Introduction

Lung cancer is one of the malignant tumors with the highest morbidity and mortality worldwide, among which non-small cell lung cancer (NSCLC) accounts for 80–85% of all lung cancers.1 Epidermal growth factor receptor (EGFR) gene mutation is one of the most common driver mutations in NSCLC, with an incidence as high as 40–50% in East Asian populations, significantly higher than that in Caucasian populations.2,3 The advent of EGFR tyrosine kinase inhibitors (TKIs) has significantly improved the prognosis of patients with EGFR-mutant NSCLC.4,5 However, almost all patients eventually develop acquired resistance after EGFR-TKI treatment, leading to disease progression, which remains one of the major clinical challenges in the treatment of EGFR-mutant NSCLC.6,7

For patients who have progressed after EGFR-TKI treatment, the traditional treatment regimen is platinum-based doublet chemotherapy, but the efficacy is limited, with a median progression-free survival (PFS) of only 4–5 months and a median overall survival (OS) of less than 12 months.8,9 Bevacizumab is a humanized monoclonal antibody against vascular endothelial growth factor (VEGF). Multiple studies have confirmed that bevacizumab plus chemotherapy can significantly improve the objective response rate (ORR) and PFS in patients with EGFR-TKI resistance, and it has become one of the standard second-line treatment regimens recommended by domestic and international guidelines.10 However, bevacizumab only blocks the VEGF pathway and cannot improve the tumor immune microenvironment, resulting in limited survival benefits. Moreover, long-term use increases the risk of adverse reactions such as hypertension and proteinuria.

In recent years, the emergence of immune checkpoint inhibitors (ICIs) has brought important advances to the treatment of NSCLC, but their efficacy in patients with EGFR-mutant NSCLC has been unsatisfactory. Multiple large Phase III clinical studies have shown that traditional PD-1/PD-L1 monoclonal antibodies plus chemotherapy fail to bring significant clinical benefits to NSCLC patients with EGFR-TKI resistance. The KEYNOTE-789 study showed that pembrolizumab plus chemotherapy compared with chemotherapy alone only prolonged median PFS by 0.1 months (5.6 months vs 5.5 months, HR=0.80, P=0.012) and median OS by 1.2 months (15.9 months vs 14.7 months, HR=0.84, P=0.036), which did not meet the prespecified statistical significance threshold.11 The CheckMate 722 study also confirmed that nivolumab plus chemotherapy failed to significantly improve PFS in patients with EGFR-TKI resistance (5.6 months vs 5.4 months, HR=0.75, P=0.053).12 Subgroup analysis of the IMpower130 study also showed limited benefit of atezolizumab plus chemotherapy in patients with EGFR mutations.13 This is mainly because EGFR-mutant NSCLC is generally considered an “immunologically cold tumor” with a small number of infiltrating effector T cells and a high proportion of immunosuppressive cells in the tumor microenvironment, leading to poor efficacy of traditional ICIs.14,15 While PD-L1 expression is typically an unreliable predictor of ICI benefit in these “cold” tumors, whether novel PD-1/VEGF bispecific antibodies can overcome this limitation and whether patients with high PD-L1 expression might still derive distinct benefits remain important subjects of investigation.

The emergence of PD-1/VEGF bispecific antibodies provides a new idea to solve this dilemma. These drugs can simultaneously block both PD-1/PD-L1 and VEGF/VEGFR pathways. On the one hand, blocking the VEGF pathway improves tumor angiogenesis, reduces tumor interstitial pressure, and promotes immune cell infiltration; on the other hand, blocking the PD-1 pathway relieves immunosuppression and activates T cell-mediated anti-tumor immune responses, resulting in synergistic anti-tumor effects.16 Ivonescimab is the world’s first approved PD-1/VEGF bispecific antibody independently developed in China. Its unique tetravalent structure design enables it to bind PD-1 and VEGF with high affinity simultaneously, forming soluble complexes and enhancing the blocking effect on both targets.17 HARMONi-A is a global multicenter, randomized, double-blind, placebo-controlled phase III clinical study designed to evaluate the efficacy and safety of ivonescimab plus chemotherapy versus chemotherapy alone in patients with advanced EGFR-mutant NSCLC who have progressed after EGFR-TKI treatment. The results showed that the median PFS was significantly longer in the ivonescimab plus chemotherapy group than in the chemotherapy alone group (7.1 months vs 4.8 months, HR=0.46, P<0.001), and the median OS was also significantly prolonged (17.1 months vs 14.5 months, HR=0.75, P=0.04).17 The Chinese subgroup analysis results were consistent with the global overall results, further confirming the efficacy and safety of ivonescimab in the Chinese population. Based on these results, it has become one of the standard treatment options for patients with advanced EGFR-mutant NSCLC after EGFR-TKI resistance.

Despite the positive results of ivonescimab studies, several clinical questions remain unanswered.18,19 First, as ivonescimab entered clinical practice more recently than bevacizumab, potential era effects may shape retrospective comparisons and require careful consideration. However, the control group in HARMONi-A was chemotherapy alone, whereas bevacizumab plus chemotherapy is the standard second-line regimen in routine clinical practice, and no head-to-head real-world evidence comparing ivonescimab plus chemotherapy with bevacizumab plus chemotherapy currently exists. Second, the enrolled population in registration studies was strictly selected, excluding patients with complex conditions such as severe comorbidities and multiple lines of treatment failure, limiting the generalizability of the results to real-world clinical practice. In addition, the efficacy differences of ivonescimab in special populations such as different PD-L1 expression levels and patients with brain metastases remain unclear, lacking sufficient evidence-based medical support. While real-world comparative studies provide valuable insights into routine clinical practice that complement randomized trials, they possess inherent limitations, particularly selection bias. Furthermore, although propensity score matching (PSM) is employed to balance baseline characteristics, it can only adjust for measured covariates and cannot fully eliminate confounding from unobserved or unrecorded clinical variables. Therefore, this multicenter real-world study was conducted to evaluate the efficacy and safety of ivonescimab plus chemotherapy versus bevacizumab plus chemotherapy in patients with advanced EGFR-mutant lung adenocarcinoma who had progressed after EGFR-TKI treatment, and to explore prognostic factors affecting efficacy, providing medical evidence for the selection of treatment regimens in clinical practice.

