PiggyBac-Engineered Membrane-Bound IL7 TILs Combined with Anti–PD-1 Antibody Demonstrates Efficacy in Recurrent Ovarian Cancer: A First-in-Human Phase I Trial

Abstract

Purpose:

Recurrent ovarian cancer (rOC) remains an unmet need. This first-in-human phase I trial evaluated GC203, an autologous tumor-infiltrating lymphocyte (TIL) product genetically modified via piggyBac transposon to overcome the immunosuppressive tumor microenvironment.

Patients and Methods:

A cohort of 18 patients with rOC who were heavily pretreated (median, 3.5 prior lines of therapy) underwent lymphodepletion with cyclophosphamide and hydroxychloroquine, followed by GC203 infusion. The primary endpoints were safety and tolerability. Secondary endpoints, assessed in the full analysis set using RECIST 1.1 guidelines, encompassed objective response rate (ORR), progression-free survival (PFS), and overall survival (OS). Exploratory analyses included the Morisita overlap index (MOI) to quantify T-cell receptor repertoire similarity between infused TILs and circulating T cells.

Results:

The GC203 regimen demonstrated a favorable safety profile. Treatment-related adverse events were primarily hematologic, with grade ≥ 3 lymphopenia and neutropenia each occurring in 57% of patients. All proved to be transient with a median resolution of 7 days. The regimen achieved an unconfirmed ORR of 33.3% (6/18 patients) alongside a disease control rate of 83.3%. Survival analyses revealed a median PFS of 7.2 months [95% confidence interval (CI), 1–13.4] and a median OS of 17.1 months (95% CI, 9.5–24.7). A key exploratory analysis identified the MOI as a significant predictor of treatment response, achieving an area under the curve of 0.79.

Conclusions:

The GC203 regimen, combining piggyBac membrane-bound IL7 autologous TILs with anti–PD-1 antibody, demonstrates favorable safety and promising efficacy in heavily pretreated patients with rOC, which supports its further development in this high-need population.

Translational Relevance .

This first-in-human phase I trial demonstrates that GC203, a piggyBac-engineered tumor-infiltrating lymphocyte (TIL) product expressing membrane-bound IL7 (mbIL7), is feasible, safe, and clinically active in heavily pretreated recurrent ovarian cancer (rOC). GC203 regimen achieved a 33.3% objective response rate and 83.3% disease control rate, with a median overall survival of 17.1 months. Mechanistically, mbIL7 enhanced T-cell persistence. The Morisita overlap index emerged as a predictive marker for response, enabling patient stratification. These findings support the continued development of cytokine-engineered TILs as a viable strategy to overcome immune barriers in refractory solid tumors, potentially expanding the application of adoptive cell therapy beyond currently approved indications.

Introduction

Ovarian cancer remains the leading cause of gynecologic cancer mortality, with a 5 years survival rate below 30% despite optimal cytoreductive surgery and platinum-based chemotherapy (1–3). Although poly-ADP-ribose polymerase inhibitors (PARPi) have demonstrated clinical benefit in BRCA-mutated populations (4–6), the rapid development of intrinsic and acquired resistance limits their utility in the majority of patients (7). In the recurrent/refractory setting, therapeutic option remains particularly constrained, as evidenced by the modest 9.6% objective response rate (ORR) observed with PD-(L)1 checkpoint inhibition in a phase IIb trial (8). Although the recent phase III KEYNOTE-B96 trial established pembrolizumab plus chemotherapy as a novel standard for platinum-resistant disease via improvement in survival outcomes (9), the persistent unmet need underscores the necessity for more effective immunotherapeutic strategies.

Conventional adoptive cell therapies have shown limited success in this challenging disease context. A previous study reported 0% ORR in three patients with recurrent ovarian cancer (rOC; ref. 10), whereas our previous investigation of IL2-depleted natural tumor-infiltrating lymphocytes (TIL) achieved only 14% ORR (1/7 patients; ref. 11). This marginA phase II trial evaluating conventional TIL therapy improvement highlights the fundamental constraints imposed by the immunosuppressive tumor microenvironment (TME) inherent to ovarian cancer. The ovarian TME presents multifaceted barriers to T-cell function, including stromal fibrosis, hypoxia, and metabolic reprogramming that collectively impair T-cell infiltration and effector activity (12, 13). Furthermore, constitutive activation of immunosuppressive pathways, such as PD-L1/PD-1, TGFβ, and indoleamine 2, 3-dioxygenase, promotes regulatory T cell and myeloid-derived suppressor cell accumulation while inhibiting TIL cytotoxicity (14, 15).

