Introduction

Worldwide, approximately 1.8 million deaths annually are attributed to lung cancer, which remains the leading cause of cancer-related mortality1,2. NSCLC accounts for approximately 85% of all lung cancer cases, and its global incidence continues to rise3,4. Treatment outcomes for NSCLC remain poor, with five-year survival rates declining sharply from 68% in stage IB to just 21% in stage IIIC. Moreover, most advanced-stage NSCLC cases acquire resistance to existing therapies, leading to disease progression5,6.

NSCLC comprises a genetically heterogeneous group of malignancies driven by various oncogenic alterations7. Several recurrent and potentially targetable oncogenes have been identified in NSCLC, including mutations in epidermal growth factor receptor (EGFR) and v-raf murine sarcoma viral oncogene homolog B1 (BRAF), as well as chromosomal rearrangements involving ROS proto-oncogene 1 receptor tyrosine kinase (ROS1) and anaplastic lymphoma kinase (ALK)8,9.

The c-MET protein, encoded by the MET proto-oncogene, is a receptor tyrosine kinase expressed on cell membranes10. Constitutively altered MET pathway activation strongly propels malignant transformation, stimulates tumor cell progression as well as angiogenesis, increases tumor invasiveness as well as metastatic potential, and mediates resistance to MET tyrosine kinase inhibitors (TKIs)10,11. MET signaling dysregulation typically arises through three distinct mechanisms: protein overexpression, MET gene amplification, and exon 14 skipping mutations (METex14)11. In NSCLC, MET exon 14 skipping mutations are established oncogenic drivers, whereas MET amplification and protein overexpression are generally associated with poorer prognosis12,13,14,15,16,17. It should be noted that MET gene fusions, though rare and actionable, were not included in the present analysis due to insufficient data in the identified studies.

Given the extensive involvement of aberrant MET signaling in NSCLC pathogenesis and therapeutic resistance, MET has emerged as a promising therapeutic target. Over the past decade, several MET-targeted therapies—including TKIs and antibody-drug conjugates (ADCs)—have entered clinical evaluation18,19,20,21,22,23,24. MET-targeting TKIs are classified into two major categories: Type I inhibitors primarily engage the ATP-binding site of the c-MET kinase domain in a characteristic “U-shaped” binding orientation and are particularly active against METex14 mutations25. Type II inhibitors adopt an “extended” binding conformation, interacting with both the ATP pocket and an adjacent hydrophobic pocket. Although they may retain activity against certain resistance mutations that render Type I inhibitors ineffective, Type II inhibitors often exhibit reduced selectivity and increased off-target effects26.

Tepotinib, a highly selective and potent Type Ib MET inhibitor, has received first-line approval for the treatment of NSCLC patients harboring METex14 skipping mutations27,28. Long-term follow-up data from the VISION trial recently demonstrated an ORR of 57.3% and a median PFS of 12.6 months in treatment-naïve patients29.

Nevertheless, the existing evidence for Tepotinib has important limitations. Most available data derive from single-arm Phase II studies, such as VISION, and small-scale trials. These studies inherently lack control arms and include limited patient numbers19,29,30.

This systematic review evaluates the clinical efficacy and safety of Tepotinib in NSCLC patients harboring MET amplification or METex14 skipping mutations. This work aims to delineate the therapeutic benefit across these distinct molecular subgroups and thereby provide robust evidence to inform clinical decision-making.

Results

Literature screening results

The study selection process is illustrated in Fig. 1. A total of 739 potentially eligible studies were identified using predefined search strategies. After removing 240 duplicates, we performed title and abstract screening, which excluded a further 462 records for irrelevance, non-English articles, or wrong formats. Full-text scrutiny then excluded a further 30 studies for incomplete outcomes, continuing trials, lack of relevant outcomes, or low patient numbers. Six studies, after meeting inclusion criteria, were included in the meta-analysis.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

PRISMA flow chart of study selection process.