Materials and MethodsStudy Design and Patients

This was a multicenter, retrospective, observational, propensity score-matched real-world cohort study that enrolled patients with advanced EGFR-mutant lung adenocarcinoma who received treatment at Tianjin Medical University Cancer Institute and Hospital and Cangzhou Hospital of Integrated Traditional Chinese and Western Medicine, Hebei Province, between January 1, 2020 and December 31, 2025. The study protocol was approved by the Institutional Review Board of Tianjin Medical University Cancer Institute and Hospital (approval no. EK20250091) and the Ethics Committee of Cangzhou Hospital of Integrated Traditional Chinese and Western Medicine (approval no. CZX2024180). The requirement for written informed consent was waived due to the retrospective nature of the study. All patient data were strictly maintained with confidentiality, and identifying information was anonymized prior to analysis.

Inclusion Criteria

1) Histopathologically or cytologically confirmed advanced (AJCC 8th edition TNM stage IV) lung adenocarcinoma; 2) Confirmed EGFR common sensitizing mutations (exon 19 deletion or exon 21 L858R point mutation) by next-generation sequencing (NGS); patients harboring rare or uncommon EGFR mutations (eg, exon 18 G719X, exon 20 insertion, exon 21 L861Q, S768I) were excluded; 3) Previously received at least one line of EGFR-TKI monotherapy and had disease progression as assessed by the investigator according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1; 4) The first-line systemic treatment after disease progression was either ivonescimab plus pemetrexed + platinum chemotherapy or bevacizumab plus pemetrexed + platinum chemotherapy; 5) No prior treatment with any anti-angiogenic agent (eg, bevacizumab, anlotinib, apatinib) or any immune checkpoint inhibitor (anti-PD-1/PD-L1 or anti-CTLA-4 antibody); 6) Eastern Cooperative Oncology Group (ECOG) performance status score of 0–1; 7) At least one measurable target lesion at baseline (longest diameter ≥10 mm, lymph node short diameter ≥15 mm according to RECIST 1.1 criteria); 8) Asymptomatic brain metastases were permitted, provided they were neurologically stable and not requiring escalation of corticosteroids; 9) No significant abnormalities in baseline blood routine, liver and kidney function, coagulation function, and no contraindications to chemotherapy, immunotherapy, or anti-angiogenic therapy; 10) Complete clinical data and traceable follow-up data.

Exclusion Criteria

1) Concomitant other driver gene mutations, including ALK, ROS1, RET, MET exon 14 skipping mutation, BRAF V600E, etc.; 2) Received antibody-drug conjugates (ADCs), EGFR-TKI rechallenge, or other systemic anti-tumor therapies as the first-line treatment after disease progression; 3) History of other primary malignant tumors; 4) Concomitant active autoimmune diseases requiring systemic immunosuppressive therapy; 5) Concomitant severe dysfunction of vital organs such as heart, lung, liver, and kidney, or active infections (such as active hepatitis B, hepatitis C, human immunodeficiency virus infection, etc).; 6) Symptomatic brain metastases or leptomeningeal disease requiring urgent local intervention or escalating doses of corticosteroids; 7) Follow-up time less than 3 months.

Case Screening Process and Baseline Data Collection

Baseline clinical data of all enrolled patients were retrospectively collected from the electronic medical record (EMR) systems of the two participating centers. A total of 457 patients with advanced lung adenocarcinoma who received ivonescimab plus chemotherapy or bevacizumab plus chemotherapy between January 2020 and December 2025 were initially retrieved. Screening was performed according to the inclusion and exclusion criteria: 28 patients with concomitant other driver gene mutations, 17 patients with previous immunotherapy or anti-angiogenic therapy, 12 patients with non-compliant second-line treatment regimens, 3 patients with other malignant tumors, 4 patients with symptomatic brain metastases, 3 patients with incomplete clinical data, and 0 patients with follow-up time less than 3 months were excluded. Finally, a total of 390 eligible patients were included, of whom 130 received ivonescimab plus chemotherapy (ivonescimab group) and 260 received bevacizumab plus chemotherapy (bevacizumab group). Variables extracted from the EMR system included age, gender, ECOG performance status score, smoking history, baseline brain metastasis status, EGFR sensitive mutation type, initial first-line EGFR-TKI treatment type, T790M mutation status, PD-L1 tumor proportion score (TPS) expression level, and platinum type used in this chemotherapy. The definitions of all baseline data were consistent between the two centers.

Treatment Regimens

Both groups received pemetrexed plus platinum-based doublet chemotherapy as the backbone, combined with either ivonescimab or bevacizumab. The chemotherapy backbone was identical between groups and consisted of pemetrexed 500 mg/m2 administered as a 10-minute intravenous infusion on day 1, plus either cisplatin 75 mg/m2 administered as a 2-hour intravenous infusion on day 1 or carboplatin AUC = 5 administered as a 30-minute intravenous infusion on day 1, repeated every 3 weeks (Q3W) for 4 to 6 induction cycles. The choice of platinum agent was determined by the investigator according to the patient’s age, performance status, renal function, and comorbidities. In the ivonescimab group, ivonescimab was administered at 20 mg/kg as an intravenous infusion on day 1 of each 3-week cycle (initial infusion duration 60 ± 10 minutes). In the bevacizumab group, bevacizumab was administered at 15 mg/kg as an intravenous infusion on day 1 of each 3-week cycle (initial infusion duration ≥ 90 minutes, shortened to 60 minutes for subsequent infusions if well tolerated). All patients received oral folic acid 400 μg daily starting 7 days before the first dose of pemetrexed, continuing until 21 days after the last dose of pemetrexed; vitamin B12 1000 μg was intramuscularly injected within 7 days before the first dose of pemetrexed and then repeated every 9 weeks until the end of treatment. Dexamethasone 4 mg was orally administered twice daily on the day before, the day of, and the day after pemetrexed administration to prevent pemetrexed-related rash and allergic reactions.

Maintenance Therapy

After completion of 4–6 induction cycles, patients without disease progression or unacceptable toxicity entered the maintenance phase. In the ivonescimab group, maintenance therapy consisted of ivonescimab 20 mg/kg intravenously plus pemetrexed 500 mg/m2 intravenously, both administered on day 1 every 3 weeks. In the bevacizumab group, maintenance therapy consisted of bevacizumab 15 mg/kg intravenously plus pemetrexed 500 mg/m2 intravenously, both administered on day 1 every 3 weeks. Maintenance was continued until disease progression, unacceptable toxicity, or loss to follow-up.

