Crizotinib or vebreltinib response and resistance in advanced non-small cell lung cancer with MET exon 14 skipping Authors: Jiao Yang, Cheng-Di Weng, Han-Min Wang, Jun-Wei Su, Quan-Quan Tan, Xiao-Cheng Lin, Mei-Mei Fang, Zi-Ji Mao, Hua-Jun Chen, Hui Li, Gengpeng Lin, Yi-Long Wu, Yubiao Guo, Jin-Ji Yang Published: 2026 ABSTRACT AbstractBackgroundMET exon 14 skipping (METex14) mutations account for 1–3% of non-small cell lung cancer (NSCLC) cases. Although MET tyrosine kinase inhibitors (TKIs) and immunotherapy have improved survival, resistance remains a challenge.MethodsThis real-world analysis includes 24 METex14-mutated NSCLC patients who underwent genetic sequencing before MET TKI treatment. RNA sequencing data from 11 METex14 mutated NSCLC patients in the TCGA database were analyzed to characterize tumor immune microenvironment composition and functional profiles. Patient-derived organoids (PDOs) were established from malignant effusion or tissue of three patients to suggest treatment responses.ResultsIn this cohort of 24 METex14-mutated NSCLC patients, the overall median progression-free survival (mPFS) was 7.0 months. While crizotinib (n = 12) showed mPFS of 6.1 months (disease control rate DCR 66.7%) with 50% progressing within 3 months, vebreltinib (n = 12) demonstrated longer mPFS of 8.7 months (DCR 91.7%), with half exceeding 7.0 months. However, these differences did not reach statistical significance. Notably, two patients benefited from sequential type Ib MET TKI therapy after type Ia failure. Resistance mechanisms included acquired D1228N MET mutation, EGFR amplification, and histologic transformation to pulmonary sarcomatoid carcinoma (n = 1). One patient with high PD-L1 expression (TPS ≥ 50%) achieved 13.0 months PFS with camrelizumab/endostar immunotherapy. RNA sequencing revealed PD-L1-high tumors harbored an immune-favorable microenvironment. Patient-derived organoid (PDO) drug test demonstrated preliminary qualitative consistency with clinical outcomes, supporting their predictive value.ConclusionsWhile vebreltinib appears clinically advantageous over crizotinib for METex14-mutated NSCLC, the therapeutic benefits did not reach statistical significance in this study. The observed differential response patterns and resistance mechanisms suggest distinct biological behaviors to type Ia and Ib MET TKIs that warrant further investigation. These findings underscore the need for larger prospective studies to validate the potential superiority of vebreltinib and to better characterize the molecular determinants in NSCLC.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-026-05183-x.Keywords: MET tyrosine kinase inhibitor, Non-small cell lung cancer, Drug resistance, Immunotherapy, Patient-derived organoid Introduction As an oncogene, MET dysregulation plays an essential role in tumorigenesis and tumor development in lung cancer. Several mechanisms underlying MET dysfunction have been identified, including MET point mutations, MET amplification, MET fusion, and MET exon 14 skipping mutations (METex14) [1]. METex14 is detected in approximately 1–3% of lung cancer patients and is frequently observed in older individuals and smokers [2–6]. Type I MET inhibitors bind to the active conformations of kinases in the ATP pocket. One type of Ia inhibitor is crizotinib, which binds to kinase through an interaction with the solvent-front residue G1163. Crizotinib is the first FDA-approved MET tyrosine kinase inhibitor (TKI). In 2016, Alexander Drilon first reported that 18 advanced non-small cell lung cancer (NSCLC) patients harboring METex14 exhibited a 100% disease control rate (DCR), and eight of them achieved a partial response (PR) [7]. Subsequently, in the expanded PROFILE-1001 trial, an objective response rate (ORR) of 32% was reported in 65 advanced NSCLC patients in 2020 [8]. Tepotinib, capmatinib, savolitinib, and vebreltinib are all classified as type Ib MET inhibitors because they bind to the kinase independently of the residue G1163 interaction [1]. Type II MET inhibitors bind to the inactive conformation of kinases in the ATP pocket, whereas type III TKIs bind outside the ATP pocket [1]. The VISION and GEOMETRY mono-1 clinical trials reported the efficacy of tepotinib and capmatinib in advanced NSCLC patients, respectively [9, 10]. Because of achieving a 10-month duration of response and a 40–60% ORR, capmatinib and tepotinib are highly recommended as front-line treatments for METex14 lung cancer patients. In 2020, the phase I study of vebreltinib (PLB-1001, CBT-101, and bozitinib) demonstrated an ORR of 66.7% in 11 enrolled MET-altered NSCLC patients [11]. Recently the phase II KUNPENG study of vebreltinib in MET-altered NSCLC revealed that the ORR was 77.1% in treatment-naïve patients, and 70.6% in previously treated patients [12]. Clinical trials involving type II MET TKIs such as cabozantinib (NCT01639508, ClinicalTrials.gov, registered on July 12, 2012) and gumarontinib (NCT02920996, ClinicalTrials.gov, registered on September 30, 2016) are currently ongoing. Further prospective studies are warranted to assess the comparative efficacy and distinct clinical profiles of crizotinib versus vebreltinib in real-world METex14-altered NSCLC populations. Promising responses to immunotherapy have been observed in METex14 NSCLC patients; however, PD-L1 expression levels and tumor mutational burden do not appear to be reliable indicators of treatment efficacy [13, 14]. Additional prospective studies are required to thoroughly evaluate the efficacy of immune checkpoint inhibitors (ICIs). Materials and methods Patient information The study cohort was retrospectively retrieved from the Lung Cancer Database at Guangdong Provincial People’s Hospital from July 1, 2016, to April 30, 2025. Treatment assignment to crizotinib or vebreltinib was determined according to clinical trial eligibility, drug availability, physician’s decision, and patient financial status, consistent with real-world clinical practice (Supplementary Table 2). This cohort included four patients from the Phase II KUNPENG clinical trial (NCT04258033, ClinicalTrials.gov, registered on February 06, 2020). The remaining 20 patients were treated in a real-world setting: they either failed clinical trial screening or were ineligible for trial participation, and subsequently received targeted therapy via compassionate use programs or self-funded purchase based on their economic conditions. The choice between crizotinib and vebreltinib was further guided by drug accessibility at different treatment periods and the treating physician’s clinical judgment. No randomization was performed. All patients were adults with histologically confirmed stage III-IV NSCLC and METex14 mutations, each with at least one measurable lesion as defined by the Response Evaluation Criteria in Solid Tumors (RECIST) v1.1. The MET mutation status was determined in all patients using next-generation sequencing (NGS). METex14 status detection,immunohistochemistry (IHC) and pathology examination To detect the METex14 status, DNA was isolated from tumor tissue, blood, or effusion specimens (Supplementary Table 2). As this is a retrospective study, patients were subjected to DNA-based targeted next-generation sequencing (NGS) assays using different commercial panels (varied across individuals) conducted by certified clinical genetic testing laboratories. For four patients, only METex14 status was available, with no additional genomic profiling data recorded. Nevertheless, almost all panels consistently covered 10 more key lung cancer driver genes, including MET, EGFR, ALK, ROS1, KRAS, NRAS, BRAF, HER2, PIK3CA, and TP53. Tissue samples were preferentially used for genomic profiling. Only two patients (P05 and P19) had only plasma NGS results, with unavailable tumor tissue, with VAF values of 49.84% and 2.2%, respectively. Gene amplification was defined as a copy number (CN) ≥ 4.0. All MET exon 14 skipping alterations were identified through these NGS assays. All METex14 variants identified in this study are well-established pathogenic splicing events that have been previously reported in the literature and clinical databases [1, 15–17]. Owing to limited residual specimens, RNA sequencing was not performed to further validate these splicing events. PD-L1 staining was performed using the DAKO 22C3 antibody (Dako, Carpinteria, CA). PD-L1 levels were determined by calculating the tumor proportion scores (TPS) that stained positive for PD-L1. MET expression was analyzed by IHC in 3 patients with available FFPE tumor tissue. Membranous MET staining was evaluated semi-quantitatively according to staining intensity and percentage of positive tumor cells; intensity was graded as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong), and the percentage of positive tumor cells was recorded as the proportion of stained tumor cells relative to the total tumor cell population. For the purpose of this study, high MET expression was defined as ≥ 50% positive tumor cells with at least moderate intensity (score ≥ 2). For patients with suspected histologic transformation after MET-TKI treatment failure, re-biopsy was performed. Histologic features were evaluated by hematoxylin and eosin (H&E) staining. Histologic diagnosis was confirmed using an immunohistochemistry panel including Napsin A, TTF-1, CK7, CK5/6, p63, and broad-spectrum cytokeratin (CK). Patient-derived organoid (PDO) establishment As previously illustrated [18–20], PDOs are derived from malignant serous effusion (MSE) or tissue specimens. MSE was obtained through thoracentesis, placed in a sterile bag or glass bottle containing heparin (10 U/mL), and then transported to the laboratory at low temperature (2 to 8 ℃) within 4h. The MSE sample was centrifuged at 300 g for 5 min at 4 ℃, and the supernatant was discarded. The tissue specimens were divided into 1 mm3 fragments, stored in cold Hank’s balanced salt solution (HBSS) with antibiotics (Lonza, Basel, Switzerland), and transported to the laboratory on ice within 1h after retrieval from the patient. After washing three times with cold HBSS and dividing once more with sterile blades, the samples were incubated in DMEM/F12 medium (Lonza) with 0.001% DNase (Sigma-Aldrich, MO, USA), 1 mg/mL collagenase/dispase (Roche, IN, USA), 200 U/mL penicillin, 200 mg/mL streptomycin, and 0.5 mg/mL amphotericin B (2% antibiotics, Sigma) at 37 ℃ for 3h. The cell pellet was resuspended in the medium (ALCM; Accurate International Biotechnology Co., Guangzhou, China). After gentle agitation and intermittent resuspension, the digested tissue suspension was repeatedly pipetted and filed with a 70 μm strainer. The cell suspension generated from MSE or tissue was centrifuged at 112g for 3 min. After resuspending the pellet into 100 μL ALCM, 200 μL Matrigel (Corning, New York, NY, USA) was added into the cell suspension for establishing organoids. The final 300 μL solution was solidified on pre-warmed six-well culture plates (Corning) at 37 ℃ for 30 min. After gelation, 3 mL basal medium (MBM) was added to each well. The medium was replaced with fresh MBM every 2–3 days. Drug