IL7 emerges as a critical cytokine capable of countering these limitations through its essential role in T-cell homeostasis. By signaling via JAK-STAT5 and PI3K-AKT pathways, IL7 enhances T-cell survival, metabolic fitness, and memory differentiation (16, 17). Building on this rationale, we developed GC203, a novel adoptive cell therapy using piggyBac transposon-engineered autologous TILs expressing membrane-bound IL7 (mbIL7). This first-in-human phase II trial (NCT05468307) evaluated GC203 in patients with heavily pretreated or refractory ovarian cancer, aiming to overcome stromal barriers and T-cell exhaustion through cytokine engineering. Here, we report the safety and efficacy in this trial, providing evidence for mbIL7 as a transformative strategy for this high-need population.

Patients and Methods

Generation of the mbIL7 TILs for clinical study

The manufacturing workflow for mbIL7 TILs utilized in both preclinical investigations and this clinical study was implemented following previously established methodology (11). Resected tumor specimens underwent standardized processing before being transferred to Shanghai Juncell Therapeutics’ GMP-compliant facility for further processing.

During the pre–rapid expansion phase (pre-REP), tumor fragments were cultured in X-VIVO 15 mediums supplemented with recombinant human IL2 (rhIL2, 2000 IU/mL, RRID: AB_2148786), IL7 (10 ng/mL, RRID: AB_2148801), and IL15 (10 ng/mL, RRID: AB_2148817). Following initial expansion, the resulting TIL seed cells were cryopreserved for subsequent processing. For the genetic modification phase, thawed cells were cultured for 2 to 7 days before being harvested and resuspended in an electroporation buffer containing the GC203 plasmid. This plasmid construct, based on the piggyBac transposon system, encodes a glycosylphosphatidylinositol-anchored membrane-bound IL7 (mbIL7-GPI), with the schematic representation available in Supplementary Fig. S1A.

Electroporation was performed using the Nucleofector 2b Device (Lonza) with program U-014, a condition optimized for TIL processing. Following electroporation, the electroporation buffer containing residual plasmid was removed, and cells were transferred into a TIL-specific medium for continued culture. The genetically modified mbIL7-GPI TILs were activated for 2 to 5 days in anti-CD3/CD28 antibody–coated plates, with each antibody used at 5 μg/mL concentration (BioLegend 317302; R&D MAB342-500). The final expansion was conducted through the rapid expansion protocol in X-VIVO 15 mediums supplemented with reduced-dose rhIL2 (300 IU/mL) to achieve clinical-scale production. Transfection efficiency was systematically validated using both flow cytometry and confocal microscopy to ensure product quality and consistency.

Infusion product characterization

Flow cytometric analysis was performed to characterize immune cell populations in infusion products and peripheral blood at designated postinfusion intervals. Sample processing involved staining with titrated fluorochrome-conjugated antibodies (30 minutes, 4°C) in BioLegend Stain Buffer (RRID: AB_2205857), followed by washing procedures to remove unbound antibodies prior to analysis on spectral flow cytometry platforms.

The comprehensive antibody panel incorporated multiple surface and intracellular markers essential for detailed immune subset analysis. CD3 (RRID: AB_2563408), CD4 (RRID: AB_2228842), and CD8α (RRID: AB_2563213) antibodies enabled T-lymphocyte subset discrimination, whereas activation markers included CD25 (RRID: AB_2561861), CD69 (RRID: AB_2924530), and CD137 (RRID: AB_2924530). Coinhibitory receptor profiling utilized antibodies targeting PD-1 (RRID: AB_2566362), LAG-3 (RRID: AB_10637051), TIGIT (RRID: AB_2632929), and TIM-3 (RRID: AB_2565886). Additional differentiation markers comprised CD28 (RRID: AB_2632607) for costimulatory signaling, CD39 (RRID: AB_2750430) for immunosuppressive activity assessment, CD44 (RRID: AB_2076578) for activation status, CD45 (RRID: AB_2561940) for pan-leukocyte identification, CD56 (RRID: AB_2565920) for NK cell detection, and CD57 (RRID: AB_2562459) for terminally differentiated populations. Secondary detection used BD APC streptavidin (RRID: AB_10050396) with appropriate validation controls. Flow cytometry data were analyzed using FCS Express 7 software (RRID: SCR_016431).

Clinical study design

From June 2022 to October 2023, 47 patients with rOC were screened for eligibility at Shanghai Tenth People’s Hospital. The clinical study was conducted according to Declaration of Helsinki and Good Clinical Practice guidelines, and the protocol was approved by the Ethics Committee of Shanghai Tenth People’s Hospital (SHSY-IEC-5.0/22K53/P01). Patients on the study provided written informed consent before enrollment. Data cutoff occurred on December 22, 2024. This single-center, first-in-human phase 1 trial (NCT05468307) assessed the safety and efficacy of mbIL7 TILs in patients with rOC.

Patient characteristics

The study enrolled adult patients (≥18 years) with histologically confirmed OC that had proven refractory to standard therapy. Key inclusion requires an Eastern Cooperative Oncology Group (ECOG) performance status from 0 to 2, indicating preserved functional capacity. Participants were required to have surgically resectable tumors and objectively measurable disease according to RECIST version 1.1 criteria, as confirmed through baseline imaging assessments conducted prior to infusion. A critical inclusion parameter was a life expectancy exceeding 3 months, ensuring patients could adequately complete the planned treatment protocol and initial follow-up assessments (detailed in Supplementary Clinical Protocols).

