Updated efficacy and predictive biomarkers of QL1706, a bifunctional PD-1/CTLA-4 dual blocker in advanced solid tumors—A phase 1/1b study

Summary

QL1706 has shown promising efficacy in solid tumors in a phase 1/1b study. Here, we report updated long-term survival outcomes and biomarker analyses. Among 468 patients treated with QL1706 (5 mg/kg), median progression-free survival (mPFS) and overall survival (mOS) are 1.5 and 14.2 months for non-small cell lung cancer (NSCLC), 1.9 and 20.2 months for nasopharyngeal carcinoma (NPC), and 4.2 and 18.6 months for cervical cancer (CC), respectively. Liver metastasis is correlated with poor progression-free survival (PFS) and overall survival (OS) in NSCLC and poor OS in CC, while elevated lactate dehydrogenase is linked to shorter PFS and OS in NPC. CDK4/11q13 diploid or the expression of GZMK^high & MYC^low distinguishes NPCs with the most favorable PFS. In NSCLC, PD-L1⁺/TIL⁺ or a low ARG1:CXCL13 ratio indicates better outcomes. QL1706 offers long-term survival benefits in solid tumors, with identified molecular markers aiding in selecting suitable candidates. This study has been registered on clinicaltrials.gov (NCT04296994 and NCT05171790).

Graphical abstract

Highlights

• •QL1706, a bifunctional PD-1/CTLA-4 dual blocker, shows long-term survival benefits
• •Clinical risk factors are identified for poor prognosis
• •CDK4/11q13 diploid or GZMK^high/MYC^low expression identifies better outcomes in NPC
• •PD-L1⁺/TIL⁺ or a low ARG1:CXCL13 ratio indicates better survival in NSCLC

Ma et al. report sustained survival benefits with QL1706, a bifunctional PD-1/CTLA-4 dual blocker, in patients with advanced solid tumors. They further identify clinical risk factors associated with unfavorable prognosis and predictive biomarkers to stratify patients with nasopharyngeal carcinoma or non-small-cell lung cancer who are likely to benefit from immunotherapy.

Introduction

In recent years, the combination of anti-programmed cell death protein 1 (PD-1) and anti-cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) antibodies has demonstrated compelling clinical efficacy across a variety of tumor types.^1,2 The introduction of bispecific antibodies that target both PD-1 and CTLA-4 receptors holds the potential to augment the immune response and mitigate cumulative toxicity. This is achieved by competitively targeting T cells that co-express PD-1 and CTLA-4 and by enhancing the affinity toward CTLA-4 receptors.^3,4 QL1706 is a bifunctional MabPair product with a mixture of anti-PD-1 and anti-CTLA-4 monoclonal antibodies in fixed combination, characterized by a truncated elimination half-life (t_1/2) for CTLA-4 (4.7–7.9 days).⁵ QL1706 differs from a cocktail of two separate antibodies targeting PD-1 and CTLA-4, such as the combination of nivolumab and ipilimumab. The ratio of 2:1 of this MabPair product is determined based on pharmacokinetics (PK) simulation of anti-PD-1 and anti-CTLA-4 in human using cynomolgus monkey PK data.⁵ Although not directly compared, our simulation (unpublished data) indicates QL1706’s anti-PD-1 and anti-CTLA-4 exposures at the 3-week interval are similar to nivolumab plus ipilimumab. Our previous reports indicated that, the profile of QL1706, marked by its reduced anti-CTLA-4 exposure concurrent with a consistent duration of anti-PD-1 exposure, showed enhanced tolerability with alleviated CTLA-4 antibody-mediated immune-related adverse events (irAEs) and promising short-term antitumor activity. On September 30, 2024, the National Medical Products Administration granted conditional approval for the market release of the injection iparomlimab and tuvonralimab (QL1706) in China, intended for the treatment of patients with recurrent or metastatic cervical cancer who have previously failed platinum-based chemotherapy. Moreover, QL1706 plus chemotherapy and bevacizumab have showed favorable antitumor activity in patients with EGFR-mutant non-small cell lung cancer (NSCLC) who failed in tyrosine kinase inhibitor (TKI) therapy in a phase 2 study.⁶ Consequently, the therapeutic potential of QL1706 in various solid tumors warranted further comprehensive investigation and validation through extended follow-up studies.

In our previous report of short-term antitumor activity, it is worth noting that a considerable portion of participants (213/468, 45.5%) showed progressive disease at the first efficacy evaluation, necessitating their early withdrawal from our trial.⁵ Similar outcomes have been observed in other trials involving bispecific dual immune checkpoint inhibitors (ICIs), where the median progression-free survival (mPFS) was only 1.5 and 1.9 months for KN046³ and AK104,⁷ respectively. The lack of reliable biomarkers for predicting responses to dual PD-1 and CTLA-4 blockade hampers the development of targeted therapeutic strategies for biomarker-selected patient populations. Therefore, there is a critical need for further biomarker research to identify the patients who are most likely to benefit from QL1706.

In patients with nasopharyngeal carcinoma (NPC), genetic variants in the granzyme, gasdermin, and interferon (IFN) pathways were associated with poorer progression-free survival (PFS) compared with the wild type.⁸ The POLARIS-02 study has revealed that patients with NPC harboring genomic amplification in the 11q13 region or possessing ETV6 genomic alterations responded poorly to toripalimab.⁹ However, to date, few studies have explored the value of immunotherapy-related biomarkers in patients with NPC. For NSCLC, PD-L1 expression¹⁰ is commonly used to predict the efficacy of immunotherapy. In addition, tumor mutational burden,¹¹ microsatellite instability,¹² mutations in polymerase epsilon and delta 1 (POLE/POLD1),¹³ Kirsten rat sarcoma viral oncogene homolog (KRAS),¹⁴ ataxia telangiectasia mutated (ATM)¹⁵ and gene expression profile of IFN-γ,¹⁶ effector T cells,¹⁷ PSME1/2, and PSMB9¹⁵ have been identified as clinically relevant to immunotherapy response in solid tumors, while HLA-LOH, B2M loss, STK11/KEAP1 loss of functions, and JAK1/2¹⁸ mutations as well as telomerase reverse transcriptase (TERT)¹⁹ amplification were associated with resistance to immunotherapy. Yet, most of these biomarkers are indicators of superiority rather than inferiority, and their predictive value varies, even among patients with the same malignancy undergoing different immunotherapeutic strategies. Thus, improving the selection of patients with NSCLC for immunotherapy, particularly identifying those resistant to such treatments, is essential.

In this report, we present updated efficacy data, encompassing PFS and overall survival (OS), supplemented by an additional 17.3 months of follow-up time beyond the primary analysis.²⁰ A comprehensive exploration of clinical risk characteristics and prognostic biomarkers predictive of QL1706’s efficacy, covering genomic and transcriptomic analyses and the tumor-immune microenvironment (TIME), were conducted.

Results

Clinical results

Patient characteristics

Between March 2020 and July 2021, 518 patients were enrolled in this study, with 99 in phase 1 and 419 in phase 1b (Figure 1). The baseline characteristics of these participants have been previously reported. The cohort included patients with NSCLC (146, 28.2%), NPC (134, 25.9%), cervical cancer (CC) (55, 10.6%), colorectal cancer (CRC) (27, 5.2%), small cell lung cancer (SCLC) (26, 5.0%), hepatocellular carcinoma (HCC) (25, 4.8%), biliary tract carcinoma (21, 4.1%), and other types (76, 14.7%).

Figure 1.

