TQB2618 plus Penpulimab, Alone or in Combination with Chemotherapy, for Recurrent or Metastatic Nasopharyngeal Carcinoma: A Multicenter, Two-Cohort, Phase II Trial

Abstract

Purpose:

The purpose of this study was to evaluate the efficacy and safety of dual immunotherapy blocking T-cell immunoglobulin and mucin domain–containing protein-3 (TIM-3) and programmed cell death protein-1 (PD-1) in recurrent/metastatic nasopharyngeal carcinoma.

Patients and Methods:

The TQB2618 plus penpulimab (TP) cohort enrolled patients with progression on prior PD-1 or programmed cell death ligand-1 blockade to receive intravenous TP every 3 weeks. The TP combined with gemcitabine–cisplatin (TPGC) cohort enrolled treatment-naïve patients to receive TQB2618, penpulimab, gemcitabine, and cisplatin every 3 weeks for four to six cycles, followed by maintenance dual immunotherapy. Primary endpoints were dose-limiting toxicity (DLT) and objective response rate for the TP cohort and progression-free survival (PFS) for the TPGC cohort. This two-cohort trial is registered with ClinicalTrials.gov (NCT05563480).

Results:

Between November 2022 and October 2023, 17 patients were enrolled in the TP cohort and 30 in the TPGC cohort. No DLTs were observed. In the TP cohort, disease control was achieved in 10 (58.8%) of 17 patients, with no complete or partial responses; the median PFS was 1.6 months [95% confidence interval (CI), 0–3.2]. In the TPGC cohort, the objective response rate was 86.2%, including 4 (13.8%) complete responses. The median PFS was 10.8 months (95% CI, 9.6–16.4), with a 12-month rate of 40.9%; the median overall survival was unreached. Grade 3 to 4 treatment-related adverse events occurred in 2 (11.8%) patients in the TP cohort and 25 (83.3%) in the TPGC cohort, predominantly hematologic toxicities.

Conclusions:

TP combined with gemcitabine–cisplatin demonstrated encouraging antitumor activity and a manageable safety profile in treatment-naïve patients. However, the chemotherapy-free dual regimen showed limited efficacy in immunotherapy-refractory disease.

Translational Relevance.

This multicenter, single-arm, two-cohort, phase II trial provides the first prospective clinical evidence on dual T-cell immunoglobulin and mucin domain–containing protein-3 (TIM-3) and programmed cell death protein-1 (PD-1) blockade in recurrent or metastatic nasopharyngeal carcinoma (NPC). By evaluating this strategy both as subsequent-line therapy in patients with established resistance to PD-1 or programmed cell death ligand-1 blockade and in combination with gemcitabine–cisplatin chemotherapy as first-line treatment, the study addresses two distinct and clinically relevant settings. The findings show that dual TIM-3 and PD-1 blockade exhibits encouraging antitumor activity when combined with chemotherapy in treatment-naïve patients, whereas this regimen has a manageable safety profile but limited efficacy in patients with immunotherapy-resistant disease. These results extend current knowledge beyond PD-1–centered strategies and provide early clinical validation of TIM-3 as a therapeutic target in NPC. Further randomized controlled trials directly comparing the safety and efficacy of this regimen with the current standard of care, PD-1 blockade plus gemcitabine–cisplatin chemotherapy, are needed to establish the clinical role of TIM-3 blockade.

Introduction

Nasopharyngeal carcinoma (NPC), a prevalent head and neck malignancy in Southeast Asia, is characterized by a high rate of distant metastasis, affecting approximately 15% of patients at initial diagnosis (1). Around 20% of patients with locoregionally advanced NPC subsequently develop recurrent or metastatic disease (R/M NPC) following definitive treatment (2). In the immunotherapy era, the addition of monoclonal antibodies targeting programmed cell death protein-1 (PD-1; refs. 3–5) or programmed cell death ligand-1 (PD-L1; ref. 6) to platinum-based chemotherapy has significantly improved clinical outcomes, leading to the establishment of chemoimmunotherapy as the standard first-line treatment for R/M NPC (7).

However, with the expanding use of PD-1 blockade in the locoregionally advanced setting, increasing concern has arisen that patients who experience disease relapse may have already developed resistance to PD-1–directed therapy, thereby complicating the role of immunotherapy in R/M NPC (8–10). For patients who progress after first-line treatment, therapeutic options remain limited to previously unused chemotherapy or PD-1 blockade monotherapy (7), both of which confer only modest benefit because of treatment intolerance and acquired resistance. These challenges highlight the urgent need for novel therapeutic strategies with distinct immunologic mechanisms. T-cell immunoglobulin and mucin domain–containing protein-3 (TIM-3) is expressed on multiple immune cell subsets and functions as a key negative regulator of immune activation (11). Upregulation of TIM-3 has been closely associated with T-cell exhaustion and tumor immune evasion and has been implicated in resistance to PD-1/PD-L1 blockade (12). Preclinical studies indicate that concurrent inhibition of TIM-3 and PD-1 may delay or overcome resistance to PD-1 blockade (13). Moreover, early exploratory evidence from solid tumors suggests that dual TIM-3 and PD-1 blockade can achieve greater antitumor activity than PD-1 blockade alone (14), supporting its development as a promising immunotherapeutic approach. To date, no clinical trial has evaluated this dual immunotherapy with TIM-3 and PD-1 blockades in NPC.