sensitivity test After culturing PDOs for over 2 weeks, they were harvested and dissociated using 1 × TrypLe (Gibco, Waltham, MA, USA). Dissociated organoids were mixed in MBM with Matrigel (1:3 ratio) and seeded in 384-well plates. After gelation, 30μL MBM was added to each well. After culturing for 48 h, a dilution series of each compound (50, 10, 2, 0.4, 0.08, and 0.016 μM) was dispensed using liquid-handling robotics, and cell viability for each cell was assessed using CellTiter-Glo (Promega, Madison, WI, USA) after 4 days with drug incubation. The plates were agitated for 30 min at room temperature before measuring luminescence. The half-maximal inhibitory concentration (IC50) values were determined using GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA, USA). mRNA sequencing (RNA-seq) analysis from the cBioPortal dataset Data for the RNA-seq analysis were retrieved from the cBioPortal for Cancer Genomics, an online data repository (https://www.cbioportal.org/). A total of 11 NSCLC patients with METex14 were included in the cBioPortal cohort (Supplementary Table 1). These patients had both mRNA expression profiling and gene mutation data available. All patients were in the early stages of the disease. The proportions of immune cells in NSCLC patients were analyzed based on this RNA-seq data. Follow-up and statistical analysis Progression-free survival (PFS) was defined as the duration from the initiation of treatment to the confirmation of disease progression or death from any cause. Patients without an event were censored at the last follow-up. Survival curves were estimated using the Kaplan–Meier method, and between-group differences were compared using the log-rank test. Categorical baseline variables were compared using the Fisher's exact test, and continuous variables were compared using the two-sample t-test. A stratified analysis by treatment line (first-line vs. second-or-later lines of MET-TKI treatment) was performed to assess the potential confounding effect of treatment line on efficacy outcomes. All statistical analyses and figures were generated using the R software (version 4.3.1). The last follow-up was conducted on May 20, 2025. Results Patient clinical characteristics Between July 1, 2016 and April 30, 2025, we identified 24 METex14-mutated NSCLC patients who received MET TKI therapy at Guangdong Provincial People's Hospital, including four participants enrolled in the Phase II KUNPENG clinical trial (NCT04258033, ClinicalTrials.gov, registered on February 06, 2020). The detailed clinical information is shown in Fig. 1A. Half of the patients received crizotinib, while the remaining half were treated with vebreltinib. The average age of all patients at diagnosis was 65.8 years (range, 54–83 years). There were 12 female and 12 male patients. Nineteen of them had adenocarcinomas, and 17 were nonsmokers. The cohort comprised 22 patients (91.7%) with stage IV disease at diagnosis. The remaining two patients (8.3%) had locally advanced disease (one each with stage IIIA and IIIB) and were deemed ineligible for curative-intent surgery or definitive radiotherapy. Thus, all enrolled patients were classified as having advanced-stage NSCLC. At the time of MET targeted therapy, 17 patients had an Eastern Cooperative Oncology Group performance status (ECOG-PS) score of 1, while the remaining three had an ECOG-PS score of 2. MET TKI treatment was administered as first-line therapy in 15 patients, second-line in 7 patients, and third-line in 2 patients. Treatment outcomes in METex14-altered patients. A Clinical characteristics and treatment responses of the patient cohort. B Kaplan–Meier analysis comparing progression-free survival (PFS) between vebreltinib- and crizotinib-treated groups. NSCLC non-small cell lung cancer, METex14 MET exon 14 skipping mutation, ADC adenocarcinoma carcinoma, PSC pulmonary sarcomatoid carcinoma, amp amplification, PFS progress-free survival, PR partial response, SD stable disease, PD progressive disease, NA not applicable, —not available, P02/P24 locally advanced NSCLC, * event indicator: 1 PD or death, 0 censored data At baseline, 5 patients (20.8%) had central nervous system (CNS) metastases, including 2 patients in the Vebreltinib arm and 3 in the Crizotinib arm. During follow-up, CNS progression occurred in 1 patient in the Vebreltinib arm and 2 patients in the Crizotinib arm; only one patient in the Vebreltinib arm achieved intracranial stable disease with no intracranial objective responses observed in either cohort (Supplementary Table 2). MET TKI treatment response in METex14 NSCLC The overall median PFS (mPFS) for MET TKI-treated patients was 7.0 months (95% CI: 3.6, 25). Crizotinib group had mPFS 6.1 months and 41.7% (5/12) patients experienced early progression (< 3 months). The vebreltinib group demonstrated mPFS of 8.7 months, with 50% (6/12) of patients achieving PFS exceeding 7 months. However, no significant differences were observed between two groups (p = 0.42, Fig. 1B). Exploratory analysis also failed to identify a statistically significant association between any co-occurring genetic alteration and PFS. The DCR was 66.7% (8/12) for crizotinib and 91.7% (11/12) for vebreltinib. No statistically significant difference was observed between the two treatments. According to RECIST v1.1 criteria, the best overall response to MET TKI treatment included 9 PR, 10 cases of stable disease (SD), and 5 instances of progressive disease (PD). There was no significant difference among the subgroups, including age, gender, pathology, stage, smoking