Comprehensive efficacy evaluations were performed every 2 cycles (every 6 weeks) during treatment, including chest enhanced CT, abdominal enhanced CT or ultrasound, head enhanced MRI, and whole-body bone scan. Efficacy was assessed according to RECIST version 1.1.20 If disease progression confirmed by the investigator or intolerable treatment-related toxicity occurred during treatment, the current treatment regimen was immediately terminated and replaced with other subsequent anti-tumor therapies. The grading of adverse events was performed according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0 (NCI-CTCAE 5.0),21 and the management of immune-related adverse events followed international guidelines.22–24 For grade 1–2 adverse events, treatment could be continued at the original dose under close monitoring with symptomatic supportive treatment; for grade 3 non-hematological toxicity or grade 4 hematological toxicity, treatment was suspended and active symptomatic treatment was given. After the adverse event recovered to ≤grade 1, treatment was continued at a reduced dose level. If the same adverse event recurred after 2 dose adjustments, or grade 4 non-hematological toxicity, grade ≥3 immune-related adverse events, or treatment-related interstitial lung disease occurred, treatment was permanently discontinued.

Study Endpoints and Assessments

The primary endpoint of this study was PFS, defined as the time from the first administration of the treatment regimen to the first occurrence of disease progression confirmed by the investigator according to RECIST 1.1 or death from any cause, whichever occurred first. Secondary endpoints included OS, ORR, disease control rate (DCR), and treatment-related adverse events (TRAE). Efficacy was evaluated using the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. Tumor responses were classified as follows: complete response (CR), defined as disappearance of all target lesions, resolution of all non-target lesions, and normalization of tumor markers; partial response (PR), defined as at least a 30% decrease in the sum of the longest diameters (SLD) of target lesions relative to baseline, with no progression of non-target lesions and no new lesions; stable disease (SD), defined as neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD, with no progression of non-target lesions and no new lesions; progressive disease (PD), defined as at least a 20% increase in the SLD of target lesions relative to the smallest SLD recorded during treatment, with an absolute increase of at least 5 mm, or the appearance of any new lesion, or unequivocal progression of non-target lesions. ORR was defined as the proportion of patients who achieved CR or PR after treatment among all evaluable patients; DCR was defined as the proportion of patients who achieved CR, PR, or SD after treatment among all evaluable patients. Efficacy evaluations were performed every 2 treatment cycles (every 6 weeks) after the start of treatment until disease progression, death, or loss to follow-up; for patients with clinically suspected disease progression, imaging evaluations could be performed in advance. All imaging assessments were retrospectively reviewed by two independent radiologists at each participating center who were blinded to treatment allocation and clinical outcomes (blinded independent review). Discrepancies between the two reviewers were resolved by consensus discussion, and unresolved cases were adjudicated by a third senior radiologist. OS was defined as the time from the first administration of the treatment regimen to death from any cause. For patients who were still alive or lost to follow-up at the last follow-up, both PFS and OS were censored at the date of the last follow-up. Safety evaluation was conducted from the first administration of the treatment regimen until 30 days after the last treatment. All observed adverse events were recorded in detail regarding their occurrence time, severity, duration, management measures, and outcomes. Severity grading was performed according to NCI-CTCAE 5.0. TRAE was defined as adverse events with a causal relationship or uncertain causal relationship with the study drugs; immune-related adverse events (irAE) were defined as special types of adverse events related to the mechanism of action of immune checkpoint inhibitors; anti-angiogenic-related adverse events were defined as special types of adverse events related to the mechanism of action of anti-angiogenic drugs, including hypertension, proteinuria, bleeding events, etc. All serious adverse events were required to be reported to the ethics committee of each center within 24 hours.

Follow-Up

All enrolled patients were followed up regularly through a combination of the electronic medical record (EMR) system and telephone follow-up. During treatment, follow-up was performed every 2 treatment cycles (every 6 weeks), including physical examination, laboratory tests, and tumor imaging evaluation (chest enhanced CT, abdominal enhanced CT or ultrasound, head enhanced MRI, whole-body bone scan, etc). After treatment completion, follow-up was performed every 3 months for the first 2 years and every 6 months thereafter until patient death, loss to follow-up, or study cutoff. The follow-up content mainly included the patient’s survival status, disease progression, subsequent anti-tumor treatment regimens, and occurrence of adverse events. The follow-up cutoff date for this study was March 31, 2026. For patients who were still alive at the last follow-up, the date of the last follow-up was used as the censoring time; for patients lost to follow-up, defined as being unable to be contacted for 3 consecutive follow-ups, the date of the last confirmed survival was used as the censoring time.

Statistical Analysis

All statistical analyses were performed using SPSS 26.0 statistical software (IBM Corp., Armonk, NY, USA) and R 4.3.1 statistical software (R Foundation for Statistical Computing, Vienna, Austria). Two-sided tests were used, and P<0.05 was considered statistically significant. A 1:1 propensity score matching (PSM) was performed to balance baseline confounding factors between the two groups. Matching variables included age, gender, ECOG performance status score, smoking history, baseline brain metastasis status, PD-L1 TPS expression level, EGFR sensitive mutation type, initial first-line EGFR-TKI treatment type, T790M mutation status, and platinum type used in this chemotherapy. The nearest neighbor matching method was used with a caliper value of 0.05. Matching quality was assessed using the standardized mean difference (SMD), and SMD<0.1 was considered to indicate good balance of baseline characteristics. For baseline characteristics comparison, categorical variables were expressed as number (percentage), and intergroup comparisons were performed using the χ2-test or Fisher’s exact test; continuous variables were expressed as mean±standard deviation or median (interquartile range), and intergroup comparisons were performed using the independent samples t-test or Mann–Whitney U-test. For efficacy analysis, ORR and DCR were expressed as number (percentage), and intergroup comparisons were performed using the χ2-test. Survival analysis was performed using the Kaplan-Meier method to draw survival curves, and intergroup differences were tested using the Log rank test; the Cox proportional hazards regression model was used to calculate the hazard ratio (HR) and its 95% confidence interval (CI). Preset subgroup analyses included age, gender, ECOG performance status score, smoking history, baseline brain metastasis status, EGFR sensitive mutation type, initial first-line EGFR-TKI treatment type, T790M mutation status, and PD-L1 TPS expression level. Given the exploratory nature of these subgroup analyses, no adjustments for multiplicity were made, and the results should be interpreted with caution due to the increased risk of false-positive findings. Interaction tests were used to assess the interaction between treatment regimens and subgroup factors. Prognostic factor analysis was performed using univariate and multivariate Cox proportional hazards regression models, and variables with P<0.1 in univariate analysis were included in the multivariate model.

