Neoadjuvant retlirafusp alfa (anti-PD-L1/TGF-β bifunctional fusion protein) with or without chemotherapy in unresectable stage III non-small cell lung cancer: updated results from the phase 2 TRAILBLAZER trial Authors: Yi Pan, Xuening Yang, Qing Zhou, Yanqiu Zhao, Guang Han, Qingsong Pang, Zhenfa Zhang, Qifeng Wang, Jun Yao, Hui Wang, Weihua Yang, Baogang Liu, Qixun Chen, Xianghui Du, Kaican Cai, Baosheng Li, Yunchao Huang, Xiao Li, Yijun Jia, Li Song, Wei Shi, Yi-Long Wu Published: 2026 ABSTRACT AbstractTRAILBLAZER is a proof-of-concept phase 2 study (ClinicalTrials.gov, NCT04580498) of neoadjuvant retlirafusp alfa (an anti-PD-L1/TGF-β bifunctional agent) in unresectable stage III non-small cell lung cancer (NSCLC) not harboring EGFR or ALK alterations. During induction, retlirafusp alfa was administered either with chemotherapy or as monotherapy, followed by surgery or radiotherapy as assessed by a local multidisciplinary team and consolidation retlirafusp alfa. Patients without high PD-L1 expression were allocated to retlirafusp alfa plus chemotherapy (arm A; n = 88); patients with high PD-L1 expression were randomly allocated at a 1:1 ratio to retlirafusp alfa combination (arm B; n = 9) or monotherapy (arm C; n = 10). In this updated analysis (data cutoff, March 24, 2025), the median follow-up was 39.4 months. The event-free survival (EFS) rate at 3 years was 50.5% (95% CI 39.0–61.0) in arm A + B and 77.1% (34.5–93.9) in arm C; the corresponding overall survival (OS) rates at 3 years were 68.3% (95% CI 57.5–77.0) and 87.5% (38.7–98.1), respectively. Among 27 patients who underwent surgery, the 3-year EFS and OS rates were 69.5% (95% CI 48.1–83.5) and 84.9% (64.5–94.0), respectively, versus 50.8% (36.5–63.4) and 70.4% (56.7–80.5), respectively, in radiotherapy-treated patients. The safety data were consistent with those in a previous report. With extended follow-up, the study regimen showed sustained survival benefits with no new safety signals. Patients who underwent surgery after induction treatment achieved clinically meaningful survival gains. Our findings support the use of neoadjuvant immunotherapy with a response-adapted surgical strategy as a promising approach for unresectable stage III NSCLC.Subject terms: Lung cancer, Cancer therapy Introduction Stage III non-small cell lung cancer (NSCLC) represents one of the most clinically challenging disease types owing to its diverse presentations in terms of tumor volume, local invasion, and extent of nodal involvement.1 The therapeutic goal in stage III NSCLC is a cure, and management is inherently multimodal, requiring multidisciplinary team (MDT) evaluation to guide treatment selection and sequencing through the integration of locoregional therapy (i.e. surgery or radiotherapy) with systemic therapy to control micrometastatic spread.2,3 For patients with stage III unresectable disease without driver mutations, the current standard treatment is definitive chemoradiation (CRT) followed by consolidation anti-PD-L1 therapy. This regimen was established by the phase 3 PACIFIC trial of durvalumab after concurrent CRT and further supported by the GEMSTONE-301 trial of sugemalimab after concurrent/sequential CRT, both of which showed significant progression-free survival (PFS) benefits.4,5 Despite the substantial progress achieved with the introduction of immunotherapy beyond the chemotherapy era, most patients eventually experience treatment failure.6,7 In addition, up to 30% of patients are ineligible for consolidation immunotherapy after CRT owing to disease progression, residual toxicity, or clinical deterioration.8–12 To improve survival and increase the population benefitting from immunotherapy, two phase 3 trials (PACIFIC-2 and CheckMate 73 L) evaluated earlier administration of immunotherapy concurrently with CRT in unresectable stage III NSCLC.13,14 However, neither demonstrated additional benefit, possibly because of radiation-induced ablation of tumor-draining lymphatics and impaired immune priming.15 Furthermore, safety remains a concern; the overlapping toxicities with concurrent administration may offset potential survival gains.13,14 In contrast, initiating immunotherapy before local therapy has consistently improved pathological complete response (pCR) and survival outcomes in patients with resectable disease,16–23 providing a compelling rationale for evaluating the neoadjuvant strategy in the