Treatment procedures

Patients received lymphodepletion with cyclophosphamide (20 mg/kg/d, days −5 to −3) and hydroxychloroquine (600 mg single dose, day −5), with dose/timing adjustments permitted per investigator discretion. Anti–PD-1 therapy (sintilimab, 100 mg; Innovent) was administered 1 hour before TIL infusion (day 0), followed by 100 mg every 6 weeks for 5 doses. Treatment schema and timelines are detailed in Supplementary Clinical Protocols.

Outcomes and follow-up

Primary endpoints focused on safety and tolerability, assessed via adverse events (AE), treatment-related AEs, and serious AEs classified using MedDRA v23.1 (RRID: SCR_003751) with CTCAE v5.0 (RRID: SCR_010296) grading. Additional assessments included physical exams, vital sign monitoring, and clinical laboratory tests (hematology, chemistry, and urinalysis).

Secondary efficacy endpoints, evaluated in the full analysis set (FAS) according to RECIST 1.1 (RRID: SCR_012298) criteria, included ORR, median progression-free survival (mPFS), median overall survival (mOS), and median duration of response (mDOR). Disease control rate (DCR) was defined as the proportion of patients achieving best overall response of complete response (CR), partial response (PR), or stable disease (SD) per RECIST 1.1. SD was strictly defined as sum of diameters (SOD) change between -30% and +20% from baseline maintained for at least 4 weeks (based on study imaging intervals of every 4–6 weeks) or until documented progression/death. This criterion ensured that DCR reflected clinically meaningful disease stabilization. Tumor burden was evaluated via baseline CT or MRI imaging, with subsequent assessments every 4 to 6 weeks during the initial 6 months, every 3 months from months 6 to 24, and every 6 months from months 24 to 30. Response duration was calculated from the infusion date for responding patients.

Exploratory analyses included longitudinal peripheral blood immune profiling by flow cytometry to characterize phenotypic evolution and cellular frequencies. Additional exploratory investigations included cytokine monitoring for cytokine release syndrome (CRS) risk, multiplex immunohistochemistry (mIHC) of pre- and post-treatment biopsies, and T-cell receptor (TCR) repertoire sequencing to determine the Morisita overlap index (MOI) between the infused TILs and circulating T cells.

T-cell clone assessing

RNA was extracted from cryopreserved peripheral blood mononuclear cells (PBMC) and TIL samples using the miRNeasy Mini Kit (Qiagen, cat. #217084). TCR profiling was performed with the QIAseq Immune Repertoire TCR Panel (Qiagen, cat. #334651) for targeted library construction, including reverse transcription with TCR constant-region primers, unique molecular identifier–integrated 5′ adapter ligation, and TCR-specific sequence enrichment. Dual-indexed libraries were quantified (Agilent 2100 Bioanalyzer, RRID: SCR_018043) and sequenced on an Illumina NovaSeq 6000 (RRID: SCR_016387) with 150-bp paired-end reads. Raw data were analyzed by V(D)J alignment using MIXCR (RRID: SCR_018725) v4.0.0, followed by clonotype dynamics analysis with VDJtools (RRID: SCR_027363) TrackClonotypes.

To assess the systemic persistence of adoptively transferred cells, TCR clonotypes present in the TIL infusion product were analyzed in serial PBMC samples. The relative frequency of each such clonotype was calculated as reads for that clonotype being divided by total TCR sequencing reads obtained from the sample, with the results expressed as percentage values.

Statistics analysis

Continuous variables were reported as the mean ± SD or median (range), and categorical variables as counts (%). The efficacy analysis of this study is based on the FAS, which includes all patients who received the GC203 regimen and completed at least one postbaseline tumor assessment. In addition, in accordance with the intention-to-treat (ITT) principle, we have provided a conservative sensitivity analysis that includes all patients who underwent lymphodepletion and received the GC203 infusion. Treatment response was defined by tumor regression. Survival outcomes (OS and PFS) and efficacy endpoints (ORR, DCR, and DOR) were analyzed using Kaplan–Meier methods with 95% confidence interval (CI). An independent samples t test was used for comparison. Statistical analyses, including baseline characteristics, AEs, and clinical outcomes, were performed using IBM SPSS Statistics (RRID: SCR_016479) v29.0 and R v4.4.1 (packages: survival and survminer). The MOI served as the quantitative metric for assessing TCR repertoire similarity between infused TIL products and peripheral circulating T cells, with values spanning from 0 (indicating minimal overlap) to 1 (representing maximal repertoire similarity). This index provided an objective measurement of TCR mobilization efficiency by linking the infused T-cell compartment to systemic immune reconstitution dynamics, based on previous validation work in natural TIL therapy contexts (11). The MOI was calculated using the R package immunarch. Graphs were generated with GraphPad Prism (RRID: SCR_002798) 8.0.2 and R v4.4.1 (ggplot2 package).