ResultsBaseline Characteristics and Follow-Up

A total of 390 patients with advanced EGFR-mutant lung adenocarcinoma who had progressed after EGFR-TKI treatment were included in this study, of whom 130 were in the ivonescimab plus chemotherapy group and 260 were in the bevacizumab plus chemotherapy group. The median follow-up time for the overall cohort was 25.5 months (interquartile range: 12.5–31.6 months). After PSM, 120 patients were included in each group, and the SMD of all baseline characteristics was <0.1, indicating well-balanced baseline characteristics between the two groups (Table 1). All subsequent efficacy and safety analyses are reported primarily for the post-PSM cohort, with pre-matching results provided for reference.

Table 1 Comparison of Baseline Characteristics Between the Ivonescimab Group and Bevacizumab Group Before and After Propensity Score Matching

Objective Efficacy

In the post-matching cohort, the ORR was 45.0% (54/120) in the ivonescimab plus chemotherapy group, significantly higher than 28.3% (34/120) in the bevacizumab plus chemotherapy group (χ2=7.177, P=0.007); the DCR was 91.7% (110/120) in the ivonescimab group, significantly higher than 81.7% (98/120) in the bevacizumab group (χ2=5.192, P=0.023). Results in the pre-matching cohort were directionally consistent (ORR 44.6% vs 27.7%, χ2=11.169, P=0.001; DCR 91.5% vs 81.9%, χ2=6.329, P=0.012) (Table 2).

Table 2 Comparison of Objective Tumor Response Efficacy Between the Two Groups Before and After Propensity Score Matching

Safety

There were no statistically significant differences in the incidence of total TRAE and grade ≥3 TRAE between the two groups (P>0.05). The incidence of all-grade TRAE was 93.3% in the ivonescimab plus chemotherapy group, with a grade ≥3 TRAE incidence of 25.8%; the incidence of all-grade TRAE was 91.7% in the bevacizumab plus chemotherapy group, with a grade ≥3 TRAE incidence of 28.3% (Table 3). In terms of hematological toxicity, there were no statistically significant differences in the incidence of neutropenia, leukopenia, anemia, and thrombocytopenia between the two groups (P>0.05), and the incidence of grade ≥3 hematological toxicity was low in both groups. Chemotherapy-related non-hematological toxicities were mainly nausea, vomiting, fatigue, anorexia, and elevated transaminases, mostly grade 1–2, with no statistically significant differences in incidence between the two groups (P>0.05). Anti-angiogenic-related specific toxicities included hypertension, proteinuria, and bleeding events, with no statistically significant differences in incidence between the two groups (P>0.05). A total of 38 cases (31.7%) of irAE occurred in the ivonescimab plus chemotherapy group, including 3 cases (2.5%) of grade ≥3 irAE, which were 1 case of thyroid dysfunction and 2 cases of immune pneumonitis; all irAE were controlled after glucocorticoid treatment, and no treatment-related deaths occurred. No immune-related adverse events were observed in the bevacizumab plus chemotherapy group.

Table 3 Comparison of Treatment-Related Adverse Events Between the Two Groups in the Post-Matching Cohort

Survival AnalysisOverall Survival Outcomes

In the pre-matching cohort, the median PFS was significantly longer in the ivonescimab plus chemotherapy group than in the bevacizumab plus chemotherapy group (mPFS 7.7 vs 5.4 months, HR=0.698, 95% CI: 0.569–0.856, log-rank χ2=11.52, P=0.001), and the median OS was also significantly prolonged (mOS 18.3 vs 14.9 months, HR=0.704, 95% CI: 0.574–0.861, log-rank χ2=11.54, P=0.001) (Figure S1A and S1B). Consistent results were observed in the post-matching cohort (mPFS 7.6 months vs 5.9 months; HR=0.770, 95% CI: 0.595–0.997; log-rank χ2=4.134, P=0.042; mOS 18.0 months vs 15.1 months; HR=0.722, 95% CI: 0.557–0.938; log-rank χ2=6.593, P=0.010) (Figure S1C and S1D).

Survival Analysis Stratified by PD-L1 Expression Level

Stratified analysis by PD-L1 TPS expression level showed that in patients with PD-L1 TPS<1%, the median PFS was 7.2 months in the ivonescimab plus chemotherapy group, 1.1 months longer than 6.1 months in the bevacizumab plus chemotherapy group, with no statistically significant difference (HR=0.716, 95% CI: 0.463–1.109, log-rank χ2=0.032, P=0.857); the median OS was 17.8 months, 3.0 months longer than 14.8 months in the control group, with no statistically significant difference (HR=0.957, 95% CI: 0.667–1.373, log-rank χ2=0.060, P=0.807) (Figure S2A and S2B). In patients with 1%≤PD-L1 TPS<50%, the median PFS was 7.5 months in the ivonescimab plus chemotherapy group, 1.8 months longer than 5.7 months in the bevacizumab plus chemotherapy group, with no statistically significant difference (HR=0.716, 95% CI: 0.463–1.109, log-rank χ2=2.439, P=0.118); the median OS was 18.0 months, 3.0 months longer than 15.0 months in the control group, with no statistically significant difference (HR=0.717, 95% CI: 0.464–1.108, log-rank χ2=2.481, P=0.115) (Figure S2C and S2D). In patients with PD-L1 TPS≥50%, the median PFS was not reached (NR) in the ivonescimab plus chemotherapy group, significantly longer than 5.6 months in the bevacizumab plus chemotherapy group (HR=0.388, 95% CI: 0.186–0.809, log-rank χ2=7.891, P=0.005); the median OS was 20.7 months, significantly longer than 14.8 months in the control group (HR=0.300, 95% CI: 0.138–0.650, log-rank χ2=16.28, P<0.001) (Figure S2E and S2F). Further analysis of the prognostic value of PD-L1 expression level in the ivonescimab plus chemotherapy cohort showed that there was an overall statistically significant difference in PFS among the three groups (log-rank χ2=9.746, P=0.008) (Figure S3A). The OS of the three groups showed a gradient prolongation trend, and the overall Log rank test showed a borderline benefit trend (χ2=5.964, P=0.051), which did not reach statistical significance (Figure S3B). The median PFS of patients with PD-L1 TPS≥50% was significantly longer than that of patients with 1%≤TPS<50% (P=0.024) and patients with TPS<1% (P=0.002) (Figure S3C and S3D); while there was no statistically significant difference in PFS between patients with TPS<1% and patients with 1%≤TPS<50% (P=0.322) (Figure S3E).

