Analysis of survival and prognostic differences in locally advanced or metastatic non-squamous non-small cell lung cancer with KRAS mutations Authors: Xiangyu Du, Shanxin Zhang, Yahui Lv, Zhefeng Liu Published: 2026 ABSTRACT AbstractBackgroundAmong Eastern populations, the occurrence of Kirsten rat sarcoma viral oncogene (KRAS) mutations in non-small cell lung cancer (NSCLC) is around 10% to 15%. Despite being approved for marketing, targeted drugs are recommended by the Chinese Society of Clinical Oncology (CSCO) Guidelines only as a second-line option, and the selection of first-line treatment regimens remains to be verified. The purpose of this study is to explore the effectiveness of various initial treatment plans for NSCLC with KRAS mutations in real-world settings.MethodsThe study reviewed the clinical data of 180 patients with KRAS mutation-positive locally advanced or metastatic non-squamous NSCLC treated at the Chinese People’s Liberation Army (PLA) General Hospital over the previous decade. The patients were categorized based on their initial treatment plans to compare therapeutic efficacy and prognosis, and the relevant influencing factors were analyzed.ResultsIn the 180 patients, median progression-free survival (mPFS) was 8.60 months and median overall survival (mOS) was 17.47 months. Immune checkpoint inhibitors plus chemotherapy (ICIs + CHE) showed the best efficacy, with mPFS 11.30 months, mOS 21.24 months, objective response rate (ORR) 39.29%, and disease control rate (DCR) 89.29%. ICIs included standard programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) antibodies, chemotherapy was platinum-based doublet therapy, and anti-angiogenic therapy was limited to bevacizumab (BEV). This study directly compared ICIs + CHE with the triple combination (ICIs + CHE + BEV) to fill an existing research gap. The ICIs + CHE group had significantly better progression-free survival (PFS) than ICIs or CHE alone, and better overall survival (OS) than CHE alone. Adding anti-angiogenic therapy to ICIs + CHE provided no significant additional benefit. Different level of PD-L1 expression did not affect PFS or OS in immunotherapy-based regimens. In univariate and multivariate analyses, age and gender did not influence PFS; high Eastern Cooperative Oncology Group (ECOG) performance status and bone metastasis were associated with OS. Baseline characteristics were generally comparable across groups, with only minor differences in age, brain metastasis, and PD-L1 expression (P<0.05), which did not independently predict survival.ConclusionsICIs + CHE is the preferred first-line treatment regimen for patients with locally advanced/metastatic KRAS-mutant non-squamous NSCLC, and the addition of anti-angiogenic agents does not improve therapeutic efficacy. Clinical treatment regimens should be selected individually based on patient characteristics and PD-L1 expression level did not serve as a survival stratification factor for immunotherapy-based treatment regimens.Keywords: Non-squamous non-small cell lung cancer (non-squamous NSCLC), Kirsten rat sarcoma viral oncogene (KRAS), first-line treatment regimens Highlight box Key findings • This real-world study enrolling 180 East Asian patients with Kirsten rat sarcoma viral oncogene (KRAS)-mutant advanced non-squamous non-small cell lung cancer (NSCLC) identifies immune checkpoint inhibitors plus chemotherapy (ICIs + CHE) as the optimal first-line option (median progression-free survival 11.30 months, median overall survival 21.24 months), superior to single-agent ICIs or chemotherapy. Adding bevacizumab confers no additional clinical benefit, and programmed death-ligand 1 (PD-L1) stratification shows no prognostic predictive value for patients receiving immunotherapy-based treatment. Eastern Cooperative Oncology Group performance status (ECOG PS) 2 and bone metastasis correlate with shortened overall survival. What is known and what is new? • KRAS mutation occurs in 10–15% East Asian NSCLC; prior data favored ICIs + CHE yet anti-angiogenic addition value remained unclear. • This head-to-head comparison proves routine bevacizumab combination is unnecessary and PD-L1 fails to stratify immunotherapy benefits. What is the implication, and what should change now? • ICIs + CHE is standardized first-line for eligible patients while bevacizumab is limited to specific subgroups such as those with STK11/KEAP1 co-mutations or symptomatic visceral metastases. PD-L1 level cannot stratify prognostic differences to guide upfront ICIs selection, with single-agent immunotherapy only for frail chemo-intolerant high-PD-L1 patients. Risk-adapted surveillance is needed for patients with poor ECOG or bone metastasis, and more clinical trials are required to advance first-line targeted therapies especially for non-G12C KRAS subtypes. Introduction Lung cancer led all malignancies in 2022 with 2.48 million new cases and 1.81 million deaths, ranking first in both incidence and mortality (1). The National Cancer Center’s latest information reveals that the incidence and mortality of malignant tumors in China are on a continuous rise, with lung cancer being the leading cause of cancer-related incidence and death (2). The majority of lung cancer cases, over 80%, is non-small cell lung cancer (NSCLC). Owing to atypical early symptoms, most patients receive a diagnosis at an advanced stage, with a 5-year survival rate of less than 15% (3). The Kirsten rat sarcoma viral oncogene (KRAS) mutation, the first discovered driver gene mutation, was first reported in 1984 (4), The mutation rate is 20% to 30% among Western populations and 10% to 15% among Eastern populations, which can be further subdivided into subtypes such as G12C, G12D G12V and other variants. KRAS mutations are often found in lung adenocarcinoma but are infrequent in lung squamous cell carcinoma (5). For