unresectable setting. Delivering immunotherapy upfront preserves the lymphatic architecture and tumor antigen exposure, which may increase immune activation and thereby facilitate tumor downstaging, elimination of micrometastases, and surgical conversion.24,25 While the utility of induction chemotherapy has historically been constrained by suboptimal complete resection rates, the integration of neoadjuvant immunotherapy has been shown to improve complete resection rates.16,21,22 Definitive surgery offers maximal local control, and together with consolidation immunotherapy, represents a potential approach to improve cure rates. TGF-β signaling acts as a crucial driver of tumor progression, promoting cell proliferation, invasion, and angiogenesis.26 It also contributes to immune evasion through mechanisms complementary to the PD-1/PD-L1 pathway. By promoting the expansion of cancer-associated fibroblasts and the deposition of a collagen-rich extracellular matrix, TGF-β creates a physical and biochemical barrier that restricts CD8 + T-cell infiltration.27 Furthermore, it suppresses antitumor immunity by promoting regulatory T-cell differentiation and impairing dendritic cell function, contributing to resistance to checkpoint blockade.28 Therefore, dual blockade of TGF-β and PD-1/PD-L1 signaling provides a strong mechanistic rationale for enhanced antitumor activity. Retlirafusp alfa (SHR-1701) is a bifunctional immunotherapy designed to inhibit both pathways through the fusion of a PD-L1-targeted monoclonal antibody to the extracellular domain of TGF-β receptor II.29–31 Retlirafusp alfa has been approved in combination with chemotherapy for the first-line treatment of advanced gastric cancer and gastroesophageal junction adenocarcinoma in China. 29 In NSCLC, the preliminary clinical activity of retlirafusp alfa was shown in a phase 1 study as a first-line treatment for PD-L1-positive advanced disease, in which the objective response rate (ORR) was numerically higher than that reported with approved PD-1/L1-targeted therapy. 32 TRAILBLAZER is a proof-of-concept phase 2 study in patients with MDT-determined, initially unresectable stage III NSCLC that evaluated a multimodal approach of neoadjuvant immunotherapy-based treatment, followed by surgery or radiotherapy, and subsequent consolidation immunotherapy.33 This adaptive-local-therapy design allowed evaluation of the feasibility and potential benefits of surgical conversion after induction immunotherapy. Given the higher tumor burden and poorer prognosis of unresectable disease, we investigated retlirafusp alfa, a potentially more active immunotherapy. The study met the dual primary endpoints with neoadjuvant retlirafusp alfa plus chemotherapy, with an ORR of 58% after induction treatment and an event-free survival (EFS) rate of 56.6% at 18 months. The safety profile was manageable, with hematological toxicity being the most frequent grade 3 or higher adverse event (AE) and relatively low rates of severe pneumonitis and radiation pneumonitis.33 Importantly, immunotherapy-based induction enabled surgery in 25% of patients, resulting in surgical (complete resection rate, 100%) and pathological outcomes (major pathological response rate, 44%; pCR rate, 26%) comparable to those of patients with upfront resectable disease. It remains to be elucidated whether surgical conversion with the neoadjuvant strategy translates to longer-term survival benefits. Here, we present an updated analysis after more than 3 years of follow-up, extending 18 months since the primary report, focusing on long-term survival outcomes. Results Patients From November 20, 2020, to January 27, 2022, a total of 107 eligible patients received study treatment. Ninety-seven patients received neoadjuvant retlirafusp alfa plus chemotherapy (arm A, n = 88; arm B, n = 9), and 10 received retlirafusp alfa monotherapy (arm C; Fig. 1). All 107 patients were analyzed for efficacy and safety. Overall, 27 (25%) patients underwent surgery, including 24 (25%) in arm A + B and 3 (30%) in arm C. Additionally, 59 (55%) patients underwent radiotherapy, including 55 (57%) in Arm A + B and 4 (40%) in Arm C. Local therapy was not received by 21 (20%) patients; the most common reasons were consent withdrawal and disease progression (8% and 4%, respectively; Supplementary Table 1). Study design. * Retlirafusp alfa (30 mg/kg) plus paclitaxel (175 mg/m2) and carboplatin (AUC 