Survival Analysis Stratified by Baseline Brain Metastasis Status

Stratified analysis by baseline brain metastasis status showed that in patients with baseline brain metastases, the median PFS was 7.0 months in the ivonescimab plus chemotherapy group, 1.8 months longer than 5.2 months in the bevacizumab plus chemotherapy group, with no statistically significant difference (HR=0.807, 95% CI: 0.534–1.219, log-rank χ2=1.098, P=0.295); the median OS was 17.4 months, 2.7 months longer than 14.7 months in the control group, with no statistically significant difference (HR=0.762, 95% CI: 0.504–1.150, log-rank χ2=1.846, P=0.174) (Figure S4A and S4B). In patients without baseline brain metastases, the median PFS was 8.5 months in the ivonescimab plus chemotherapy group, 1.8 months longer than 6.7 months in the bevacizumab plus chemotherapy group, with no statistically significant difference (HR=0.758, 95% CI: 0.544–1.056, log-rank χ2=2.850, P=0.091); the median OS was 19.9 months, significantly longer than 15.8 months in the control group (HR=0.682, 95% CI: 0.489–0.952, log-rank χ2=5.591, P=0.018) (Figure S4C and S4D). Further analysis of the prognostic value of baseline brain metastasis status in the ivonescimab plus chemotherapy cohort showed that the median PFS of patients with baseline brain metastases was 7.0 months, significantly shorter than 8.5 months in patients without baseline brain metastases (HR=1.500, 95% CI: 1.006–2.235, log-rank χ2=4.704, P=0.030); the median OS was 17.4 months, significantly shorter than 19.9 months in patients without baseline brain metastases (HR=1.499, 95% CI: 1.011–2.223, log-rank χ2=4.843, P=0.028) (Figure S5A and S5B).

Preset Subgroup Analysis

Preset subgroup analysis showed that in all evaluated subgroups, a consistent benefit trend of ivonescimab plus chemotherapy in reducing the risk of disease progression was observed, and no significant interaction between treatment regimens and subgroup factors was found (all interaction P>0.05), indicating no statistically significant heterogeneity in treatment effect across these subgroups. Although patients with PD-L1 TPS≥50% exhibited a numerically more pronounced PFS benefit from ivonescimab plus chemotherapy (HR=0.39, 95% CI: 0.19–0.81), the absence of a significant interaction, the small sample size, and the lack of multiplicity adjustment mean this finding must be viewed strictly as hypothesis-generating and requires further verification in larger prospective studies (Figure S6).

Prognostic Factor AnalysisPrognostic Factor Analysis for PFS

Univariate Cox regression analysis showed that ECOG performance status score 1 (HR=1.749, 95% CI: 1.199–2.551, P=0.004), baseline brain metastasis (HR=1.557, 95% CI: 1.113–2.179, P=0.008), T790M mutation positive (HR=1.387, 95% CI: 0.958–2.007, P=0.087), PD-L1 TPS≥50% (HR=0.523, 95% CI: 0.315–0.869, P=0.012), and ivonescimab plus chemotherapy (HR=0.701, 95% CI: 0.499–0.984, P=0.040) were associated with PFS. Variables with P<0.1 in univariate analysis were included in the multivariate Cox regression model. The results showed that ECOG performance status score 1 (HR=1.578, 95% CI: 1.075–2.317, P=0.020) and baseline brain metastasis (HR=1.423, 95% CI: 1.010–2.006, P=0.044) were independent poor prognostic factors for PFS; PD-L1 TPS≥50% (HR=0.479, 95% CI: 0.288–0.798, P=0.005) and ivonescimab plus chemotherapy (HR=0.685, 95% CI: 0.486–0.967, P=0.031) were independent protective prognostic factors for PFS (Table 4).

Table 4 Univariate and Multivariate Cox Regression Analysis of Progression-Free Survival in the Post-Matching Cohort

Prognostic Factor Analysis for OS

Univariate Cox regression analysis showed that ECOG performance status score 1 (HR=1.825, 95% CI: 1.249–2.673, P=0.002), baseline brain metastasis (HR=1.592, 95% CI: 1.138–2.231, P=0.007), PD-L1 TPS≥50% (HR=0.489, 95% CI: 0.295–0.811, P=0.006), and ivonescimab plus chemotherapy (HR=0.715, 95% CI: 0.543–0.940, P=0.016) were associated with OS. Multivariate Cox regression analysis showed that ECOG performance status score 1 (HR=1.692, 95% CI: 1.150–2.490, P=0.008) and baseline brain metastasis (HR=1.476, 95% CI: 1.053–2.069, P=0.023) were independent poor prognostic factors for OS; PD-L1 TPS≥50% (HR=0.452, 95% CI: 0.272–0.752, P=0.002) and ivonescimab plus chemotherapy (HR=0.698, 95% CI: 0.528–0.922, P=0.011) were independent protective prognostic factors for OS (Table 5).

Table 5 Univariate and Multivariate Cox Regression Analysis of Overall Survival in the Post-Matching Cohort

Discussion

To our knowledge, this is the first real-world, multicenter, head-to-head comparative study evaluating ivonescimab plus chemotherapy versus bevacizumab plus chemotherapy in patients with advanced EGFR-mutant lung adenocarcinoma who progressed after EGFR-TKI treatment. After balancing baseline confounding factors with 1:1 propensity score matching, the median PFS and OS were significantly longer in the ivonescimab plus chemotherapy group than in the bevacizumab plus chemotherapy group, with comparable safety between the two groups. Because the comparator arm in our study is the current real-world standard of care—rather than chemotherapy alone as in the registration HARMONi-A trial—our findings provide a level of clinical evidence that is more directly relevant to routine practice than conventional retrospective cohorts, and they carry direct implications for treatment selection following the reimbursement of ivonescimab in China and for future updates of clinical practice guidelines.