status, PS score, lines of MET-TKI treatment, best response rate, etc. (Table 1). Treatment trajectories and durations of response for all patients are illustrated in Supplementary Fig. 1A. Table 1Patient subgroup analysesALLCrizotinibVebreltinibp.overallN = 24N = 12N = 12Age65.8 (7.93)65.3 (9.37)66.2 (6.58)0.784Gender0.683 Female12 (50.0%)5 (41.7%)7 (58.3%) Male12 (50.0%)7 (58.3%)5 (41.7%)Pathology1.000 ADC*19 (79.2%)9 (75.0%)10 (83.3%) PSC*2 (8.33%)1 (8.33%)1 (8.33%) SCC*1 (4.17%)1 (8.33%)0 (0.00%) NA*2 (8.33%)1 (8.33%)1 (8.33%)Smoking1.000 No17 (70.8%)8 (66.7%)9 (75.0%) Yes7 (29.2%)4 (33.3%)3 (25.0%)Stage1.000 III2 (8.33%)1 (8.33%)1 (8.33%) IV22 (91.7%)11 (91.7%)11 (91.7%)PS score*0.371 117 (70.8%)7 (58.3%)10 (83.3%) 27 (29.2%)5 (41.7%)2 (16.7%)PD-L1% ≧50%10 (41.7%)4 (33.3%)6 (50.0%)0.925 < 506 (25.0%)3 (25.0%)3 (25.0%) Negative2 (8.33%)1 (8.33%)1 (8.33%) NA6 (25.0%)4 (33.3%)2 (16.7%)Lines of MET-TKI1.000 115 (62.5%)8 (66.7%)7 (58.3%) 27 (29.2%)3 (25.0%)4 (33.3%) 32 (8.33%)1 (8.33%)1 (8.33%)PFS/m0.069 < 37 (29.2%)6 (50.0%)1 (8.33%) ≧317 (70.8%)6 (50.0%)11 (91.7%)Best response0.100 PR*9 (37.5%)2 (16.7%)7 (58.3%) SD*10 (41.7%)6 (50.0%)4 (33.3%) PD*5 (20.8%)4 (33.3%)1 (8.33%)*: ADC adenocarcinoma carcinoma, PSC pulmonary sarcomatoid carcinoma, SCC squamous cell carcinoma, NA not applicable, PS performance status, PR partial response, SD stable disease, PD progressive disease Patient subgroup analyses *: ADC adenocarcinoma carcinoma, PSC pulmonary sarcomatoid carcinoma, SCC squamous cell carcinoma, NA not applicable, PS performance status, PR partial response, SD stable disease, PD progressive disease A stratified analysis by treatment line revealed no significant differences in PFS between vebreltinib and crizotinib in either the first-line (HR = 0.11, 95% CI: 0.32–3.89, P = 0.863) or second/later subgroups (HR = 0.96, 95% CI: 0.43–15.83, P = 0.296; Supplementary Fig. 1B). The overall HR for PFS was 0.36 (95% CI: 0.57–3.65, P = 0.445), consistent with the stratified findings. Owing to the small sample size in each subgroup, the confidence intervals were wide, and no definitive conclusions could be drawn from the subgroup analyses. Patient 14 exhibited an extremely rare molecular profile with concurrent METex14 skipping variant, ALK fusion, and ROS1 fusion. The patient received first-line crizotinib therapy and achieved a durable progression-free survival of 23.3 months before disease progression, which was notably longer than the median PFS reported for crizotinib in patients with METex14 alteration (7.3 months), ALK fusion (10.9 months), or ROS1 fusion (15.9 months) [8, 21–23]. Notably, two patients (P05 and P19) underwent baseline plasma NGS only because of inaccessible tissue specimens, with METex14 VAF values of 49.84% and 2.2%, respectively. These results should be interpreted with caution. MET expression levels appear to be unrelated to MET TKI response MET expression levels in patients P01, P02, and P23 were assessed through IHC (Fig. 1A). Although all three patients exhibited high levels of MET expression, their responses to MET TKI treatment varied. Patients P01 and P02 had a PFS of 70 and 25 months, respectively, whereas patient P23 did not benefit from MET TKI treatment. Patients P01 and P02 achieved durable clinical benefits with long progression‑free survival. In contrast, patient P23 exhibited primary resistance and no clinical benefit from MET‑TKI treatment. These results demonstrate that high MET protein expression alone is not a reliable predictive marker for response to MET‑TKI therapy in METex14 NSCLC. Molecular profiling of genetic alterations in METex14 in NSCLC All patients underwent METex14 mutation testing and were diagnosed METex14 before MET TKI treatment. Among them, 20 patients had available molecular profiling reports, including either a 14-gene driver panel or more comprehensive genomic testing. In addition to METex14 alterations, co-occurring genomic aberrations were identified, including EGFR amplification, ALK rearrangements, NRAS mutations, PIK3CA mutations, and ROS1 alterations (Fig. 2). As expected, TP53 mutation (45%) was the most popular co-occurred mutant gene (Supplementary Table 3). Genetic variants of METex14 and co-mutant genes. Variable genetic variants of MET exon 14 skipping are color-coded. Somatic mutation characteristics of NSCLC cases with METex14. Each column represents one patient. Alteration types are color-coded and shown on the right panel METex14 mutations demonstrated diverse genetic variants, with intronic base substitution adjacent to splice donor site being the most prevalent (40.0%). Other variant types included intronic indel adjacent to splice acceptor site, intronic del adjacent to splice acceptor site, intronic base substitution adjacent to splice acceptor site, and base substitution at splice donor site. No significant correlation was observed between specific METex14 variant types and clinical outcomes. No definitive correlation between co-occurring genetic alterations and PFS could be identified owing to the small sample size and high interpatient molecular heterogeneity. Switching from type Ia to Ib MET TKI may overcome drug resistance For patients receiving consecutive MET TKIs (n = 2), only the first agent was included in the PFS analysis comparing crizotinib versus vebreltinib (Fig. 1). Patients P17 and P23 received two MET TKI treatments (Fig. 3A and B). P17 received crizotinib as first-line treatment and had a PFS of 6.4 months, with the best response being SD. After the failure of MET TKI and subsequent chemotherapy, P17 