TQB2618 is a monoclonal antibody that inhibits the interaction between TIM-3 and its ligands. In this study, we report the results of a two-cohort, phase II trial evaluating the safety and preliminary antitumor activity of TQB2618 combined with PD-1 blockade using penpulimab. This regimen was administered either as dual immunotherapy [TQB2618 plus penpulimab (TP) cohort] in the subsequent-line setting or in combination with gemcitabine–cisplatin chemotherapy (TPGC cohort) as first-line treatment for patients with R/M NPC.

Patients and Methods

Study design and patients

This multicenter, single-arm, phase II trial comprised two parallel cohorts designed to evaluate the safety and preliminary antitumor activity of TP in patients with R/M NPC. The representativeness of study participants is shown in Supplementary Table S1. The TP cohort assessed this dual immunotherapy as subsequent-line treatment in three hospitals in China, whereas the TPGC cohort evaluated the same immunotherapy combined with gemcitabine–cisplatin chemotherapy as first-line treatment in five hospitals in China (Supplementary Table S2). This two-cohort design was determined at the study’s inception, with the aim of comprehensively assessing the value of TIM-3 blockade in R/M NPC. As such, the two cohorts are reported as an integrated whole in this report. The trial was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice guidelines. The study protocol and all amendments were approved by the ethics committee of the leading center, Sun Yat-sen University Cancer Center (A2022-103-X01), as well as by the ethics committees of all participating institutions. Written informed consent was obtained from all patients before enrollment. The authenticity of this article has been validated by uploading the standardized patient-level data onto the Research Data Deposit platform (www.researchdata.org.cn) after deidentification, with the accession number RDDA2026486326.

The eligibility criteria common to both cohorts included histologically confirmed NPC, an Eastern Cooperative Oncology Group performance status of 0 or 1, an estimated life expectancy of at least 3 months, and the presence of at least one measurable lesion according to Response Evaluation Criteria in Solid Tumors (RECIST; RRID: SCR_026435) version 1.1. Adequate organ function was required, defined as an absolute neutrophil count of at least 1.5 × 10⁹/L, hemoglobin concentration of at least 90 g/L, platelet count of at least 100 × 10⁹/L, total bilirubin concentration less than 1.5 times the upper limit of normal (ULN), aspartate aminotransferase and alanine aminotransferase concentrations less than 2.5 times ULN (or up to 5 times ULN in the presence of liver metastases), and a creatinine clearance rate greater than 60 mL/minute calculated using the Cockcroft–Gault formula.

The patients enrolled in the TP cohort were 18 to 75 years of age and had radiographically confirmed disease progression according to RECIST version 1.1 following at least one line of platinum-based chemotherapy combined with PD-(L)1 blockade. These patients could have received up to two prior lines of immunotherapy for R/M NPC, and the most recent immunotherapy was required to have achieved stable disease for at least 6 weeks, a confirmed partial response, or a treatment duration of at least 12 weeks. The patients enrolled in the TPGC cohort were 18 to 70 years of age and were required to be treatment-naïve for R/M NPC. Previous chemoradiotherapy for locoregionally advanced NPC was not considered prior to systemic treatment unless recurrence or distant metastasis occurred within 6 months after completion of therapy. Previous exposure to immunotherapy in the radical treatment setting was permitted, provided that no more than one immune checkpoint inhibitor, excluding bispecific antibodies or the study agents used in this trial, had been administered.

The key exclusion criteria included a history of malignancy within 5 years before enrollment (with the exception of curatively treated basal cell carcinoma of the skin, cervical carcinoma in situ, or papillary thyroid carcinoma), unresolved adverse events of grade 1 or higher severity, active autoimmune disease or immunodeficiency, medical conditions requiring systemic immunosuppressive therapy, other active malignancies, active infection with hepatitis B or C virus, syphilis, or COVID-19, active central nervous system metastases, prior exposure to TIM-3 blockade, radiologic evidence of tumor invasion into major blood vessels, thrombotic or cardiovascular events within 6 months before the first dose, major surgery (excluding needle or endoscopic biopsy), traumatic injury, or anticancer therapy within 28 days before the first dose, and a history of re-irradiation for recurrent NPC. Full inclusion and exclusion criteria are provided in the Study Protocol (pp 34–39).

Procedures

Both cohorts consisted of a safety lead-in phase followed by an expansion phase. In the TP cohort, a safety lead-in phase using a standard 3 + 3 dose-escalation design was conducted to determine the recommended dose of TQB2618 (1,200 or 1,500 mg) in combination with a fixed dose of penpulimab (200 mg; RRID: AB_3694946). Both agents were administered intravenously on day 1 of each 21-day cycle. Only patients who received the planned doses and completed all scheduled assessments during cycle 1 were considered evaluable for dose escalation. If a patient did not meet these criteria, an additional patient was enrolled as a replacement, whereas the original patient continued treatment until withdrawal or study completion. Treatment was continued until disease progression, unacceptable toxicity, or for a maximum duration of 2 years.

