR	8	3.0

R, responder; NR, non-responder; OS, overall survival; PFS, progression-free survival

The safety profile of TIL monotherapy was consistent with previously reported effects associated with lymphodepletion and IL-2 administration. The most common adverse events included fever, anemia, nausea, hypotension, hyponatremia, and lymphopenia (Fig. 2C). All adverse events were manageable and reversible with supportive care, resolving within 1 week after completion of IL-2 administration. No unexpected grade 3–4 toxicities were observed, and there were no treatment-related deaths. Notably, no gene-edited TIL used in downstream studies were administered in this trial phase, ensuring that safety data reflect non-edited TIL therapy. These clinical data demonstrate that autologous TIL monotherapy is feasible, safe, and capable of inducing durable objective responses in a subset of patients with advanced solid tumors. The heterogeneous response patterns observed across patients underscore the need to identify predictive biomarkers of response and resistance, providing rationale for further mechanistic and translational investigations.

Durable clinical benefit and biomarker dynamics

To characterize the durability of clinical benefit following TIL monotherapy, we performed longitudinal survival analyses and monitored dynamic changes in tumor-associated biomarkers and immune parameters in peripheral blood. Kaplan-Meier analysis revealed a median PFS of 3 months among the 12 enrolled patients (Fig. 3A). Notably, a subset of patients achieved sustained disease control beyond 6 months, particularly those attaining CR or PR. OS analysis showed a median OS of 8.75 months (Fig. 3B; Table 2), with several patients remaining alive and free of disease progression at the data cutoff date (December 31, 2024). These survival outcomes are encouraging given the advanced disease stage and limited prior therapeutic options in this patient population.

Fig. 3.

Clinical efficacy outcomes and biomarker dynamics from the trial. (A) Progression-free survival (PFS) curve demonstrating a median survival duration of 3 months in the patient cohort. (B) Overall survival (OS) curve showing a median survival of 8 months, calculated from the enrollment date to December 31, 2024, or the date of death. (C) Temporal profile of lactate dehydrogenase (LDH) levels in melanoma patients, showing a progressive decline in responders. (D) Dynamic changes in the tumor biomarker cancer antigen 125 (CA125) among cervical cancer patients. Longitudinal analysis of CD3 + T cell proportions in peripheral blood mononuclear cells (PBMCs) across the study period. Survival analyses were performed using the Kaplan-Meier method and overall survival (OS) was measured from the date of enrollment to December 31, 2024, or the date of death, whichever occurred first

We next evaluated peripheral blood biomarkers longitudinally to identify potential correlates of clinical response. In patients with melanoma, serum LDH levels—a marker associated with tumor burden—were measured at multiple time points before and after TIL infusion. Among responders (CR/PR), LDH levels progressively declined over time, whereas non-responders exhibited persistently elevated or fluctuating LDH levels (Fig. 3C). These data suggest that serial monitoring of LDH may serve as a serum biomarker of tumor burden reduction in melanoma patients undergoing TIL therapy. Similarly, in cervical cancer patients, serial measurements of serum cancer antigen 125 (CA125) were obtained. Patients achieving clinical response demonstrated marked reductions in CA125 levels following TIL infusion, which correlated with radiographic evidence of tumor regression (Fig. 3D). These findings further support the utility of tumor biomarkers in non-invasively indirectly assessing therapeutic efficacy of TIL treatment.

Concurrently, we assessed the systemic immune response by evaluating the proportion of CD3⁺ T cells among total PBMCs at multiple time points post-treatment. Following TIL infusion, a rapid expansion of circulating CD3⁺ T cells was observed in the majority of patients (Fig. 3E, Supplemental Fig. 2). Collectively, these results indicate that autologous TIL therapy not only induces objective tumor responses in solid tumors but is also associated with systemic immune activation and favorable biomarker dynamics that correlate with clinical benefit. These longitudinal biomarker assessments provide valuable tools for monitoring treatment response and identifying early predictors of therapeutic success.

T cell phenotyping reveals CD8 + T dominance post TIL treatment, and cytokine profiles correlate with clinical response

To explore the immunophenotypic features associated with clinical response to TIL therapy, we performed comprehensive flow cytometry analyses on infused TIL products and longitudinally collected peripheral blood samples from treated patients. The objective was to evaluate the composition, activation status, and differentiation profiles of T cell subsets—particularly CD4⁺ and CD8⁺ T cells—and to determine their correlation with clinical outcomes. Expanded TIL infusion products from all 12 patients were analyzed by multicolor flow cytometry for surface expression of canonical T cell markers, including CD3, CD4, CD8, TCRα/β, and EpCAM (Table 3). Over 95% of infused TIL were CD3⁺ T cells, confirming successful expansion of tumor-reactive T cells. Notably, EpCAM—a marker indicative of epithelial contamination—was undetectable, supporting the purity of the TIL products. Among CD3⁺ T cells, the majority were CD8⁺ cytotoxic T lymphocytes (CTLs), with a smaller proportion comprising CD4⁺ helper T cells (Table 3). This CD8-dominant phenotype was more prominent in TIL products administered to responders (CR/PR) compared to non-responders (SD/PD), although the difference did not reach statistical significance (Supplemental Fig. 3A–C). Nevertheless, the trend toward a lower CD4⁺/CD8⁺ ratio in responders suggests a potential contribution of cytotoxic effector cells to anti-tumor activity.