received savolitinib as the fourth-line treatment. Surprisingly, this resulted in a 14-month PFS, with the best response being SD (Fig. 3A). Sequential MET-targeted therapy in METex14-altered patients. A Corresponding treatment history and response patterns for Patient P17. B Treatment timeline of P23 with correlative PDO drug response following MET inhibition. METex14 MET exon 14 skipping mutation, PR partial response, SD Stable disease, bev bevacizumab, AP pemetrexed + carboplatin Patient P23 was also treated with crizotinib as first-line therapy but did not respond and progressed rapidly within 3 weeks. As third-line treatment, P23 received vebreltinib plus bevacizumab, achieving PR as the best outcome. Treatment with this second MET TKI (vebreltinib) yielded a PFS of 5 months (Fig. 3B). Pathological transformation is a potential mechanism of resistance to MET TKIs Six patients underwent genetic testing after MET TKI failure, paired baseline and progression samples were available for these cases. An acquired MET D1228N mutation was identified in P02, and P17 was observed to be acquired EGFR amplification. The MET D1228N mutation in patient P02 and EGFR amplification in patient P17 were not detected at baseline, indicating that these were acquired resistance mechanisms following MET-TKI therapy. P18, who initially had adenocarcinoma histology prior to MET TKI treatment, subsequently transformed into pulmonary sarcomatoid carcinoma following MET TKI therapy failure (Fig. 4A). Two tissue specimens from this patient 18 were pathologically examined at different time points (Supplementary Fig. 2). The 2018 specimen showed partial positivity for NapsinA and TTF-1 (adenocarcinoma-associated markers) and negativity for CK5/6 and P40 (squamous/sarcomatoid-associated markers), leading to a diagnosis of poorly differentiated adenocarcinoma. In contrast, the 2019 specimen exhibited partial CK positivity (epithelial differentiation), focal CK5/6 and P63 positivity (squamous differentiation features), and negativity for TTF-1 and NapsinA (excluding adenocarcinoma); combined with sarcomatoid morphological features (pleomorphic nuclei, giant tumor cells) on H&E staining, the diagnosis was sarcomatoid carcinoma (Supplementary Fig. 2). These findings reflect the pathological transformation of the tumor over time. Genetic re-sequencing of post-progression specimens from Patients P01, P06, and P23 failed to identify established molecular mechanisms of MET TKI resistance, suggesting alternative resistance pathways may be involved. Pathological transformation post targeted therapy and treatment responses in METex14-altered patients. A Histologic transformation from adenocarcinoma to pulmonary sarcomatoid carcinoma in P18 after crizotinib failure. B Therapeutic timeline and clinical outcomes for P05. METex14 MET exon 14 skipping mutation, PR partial response, SD Stable disease, bev bevacizumab, AP pemetrexed + carboplatin, scale bar 100μm High PD-L1 expression in METex14 NSCLC appears to be favorable for immunotherapy PD-L1 expression analysis was performed in 18 patients (Fig. 1A and Table 1). The distribution of PD-L1 expression level was as follows: 10 patients (55.6%) showed high expression (TPS ≥ 50%), 6 patients (33.3%) demonstrated low expression (1–49%), and 2 patients (11.1%) were negative (TPS < 1%). Among the cohort, only two patients (P05 and P17) had received immune checkpoint inhibitor therapy (Figs. 3A and 4B). Patient P05, with high PD-L1 expression (TPS 85%), received second-line camrelizumab plus endostar following MET TKI failure, achieving a PFS of 13 months with a PR as best response (Fig. 4B). In contrast, patient P17 (PD-L1 negative, TPS < 1%) was treated with fifth-line pembrolizumab plus docetaxel, demonstrating a PFS of 3 months with SD as best response (Fig. 3A). To investigate the immune microenvironment in METex14 NSCLC, we analyzed RNA-seq data from 11 METex14 NSCLC patients in TCGA (Supplementary Table 1). PD-L1 expression levels were classified into high and low groups based on the median expression value. By transcriptomic analysis, we observed that patients with high PD-L1 expression showed: significantly increased infiltration of activated memory CD4+ T cells and M2 macrophages and markedly reduced proportions of plasma cells (p < 0.05) (Fig. 5A). Furthermore, the high PD-L1 group demonstrated enhanced infiltration of immunologically active cell populations, including CD8+ T cells, M2 macrophages, and activated mast cells (Fig. 5B). These findings suggest that METex14 NSCLC tumors with elevated PD-L1 expression maintain a more immunologically active tumor microenvironment. Immune microenvironment characterization in METex14-altered patients using TCGA RNA-sequencing data. A Differential immune cell infiltration associated with PD-L1 mRNA-high tumors, showing significant variations in plasma cells, activated memory CD4+ T cells, and M2 macrophage populations. B Integrated heatmap analysis of immune cell composition correlated with clinical variables (age, gender) and PD-L1 expression levels. ** p < 0.01, ADC adenocarcinoma carcinoma, SCC squamous cell carcinoma PDOs hold promise for predicting clinical efficacy In this study, only patients P06, P22, and P23 had enough samples and were successful established PDO cultures for the further experiment (Figs. 6A, 6B and 3B). After resistance to MET TKIs, pleural effusions from P06 and P23 were used for PDO cultures. The dose–response curves for P06 PDOs treated with anlotinib, cabozantinib, and imatinib are shown in Fig. 6A. All IC50 values