In the TPGC cohort, the patients received four to six cycles of combination therapy comprising TQB2618 1,200 mg (day 1), penpulimab 200 mg (day 1), gemcitabine 1 g/m² (days 1 and 8), and cisplatin 75 mg/m² (day 1), administered once every 3 weeks. Upon completion of combination therapy, the patients proceeded to maintenance dual immunotherapy with TP every 3 weeks until disease progression, unacceptable toxicity, or a maximum treatment duration of 2 years.

Dose reductions of TQB2618 or penpulimab were not permitted. Dose modifications of gemcitabine and cisplatin were allowed and were implemented in accordance with the prespecified protocol. No concomitant antitumor therapy was permitted during the study period, except for palliative local radiotherapy or surgery for nontarget lesions present at baseline. Administration of either immunotherapy agent could be delayed for up to 12 weeks to allow recovery from immune-related adverse events before treatment resumption. If TQB2618 was permanently discontinued because of immune-related toxicity, penpulimab could be continued as monotherapy, and vice versa. Early termination of the study could be initiated by the sponsor, ethics committees, or investigators for reasons including severe adverse events posing substantial risk to patients, administrative withdrawal, or sponsor decision.

Assessment

Physical examination, hematologic testing, serum biochemical profiling, urinalysis, and electrocardiography were performed before treatment on day 1 of each treatment cycle. Cell-free Epstein–Barr virus (EBV) DNA levels were quantified using qRT-PCR from baseline and serial on-treatment blood samples. Tumor PD-L1 membrane expression was assessed exclusively in patients in the TPGC cohort using formalin-fixed, paraffin-embedded tumor samples obtained on a voluntary basis. All analyses were performed at a central laboratory using a validated immunohistochemistry assay (Dako 22C3 pharmDx; Agilent Technologies; RRID: AB_2833074). PD-L1 positivity was defined as a tumor proportion score of at least 1%. Tumor responses were evaluated by site investigators according to RECIST version 1.1 using computed tomography or magnetic resonance imaging. Imaging assessments were performed at baseline, every two treatment cycles during the first year, and every three cycles thereafter until disease progression, study discontinuation, withdrawal of informed consent, initiation of alternative antitumor therapy, or death, whichever occurred first. Adverse events were monitored throughout the study and graded according to the Common Terminology Criteria for Adverse Events (CTCAE; RRID: SCR_010296) version 5.0.

Outcomes

The primary endpoints for the TP cohort were dose-limiting toxicity (DLT) and objective response rate. The DLT was defined as the occurrence of treatment-related adverse events (trAE) during the first 21-day treatment cycle, graded according to the CTCAE version 5.0. The DLTs included, but were not limited to, grade 4 or higher hematologic toxicities, grade 3 or higher nonhematologic toxicities, or grade 3 infusion-related reactions that did not resolve within 6 hours after infusion discontinuation. The objective response rate was defined as the proportion of patients who achieved a radiologically confirmed complete response or partial response. The primary endpoint for the TPGC cohort was progression-free survival (PFS), assessed as the median PFS and the PFS rate at 12 months. The PFS was defined as the time from enrollment to radiologic disease progression, death from any cause, or censoring at the date of the last disease assessment. For the TPGC cohort, overall survival (OS), defined as the time from enrollment to death, was regarded as a secondary endpoint, and the association between PD-L1 expression and PFS was regarded as a prespecified exploratory endpoint. The secondary endpoints for both cohorts included the duration of response, defined as the time from the first documentation of complete or partial response to disease progression or death, and disease control rate, defined as the proportion of patients achieving complete response, partial response, or stable disease, and safety.

Statistical analysis

In the TP cohort, three additional patients were enrolled at a given dose level if a DLT was observed among the first three patients. The safety lead-in phase enrolled 6 to 12 patients and was followed by an expansion phase planned to include 18 to 24 patients. In the TPGC cohort, if the treatment regimen was tolerated in the first three to six evaluable patients, an expansion phase enrolling 24 to 27 patients was initiated. Both cohorts were planned to enroll a total of 30 patients. Safety analyses were conducted in all patients who received at least one dose of study treatment and had available safety data (safety analysis set). Antitumor activity analyses were performed in patients who received at least one dose of treatment and had at least one evaluable efficacy assessment (efficacy analysis set). Survival outcomes were analyzed in all enrolled patients according to the intention-to-treat principle (full analysis set). The PFS, OS, and duration of response were estimated using the Kaplan–Meier method (RRID: SCR_024521), with the corresponding 95% confidence intervals (CI). Patients who remained alive without disease progression at the time of analysis or who were lost to follow-up were censored at the date of the last disease assessment. Prespecified subgroup analyses were performed to explore the association between PD-L1 expression and PFS. Post hoc analyses were performed to investigate the relationship between the cell-free EBV DNA and PFS. The Kaplan–Meier method was used to estimate survival, which was compared using a log-rank test. An unstratified Cox proportional hazards model was used to estimate the hazard ratio (HR) and 95% CI. No interim analyses or sensitivity analyses were planned. The last follow-up date was September 17, 2024, for the TP cohort and December 20, 2024, for the TPGC cohort. Owing to limited early efficacy, enrollment in the TP cohort was halted during the expansion phase, and the cohort was formally terminated on February 26, 2025. All analyses were conducted after database lock on June 30, 2025, using PRISM version 10.5.0 (GraphPad Software Inc.; RRID: SCR_002798) and SPSS version 31.0 (IBM; RRID: SCR_016479). This two-cohort study is registered with ClinicalTrials.gov (NCT05563480).