Table 3.

Cell compositions of infused TIL in 12 studied patients

Pt. ID	Efficacy	Infused amount (1010)	CD3 + T%	CD3 + CD8+T%	CD3 + CD4+T%	CD3+ TCRα/β + T%	EpCAM+%
Pt.1	SD	4.2	96.4	64	23.1	89.2	0.01
Pt.2	PD	1.8	96.8	80.7	10.3	95.9	0
Pt.3	PR	0.3	97	56.6	29.5	88.9	0.06
Pt.4	PD	13	99.7	84	5.4	94.9	0.07
Pt.5	SD	0.51	99.6	62.8	11.7	90.2	0.01
Pt.6	PD	0.6	97.7	61.6	31.5	93.7	0.1
Pt.7	SD	11	99.7	86.4	10	94.4	0.04
Pt.8	CR	2.7	97.5	62.7	14.7	86.5	0.06
Pt.9	PR	3.5	99.3	90.3	2.5	96.2	0.11
Pt.10	PD	0.89	99.7	52.2	19.5	89.3	0.1
Pt.11	PR	8.5	99.9	74.4	24.1	84.6	0.15
Pt.12	SD	7.4	99.4	24.6	70.3	82.1	0

CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease

In addition to characterizing the infused TIL products, we assessed longitudinal T cell dynamics in peripheral blood. Flow cytometric analyses at days 7, 0, 1, 4, 8, 15, 21, 28, 42, 84, and 168 post-infusion revealed sustained elevation of circulating CD3⁺ T cells in most patients, with CD8⁺ T cells representing a substantial proportion of this expanded population (Fig. 3E and Supplemental Fig. 4). Notably, CD8⁺ T cells exhibited an increasing trend over time following TIL infusion, whereas CD4⁺ T cells showed a decreasing trend (Supplemental Fig. 4A). The CD4⁺/CD8⁺ ratio, along with frequencies of CD103⁺ cells, regulatory T cells (Tregs), and CCR7⁺ T cells, also declined post-treatment (Supplemental Fig. 4A). No significant differences were observed between responders and non-responders in CD8⁺ T cells, CD4⁺ T cells, CD4⁺/CD8⁺ ratio, CD103⁺ cells, Tregs, CD8⁺CD39⁺ T cells, or CCR7⁺ T cells during follow-up (Supplemental Fig. 4B). Despite inter-individual variability, responders tended to maintain higher levels of PD-1⁺ T cells at 28 and 42 days post-treatment and higher frequencies of terminally exhausted T cells (NTB-A⁺CD69⁺) at 8 days post-infusion compared to non-responders (Supplemental Fig. 4B), suggesting that PD-1 expression and terminal exhaustion may not impair or reflect treatment efficacy. Larger cohort studies are required to validate these observations, and underlying molecular mechanisms warrant further investigation. Overall, these findings indicate that the composition and functional potential of infused TIL—particularly enrichment of CD8⁺ T cells—may contribute to clinical response.

To further characterize the systemic immune landscape following TIL infusion, we conducted longitudinal profiling of serum cytokines in all 12 patients using multiplex ELISA. Cytokines were selected to capture both effector (pro-inflammatory) and regulatory (immunosuppressive) immune responses, thereby providing insight into the functional state of the immune system after TIL therapy. The cytokines analyzed included IFN-γ, TNF-α, IL-2, IL-6, IL-7, IL-13, IL-21, and TGF-β1. Among the pro-inflammatory cytokines, IFN-γ and TNF-α—both key mediators of T-cell cytotoxic function—showed marked increases in responders (patients with CR or PR) compared to non-responders (SD or PD) (Supplemental Fig. 5A). These elevations were typically observed by day 7 post-infusion and were sustained through week 6, suggesting that infused TIL were functionally active in vivo and capable of producing Th1-type effector cytokines upon tumor recognition. Serum IL-2 levels remained relatively stable throughout treatment in most patients; however, IL-2 levels transiently increased in responders, consistent with high-dose exogenous IL-2 administration post-infusion. Moreover, the magnitude and persistence of IL-2 elevation were significantly greater in responders at 8 and 15 days post-TIL infusion compared with non-responders (Supplemental Fig. 5B,P = 0.044 and P = 0.048), potentially reflecting enhanced TIL proliferation and survival in these individuals.