Basic characteristics

Table 1 provides an overview of characteristics derived from the selected studies. These studies, appearing between 2022 and 2025, collectively encompassed a total of 546 patients. Most studies were Phase II clinical trials29,33,34,36,39, accompanied by one retrospective study40. All patients had metastatic or advanced NSCLC with MET alterations, including amplification and/or exon 14 skipping mutations. All patients enrolled presented with metastatic or advanced-stage NSCLC. Female patient representation ranged significantly from 12.5% to 75.0%, with average patient age varying from 59.3 to 74.0 years. The majority demonstrated an Eastern Cooperative Oncology Group (ECOG PS) performance status of 0 or 1. Smoking history varied, including current, former, and never-smokers. Tepotinib monotherapy (500 mg once daily) was used in most studies. One study investigated Tepotinib combined with gefitinib36, and another examined Tepotinib combined with osimertinib34.

Table 1 Baseline clinical characteristics of included studies.
Full size table

Quality assessment

Assessment of non-randomized studies’ methodological rigor was conducted using the Methodological Index for Non-randomized Studies (MINORS) scale35, and detailed findings are summarized in Supplementary Table S2. The MINORS scale was selected because it is specifically validated for assessing the methodological quality of non‑comparative (single‑arm) interventional studies, which constituted the majority of our included trials. While ROBINS‑E is designed for exposure studies, MINORS better aligns with the interventional nature of our meta‑analysis. The lone randomized controlled trial included underwent evaluation through the Cochrane Risk of Bias Tool (RoB 2)36,37, revealing a high overall bias risk; detailed results for this assessment are provided in Supplementary Table S3.

Analysis of efficacy

The primary endpoint was the ORR, with secondary outcomes including DCR, DOR, PFS, and OS. Random-effects modeling was applied under conditions of notable heterogeneity; otherwise, fixed-effects modeling was utilized. The pooled analysis showed an ORR of 52% (95% CI: 48%–56%; I2 = 34.4%), a DCR of 76% (95% CI: 72%–80%; I2 = 37.6%), a mean DOR of 12.62 months (95% CI: 8.73–18.24 months; I² = 91.1%), a mean PFS of 10.16 months (95% CI: 7.60–13.58 months; I2 = 72.8%), and a mean OS of 14.67 months (95% CI: 9.60–22.43 months; I2 = 93.5%). It should be noted that this pooled ORR includes both treatment‑naive and previously treated patients, as the included studies did not consistently stratify outcomes by line of therapy. Significant heterogeneity was observed for DOR, PFS, and OS, while ORR and DCR exhibited low to moderate heterogeneity. The efficacy analysis results are shown in Figs. 2, 3, 4 and 5, and 6.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Forest plot of ORR.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Forest plot of DCR.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Forest plot of mDOR.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Forest plot of mOS.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

Forest plot of mPFS.

Subgroup analysis

Subgroup analysis by MET dysregulation type revealed an ORR of 53% (95% CI: 45%–60%) for MET amplification and 52% (95% CI: 47%–57%) for MET exon 14 skipping, without significant differences between subgroups (χ² = 0.01, p = 0.9054) (Fig. 7). Subgroup analysis by therapy type showed an ORR of 51% (95% CI: 46%–56%) for monotherapy and 56% (95% CI: 48%–64%) for combination therapy (i.e., Tepotinib plus an EGFR‑TKI such as gefitinib or osimertinib), also without significant differences (χ² = 1.37, p = 0.2421) (Fig. 8).

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

Forest plot of ORR in subgroup analysis on MET dysregulation.

Fig. 8
Fig. 8The alternative text for this image may have been generated using AI.
Full size image

Forest plot of ORR in subgroup analysis on therapy type.