Results

Patients and treatment

In the TP cohort, 17 patients were enrolled from 22 screened individuals between November 2022 and September 2023, including six patients in the safety lead-in phase and 11 in the expansion phase. All 17 patients discontinued study treatment and were included in the safety, efficacy, and full analysis sets. In the TPGC cohort, between February 2023 and October 2023, 30 patients were enrolled from 37 screened individuals, including 3 in the safety lead-in phase and 27 in the expansion phase. Eighteen patients discontinued treatment, including 13 because of disease progression and 5 owing to withdrawal. All 30 patients were included in the safety and full analysis sets, except for one patient who lacked postbaseline efficacy assessment owing to loss to follow-up and was therefore excluded from the efficacy analysis set (Fig. 1).

Figure 1.

Study flow diagram. The last follow-up dates were September 17, 2024, and December 20, 2024, for the TP and TPGC cohorts, respectively. ^aThe expansion phase was originally planned to enroll 18–24 patients; however, owing to limited efficacy observed in an early analysis, enrollment in this cohort was terminated on February 26, 2025. ^bSafety analyses included all patients who received at least one dose of the study agents and had available safety assessment records (safety analysis set). ^cAntitumor activity analyses included all enrolled patients who received at least one dose of the study agents and had at least one evaluable efficacy assessment (efficacy analysis set). ^dSurvival outcomes were analyzed in all enrolled patients according to the intention-to-treat principle (full analysis set).

The median [interquartile range (IQR)] age was 49 years (44–58 years) in the TP cohort and 51 years (33–70 years) in the TPGC cohort, with male-to-female ratios of 10:7 and 5:1, respectively (Table 1). In the TP cohort, 88.2% (15/17) of the patients had distant metastases, and 47.1% (8/17) and 17.6% (3/17) had received two and three prior lines of therapy for advanced disease, respectively. All patients were EBV DNA–positive at enrollment and had previously received PD-1 [82.4% (14/17)] or PD-L1 [17.7% (3/17)] blockade within two prior lines of treatment. The initial response to prior PD-(L)1 blockade was dominated by partial response (58.8%) and stable disease (23.5%). The median (IQR) duration of the initial response and the number of treatment cycles were 9 (7–15) months and 9 (6–12) cycles, respectively. The median (IQR) interval from the last cycle of PD-(L)1 blockade to trial enrollment was 8 (5–18) weeks. In the TPGC cohort, 56.7% (17 of 30) of the patients had distant metastases, and 76.7% (23 of 30) were EBV DNA–positive; among patients with metastasis, 88.2% (15/17) had polymetastatic disease. Most patients had more than one involved site recorded at enrollment, with 63.3% (19/30) and 26.7% (8/30) of the patients having two and three involved sites, respectively, and 56.7% (17/30) of the patients had PD-L1–positive tumors.

Table 1.

Baseline characteristics.

Characteristic	TP cohort (N = 17)	TPGC cohort (N = 30)
Age, years	​	​
Median (IQR)	49 (44–58)	51 (33–70)
Sex assigned at birth	​	​
Male	10 (58.8)	25 (83.3)
Female	7 (41.2)	5 (16.7)
ECOG performance status score	​	​
0	0 (0)	3 (10)
1	17 (100)	27 (90)
Pathology	​	​
Differentiated nonkeratinized	4 (23.5)	3 (10)
Undifferentiated nonkeratinizing	13 (76.4)	27 (90)
Cell-free EBV DNA at enrollment, copies/mL	​	​
Median (IQR)	9,195 (509–30,000)	1,190 (115–8,763)
Positive	17 (100)	23 (76.7)
Negative	0 (0)	7 (23.3)
Stage at enrollment	​	​
Distant metastases	14 (82.4)	17 (56.7)
Recurrence	2 (9.1)	13 (43.3)
Recurrence with distant metastasis	1 (5.9)	0 (0)
Distant metastasisa	​	​
Lung	7 (46.7)	8 (47.1)
Liver	7 (46.7)	10 (58.8)
Bone	4 (26.7)	14 (82.4)
Lymph nodes	14 (93.3)	9 (52.9)
Polymetastasis	8 (53.3)	15 (88.2)
Number of involved sites recorded at enrollment	​	​
1	—	2 (6.7)
2	—	19 (63.3)
3	—	8 (26.7)
4	—	1 (3.3)
Prior lines for advanced disease	​	​
1	6 (35.5)	—
2	8 (47.1)	—
3	3 (17.6)	—
Previous immunotherapy for advanced disease	​	​
PD-1 blockade	14 (82.4)	—
PD-L1 blockade	3 (17.7)	—
Initial response to previous immunotherapy	​	​
Complete response	1 (5.9)	—
Partial response	10 (58.8)	—
Stable disease	4 (23.5)	—
Progressive disease	2 (9.1)	—
Duration of the initial response	​	​
Median (IQR) of time interval, months	9 (7–15)	—
Median (IQR) of immunotherapy cycles	9 (6–12)	—
Interval from last immunotherapy to enrollment	​	​
Median (IQR), weeks	8 (5–18)	—
Treatment history during locoregionally advanced period	​	​
Chemotherapy	—	12 (40)
Radiotherapy	—	12 (40)
Targeted therapy	—	3 (10)
Immunotherapy	—	3 (10)
PD-L1 expressionb	​	​
Positive	—	17 (56.7)
Negative or unknown	—	13 (43.3)

Abbreviation: ECOG, Eastern Cooperative Oncology Group.