Notably, IL-7, a cytokine involved in T-cell homeostasis and memory formation, was modestly elevated in responders at 8 and 15 days post-TIL infusion compared with non-responders (Supplemental Fig. 5B, P = 0.014 and P = 0.015), further supporting the hypothesis that a favorable cytokine environment may contribute to long-term TIL persistence and anti-tumor efficacy. There were no significant differences in the expression levels of IL-21, IL-6, IL-10, or IL-13 between responders and non-responders during treatment (Supplemental Fig. 5B). Collectively, these findings indicate that specific cytokine signatures correlate with clinical outcomes following TIL therapy. A Th1-skewed response characterized by high IFN-γ and TNF-α, low IL-10 and TGF-β1, and moderate IL-7 and IL-21 may serve as an immunological fingerprint of response. Future studies will focus on validating these cytokine profiles as early predictive biomarkers of efficacy in larger cohorts.

TCR repertoire remodeling dynamics post-TIL therapy

To evaluate the clonal dynamics and persistence of tumor-reactive T cells following TIL therapy, we performed both bulk RNA and single-cell T-cell receptor (TCR) sequencing on infused TIL products and PBMCs collected at multiple time points post-infusion (Figs. 4 and 5). This analysis aimed to characterize clonal expansion, TCR diversity, and the long-term fate of infused T-cell populations in vivo. Bulk TCR-β sequencing was conducted for all 12 patients. TCR repertoire data were generated from the infused TIL and from PBMCs at day 15 and week 6 post-infusion, except for patient 1, who had an additional time point at week 12, and patient 2, who lacked a TIL sample for sequencing (Figs. 4 and 5 and Supplemental Figs. 6–8). We first quantified the number of shared TCR clonotypes between infused TIL and post-treatment PBMCs. The majority of patients exhibited persistence of infused TCR clones in circulation, with detectable overlap between the infused product and blood-derived T cells at both time points (Fig. 4A).

Fig. 4.

T cell receptor (TCR) sequencing analysis revealed that the number of shared TCR clones was associated with progression-free survival. (A) Number of TCR clones at different time points (pre-infusion, 15 days, 6 weeks, 12 weeks) in 12 enrolled patients. (B) Responders (R) exhibited a decreased number of shared TCR clones compared with non-responders (NR). (C) No difference in TCR clone numbers was observed between patients with long versus short overall survival (OS). (D) Patients with longer progression-free survival (PFS) had a significantly lower number of TCR clones than those with shorter PFS (P = 0.029). Statistical significance between groups was analyzed using two-tailed Student’s t-test or Mann-Whitney U test. A p value ≤ 0.05 was considered statistically significant

Fig. 5.

T cell receptor (TCR) sequencing analysis suggests that the clonality index correlates with overall survival. (A) Proportions of the top 20 individual TCR sequences among 12 enrolled patients, indicating that most responders harbored dominant TCR clones. (B) Patients with longer OS exhibited a significantly lower TCR clonality index compared with those with shorter OS (P = 0.0243). (C) Single-cell TCR sequencing for patient 1 revealed detectable TCR clones at 6 weeks post-TIL infusion. Statistical significance between groups was analyzed using two-tailed Student’s t-test. A p value ≤ 0.05 was considered statistically significant

Interestingly, patients who achieved clinical responses (CR/PR) exhibited fewer shared TCR clones compared to non-responders (SD/PD) (Fig. 4B, P = 0.091). This counterintuitive observation reflects the preferential expansion of a limited number of dominant clones in responders, rather than the broad persistence of the entire infused T-cell repertoire. Moreover, responders demonstrated reduced clonal diversity, suggesting selective clonal focusing toward tumor-reactive specificities. In contrast, non-responders maintained more polyclonal profiles, potentially indicative of non-productive or exhausted T-cell repertoires. Stratification by clinical outcome revealed additional associations: patients with longer progression-free survival (PFS) had significantly fewer shared clones at post-treatment time points compared to those with shorter PFS (Fig. 4D, P = 0.029), whereas overall survival (OS) did not show a significant correlation with shared clone frequency (Fig. 4C–D). These findings suggest that durable TIL efficacy may be driven by the focused expansion of high-affinity, tumor-specific clones rather than the maintenance of a diverse repertoire.