were above 10 μM, indicating drug resistance. Similarly, anlotinib and cabozantinib were tested in P23 PDO cultures, and both showed very high IC50 values (Fig. 3B). Additionally, cabozantinib combined with cetuximab or 10,058-F4 (a novel c-Myc TKI) did not overcome drug resistance in P23 PDO cultures. Consistent with the PDO results, P06 and P23 rapidly progressed after anlotinib treatment. Treatment-naïve P22 tissues were obtained to establish PDO cultures, which treated with crizotinib or savolitinib. The results showed an IC50 of 4.9 μM for crizotinib, indicating moderate sensitivity. Since moderate sensitivity does not predict robust clinical benefit, this finding is consistent with the patient’s lack of response to first-line crizotinib treatment (Fig. 6B). However, these findings are based on only 3 cases and should be considered preliminary. Drug sensitivity test in patient-derived organoid. A and B showed the dose–response curves of patient-derived organoid cultures from patient P06 and P22 respectively Safety Safety and tolerability were evaluated in all 24 patients who received MET-TKI treatment. One patient in each treatment arm developed a grade 3/4 treatment-related ALT elevation that required dose interruption or temporary discontinuation of the MET-TKI. No other grade ≥ 3 toxicities or treatment discontinuations due to adverse events were reported (Supplementary Table 2). Discussion We observed that METex14 mutations were more frequently identified in older NSCLC patients, which is consistent with previous reports [4, 10]. The clinical and functional significance of different METex14 splicing variants has not been fully characterized or validated [15, 17]. In the present study, five genetic variants of METex14 were identified. Our results showed no significant relationship between the METex14 variants and PFS with MET TKI treatment. Further research is required to explore this relationship. The PROFILE 1001 clinical trial revealed a 7.3-month PFS for crizotinib in METex14 NSCLC patients [8]. Similarly, the mPFS after MET TKI therapy among the 24 patients included in our study was 7.0 months. A phase I study of vebreltinib in MET mutated NSCLC patients showed a 66.7% ORR [11]. The subsequent phase II study revealed a 14.5-month median PFS in 35 treatment-naïve patients and a 7.7-month median PFS in previously treated MET altered patients [12, 24]. Our study revealed distinct patterns of treatment response between the two MET inhibitors: six vebreltinib-treated patients (50%) achieved PFS exceeding 7 months, including one exceptional responder with 70-month PFS. Conversely, almost half of crizotinib-treated patients (5/12) developed rapid disease progression, suggesting frequent primary resistance to crizotinib in METex14-mutated NSCLC. Notably, despite these divergent response patterns, the overall PFS comparison between crizotinib (median 6.1 months) and vebreltinib (median 8.7 months) did not reach statistical significance. Notably, baseline imbalances were observed between treatment groups: the crizotinib cohort included more patients with ECOG PS = 2 and showed a trend toward more frequent early progression (< 3 months). These differences may have influenced the observed between-group differences in PFS. These baseline differences should be considered when interpreting the comparative efficacy results. Of note, patient P19 exhibited a low plasma VAF of 2.2% for the METex14 skipping alteration, which may reflect low circulating tumor DNA abundance in peripheral blood. Duo to plasma-based genotyping, the findings should be carefully interpreted. Of note, Patient 14 in our cohort harbored the unprecedented concurrent METex14 skipping variant, ALK fusion, and ROS1 fusion, and received first-line crizotinib with an exceptional PFS of 23.3 months. Crizotinib is a first-generation multi-TKI targeting MET, ALK, and ROS1, and is recommended as first-line standard therapy for both ALK- and ROS1-rearranged NSCLC. Emerging evidence also supports the clinical activity of crizotinib in METex14-altered NSCLC. The concurrent presence of three distinct actionable drivers in this patient is extremely rare, with limited real-world data available. Notably, the 23.3-month PFS observed herein is substantially longer than the median PFS of 7.3 months for METex14-mutant NSCLC, 10.9 months for ALK-positive NSCLC, and 15.9 months for ROS1-positive NSCLC treated with single-agent crizotinib. This remarkable outcome suggests that crizotinib’s simultaneous inhibition of all three oncogenic drivers may underlie the durable disease control, highlighting the potential value of multi-target inhibition in patients with overlapping oncogenic alterations. This rare case underscores the complexity of molecular profiling in NSCLC and the clinical benefit of selecting a multi-target agent that covers all identified actionable alterations in the first-line setting. In our study, MET IHC from three patients showed high MET expression; however, their responses to MET inhibitors varied. Previous reports have indicated that the clinical response to MET TKI treatment is independent of MET expression levels [25, 26]. Moreover, high MET expression levels did not predict the presence of METex14 mutations [26]. Therefore, MET IHC is a poor predictor of MET-targeted therapy efficacy [26]. MET TKI resistance often arises from on-target genomic alterations (such as secondary MET kinase domain mutations and MET amplification) and off-target mechanisms (including ERBB family receptor tyrosine kinase gene amplification, MAPK pathway