Results

Patient characteristics

Between June 2022 and October 2023, a total of 47 patients with heavily pretreated rOC were screened. Following initial evaluation, 13 patients (27.7%) were excluded because of an ECOG performance status score of 3. The remaining 34 patients underwent tumor tissue sampling for TIL manufacturing.

From this cohort, 25 completed lymphodepletion conditioning regimen comprising cyclophosphamide and hydroxychloroquine, subsequently receiving GC203 infusion to establish the ITT population (Supplementary Table S1). The cohort consisted exclusively of Chinese Han patients with advanced disease characteristics, including heavily pretreated (median 3 prior lines) and predominantly platinum-resistant (80%; Supplementary Table S2). The ITT population experienced one fatal event within 30 days after infusion involving a patient who, following ileal lesion resection with anastomosis, developed grade 4 myelosuppression with agranulocytosis culminating in Gram-negative septicemia despite initial stable preinfusion status. Additional exclusions from the ITT group included four patients who initiated new antitumor therapy within 6 weeks after infusion and three patients lacking measurable target lesions at baseline radiographic assessment before infusion. Among these three, two underwent cytoreductive surgery that eliminated measurable lesions following TIL generation tissue procurement, whereas one presented with diffuse peritoneal thickening incompatible with RECIST measurement criteria.

After these exclusions, 18 patients comprised the FAS for efficacy evaluation (Fig. 1), with a median follow-up of 20.4 months (range, 3–29.3) by the December 22, 2024 data cutoff. The challenging disease trajectory was evidenced by only eight patients (44.4%) remaining alive at last follow-up, highlighting the poor prognosis inherent to this heavily pretreated cohort.

Figure 1.

CONSORT flow diagram. Flow diagram shows patient disposition: tumor resection for TIL harvest, lymphodepletion, GC203 infusion, and follow-up. Of 47 screened patients, 34 underwent tumor sampling for TIL manufacturing; nine did not proceed to lymphodepletion. ITT population: 25 patients received lymphodepletion (hydroxychloroquine 600 mg day −5; cyclophosphamide 20 mg/kg/day days −5 to −3) and then GC203 infusion (day 0). Seven were excluded from FAS: four not assessed for efficacy (one with fatal infection <30 days after infusion and three with new antitumor therapy before the first posttreatment assessment) and three with no evaluable target lesions at baseline (post-TIL harvest resection).

Baseline characteristics of the FAS cohort further underscore the refractory nature of their disease, with a median age of 52 years (range, 32–70) and 55.6% exhibiting ECOG performance status 1. All patients presented with metastatic lesions (median 3.3 lesions) and substantial tumor burden (median target lesion diameter 58.7 mm). Prior treatment exposure was extensive (median 3.5 lines, range: 1-10), including platinum-based chemotherapy (100%), PARPis (66.7%), and immunotherapy (27.8%), with 66.7% (12/18) meeting criteria for platinum-resistant disease (Table 1).

Table 1.

Characteristics of participants.

Characteristics	Total (n = 18)
Median age (range), years	52 (32–70)
Median time since initial diagnosis (range), months	34.8 (13.5–148.6)
Median no. of previous systemic therapies (range)	3.5 (1–10)
Median SOD (range), mm	57 (13–146)
ECOG score, n (%)	​
1	10 (55.6)
2	8 (44.4)
FIGO stage at initial diagnosis, n (%)	​
I	1 (5.6)
II	2 (11.1)
III	13 (72.2)
IV	2 (11.1)
Histology, n (%)	​
High-grade serous carcinoma	12 (66.7)
Ovarian clear cell carcinoma	2 (11.1)
Ovarian endometrioid carcinoma	2 (11.1)
Granulosa cell tumor	1 (5.55)
Low-grade mucinous carcinoma	1 (5.55)
Recurrent/metastatic site, n (%)	​
Local infiltration (including peritoneal implantation)	15 (83.3)
Lymph node	8 (44.4)
Hematogenous spread	8 (44.4)
Recurrent/metastatic lesions ≥3, n (%)	9 (50)
SOD ≥60 mm, n (%)	9 (50)
Previous systemic therapy, n (%)	​
Chemotherapy	18 (100)
Targeted therapy	16 (88.9)
PARPis	12 (66.7)
Immunotherapy	5 (27.8)
Radiotherapy	2 (11.1)
No. of previous surgical intervention, n (%)	​
1	9 (50.0)
2	7 (38.9)
3	2 (11.1)
Platinum-based chemotherapy sensitivity, n (%)	​
Platinum-sensitive	6 (33.3)
Platinum-resistant	12 (66.7)
CA125 levels, n (%)	​
<35	8 (44.4)
35–70	4 (22.2)
>70	6 (33.3)

In terms of treatment delivery, the median infused TIL dose was 2.7 × 10⁹ cells (range: 1.2 × 10⁹–6.5 × 10⁹), with 64.5% of cells expressing mbIL7 (range, 10.7%–98.5%). Flow cytometry analysis of thawed products confirmed the presence of CD4⁺ and CD8⁺ T-cell subsets and revealed a stem-like exhausted phenotype (CD62L⁺/LAG3⁺/TIM3⁺; Supplementary Fig. S2), a profile that may influence response to mbIL7-driven reinvigoration.