Single-cell TCR sequencing (scTCR-seq) was performed in three patients (Patients 1, 2, and 11), using both infused TIL and PBMCs collected at Week 6 or Week 24. Clonotype tracking in Patient 1 revealed that dominant clones present in the infused TIL product persisted in peripheral blood at 6 weeks post-treatment (Fig. 5C), confirming long-term engraftment of functional T cells. Similarly, the top 100 TIL clones remained detectable in Week 6 PBMCs of responders, whereas their representation declined over time in non-responders (Fig. 5A). TCR clonality indices were calculated for each patient. Responders displayed higher TCR clonality than non-responders, with dominant clones comprising a greater proportion of the circulating repertoire. Patients with longer OS tended to have lower TCR clonality compared to those with shorter OS, a trend that reached statistical significance (P = 0.0243) (Fig. 5B). Collectively, these results indicate that effective TIL therapy is associated with the in vivo persistence and expansion of dominant tumor-specific TCR clones. This clonal remodeling likely reflects both the quality of the infused T-cell product and the permissiveness of the host immune environment. Monitoring TCR dynamics may therefore serve as a valuable biomarker for assessing therapeutic response and elucidating mechanisms underlying TIL efficacy.

Transcriptomic profiling identifies Argininosuccinate synthase 1 (ASS1) and centrosomal protein 20 (CEP20) as negative regulators

To identify intrinsic molecular determinants of TIL associated with therapeutic efficacy, bulk RNA sequencing was performed on infused TIL products from all 12 patients (Fig. 6 and Supplemental Fig. 9). Differential gene expression analysis was conducted between TIL from responders (CR/PR; n = 4) and non-responders (SD/PD; n = 8) to identify potential regulators of TIL function. Unsupervised hierarchical clustering based on the most differentially expressed genes revealed a distinct transcriptomic signature separating the two groups (Fig. 6A). Among the genes significantly downregulated in the responded group, nearly 20 genes were selected based on top fold change, lowest p value, or known relevance to tumor immunology for evaluation through genetic editing. ASS1 and CEP20 were prioritized as candidate targets for further validation of their role in modulating TIL antitumor function based on the preliminary testing results of ASS1 KO and CPE20 KO TIL showing improved anti-tumor activity, high TIL proliferation rate, high gene editing efficiency and reserved TIL cell phenotype. ASS1 and CEP20 were of particular interest also due to their established roles in cellular metabolism and cell cycle regulation, respectively. It was reported that ASS1 production in macrophages allows T-cell suppression and is regulated by the transcription factor NF-kappaB [24]. CEP20 is related to centrosome function that centrosomes play a vital role in cell division, including the proper segregation of chromosomes during mitosis, and therefore CEP20 may contribute to the accurate and efficient cell division of TIL [25].Both genes were consistently overexpressed in TIL from non-responders, suggesting a potential inhibitory role in T-cell functionality or survival within the tumor microenvironment. However, the molecular mechanisms by which ASS1 and CEP20 regulates TIL function require further in-depth investigation.

Fig. 6.

Potential targets identified through bulk RNA sequencing analysis of enrolled patients. (A) Heatmap of the top 30 downregulated genes in infused TIL from responders compared to non-responders, as determined by bulk RNA sequencing. ASS1 and CEP20 genes are indicated by arrows. (B) CRISPR/Cas9-mediated knockout (KO) efficiencies of ASS1 and CEP20 in TIL were 61% and 65%, respectively, as measured by Sanger sequencing. (C) Flow cytometric analysis of memory (CD127+) and exhaustion (PD-1 + CD39+) phenotypes in ASS1- and CEP20-knockout TIL. (D) ASS1- or CEP20-knockout TIL exhibited slightly reduced proliferation capacity compared to non-edited (NE) control TIL. (E–F.) In vitro cytolysis assays demonstrated that ASS1-knockout TIL exhibited significantly enhanced anti-tumor activity compared to NE control TIL and CEP20-knockout TIL at effector-to-target (E: T) ratios of 40:1 and 20:1 (P = 0.0016 and P = 0.0003, respectively), whereas CEP20-knockout TIL showed comparable cytolytic activity to NE controls (P = 0.9485 and P = 0.7298). (G–H.) Intracellular cytokine staining revealed that the proportion of TIL co-producing IFN-γ and TNF-α following co-culture with the HLA-matched HYM-1 cell line (sharing HLA-A*24:02) was significantly higher in ASS1-knockout TIL than in NE controls (P = 0.0002), while CEP20-knockout TIL showed a modest increase (P = 0.0622). No reactivity was observed against the HLA-mismatched A375 cell line in any TIL group (P = 0.7610 and P = 0.7786). In vitro experiments were set up as triplicates and were done in two or three independent experiments. Statistical significance was assessed using two-tailed Student’s t-test, Mann-Whitney U test, two-way ANOVA with multiple comparisons adjustment (Dunnett’s test), or one-way ANOVA. A p value ≤ 0.05 was considered statistically significant. TILs, Tumor Infiltrating Lymphocytes