decades, KRAS was considered an “undruggable” target due to the poor druggability of its encoded protein, diverse mutation subtypes, strong tumor heterogeneity, and drug research and development barriers (6), leaving platinum-based chemotherapy (CHE) as the main first-line treatment for KRAS-mutant NSCLC for a long time. In recent years, immunotherapy and targeted therapy have emerged as core strategies complementing chemotherapy for KRAS-mutant NSCLC. Both basic and clinical studies have confirmed that patients with KRAS-mutant NSCLC can benefit from immunotherapy. Therefore, immunotherapy has also been recommended as a first-line treatment option in the Chinese Society of Clinical Oncology (CSCO) Guidelines, including various treatment modalities such as immunotherapy monotherapy and immunotherapy combined with chemotherapy. Although sotorasib, adagrasib, and fulzerasib have been approved for subsequent-line KRAS G12C-mutant advanced NSCLC, they are only recommended for second and later-line therapy owing to limited first-line evidence and restriction to the G12C subtype. Despite these advances, key clinical questions remain unresolved for first-line treatment of KRAS-mutant NSCLC: the superiority of immune checkpoint inhibitors (ICIs) plus CHE in real-world Eastern populations needs further verification, the clinical value of adding anti-angiogenic drugs to ICIs + CHE is controversial due to the lack of direct head-to-head comparisons, and the predictive role of programmed death-ligand 1 (PD-L1) expression for ICIs-based therapy is unclarified. This study retrospectively analyzed the clinical efficacy and survival outcomes of 180 patients with KRAS-mutant locally advanced or metastatic non-squamous NSCLC who received different first-line treatments at the Chinese People’s Liberation Army (PLA) General Hospital over the past decade, aiming to identify the optimal first-line therapeutic strategy and provide evidence-based guidance for clinical individualized treatment. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/rc). Methods Study subjects Patients with KRAS mutation-positive NSCLC who were treated in the 1st, 3rd and 5th Medical Centers of Chinese PLA General Hospital from May 2015 to May 2025 were enrolled. Inclusion criteria: (I) definite diagnosis of NSCLC confirmed by histopathology or cytology; (II) diagnosed as stage IIIB/IIIC (locally advanced) or stage IV (distant metastasis) according to the 8th edition of the Lung Cancer TNM Staging System issued by the International Association for the Study of Lung Cancer (IASLC); (III) KRAS mutation detected in tissue samples (such as primary lung lesions and tumor metastases) or body fluid samples (such as pleural effusion and peripheral blood) by detection technologies including next generation sequencing (NGS) or polymerase chain reaction (PCR); (IV) presence of at least one measurable lesion in accordance with version 1.1 of the Response Evaluation Criteria in Solid Tumors (RECIST); (V) Eastern Cooperative Oncology Group (ECOG) performance status (PS) score of 0 to 2; (VI) no prior systemic therapy for locally advanced or metastatic lesions, being first-line treatment patients. Exclusion criteria: (I) complication with other malignant tumors; (II) incomplete baseline clinical data, treatment information or follow-up data, with missing key efficacy and prognostic information; (III) irregular first-line treatment regimen or excessively short treatment cycle, making it impossible to evaluate efficacy; (IV) complication with severe dysfunction of important organs such as heart, liver and kidney, or active infection, making it impossible to tolerate systemic therapy. All first-line treatment regimens were formulated through multidisciplinary team (MDT) discussions involving oncologists, pathologists and radiotherapists, and strictly abided by the CSCO Guidelines for the Diagnosis and Treatment of NSCLC corresponding to the treatment period. For KRAS-mutant stage Ⅳ NSCLC patients with ECOG PS 0–1, the regimens were selected from the guideline-recommended first-line options including ICIs, ICIs + CHE, CHE, and CHE combined with anti-angiogenic therapy [bevacizumab (BEV), as the most commonly used representative]. For patients with ECOG PS 2, first-line chemotherapy was the only recommended regimen in accordance with the guidelines. All eligible patients received guideline-recommended regimens prioritized if there were no contraindications to the corresponding drugs. A standardized missing data management strategy was applied in statistical analysis. For missing baseline categorical variables (e.g., PD-L1 expression), an independent “unknown” subgroup was set for analysis to avoid deletion bias. Patients with incomplete treatment/efficacy data or unevaluable efficacy due to short treatment cycles were excluded per the exclusion criteria and not included in final analysis. Patients lost to follow-up with missing progression-free survival (PFS)/overall survival (OS) data were excluded during screening, and all included patients had complete survival outcome data. A supplementary sensitivity analysis was conducted for the small proportion of missing PD-L1 expression data, confirming this missing data had no independent impact on the primary study conclusions, further verifying the reliability of results. The grouping and comparison of the above regimens were designed based on two core scientific questions to address the existing research gaps in this field: (I) to compare the efficacy differences among ICIs, CHE, and ICIs + CHE, which not only verifies the clinical efficacy of ICIs for KRAS-mutant NSCLC but also further explores whether ICIs + CHE can bring additional survival benefits compared with ICIs; (II) to verify the real-world clinical value of the triple combination regimen (ICIs + CHE + BEV), since the IMpower150 