Safety and tolerability of the GC203 regimen

The GC203 regimen demonstrated a favorable safety profile with no long-term complications during the study period. All 18 patients experienced AEs (Table 2), consistent with established patterns from previous investigations (11). No unexpected or novel safety concerns emerged during follow-up.

Table 2.

AEs.

Preferred terma	AE, n (%)
	Grade 2	Grade 3	Grade 4	Grade 3/4 total
Any AEb	15 (83)	13 (72)	4 (22)	17 (94)
Lymphocyte count decreased	7 (39)	7 (39)	3 (17)	10 (57)
Neutrophil count decreased	4 (22)	7 (39)	3 (17)	10 (57)
Leukocyte count decreased	5 (28)	7 (39)	2 (11)	9 (50)
Anemia	4 (22)	3 (17)	0 (0)	3 (17)

^aMedical Dictionary for Regulatory Activities version 23.1, graded according to CTCAE version 5.0.

^bAEs occurred in ≥25% of patients or occurred with at least one grade 2 or more AEs in the patients who received the GC203 regimen.

The most common grade 3 or higher AEs were hematologic toxicities related to preconditioning, including lymphopenia and neutropenia (each in 56%, 10 of 18 patients), leukopenia (50%, 9 of 18), and anemia (17%, 3 of 18). These hematologic toxicities, occurring within 30 days after infusion, typically resolved to grade 2 within a median of 7 days (range: 4–25; Supplementary Fig. S3; Supplementary Table S3).

Cytokine analyses referenced immediate pre-TIL infusion (day 0) as the primary baseline, IL7 and IFNγ showed significant increases, with peak levels observed within 30 days. In contrast, IL10, IL8, and IL6 remained stable, suggesting selective modulation of proinflammatory pathways without systemic hyperactivation (Supplementary Fig. S3).

Efficacy of the GC203 regimen

The primary efficacy evaluation used RECIST 1.1 criteria to assess ORR, mPFS, mOS, and mDOR (18). A distinctive aspect of our response classification system defined responders as patients exhibiting any degree of tumor size reduction or no change relative to baseline, whereas nonresponders demonstrated tumor progression. This approach addressed limitations observed in previous investigations in which conventional TIL therapy reported 0% ORR despite 66.7% transient SD in patients with ovarian cancer (n = 3; ref. 10), highlighting a critical limitation of conventional RECIST thresholds in immunotherapy trials for immunosuppressive tumors like ovarian cancer.

In the FAS of 18 efficacy-evaluable patients, the GC203 regimen demonstrated substantial antitumor activity with an unconfirmed ORR of 33.3% (95% CI, 16.3–56.3), comprising two CRs (11.1%) and four PRs (22.2%). The DCR reached 83.3% (95% CI, 60.8–94.2), with one CR and one PR subsequently confirmed with sustained remission. In the ITT population (n = 25), unconfirmed ORR was 28% (7/25; 95% CI, 12.1–49.4; Fig. 2A and B; Supplementary Table S1). Supplemental Figure S4A presented the SOD of targeted lesions in 18 patients prior to their discontinuation from the study.

Figure 2.

Response after GC203 regimen. A, Swimmer plot showing time-to-event outcomes from TIL infusion, with vertical markers indicating response assessment and long-term follow-up. Patients are stratified by best overall response (CR, PR, SD, and PD). B, Waterfall plot of maximum tumor burden reduction from baseline (RECIST v1.1). Asterisk denotes pseudoprogression in patient P03, characterized by transient tumor enlargement preceding immune-mediated tumor necrosis. C, Kaplan–Meier analysis of OS after GC203 infusion. EOT, end of treatment; PD, progressive disease; SD, stable disease.

Survival outcomes further supported the regimen’s clinical utility, with mPFS of 7.2 months (95% CI, 1–13.4; Supplementary Fig. S4B) and mOS of 17.1 months (95% CI, 9.5–24.7; Fig. 2C). Survival rates at 6, 12, and 24 months were 72.2% (95% CI, 54.2–96.2), 55.6% (95% CI, 36.8–84), and 49.4% (95% CI, 30.8–79.3), respectively. The mDOR was 9.1 months (95% CI, 7.3–10.9) and median duration of clinical benefit was 9.1 months (95% CI, 7.3–11; Table 3).

Table 3.

Clinical response.