Flowchart of the study

Survival outcomes updates

As of the data cutoff on August 1^st, 2023, 419/518 (80.9%) patients had experienced disease progression, and 295/518 (56.9%) patients had died. The median follow-up time was 26.8 months. The objective response rate (ORR) and median duration of response (mDoR) were 19.2% (19/99, 95% confidence interval [CI] 12.0–28.3%) and 10.5 months (95% CI 5.9–18.7) for phase 1, 16.2% (68/419, 95% CI 12.8–20.1%) and 12.9 months (95% CI 11.0-not reached [NR]) for phase 1b, and 16.8% (87/518, 95% CI 13.7–20.3%) and 12.7 months (95% CI 10.8-NR) for all patients, respectively (Table S1). The mPFS and median overall survival (mOS) were 2.8 months (95% CI 1.4–4.7) and NR (95% CI 25.2-NR) for phase 1, 1.5 months (95% CI 1.4–2.2) and 12.7 months (95% CI 11.6–16.1) for phase 1b, and 1.5 months (95% CI 1.4–2.6) and 16.1 months (95% CI 12.7–18.8) in all patients, respectively.

A total of 468 patients in phases 1 and 1b received QL1706 at the recommended phase 2 dose (RP2D) of 5 mg/kg every 3 weeks, resulting in an ORR and mDoR of 16.9% (79/468, 95% CI 13.6–20.6%) and 12.4 months (95% CI 10.5-NR), respectively (Table S2). The mPFS and mOS were 1.5 months (95% CI 1.4–2.4) and 14.6 months (95% CI 12.2–17.6), respectively (Figures 2A and 2B). QL1706 exhibited promising antitumor activity, especially in NSCLC, NPC, and CC, with mDoRs of 11.1 months (95% CI 3.5-NR) in NSCLC, 11.0 months (95% CI 8.3-NR) in NPC, and NR (95% CI 5.6-NR) in CC. The mPFS and mOS were 1.5 months (95% CI 1.3–2.8) and 14.2 months (95% CI 10.9–19.9) in NSCLC (Figures 2C and 2D), 1.9 months (95% CI 1.3–2.9) and 20.2 months (95% CI 12.5–26.7) in NPC (Figures 2E and 2F), 4.2 months (95% CI 1.5–5.5) and 18.6 months (95% CI 9.5-NR) in CC, respectively (Figures 2G and 2H). Efficacy assessments for the other tumor types in patients receiving QL1706 at the RP2D are detailed in Table S2. The best overall responses of the target lesions from baseline, along with the duration of treatment for NSCLC, NPC, and CC treated with QL1706 at the RP2D, are shown in Figure S1.

Figure 2.

Kaplan-Meier curves of PFS and OS for patients treated with QL1706 at the RP2D

(A) Kaplan-Meier curves of PFS for all patients.

(B) Kaplan-Meier curves of OS for all patients.

(C) Kaplan-Meier curves of PFS for patients with NSCLC.

(D) Kaplan-Meier curves of OS for patients with NSCLC.

(E) Kaplan-Meier curves of PFS for patients with NPC.

(F) Kaplan-Meier curves of OS for patients with NPC.

(G) Kaplan-Meier curves of PFS for patients with CC.

(H) Kaplan-Meier curves of OS for patients with CC.

PFS, progression-free survival; OS, overall survival; NSCLC, non-small-cell lung cancer; NPC, nasopharyngeal carcinoma; CC, cervical cancer; RP2D, recommended phase 2 dose. See also Figure S1, Tables S1–S3.

Immunotherapy-naive patients receiving QL1706 at the RP2D experienced a mDoR of 12.9 months (95% CI 10.8-NR) across all patients, with specific figures of 11.8 months (95% CI 3.5-NR) in NSCLC, 11.0 months (95% CI 8.1-NR) in NPC, and NR in CC (Table S3). The mPFS and mOS were recorded at 2.5 months (95% CI 1.5–2.8) and 17.7 months (95% CI 14.9–20.3) across all patients, 1.5 months (95% CI 1.3–4.1) and 21.1 months (95% CI 12.7-NR) in NSCLC, 2.8 months (95% CI 1.4–8.5) and 26.7 months (95% CI 13.5-NR) in NPC, and 4.2 months (95% CI 1.7–6.9) and 18.6 months (95% CI 9.5-NR) in CC, respectively. For patients previously treated with immunotherapy and then treated with QL1706 at the RP2D, the mDoRs were 10.4 months (95% CI 4.2-NR) in all patients, 4.7 months (95% CI NR-NR) in NSCLC, 10.4 months (95% CI 5.9-NR) in NPC, and NR in CC. The mPFS and mOS were 1.4 months (95% CI 1.3–1.4) and 11.5 months (95% CI 9.3–12.7) in all patients, 1.4 months (95% CI 1.3–2.8) and 10.9 months (95% CI 5.5–14.2) in NSCLC, 1.3 months (95% CI 1.3–2.0) and 13.1 months (95% CI 10.4–25.2) in NPC, and 1.4 months (95% CI 1.4-NR) and 9.9 months (95% CI 7.0-NR) in CC, respectively.

Risk factors for clinical outcomes

We conducted univariate (Figure 3) and multivariate Cox regression analyses (Figure S2) to evaluate the clinical risk factors associated with unfavorable outcomes (PFS ≤ mPFS, no durable benefit [NDB] and OS ≤ mOS) in NSCLC, NPC, and CC cohorts receiving QL1706 at the RP2D. In NSCLC, liver metastasis was associated with shorter PFS (hazard ratio [HR], 2.09; 95% CI, 1.28–3.40; p = 0.003). According to the multivariate Cox regression analyses, patients with prior immunotherapy (HR, 1.68; 95% CI, 1.03–2.74; p = 0.038) and liver metastasis (HR, 2.48; 95% CI, 1.50–4.11; p < 0.001) had significantly poorer OS (Figure S2A). In NPC, a history of prior immunotherapy (PFS: HR, 2.28; 95% CI, 1.34–3.89; p = 0.002. NDB: HR, 2.15; 95% CI, 1.30–3.56; p = 0.003, Figure S2B) and lactate dehydrogenase (LDH) greater than the upper limit of normal (ULN) (PFS: HR, 2.23; 95% CI, 1.30–3.83; p = 0.004. NDB: HR, 1.89; 95% CI, 1.14–3.13; p = 0.013, Figure S2B) were associated with a higher risk of experiencing short PFS and NDB. Conversely, female patients exhibited a lower risk (PFS: HR, 0.36; 95% CI, 0.15–0.85; p = 0.020. NDB: HR, 0.44; 95% CI, 0.21–0.89; p = 0.024, Figure S2B). LDH levels greater than ULN also worsened OS in NPC (HR, 1.78; 95% CI, 1.09–2.92; p = 0.022) (Figure S2B). For patients with CC, a derived neutrophils/(leukocytes minus neutrophils) ratio (dNLR) > 3 (HR, 2.73; 95% CI, 1.27–5.85; p = 0.010) and liver metastasis (HR, 3.97; 95% CI, 1.53–10.26; p = 0.004) were significantly associated with shorter OS (Figure S2C).

Figure 3.

Forest plot of univariate Cox regression analysis

(A) Forest plot of univariate Cox regression analysis of mPFS, NDB vs. DCB, and mOS in NSCLC; (B) forest plot of univariate Cox regression analysis of mPFS, NDB vs. DCB, and mOS in NPC; (C) forest plot of univariate Cox regression analysis of mPFS, NDB vs. DCB, and mOS in CC. Significant differences are denoted by p < 0.05, p < 0.01 and p < 0.001. A p value less than 0.05 is considered to be statistically significant. mPFS, median progression-free survival; DCB, durable clinical benefit; NDB, no durable benefit; mOS, median overall survival; NSCLC, non-small-cell lung cancer; NPC, nasopharyngeal carcinoma; CC, cervical cancer; HR, hazard ratio; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; dNLR, derived neutrophils/(leukocytes minus neutrophils) ratio; LDH, lactate dehydrogenase; ULN, upper limit of normal.

See also Figure S2.