study (the first to confirm the efficacy of triple combination) lacked a direct comparison between ICIs plus chemotherapy and triple combination, and this study intended to fill this gap by comparing ICIs + CHE, CHE + BEV, and ICIs + CHE + BEV. All therapeutic agents and administration principles in this study were strictly standardized and unified across the three research centers to ensure the homogeneity of treatment interventions: (I) ICIs: the ICIs used were mainly pembrolizumab, tislelizumab, nivolumab, and sintilimab. Relevant previous studies have confirmed no statistically significant difference in the clinical efficacy of these programmed cell death protein 1 (PD-1)/PD-L1 monoclonal antibodies for advanced NSCLC (7), so they were classified as a unified category in statistical analysis; (II) chemotherapy: all regimens were platinum-based doublet chemotherapy, mainly combined with pemetrexed or paclitaxel. The dosage was strictly determined according to the patient’s body surface area, area under the curve (AUC) formula, and liver/kidney function, with unified administration cycles and supportive care measures; (III) anti-angiogenic therapy: only BEV was included in this study, as it is the main recommended anti-angiogenic drug for first-line treatment of advanced non-squamous NSCLC in clinical practice, and no other anti-angiogenic agents were used for enrolled patients. All patients meeting the inclusion and exclusion criteria were divided into the ICIs group, CHE group, ICIs + CHE group, CHE + BEV group, ICIs + CHE + BEV group and other group according to their first-line treatment regimens. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of the Chinese PLA General Hospital (approval No. ethical review 2021-462-01) and individual consent for this retrospective analysis was waived. Study methods The clinical information of patients who satisfied the inclusion and exclusion criteria was gathered retrospectively. The follow-up period was from the initial treatment to disease progression, occurrence of intolerable adverse reactions or death, with the final follow-up date of September 30, 2025. The recorded data included: baseline general information before treatment, gene detection results, chest computed tomography (CT), cranial magnetic resonance imaging (MRI), abdominal and pelvic CT, positron emission tomography-CT (PET-CT), bone scan, tumor markers, blood routine, liver and kidney function, KRAS mutation subtypes, co-mutant genes, diagnosis date, progression date, death date, etc. During the treatment, the size of target lesions was dynamically compared based on reexamination results to evaluate the therapeutic efficacy. PD-L1 expression detection PD-L1 expression was detected by immunohistochemistry (IHC)—the gold standard approved by the National Medical Products Administration (NMPA) and recommended by the 2025 CSCO NSCLC Guidelines—at the Pathology Departments of each Medical Center of the Chinese PLA General Hospital or qualified third-party testing institutions. All procedures strictly followed the national clinical principle of one drug corresponding to one dedicated reagent, with only NMPA-approved IHC reagents (22C3, 28-8, SP263) matching the administered PD-1/PD-L1 inhibitors used (no cross-use of reagents). Tumor Proportion Score (TPS) was the exclusive interpretation standard, with cutoff values defined per the 2025 CSCO Guidelines: TPS <1% (negative), 1%≤ TPS <50% (moderate/low expression), and TPS ≥50% (high expression). Efficacy evaluation and criteria The therapeutic efficacy was evaluated according to RECIST 1.1. The efficacy was classified based on lesion size as complete response (CR), partial response (PR), stable disease (SD) or progressive disease (PD). OS was defined as the time from the initiation of treatment to death from any cause; PFS was defined as the time from the initiation of treatment to disease progression or death (whichever occurred first); disease control rate (DCR) was defined as the percentage of patients with treatment response (CR + PR + SD) in the total number of patients; objective response rate (ORR) was defined as the percentage of patients with CR + PR in the total number of patients. Statistical analysis In statistical analysis, ICIs with consistent clinical efficacy, standardized platinum-based doublet chemotherapy, and BEV-based anti-angiogenic therapy were classified into unified groups respectively, to eliminate the heterogeneity of therapeutic agents within the same treatment category and ensure the validity of intergroup comparison. Statistical analysis was performed using SPSS 26.0 software. The baseline characteristics of patients were described by descriptive statistics, and categorical variables were expressed as percentages. PFS and OS were calculated using the Kaplan-Meier method, and survival curves were generated with R language. To compare survival data, the log-rank test was applied, considering a two-sided P value under 0.05 as statistically significant. Results General clinical data of patients A total of 180 cases of stage IIIB, IIIC and IV KRAS mutation-positive NSCLC patients were collected from the 1st, 3rd and 5th Medical Centers. By the final follow-up date, 152 patients had died and 28 were still alive, including 143 males and 37 females. Patients aged over 65 years accounted for 39.44% of the total. There were 15 cases (8.33%) at stage IIIB/IIIC and 165 cases (91.67%) at stage IV. The patients were grouped according to first-line treatment regimens: ICIs monotherapy group (9 cases, 5.00%), CHE monotherapy group (34 cases, 18.89%), ICIs + CHE group (56 cases, 31.11%), CHE + BEV group (33 cases, 18.33%), and ICIs + CHE + BEV group (31 cases, 17.22%). The most common KRAS mutation subtypes were G12C (39 cases, 21.67%), G12D (25 cases, 13.88%), G12V (21 cases, 11.67%) and G12A (12 cases, 