Variable	n = 18
Best overall response	​
CR, n (%)	2 (11.1)
PR, n (%)	4 (22.2)
SD, n (%)	9 (50)
PD, n (%)	3 (16.7)
ORR, n (%) [95% CI]	6 (33.3) [16.3, 56.3]
DCR, n (%) [95% CI]	15 (83.3) [60.8, 94.2]
TTP (months) [95% CI]	2.6 [0.9, 5]
DOR (months) [95% CI]	9.1 [7.3, 10.9]
mDoCB (months) [95% CI]	9.1 [7.3, 11]
mPFS (months) [95% CI]	7.2 [1, 13.4]
mOS (months) [95% CI]	17.1 [9.5, 24.7]
6-month OS rate (%) [95% CI]	72.2 [54.2, 96.2]
12-month OS rate (%) [95% CI]	55.6 [36.8, 84]
24-month OS rate (%) [95% CI]	49.4 [30.8, 79.3]

Abbreviations: mDoCB, median duration of clinical benefit; PD, progressive disease; TTP, time to progress.

Notably, efficacy persisted in high-risk subgroups, including patients with >4 prior lines of therapy achieving an ORR of 40% and DCR of 80%, despite a shorter mPFS of 3.4 months, confirming activity in heavily pretreated disease. The platinum-resistant subset (n = 12) maintained 25% ORR, supporting the utility of GC203 regimen in refractory settings. Response to bridging therapy emerged as a significant prognostic factor, with SD patients exhibiting 35.7% ORR versus 0% in patients with progression.

Histologic subgroup analyses revealed consistent antitumor activity of GC203 regimen across various ovarian cancer subtypes. The high-grade serous carcinoma (HGSC) cohort (n = 12) demonstrated 25% ORR and 83.3% DCR with mPFS of 5.3 months. Promising activity was also observed in rare histologic subtypes (detailed in Supplementary Table S4), collectively supporting the promising antitumor activity of the GC203 regimen.

Mechanistic insights from longitudinal case analyses

Patient 01 presented with heavily pretreated rOC that had proven refractory to multiple prior therapies, including natural TIL therapy (NCT04766320). This case demonstrated metastatic lesion regression accompanied by dynamic TCR clonotype evolution following GC203 infusion. TCRβ sequencing revealed initial expansion of infused TIL clones, followed by sustained proliferation of noninfused, tumor-reactive T-cell clonotypes within the host repertoire (Fig. 3A and B). This pattern of sequential immune reconstitution suggests that mbIL7 may sustain the survival and activation of preexisting host-derived clones, potentially contributing to the observed clinical response.

Figure 3.

Mechanistic insights from two case analyses. A, Serial CT imaging of P01 (baseline to day 90) demonstrating sustained regression of metastatic lung/liver lesions, summed target diameters: 146 mm (baseline) to 70 mm (day 30). B, TCR clonotype dynamics in P01 at day 90 after infusion: exogenous (blue) and endogenous (red) clonotype frequencies quantified via sequencing. C, Pseudoprogression in P03: transient tumor enlargement (day 130) followed by necrotic clearance (day 210), confirmed histologically. D, Immunofluorescence mapping of immune infiltration: the residual tumor posttreatment CD4⁺/CD8⁺ T-cell and CD20⁺ B-cell enrichment vs. stromal-restricted tumor for TIL.

Patient 03 exhibited pseudoprogression in hepatic and splenic metastases that transitioned to necrosis at 6 months despite RECIST-defined progression. Analysis of the hepatic TME revealed substantial intratumoral mbIL7 enrichment, accompanied by upregulation of proinflammatory chemokines CXCL9, CXCL10, and CCL18, with concurrent downregulation of myeloid-recruiting chemokines CXCL1, CXCL3, CXCL5, CXCL8, CXCL17, CCL11, CCL20, and CCL21 (Fig. 3C; Supplementary Fig. S5A and S5B). This chemokine profile alteration correlated with spatial reorganization of immune cell populations, characterized by spatial redistribution of PD-1⁺CD8⁺ T cells, CD20⁺ B cells, and CD4⁺ T cells to tumor cores (Fig. 3D). The observed immune cell redistribution and tertiary lymphoid structure formation suggest enhanced intratumoral immune activity. These preliminary findings indicate GC203 regimen may modulate chemokine networks and affect in situ T-cell activation but require further validation to confirm mechanistic links.

Determinants of GC203 regimen efficacy

The antitumor efficacy of regimen GC203 is mediated through a coordinated interaction of TIL product dynamics, TME status, and tumor-intrinsic features, with these elements contributing to response. Longitudinal flow cytometric analysis of peripheral blood revealed dynamic T-cell reconstitution after GC203 infusion, in which CD8⁺ T cells undergo rapid expansion within 24 hours, transient contraction by day 3, and subsequent progressive expansion through day 60, suggesting sustained peripheral persistence and potential tumor trafficking of the administered TIL products (Fig. 4A).

Figure 4.