Data are shown as n (%) unless otherwise specified.

^aSpecific metastatic sites and percentages were analyzed only in patients with distant metastasis from the TP cohort (n = 15) and TPGC cohort (n = 17). Bone metastases are not considered target lesions.

^bPositive PD-L1 expression was defined as a tumor proportion score of at least 1%.

Efficacy

In the TP cohort, after a median follow-up of 14.6 months (range, 13–22 months), none of the 17 patients achieved a complete or partial response, although 7 patients showed reductions in target lesion size. Ten (58.8%) patients achieved stable disease lasting at least 4 weeks, resulting in a disease control rate of 58.8%, whereas seven (41.2%) patients experienced progressive disease (Fig. 2A). The median duration of response was 2.2 months. The median PFS was 1.6 months (95% CI, 0–3.2 months), and the 12-month PFS rate was 9.1% (Fig. 2B; Table 2).

Figure 2.

Tumor response and survival in the TP cohort. A, Waterfall plot depicting the best percentage change from baseline in the sum of diameters of target lesions for each patient. B, Kaplan–Meier estimates of investigator-assessed PFS. The asterisk in A indicates the appearance of new tumor lesions. PR, progressive disease; SD, stable disease.

Table 2.

Response evaluation and survival.

Variable	TP cohort (N = 17)	TPGC cohort (N = 30)
Median follow-up, monthsa	14.6	18.6
Median duration of response, months	2.2	9.2
Tumor assessment, n (%)b	​	​
Complete response	0	4 (13.8)
Partial response	0	21 (72.4)
Stable disease	10 (58.8)	4 (13.8)
Progressive disease	7 (41.2)	0
Objective response rate, n (%)	0	25 (86.2)
Disease control rate, n (%)	10 (58.8)	29 (100)
PFS	​	​
Events, n (%)	13 (76.5)	17 (56.7)
Median (95% CI), months	1.6 (0–3.2)	10.8 (9.6–16.4)
Estimate at 12 months, %	9.1	40.9
PFS for PD-L1–positive population	​	​
Events, n/total n (%)	—	10/17 (58.8)
Median (95% CI), months	—	13.6 (8.4–16.6)
Estimate at 12 months, %	—	52.7
PFS for PD-L1–negative/unknown population	​	​
Events, n/total n (%)	—	7/13 (53.8)
Median (95% CI), months	—	9.6 (8.5–10.8)
Estimate at 12 months, %	—	22.9

^aThe cutoff dates were September 17, 2024, for the TP cohort and December 20, 2024, for the TPGC cohort.

^bIn the TPGC cohort, one patient had no evaluable efficacy data because of loss to follow-up; therefore, 29 patients were included in the efficacy analysis set.

In the TPGC cohort, after a median follow-up of 18.6 months (range, 2.8–21 months), 25 of 29 evaluable patients achieved an objective response rate of 86.2%, including four complete responses (13.8%) and 21 partial responses (70%). The disease control rate was 100% (29 of 29), with four (13.8%) patients maintaining stable disease. All 29 patients with at least one postbaseline tumor assessment showed a reduction from baseline in the size of their target lesions (Fig. 3A and B; Table 2). At the data cutoff, 12 (40%) patients remained on treatment, and the median duration of response was 9.2 months. The median PFS was 10.8 months (95% CI, 9.6–16.4 months), with a 12-month PFS rate of 40.9% (Fig. 3C). Two deaths (6.7%) occurred, and the median OS was not reached, with a 12-month OS rate of 93.3% (Fig. 3D). The longitudinal changes in target lesion size for individual patients in the TP and TPGC cohorts are shown in Supplementary Figs. S1 and S2.

Figure 3.

Tumor response and survival in the TPGC cohort. A, Swimmer plot showing baseline tumor burden, treatment exposure, and response duration for individual patients. B, Waterfall plot depicting the best percentage change from baseline in the sum of diameters of target lesions for each patient. C, Kaplan–Meier estimates of investigator-assessed PFS. D, Kaplan–Meier estimates of investigator-assessed OS.

Safety

In the TP cohort, no DLTs were observed during the safety lead-in phase at either dose level of TQB2618. Of the 17 enrolled patients, 13 (76.5%) experienced trAEs of any grade. The most frequently reported trAEs were hypothyroidism (41.2%), increased aspartate aminotransferase concentrations (23.5%), anorexia (17.6%), and constipation (17.6%). Grade 3 to 4 trAEs occurred in two (11.8%) patients, comprising one case of leukopenia (5.9%) and one case of thrombocytopenia (5.9%). No treatment-related deaths were reported in this cohort.