To evaluate the functional relevance of these targets, we utilized CRISPR/Cas9-mediated gene editing to knockout ASS1 or CEP20 in tumor-infiltrating lymphocytes (TIL) derived from fresh tumor specimens. Gene disruption efficiency was confirmed by Sanger sequencing and quantified at 61% for CEP20 and 65% for ASS1 (Fig. 6B). Following genetic modification, cells underwent rapid expansion using the same GMP-compliant protocol applied for clinical-grade TIL production. Phenotypic analysis revealed that ASS1- and CEP20-knockout TIL exhibited increased proportions of memory-associated CD127 + T cells and non-exhausted T cells (PD-1 − CD39−) compared to non-edited control TIL (P = 0.0013,P = 0.0099) (Fig. 6C and Supplementary Fig. 9). This shift in differentiation profile suggests that loss of ASS1 or CEP20 promotes a less differentiated, stem-like phenotype associated with enhanced persistence and anti-tumor functionality. Proliferation assays demonstrated that edited TIL maintained robust expansion capacity, with only a modest reduction in total cell yield relative to controls (Fig. 6D), indicating that gene editing did not compromise TIL viability or scalability.

Functionally, ASS1-knockout TIL exhibited significantly enhanced cytolytic activity in vitro. In co-culture assays with HLA-matched tumor targets, ASS1-deficient TIL mediated greater tumor lysis compared to both *CEP20*-knockout and non-edited TIL (P = 0.0003), whereas CEP20-knockout TIL showed no significant difference from controls (P = 0.3829) (Fig. 6E–F). Concurrently, intracellular cytokine staining revealed a higher frequency of double-positive IFN-γ⁺/TNF-α⁺ T cells in ASS1-knockout TIL upon stimulation with tumor antigens in co-culture with an HLA-A\24:02-matched HYM-1 cell line (P = 0.0002) (Fig. 6G–H). No cytokine production was detected in response to the HLA-mismatched A375 tumor cell line, confirming antigen specificity (Fig. 6G–H). These findings indicate that ASS1 acts as a metabolic checkpoint restraining the effector function of TIL, and its targeted disruption may augment their therapeutic potential. In contrast, CEP20, while linked to phenotypic modulation, exerted a more limited functional effect. Collectively, this transcriptome-guided, reverse-translational approach identifies ASS1 as a novel and actionable target for the genetic improvement of TIL.

Genetically enhanced TIL show improved anti-tumor effects in vivo

To validate the enhanced anti-tumor potential of ASS1- and CEP20-edited TIL in vivo, we conducted functional studies in an independent cervical cancer PDX mouse model (model 006), which was established by engrafting fresh human tumor tissue into NOD-scid IL2rg⁻/⁻(NSG) mice—immunodeficient hosts permissive for human T-cell engraftment (Supplementary Figs. 10–11). This model has been validated in large-scale experiments, with no evidence of autoimmune responses or graft-versus-host disease observed (Supplementary Fig. 11). In vivo distribution studies and in vitro killing assays further demonstrated potent anti-tumor activity of TIL in this PDX model (Supplementary Fig. 12). Using this system, we show that adoptively transferred TIL efficiently home to tumor tissues, persist in circulation, and exert strong cytotoxicity against autologous tumor targets, confirming the in vivo and in vitro therapeutic potential of TIL (Supplementary Figs. 10–12).

PDX-bearing mice (model 006) with tumors measuring 50–100 mm³ were randomized into four treatment groups: (1) vehicle control, (2) unedited TIL (2 × 10⁷ cells), (3) ASS1 knockout (KO) TIL (2 × 10⁷ cells), and (4) CEP20 KO TIL (2 × 10⁷ cells). All mice received daily intraperitoneal injections of human IL-2 (10,000 IU) for 7 days post-infusion to support T-cell survival and activity (Fig. 7A). Tumor growth was monitored twice weekly. Compared with vehicle-treated controls, both ASS1 KO and CEP20 KO TIL induced significant tumor suppression over time (P = 0.04 and P = 0.0013, respectively), with ASS1 KO TIL achieving the most durable and pronounced tumor regression (Fig. 7B). Unedited TIL also delayed tumor progression but were less effective than gene-edited TIL. These results confirm that deletion of ASS1 or CEP20 enhances the in vivo potency of TIL. Notably, no significant differences in body weight were observed among treatment groups throughout the study (Fig. 7C), indicating that the edited TIL were well tolerated and did not cause systemic toxicity.

Fig. 7.