Biomarkers

To further differentiate between patients unlikely to benefit from QL1706, we examined the relationship between patient responses and specific biomarker characteristics. Thirty-three patients with NPC and 22 patients with NSCLC, possessing sufficient tumor tissue samples, were selected for sequencing. A comparison of baseline characteristics between patients with available samples and the entire NPC and NSCLC cohorts was displayed in Tables S4 and S5. The characteristics of patients with NPC were evenly distributed between the subgroups—those who achieved durable clinical benefit (DCB) and those with NDB—except for the number of therapy lines and ICI response (refer to the methodology section), as detailed in Table S6. In the NSCLC cohort, the distribution of patient characteristics was not significantly different between the DCB and NDB groups, with the exception of ICI response (Table S7).

Patients with NPC exhibiting CDK4/11q13 diploid showed sensitivity to QL1706 treatment

Genomic analysis revealed the most frequent somatic variants were KMT2D, LRP1B, TP53, and KMT2C (Figure 4A). Diploid in CDK4 and 11q13 (containing FGF3, FGF19, and CCND1) were identified (Figure 4A) and significantly prevalent among patients with a long-term response to QL1706 (Figure 4B, p = 0.025; Figures S3A–S3C, CDK4/11q13 diploid: patients with non-PD = 56.5% [13/23]; patients with DCB = 52.2% [12/23]). Such diploid in CDK4 or 11q13 were significantly linked to increased PFS (Figure 4C; 11q13 group, p = 0.05; Figure S3D; CDK4 group, p = 0.039; Figure S3E). Notably, CDK4/11q13 diploid marked a subgroup of NPC without rapid disease progression (mPFS 4.665 vs. 1.248 months, p = 0.0039, Figure 4D), more so than those with a CDK4/11q13 gain status. Moreover, NPC cases with CDK4/11q13 diploid exhibited an immune-activated microenvironment, characterized by enhanced infiltration of proinflammatory cells, activated CD4/CD8, and diminished myelocytomatosis viral oncogene homolog (MYC) and cell-cycle pathways (Figures 4E and 4F). These findings indicate that CDK4/11q13 alterations are crucial for the sensitivity and response mechanisms to QL1706 in patients with NPC.

Figure 4.

Genetic alterations in patients with NPC in response to QL1706 treatment

(A) Overview of molecular and clinical characteristics of the 32 patients with NPC.

(B) Waterfall and boxplots showing tumor response and size variation in NPC with gains in different CNVs. Each point represents an individual subject (n = 1). The box section of the boxplot represents the quartile range of the pathological regression, namely the 25% and 75% quartiles.

(C) Forest plot displaying HRs for PFS across various CNV alterations.

(D) PFS comparison between patients with CDK4/11q13 diploid and those with CDK4/11q13 gain.

(E) Heatmap of 28 immune cell scores in NPC with CDK4/11q13 CNV diploid vs. gain.

(F) GSEA highlighting genes upregulated in the NPC cohort, with ES, NES, and FDR-adjusted q value indicated. The black horizontal bar indicates the genes present in the gene set. The highest enrichment score indicates the enrichment.

Significant differences are denoted by p < 0.05, p < 0.01 and p < 0.001. A p value less than 0.05 is considered to be statistically significant.

NPC, nasopharyngeal carcinoma; CNV, copy number variations; HR, hazard ratio; PFS, progression-free survival; GSEA, gene set enrichment analysis; ES, enrichment score; NES, normalized enrichment score; FDR, false discovery rate-adjusted q value. See also Figure S3 and Table S4.

MYC^high/GZMK^low distinguished a favorable response to QL1706 in patients with NPC

Previous studies found that FGF19 induces elevated CCND1,²¹ and CDK4 and CCND1 were found to be transactivated by MYC.²² Here, in support of the results derived from transcriptomic analysis and omics studies, differentially expressed genes (DEGs) were enriched in the cell cycle-related pathways, particularly “HALLMARK_MYC_TARGETS_V1 and V2,” within the NDB group (Figures 5A and 5B). Gene set enrichment analysis underscored a significant concentration of MYC target genes (Figure 5C). Low expression of MYC (p < 0.05) was correlated with DCB patients (Figure 5D). Additionally, high MYC levels were inversely correlated with the presence of various immune cells, including macrophages (p < 0.01), mast cells (p < 0. 05), myeloid-derived suppressor cells (p < 0. 05), monocytes (p < 0. 05), neutrophils (p < 0. 05), and plasmacytoid dendritic cell (p < 0. 05) infiltration in NPC (Figure S4A). The DCB group displayed an immunologically activated status (Figure 5E), notably with Granzyme K (GZMK) expression significantly linked to immune activation markers such as activated B cells (p < 0.001), activated CD4 T cells (p < 0.01), activated CD8 T cells (p < 0.001), and IFN gama_6 (p < 0.05) (Figures S4B and S4C). Of note, GZMK (p < 0.01) was highly expressed in DCB patients (Figure 5D). Patients with NPC exhibiting co-expression levels of GZMK^high & MYC^low have an immune activating property (Figures S4D and S4E). The same results were seen in NPC cohorts from RATIONALE-309 dataset (Figures S4F–S4I).

Figure 5.

Association of MYC and GZMK expression with QL1706 efficacy in NPC

(A) Differential gene expression analysis shows genes highly expressed in DCB versus NDB groups, with significance indicated by dot size (fold change > 1.5, p < 0.05).

(B) Pathway enrichment analysis differentiates between DCB and NDB samples.

(C) GSEA demonstrates activated MYC_TARGETS in the NDB group.

(D) Heatmap illustrating the expression levels of genes associated with MYC, adaptive immune system and cell cycle across 32 NPC samples stratified by clinical outcome. The right panel displays the Cox regression results for these genes, including HR and 95% CI.

(E) Heatmap displaying mRNA levels of Cell_cycle, Immune, or MYC-related genes in DCB and NDB groups.

(F) Pathological regression and PFS status in patients under immune checkpoint therapy, categorized by MYC and/or GZAMK expression levels.

(G) Kaplan-Meier survival plots for PFS in QL1706-treated NPC cohort, segmented by GZMK^high and GZMK^low.

(H) Kaplan-Meier survival plots for PFS in QL1706-treated NPC cohort, segmented by MYC^high and MYC^low.

(I) Kaplan-Meier survival plots for PFS in QL1706-treated NPC cohort, segmented by MYC^low&GZMK^high, MYC^high&GZMK^low, and other groups.

Significant differences are denoted by p < 0.05, p < 0.01 and p < 0.001. A p value less than 0.05 is considered to be statistically significant.

NPC, nasopharyngeal carcinoma; DCB, durable clinical benefit; NDB, no durable benefit; GSEA, gene set enrichment analysis; PFS, progression-free survival. See also Figures S4–S9 and Table S6.

Then a significant portion of tumor regression and 1-year non-progression was observed predominantly in the GZMK^high or MYC^low group (Figure 5F). The GZMK^high and/or MYC^low profile was predominantly observed in patients with non-PD/DCB (Figures S5A–S5C). The area under the receiver operating characteristic curve (AUC) for GZMK and MYC genes was 0.735 and 0.808, respectively, for predicting clinical benefit (DCB/NDB) (Figure S5D). GZMK^high or MYC^low patients had significantly longer PFS than those with GZMK^low or MYC^high patients (GZMK^high: p = 0.013, Figure 5G; MYC^low: p < 0.0001; Figure 5H). Moreover, the combination of GZMK and MYC provided a more accurate discrimination of survival benefits for NPC (mPFS 12.945 vs. 1.216 months, p < 0.0001, Figure 5I). A consistent molecular pattern result could be obtained in the RATIONALE-309-immno dataset (Figures S6A–S6G). In particular, a single GZMK or MYC at mRNA level or a dual mRNA marker GZMK & MYC were superior to PD-L1 immunohistochemistry (IHC) result or Epstein-Barr Virus (EBV) expression level in predicting response to immunotherapy (Figure S6A).