6.66%); other detected subtypes included G12R, G12S, G13C, G13D, Q61H, etc. The common co-mutations were TP53 (60 cases, 33.33%), STK11 (30 cases, 16.67%), PIK3CA (15 cases, 8.33%), KEAP1 (10 cases, 5.55%) and RET (5 cases, 2.77%); other detected co-mutations included NTRK1, ERBB2, PTEN, CDKN2A, ALK, BRAF, etc. Regarding PD-L1 expression, 36 cases (20.00%) had PD-L1 <1%, 70 cases (38.89%) had 1%≤ PD-L1 <50%, 49 cases (27.22%) had PD-L1 ≥50%, and 25 cases (13.89%) had unknown or undetected PD-L1 expression. Among the 180 enrolled patients, 162 cases (90.00%) had ECOG PS 0–1, and 18 cases (10.00%) had ECOG PS 2, which was consistent with the main applicable population of CSCO guideline-recommended first-line regimens for advanced NSCLC. Comprehensive statistical comparison of core baseline indicators showed that the overall baseline characteristics of each treatment group were comparable: no statistically significant differences were observed in gender, smoking history, tumor stage, and most metastatic sites (intrapulmonary, bone, liver, pleural) across groups (all P>0.05). Only age, brain metastasis status and PD-L1 expression level presented intergroup statistical differences (all P<0.05), and subsequent survival analysis confirmed these factors had no independent prognostic impact on PFS and OS. The baseline characteristics of the patients are shown in Table 1 and Figure 1. Table 1General clinical data of patientsVariablesTotal (n=180)CHE (n=34)CHE + BEV (n=33)ICIs (n=9)ICIs + CHE (n=56)ICIs + CHE + BEV (n=31)Others (n=17)PECOG0.09 0–1168 (93.33)34 (100.00)30 (90.91)8 (88.89)54 (96.43)28 (90.32)14 (82.35) 212 (6.67)0 (0.00)3 (9.09)1 (11.11)2 (3.57)3 (9.68)3 (17.65)Age0.01* ≤65 years109 (60.56)21 (61.76)23 (69.70)1 (11.11)35 (62.50)22 (70.97)7 (41.18) >65 years71 (39.44)13 (38.24)10 (30.30)8 (88.89)21 (37.50)9 (29.03)10 (58.82)Gender0.63 Male143 (79.44)26 (76.47)29 (87.88)7 (77.78)41 (73.21)26 (83.87)14 (82.35) Female37 (20.56)8 (23.53)4 (12.12)2 (22.22)15 (26.79)5 (16.13)3 (17.65)Smoking0.20 No54 (30.00)10 (29.41)6 (18.18)4 (44.44)22 (39.29)6 (19.35)6 (35.29) Yes126 (70.00)24 (70.59)27 (81.82)5 (55.56)34 (60.71)25 (80.65)11 (64.71)Stage0.34 315 (8.33)6 (17.65)1 (3.03)0 (0.00)4 (7.14)2 (6.45)2 (11.76) 4165 (91.67)28 (82.35)32 (96.97)9 (100.00)52 (92.86)29 (93.55)15 (88.24)Metastasis Lung68 (37.78)14 (41.18)11 (33.33)2 (22.22)24 (42.86)11 (35.48)6 (35.29)0.84 Brain47 (26.11)6 (17.65)13 (39.39)2 (22.22)9 (16.07)13 (41.94)4 (23.53)0.045* Bone78 (43.33)14 (41.18)13 (39.39)7 (77.78)27 (48.21)9 (29.03)8 (47.06)0.17 Adrenal26 (14.44)7 (20.59)3 (9.09)2 (22.22)8 (14.29)6 (19.35)0 (0.00)0.27 Liver17 (9.44)2 (5.88)3 (9.09)3 (33.33)5 (8.93)2 (6.45)2 (11.76)0.30 Pleural52 (28.89)6 (17.65)8 (24.24)3 (33.33)18 (32.14)13 (41.94)4 (23.53)0.35PD-L10.002* <1%36 (20.00)5 (14.71)11 (33.33)0 (0.00)9 (16.07)9 (29.03)2 (11.76) 1–49%70 (38.89)13 (38.24)11 (33.33)0 (0.00)24 (42.86)11 (35.48)11 (64.71) ≥50%49 (27.22)7 (20.59)7 (21.21)9 (100.00)16 (28.57)6 (19.35)4 (23.53) Unknown25 (13.89)9 (26.47)4 (12.12)0 (0.00)7 (12.50)5 (16.13)0 (0.00)Data are presented as n (%). *, P<0.05. BEV, bevacizumab; CHE, chemotherapy; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; PD-L1, programmed death-ligand 1. General clinical data of patients Data are presented as n (%). *, P<0.05. BEV, bevacizumab; CHE, chemotherapy; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; PD-L1, programmed death-ligand 1. Heatmap showing the clinical characteristics, treatment outcomes, and mutation profiles of samples analyzed with multigene NGS testing. BEV, bevacizumab; CHE, chemotherapy; CR, complete response; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; NGS, next-generation sequencing; PD, progressive disease; PD-L1, programmed death-ligand 1; PR, partial response; SD, stable disease. Analysis of differences in different first-line treatment regimens PFS data of first-line treatment were available for all 180 patients, with a mPFS of 8.60 months [95% confidence interval (CI): 7.53–10.13] and a mOS of 17.47 months (95% CI: 14.60–19.88). The clinical outcomes of each group were as follows: ICIs monotherapy group: mPFS 7.53 months [95% CI: 5.27–not available (NA)], mOS 14.77 months (95% CI: 12.63–NA), ORR 22.22%, DCR 77.78%, 2-year OS rate 22.22%, 3-year OS rate 11.11%; CHE monotherapy group: mPFS 5.83 months (95% CI: 4.67–10.13), mOS 13.22 months (95% CI: 11.26–19.88), ORR 35.29%, DCR 70.59%, 2-year OS rate 17.65%, 3-year OS rate 5.88%; ICIs + CHE group: mPFS 11.30 months (95% CI: 9.03–12.97), mOS 21.24 months (95% CI: 14.20–29.13), ORR 39.29%, DCR 89.29%, 2-year OS rate 39.29%, 3-year OS rate 17.86%; CHE + BEV group: mPFS 8.53 months (95% CI: 5.22–12.50), mOS 15.57 months (95% CI: 12.84–21.00), ORR 54.55%, DCR 93.94%, 2-year OS rate 18.18%, 3-year OS rate 6.06%; ICIs + CHE + BEV group: mPFS 8.67 months (95% CI: 7.26–11.80), mOS 20.00 months (95% CI: 17.17–41.67), ORR 45.16%, DCR 83.87%, 2-year OS rate 35.48%, 3-year OS rate 25.81%. The other group mainly included patients who participated in clinical trials with antibody-drug conjugate (ADC) combined regimens or those with poor baseline conditions who could not tolerate conventional treatment. No statistical analysis was performed for this group due to the small sample size. The survival data of each group are shown in Table 2. Table 2Survival analysis data of each subgroupVariablesICIsCHEICIs + CHECHE + BEVICIs + CHE + BEVmPFS (months)7.535.8311.308.538.67mOS (months)14.7713.2221.2415.5720.00ORR (%)22.2235.2939.2954.5545.16DCR (%)77.7870.5989.2993.9483.872-year OS rate (%)22.2217.6539.2918.1835.483-year OS rate (%)11.115.8817.866.0625.81ICIs include pembrolizumab, tislelizumab, nivolumab and sintilimab (unified category for consistent efficacy). Chemotherapy = standardized platinum-based doublet therapy (pemetrexed/ paclitaxel + platinum) with dosage determined