In vivo evaluation of ASS1- and CEP20-knockout TIL in a patient-derived xenograft (PDX) mouse model confirms robust anti-tumor responses. (A) Experimental timeline. (B) Tumor measurements showed significantly lower tumor burden in mice treated with ASS1-knockout and CEP20-knockout TIL 50 days post-infusion (P = 0.04 and P = 0.0013, respectively; n = 6 animals per group). Due to unavailability of autologous TIL for PDX model 006, TIL from a donor sharing HLA-A*24:02 with the autologous TIL were used. (C) Body weight monitoring revealed no significant differences among treatment groups throughout the study. (D) Flow cytometry analysis of peripheral blood showed increased percentages of hCD45 + CD3+ cells among lymphocytes in mice treated with ASS1- or CEP20-knockout TIL at 10 and 20 days post-treatment compared to vehicle controls. (E) ELISA analysis of serum from peripheral blood demonstrated significantly higher levels of IFN-γ in mice receiving ASS1- or CEP20-knockout TIL at 20 and 30 days post-infusion compared to vehicle-treated mice. In vitro experiments were set up as triplicates and were done in two or three independent experiments. Statistical significance was evaluated using two-tailed Student’s t-test, Mann-Whitney U test, two-way ANOVA with multiple comparisons adjustment (Dunnett’s test), or one-way ANOVA. A p value ≤ 0.05 was considered statistically significant. * p < 0.05, ** p < 0.01

To evaluate the in vivo persistence of human T cells, peripheral blood was collected at days 10, 20, and 30 post-infusion and analyzed by flow cytometry. Mice receiving ASS1 KO or CEP20 KO TIL showed significantly higher proportions of hCD45⁺CD3⁺ cells among circulating lymphocytes at days 10 and 20 post-TIL infusion compared with vehicle-treated controls (Fig. 7D), suggesting enhanced engraftment or expansion of edited TIL in vivo. Serum IFN-γ levels were measured longitudinally by ELISA. Both ASS1 KO and CEP20 KO TIL-treated mice exhibited elevated IFN-γ concentrations at days 20 and 30 post-treatment, significantly exceeding those in control animals (Fig. 7E). These findings are consistent with an activated T-cell phenotype and corroborate the in vitro functional data previously described. Collectively, these results demonstrate that CRISPR/Cas9-mediated disruption of ASS1 or CEP20 enhances the antitumor efficacy, in vivo persistence, and cytokine production of TIL. This genetically engineered TIL approach provides proof-of-concept for rational improvement of TIL products to overcome intrinsic functional limitations and improve therapeutic outcomes in solid tumors.

Discussion

This phase I investigator-initiated clinical trial demonstrates the safety, feasibility, and preliminary efficacy of autologous TIL therapy in patients with advanced, treatment-refractory solid tumors. Beyond melanoma—the historical foundation of TIL-based immunotherapy [26]—our findings extend the clinical applicability of TIL therapy to cervical, lung, and head and neck cancers. An objective response rate of 33.3% and a disease control rate of 75% were achieved in a heterogeneous, heavily pretreated patient cohort, including one complete response and three partial responses. These results underscore the potential for broader application of TIL therapy in epithelial malignancies that have traditionally been resistant to immunotherapies. Preclinical studies using patient-derived xenograft models provided foundational evidence of TIL-mediated antitumor activity and established key parameters for dosing, trafficking, and persistence [27–29]. Autologous TIL outperformed HLA-matched allogeneic T cells, highlighting the importance of tumor-specific antigen recognition. Biodistribution analyses demonstrated rapid and preferential homing of infused TIL to tumor sites, while in vitro cytotoxicity assays confirmed functional reactivity against patient-derived organoids and HLA-matched tumor cell lines.

The clinical data were further supported by dynamic immunological analyses. Expansion and persistence of CD3⁺ T cells in the periphery, along with longitudinal reductions in LDH and CA125, served as pharmacodynamic indicators of clinical response. Flow cytometry revealed a CD8⁺ T cell–dominant phenotype in responders, suggesting that effector cytotoxic T cells may be the primary mediators of tumor regression. Concurrently, cytokine profiling demonstrated that a T-helper 1 (Th1)–skewed systemic response—characterized by elevated interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) and reduced interleukin-10 (IL-10)—correlated with durable clinical benefit. These immune correlates provide a mechanistic framework for understanding TIL activity in vivo and may serve as potential biomarkers for clinical response. A central innovation of this study lies in the integration of transcriptomic profiling [30] with functional gene editing to identify and overcome intrinsic limitations of TIL therapy. By comparing TIL products from responders and non-responders, we identified ASS1 and CEP20 as candidate negative regulators. CRISPR/Cas9-mediated knockout [31] of these genes induced a favorable shift toward memory-like phenotypes [32], enhanced cytokine production, and improved cytotoxicity. ASS1 [33–34], in particular, emerged as a metabolic checkpoint that restricts TIL function. Its deletion significantly enhanced antitumor responses both in vitro and in vivo, demonstrating the therapeutic potential of rationally engineered TIL products. These findings position gene-edited TIL as a next-generation cell therapy platform [35]capable of overcoming T cell–intrinsic barriers to efficacy. While recent clinical advances have explored TIL therapy in combination with immune checkpoint inhibitors or IL-2 variants, our study provides a blueprint for genetically enhancing TIL themselves to unlock their full therapeutic potential.