In the TPGC cohort, all 30 enrolled patients (100%) experienced trAEs of any grade, with hematologic toxicities being the most common, including leukopenia (96.7%), anemia (93.3%), and neutropenia (90%). Grade 3 to 4 trAEs were observed in 25 (83.3%) patients, most frequently leukopenia (40%), neutropenia (36.7%), anemia (33.3%), and thrombocytopenia (33.3%). Two deaths, classified as grade 5 adverse events, occurred because of disease progression (Table 3).

Table 3.

trAEs.

Adverse event	TP cohort (N = 17)		TPGC cohort (N = 30)
	All grades	Grades 3–4	All gradesa	Grades 3–4
Any adverse eventsb	13 (76.5)	2 (11.8)	30 (100)	25 (83.3)
Asthenia	2 (11.8)	0	17 (56.7)	1 (3.3)
Anorexia	3 (17.6)	0	22 (73.3)	0
Weight loss	0	0	7 (23.3)	0
Anaphylaxis	0	0	1 (3.3)	0
Hypertriglyceridemia	0	0	2 (6.7)	0
Hyponatremia	1 (5.9)	0	6 (20)	0
Epistaxis	2 (11.8)	0	3 (10)	0
Pneumonitis	0	0	1 (3.3)	0
Cough	2 (11.8)	0	10 (33.3)	0
Dyspnea	1 (5.9)	0	1 (3.3)	0
Sore throat	1 (5.9)	0	3 (10)	0
Vomiting	1 (5.9)	0	17 (56.7)	0
Nausea	0	0	23 (76.7)	0
Dry mouth	0	0	2 (6.7)	0
Constipation	3 (17.6)	0	8 (26.7)	0
Leukopenia	2 (11.8)	1 (5.9)	29 (96.7)	12 (40)
Neutropenia	1 (5.9)	0	27 (90)	11 (36.7)
Anemia	2 (11.8)	1 (5.9)	28 (93.3)	10 (33.3)
Thrombocytopenia	1 (5.9)	0	17 (56.7)	10 (33.3)
Lymphocytopenia	0	0	6 (20)	0
ALT increased	1 (5.9)	0	7 (23.3)	0
AST increased	4 (23.5)	0	7 (23.3)	0
Creatinine increased	2 (11.8)	0	9 (30)	0
Proteinuria	1 (5.9)	0	6 (20)	0
Headache	2 (11.8)	0	7 (23.3)	0
Insomnia	0	0	7 (23.3)	0
Amylase elevation	0	0	4 (13.3)	0
Lipase elevation	1 (5.9)	0	1 (3.3)	1 (3.3)
Diarrhea	0	0	9 (30)	0
Abdominal pain	1 (5.9)	0	3 (10)	0
Myalgia	2 (11.8)	0	1 (3.3)	0
Arthralgia	0	0	1 (3.3)	0
Localized edema	1 (5.9)	0	1 (3.3)	0
Rash	0	0	3 (10)	0
Pruritus	0	0	4 (13.3)	0
Cardiac troponin I increased	0	0	1 (3.3)	0
Hypothyroidism	7 (41.2)	0	6 (20)	0
Hyperthyroidism	0	0	3 (10)	0
Hypoesthesia	0	0	7 (23.3)	0
Alopecia	0	0	6 (20)	0
Musculoskeletal pain in the chest	2 (11.8)	0	1 (3.3)	0
Blurred vision	2 (11.8)	0	1 (3.3)	0

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase.

Data are shown as n (%). trAEs are listed in descending order of frequency in the TP cohort.

^aTwo patients were categorized as grade 5 adverse events because of death from disease progression.

^btrAEs of any grade occurring in at least 3% of patients are shown. All grade 3–4 adverse events are reported irrespective of frequency.

Subgroup and post hoc analyses

The subgroup analyses according to PD-L1 expression were performed in all 30 patients enrolled in the TPGC cohort. The median PFS was 13.6 months (95% CI, 8.4–16.6) in the PD-L1–positive group (n = 17) and 9.6 months (95% CI, 8.5–10.8) in the PD-L1–negative or unknown group (n = 13; Table 1). Although the 12-month PFS rates were numerically higher in the PD-L1–positive group than in the PD-L1–negative or unknown group (52.7% vs. 22.9%), no statistically significant difference between groups was observed (P = 0.440; Supplementary Fig. S3).

The dynamic analysis of cell-free EBV DNA demonstrated that patients in both the TP and TPGC cohorts could be grouped into high and low levels at baseline and exhibited distinct changes after the first treatment cycle, with EBV DNA levels increasing, decreasing, or remaining undetectable (Supplementary Figs. S4 and S5). Based on the data from 47 patients across the two cohorts, those who achieved a reduction of at least one order of magnitude in EBV DNA after the first cycle or maintained undetectable levels were classified as the EBV DNA dynamics–based low-load group (n = 24), whereas those without a meaningful decline or with an increase in EBV DNA were classified as the high-load group (n = 23). The low-load group showed significantly better PFS, with a higher 12-month rate (47.1% vs. 15.1%) and a longer median PFS (11.1 vs. 5.5 months) than the high-load group (HR, 0.37; 95% CI, 0.18–0.79; P = 0.007; Supplementary Fig. S6).