gene amplification, and KRAS mutations) [27–29]. On-target and off-target genetic changes did not exhibit significantly different effects on the MET TKI response [29]. We consistently identified the acquisition of a secondary MET kinase domain mutation, D1228N, and EGFR amplification in some patients after MET TKI failure. Sarcomatoid transformation has been reported as a potential resistance mechanism to tyrosine kinase inhibitor treatments, such as EGFR and ROS1 inhibitors [30]. We observed that one patient initially diagnosed with adenocarcinoma transformed to sarcomatoid carcinoma following MET-targeted therapy failure. Jamme et al. reported that patients with both METex14 mutations and co-alterations in the PI3K pathway exhibited primary resistance to MET TKIs and that PI3K inhibitors could overcome this drug resistance in vitro [31]. Conversely, in our study, a patient with both METex14 mutation and PIK3CA H1047R co-mutation achieved an 6.1-month PFS with MET TKI therapy. Therefore, further studies with larger sample sizes are required to elucidate molecular resistance mechanisms. Given the small cohort size, the exploration of resistance mechanisms remains preliminary and requires further validation in larger cohorts. To overcome MET TKI resistance, recent therapeutic strategies have focused on sequential or combinatorial approaches [29, 31, 32], including switching between type I and II MET TKIs [16, 29]. Although crizotinib, vebreltinib, and salvolitinib are type I MET TKIs, our study demonstrated that type Ib MET TKI (vebreltinib and salvolitinib) can overcome type Ia resistance (crizotinib). Overall, different types of MET TKIs are promising for a second attempt at MET-targeted therapy in heavily treated METex14 mutant patients after initial MET TKI failure. Of note, even after multiple lines of MET-TKI therapy, re-challenge with an alternative type Ib MET TKI may still provide durable clinical benefit, as observed in patient P17. PD-L1 expression serves as a predictive biomarker for immunotherapy, with survival advantages observed in groups with higher PD-L1 expression [33]. Previous research indicated that patients with METex14 patients might derive less benefit from immunotherapy compared with those harboring MET ampplication [34]. In this study, METex14 NSCLC patients with high PD-L1 expression levels did benefit from immunotherapy following MET TKI failure. From the RNA-seq analysis, higher PD-L1 expression was associated with a higher proportion of immune-active cells, such as activated memory CD4+ T cells, M2 macrophages, CD8+ T cells, and mast cells. Conversely, there were fewer immunosuppressive cells, like plasma cells. All this evidence supports the existence of a favorable immune microenvironment. Thus, immunotherapy has been suggested as an promising and accepted treatment option for advanced-stage METex14 NSCLC patients, especially those with high PD-L1 expression. In the TCGA cohort, PD-L1 high and low subgroups were stratified by median mRNA expression, which is not equivalent to clinically used PD-L1 TPS based on IHC. mRNA levels do not directly correspond to protein expression, and this important limitation should be considered when interpreting these findings. PDOs represent a promising approach for predicting clinical response to chemotherapy or targeted treatments in lung cancer patients. Our previous studies have demonstrated a high success rate in establishing PDO cultures, with substantial concordance in treatment response [20]. In our study, PDO cultures were successfully established for three patients, and the majority of the therapeutic regimens tested in PDOs exhibited drug resistance, consistent with clinical observations. Thus, PDOs are highly recommended as predictive tools for the treatment response in cancer. The optimal drug choice should prioritize the specific characteristics of each patient, including genetic mutations and other relevant factors. The drug sensitivity results from PDOs could guide therapeutic decisions and inform clinical treatment regimens. PDO cultures also offer a promising model for testing new drugs prior to clinical trials and study the tumor immune microenvironment [35]. It is possible that PDO drug sensitivity may eventually serve as a key principle in clinical drug diagnosis and individualized treatment plans. However, patient-derived organoids were successfully established in only 3 patients, which represents a major limitation of this study. Therefore, the preliminary findings should be interpreted with caution, and the predictive value of PDOs requires further validation in larger cohorts. Overall, PDOs represent a promising tool for advancing precision medicine and standardizing clinical approaches in the future. Our study has certain limitations. First, the single-center, retrospective study included few participants. Second, the different lines and drugs of MET TKIs treatment were biased in subsequent analysis. Third, owing to insufficient samples and difficulties in the technique for PDO cultures, few drugs were selected for testing. The results of the PDO cultures need to be carefully interpreted. Fourth, financial issues and the low incidence of METex14 in NSCLC patients limited the study group size and perspective. Fifth, only DNA-based sequencing was used to identify METex14, without RNA-based confirmation. Although highly dependent on RNA quality, RNA-based analysis showed a higher proportion of METex14 detection than DNA-based analysis [15]. Further studies with larger sample sizes and in vitro models are