The HARMONi-A study was the first global multicenter phase III registration study to confirm the significant survival benefit of ivonescimab plus chemotherapy versus chemotherapy alone in patients with advanced NSCLC with EGFR-TKI resistance, with median PFS of 7.1 months and 4.8 months (HR=0.46, P<0.001) and median OS of 17.1 months and 14.5 months (HR=0.75, P=0.04), respectively.17 Based on these results, ivonescimab has become one of the standard treatment options recommended by domestic and international guidelines for patients after EGFR-TKI resistance. However, the control group of the HARMONi-A study was chemotherapy alone, while bevacizumab plus chemotherapy is the more commonly used standard second-line treatment regimen in clinical practice, and head-to-head comparison data between ivonescimab plus chemotherapy and this standard regimen are lacking.25 The present study addresses this critical evidence gap by directly comparing the two regimens in a real-world setting. The results showed that the median PFS was prolonged by 1.7 months in the ivonescimab plus chemotherapy group compared with the bevacizumab plus chemotherapy group (7.6 months vs 5.9 months, HR=0.77, P=0.042), and the median OS was prolonged by 2.9 months (18.0 months vs 15.1 months, HR=0.72, P=0.010). Notably, the efficacy of the bevacizumab plus chemotherapy group in this study (mPFS 5.9 months, mOS 15.1 months) was superior to that of the chemotherapy alone group in the HARMONi-A study (mPFS 4.8 months, mOS 14.5 months), further confirming the standard treatment status of bevacizumab plus chemotherapy in this population. On this basis, ivonescimab plus chemotherapy still brought survival benefits, suggesting that PD-1/VEGF bispecific antibody plus chemotherapy has better anti-tumor activity than traditional anti-angiogenic drugs plus chemotherapy. Given that ivonescimab was recently included in the Chinese National Reimbursement Drug List, these head-to-head data provide timely and clinically actionable evidence to inform regimen selection in daily practice and to support its preferred positioning in future guideline recommendations for EGFR-TKI-resistant NSCLC.

The ORIENT-31 study is another important phase III clinical study that confirmed the significant benefit of sintilimab plus bevacizumab and chemotherapy versus chemotherapy alone in patients with EGFR-TKI resistance, with median PFS of 7.2 months and 4.3 months (HR=0.46, P<0.0001) and median OS of 21.1 months and 19.2 months (HR=0.79, P=0.059), respectively.26 This regimen has also become one of the standard treatment options after EGFR-TKI resistance. The efficacy of ivonescimab plus chemotherapy in this study was comparable to that of sintilimab + bevacizumab + chemotherapy in the ORIENT-31 study (mPFS 7.6 months vs 7.2 months). While cross-trial comparisons must be interpreted with extreme caution due to inherent differences in study populations and unmeasured confounders, as a PD-1/VEGF bispecific antibody, ivonescimab only requires a single infusion to simultaneously block both targets, which is more convenient to administer, has higher patient compliance, and may reduce the risk of infusion-related adverse reactions compared with the combination of two monoclonal antibodies. In addition, the incidence of grade ≥3 treatment-related adverse events in the ivonescimab plus chemotherapy group in this study was 25.8%, which was lower than that of the four-drug combination regimen in the ORIENT-31 study (37.9%), suggesting that bispecific antibody plus chemotherapy may have a better safety profile. In sharp contrast, multiple large phase III clinical studies have confirmed that traditional PD-1/PD-L1 monoclonal antibodies plus chemotherapy fail to bring significant clinical benefits to patients with EGFR-TKI resistance.11–13 The KEYNOTE-789 study showed that pembrolizumab plus chemotherapy compared with chemotherapy alone only prolonged median PFS by 0.1 months (5.6 months vs 5.5 months, HR=0.80, P=0.012) and median OS by 1.2 months (15.9 months vs 14.7 months, HR=0.84, P=0.036), neither of which met the prespecified statistical significance thresholds. The CheckMate 722 study also confirmed that nivolumab plus chemotherapy failed to significantly improve PFS in this population (5.6 months vs 5.4 months, HR=0.75, P=0.053). These results suggest that combining immune checkpoint inhibitors with chemotherapy alone is insufficient to overcome resistance in EGFR-mutant NSCLC, and simultaneous combination with anti-angiogenic drugs is the key to improving the efficacy of immunotherapy. The mechanism may be that anti-angiogenic drugs not only inhibit tumor angiogenesis but also improve the tumor immune microenvironment, reduce tumor interstitial pressure, promote immune cell infiltration, and produce synergistic anti-tumor effects with immune checkpoint inhibitors.27 PD-1/VEGF bispecific antibodies can simultaneously target both pathways and have a unique tetravalent structure design that can form soluble complexes, enhance the blocking effect on both targets, and further amplify the synergistic effect.28

The preset subgroup analysis of this study showed that in all evaluated subgroups, a consistent benefit trend of ivonescimab plus chemotherapy in reducing the risk of disease progression was observed, and no significant interaction between treatment regimens and subgroup factors was found, suggesting that the benefit of ivonescimab may have broad population applicability. However, given the limited statistical power of these analyses, statements regarding broad applicability across all patient subgroups should be interpreted cautiously. Notably, patients with PD-L1 TPS≥50% exhibited a numerically pronounced survival advantage from ivonescimab plus chemotherapy, with a median PFS of not reached (NR) in the ivonescimab plus chemotherapy group, which was significantly longer than the 5.6 months in the bevacizumab plus chemotherapy group; the median OS was 20.7 months, 5.9 months longer than that in the control group. However, it is crucial to recognize that this subgroup consisted of a very small sample size (only 18 patients per arm) and was not supported by a significant interaction test, making the resulting hazard ratio estimates statistically unstable Furthermore, this observation directly contradicts the typical “cold-tumor” immunophenotype of EGFR-mutant disease discussed earlier, where high PD-L1 expression has not reliably predicted benefit from traditional immune checkpoint inhibitors. The exact mechanism by which a PD-1/VEGF bispecific agent might overcome this intrinsic resistance in high PD-L1 expressors remains unclear, and this tension highlights that our finding must be interpreted strictly as hypothesis-generating rather than an established biomarker effect.29 Further analysis within the ivonescimab group showed that the PFS and OS of patients with PD-L1 TPS≥50% were significantly longer than those of patients with low expression and negative expression, suggesting that for patients with EGFR-TKI resistance and high PD-L1 expression, ivonescimab plus chemotherapy may be the optimal treatment option. Baseline brain metastasis is a common poor prognostic factor in patients with advanced EGFR-mutant NSCLC. The proportion of patients with baseline brain metastases in this study was 38.3%, which was consistent with the proportion in real-world clinical practice. The results showed that in patients with baseline brain metastases, the median PFS was prolonged by 1.8 months (7.0 months vs 5.2 months) and the median OS was prolonged by 2.7 months (17.4 months vs 14.7 months) in the ivonescimab plus chemotherapy group compared with the bevacizumab plus chemotherapy group, showing a benefit trend although not reaching statistical significance. This result may be related to the small sample size and insufficient statistical power of this subgroup. As a PD-1/VEGF bispecific antibody, ivonescimab’s anti-angiogenic effect can improve blood-brain barrier permeability, promote drug entry into intracranial lesions, and its immune activation effect can eliminate intracranial tumor cells,30 suggesting its potential application value in patients with brain metastases, which needs further verification in larger prospective studies. However, it is important to note that our subgroup analyses are constrained by small sample sizes and are statistically underpowered. Therefore, these subgroup findings should be interpreted as exploratory rather than confirmatory, and they require rigorous validation in larger, adequately powered prospective trials. The multivariate Cox regression analysis of this study showed that ECOG performance status score 1 and baseline brain metastasis were independent poor prognostic factors for both PFS and OS, while PD-L1 TPS≥50% and ivonescimab plus chemotherapy were independent protective prognostic factors. This result is consistent with the conclusions of multiple previous studies,31 further verifying the reliability of the results of this study.