by body surface area, AUC formula and liver/kidney function; anti-angiogenic therapy = only bevacizumab included. AUC, area under the curve; BEV, bevacizumab; CHE, chemotherapy; DCR, disease control rate; ICI, immune checkpoint inhibitor; mOS, median overall survival; mPFS, median progression-free survival; ORR, objective response rate; OS, overall survival. Survival analysis data of each subgroup ICIs include pembrolizumab, tislelizumab, nivolumab and sintilimab (unified category for consistent efficacy). Chemotherapy = standardized platinum-based doublet therapy (pemetrexed/ paclitaxel + platinum) with dosage determined by body surface area, AUC formula and liver/kidney function; anti-angiogenic therapy = only bevacizumab included. AUC, area under the curve; BEV, bevacizumab; CHE, chemotherapy; DCR, disease control rate; ICI, immune checkpoint inhibitor; mOS, median overall survival; mPFS, median progression-free survival; ORR, objective response rate; OS, overall survival. Survival analysis of ICIs monotherapy, CHE monotherapy and ICIs + CHE The mPFS of patients receiving first-line ICIs monotherapy, CHE monotherapy and ICIs + CHE were 7.53, 5.83 and 11.30 months, respectively, with corresponding mOS of 14.77, 13.22 and 21.24 months. Pairwise comparisons revealed that the PFS of the ICIs + CHE group was notably better than that of the ICIs and CHE monotherapy groups, with statistical significance, and the OS of the ICIs + CHE group surpassed that of the CHE monotherapy group (see Figure 2). Survival analysis of ICIs monotherapy, CHE monotherapy and ICIs + CHE. BEV, bevacizumab; CHE, chemotherapy; ICI, immune checkpoint inhibitor. Survival analysis of ICIs + CHE, CHE + BEV and ICIs + CHE + BEV The mPFS of patients receiving first-line ICIs + CHE, CHE + BEV and ICIs + CHE + BEV were 11.30, 8.53 and 8.67 months, respectively, with corresponding mOS of 21.24, 15.57 and 20.00 months. Statistical analysis showed that the addition of anti-angiogenic agents to ICIs + CHE did not yield statistically significant additional benefits (see Figure 3). Survival analysis of ICIs + CHE, CHE + BEV and ICIs + CHE + BEV. BEV, bevacizumab; CHE, chemotherapy; ICI, immune checkpoint inhibitor. Survival analysis of patients with different PD-L1 expression levels Among patients receiving immunotherapy-based regimens (including ICIs monotherapy, ICIs + CHE, and ICIs + CHE + BEV), the mPFS of patients with PD-L1 <1%, 1%≤ PD-L1 <50% and PD-L1 ≥50% were 8.67 months (95% CI: 7.73–15.90), 7.85 months (95% CI: 6.70–10.46) and 9.37 months (95% CI: 7.53–16.88), respectively, with corresponding mOS of 19.50 months (95% CI: 15.57–41.17), 18.67 months (95% CI: 13.03–22.90) and 24.77 months (95% CI: 14.77–NA). Pairwise comparisons showed no statistically significant differences in PFS and OS among the three groups (see Figure 4). Survival analysis of patients with different PD-L1 expression levels. PD-L1, programmed death-ligand 1. Analysis of factors influencing PFS Univariate and multivariate Cox regression analyses for PFS in KRAS mutation-positive non-squamous NSCLC patients showed that all included variables (age, gender, smoking history, tumor stage, ECOG PS, distant metastasis sites including intrapulmonary, brain and bone metastasis, and PD-L1 expression level) had P>0.05 and hazard ratio (HR) close to 1 in the univariate analysis, indicating no significant effects. After multivariate adjustment, all variables still had P>0.05, and the 95% CIs all included 1. No independent risk factors or protective factors for PFS were identified (see Table 3). Table 3Cox regression analysis of factors influencing PFSVariablesUnivariateMultivariateHR (95% CI)PHR (95% CI)PAge ≤65 years1.00 (reference)1.00 (reference) >65 years1.04 (0.76–1.43)0.811.02 (0.71–1.47)0.92Gender Male1.00 (reference)1.00 (reference) Female1.14 (0.78–1.67)0.501.19 (0.69–2.06)0.54Smoking history No1.00 (reference)1.00 (reference) Yes0.88 (0.63–1.23)0.470.99 (0.61–1.61)0.98Stage III1.00 (reference)1.00 (reference) IV1.05 (0.62–1.80)0.841.09 (0.62–1.91)0.77ECOG performance status 0–11.00 (reference)1.00 (reference) 21.28 (0.71–2.31)0.411.32 (0.70–2.50)0.39Intrapulmonary metastasis1.01 (0.74–1.39)0.941.04 (0.73–1.48)0.83Brain metastasis0.91 (0.64–1.30)0.620.92 (0.63–1.36)0.68Bone metastasis1.09 (0.79–1.48)0.611.08 (0.77–1.51)0.65Adrenal metastasis0.88 (0.56–1.36)0.560.80 (0.49–1.31)0.38Liver metastasis1.02 (0.61–1.71)0.951.21 (0.67–2.19)0.53Pleural metastasis/effusion0.88 (0.63–1.25)0.480.81 (0.55–1.19)0.29PD-L1 expression <1%1.00 (reference)1.00 (reference) 1–49%1.03 (0.68–1.56)0.901.00 (0.65–1.55)0.99 ≥50%0.91 (0.59–1.42)0.680.86 (0.53–1.40)0.55CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; PD-L1, programmed death-ligand 1; PFS, progression-free survival. Cox regression analysis of factors influencing PFS CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; PD-L1, programmed death-ligand 1; PFS, progression-free survival. Analysis of factors influencing OS Univariate and multivariate Cox regression analyses for OS were performed similarly. Univariate analysis showed that ECOG PS 2 and bone metastasis were risk factors for OS (both P<0.05), while intrapulmonary metastasis was a protective factor for OS (P<0.05). Other variables including age, gender, smoking history and tumor stage had P>0.05 with no significant effects. After multivariate adjustment, all variables had P>0.05 with no independent statistical effects (see Table 4). Table 4Cox regression analysis of factors influencing OSVariablesUnivariateMultivariateHR (95% CI)PHR (95% CI)PAge ≤65 years1.00 (reference)1.00 (reference) >65 years1.21 (0.88–1.67)0.251.31 (0.91–1.89)0.15Gender Male1.00 (reference)1.00 (reference) Female0.94 (0.62–1.43)0.781.11 (0.62–1.96)0.73Smoking history No1.00 (reference)1.00 (reference) Yes1.01 (0.71–1.43)0.971.12 (0.68–1.83)0.66Stage III1.00 (reference)1.00 (reference) IV1.07 (0.63–1.83)0.811.03 (0.58–1.82)0.92ECOG performance status 0–11.00 (reference)1.00 (reference) 21.85 (1.02–3.35)0.043*1.63 (0.86–3.08)0.13Intrapulmonary metastasis0.69 (0.49–0.96)0.03*0.69 (0.48–1.00)0.053Brain metastasis1.04 (0.73–1.49)0.821.06 (0.70–1.58)0.79Bone metastasis1.43 (1.04–1.97)0.03*1.31 (0.93–1.84)0.12Adrenal metastasis0.87 (0.55–1.36)0.540.84 (0.51–1.38)0.49Liver metastasis0.75 (0.42–1.32)0.310.94 (0.48–1.81)0.84Pleural metastasis/effusion1.04 (0.73–1.47)0.841.13 (0.76–1.67)0.54PD-L1 expression <1%1.00 (reference)1.00 (reference) 1–49%0.98 (0.63–1.51)0.931.05 (0.68–1.65)0.82 ≥50%0.76 (0.47–1.23)0.260.75 (0.45–1.25)0.28*, P<0.05. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; OS, overall survival; PD-L1, programmed death-ligand 1. Cox regression analysis of factors influencing OS *, P<0.05. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; OS, overall survival; PD-L1, programmed death-ligand 1. Discussion The study systematically evaluated the efficacy of various first-line treatment regimens based on real-world data from 180 patients with KRAS mutation-positive locally advanced or metastatic NSCLC. All treatment regimens were formulated in accordance with CSCO Guidelines and MDT discussions, and the overall comparability of baseline characteristics across groups ensured the reliability of efficacy comparison among different first-line regimens. The grouping comparison was designed to address two core research gaps: verifying the additional survival benefit of ICIs + CHE over ICIs monotherapy, and exploring the real-world value of triple combination regimen (ICIs + CHE + BEV) by direct comparison with ICIs + CHE, which is a key deficiency of the IMpower150 study. All therapeutic agents were strictly standardized, ensuring the reliability of intergroup efficacy comparison. Platinum-based chemotherapy was the primary treatment option for KRAS mutation-positive NSCLC patients before immunotherapy emerged, but the efficacy was unsatisfactory. Multiple studies have shown that the PFS and OS of KRAS mutation-positive patients after first-line chemotherapy are significantly shorter than those of patients with driver gene-negative or KRAS wild-type NSCLC. For instance, Hames and his group reported mPFS and mOS of 4.5 and 8.8 months, respectively, for KRAS-mutant patients on chemotherapy, lower than in the wild-type population (8). Domestic studies have also confirmed that the mPFS of KRAS mutation-positive patients treated with pemetrexed plus platinum was 6.4 months, significantly lower than 9.9 months of the wild-type population (9). In this study, the mPFS and mOS of the CHE monotherapy group were only 5.83 and 13.22 months, respectively, which were consistent with the above research results and further verified the limitations of conventional chemotherapy in the KRAS mutation-positive population. Basic research indicates that KRAS mutations can upregulate PD-L1 via the RAF-MEK-ERK pathway, mediating immune escape. Additionally, KRAS-mutant tumors often exhibit a high tumor mutational burden (TMB). These characteristics provide the molecular basis for immunotherapy sensitivity (10,11). A meta-analysis integrating three first-line trials pointed out that for KRAS mutation-positive NSCLC, compared with chemotherapy alone, the use of immunotherapy, whether combined with chemotherapy or not, was linked to a significant extension in OS [HR 0.59 (95% CI: 0.49–0.72); P<0.00001] and PFS [HR 0.58 (95% CI: 0.43–0.78); P=0.0003], with OS benefits observed in both first-line and second-line trials (12-15). Mechanistically, chemotherapeutic drugs induce tumor cell death by damaging tumor cell DNA or interfering with the cell cycle, and release a large number of tumor-associated antigens at the same time, providing “targets” for the immune response. Meanwhile, in the tumor microenvironment, they selectively eradicate immune-suppressive cells, including myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs), break the immune-suppressive state, and “clear the obstacles” for effector T cells activated by ICIs. CheckMate-9LA data showed that nivolumab plus ipilimumab with chemotherapy extended mOS by 8.3 months compared to chemotherapy alone (16), In our study, ICIs + CHE extended mPFS by 3.77 and 5.47 months, and mOS by 6.47 and 8.02 months compared to ICIs and CHE monotherapy, respectively. Although the OS benefit of the combined therapy did not reach statistical significance, the favorable trend aligns with previous research, confirming the synergistic value of combined therapy. From a clinical practice perspective, this result has important guiding significance for the stratification of first-line treatment for KRAS mutation-positive NSCLC: for patients with PD-L1 expression ≤49% or high tumor burden, ICIs monotherapy may be difficult to control disease progression rapidly, while ICIs + CHE can achieve dual benefits of short-term efficacy and long-term survival through the “rapid tumor shrinkage” effect of chemotherapy and the “long-term tumor control” advantage of immunotherapy. For patients with high PD-L1 expression (TPS ≥50%), ICIs + CHE remains preferred if PS is good and high tumor burden; ICIs monotherapy can be a “high-efficiency and low-toxicity” option for elderly patients who cannot tolerate chemotherapy; for patients with immune-negative co-mutations such as STK11/KEAP1, ICIs monotherapy should be avoided (17,18). Anti-angiogenic agents (e.g., BEV) inhibit VEGF, normalizing tumor vasculature and promoting immune cell infiltration, theoretically synergizing with ICIs + CHE (19,20). The IMpower150 trial was the first phase III clinical trial to confirm the efficacy of the “immunotherapy + anti-angiogenic therapy + chemotherapy” regimen. In patients with KRAS mutations, subgroup analysis revealed that the ABCP regimen (atezolizumab + BEV + carboplatin + paclitaxel) extended OS to 19.8 months compared to 9.9 months with the BCP regimen (BEV + carboplatin + paclitaxel), with an HR of 0.50, indicating that combining anti-angiogenic agents with ICIs and CHE could enhance effectiveness for patients with KRAS mutations (21). However, there was no direct comparison between ABCP and ACP. According to this, research conducted by Wang et al. demonstrated that the PFS for chemotherapy paired with immunotherapy and anti-angiogenic therapy was better than that for chemotherapy with immunotherapy alone (14.1 months compared to 7.7 months, P=0.049), and OS also indicated a trend of difference (31.9 vs 19.3 months, P=0.158), suggesting that the triple combination regimen may bring more benefits to KRAS mutation-positive NSCLC patients (22). However, this result has not been replicated in multiple subsequent clinical trials. For example, the ASTRUM-002 study found that the addition of a BEV biosimilar to ICIs + CHE did not significantly improve the PFS and OS of advanced non-squamous NSCLC patients (23). In this study, the CHE + BEV group had a high ORR of 54.55% and a DCR of 93.94%, the results were significantly higher compared to the CHE monotherapy group and the ICIs monotherapy group, affirming the efficacy benefits of combining chemotherapy and anti-angiogenic agents. However, the mPFS and mOS of this group were still lower than those of the ICIs + CHE group, indicating limited survival benefits. No significant differences in PFS and OS were observed between the ICIs + CHE + BEV group and the ICIs + CHE group, indicating that anti-angiogenic agents did not enhance the benefits of ICIs + CHE, but may affect patient tolerance due to increased toxicity. Notably, the triple combination regimen (ICIs + CHE + BEV) is not a first-line preferred option for KRAS-mutant advanced NSCLC in the 2025 CSCO Guidelines, due to insufficient definitive efficacy data, potential adverse effect superposition and increased economic burden on patients and medical insurance. Its clinical use is strictly restricted to highly selected subgroups (STK11/KEAP1 co-mutations, symptomatic brain/liver metastasis), which explains why only 31 patients received this regimen as first-line treatment across multiple medical centers of the Chinese PLA General Hospital over a decade. Retrospective studies have found that such patients receiving the triple regimen of immunotherapy + chemotherapy + BEV had a trend of prolonged PFS and OS compared with chemotherapy + BEV. Basic research also suggests that anti-angiogenic agents (e.g., BEV) can promote the normalization of tumor blood vessels, increase the infiltration of effector lymphocytes, and may help improve the microenvironment of “immune-cold” tumors such as KL subtype (24,25). For patients with brain edema, BEV can reduce edema and intracranial pressure and improve clinical symptoms (26). Patients with liver metastasis are relatively resistant to immunotherapy and have a poor prognosis (27,28), the use of anti-angiogenic agents in addition can lead to a better prognosis for these patients. KRAS-targeted therapy has experienced a process from “undruggable” to direct targeted breakthrough. A 2013 study found that a druggable pocket exists in the Switch II region of the KRAS G12C mutant protein in the GDP-bound state, which promoted the research and development of direct targeted inhibitors (29). Since 2021, sotorasib and adagrasib, both G12C inhibitors, have received approval for subsequent-line therapy. In 2024, fulzerasib was conditionally approved in China for the treatment of advanced NSCLC patients with G12C mutations who have received previous systemic therapy, with an ORR of 49.1%, mPFS of 9.7 months, favorable efficacy in the brain metastasis subgroup, and controllable safety. The CodeBreaK 200 study confirmed that sotorasib had a better PFS, higher ORR and better safety than docetaxel in subsequent-line therapy (30). Based on this, the 2025 version of the CSCO Guidelines recommended fulzerasib and garsorasib as grade I recommendations for subsequent-line therapy based on domestic drug accessibility. However, no G12C inhibitor monotherapy has been approved for first-line treatment at present. Early studies have shown that the first-line exploration of G12C inhibitors combined with immunotherapy or chemotherapy has shown considerable efficacy. For example, adagrasib combined with pembrolizumab achieved an ORR of 63% in treatment-naive patients with PD-L1 ≥50%, but some combined regimens have the problem of increased hepatotoxicity, and relevant phase III clinical trials are still ongoing (31). In this study, fulzerasib was not included in the efficacy comparison due to the short clinical application time and immature data of enrolled patients by the follow-up cutoff date. As of January 31, 2026, a total of 11 patients were enrolled for first-line treatment, 12 for second-line treatment, and 3 for subsequent-line treatment. Follow-up of these patients will be continued to provide evidence-based evidence for the first-line application of KRAS G12C targeted drugs. For patients with non-G12C KRAS mutation subtypes, there is still a lack of exclusive targeted drugs, and ICIs + CHE remains the core first-line treatment regimen. The current study integrated all KRAS mutation subtypes for primary analysis, which was based on two key considerations. First, multiple clinical studies focusing on advanced KRAS-mutant NSCLC have confirmed that there are no statistically significant differences in PFS or OS among major KRAS subtypes (G12A, G12C, G12D, G12V, G13C), and the biological heterogeneity of subtypes has not been shown to lead to divergent responses to traditional first-line regimens (CHE, ICIs, ICIs + CHE) (32-34). Second, stratified analysis by KRAS subtype plus first-line treatment regimen would drastically reduce the sample size of each subgroup (e.g., the G12C subtype with the largest sample size would have an average of less than 8 cases per treatment subgroup), resulting in insufficient statistical power and increased risk of type I/II errors, which would affect the scientific validity of intergroup comparisons. This study is a multicenter retrospective study with inherent biases. This study is a multicenter retrospective study with inherent biases: minor information bias exists due to undetected PD-L1 expression and missing baseline data in a small number of patients; however, the selection of treatment regimens was standardized by CSCO Guidelines and MDT discussions, and baseline characteristics were generally comparable across groups. The potential confounding effects of age, brain metastasis and PD-L1 expression (with intergroup differences) were fully excluded by univariate and multivariate Cox regression analyses, which confirmed no independent prognostic impact of these factors on survival outcomes. In addition, the multicenter design of three independent tertiary oncology centers effectively reduced selection bias caused by single-center population characteristics. Selection bias exists as only patients from the Chinese PLA General Hospital were enrolled, who were mainly from populations with high medical resource accessibility, and thus cannot represent the overall characteristics of Eastern populations or patients in primary hospitals. The sample size of some treatment groups is small (e.g., only 9 cases in the ICIs monotherapy group), resulting in limited statistical power and inability to accurately explore the efficacy differences among subgroups. The variations in effectiveness among various ICIs, chemotherapy treatments, and anti-angiogenic medications were not categorized, and the heterogeneity of treatment regimens was not controlled. In addition, the post-hoc subgroup analysis of KRAS subtypes is limited by the small sample size of rare subtypes, and the efficacy of first-line regimens in these subtypes cannot be further explored, which needs to be verified by prospective multicenter studies with larger sample sizes. Conclusions Based on 10-year real-world data from the Chinese PLA General Hospital, this study validated in 180 patients of Eastern ancestry that for patients with KRAS mutation-positive locally advanced or metastatic non-squamous NSCLC, the first-choice treatment is a combination of ICIs and chemotherapy. This regimen demonstrated significant superiority over immune monotherapy, chemotherapy monotherapy and chemotherapy combined with anti-angiogenic therapy in prolonging PFS and improving DCR. Including anti-angiogenic agents with immunotherapy and chemotherapy did not result in statistically significant extra benefits, and PD-L1 expression level did not serve as a survival stratification factor for immunotherapy-based treatment regimens. The conclusions of this study are highly consistent with the guidance of the 2025 CSCO Guidelines for the Diagnosis and Treatment of Non-Small Cell Lung Cancer, providing crucial real-world evidence for the Eastern population to support stratified decision-making in clinical first-line treatment and also defining the applicable scenarios for the precise administration of anti-angiogenic agents. In the future, research on KRAS mutation-positive NSCLC needs to conduct prospective multicenter randomized controlled clinical trials, subdivide the efficacy differences of different drugs, and incorporate indicators such as treatment tolerance and quality of life to break through the limitations of retrospective studies. It is necessary to carry out subgroup studies based on different KRAS mutation subtypes and co-mutation characteristics, and combine multi-omics technologies to explore efficacy-predictive biomarkers for the realization of precise and individualized treatment. Continuous follow-up of clinical application data of KRAS G12C targeted drugs is required to investigate the first-line efficacy of targeted monotherapy and combination therapy, study drug resistance mechanisms and construct a whole-course treatment system. The coverage of real-world research should be expanded, in collaboration with primary medical institutions, to establish a long-term follow-up system for evaluating long-term survival benefits. Focus should be placed on non-G12C KRAS mutation subtypes to conduct clinical translational research on novel targeted drugs, optimize ICIs + CHE regimens and fill the gaps in therapeutic evidence. With the development of molecular biology technologies and the advancement of tumor treatment concepts, the treatment of KRAS mutation-positive NSCLC will continue to move towards the direction of “precision, individualization and multimodal combination”. It is expected that in the future, through molecular subtyping guidance and clinical characteristic stratification, optimal treatment regimens can be formulated for patients, breaking the long-standing therapeutic predicament and further improving the survival benefits and quality of life of patients. Supplementary The article’s supplementary files as Acknowledgments None. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of Chinese PLA General Hospital (approval No. ethical review 2021-462-01) and individual consent for this retrospective analysis was waived. Footnotes Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/rc Funding: None. Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/coif). The authors have no conflicts of interest to declare. Data Sharing Statement Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/dss References Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/prf Supplementary Materials The article’s supplementary files as Data Availability Statement Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0522/dss