However, in the RATIONALE-309-GP cohort of NPC-treated with GP, neither GZMK nor MYC alone, nor in combination, was able to distinguish efficacy and prognosis (Figures S7A–S7G). Similarly, another cohort of NPC (GSE102349) could not distinguish prognosis with this molecular pattern of GZMK and MYC (Figures S8A–S8E), indicating that the GZMK^high & MYC^low profile is a specific signal for improved efficacy of immunotherapy.

In an independent NPC immunotherapy cohort receiving nivolumab plus ipilimumab (GSE224450), the molecular pattern defined by GZMK and MYC expression could distinguish treatment response. We also compared with known prognostic and predictive biomarkers, including CD274, IFN-γ, and effector T cell signatures (Figures S9A–S9C). These results imply that GZMK and/or MYC expression are crucial in determining the sensitivity of immunotherapy, including the regimen QL1706.

Tumor-infiltrating lymphocytes and PD-L1 subtyping predicted therapeutic efficacy of QL1706 in NSCLC

The mutation profiles of NSCLC (Figure 6A) were consistent with a previous report.²³ Given the absence of genomic biomarkers available to identify patients who do not respond to QL1706, we attempted to analyze the associations between the TIME and the clinical benefit of QL1706 in patients with NSCLC. We classified TIME based on PD-L1 expression and CD8⁺ tumor-infiltrating lymphocyte (TIL) levels into four categories: PD-L1^– and TIL^–; PD-L1⁺ and TIL⁺, PD-L1^– and TIL⁺, and PD-L1⁺ and TIL^–. Among the 11 patients with NSCLC screened, fewer response events were observed among those with PD-L1^–&TIL^– profiles (DCB, PR, and SD; Figure S10A). Patients with PD-L1⁺ or (/) TIL⁺ had significantly longer PFS and OS following QL1706 treatment compared to those with PD-L1^–&TIL^– profiles (mPFS, 10.9 vs. 2.6 months, p = 0.021, Figure 6B; mOS, NA vs. 11 months, p = 0.01; Figure S10B). This observation was corroborated in the OAK immunotherapy cohort dataset (OAK_NSCLC-ICI) where PD-L1^–&TIL^– versus PD-L1⁺/TIL⁺ showed a significant difference (p < 0.0001, Figure 6C). However, no significant prognostic impact was noted in the OAK chemotherapy cohort (PD-L1^–&TIL^– vs. PD-L1⁺ or TIL⁺, p = 0.41; Figure 6D) or the TCGA_NSCLC cohort (PD-L1^–&TIL^– vs. PD-L1⁺ or TIL⁺; PFS, p = 0.67; OS, p = 0.069, Figures S10C and S10D).

Figure 6.

Genetic alterations and tumor immune microenvironment in patients with NSCLC receiving QL1706

(A) Summary of key molecular and clinical features of the 22 patients with NSCLC.

(B) Kaplan-Meier survival analyses of PFS for the QL1706_ NSCLC cohort.

(C) Kaplan-Meier survival analyses of PFS for the OAK_NSCLC-ICI cohort.

(D) Kaplan-Meier survival analyses of PFS for the OAK_NSCLC-chemo cohort.

(E) Correlation analysis of PD-L1 & TILs with macrophages and neutrophils across QL1706_NSCLC, OAK_NSCLC-ICI, and TCGA_NSCLC cohorts. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(F) Sankey diagram illustrating the relationships among Neutrophil-M1-CIBERSORT, PD-L1-TIL, and Neutrophil-M1-quantiseq in NSCLC.

Significant differences are denoted by p < 0.05, p < 0.01 and p < 0.001. A p value less than 0.05 is considered to be statistically significant.

NSCLC, non-small-cell lung cancer; PFS, progression-free survival; ICI, immune checkpoint inhibitor; PD-L1, programmed cell death-ligand 1; TIL, tumor-infiltrating lymphocyte. The “Other” group comprised patients characterized by both high neutrophil and M1 expression (Neu^high & M1^high) as well as those with both low neutrophil and M1 expression (Neu^low & M1^low). See also Figure S10 and Tables S5 and S7.

Additionally, PD-L1^–&TIL^– NSCLC was characterized by an abundance of neutrophils (p = 0.019) and a reduction in M1 macrophages (p = 0.00055) (Figure S10E). This pattern was consistent across the OAK and TCGA_NSCLC datasets (Figure 6E). In QL1706 cohort, both CIBERSORT and quantiseq analysis methods in Sankey diagram showed the similar results (Figure 6F).

A low ARG1:CXCL13 ratio delineated a subpopulation of patients with NSCLC who might benefit from QL1706 treatment

To enable internal normalization of gene expression in M1 macrophages and neutrophils, we used the pairwise log ratio method, followed by the construction of a pairwise signature using an elastic network classifier²⁴ that was trained on overall patient survival. The expression ratio of ARG1, representing neutrophil cells, to CXCL13, representing M1 macrophages, was used as a biomarker to distinguish the survival benefit of QL1706 for NSCLC (Figure 7A).

Figure 7.

The association of the ARG1:CXCL13 ratio with prognosis in patients with NSCLC receiving QL1706

(A) Analysis of ARG1 and CXCL13 correlations with M1 macrophages and neutrophils in QL1706 and OAK_NSCLC-ICI cohorts. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(B) A heatmap showing immune signature associations with high and low ARG1:CXCL13 ratios.

(C) Forest plots detailing the ARG1:CXCL13 ratio’s effect on PFS post-QL1706 therapy.

(D) Kaplan-Meier plots for PFS in the QL1706 NSCLC group.

(E) Forest plots for the ARG1:CXCL13 ratio’s impact on OS after QL1706 treatment.

(F) Kaplan-Meier plots for OS in the QL1706 NSCLC cohort.

Significant differences are denoted by p < 0.05, p < 0.01 and p < 0.001. A p value less than 0.05 is considered to be statistically significant.

NSCLC, non-small-cell lung cancer; ICI, immune checkpoint inhibitor; PFS, progression-free survival; OS, overall survival. See also Figure S11.

Analysis of the immune microenvironment showed that a high ARG1:CXCL13 ratio was positively correlated with pro-tumor immunity, angiogenesis, fibrosis, and malignant cell properties, whereas a low ratio indicated antitumor immunity (Figure 7B). Further analysis revealed that both PD-L1^–&TIL^– and PD-L1⁺/TIL⁺ patients could be further categorized based on the ARG1:CXCL13 ratio (Figure S11A), with a low ratio associated with longer PFS (p = 0.066, Figure 7C; mPFS 10.9 vs. 2.6 months, p = 0.034; Figure 7D) and OS (p = 0.009, Figure 7E; mOS NA vs. 11 months, p = 0.0051; Figure 7F). The correlation between CD274 (PD-L1) expression and PFS was marginally significant (Figure 7C). Additionally, a higher incidence of DCB and PR + SD was observed in patients with a low ARG1:CXCL13 ratio (Figure S11A).

In the OAK_NSCLC-ICI cohort, the 1-year progression rate differed significantly among the two subgroups divided by the ARG1:CXCL13 ratio (p = 0.01, Figure S11B). A similar pattern of survival benefit was observed in the OAK_NSCLC-ICI (Figure S11B) and POPLAR_NSCLC-ICI (Figure S12A) cohorts, but not in the OAK_NSCLC-chemo (Figure S11C) or POPLAR_NSCLC-chemo cohorts (Figure S12B).