Moreover, there are several possible interpretations of fewer shared clones as favorable we observed in the study. Firstly, a lower number of shared clones may indicate a more tumor-specific immune response. When there are fewer shared clones, it implies that the TIL population may has a more targeted and personalized reaction against the tumor. Each tumor often has its unique set of neoantigens, and a focusing of the T cell response towards these specific antigens can lead to a more effective anti-tumor attack. Fewer shared tumor-specific clones are more likely to recognize and eliminate tumor cells precisely, thus potentially leading to better treatment outcomes [36–37]. Secondly, it might be related to sampling depth that if the sampling depth is sufficient, and yet we observe fewer shared clones, it could mean that the TIL are not just randomly distributed or a result of general immune activation, but rather a concentrated, specific response to the tumor. However, it is important to note that if the sampling depth is insufficient, we may misinterpret the situation as a small number of shared clones when in fact there is a larger, more homogenous population that we have failed to fully capture [36–37]. Thirdly, differences in TIL product composition might also play a role. A TIL product with fewer shared clones may have a higher proportion of effector T cells that are specific to the tumor. These effector T cells are often more potent in killing tumor cells [36–37]. However, whether this reflects true tumor-specific focusing, sampling depth, or differences in TIL product composition requires further in-depth investigation. Future studies could involve more comprehensive profiling of the TIL, increasing the sampling depth, and analyzing the functional characteristics of different T cell clones to better understand the underlying mechanisms.

Importantly, the present study was set as an Asian multi-tumor cohort which is a significant departure as the Asian population may has unique genetic, environmental, and lifestyle characteristics. Genetic polymorphisms in Asian individuals may lead to different immune responses and TIL profiles, and lifestyle factors like diet can also impact the tumor microenvironment and TIL infiltration in Asian patients more distinctively compared to other ethnic groups. This cohort allows for a more targeted understanding of TIL behavior in Asian patients across multiple tumor types, which was not comprehensively explored in prior studies. Additionally, the clonal remodeling patterns observed in this Asian TIL trial might be novel, and this multi-tumor cohort could have unique tumor microenvironments that drive different clonal evolution of TIL. The patterns we observed might be crucial for understanding how TIL adapt and function in the context of Asian tumors. The identification of ASS1 and CEP20 as candidate regulators is a new finding that in prior TIL studies, these genes were not recognized as having significant roles in TIL biology. CEP20 is related to centrosome function, and its role in TIL may suggest a connection between cell division and TIL activation or proliferation. However, the phenotype of CEP20 KO TIL we observed was weaker than ASS1 KO TIL, and additional mechanistic work are needed to further confirm its role.

It was worth to note that ASS1 is involved in the urea cycle, and its potential link to TIL function could imply a previously unknown metabolic-immune interaction. Urea-cycle metabolism can influence gene expression in T cells. Nuclear factor-kappa B (NF-κB), a key transcription factor involved in cell survival signaling, can be modulated by arginine availability [38]. Proper arginine levels and urea-cycle activity can ensure the normal activation and function of NF-κB, which in turn promotes the expression of anti-apoptotic genes and enhances T cell persistence [38–40]. Arginine is involved in the metabolic reprogramming required for memory T cell formation. During the differentiation of naive T cells into memory T cells, arginine metabolism shifts to support the long-term survival and self-renewal properties of memory T cells. Additionally, Arginine is a precursor for the synthesis of nitric oxide (NO) and ornithine, which are important for the production of effector molecules in CTLs [38–40]. Therefore, ASS1 may functionally regulates urea-cycle metabolism and arginine availability to play crucial roles in T cell persistence, memory formation, and cytotoxicity through multiple mechanisms. Future studies are required for in-depth exploration and confirmation in terms of the molecular mechanisms by which ASS1 and CEP20 regulates TIL function. The discovery of these genes as candidate regulators opens up new avenues for understanding the molecular mechanisms underlying TIL T-cell persistence, memory formation, cytotoxicity, proliferation and viability.