TIL kinetics and persistence in peripheral blood. A, CD8⁺ T-cell kinetics in peripheral blood after GC203 infusion, showing expansion (days 0–14) followed by sustained persistence (days 14–90). B, Frequency indicates the relative abundance of individual TCR clonotypes (from the infused TIL product) within the total TCR repertoire of postinfusion PBMCs, reflecting their systemic persistence. C, Box plots illustrate day 14 MOI comparison in nonresponders and responders by t test (P = 0.038). D, Kaplan–Meier analysis revealed that responders exhibited significantly improved OS compared with nonresponders, with a log-rank P value of 0.014. E, ROC curve of the diagnostic model, showing sensitivity vs. 1− specificity. The AUC is 0.79 (95% CI, 0.54–1). F, Box plots showing the percentage of PD-1 expression on CD3⁺ T cells in nonresponders vs. responders was determined by t test (P = 0.005). * AUC, area under the curve; ROC, receiver operating characteristic.

Complementary TCRβ sequencing analysis of longitudinal PBMCs tracked the fate of infused TIL clones, revealing a biphasic kinetic pattern characterized by peak representation in the circulating TCR repertoire at day 7 followed by a contraction phase (Fig. 4B). This biphasic trajectory with early expansion and subsequent contraction paralleled peripheral T-cell expansion kinetics and indicated robust in vivo clonal amplification prior to tumor infiltration.

Longitudinal analysis of the MOI at postinfusion time points (days 0, 7, 14, and 30) revealed that day 14 represented the critical juncture in which responders exhibited significantly higher MOI values than nonresponders (P = 0.038; Fig. 4C). This enhanced TCR alignment correlated with superior clinical outcomes, including significantly improved ORR with 80% ORR in responders and 28.6% ORR in nonresponders, and prolonged OS (P = 0.014; Fig. 4D). Receiver operating characteristic analysis confirmed the MOI’s robust predictive capacity for GC203 regimen response, with an area under the curve of 0.79 (Fig. 4E).

Preinfusion peripheral T cells from responders exhibited significantly lower PD-1 expression on circulating T cells than those from nonresponders (P = 0.005; Fig. 4F), suggesting reduced immunosuppressive TME pressure at baseline. Responders further displayed a stem-like exhausted TIL phenotype characterized by elevated lymph node homing marker CD62L (19), costimulatory receptors CD28/CD137 (20, 21), and activation marker CD25 (22), alongside exhaustion markers (23) (Supplementary Fig. S6). This phenotype implied enhanced homing to the TME and sustained activation within immunosuppressive niches.

Genomic analysis identified differential mutation patterns between response groups, with responders showing mutations in CEP90, HKDC1, and ASPM (Supplementary Fig. S7), linked to neoantigen exposure and cell metabolism (24–26). In contrast, nonresponders displayed higher tumor burden at baseline (SOD ≥60 mm) and mutations in AHNAK2, ANK3, FASN, and BTAF1 (Supplementary Fig. S7), which might be associated with immunosuppression via myeloid niche stabilization and antigen exclusion (27–29).

These data collectively establish a multidimensional framework characterizing GC203 regimen efficacy, in which high MOI levels at day 14, low preinfusion PD-1 expression, favorable tumor genetics, and tumor size were observed to correlate with enhanced therapeutic response. These findings provide a biomarker-driven strategy for optimizing patient selection and predicting outcomes in TIL-based immunotherapy.

Discussion

This first-in-human phase I trial demonstrates that GC203, a novel piggyBac-transposon–engineered TIL product expressing mbIL7, represents a promising therapeutic strategy for heavily pretreated rOC. The regimen exhibited a favorable safety profile, with no TIL-related grade 4/5 AEs, and immune-related AEs, including CRS, were low-grade and manageable. Although one grade 5 event of COVID-19 pneumonia was deemed unrelated to the investigational product, the potential contribution of lymphodepletion-related cytopenia to infection susceptibility cannot be entirely excluded.

The GC203 regimen demonstrated clinically meaningful efficacy with an ORR of 33.3% and a DCR of 83.3%, outperforming historic outcomes of conventional TIL therapy in comparable patient populations (14% ORR; ref. 11). The responses proved durable responses with a median duration of 9.1 months. The manufacturing process achieved 100% success rate in TIL product production, supporting the feasibility of this approach.

This trial omitted a traditional dose-escalation design, based on two key considerations: First, prior clinical evidence demonstrated that autologous TILs can be safely expanded and administered across a broad dose range without dose-limiting toxicities, supporting the feasibility of an expansion-driven dosing approach (11). Second, GC203 is specifically engineered to express mbIL7, which preclinically enhances T-cell persistence and fitness at feasible doses without additional toxicity, obviating the need for exploring efficacy through dose gradients. Consequently, we administered the maximum feasible dose of GC203 (median 2.7 × 10⁹ viable cells) from each manufacturing run to ensure delivery of a biologically active product while avoiding unnecessary dose escalation in this heavily pretreated patient population.