As successfully validated in the OAK and POPLAR cohorts, a low ARG1:CXCL13 ratio could serve as an efficient biomarker for identifying favorable outcomes to immunotherapy (Figures S11D–S11F; Figures S12C and S12D), but not chemotherapy (Figures S11G and S11H; Figure S12E). Moreover, in the POPLAR study, immunotherapy provided significantly better long-term survival benefits compared to chemotherapy in patients with NSCLC exhibiting a low ARG1:CXCL13 ratio (Figures S12F and S12G). However, in the TCGA_NSCLC cohort, the therapeutic efficacy was not correlated with the ARG1:CXCL13 ratio (Figure S12H), underscoring the significant role of ARG1:CXCL13 ratio alterations in distinguishing DCB patients receiving immunotherapy.

Discussion

This present study reports to date the largest cohort of long-term survival outcomes, results of subgroup analyses, and biomarkers of a bifunctional MabPair product (QL1706) with dual blockade of PD-1 and CTLA-4 in advanced solid tumors. Overall, our results demonstrate that QL1706 exhibits enduring survival advantages across various types of advanced solid tumors, notably in heavily treated NPC, CC, and NSCLC. Clinical risk features were analyzed in a large cohort of 468 patients at the RP2D to identify the population that may benefit from QL1706 treatment. Moreover, several biomarkers were associated with the therapeutic outcomes of QL1706, but not with chemotherapy, in NPC and NSCLC. These associations have been confirmed in validation cohorts from independent public datasets. Here, we found that NPC with CDK4/11q13 diploid and GZMK^high and/or MYC^low had a favorable mPFS. These biomarkers could favor individualized treatment selection to achieve long-term survival benefits conferred by QL1706. Furthermore, the presence of both PD-L1⁺ and TIL⁺ profiles and a low ARG1:CXCL13 ratio were strongly linked to favorable survival outcomes in NSCLC. These biomarkers could favor individualized treatment selection to achieve long-term survival benefits conferred by QL1706. This study presents the comprehensive landscape of genomes and transcriptomes in advanced NSCLC and NPC, revealing predictive signatures putatively aiding patient stratification for immunotherapy.

QL1706 exhibited potency in improving survival benefit of recurrent/metastatic (R/M) NPC, particularly for heavily pretreated immunotherapy-naive patients. Notably, the mOS within the study was 20.2 months for all patients with NPC and elevated 26.7 months for immunotherapy-naive patients. For patients with NPC subjected to second line chemotherapy, the mOS estimated was 11.5–12.5 months²⁵ Besides, the mOS of other anti-PD-1 drugs were 17.4 months for toripalimab and camrelizumab, 16.5 months for pembrolizumab, and 17.1 months for nivolumab in the second-line or later setting.²⁶ Documented studies of PD-(L)1/CTLA-4 dual-blockade combinations administered to immunotherapy-naive patients with R/M NPC showed that the mOS were 24.7 months in bispecific antibody KN046 treatment³ and 19.5 months in the nivolumab and ipilimumab combination.²⁷ Caution is warranted when interpreting cross-trial comparisons due to differences in study designs, patient populations, and treatment settings. The promising findings noted in the current research justify the need for additional clinical investigations in patients with R/M NPC. A phase 3 trial of QL1706 combined with chemotherapy (NCT05576272) in NPC is currently in progress.

However, the benefit of QL1706 was not observed across the entire population, a phenomenon similarly noted with other dual blockades targeting PD-1/PD-L1 and CTLA-4.^3,27 We endeavored to pinpoint individuals with high-risk clinical factors or biomarkers demonstrating resistance to QL1706, in order to accurately discern the candidate population most likely to benefit from QL1706 therapy. The specific biomarkers for predicting QL1706 response in NPC has never been reported. Based on genomic and transcriptomic analysis, results from the NPC with CDK4/11q13 diploid and GZMK^high and/or MYC^low suggested the clinical benefit of QL1706, with a favorable response rate and a better mPFS. In parallel, copy-number diploid of CDK4 or 11q13 was reported to correlate with sensitivity to mono-immunotherapy and chemo-immunotherapy in multiple types of solid tumors, such as NPC,⁹ esophageal squamous cell carcinoma,²⁰ melanoma,²⁸ and HCC.⁷ Given that CCND1 is located in 11q13⁹ and CDK4 mediates cell cycle progression,²⁹ NPCs with CDK4 or 11q13 diploid may exhibit a weaker proliferation capacity, hindering survival under immune cell attack. This aligns with the observation that NPC with CDK4 or 11q13 diploid had more infiltrated proinflammatory cells.

CDK4 and CCND have been demonstrated to be regulated by MYC.²² FGF19 and CCND1 was demonstrated to be involved in cell cycle.²¹ Our study found that MYC-targeted pathways and cell cycle-related signaling were enriched in NPCs with CDK4/11q13 gain. Clearly, this question requires much investigation. Going further, MYC expression indicated rapid disease progression in response to QL1706 in this study. Similarly, High MYC expression in breast cancer³⁰ and SCLC³¹ were found to be strongly associated with poorer prognosis and immunotherapy responses. Previous reports have shown that high MYC expression leads to accelerated cell cycle progression³² and enhances tumor immune escape via the putative inhibition of IFN signaling,³³ thereby hampering the efficacy of anti-PD-1 as a monotherapy.³⁰ This study also supports the notion that inhibiting MYC or CDK4/6 may enhance antitumor immunity,^34,35 potentially reversing the efficacy of ICIs.³³

Conversely, high expression of GZMK was associated with better efficacy and prognosis in patients with NPC receiving QL1706. The possible reason for this is that the immune response to NPC, a malignant tumor associated with the latent EBV infection, relies largely on innate immunity rather than adaptive immunity.⁸ Whereas GZMK is a component of innate immunity,³⁶ and recent studies have found that high levels of GZMK typically indicate a robust intratumoral T cell response and have been linked to improved OS and better response to immunotherapy in intrahepatic cholangiocarcinoma,³⁷ HCC,³⁸ and NSCLC.³⁹ GZMK^high expression defines a core subset of tissue-resident, inflamed CD8⁺ T cells⁴⁰ across diseases and human tissues. Furthermore, GZMK⁺ CD8⁺ T cells, as part of the effector memory CD8⁺ T cell subset, can be activated into an effector phenotype during the treatment of ICIs.⁴¹ This activation is also associated with a long-term response to ICIs.^42,43,44,45 Combined biomarker analyses revealed that GZMK^low and MYC^high expression reflect the imbalance of proliferated cancer cells and less infiltrated effector CD8⁺ T cells, which undoubtedly led to an unfavorable response to immunotherapy, including QL1706. However, the detailed mechanism of how these two biomarkers regulate each other reciprocally is still unknown. This implies that GZMK/MYC can be modulated through controlling immune responses or altering the tumor immunosuppressive microenvironment. These biomarkers, which have been confirmed in validation cohorts from independent datasets, were associated with the therapeutic outcomes of immunotherapy, but not with chemotherapy, in NPC. Specifically, GZMK/MYC biomarkers outperformed conventional biomarkers such as PD-L1 and EBV levels in NPC. Therefore, these biomarkers could favor individualized treatment selection to achieve the long-term survival benefits conferred by QL1706.

In our study, immunotherapy-naive patients with NSCLC receiving QL1706 monotherapy had a mOS of 21.1 months, which is a satisfactory outcome considering that most of these were heavily pretreated patients. According to previous reports, the mOS for immunotherapy-naive patients treated with other dual blockades targeting PD-1/PD-L1 and CTLA-4 ranged from 10 to 19.6 months^7,46 A pooled analysis of long-term survival in patients with advanced NSCLC treated with first-line nivolumab in combination with ipilimumab revealed a mOS of 18.6 months⁴⁷ While caution must be exercised when comparing results across different trials, our findings suggest the promising potential of QL1706 in enhancing the long-term survival of NSCLC. A phase 3 study on the efficacy and safety of QL1706 with chemotherapy as first-line therapy for PD-L1 negative advanced or metastatic NSCLC (NCT05690945) is being investigated.