AEs

Safety evaluations of AEs were performed for all patients who received Tepotinib, but as a result of incomplete response, certain patients were excluded from the efficacy analyses. Amongst six studies that were included, there were five studies involving 511 patients with 449 treatment-related adverse event (TRAE) reports, with most TRAE being mild in nature. One study36 failed to report relevant TRAE data. Adverse events that were attributed to Tepotinib were graded by criteria established by the National Cancer Institute (National Toxicity Criteria for Adverse Events), with a detailed description given in Table 2. Pooled incidence rates for TRAEs are summarized in Table 3, indicating that most TRAEs were categorized as Grade 1–2. The most frequently reported any-grade TRAEs included peripheral edema (50%, 95% CI: 34%–65%), diarrhea (36%, 95% CI: 18%–54%), hypoalbuminemia (22%, 95% CI: 18%–25%), nausea (22%, 95% CI: 18%–26%), and elevated blood creatinine levels (18%, 95% CI: 10%–25%). Increases in alanine aminotransferase and aspartate aminotransferase were each reported in 13% of patients. Severe (Grade ≥ 3) TRAEs were uncommon, with peripheral edema (8%, 95% CI: 6%–11%) being the most frequent. Grade ≥ 3 events were uncommon (2% for hypoalbuminemia and ALT elevation) and were not reported for diarrhea, creatinine increase, or AST elevation. Estimates of other TRAEs showed low to moderate heterogeneity.

Table 2 Adverse events in evaluated NSCLC patients.
Full size table
Table 3 Pooled results of common TRAEs of any grade and ≥ grade.
Full size table

Sensitivity analysis

Sensitivity analysis using the leave-one-out method assessed the robustness of study findings. Each study was sequentially removed, and pooled estimates were recalculated to verify stability. Removing individual studies did not significantly alter the overall results, confirming the reliability of findings. Results of sensitivity analyses for ORR, DCR, DOR, PFS, and OS are presented in Figures S2, S3, S4, and S5, respectively.

Discussion

This systematic review and meta-analysis comprehensively evaluated the efficacy of Tepotinib in 546 patients with MET-altered advanced NSCLC. The analysis confirmed prominent clinical efficacy, yielding a pooled ORR of 52% and a DCR of 76%. The pooled efficacy estimates encompass both treatment‑naive and pretreated populations, as the included studies did not consistently stratify outcomes by line of therapy, which is an important consideration when interpreting these results. Additionally, median DOR reached 12.62 months, PFS averaged 10.16 months, and OS was 14.67 months. Analysis of time-to-event metrics revealed notable heterogeneity.

These findings closely parallel results from significant clinical trials. Notably, the VISION trial, a pivotal Phase II study investigating Tepotinib in METex14-altered NSCLC patients, demonstrated an ORR of 57.3% and median PFS of 12.6 months in treatment-naïve subjects, aligning with our current analysis29. This consistency underscores Tepotinib’s effectiveness in clinical practice. Tepotinib exhibited comparable efficacy when contrasted with other selective MET inhibitors. For instance, the GEOMETRY mono-1 trial examined capmatinib—another selective type Ib MET inhibitor—reporting an ORR of 68% (95% CI: 55.0–79.7%) in 60 previously untreated patients and 44% (95% CI: 34.1–54.3%) in 100 pretreated cases. However, cross-trial comparisons are limited by differences in patient populations and study designs41,42. The present meta-analysis contributes by quantitatively synthesizing multiple studies, providing more precise efficacy estimates and confirming reproducible Tepotinib activity.

Subgroup analyses yielded clinically relevant insights. Comparable efficacy was noted between METex14 skipping (ORR: 52%) and MET amplification (ORR: 53%) subgroups. While METex14 skipping mutations are established oncogenic drivers sensitive to MET inhibition, the notable activity in MET-amplified tumors highlights the broader therapeutic potential of Tepotinib. This finding challenges earlier assumptions regarding the limited responsiveness of MET amplification and indicates that high-level MET amplification may represent a potent, targetable driver, particularly with highly selective inhibitors such as Tepotinib.