5), all administered on day 1 of each cycle. † Retlirafusp alfa (30 mg/kg). ‡ Radiotherapy (60 Gy/30 fractions) plus concurrent cisplatin (30 mg/m2, once weekly). § Treatment beyond disease progression was permitted if the investigator considered the patient to derive clinical benefit. MDT, multidisciplinary team; Q3W, every 3 weeks; R, randomization; TPS, tumor proportion score The study population was predominantly male (94%) and had a smoking history (90%); 74% presented with squamous-cell carcinoma. At study entry, most patients had stage IIIB/IIIC NSCLC (44%/30%), and 18% had high PD-L1 expression (Table 1). Table 1Patient and tumor characteristics at baselineArm A (n = 88)Arm B (n = 9)Arm C (n = 10)Arm A + B (n = 97)All patients (n = 107)Age, median (range), y59 (35–70)58 (31–66)58 (47–69)59 (31–70)59 (31–70)Male84 (95)7 (78)10 (100)91 (94)101 (94)ECOG performance status029 (33)5 (56)2 (20)34 (35)36 (34)159 (67)4 (44)8 (80)63 (65)71 (66)HistologySquamous65 (74)7 (78)7 (70)72 (74)79 (74)Nonsquamous23 (26)2 (22)3 (30)25 (26)28 (26)Disease stageIIIAa20 (23)3 (33)4 (40)23 (24)27 (25)IIIB41 (47)3 (33)3 (30)44 (45)47 (44)IIIC26 (30)3 (33)3 (30)29 (30)32 (30)Other1 (1)001 (1)1 (1)Smoking statusCurrent or former smoker79 (90)7 (78)10 (100)86 (89)96 (90)Never smoked9 (10)2 (22)011 (11)11 (10)PD-L1 TPS<1%48 (55)0048 (49)48 (45)1–49%40 (45)0040 (41)40 (37)≥50%09 (100)10 (100)9 (9)19 (18)Dara are n (%) or otherwise specifiedECOG Eastern Cooperative Oncology Group, TPS tumor proportion scoreaJudged as unresectable because of tumor invasion (n = 7), multistation N2 disease (n = 8), bulky N2 disease (n = 4), and R0 resection not achievable without pneumonectomy (n = 8) Patient and tumor characteristics at baseline Dara are n (%) or otherwise specified ECOG Eastern Cooperative Oncology Group, TPS tumor proportion score aJudged as unresectable because of tumor invasion (n = 7), multistation N2 disease (n = 8), bulky N2 disease (n = 4), and R0 resection not achievable without pneumonectomy (n = 8) Efficacy In this updated analysis (March 24, 2025), the median follow-up was 39.4 months. At the data cutoff, 42 (39%) of the 107 patients experienced EFS, including 40 (41%) of 97 in arm A + B and 2 (20%) of 10 in arm C. The median EFS was not reached (NR) in all patients or in either treatment arm (Fig. 2a–c). The 2- and 3-year EFS rates were 55.5% (95% CI 44.5–65.2) and 52.9% (41.9–62.7) in the overall population; 53.5% (42.0–63.7) and 50.5% (39.0–61.0) in arm A + B; and 77.1% (34.5–93.9) at both time points in arm C, respectively. Overall, 72.6% of patients (95% CI 61.4–81.0) were estimated to be free from distant metastasis or death at 2 years, and 70.8% (59.4–79.6) at 3 years. A total of 31 (29%) patients died, including 30 (31%) in arm A + B and 1 (10%) in arm C. The 2- and 3-year overall survival (OS) rates were 78.6% (95% CI 69.1–85.5) and 70.0% (59.7–78.1) in the overall population; 77.8% (67.7–85.1) and 68.3% (57.5–77.0) in arm A + B; and 87.5% (38.7–98.1) at both time points in arm C (Fig. 3a–c). Data on subsequent anticancer systemic therapy after the end of the study are provided in Supplementary Table 2. K‒M analysis of EFS in arm A + B (a), arm C (b), all patients (c) and patients receiving local therapy (d). EFS, event-free survival; RT, radiotherapy K‒M analysis of OS in arm A + B (a), arm C (b), all patients (c) and patients receiving local therapy (d). OS, overall survival; RT, radiotherapy In a post hoc analysis, 93 of 107 (87%) patients (including 24 of 27 [89%] who underwent surgery) were classified as having unresectable stage III NSCLC per the European Organization for Research and Treatment of Cancer (EORTC) consensus criteria. Among them, EFS events occurred in 37 (40%) patients. The median EFS was NR; the EFS rate was 55.5% (95% CI 43.7–65.8) at 2 years and 52.5% (40.7–63.0) at 3 years. The OS data were 30% mature (28/93 deaths). The median OS was NR; the OS rate was 76.8% (95% CI 66.3–84.4) at 2 years and 68.2% (57.1–77.0) at 3 years. Survival outcomes by treatment arm are summarized in Supplementary Table 3. Local therapy outcomes Overall, 27 patients underwent surgery, and 59 received radiotherapy. PET‒CT was performed in 54 of 107 (50%) patients after induction therapy, including 14 (52%) who received surgery and 32 (54%) who received radiotherapy. The baseline characteristics of the surgical patients were consistent with those of the overall population (Supplementary Table 4). EFS events occurred in 8 (30%) surgically treated patients and 26 (44%) patients treated with radiotherapy. As of the data cutoff, the median EFS was NR in both subgroups. The 2- and 3-year EFS rates were 69.5% (95% CI 48.1–83.5) in the surgery group and 55.1% (40.7–67.3) and 50.8% (36.5–63.4), respectively, in the radiotherapy group (Fig. 2d). Five (19%) surgically treated patients and 18 (31%) radiotherapy-treated patients died. The 2-year and 3-year OS rates were 92.6% (95% CI 73.5–98.1) and 84.9% (64.5–94.0), respectively, in the surgery group and 77.7% (64.7–86.4) and 70.4% (56.7–80.5), respectively, in the radiotherapy group (Fig. 3d). Safety By the time of the primary EFS analysis (Sep 30, 2023), all patients had completed or discontinued the study treatment.33 The safety data were consistent with those of previous reports, with no new grade 3–5 treatment-related pneumonitis or treatment-related deaths. All treatment-related AEs of grade 3 or higher and an incidence of ≥5% were hematological toxicities, except for hypokalemia (6%). Detailed data on the AEs related to the study agents, surgery and radiotherapy are provided in Supplementary Tables 5–8. Discussion The ultimate therapeutic goal in stage III NSCLC is a cure. Because immunotherapy enhances tumor downstaging, facilitates R0 resection and improves distant control,17 optimizing locoregional control through surgery may become increasingly critical to achieve this goal. However, as highlighted by the 2024 EORTC consensus, the role of postdownstaging surgery in initially unresectable disease is uncertain.34 Long-term outcomes are essential for fully evaluating oncologic resectability. In this context, we report, to our knowledge, the first >3-year outcomes from a prospective study of immunotherapy-based induction followed by adaptive-local therapy in unresectable stage III NSCLC. In patients treated with neoadjuvant retlirafusp alfa plus chemotherapy (primary cohort), the EFS and OS benefits were sustained over time, with 3-year survival rates of 50.5% and 68.3%, respectively. Favorable outcomes in the overall population indicate that deferring local therapy within a neoadjuvant framework did not compromise disease control. Among patients who underwent surgery, the EFS and OS curves remained separate from those of patients who received definitive radiotherapy, with EFS rates of 69.5% vs 50.8% and OS rates of 84.9% vs 70.4%, respectively, at 3 years. Although a longer follow-up is pending, the EFS plateau beyond 2 years suggests a durable OS tail and holds the potential for an improved cure rate. The safety outcomes remained consistent with those of previous reports, with a modest rate of severe pulmonary toxicity. Collectively, our 3-year data provide a clinically meaningful benchmark supporting the strategy of chemoimmunotherapy induction, highlight the curative potential of surgery as a response-adapted local therapy in patients with upfront unresectable stage III NSCLC and underscore the need to extend the definition of resectability in the immunotherapy era. In CRT-pretreated patients without progression or residual toxicity, PACIFIC reported PFS and OS rates at 3 years of 39.7% and 56.7%, respectively, with durvalumab after concurrent CRT,6 and GEMSTONE-301 reported rates of 26.1% and 55.8%, respectively, with sugemalimab after concurrent or sequential CRT.7 In the present study, in previously untreated patients, the EFS and OS rates at 3 years were 50.5% and 68.3%, respectively, in those receiving neoadjuvant retlirafusp alfa plus chemotherapy. Although cross-trial comparisons in stage III NSCLC are inherently limited by heterogeneity in terms of the eligibility criteria and different time origins for survival analyses, our results appear favorable compared with those of approved regimens, given the higher baseline burden of poor prognostic features, including higher proportions of stage IIIC disease (30% versus 0–21%) and squamous histology (74% versus 47–69%).4,5,14 Furthermore, our analyses included all treated patients, irrespective of response or tolerability. The prevalence of squamous histology in our study was reflective of the East Asian patient population lacking actionable genomic alterations, with EGFR mutations occurring more frequently in adenocarcinoma patients and Asians.5,22 Notably, the early plateau beyond 2 years in EFS in this study has not been observed in other unresectable NSCLC trials using consolidative immunotherapy alone. Considering that distant metastasis is a main failure mode in locally advanced disease, early systemic therapy intensification is biologically rational; our data make a compelling case for a neoadjuvant strategy. The induction–consolidation immunotherapy approach may also help mitigate toxicity by avoiding full concurrent overlap with CRT. The CheckMate 73 L trial reported higher rates of severe toxicity with simultaneous CRT and immunotherapy, which may have influenced survival outcomes.14 Favorable outcomes were also observed in arm C. While the sample size was small because of slow accrual and warrants cautious interpretation, our findings, together with those of the SPRINT and AF-16 trials,35,36 suggest that single-agent immunotherapy induction may be considered for selected patients. Taken together, our regimen is intended to offer an alternative option for candidates for cCRT and immunotherapy, including patients eligible for the PACIFIC approach and broader populations. Although 20% of patients did not proceed to local therapy, this was multifactorial, as has also been reported in other trials that incorporate postinduction definitive local treatment.19,20,37,38 The favorable long-term survival outcomes seen in the overall population support this induction strategy, and further studies are warranted to optimize the target patient population. In the present analysis, the long-term survival benefit of the neoadjuvant regimen was supported by the surgical subset, yielding absolute 3-year improvements of 18.7% in EFS and 14.5% in OS for resected patients vs radiotherapy-treated patients, with survival curves diverging after 1 year. Notably, these improvements were achieved with baseline disease characteristics in the surgical cohort comparable to those of the overall population and generally in alignment with the EORTC consensus criteria for unresectable disease.33 In phase 3 trials of resectable NSCLC (CheckMate-816, AEGEAN and KEYNOTE-671),19,37,38 neoadjuvant or perioperative anti-PD-1/PD-L1 antibody plus chemotherapy led to 3-year EFS rates ranging from 54–60.1% and OS rates ranging from 67.1–78%. In this study, patients whose disease was initially unresectable converted to resectable disease after retlirafusp alfa-based induction achieved comparable survival outcomes, despite having more advanced disease at baseline (stage IIIB/IIIC, 63% vs 0–24%; N2/3, 85% vs 42.3–49.5%). These data extend prior observations of similar surgical outcomes and highlight the robust activity of retlirafusp alfa against micrometastatic disease, whose burden increases with advancing stage. TGFβ signaling drives tumor growth and spread by promoting fibrosis, epithelial–mesenchymal transition, and angiogenesis and contributes to an immunosuppressive tumor microenvironment that limits the efficacy of immunotherapy;26 thus, dual TGFβ and PD-1/PD-L1 blockade may potentiate the activity of retlirafusp alfa. Long-term outcomes comparable to those of upfront resectable patients, together with superior survival in patients who underwent resection versus those who received radiotherapy, further support the incorporation of surgery after induction immunotherapy. At the time of protocol development, there was no universally accepted definition of resectability for stage III NSCLC, a highly heterogeneous disease group. The protocol, therefore, prespecified pragmatic criteria to support consistent decision-making (anticipated complete resection without the need for pneumonectomy). To ensure consistency, resectability was adjudicated by the same local MDT at baseline and after induction therapy. Our baseline classification showed high concordance (87%) with the EORTC consensus, with discordance largely driven by cases in which the anticipated need for pneumonectomy rendered disease unresectable. In a post hoc analysis of patients classified as unresectable by the EORTC criteria, survival outcomes were similar, supporting the robustness of our findings. One major limitation of this study is the absence of a standard-therapy control group, which precludes direct comparison. In addition, the small size of the high PD-L1 expression cohort limits the evaluation of immunotherapy monotherapy as an induction treatment. Moreover, biomarker analyses of the TGF-β pathway were not conducted. Furthermore, although PET‒CT was the preferred imaging modality for restaging per protocol, access could be limited by resource constraints and patients’ clinical condition, as in real-world practice. Finally, for pragmatic reasons, treatment allocation between surgery and radiotherapy following induction treatment was determined by local MDTs. Although guided by the protocol, these decisions also relied on surgeon experience and institutional practice and therefore could vary across sites. The generalizability of our results warrants further confirmation. In summary, in this updated analysis of TRAILBLAZER, neoadjuvant retlirafusp alfa with or without chemotherapy, followed by MDT’s choice of surgery or radiotherapy, produced a sustained survival benefit, with no new safety signals. With extended follow-up, patients who underwent postinduction surgery achieved clinically meaningful survival gains. These findings suggest that early initiation of immunotherapy, such as induction followed by response-adapted local therapy, including surgery when feasible, is a promising approach for unresectable stage III NSCLC, which merits confirmation in randomized trials. Materials and methods Study design and patients This open-label, phase 2 study (ClinicalTrials.gov, NCT04580498) was performed at 20 clinical sites in China. The protocol and its amendments were approved by ethics committees at all the sites (Supplementary Table 9). The study adhered to the Declaration of Helsinki and the Good Clinical Practice guidelines, with written informed consent obtained from all participants. Patients with histologically or cytologically confirmed stage III NSCLC, according to the American Joint Committee on Cancer 8th edition, were enrolled. Surgical unresectability was determined by a local MDT that included at least one radiologist, a thoracic surgeon, and a medical oncologist. Recommended criteria for unresectable nodal disease in Stage IIIA patients, including bulky or multistation involvement or complete resection, were not feasible.39 Disease was deemed resectable if R0 surgery (excluding pneumonectomy) could be achieved alongside systematic lymphadenectomy. Pneumonectomy was excluded because of its high perioperative mortality risk.40 Additional criteria included age 18–70 years, an Eastern Cooperative Oncology Group performance status of 0–1, and adequate lung function (forced expiratory volume in 1 s greater than 1.0 L and more than 40% of the predicted normal volume). Patients were excluded if they had EGFR or ALK genomic alterations, previous systemic therapy for NSCLC or radiation therapy for thoracic disease. Additional exclusion criteria included malignant pleural effusion; recent use of immunosuppressive ( < 2 weeks) or immunostimulatory ( < 4 weeks) agents; a history of autoimmune disease; and active or suspected interstitial lung disease. The full protocol is available in the Supplement file. Treatment The study protocol followed a three-phase structure: induction, local therapy, and consolidation. During induction, patients were treated with retlirafusp alfa (30 mg/kg), either as a single agent or combined with paclitaxel (175 mg/m²) and carboplatin (AUC 5). All the agents were administered intravenously on day 1 of each 3-week cycle for a total of 3 cycles. Treatment allocation was determined by PD-L1 expression: those with a tumor proportion score (TPS) less than 50% were assigned to combination therapy (arm A), whereas patients with a TPS of 50% or greater were randomly assigned a 1:1 ratio to receive either combination therapy (arm B) or retlirafusp alfa monotherapy (arm C). Block randomization was performed using a sequence produced by an independent entity. Following induction, resectability was reassessed by the same local MDT. Patients considered eligible for definitive surgery underwent resection; those deemed unresectable received definitive radiotherapy (60 Gy over 30 fractions at 5 fractions per week), together with weekly cisplatin (30 mg/m²), all within 6 weeks. Consolidation retlirafusp alfa (30 mg/kg) was administered in 3-week cycles for up to 16 cycles. Investigators could allow treatment to continue beyond initial disease progression at their discretion. Outcomes The trial utilized two primary endpoints, namely, the ORR after induction therapy and the EFS rate at 18 months (defined as the interval from the first study dose until the first disease progression or recurrence or death from any cause), both of which were evaluated by an investigator according to RECIST version 1.1. The secondary endpoints were time to death or distant metastasis (TDDM; defined as the interval from the first dose to distant progression per investigator or death from any cause), OS (defined as the interval from the first dose to death from any cause), and safety. A tumor scan was conducted at baseline with PET-CT and brain MRI. At restaging following induction therapy, PET-CT was recommended whenever feasible; otherwise, response evaluation was based on contrast-enhanced CT. A brain MRI was performed as clinically indicated. Tumor assessments were also performed 2 to 4 weeks after the end of local treatment and at 9-week intervals during consolidation treatment with contrast-enhanced CT. During follow-up, imaging assessments were performed with contrast-enhanced CT at 3-month intervals in year 1 and at 6-month intervals in years 2–3, followed by annual CT or radiography thereafter. Statistical analysis As previously detailed,33 a total of 94 patients were planned for the primary cohort receiving neoadjuvant chemoimmunotherapy. This target provides 90% power to identify an EFS rate of 45.9% at 18 months, with a one-sided α of 0.025. The calculation adopted an exponential assumption for EFS and used a historical 18-month EFS rate of 28.7% for cCRT in stage III NSCLC, which was derived from evidence at the time of study design.4,40,41 Analyses of clinical activity and safety were conducted for all treated patients. Arm A + B was the primary study cohort and included all patients who received neoadjuvant chemoimmunotherapy, irrespective of PD-L1 expression. EFS, OS and TDDM were analyzed with the Kaplan‒Meier method. We estimated 95% CIs for landmark rates using the normal approximation on the log(–log) scale, followed by back-transformation to the original scale. Safety was summarized descriptively. All analyses were carried out using SAS, v9.4. Supplementary information Supplementary Data Study protocol Acknowledgements This work was sponsored by Jiangsu Hengrui Pharmaceuticals. The authors express their sincere appreciation to the patients, their families, and the investigators involved in this work. We also acknowledge You Li (Jiangsu Hengrui Pharmaceuticals) for providing project administration and overseeing data collection. Editorial assistance was provided by Xiuzhi Wu (Jiangsu Hengrui Pharmaceuticals). Author contributions Y.-L.W. and W.S. conceived and designed the study. Y.-L.W., Y.P., X.Y., Q.Z., Y.Z., G.H., Q.P., Z.Z., Q.W., J.Y., H.W., W.Y., B.Liu., Q.C., X.D., K.C., B.Li., and Y.H. contributed to patient enrollment and conducted the investigation. Y.J. and L.S. were responsible for project administration. All the authors contributed to data interpretation and manuscript writing and revision. All the authors have read and approved the article. Data availability All the data supporting the conclusions of this study are contained within this article or its accompanying supplementary files. For inquiries regarding materials or further information, please contact the corresponding author, Y.-L.W. (syylwu@live.cn). Competing interests Disclosures for Y.-L.W. include honoraria received from AstraZeneca, Roche, Pfizer, Boehringer Ingelheim, Lilly, MSD Oncology, Bristol-Myers Squibb, and Hengrui. Additionally, Y.-L.W. served in a consulting or advisory capacity for AstraZeneca, Boehringer Ingelheim, Roche, and Takeda while receiving research support from Roche, Boehringer Ingelheim, Pfizer, and Bristol-Myers Squibb. Q.Z. received honoraria from AstraZeneca, Roche, Pfizer, Lilly, Boehringer Ingelheim, Bristol-Myers Squibb, MSD Oncology, and Hengrui. X.L., Y.J., L.S., and W. S. declare employment at Jiangsu Hengrui Pharmaceuticals. All remaining authors have no competing interests to disclose. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Yi Pan, Xuening Yang, Qing Zhou Supplementary information The online version contains supplementary material available at 10.1038/s41392-026-02779-1. References Associated Data Supplementary Materials Supplementary Data Study protocol Data Availability Statement All the data supporting the conclusions of this study are contained within this article or its accompanying supplementary files. For inquiries regarding materials or further information, please contact the corresponding author, Y.-L.W. (syylwu@live.cn).