Due to the complexity of the immune system and the dynamic, heterogeneous nature of tumors, single biomarkers frequently fall short in predictive accuracy. This highlights the need for continued research into additional biomarkers and clinical prognostic factors. By exploring molecular biomarkers, the combination of PD-L1 expression and TILs can predict the efficacy of QL1706 in NSCLC, thus validating a previous finding proposed by our group.⁴¹ In addition, in-depth exploration suggested a decreased neutrophil, but an increase in M1 macrophage infiltration in PD-L1⁺/TIL⁺ NSCLC. The ARG1:CXCL13 ratio, mimicking the infiltration pattern of neutrophils and M1 macrophages, can further identify the patients who are most likely to benefit from QL1706 treatment. Approximately 40% of tumor-associated neutrophils actively transcribe ARG1 mRNA, with extracellular ARG1 localized in neutrophil extracellular traps,⁴⁸ which protect tumor cells by forming a physical barrier at the tumor-stroma interface, thereby preventing CD8⁺ T cell infiltration.⁴⁹ Furthermore, high ARG1 also defines an immunosuppressive subset of tumor-associated macrophages with M2-polarization.⁵⁰ Targeting ARG1 might reverse and enhance the efficacy of immunotherapy.⁵¹ CXCL13 was significantly positively associated with M1 macrophages⁵² and defines neoantigen reactive T lymphocytes⁵³ interacting with antigen-presenting cells in the tumor microenvironment.⁵⁴ CXCL13 can also recruit B cells mediating the maturation of tertiary lymphoid structures.⁵⁵ As suggested in Liu et al.’s study, CXCL13 was able to indicate a response to ICIs across solid tumors of different organs and locations.⁵⁶ Validated by external cohorts, the PD-L1⁺&TIL⁺ microenvironment and its derived ARG1:CXCL13 ratio can potentially define subsets of patients with NSCLC responsive to immunotherapy. However, this needs to be validated in further clinical trials.

Efforts are underway to improve the prognosis for patients with R/M CC. Combination therapy with balstilimab and zalifrelimab achieved an mPFS and mOS of 2.7 and 12.7 months, respectively.⁴² Cadonilimab, a PD-1/CTLA-4 blocker approved for treating patients with R/M CC, demonstrated a mPFS of 3.71 months and a mOS that was not reached when evaluated at 6 and 18 months²⁸ Notably, QL1706 showed encouraging activity in the treatment of CC, especially in terms of long-term survival benefit, with a mPFS and mOS of 4.2 and 18.6 months, respectively. Phase 2 study DUBHE-C-206 further validated the efficacy and safety of QL1706 in patients with recurrent or metastatic CC, achieving an ORR of 33.8%, a mPFS of 5.4 months, and a 12-month OS rate of 65.4%, leading to the approval of QL1706 for the indication of second-line and subsequent treatments for recurrent/metastatic CC.⁵⁷ Clinical features were assessed to predict the survival of CC treated with QL1706. An elevated dNLR, indicative of lower survival rates in patients with melanoma and NSCLC receiving immunotherapy,^58,59 served as a straightforward and accessible marker to identify patients with CC unlikely to derive benefit from QL1706 treatment.

In conclusion, this large-sample phase 1 trial revealed that QL1706, an inaugural bifunctional MabPair product targeting PD-1 and CTLA-4, demonstrated significant long-term survival benefits in advanced solid tumors. CDK4/11q13 diploid and GZMK^high and/or MYC^low expression, reflecting a state of immune activation, could reliably distinguish patients with NPC who are responsive to QL1706. Similarly, the combination of PD-L1⁺/TIL⁺ profiles and a low ARG1:CXCL13 ratio could identify patients with NSCLC likely to respond to QL1706. Validated by external data, these biomarkers serve as important predictors for identifying patients who are sensitive to immunotherapy, but not chemotherapy. Ongoing phase 3 studies are further investigating QL1706-based regimens across various solid tumors (NCT05576272, NCT05446883, NCT05690945, and NCT05487391).

Limitations of the study

Several limitations of this study are noteworthy. First, due to the limited availability of patient specimens, our research on biomarkers focused mainly on NPC and NSCLC. This may have introduced selection bias, as only a small fraction of the entire cohort had accessible tissue specimens for correlative analyses. Secondly, further basic research is needed to fully understand how the identified biomarkers predict the efficacy of QL1706. Additionally, the clinical applicability of these biomarkers as therapeutic or prognostic tools requires further validation through larger prospective studies and clinical trials. Particularly, translating transcriptional TIL signature to IHC based biomarker(s) was warranted in future studies with larger population.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Hongyun Zhao (zhaohy@sysucc.org.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Qilu Pharmaceutical Co., Ltd., will grant access to individual de-identified participant data, protocols, and statistical analysis plans, subject to particular criteria, conditions, and exceptions. Requests for data may be sent to the lead contact.
• •This paper does not report any original code.
• •Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

Acknowledgments

This trial was conducted in 41 Grade A class 3 hospitals in China, including Sun Yat-sen University Cancer Center, Affiliated Hospital of Hebei University, Jilin Cancer Hospital, the Cancer Hospital of the University of Chinese Academy of Sciences (Zhejiang Cancer Hospital), the first Affiliated Hospital of Gannan Medical University, the Second Affiliated Hospital of Anhui Medical University, Guangxi Medical University Cancer Hospital, Fujian Cancer Hospital, Affiliated Cancer Hospital and Institute of Guangzhou Medical University, Shandong Cancer Hospital and Institute, Affiliated Hospital of Chengde Medical University, Second Xiangya Hospital, Affiliated Hospital of Inner Mongolia Medical University, Tianjin Medical University Cancer Institute and Hospital, The People’s Hospital of Guangxi Zhuang Autonomous Region, Hainan General Hospital, Yuebei People’s Hospital, the People’s Hospital of Guangxi Zhuang Autonomous Region, the First Affiliated Hospital of Nanchang University, the First People’s Hospital of Foshan, Liaocheng People’s Hospital, the Comprehensive Cancer Centre of Drum Tower Hospital, the Affiliated Cancer Hospital of Guizhou Medical University, the First Affiliated Hospital of Kunming Medical University, the Affiliated Hospital of Qingdao University, the First Affiliated Hospital of Fujian Medical University, Fudan University Shanghai Cancer Center, Liaoning Cancer Hospital & Institute, Sichuan Cancer Hospital & Institute, the First Affiliated Hospital of Soochow University, Chongqing University Cancer Hospital, Anhui Provincial Hospital, the First Affiliated Hospital of Bengbu Medical College, Henan Cancer Hospital, Nanfang Hospital, the First Affiliated Hospital of Xiamen University, the First Affiliated Hospital of Xi’an Jiao Tong University, Xinqiao Hospital Army Medical University, Union Hospital Tongji Medical College Huazhong University of Science and Technology, Suzhou Municipal Hospital, and the First Affiliated Hospital of Xinjiang Medical University. We thank all investigators for their efforts to this trial. Also, we thank all patients and their caregivers for their invaluable participation in this study. We thank Qilu Pharmaceutical Co., Ltd., for sponsoring this study.

This work was supported by Noncommunicable Chronic Diseases-National Science and Technology Major Project (2024ZD0519700 for L.Z.), the National Nature Science Foundation of China (82473346 for H.Z., 81872449 for L.Z., and 82002409 for Y.M.), Guangdong Basic and Applied Basic Research Foundation (2020A1515010020 for Y.M.), the Chinese National Natural Science Foundation project (82303807 for F.L.), Natural Science Foundation of Guangdong Province of China (2024A1515013011 for F.L.), Fundamental Research Funds for the Central Universities, Sun Yat-sen University (24qnpy283 for F.L.), and Young Talents Program of Sun Yat-sen University Cancer Center (no. YTP-SYSUCC-0104 for F.L.).