Analysis by treatment regimen showed similar ORR for monotherapy (51%) and combination therapy (56%). However, a significant advantage emerged for combination therapy in PFS (16.60 months versus 9.35 months for monotherapy, p = 0.034). This discrepancy suggests that combination therapy, while not substantially increasing response rates, may enhance response durability. A numerical but non-significant improvement in OS for combination therapy (25.14 months versus 11.84 months, p = 0.078) further supports this notion. These findings align with preclinical evidence demonstrating synergistic effects when MET inhibitors combine with EGFR-targeted agents in MET- and EGFR-activated tumor models43,44,45.

Substantial heterogeneity observed in time-to-event outcomes (I2 > 70% for PFS, OS, and DOR) warrants attention. Potential causes include differences in patient characteristics (e.g., prior treatments, ECOG performance status, metastatic burden), methodological diversity (single-arm trials versus observational studies), and variations in defining MET amplification (especially gene copy-number thresholds). Furthermore, methodological differences in detecting MET amplification—such as variations in assay platforms and gene-copy-number thresholds—likely contributed to the observed heterogeneity. High heterogeneity in the amplification subgroup (I2 up to 94.7% for OS) reflects diverse responses, possibly due to the continuous nature of amplification compared to the binary presence of METex14 mutations.

Despite this heterogeneity, sensitivity analyses using leave-one-out methods confirmed the robustness of pooled estimates. Removing individual studies did not substantially affect conclusions, suggesting consistent central effects rather than contradictory findings. Prediction intervals for key outcomes offer clinically realistic expectations, aiding clinical decision-making.

Collectively, the findings solidify Tepotinib as a crucial therapeutic option for MET-altered NSCLC. Consistent efficacy across METex14 and amplification subgroups supports comprehensive MET testing, including both mutation and amplification analyses. The promising results of combination therapy, particularly significant PFS improvement, highlight the need for randomized trials evaluating combination approaches, especially in patients with concurrent EGFR mutations or EGFR-TKI resistance. To our knowledge, this is the first meta-analysis that quantitatively synthesizes and directly compares the efficacy of Tepotinib between MET exon 14 skipping and MET amplification subgroups, while also evaluating the differential outcomes of monotherapy versus combination regimens. Our findings provide the most robust evidence to date on this topic, addressing a critical gap in the existing literature, which primarily consists of single-arm trials.

Future research should address several knowledge gaps. Firstly, optimal patient selection criteria within the MET-altered population require refinement, especially regarding MET amplification levels. Secondly, acquired resistance mechanisms to Tepotinib need systematic investigation to inform sequential therapy strategies. Thirdly, integrating Tepotinib into earlier therapy lines and combinations with emerging modalities, such as immunotherapy, warrants exploration. Lastly, comparative studies with other MET inhibitors would further contextualize Tepotinib’s position among MET-targeted therapies.

Several limitations must be acknowledged. Firstly, the predominance of single-arm studies and the absence of randomized controlled trials prevent direct efficacy comparisons and may introduce confounding biases. Secondly, the inclusion of both prospective and retrospective study designs, while intended to synthesize all available evidence, may introduce additional confounding. However, sensitivity analyses excluding the retrospective study demonstrated robustness of the pooled estimates. Thirdly, the included studies did not systematically report coexisting genetic alterations, particularly whether patients with METex14 mutations also harbored concurrent MET amplification. Consequently, we were unable to perform subgroup analyses to evaluate the potential prognostic and predictive impact of such co‑alterations. Fourthly, limited subgroup sizes, particularly for combination therapy and MET amplification cohorts, restrict statistical power. Fifthly, variations in MET amplification testing methods and thresholds across studies may introduce misclassification bias. Finally, limited patient-level data reporting precluded deeper analyses of potential effect modifiers.