required to validate our findings. Sixth, the immune microenvironment analysis based on the TCGA database was limited to only 11 patients with METex14 alterations, owing to the paucity of publicly available samples with matched RNA-seq and clinical data. Furthermore, the TCGA cohort mainly comprised early-stage NSCLC patients, whereas our clinical cohort focused on advanced-stage (III–IV) disease. Accordingly, these transcriptomic findings should be considered exploratory, interpreted with caution, and further validated in larger, stage-matched, independent cohorts. Conclusion Although vebreltinib demonstrated numerically longer mPFS and higher DCR compared to crizotinib (8.7 vs 6.1 months and 91.7% vs 66.7%, respectively), these differences did not reach statistical significance in the treatment of METex14. Transformation to pulmonary sarcomatoid carcinoma is a potential resistance mechanism following MET-targeted therapy, highlighting the need for further investigation into overcoming resistance. METex14-mutated NSCLC patients with high PD-L1 expression may benefit from immune checkpoint inhibitors combined with anti-angiogenic therapy, suggesting a promising direction for personalized treatment strategies. PDOs hold significant potential for predicting clinical efficacy, offering a powerful tool to guide treatment selection and improve patient outcomes. As MET-targeted therapies evolve, these findings provide a foundation for future research, contributing to the development of more effective, individualized therapeutic approaches and advancing precision medicine in METex14-mutated NSCLC. Clinical trial registration Clinical trial number of this study: not applicable. The clinical trials referenced in the manuscript (NCT01639508 and NCT02920996) are external studies registered at ClinicalTrials.gov and are cited for background purposes only. Supplementary Information Additional file 1. Supplementary Fig. 1. Swimmer plot of treatment trajectories and forest plot of PFS stratified by treatment line. Additional file 2. Supplementary Fig. 2. Representative IHC staining images of tumor tissues from Patient 18. Additional file 3. Supplementary Table 1. Patients information from TCGA database. Additional file 4. Supplementary Table 2. Patients basic information. Additional file 5. Supplementary Table 3. METex14 variants and co-mutatant genes. Acknowledgements All authors are grateful to the patients and their families. The authors also appreciate the assistance provided by the staff at the Guangdong Lung Cancer Institute, Guangdong Provincial People’s Hospital, Guangdong Academy of Medical Sciences. Jin-Ji Yang—Lead contact. Author contributions Conceptualization: J.J.Y., and Y.G.; Methodology: J.J.Y., G.L., and H.L.; Data collection: J.Y., C.D.W., Z.J.M., and H.M.W.; Investigation: J.Y., H.M.W., J.W.S, Z.J.M., and Q.Q.T.; Visualization: H.L., J.Y., and J.J.Y.; Project administration: Y.L.W., J.J.Y., and H.J.C.; Validation: Y.L.W., J.J.Y., H.J.C., X.C.L., and M.M.F; Supervision: Y.L.W., J.J.Y., Y.G., and H.J.C.; Writing—original draft: J.Y., C.D.W., and H.M.W.; Writing—review & editing: Y.L.W., J.J.Y., and Y.G. Funding This study was funded by the High-level Hospital Construction Project of Guangdong Provincial People'sHospital, China [Grant Number DFJH201809, to Dr. Jin-Ji Yang]; the Natural Science Foundation of Guangdong Province, China [Grant Number 2019A1515010931, to Dr. Jin-Ji Yang]; the National Natural Science Foundation of China [Grant Number 81972164, to Dr. Jin-Ji Yang]; the Key Lab System Project of Guangdong Science and Technology Department—Guangdong Provincial Key Laboratory of Translational Medicine in Lung Cancer, China [Grant Number 2017B030314120, to Dr. Yi-Long Wu] and the Postdoctoral Foundation of Guangdong Provincial People's Hospital, China [Grant Number BY012022043, to Dr. Jiao Yang]. This research did not receive any other specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data availability Raw data of DNA/RNA sequencing was not applicable. Processed data is provided within the manuscript or supplementary information files (Supplementary Table 1). Declarations Ethics approval and consent to participate This study was conducted according to the principles of the Declaration of Helsinki and was approved by the Ethical Review Board: Guangdong Provincial People’s Hospital, and Guangdong Academy of Medical Sciences (reference number: KY-Z-2021—394—02). Informed consent to participate was obtained from the patients for the study. Consent for publication Patients signed informed consent regarding publishing their data. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jiao Yang, Cheng-Di Weng and Han-Min Wang contributed equally. Contributor Information Yi-Long Wu, Email: syylwu@live.cn. Yubiao Guo, Email: guoyb@mail.sysu.edu.cn. Jin-Ji Yang, Email: yangjinji@gdph.org.cn. References Associated Data Supplementary Materials Additional file 1. Supplementary Fig. 1. Swimmer plot of treatment trajectories and forest plot of PFS stratified by treatment line. Additional file 2. Supplementary Fig. 2. Representative IHC staining images of tumor tissues from Patient 18. Additional file 3. Supplementary Table 1. Patients information from TCGA database. Additional file 4. Supplementary Table 2. Patients basic information. Additional file 5. Supplementary Table 3. METex14 variants and co-mutatant genes. Data Availability Statement Raw data of DNA/RNA sequencing was not applicable. Processed data is provided within the manuscript or supplementary information files (Supplementary Table 1).