Author contributions

H.Z., L.Z., and Y.H.: conceptualization, formal analysis, supervision, validation, investigation, visualization, methodology, writing—original draft, project administration, and writing—review and editing. Y.M., S.L., Q.C., J.X., and Y.Y.: conceptualization, formal analysis, validation, investigation, visualization, methodology, writing—original draft, and writing—review and editing. A.Z., Y.C., Y.Z., X.W., and Z.C.: conceptualization, validation, investigation, methodology, and writing—review and editing. S.Q., J.H., C.C., C.J., D.Z., Qingshan Li, X.L., W.S., Y.B., and F.L.: investigation, methodology, writing—review and editing. W.L., Qian Li, C.Z., and Z.H.: conceptualization, formal analysis, validation, methodology, and writing—review and editing. X.K., S.X., H.L., and C.W.: conceptualization, formal analysis, validation, methodology, and writing—review and editing. All authors read and approved the final manuscript.

Declaration of interests

L.Z. reports receiving research support from Jiangsu Hengrui Pharmaceuticals, Eli Lilly, Novartis, Roche, and Bristol-Myers Squibb. W.L., Qian Li, C.Z., and Z.H. are employees of Amoy Diagnostics Co., Ltd. X.K., S.X., H.L., and C.W. are employees of Qilu Pharmaceutical, Ltd.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Nasopharyngeal Carcinoma (tumor tissue, FFPE)	This paper	NCT04296994 and NCT05171790
Critical commercial assays
MagPure FFPE DNA LQ Kit	Magen, China	cat# D6323-02B
AmoyDx® FFPE RNA Extraction Kit	Amoy Diagnostics Co., Ltd. (AmoyDx),AmoyDx	cat# 8.02.0019
DNeasy Blood and Tissue Kit	Qiagen MD, USA	cat# 55114
dsDNA/RNA HS Assay Kit	Promega, Madison, WI, USA	QuantiFluor® dsDNA System(E2670), QuantiFluor® RNA System(E3310)
DNA/RNA HS Kit	Agilent Technologies, Santa Clara, CA, USA	Agilent High Sensitivity DNA Kit (5067-4626), Agilent RNA 6000 Pico Kit (5067-1513)
NEBNext® Ultra™ II DNA Prep Kit	NEB, Beverly, MA, USA	cat# E7645L
NEBNext® Ultra™ II Directional RNA Library Prep Kit	NEB, Beverly, MA, USA	cat# E7760L
Software and algorithms
Single-sample gene set enrichment analysis (ssGSEA) algorithm	Bindea, G. et al. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity 39, 782–795, doi: https://doi.org/10.1016/j.immuni.2013.10.003 (2013)	https://www.genepattern.org/modules/docs/ssGSEAProjection/4; RRID: SCR_026610
CIBERSORT	Bagaev, A. et al. Conserved pan-cancer microenvironment subtypes predict response to immunotherapy. Cancer Cell 39, 845–865 e847, doi: https://doi.org/10.1016/j.ccell.2021.04.014 (2021).	https://cibersortx.stanford.edu/; RRID: SCR_016955
R software	Bagaev, A. et al. Conserved pan-cancer microenvironment subtypes predict response to immunotherapy. Cancer Cell 39, 845–865 e847, doi: https://doi.org/10.1016/j.ccell.2021.04.014 (2021).	https://www.r-project.org/; R version 4.3.0; RRID: SCR_001905
xCell	Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 18, 220. doi: https://doi.org/10.1186/s13059-017-1349-1 (2017).	https://github.com/dviraran/xCell; RRID: SCR_026446
quanTIseq	Finotello F et al. Molecular and pharmacological modulators of the tumor immune contexture revealed by deconvolution of RNA-seq data. Genome Med. 11, 34. doi: https://doi.org/10.1186/s13073-019-0638-6 (2019).	https://icbi.i-med.ac.at/software/quantiseq/doc/index.html; RRID: SCR_022993
RSEM	Dewey BLCN. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinf. 12:323. doi: https://doi.org/10.1186/1471-2105-12-323 (2011).	version v1.2.28; RRID: SCR_000262
STAR	Dobin A et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 29:15–21. doi: https://doi.org/10.1093/bioinformatics/bts635 (2013)	https://github.com/alexdobin/STAR; version 2.7.4; RRID: SCR_004463

Experimental model and study participant details

Samples and ethical statement

The human subjects were obtained from the multicenter, open-label phase I/Ib clinical trial (identifier: NCT04296994 and NCT05171790) of QL1706, a bifunctional PD-1/CTLA-4 dual blockade in patients with advanced solid tumors. A total of 518 patients were enrolled in this single-arm trial and received QL1706. The study protocol and amendments were approved by the Institutional Review Board/Independent Ethics Committee of all participating institutions. All participants provided written informed consent before enrollment. The research was conducted in alignment with the Declaration of Helsinki and the principles of Good Clinical Practice Guidelines.

Method details

Study design, participants, and treatment

The procedures of this multicenter, open-label phase I/Ib clinical trial (identifier: NCT04296994 and NCT05171790) have been previously published.⁵ Eligible participants for phase I included individuals 18 years of age or older (both male and female) with a pathologically confirmed diagnosis of advanced solid tumors, who had failed or were ineligible for standard antitumor therapies, exhibited measurable disease as per the Response Evaluation Criteria in Solid Tumors, version 1.1 (RECIST 1.1), had an Eastern Cooperative Oncology Group (ECOG) performance status of ≤1, and a life expectancy of at least three months.

The eligibility criteria for phase Ib were broader than those for phase I. Patients qualified for phase Ib if they had histologically or cytologically confirmed metastatic or recurrent NPC, NSCLC, SCLC, CC, ovarian cancer, fallopian tube cancer, endometrial carcinoma, breast cancer, esophageal carcinoma, adenocarcinoma of the esophago-gastric conjunction, gastric cancer, HCC, cholangiocarcinoma, kidney cancer, CRC, urothelial carcinoma, or melanoma. Additionally, patients were required to have an absolute platelet count >75 × 109/L, and those with HCC needed to be classified as Child–Pugh class A or B. Key exclusion criteria included active or previous autoimmune or interstitial lung diseases, other conditions necessitating the long-term use of immunosuppressive drugs or corticosteroids, grade ≥3 irAEs from prior immunotherapy, or previous immunotherapy combining anti-CTLA-4 antibodies with anti-PD-(L)1 antibodies.

During the phase I dose escalation stage, patients received QL1706 at doses of 0.3, 1.0, 3.0, 5.0, 7.5 and 10.0 mg/kg intravenously every three weeks. All patients in phase Ib were administered QL1706 at 5mg/kg every three weeks (RP2D). Treatment in all arms continued until disease progression, completion of the study (up to two years), the occurrence of unacceptable toxicity, initiation of a new anticancer therapy, or withdrawal of informed consent, whichever came first.

The sample size for the dose expansion phase was adjusted in a timely manner according to the efficacy and safety signals of QL1706 for a specific tumor.

Study objectives and efficacy assessments

The primary objectives of phase I were to evaluate the safety, tolerability and RP2D of QL1706. Secondary objectives included immunogenicity, pharmacokinetics, and pharmacodynamics. The primary objectives of phase Ib were to estimate the preliminary efficacy of QL1706 in malignancies such as NSCLC, NPC, CC at the RP2D. Secondary objectives included safety and population-pharmacokinetics. Efficacy was evaluated based on the ORR, DoR, DCR, PFS and OS. PFS was defined as the time from the start of treatment until disease progression or death from any cause, whichever occurred first. OS referred to the time from the start of the treatment until death from any cause. In the NPC group, DCB was defined as a PFS ≥3 months, whereas NDB was defined as disease progression within three months and a survival status of 1 (1 = endpoint event). In the NSCLC group, DCB was defined as a PFS of ≥6; NDB was defined as PFS <6 and survival status as 1 (1 = endpoint event). Tumor assessments were conducted at baseline, every six weeks (two cycles) during the first four cycles, and every nine weeks (three cycles) thereafter.

Identification of risk factors for clinical outcomes

We conducted univariate and multivariate Cox regression analyses to examine the relationship between clinical features and outcomes in NSCLC, NPC and CC treated with QL1706 at the RP2D. Clinical data collected at baseline included age (≤ median vs. >median), gender (female vs. male), ECOG (1 vs. 0), number of metastatic sites (0–1 vs. ≥ 2), history of prior immunotherapy (yes vs. no), number of previous lines of therapy (0–1 vs. ≥ 2), dNLR (>3 vs. ≤ 3), LDH level (> ULN vs. ≤ ULN), albumin level (<3.5 g/dL vs. ≥ 3.5 g/dL), liver metastasis (yes vs. no), and prior radiotherapy (yes vs. no). The dNLR was calculated using the formula: absolute neutrophil count/(white blood cell concentration − absolute neutrophil count), and the ULN for LDH was defined according to the limit of each center. Cox proportional hazards regression models were built to identify factors independently associated with poor clinical outcomes (PFS ≤ mPFS, NDB and OS ≤ mOS).

DNA/RNA extraction, sequencing and gene expression Estimation

According to the protocol, all tumor samples were collected within 28 days before the first dose. All tumor tissue analyses were performed after the data cut-off on August 1^st, 2023. Baseline patient characteristics was provided in Tables S4–S7. DNA and RNA extractions were performed using the MagPure FFPE DNA LQ Kit (Magen) for DNA and the AmoyDx FFPE RNA Extraction Kit (AmoyDx) for RNA. DNA from FFPE samples was extracted using the DNeasy Blood and Tissue Kit (Qiagen, USA). The concentrations of DNA and RNA were quantified using the Quantus fluorometer alongside the Quantus dsDNA/RNA HS Assay Kit (Promega). The integrity of DNA and RNA was assessed using the DNA/RNA HS Kit and the Agilent Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA, USA).

For DNA library preparation, DNA was fragmented to 200–250 bp using the M220 sonicator (Covaris). Indexed next-generation sequencing (NGS) libraries were then prepared through end-repairing, A-tailing, adaptor ligation, and amplification procedures using the NEBNext Ultra II DNA Prep Kit (NEB, Beverly, MA, USA). The quality of the libraries was assessed using an Agilent Bioanalyzer DNA 1000 kit and the Qubit DNA HS fluorescence kit. DNA libraries for hybrid capture and sequencing were captured separately using the AmoyDx Master Panel, which targets a panel of cancer-related genes (listed in Tables S8–S10).

For RNA library preparation, RNA was first fragmented using RNA fragmentation reagent (Ambion) before undergoing reverse transcription to synthesize complementary DNA (cDNA). The NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (NEB, Beverly, MA, USA) was used for cDNA synthesis and library preparation. The resulting cDNA libraries were sequenced on the ILLUMINA NOVASEQ 6000 platform (Illumina Inc., CA, USA).

Paired-end RNA-seq reads were aligned to the Homo sapiens genome assembly GRCh37 (hg19) using STAR 32 (version 020, 201). The transcriptome annotation was sourced from GENCODE version 20. Gene quantification was performed using RSEM 33 (version v1.2.28) (RSEM 33), which facilitated the calculation of Transcripts Per Million (TPM) values at the gene level by counting reads within coding regions and accounting for variations in library preparation methods.

For the differential gene expression and pathway analysis, the Wilcoxon rank-sum test, a non-parametric statistical method, was used to compare gene expression levels between two groups. Differentially expressed genes (DEGs) were defined by a Fold Change of ≥1.5 and a p-value of <0.05 in the comparison between the DCB and NDB groups. ESTIMATE enrichment analyses were conducted using the R package “cluster Profiler” to determine the functional roles of DEGs. These analyses included Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), Hallmark, Wikipathways, and Reactome pathways. The gene sets for these analyses were obtained from the Human Molecular Signatures Database.

Evaluation of gene signatures, TME, tumor infiltrating lymphocytes (TILs) and PD-L1

Gene signatures were employed to calculate single-sample gene set enrichment analysis (ssGSEA) scores, including TGFβ (ACTA2, ACTG2, ADAM12, ADAM19, CNN1, COL4A1, CTGF, CTPS1, FAM101B, FSTL3, HSPB1, IGFBP3, PXDC1, SEMA7A, SH3PXD2A, TAGLN, TGFBI, TNS1, TPM1), IFN_gama_6 (IDO1, CXCL10, CXCL9, HLA-DRA, IFNG, STAT1), Chemokine (CCL2, CCL3, CCL4, CCL5, CCL8, CCL18, CCL19, CCL21, CXCL9, CXCL10, CXCL11, CXCL13), IFN_gama/Effector T cell (CD8A, GZMA, GZMB, IFNG, EOMES, CXCL9, CXCL10, TBX21), Effector T cell (GZMA, GZMB, PRF1, IFNG, EOMES, CD8A) and T_cell_surv (LTBR, DBI, VSTM1, CD59, IL23A, IL12B, MAPK3, ADA, BATF).⁶⁰ Single-sample gene set enrichment analysis (ssGSEA) algorithm was used to evaluate the relative abundance of infiltrating immune cells in the TME of NPC and NSCLC. Gene sets representing TME-infiltrating immune cells were derived from prior datasets.⁶¹ The R package "GSVA" was utilized to calculate enrichment scores, which indicated the relative abundance of each type of TME-infiltrating cells in NPC and NSCLC samples. Additionally, the xCell⁶² tool was used to calculate scores for TILs, including B cells, CD4⁺ T cells, CD8⁺ T cells and natural killer cells.⁶³ The QuanTIseq⁶⁴ was also utilized to assess the TIL density within the tumor samples. The threshold for TIL was adopted from published literature.⁶³ The threshold for positive PD-L1 expression was defined as CD274 transcripts per million (TPM) < 23.6. This threshold was determined using receiver operating characteristic (ROC) curve analysis, with OS_status as the outcome variable.

Validation datasets

Transcriptomic data and clinical data for the validation cohort were collected from the The Cancer Genome Atlas (TCGA) database (https://tcga-data.nci.nih.gov/tcga/), the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/, GEO GSE102349). published literature (OAK, POPLAR, GEO: GSE224450 and RATIONALE-309).^10,27,65,66 The clinical information is presented in Table S11.

Quantification and statistical analysis

Efficacy analyses were conducted on the population that underwent at least one efficacy evaluation. The data cutoff date for this report was August 1^st, 2023. The ORR and DCR were estimated, with their 95% CI calculated using the Clopper-Pearson method. The Kaplan-Meier method was used to evaluate OS, PFS and DOR and the 95% CI of the median value was calculated using the Brookmeyer-Crowley method. Patient demographic characteristics were summarized using descriptive statistical methods. The chi-square or Fisher’s exact probability tests were used to evaluate the significance of differences in rates or percentages. Variables identified as significant at the p < 0.05 level in the univariate analysis were included in the Cox multivariate regression analysis. Non-parametric Wilcoxon rank-sum tests were utilized to compare medians between two datasets, while ANOVA tests were used for comparisons among three or more groups. Receiver operating characteristic (ROC)-AUC analysis was performed using the R package "pROC" and "survival". All statistical analyses were conducted using R Project (version 4.1.2; https://www.r-project.org/). p-values of <0.05 were considered statistically significant.

Additional resources

This study has been registered on clinicaltrials.gov (identifier: NCT04296994 and NCT05171790).

Published: October 3, 2025