Limitation

The study cohort was small and heterogeneous, which may confound the identification of robust predictive biomarkers. The small size study may limits the inferential confidence for response biomarkers and survival correlations, and the results of the correlation analysis of the shared TCR clones, clonality, cytokines, and response were exploratory and need to be confirmed by large population studies in the future. Additionally, long-term follow-up is required to assess response durability and survival benefit. The functional role of CEP20 [41] remains incompletely characterized and warrants further investigation. Although our preclinical in vivo studies used cervical cancer–derived patient-derived xenograft (PDX) models, both male and female NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NTG) mice were employed to evaluate the generalizability of TIL efficacy across solid tumors. While this approach may raise concerns regarding anatomical relevance for cervical tumors, our intent was to model solid tumor immunotherapy broadly rather than limit interpretation to gynecological cancers. Furthermore, although certain immune checkpoint molecules (PD-1/PD-L1) or T cell activation markers (CD69, CD25) were commonly expressed across different malignancies. The immune-phenotypic findings we found were only tested in melanoma and cervical cancer as they have a higher prevalence in the present study cohort. Improved function of ASS1-KO and CEP20 KO TIL may be a shared immune-phenotypic feature across multiple tumor types. However, it requires in-depth analysis regarding whether the immune-phenotypic findings are shared across all other tumor types. To determine this context, a comprehensive statistical analysis of ASS1-KO and CEP20 KO TIL from all tumor types is needed.

With the rapid development of medical technology, the discovery of innovative targets, the emergence of small molecule inhibitors, and new detection techniques and methods have provided strong support for anti-tumor treatment discoveries [42–47].

The safety of genetically edited TIL must be evaluated in a dedicated clinical trial prior to therapeutic implementation. There are several barriers to concur for the clinical trials of the gene-edited TIL. First, manufacturing complexity is a major challenge. Producing gene-edited TIL consistently and at a large scale requires advanced technology and strict quality control. Second, editing efficiency need to maintain at a high level. Low efficiency may result in a large number of non-edited TIL in the final product, reducing the overall therapeutic effect. Third, there may strict regulatory requirements to ensure that gene-edited TIL are safe and effective before allowing them into the market. Meeting these regulatory controls can be time-consuming and costly. Four, off-target effects of gene editing may lead to unexpected mutations in the genome of TIL, which might trigger autoimmune responses. Therefore, the safety and translational feasibility of the edited TIL are required to be studied in future clinical trials. Although gene-edited TIL hold great potential, a comprehensive evaluation of safety and translational feasibility is necessary before they can be widely applied in clinical practice.

Conclusion

This translational study demonstrates that autologous TIL therapy is clinically feasible across solid tumors and that CRISPR-based functional enhancement of TIL can further augment their efficacy. By identifying and validating actionable targets through reverse translational research, we establish a foundation for a new generation of precision-engineered TIL therapies designed to overcome tumor-specific resistance mechanisms. These findings support the continued development of customized, gene-edited TIL products for solid tumor immunotherapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

Conceptualization: FL; Methodology: FL, JY, and YX; Validation: FL, JY, YX, XZ, XG, ML, WL, YW, WF, CM, CZ, and YM; Formal analysis: FL, WL, and YX; Investigation: FL, JY, WL, YW, WF, CM, and YX; Data curation: FL, CZ, and CM; Writing – original draft: YM; Writing-review, editing, and revision: FL, CZ, JY, YX, XZ, XG, ML, WL, YW, WF, CM, and YM; Visualization: FL; Supervision: FL and CM; Project administration/oversight: FL and CM; Funding acquisition: YM, JY, and FL.

Funding

This research was funded by The National Natural Science Foundation of China (NSFC) grants 82272767 and 82072691 to Y.M., and 82473477 to Jilong Yang. This work was also supported by the Key Research Project of the First Affiliated Hospital of Nankai University (grant no.: 2025YJZD002), the horizontal research fundings (Funding No.: HKRA2025080246, 320.6750.2025-20-31), the hospital horizontal research fundings (Funding No.: 2019SHHLMCH, 2024YJ023, 2019ZGJCJJLM, 2024YJ020), the Natural Science Foundation of Tianjin established independently by Nankai University (no.: 24JCYBJC01590), and the Tianjin Medical Key Discipline Construction Funding (no.: TJYXZDXK-3–003 A-5),and Funded by Tianjin Key Specialty Construction Project for combining traditional Chinese and Western medicine (No. ZDZKKF02).

Data availability

Raw data will be made available by the corresponding authors upon reasonable request.

Declarations

Ethical approval

This study conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the Ethics committee of Tianjin Beichen Hospital.

Consent to participate

Written informed consent was obtained from the patient to participate in the study.

Consent for publication

Written informed consent was obtained from the patient for publication of the details of their medical case and any accompanying images.

Conflict of interest

Author Y.X. was employed by and F.L. was one of the shareholders of Suzhou Lanma Biotechnology Co.. All the other authors have no conflict of interest to declare.