The combinatorial lymphodepletion and immunomodulatory regimen was designed to tackle critical limitations of TIL therapy. Low-dose cyclophosphamide lymphodepletion strategically balances immunosuppression reduction and effector T-cell preservation to potentiate antitumor responses (30). Hydroxychloroquine inhibits autophagy-mediated MHC downregulation by blocking lysosomal degradation of antigen-presenting machinery (31). For PD-1 blockade, we departed from sintilimab’s standard 200 mg every 3 weeks schedule, giving 100 mg before TIL infusion (day 0) followed by 100 mg every 6 weeks for 5 doses. This deliberate adjustment balanced mitigating PD-1/PD-L1–mediated TME immunosuppression.

Our preclinical studies supported the mechanistic rationale that mbIL7 expression promotes T-cell persistence with superior cytotoxic function and enhances therapeutic efficacy in xenograft models, outperforming unmodified TILs (Supplementary Fig. S1). Unlike systemic IL2, mbIL7’s GPI anchor restricts bioavailability to the TME, avoiding vascular leak syndrome and capillary leak syndrome (32, 33). Spatial transcriptomics revealed that mbIL7 upregulated proinflammatory chemokines CXCL9/10 with concurrent suppression of immunosuppressive CXCL1/3, recruiting PD1⁺CD8⁺ T cells into tumor nests (Fig. 3D). The MOI emerged as a robust predictive marker, with day 14 values significantly discriminating responders from nonresponders and correlating with improved clinical outcomes.

Unlike conventional bridging therapy, designed to control disease pending product manufacture, our study used a deferred-use strategy, in which some patients received subsequent standard-of-care lines for progression after tumor tissue procurement, with premanufactured TIL product banked for later-line administration. This strategy, involving tumor tissue during clinically indicated surgery, banking the TIL product for future use upon progression, enhances accessibility via early cell procurement and product readiness for later-line use. Beyond flexibility, this approach demonstrated clinical merit, where one ITT platinum-resistant recurrent patient with no measurable target lesion after surgery remained lesion-free for >2 years after GC203 infusion. This preliminary observation suggests earlier intervention with TILs during minimal residual disease may yield superior outcomes, a hypothesis warranting investigation in earlier-line–focused trials.

Several limitations characteristic of early-phase studies should be acknowledged, including the modest cohort size and the integrated nature of the treatment regimen. which limits precise determination of the engineered TIL product’s independent contribution. A key limitation is the restricted genomic analysis, which focused on tumor mutational burden–associated genes and omitted hallmark alterations (e.g., TP53, BRCA1/2, NF1, CCNE1, and MYC) typical of the majority HGSC cohort, limiting interpretability and completeness of response-genetic correlations. Future studies should focus on validating the predictive value of the MOI in larger cohorts, optimizing lymphodepletion strategies, and prioritizing comprehensive genomic landscape mapping to further elucidate mechanisms of response and resistance.

The promising safety and efficacy data, coupled with the mechanistic insights into mbIL7’s mode of action, support further clinical development of GC203 in advanced ovarian cancer and potentially other solid tumors in which TIL therapy has shown limited efficacy to date.

Conclusion

The GC203 regimen represents a paradigm shift in TIL therapy, combining cytokine engineering with precision immunomodulation to overcome immunosuppressive barriers in rOC. Whereas early-phase limitations preclude definitive conclusions, the observed durability and safety profile justify phase II exploration. By addressing key shortcomings of conventional TIL therapy of systemic toxicity and suboptimal TME engagement, this study advances the frontier of adoptive cell therapy for refractory solid tumors.

Supplementary Material

Supplementary Data 1

Supplementary materials and methods, Figure S1-S7, supplementary figure legends, and Table S1-S4.

Data Availability

The raw whole-exome sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject number PRJNA1467603 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1467603). Raw TCR repertoire sequencing data generated and analyzed in this study have been deposited in the NCBI Gene Expression Omnibus under accession number GSE331278.

Authors’ Disclosures

H. Jin reports grants from the Science and Technology Commission of Shanghai Municipality during the conduct of the study and nonfinancial support from Shanghai Juncell Therapeutics, Co., Ltd. outside the submitted work. No disclosures were reported by the other authors.

Authors’ Contributions

J. Guo: Conceptualization, resources, data curation, formal analysis, writing–original draft. Y. Wu: Data curation, software, formal analysis, methodology. W. Huang: Data curation, formal analysis. G. Ai: Data curation, validation. C. Wang: Software, visualization. N. Luo: Resources, methodology. J. Zhu: Project administration. Y. Zhou: Validation. W. Shi: Project administration. J. Ding: Data curation. Y. Ge: Data curation. W. Feng: Supervision. H. Jin: Funding acquisition, methodology, project administration. B. Zhao: Supervision, validation, visualization. Z. Cheng: Conceptualization, resources, supervision, investigation, project administration.