This study has certain limitations. First, this was a retrospective study, and although propensity score matching was used to balance baseline confounding factors, it inherently cannot adjust for unobserved variables, meaning residual confounding remains unavoidable. Second, this was a two-center study with a relatively small sample size, especially the limited sample size for subgroup analysis, which may lead to insufficient statistical power. Consequently, all subgroup analyses must be regarded as strictly exploratory. Furthermore, the findings currently lack external validation in independent cohorts, and thus these results need further verification in larger prospective studies. Third, imaging response was assessed by two independent blinded radiologists at each participating site rather than by a centralized independent review committee, which may introduce a degree of inter-site variability; future prospective studies with central imaging review are warranted. Fourth, although the OS benefit observed in our cohort reached statistical significance, the OS data should be regarded as immature: a substantial proportion of patients in both groups remained alive and on follow-up at the data cutoff, the planned number of death events had not been fully reached, and longer follow-up is therefore needed to confirm the magnitude and durability of the OS advantage. Additionally, the patient enrollment period spanned from 2020 to 2025. Because ivonescimab became available more recently than bevacizumab, the two treatment arms differ systematically in calendar era. This era effect could influence the availability of subsequent-line therapies and introduce biases that potentially impact the overall survival analysis. Moreover, because ivonescimab only became available recently, the two treatment arms differ systematically in calendar era. This era confounding likely inflates the observed OS benefit, as patients in the contemporary ivonescimab arm had access to different and potentially more advanced subsequent-line options compared to those treated in earlier years within the bevacizumab arm. Given this confounding and the immaturity of the data, the overall survival findings should be viewed as hypothesis-generating. Moreover, because ivonescimab only became available recently, the two treatment arms differ systematically in calendar era. This era confounding likely inflates the observed OS benefit, as patients in the contemporary ivonescimab arm had access to different subsequent-line options compared to those treated in earlier years within the bevacizumab arm. Another crucial limitation is the sequencing safety concern: utilizing a PD-1-containing regimen introduces documented risks of severe immune-related adverse events, such as interstitial lung disease (ILD) or pneumonitis, particularly if these patients subsequently return to EGFR-TKI rechallenge or receive antibody-drug conjugates (ADCs) in later lines. This safety consideration must be carefully weighed in clinical practice. Finally, as a real-world study, heterogeneity in treatment sequencing and subsequent therapies after disease progression cannot be entirely controlled for, which may also influence the OS analysis.

Conclusion

In summary, the results of this study indicate that compared with bevacizumab plus chemotherapy, ivonescimab plus chemotherapy significantly prolongs PFS in patients with advanced EGFR-mutant lung adenocarcinoma who have progressed after EGFR-TKI treatment. Overall survival serves as a supportive but still-maturing signal. While safety is generally manageable, the added risk of immune-related toxicities relevant to subsequent TKI re-challenge must be carefully considered. Patients with PD-L1 TPS≥50% derive more significant benefits, although this finding remains strictly exploratory given the small sample size and the lack of a significant interaction test. Overall, this retrospective study demonstrates an association between ivonescimab plus chemotherapy and improved clinical outcomes compared to bevacizumab plus chemotherapy. This study provides important real-world evidence for the clinical application of ivonescimab in this population and supports ivonescimab plus chemotherapy as one of the potential treatment options for patients with advanced EGFR-mutant lung adenocarcinoma after EGFR-TKI resistance; however, these findings cannot establish definitive superiority and require confirmation in large, prospective randomized studies before influencing routine clinical practice.

Abbreviations

ADC, Antibody-drug conjugate; AJCC, American Joint Committee on Cancer; CI, Confidence interval; CR, Complete response; CTCAE, Common Terminology Criteria for Adverse Events; DCR, Disease control rate; ECOG, Eastern Cooperative Oncology Group; EGFR, Epidermal growth factor receptor; EMR, Electronic medical record; HR, Hazard ratio; ICI, Immune checkpoint inhibitor; irAE, Immune-related adverse event; NCI, National Cancer Institute; NGS, Next-generation sequencing; NSCLC, Non-small cell lung cancer; ORR, Objective response rate; OS, Overall survival; PD, Progressive disease; PFS, Progression-free survival; PSM, Propensity score matching; PR, Partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SD, Stable disease; SMD, Standardized mean difference; TKI, Tyrosine kinase inhibitor; TRAE, Treatment-related adverse event; TPS, Tumor proportion score; VEGF, Vascular endothelial growth factor.

Data Sharing Statement

The datasets generated and analyzed during the current study are not publicly available due to patient privacy and ethical restrictions, but are available from the corresponding author (Peng Chen) on reasonable request.

Ethics Approval and Consent to Participate

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Tianjin Medical University Cancer Institute and Hospital (approval no. EK20250091) and the Ethics Committee of Cangzhou Hospital of Integrated Traditional Chinese and Western Medicine (approval no. CZX2024180). The requirement for written informed consent was waived due to the retrospective nature of the study.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This work was supported by the Self-funded Project under the Key Research and Development Program of Cangzhou City (grant number 23244102170).

Disclosure

The authors declare no conflicts of interest.

References

1. Thai AA, Solomon BJ, Sequist LV, Gainor JF, Heist RS. Lung cancer. Lancet. 2021;398:535–17. doi:10.1016/S0140-6736(21)00312-3

2. Yang JC, Lu S, Hayashi H, et al. Overall survival with amivantamab-lazertinib in EGFR-mutated advanced NSCLC. N Engl J Med. 2025;393:1681–1693. doi:10.1056/NEJMoa2503001

3. Oxnard GR, Chen R, Pharr JC, et al. Germline EGFR mutations and familial lung cancer. J Clin Oncol. 2023;41:5274–5284. doi:10.1200/JCO.23.01372

4. Jänne PA, Planchard D, Kobayashi K, et al. Survival with osimertinib plus chemotherapy in EGFR-mutated advanced NSCLC. N Engl J Med. 2026;394:27–38. doi:10.1056/NEJMoa2510308

5. Lu S, Kato T, Dong X, et al. Osimertinib after chemoradiotherapy in stage III EGFR-mutated NSCLC. N Engl J Med. 2024;391:585–597. doi:10.1056/NEJMoa2402614

6. Passaro A, Wang J, Wang Y, et al. Amivantamab plus chemotherapy with and without lazertinib in EGFR-mutant advanced NSCLC after disease progression on osimertinib: primary results from the phase III MARIPOSA-2 study. Ann Oncol. 2024;35:77–90. doi:10.1016/j.annonc.2023.10.117

7. Fang W, Wu L, Meng X, et al. Sacituzumab tirumotecan in EGFR-TKI-resistant, EGFR-mutated advanced NSCLC. N Engl J Med. 2026;394:13–26. doi:10.1056/NEJMoa2512071

8. Girard N. New strategies and novel combinations in EGFR TKI-resistant non-small cell lung cancer. Curr Treat Options Oncol. 2022;23:1626–1644. doi:10.1007/s11864-022-01022-7

9. Park S, Kim TM, Han JY, et al. Phase III, randomized study of atezolizumab plus bevacizumab and chemotherapy in patients with EGFR- or ALK-mutated non-small-cell lung cancer (ATTLAS, KCSG-LU19-04). J Clin Oncol. 2024;42:1241–1251. doi:10.1200/JCO.23.01891

10. Hong S, Yu N, Cho JY, et al. VEGF signal complexity confers resistance to atezolizumab, bevacizumab, carboplatin, and paclitaxel in EGFR-tyrosine kinase inhibitor-resistant non-small cell lung cancer. MedComm. 2025;6:e70335. doi:10.1002/mco2.70335

11. Yang JC, Lee DH, Lee JS, et al. Phase III KEYNOTE-789 study of pemetrexed and platinum with or without pembrolizumab for tyrosine kinase inhibitor‒resistant, EGFR-mutant, metastatic nonsquamous non-small cell lung cancer. J Clin Oncol. 2024;42:4029–4039. doi:10.1200/JCO.23.02747

12. Mok T, Nakagawa K, Park K, et al. Nivolumab plus chemotherapy in epidermal growth factor receptor-mutated metastatic non-small-cell lung cancer after disease progression on epidermal growth factor receptor tyrosine kinase inhibitors: final results of CheckMate 722. J Clin Oncol. 2024;42:1252–1264. doi:10.1200/JCO.23.01017

13. West H, McCleod M, Hussein M, et al. Atezolizumab in combination with carboplatin plus nab-paclitaxel chemotherapy compared with chemotherapy alone as first-line treatment for metastatic non-squamous non-small-cell lung cancer (IMpower130): a multicentre, randomised, open-label, Phase 3 trial. Lancet Oncol. 2019;20:924–937. doi:10.1016/S1470-2045(19)30167-6

14. Rahal Z, El Darzi R, Moghaddam SJ, Cascone T, Kadara H. Tumour and microenvironment crosstalk in NSCLC progression and response to therapy. Nat Rev Clin Oncol. 2025;22:463–482. doi:10.1038/s41571-025-01021-1

15. Peng S, Wang R, Zhang X, et al. EGFR-TKI resistance promotes immune escape in lung cancer via increased PD-L1 expression. Mol Cancer. 2019;18:165. doi:10.1186/s12943-019-1073-4

16. Zhao Y, Chen G, Chen J, et al. AK112, a novel PD-1/VEGF bispecific antibody, in combination with chemotherapy in patients with advanced non-small cell lung cancer (NSCLC): an open-label, multicenter, phase II trial. EClinicalMedicine. 2023;62:102106. doi:10.1016/j.eclinm.2023.102106

17. Fang W, Zhao Y, Luo Y, et al. Ivonescimab plus chemotherapy in non-small cell lung cancer with EGFR variant: a randomized clinical trial. Jama. 2024;332:561–570. doi:10.1001/jama.2024.10613

18. Zhao J, Xu W, Zhou F, et al. Navigating the landscape of EGFR TKI resistance in EGFR-mutant NSCLC - mechanisms and evolving treatment approaches. Nat Rev Clin Oncol. 2026;23:63–83. doi:10.1038/s41571-025-01085-z

19. Ma H, Li L, Jiao C, et al. Immunotherapy after EGFR-TKI treatment in advanced non-small cell lung cancer: current status and future perspectives. Oncol Rep. 2026;55:44. doi:10.3892/or.2026.9049

20. Schwartz LH, Litière S, de Vries E, et al. RECIST 1.1-Update and clarification: from the RECIST committee. Eur J Cancer. 2016;62:132–137. doi:10.1016/j.ejca.2016.03.081

21. United States Department of Health and Human Services. Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. (2017). Available from: https://ctep.cancer.gov/protocoldevelopment/electronic_applications/docs/CTCAE_v5_Quick_Reference_8.5x11.pdf. Accessed August31, 2026.

22. Thompson JA, Schneider BJ, Brahmer J, et al. NCCN guidelines® insights: management of immunotherapy-related toxicities, version 2.2024. J Natl Compr Canc Netw. 2024;22:582–592. doi:10.6004/jnccn.2024.0057

23. Schneider BJ, Naidoo J, Santomasso BD, et al. Management of immune-related adverse events in patients treated with immune checkpoint inhibitor therapy: ASCO guideline update. J Clin Oncol. 2021;39:4073–4126. doi:10.1200/JCO.21.01440

24. Hendriks LE, Kerr KM, Menis J, et al. Non-oncogene-addicted metastatic non-small-cell lung cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol. 2023;34:358–376. doi:10.1016/j.annonc.2022.12.013

25. Xiong A, Wang L, Chen J, et al. Ivonescimab versus pembrolizumab for PD-L1-positive non-small cell lung cancer (HARMONi-2): a rando