Discussion

This study represents the first phase II clinical trial to evaluate the safety and efficacy of dual immunotherapy targeting TIM-3 and PD-1 in patients with R/M NPC. The combination of TQB2618 and penpulimab with gemcitabine–cisplatin chemotherapy demonstrated encouraging antitumor activity in the first-line setting, achieving an objective response rate of 86.2% and a median PFS of 10.8 months. Importantly, the addition of TIM-3 blockade did not result in unexpected safety signals beyond those typically associated with PD-1 blockade and chemotherapy. Despite exhibiting a manageable safety profile, this dual immunotherapy provided limited clinical benefit in patients with R/M NPC refractory to prior PD-(L)1 blockade.

PD-1 blockade combined with gemcitabine–cisplatin is the established first-line standard of care for R/M NPC, with reported objective response rates of 69.5% to 88.1% and a final median PFS of 9.6 to 21.4 months in the previous three landmark clinical trials (3–5). Strategies to further improve the efficacy of this regimen, therefore, remain a clinical priority. The rationale for dual immune checkpoint inhibition is supported by the biology of NPC, in which PD-1 and TIM-3 are frequently co-overexpressed on tumor-infiltrating CD8⁺ T cells (15). The TPGC cohort in this study yielded an objective response rate of 86.2% and a median PFS of 10.8 months, which shows no obvious numerical advantage compared with current standard-of-care regimens. Notably, as the TP and TPGC cohorts in this study were conducted in parallel in line with the protocol, the TPGC cohort adopted a relatively low dose level of 1,200 mg of TQB2618. This comparatively conservative design in treatment intensity, driven by safety considerations, may have underestimated the potential therapeutic efficacy of this regimen to a certain extent. Moreover, patients in the TPGC cohort presented with a substantially higher tumor burden than those enrolled in the three aforementioned trials, as reflected by a polymetastasis rate of 88.2%. Approximately 10% of the patients in the TPGC cohort had previously received immunotherapy in the locoregionally advanced setting, which was not permitted in those trials; this prior exposure may have partially attenuated the observed benefit of dual immunotherapy (Supplementary Table S3). Therefore, the median PFS of 10.8 months observed in the TPGC cohort, which predominantly consisted of patients with polymetastatic NPC, represents a clinically meaningful outcome worthy of attention, especially in the relatively high-risk population. A trend toward longer PFS was observed in patients with PD-L1–positive tumors (Supplementary Fig. S3), consistent with previous reports indicating that PD-L1 expression may be associated with greater benefit from immunotherapy in the metastatic setting of NPC (4, 5, 16). Although this difference did not reach statistical significance owing to the limited sample size, these findings provide a strong rationale for further evaluation of this regimen in larger, randomized controlled trials.

The patients with R/M NPC who progress after first-line PD-1–based chemoimmunotherapy usually have a poor prognosis and limited therapeutic options. In the TP cohort, which enrolled heavily pretreated, immunotherapy-resistant patients, TP achieved a manageable safety profile but limited efficacy, which is consistent with the observations in other refractory solid tumors. In patients with non–small cell lung cancer treated with this dual immunotherapy, an objective response rate of 2.7% and a disease control rate of 40.5% were reported (16). Similarly, limited activity has been observed in PD-(L)1 blockade–resistant or refractory colorectal and endometrial cancers, in which only 4.5% of the patients achieved a partial response (17). The limited efficacy of TP in refractory disease may be explained by the biology of T-cell exhaustion. Immune checkpoint inhibitors predominantly reinvigorate progenitor exhausted (PD-1⁺ TCF1⁺ TIM-3⁻) CD8⁺ T cells, whereas terminally exhausted (PD-1⁺ TCF1⁻ TIM-3⁺) T cells are substantially less responsive (18, 19). In addition, TIM-3 has a complex functional role, not only serving as a marker of terminal exhaustion but also participating in effector T-cell responses (20). Beyond TIM-3, other coinhibitory receptors, including lymphocyte activation gene–3 (LAG-3), are frequently coexpressed with PD-1 and represent rational targets for combinatorial immunotherapy (21). Dual inhibition of PD-1 and LAG-3 combined with chemotherapy has yielded an objective response rate of 83.3% and a disease control rate of 97.6% in treatment-naïve advanced NPC (22). In the immunotherapy-resistant or refractory setting, PD-1 and LAG-3 blockade have demonstrated encouraging activity, with an objective response rate of 11.8% and a disease control rate of 64.7% (23). Emerging evidence further suggests that triple checkpoint blockade targeting TIM-3, PD-(L)1, and LAG-3 may provide greater antitumor activity than single-agent or dual approaches (24). In parallel, novel therapeutic modalities such as becotatug vedotin (MRG003), an epidermal growth factor receptor–targeted antibody–drug conjugate, have shown substantial efficacy in NPC following failure of platinum-based chemotherapy and PD-(L)1 blockade (25). Given the manageable safety profile of TP observed in this study, future investigations could explore its combination with antibody–drug conjugates as a strategy to synergistically overcome immunotherapy resistance.