In conclusion, this meta-analysis consolidates Tepotinib as a cornerstone therapy for MET-altered NSCLC, demonstrating consistent efficacy regardless of the alteration type. The significant progression-free survival benefit observed with combination therapy highlights its potential to enhance response durability, warranting validation in prospective randomized controlled trials. These results should inform clinical guidelines to promote comprehensive MET testing and position Tepotinib as a key therapeutic option, both as monotherapy and in rational combinations.

Materials and methods

This study was prospectively registered in the PROSPERO database (CRD420251013134), and strict adherence was made to PRISMA criteria during planning of studies, execution, and preparation of manuscript31.

Literature retrieval

A comprehensive literature search was performed using electronic databases and manual screening. Two independent reviewers (JYX and QYC) thoroughly searched the PubMed, EMBASE, Web of Science, and Cochrane Library libraries. Searches were conducted with combinations of “non-small-cell lung cancer” and “MET dysregulation”, as well as “Tepotinib”. This literature search was finalized on March 20, 2025, and further information on the exact search methodology may be obtained from Supplementary Table S1.

Eligibility criteria

Inclusion criteria for studies were clinical studies reported in English that met the following explicit requirements: (1) patients with histologically proven locally advanced or metastatic NSCLC with dysregulation of the MET signaling pathway; (2) treatment with Tepotinib; (3) objective response rate (ORR) expressed, with specific definition as the proportion of patients with complete or partial responses by RECIST criteria version 1.132; and (4) AEs graded according to investigator recording with the National Cancer Institute’s Common Terminology Criteria for Adverse Events. Abstract-only articles were excluded from this systematic review.

Study selection and data extraction

Two independent reviewers (QYC and JYX) conducted literature searching, initial screening, and subsequent data extraction. Disputes during the course were addressed by discussion and agreement. Initial screening of studies was conducted by scanning titles and abstracts of articles with predefined criteria, with subsequent detailed assessments of full text conducted to validate eligibility. Data extracted from studies were (1) first author surname; (2) year of publication; (3) study design; (4) baseline patient characteristics such as number of participants, MET dysregulation categories, gender distribution, and age; (5) treatment received details (dosage, administration schedule, and regimen detail); (6) Eastern Cooperative Oncology Group performance status (ECOG PS); (7) clinical effectiveness outcomes; and (8) occurring AEs. Some studies29,32,33,34 reported results separately for two genetically different patient groups, allowing for subgroup analyses; as such, each group was independently analyzed. All other studies that were included were portrayed as a single dataset.

Quality assessment

Quality appraisal for non-randomized studies was performed using the Methodological Index for Non-randomized Studies (MINORS) scale35. The single included randomized controlled trial underwent bias evaluation employing the Cochrane Risk of Bias Tool (RoB 2)36,37.

Statistical analysis

Patient demographics, efficacy outcomes, and adverse event data extracted from selected studies underwent comprehensive statistical evaluation. Weighted mean proportions (e.g., for ORR and DCR) were calculated using inverse-variance weighting within random- or fixed-effects meta-analysis models in NSCLC patients with MET pathway dysregulation and presented with corresponding 95% confidence intervals (95% CI). Study heterogeneity was assessed statistically using the chi-square test (Q statistic) and quantified through the I² statistic, interpreted as follows: minimal (≤ 25%), mild (26–50%), moderate (51–75%), or substantial (> 75%). In cases where significant heterogeneity (p < 0.1 or I2 > 50%) was observed, random-effects models were utilized; otherwise, fixed-effects models were applied38. Subgroup analyses were further performed based on the type of MET alteration and the therapeutic approach (monotherapy vs. combination therapy). Sensitivity analyses involved the sequential exclusion of studies contributing notably to heterogeneity. The overall safety profile was evaluated by synthesizing reported AEs across studies. Meta-analytic calculation of raw proportions employed logit transformations, with an added continuity correction (0.5) applied to studies reporting zero events. All statistical computations were executed using R statistical software (version 4.5.0, R Foundation for Statistical Computing, Vienna, Austria; https://www.R-project.org/), specifically the meta package (version 4.5.0).