The safety profile observed in both cohorts was comparable with that reported in previous studies, in which grade 3 to 4 trAEs typically occurred in approximately 7.6% to 14.2% of the patients receiving immunotherapy alone (26–28), and in around 80.9% to 94% of those treated with gemcitabine–cisplatin chemotherapy in combination with PD-1 blockade (3–5). The addition of TIM-3 blockade did not expand the expected toxicity spectrum beyond that commonly associated with PD-1 blockade and chemotherapy. Thyroid and hepatic dysfunction predominated among patients receiving PD-1 blockade, whereas hematologic toxicities remained the most frequent adverse events in those treated with PD-1 blockade plus gemcitabine–cisplatin chemotherapy (4, 27, 28). The relatively favorable safety profile of TIM-3 blockade may be attributable to its more restricted expression on terminally differentiated interferon-γ–producing T cells, in contrast to the broader expression of PD-1 on activated T cells (29). Additionally, the absence of rash (0%) in the TP cohort is most likely caused by the small sample size (n = 17), which can introduce substantial variability in event rates due to random fluctuation.

This study has several limitations. First, the single-arm design without a control group and the relatively small sample size inevitably introduce uncertainty in the interpretation of treatment efficacy. Second, enrollment was restricted to Chinese patients with nonkeratinizing histology, which may limit the generalizability of the findings to other ethnic populations or histologic subtypes. Third, the absence of tumor tissue for comprehensive biomarker analyses of TIM-3 expression constrained our ability to identify patient subgroups most likely to benefit from this therapeutic strategy.

In summary, the combination of TQB2618 and penpulimab demonstrated a manageable safety profile but limited antitumor activity in patients with R/M NPC who had developed resistance to prior immunotherapy. In contrast, the combination of TIM-3 and PD-1 blockade with gemcitabine–cisplatin showed encouraging efficacy, supporting its potential therapeutic value in the treatment-naïve setting. It is necessary to conduct further randomized controlled trials that directly compare the dual TIM-3 and PD-1 blockade plus gemcitabine–cisplatin with PD-1 blockade plus gemcitabine–cisplatin to determine the clinical value of TIM-3 blockade.

Supplementary Material

Supplementary Data S1

This file contains the protocol and statistical analysis plan.

Supplementary Table S1

Table S1 shows the representativeness of study participants.

Supplementary Table S2

Table S2 details the information of recruitment sites.

Supplementary Table S3

Table S3 summarizes clinical trials that explore PD-1 blockade combined with GP chemotherapy in R/M NPC.

Supplementary Figure S1

Figure S1 shows percentage change from baseline in target lesion size over time for individual patients in the TP cohort.

Supplementary Figure S2

Figure S2 shows percentage change from baseline in target lesion size over time for individual patients in the TPGC cohort.

Supplementary Figure S3

Figure S3 depicts Kaplan–Meier curves of progression-free survival according to PD-L1 expression status in the TPGC cohort.

Supplementary Figure S4

Figure S4 shows cell-free EBV DNA dynamics of the TP cohort.

Supplementary Figure S5

Figure S5 shows cell-free EBV DNA dynamics of the TPGC cohort.

Supplementary Figure S6

Figure S6 depicts Kaplan–Meier curves of progression-free survival according to EBV DNA dynamics-based group in patients from the two cohorts.

Data Availability

The data generated in the study are not publicly available because of restrictions related to patient privacy; however, deidentified participant data are available immediately upon publication on reasonable request from the corresponding author (Q.-Q. Cai; E-mail: caiqq@sysucc.org.cn).

Authors’ Disclosures

H. Chen reports grants from Chia Tai Tian Qing Pharmaceutical Group Co. Ltd. outside the submitted work. No disclosures were reported by the other authors.

Authors’ Contributions

C. Xu: Conceptualization, data curation, software, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. S.-Y. Wang: Resources, data curation, validation, investigation, project administration. M. Nie: Software, formal analysis, funding acquisition, investigation, visualization, writing–review and editing. K.-Y. Yang: Resources, data curation, validation, investigation, project administration. X.-Q. Huang: Data curation, software, validation, investigation, project administration. S. Qu: Data curation, software, validation, investigation, project administration. H. Chen: Resources, validation. L.-F. Shen: Data curation, software, validation, investigation. J. Huang: Data curation, investigation. F. Zhang: Data curation, investigation. Y.-P. Peng: Data curation, validation. L.-L. Shi: Data curation, investigation. X.-H. Hong: Data curation, investigation. Z.-J. Zhang: Data curation, investigation. L.-L. Tang: Resources, investigation. L. Guo: Resources, investigation. P.-Y. Ouyang: Resources, investigation. J.-M. Gao: Resources, investigation. Y.-P. Mao: Resources, visualization. Y. Huang: Software, investigation. R. Guo: Resources, investigation. L.-Z. Liu: Supervision, visualization, methodology. L. Tian: Supervision, validation, visualization, methodology. H.-J. Li: Validation, methodology. J.-B. Li: Software, formal analysis, investigation, methodology. J. Ma: Conceptualization, resources, supervision, funding acquisition, validation, visualization, writing–original draft, project administration. Y. Xia: Conceptualization, resources, supervision, funding acquisition, methodology, writing–original draft, project administration. Q.-Q. Cai: Conceptualization, resources, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing.