Abstract
Background
Globally, lung cancer is the foremost contributor to cancer-associated fatalities. Of its various histological sub-types, non-small-cell lung cancer (NSCLC) constitutes the vast majority, accounting for more than 85% of diagnoses. A key driver in this pathology is the mesenchymal–epithelial transition (MET) receptor tyrosine kinase, whose biological activity is governed by the hepatocyte growth factor (HGF) ligand. MET plays a pivotal role in regulating tumour cell proliferation and migration, and aberrant activation of MET signalling is linked to the pathogenesis of NSCLC.
Methods
A retrospective analysis of the FAERS database (2016–2025) was conducted to obtain adverse event data pertaining to four MET-targeted therapeutic agents, namely amivantamab, savolitinib, capmatinib and tepotinib. A systematic disproportionality analysis was performed to identify adverse events.
Results
Statistically significant differences were observed across most baseline clinical characteristics. Patients taking tepotinib were significantly older (mean 74.7 years) and experience a higher mortality (21%) than those taking amivantamab, savolitinib or capmatinib. In signal detection, amivantamab caused severe cutaneous/mucosal and infusion-related toxicities. The three c-Met inhibitors led to oedema, with savolitinib distinguished by haematological risks, capmatinib by severe oedema and tepotinib by pulmonary/renal toxicities. In time-to-onset analysis, amivantamab’s infusion-related reaction occurred earliest with the highest incidence; savolitinib (pyrexia/hepatic abnormalities) and capmatinib (nausea) appeared early with relatively high incidences; tepotinib’s peripheral swelling was early but had a low incidence. High death-associated adverse drug reactions included dyspnoea, peripheral oedema, decreased appetite and increased creatinine, warranting further investigation. In cluster analysis, key adverse drug reactions (e.g. infusion-related reactions for amivantamab, hepatotoxicity for savolitinib, peripheral swelling for capmatinib and peripheral oedema for tepotinib) primarily occur as isolated events with limited co-occurrence of multiple toxicities.
Conclusion
These findings support individualized monitoring and evidence-based treatment selection for older patients, those with comorbidities or patients at high risk with MET-altered NSCLC.
Keywords: FDA Adverse Event Reporting System (FAERS), mesenchymal/epithelial transition (MET), non-small-cell lung cancer (NSCLC), pharmacovigilance
Plain Language Summary
Why was this study done?
Lung cancer is the world’s leading cause of cancer death. The most common type is non-small-cell lung cancer (NSCLC). In some patients, NSCLC is driven by changes in a gene called MET. Several targeted drugs that block MET are now available but their side effects are not well compared in everyday clinical practice. This study aimed to identify the unique safety risks of each drug to help doctors choose the best option for each patient.
How was the study conducted?
We analyzed real-world side-effect reports from the U.S. FDA database (FAERS) for four MET-targeted drugs: amivantamab, savolitinib, capmatinib and tepotinib. We looked at which side effects occurred, how early they appeared and how serious they were.
What did we find?
Each drug had a distinct safety profile:
Amivantamab – most commonly caused skin rashes and reactions during the infusion. These infusion reactions happened earlier and more frequently than side effects with other drugs.
Savolitinib – main risks were blood disorders (e.g. low blood cell counts) and liver injury. Fever and liver problems often appeared early.
Capmatinib – strongly linked to severe swelling in the arms and legs. Nausea was also common and tended to occur early.
Tepotinib – required caution for lung inflammation and kidney damage (shown by rising creatinine levels). Swelling could occur early but was less common.
We also identified that shortness of breath, swelling, poor appetite, and kidney function decline were more frequently seen in patients who died, so these need close attention.
Encouragingly, each drug’s main side effects usually occurred alone, rather than together with other severe toxicities. This may make side-effect management simpler in practice.
What does this mean for patients?
These findings help doctors personalize treatment. For example, a patient with pre-existing liver disease might avoid savolitinib, while someone with kidney issues may not be suited for tepotinib. Those prone to swelling may need extra monitoring if given capmatinib.
Conclusion
This study provides clear, practical guidance for selecting MET-targeted therapies based on each patient’s age, organ function and overall health. Individualized drug choice and targeted monitoring can improve both safety and treatment outcomes for patients with MET-altered NSCLC.
Introduction
Worldwide, the most pronounced survival benefits in patients with cancer have been observed in those presenting with aggressive cancers at advanced stages, exemplified by a 20–37% increase in survival for those with regional lung cancer and a 2–10% increase for those with metastatic disease.1 Yet, even with these therapeutic advances, lung cancer remains on track to account for a greater number of deaths in 2026 than the combined mortality of the next two leading cancer types, namely colorectal and pancreatic cancer.1 Thus, lung cancer remains the leading cause of cancer-related deaths worldwide, with non-small cell lung cancer (NSCLC) accounting for more than 85% of cancer cases worldwide.2 The development of targeted therapy with small-molecule tyrosine kinase inhibitors (TKIs) and immunotherapy with immune checkpoints inhibitors has ushered in the era of precision medicine in the treatment of lung cancer.
Mesenchymal–epithelial transition (MET) amplification constitutes the most common off-target, driver-independent resistance mechanism identified to date. The MET receptor, activated by its ligand hepatocyte growth factor (HGF), plays a pivotal role in regulating tumour cell proliferation and migration, and its dysregulation is associated with several cancers, including NSCLC.3 The skipping of MET exon 14 gives rise to deletion of the juxtamembrane domain, thereby abrogating the autoinhibitory function of the MET receptor and driving constitutive oncogenic signalling activation.2
The advent of targeted therapies has fundamentally transformed the treatment landscape and prognostic outlook for patients with oncogene-driven NSCLC, with an expanding repertoire of oncogenic driver-directed agents now available. Amongst the various molecular subsets, MET-dysregulated NSCLC – encompassing MET exon 14 skipping mutations and MET gene amplification, the latter being one of the most prevalent bypass resistance mechanisms in oncogene-addicted NSCLC – has garnered particular attention. In recent years, several anti-MET therapies have received regulatory approval, and numerous additional candidates are currently under active clinical investigation.4
Dysregulation of the MET proto-oncogene has been implicated in tumorigenesis, local invasion and metastatic dissemination across a broad spectrum of solid malignancies. Amivantamab is a MET×EGFR bispecific antibody developed to target MET alterations. Several MET-TKIs, including capmatinib, tepotinib and savolitinib, have received regulatory approval for the treatment of advanced-stage NSCLC harbouring MET exon 14 skipping mutations. However, the clinical utility of capmatinib is frequently constrained by the emergence of acquired resistance mechanisms.5
Although these MET-targeted drugs have brought significant survival benefits to patients with NSCLC, their adverse drug reaction (ADR) profiles still require further in-depth understanding. In the VISION trial, tepotinib exhibited favourable antitumour activity amongst pretreated patients.6 Savolitinib has demonstrated clinical efficacy in both first-line and second-line therapeutic settings, including in patients with aggressive pulmonary sarcomatoid carcinoma.6
The utilization of worldwide public adverse event (AE) databases represents a novel approach for post-marketing drug safety surveillance.7 Therefore, an assessment of the ADR profiles of amivantamab, savolitinib, capmatinib and tepotinib is essential for guiding rational clinical use, optimizing AE management and improving patient quality of life. This study utilizes the FDA Adverse Event Reporting System (FAERS) database to mine and analyze ADR signals of these drugs, aiming to provide evidence for clinical medication safety.
Materials and methods
Data resource
We extracted data from the FAERS database (https://www.fda.gov) covering the period from 2016 to 2025. This repository captures AE reports, medication error submissions and product quality complaints associated with adverse outcomes that have been reported to the FDA. The FAERS database constitutes a cornerstone of the FDA’s post-marketing surveillance infrastructure for pharmaceutical agents and therapeutic biologic products.
AE signals were mapped to standardized categories and anatomical sites using the Medical Dictionary for Regulatory Activities (MedDRA) version 28.1 (September 2025), with coding and classification performed at the system organ class (SOC) and preferred term (PT) levels.
Study design
A retrospective disproportionality pharmacovigilance study was conducted. Data for the four drugs were extracted from original data of Drug Information (DRUG) via a keyword in Product Active Ingredient (prod_ai), then the extracted data were combined with the further six original data streams (Patient Demographics (DEMO), Adverse Reaction (REAC), Patient Outcome (OUTC), Indication for Use (INDI), Reporting Source (RPSR) and Therapeutic Use (THER)) to produce analytical data (Figure 1).
Figure 1.
Study flow diagram.
Reports listing amivantamab, capmatinib, savolitinib and tepotinib as prod_ai were included.
Outcome definition: death and non-death (comprising all outcomes with codes other than death).
Data collection and deduplication
Original data from the seven files of FAERS via case id, report year and season were used. These three items were combined into a new ID, based on which the same ID was considered as duplication.
Statistical analysis
All statistical analyses were performed using R software (version 4.4.3 for Windows). Categorical variables were summarized as frequencies with corresponding percentages, whilst continuous variables were expressed as mean ± standard deviation (SD) and median with range. Disproportionality analysis was conducted by calculating the reporting odds ratio (ROR) with 95% confidence intervals (CI). The ROR was defined as the ratio of the odds of a specific AE occurring with a given drug to the odds of the same event occurring with all other drugs in the extracted dataset. Additionally, incidence rates were derived from the total number of reported cases.
Results
Baseline characteristics of safety report populations
Comparisons of baseline characteristics between the four MET/EGFR-targeted agents showed statistically significant differences in all demographic, medication-related and clinical outcome indicators (Table 1).
Table 1.
Baseline characteristics.
| Amivantamab (n=5431) | Capmatinib (n=4112) | Savolitinib (n=184) | Tepotinib (n=1007) | p value# | |
|---|---|---|---|---|---|
| Year | |||||
| 2025 | 3066 (56.5%) | 214 (5.2%) | 21 (11.4%) | 200 (19.9%) | <0.001 *** |
| 2024 | 918 (16.9%) | 287 (7.0%) | 10 (5.4%) | 198 (19.7%) | <0.001 *** |
| 2023 | 732 (13.5%) | 896 (21.8%) | 35 (19.0%) | 308 (30.6%) | <0.001 *** |
| 2022 | 575 (10.6%) | 1322 (32.1%) | 13 (7.1%) | 159 (15.8%) | <0.001 *** |
| 2021 | 140 (2.6%) | 1044 (25.4%) | 7 (3.8%) | 137 (13.6%) | <0.001 *** |
| 2020 | 0 | 344 (8.4%) | 14 (7.6%) | 5 (0.5%) | <0.001 *** |
| 2019 | 0 | 4 (0.1%) | 7 (3.8%) | 0 | <0.001 *** |
| 2018 | 0 | 1 (0.0%) | 31 (16.8%) | 0 | <0.001 *** |
| 2017 | 0 | 0 | 24 (13.0%) | 0 | <0.001 *** |
| 2016 | 0 | 0 | 22 (12.0%) | 0 | <0.001 *** |
| Sex | |||||
| Female | 2720 (50.1%) | 2033 (49.4%) | 113 (61.4%) | 524 (52.0%) | 0.01 ** |
| Male | 1521 (28.0%) | 1749 (42.5%) | 68 (37.0%) | 439 (43.6%) | <0.001 *** |
| Missing | 1190 (21.9%) | 330 (8.0%) | 3 (1.6%) | 44 (4.4%) | |
| Weight (kg) | <0.001 *** | ||||
| Mean (SD) | 66.0 (14.2) | 70.6 (17.4) | 59.6 (10.9) | 62.4 (15.2) | |
| Median [Min, Max] | 65.0 [33.0, 138] | 69.5 [12.0, 144] | 55.0 [45.5, 89.5] | 62.0 [36.0, 100] | |
| Missing | 3099 (57.1%) | 3150 (76.6%) | 131 (71.2%) | 855 (84.9%) | |
| Age (years) | <0.001 *** | ||||
| Mean (SD) | 64.4 (11.2) | 71.9 (10.4) | 63.2 (13.7) | 74.7 (10.4) | |
| Median | 66.0 | 72.0 | 64.9 | 76.0 | |
| Missing | 2165 (39.9%) | 2307 (56.1%) | 7 (3.8%) | 178 (17.7%) | |
| Age groups | |||||
| <18 | 1 (0.0%) | 2 (0.0%) | 0 | 0 | 0.701 |
| 18–64 | 1339 (24.7%) | 308 (7.5%) | 78 (42.4%) | 108 (10.7%) | <0.001 *** |
| 64–85 | 1887 (34.7%) | 1355 (33.0%) | 93 (50.5%) | 590 (58.6%) | <0.001 *** |
| ≥85 | 39 (0.7%) | 140 (3.4%) | 6 (3.3%) | 131 (13.0%) | <0.001 *** |
| Missing | 2165 (39.9%) | 2307 (56.1%) | 7 (3.8%) | 178 (17.7%) | |
| Status of report | |||||
| Initial | 3089 (56.9%) | 1802 (43.8%) | 59 (32.1%) | 639 (63.5%) | <0.001 *** |
| Follow-up | 2342 (43.1%) | 2310 (56.2%) | 125 (67.9%) | 368 (36.5%) | <0.001 *** |
| Drug sequence | |||||
| 1 | 3285 (60.5%) | 2787 (67.8%) | 29 (15.8%) | 772 (76.7%) | <0.001 *** |
| 2 | 802 (14.8%) | 688 (16.7%) | 48 (26.1%) | 137 (13.6%) | <0.001 *** |
| 3 | 519 (9.6%) | 266 (6.5%) | 30 (16.3%) | 41 (4.1%) | <0.001 *** |
| 4 | 266 (4.9%) | 130 (3.2%) | 24 (13.0%) | 17 (1.7%) | <0.001 *** |
| 5 | 164 (3.0%) | 65 (1.6%) | 17 (9.2%) | 13 (1.3%) | <0.001 *** |
| others | 395 (7.3%) | 176 (4.3%) | 36 (19.6%) | 27 (2.7%) | <0.001 *** |
| Drug’s reported role in event | |||||
| Primary Suspect | 3147 (57.9%) | 2814 (68.4%) | 12 (6.5%) | 772 (76.7%) | <0.001 *** |
| Secondary Suspect | 2218 (40.8%) | 1229 (29.9%) | 156 (84.8%) | 197 (19.6%) | <0.001 *** |
| Concomitant | 66 (1.2%) | 60 (1.5%) | 16 (8.7%) | 38 (3.8%) | <0.001 *** |
| Interacting | 0 | 9 (0.2%) | 0 | 0 | 0.005 ** |
| Cumulative dose to first reaction (mg) | <0.001 *** | ||||
| Mean (SD) | 13600 (27200) | 65100 (124000) | 40000 (68600) | 30800 (9700) | |
| Median [Min, Max] | 7350 [350, 147000] | 24600 [252, 842000] | 12600 [600, 281000] | 37000 [16200, 37100] | |
| Missing | 5322 (98.0%) | 3644 (88.6%) | 120 (65.2%) | 1001 (99.4%) | |
| Dechallenge | |||||
| Positive | 1403 (25.8%) | 1057 (25.7%) | 99 (53.8%) | 210 (20.9%) | <0.001 *** |
| Negative | 238 (4.4%) | 191 (4.6%) | 21 (11.4%) | 56 (5.6%) | <0.001 *** |
| Unknown | 1620 (29.8%) | 1216 (29.6%) | 24 (13.0%) | 262 (26.0%) | <0.001 *** |
| Does not apply | 502 (9.2%) | 744 (18.1%) | 4 (2.2%) | 195 (19.4%) | <0.001 *** |
| Missing | 1668 (30.7%) | 904 (22.0%) | 36 (19.6%) | 284 (28.2%) | |
| Rechallenge | |||||
| Positive | 59 (1.1%) | 94 (2.3%) | 5 (2.7%) | 9 (0.9%) | <0.001 *** |
| Negative | 66 (1.2%) | 66 (1.6%) | 0 | 6 (0.6%) | 0.015 * |
| Unknown | 1917 (35.3%) | 225 (5.5%) | 58 (31.5%) | 188 (18.7%) | <0.001 *** |
| Does not apply | 15 (0.3%) | 75 (1.8%) | 0 | 21 (2.1%) | <0.001 *** |
| Missing | 3374 (62.1%) | 3652 (88.8%) | 121 (65.8%) | 783 (77.8%) | |
| Duration/length of therapy (day) | <0.001 *** | ||||
| Mean (SD) | 38.5 (89.3) | 75.3 (118) | 34.1 (75.7) | 77.6 (106) | |
| Median | 1.00 | 34.0 | 17.0 | 40.0 | |
| Missing | 3679 (67.7%) | 3314 (80.6%) | 64 (34.8%) | 758 (75.3%) | |
| Time from medication to symptom (day) | |||||
| <10 | 1007 (18.5%) | 149 (3.6%) | 13 (7.1%) | 46 (4.6%) | <0.001 *** |
| [10, 30) | 296 (5.5%) | 234 (5.7%) | 55 (29.9%) | 87 (8.6%) | <0.001 *** |
| [30, 60) | 238 (4.4%) | 172 (4.2%) | 22 (12.0%) | 40 (4.0%) | <0.001 *** |
| [60, 90) | 99 (1.8%) | 82 (2.0%) | 4 (2.2%) | 35 (3.5%) | 0.015 * |
| [90, 365) | 364 (6.7%) | 142 (3.5%) | 8 (4.3%) | 60 (6.0%) | <0.001 *** |
| ≥365 | 66 (1.2%) | 38 (0.9%) | 8 (4.3%) | 10 (1.0%) | 0.006 ** |
| Missing | 3361 (61.9%) | 3295 (80.1%) | 74 (40.2%) | 729 (72.4%) | |
| Death cases | |||||
| Death Cases | 408 (7.5%) | 653 (15.9%) | 12 (6.5%) | 211 (21.0%) | <0.001 *** |
| Non-death Cases | 2539 (46.8%) | 1208 (29.4%) | 93 (50.5%) | 480 (47.7%) | <0.001 *** |
| Missing | 2484 (45.7%) | 2251 (54.7%) | 79 (42.9%) | 316 (31.4%) | |
| Occupation | |||||
| Consumer | 458 (8.4%) | 1865 (45.4%) | 4 (2.2%) | 286 (28.4%) | <0.001 *** |
| Health professional | 1106 (20.4%) | 445 (10.8%) | 9 (4.9%) | 81 (8.0%) | <0.001 *** |
| Pharmacist | 978 (18.0%) | 127 (3.1%) | 0 | 74 (7.3%) | <0.001 *** |
| Physician | 2824 (52.0%) | 1570 (38.2%) | 161 (87.5%) | 492 (48.9%) | <0.001 *** |
| Other health-professional | 0 | 3 (0.1%) | 0 | 0 | 0.169 |
| Missing | 65 (1.2%) | 102 (2.5%) | 10 (5.4%) | 74 (7.3%) | |
| Occrring countries | |||||
| United States | 1828 (33.7%) | 2407 (58.5%) | 22 (12.0%) | 278 (27.6%) | <0.001 *** |
| France | 771 (14.2%) | 337 (8.2%) | 0 | 1 (0.1%) | <0.001 *** |
| Japan | 557 (10.3%) | 309 (7.5%) | 1 (0.5%) | 237 (23.5%) | <0.001 *** |
| European Union | 751 (13.8%) | 8 (0.2%) | 0 | 19 (1.9%) | <0.001 *** |
| China | 377 (6.9%) | 50 (1.2%) | 33 (17.9%) | 6 (0.6%) | <0.001 *** |
| Canada | 207 (3.8%) | 33 (0.8%) | 0 | 7 (0.7%) | <0.001 *** |
| Germany | 61 (1.1%) | 112 (2.7%) | 0 | 20 (2.0%) | <0.001 *** |
| South Korea | 91 (1.7%) | 18 (0.4%) | 58 (31.5%) | 5 (0.5%) | <0.001 *** |
| Brazil | 128 (2.4%) | 31 (0.8%) | 0 | 2 (0.2%) | <0.001 *** |
| Taiwan (Province of China) | 113 (2.1%) | 31 (0.8%) | 5 (2.7%) | 12 (1.2%) | <0.001 *** |
| Spain | 47 (0.9%) | 56 (1.4%) | 0 | 7 (0.7%) | 0.048 * |
| Italy | 18 (0.3%) | 79 (1.9%) | 0 | 9 (0.9%) | <0.001 *** |
| Belgium | 41 (0.8%) | 54 (1.3%) | 0 | 10 (1.0%) | 0.029 * |
| Others | 441 (8.1%) | 587 (14.3%) | 65 (35.3%) | 394 (39.1%) | <0.001 *** |
p values were adjusted using the Holm method.
p<0.05,
p<0.01,
p<0.001.
ROR-based disproportionality analysis identified distinct ADR profiles for each drug (Figure 2). Severe cutaneous and mucosal toxicities, as well as infusion-related reactions, were the key adverse signals of amivantamab. The forest plot showed that paronychia and infusion-related reactions yielded the highest reporting odds ratios with substantial case numbers. A variety of multi-organ toxicities were also observed. Savolitinib showed unique life-threatening haematological risks (disseminated intravascular coagulation, haemolysis) requiring intensive monitoring. Capmatinib mainly caused oedema/fluid retention (peripheral/generalized oedema, large case numbers). Tepotinib also caused oedema but with stronger reactions and more obvious pulmonary/renal toxicities than capmatinib. Overall, amivantamab combined EGFR and antibody-related toxicities, whilst the three c-Met inhibitors led to oedema as a class effect, with savolitinib distinguished by haematological toxicity, capmatinib by severe peripheral oedema and tepotinib by pulmonary/renal toxicities.
Figure 2.
Forest plots of ROR for adverse drug reactions associated with four targeted antitumour agents.
(A) Amivantamab, (B) savolitinib, (C) capmatinib and (D) tepotinib. The vertical axis lists adverse events sorted in descending order of reporting odds ratio (ROR). The dot-size is proportional to the number of reported cases. The vertical dashed line indicates the ROR=1 reference line. A signal was defined as statistically significant if the ROR value was >1 and the 95% CI did not include 1.
Figure 3 presents the temporal characteristics and clinical outcomes of key ADRs for the four targeted agents. For amivantamab, infusion-related reaction occurred the earliest with the highest incidence rate. For savolitinib, pyrexia and impaired hepatic function presented abnormally early with relatively high incidence rates. For capmatinib, nausea emerged early with a high incidence rate, whilst for tepotinib, peripheral swelling occurred early but with a low incidence rate. Notably, ADRs with relatively high mortality included dyspnoea in amivantamab-treated patients, peripheral oedema in savolitinib-treated patients, decreased appetite in capmatinib-treated patients, and increased blood creatinine in tepotinib-treated patients. Although no direct causal relationship can be confirmed between these ADRs and death, their relatively high death-associated proportions indicate a strong potential association, warranting further investigation into the underlying mechanisms and clinical risk factors.
Figure 3.
Temporal characteristics and clinical outcomes of key ADRs for four targeted antitumour agents.
(A) Amivantamab, (B) savolitinib, (C) capmatinib, (D) tepotinib. (a) Adverse drug reaction (ADR) timeline and rate: the x-axis indicates the mean onset time of ADRs (days), and the y-axis represents the reported death percentage. (b) Patient outcome percentage: stacked bars show the proportion of different clinical outcomes for each ADR.
Figure 4 presents the top ADR clusters for the four targeted agents, analysed via intersection set analysis to characterize co-occurrence patterns of the 10 most frequently reported ADRs for each drug.
Figure 4.
Top ADR cluster analysis for four targeted antitumour agents.
(A) Amivantamab, (B) savolitinib, (C) capmatinib, (D) tepotinib. The red bar plots show the intersection size (number of cases) for each adverse drug reaction (ADR) co-occurrence combination. The blue horizontal bars represent the set size (total reported cases) for each individual ADR. The dot plots below indicate the specific ADRs included in each intersection combination.
Notably, for all four agents, the highest intersection sizes were observed for single-ADR combinations, indicating that many ADR reports occurred as isolated events rather than concurrent multiple toxicities. For amivantamab, the largest intersection corresponded to infusion-related reaction as a standalone event, consistent with its status as the most frequently reported ADR. For savolitinib, the top intersections were dominated by single events of drug-induced liver injury and hepatic function abnormalities, reflecting the drug’s prominent hepatotoxicity profile. For capmatinib, the highest intersection size was observed for peripheral swelling as an isolated ADR, aligning with its core fluid retention-related toxicity. For tepotinib, the largest intersection was attributed to peripheral oedema as a single event, further confirming oedema as its hallmark AE. Collectively, these findings demonstrate that the key ADRs for each agent primarily occur as independent events, with limited co-occurrence of multiple toxicities in clinical reports.
Infusion-related reaction was uniquely observed in amivantamab-treated patients (7.8%), with no reports in patients treated with savolitinib, capmatinib or tepotinib (p<0.001), highlighting this as a drug-specific toxicity requiring intensive clinical monitoring during amivantamab administration. No statistically significant inter-drug differences were detected for the following ADRs: vomiting, pneumonia, pruritus, hypotension, constipation, general physical health deterioration and acute kidney injury (p>0.05), indicating that these events occurred at comparable frequencies across all four agents (Table 2; the report number in Table 2 involves data containing preferred terms only, not the same to Table 1 (total sample)).
Table 2.
Preferred term for the most common adverse drug reactions.
| No | Preferred term | Amivantamab | Savolitinib | Capmatinib | Tepotinib | p value# |
|---|---|---|---|---|---|---|
| 1 | Report number | 9777 (100.0%) | 249 (100.0%) | 9779 (100.0%) | 2461 (100.0%) | |
| 2 | Infusion-related reaction | 766 (7.8%) | 0 | 0 | 0 | <0.001*** |
| 3 | Rash | 491 (5.0%) | 1 (0.4%) | 67 (0.7%) | 16 (0.7%) | <0.001*** |
| 4 | Oedema peripheral | 82 (0.8%) | 4 (1.6%) | 330 (3.4%) | 104 (4.2%) | <0.001*** |
| 5 | Fatigue | 82 (0.8%) | 1 (0.4%) | 374 (3.8%) | 61 (2.5%) | <0.001*** |
| 6 | Nausea | 119 (1.2%) | 3 (1.2%) | 324 (3.3%) | 58 (2.4%) | <0.001*** |
| 7 | Peripheral swelling | 18 (0.2%) | 0 | 389 (4.0%) | 38 (1.5%) | 0.03* |
| 8 | Dyspnoea | 136 (1.4%) | 8 (3.2%) | 197 (2.0%) | 23 (0.9%) | 0.03* |
| 9 | Oedema | 41 (0.4%) | 1 (0.4%) | 164 (1.7%) | 77 (3.1%) | 0.03* |
| 10 | Diarrhoea | 99 (1.0%) | 1 (0.4%) | 87 (0.9%) | 75 (3.0%) | 0.03* |
| 11 | Asthenia | 65 (0.7%) | 0 | 166 (1.7%) | 8 (0.3%) | 0.03* |
| 12 | Decreased appetite | 66 (0.7%) | 0 | 134 (1.4%) | 33 (1.3%) | 0.03* |
| 13 | Vomiting | 73 (0.7%) | 1 (0.4%) | 104 (1.1%) | 18 (0.7%) | 0.381 |
| 14 | Product dose omission issue | 62 (0.6%) | 0 | 105 (1.1%) | 7 (0.3%) | 0.03* |
| 15 | Pleural effusion | 43 (0.4%) | 2 (0.8%) | 92 (0.9%) | 22 (0.9%) | 0.03* |
| 16 | Dizziness | 44 (0.5%) | 0 | 103 (1.1%) | 8 (0.3%) | 0.03* |
| 17 | Pyrexia | 68 (0.7%) | 10 (4.0%) | 65 (0.7%) | 9 (0.4%) | 0.03* |
| 18 | Increased blood creatinine | 11 (0.1%) | 1 (0.4%) | 101 (1.0%) | 38 (1.5%) | 0.03* |
| 19 | Renal impairment | 26 (0.3%) | 0 | 49 (0.5%) | 75 (3.0%) | 0.03* |
| 20 | Pulmonary embolism | 122 (1.2%) | 2 (0.8%) | 19 (0.2%) | 6 (0.2%) | 0.03* |
| 21 | Paronychia | 143 (1.5%) | 1 (0.4%) | 2 (0.0%) | 1 (0.0%) | 0.03* |
| 22 | Joint swelling | 2 (0.0%) | 0 | 127 (1.3%) | 9 (0.4%) | 0.03* |
| 23 | Cough | 43 (0.4%) | 0 | 84 (0.9%) | 8 (0.3%) | 0.03* |
| 24 | Pneumonitis | 90 (0.9%) | 1 (0.4%) | 36 (0.4%) | 8 (0.3%) | 0.03* |
| 25 | Thrombocytopenia | 105 (1.1%) | 0 | 15 (0.2%) | 8 (0.3%) | 0.03* |
| 26 | Pneumonia | 56 (0.6%) | 3 (1.2%) | 45 (0.5%) | 12 (0.5%) | 0.495 |
| 27 | Pruritus | 53 (0.5%) | 0 | 43 (0.4%) | 20 (0.8%) | 0.381 |
| 28 | Hypotension | 68 (0.7%) | 1 (0.4%) | 36 (0.4%) | 10 (0.4%) | 0.105 |
| 29 | Neutropenia | 97 (1.0%) | 0 | 10 (0.1%) | 1 (0.0%) | 0.03* |
| 30 | Dermatitis acneiform | 103 (1.1%) | 0 | 3 (0.0%) | 1 (0.0%) | 0.03* |
| 31 | Constipation | 37 (0.4%) | 0 | 48 (0.5%) | 19 (0.8%) | 0.381 |
| 32 | Dysphagia | 8 (0.1%) | 0 | 92 (0.9%) | 3 (0.1%) | 0.03* |
| 33 | Back pain | 30 (0.3%) | 0 | 65 (0.7%) | 7 (0.3%) | 0.03* |
| 34 | Oxygen saturation decreased | 89 (0.9%) | 0 | 10 (0.1%) | 2 (0.1%) | 0.03* |
| 35 | Interstitial lung disease | 51 (0.5%) | 2 (0.8%) | 17 (0.2%) | 27 (1.1%) | 0.03* |
| 36 | Weight increased | 6 (0.1%) | 0 | 73 (0.7%) | 15 (0.6%) | 0.03* |
| 37 | Swelling | 5 (0.1%) | 0 | 77 (0.8%) | 11 (0.4%) | 0.03* |
| 38 | Fluid retention | 3 (0.0%) | 0 | 69 (0.7%) | 18 (0.7%) | 0.03* |
| 39 | Stomatitis | 78 (0.8%) | 0 | 10 (0.1%) | 1 (0.0%) | 0.03* |
| 40 | General physical health deterioration | 45 (0.5%) | 0 | 30 (0.3%) | 12 (0.5%) | 0.495 |
| 41 | Flushing | 83 (0.8%) | 0 | 2 (0.0%) | 0 | 0.03* |
| 42 | Skin toxicity | 84 (0.9%) | 0 | 1 (0.0%) | 0 | 0.03* |
| 43 | Hypoalbuminaemia | 56 (0.6%) | 0 | 15 (0.2%) | 13 (0.5%) | 0.03* |
| 44 | Febrile neutropenia | 74 (0.8%) | 1 (0.4%) | 2 (0.0%) | 3 (0.1%) | 0.03* |
| 45 | Mucosal inflammation | 75 (0.8%) | 0 | 4 (0.0%) | 0 | 0.03* |
| 46 | Erythema | 61 (0.6%) | 0 | 11 (0.1%) | 6 (0.2%) | 0.03* |
| 47 | Alanine aminotransferase increased | 14 (0.1%) | 7 (2.8%) | 43 (0.4%) | 12 (0.5%) | 0.03* |
| 48 | Hypoacusis | 2 (0.0%) | 0 | 69 (0.7%) | 4 (0.2%) | 0.03* |
| 49 | Acute kidney injury | 28 (0.3%) | 3 (1.2%) | 28 (0.3%) | 11 (0.4%) | 0.381 |
| 50 | Deafness | 8 (0.1%) | 0 | 56 (0.6%) | 6 (0.2%) | 0.03* |
| 51 | Deep vein thrombosis | 58 (0.6%) | 0 | 6 (0.1%) | 3 (0.1%) | 0.03* |
| 52 | Generalized oedema | 13 (0.1%) | 0 | 41 (0.4%) | 13 (0.5%) | 0.03* |
| 53 | Arthralgia | 16 (0.2%) | 2 (0.8%) | 36 (0.4%) | 11 (0.4%) | 0.04* |
| 54 | Myelosuppression | 53 (0.5%) | 3 (1.2%) | 5 (0.1%) | 4 (0.2%) | 0.03* |
| 55 | Aspartate aminotransferase, increased | 10 (0.1%) | 7 (2.8%) | 30 (0.3%) | 11 (0.4%) | 0.03* |
| 56 | Hepatic function, abnormal | 15 (0.2%) | 10 (4.0%) | 25 (0.3%) | 8 (0.3%) | 0.03* |
| 57 | Respiratory failure | 24 (0.2%) | 4 (1.6%) | 17 (0.2%) | 9 (0.4%) | 0.04* |
| 58 | Anaphylactic reaction | 41 (0.4%) | 4 (1.6%) | 1 (0.0%) | 0 | 0.03* |
| 59 | Drug-induced liver injury | 5 (0.1%) | 11 (4.4%) | 16 (0.2%) | 4 (0.2%) | 0.03* |
| 60 | Abdominal pain upper | 9 (0.1%) | 2 (0.8%) | 14 (0.1%) | 9 (0.4%) | 0.04* |
| 61 | Pulmonary toxicity | 6 (0.1%) | 0 | 5 (0.1%) | 14 (0.6%) | 0.03* |
| 62 | Renal disorder | 2 (0.0%) | 0 | 8 (0.1%) | 12 (0.5%) | 0.03* |
| 63 | Drug hypersensitivity | 1 (0.0%) | 3 (1.2%) | 3 (0.0%) | 0 | 0.03* |
| 64 | Subdural haematoma | 2 (0.0%) | 3 (1.2%) | 0 | 0 | 0.03* |
| 65 | Disseminated intravascular coagulation | 1 (0.0%) | 2 (0.8%) | 1 (0.0%) | 0 | 0.03* |
| 66 | Haemolysis | 0 | 2 (0.8%) | 0 | 0 | 0.03* |
p values were adjusted using the Holm method.
p<0.05,
p<0.01,
p<0.001.
Amongst all SOC categories, no statistically significant inter-drug differences were observed for the following: respiratory, thoracic and mediastinal disorders, nervous system disorders, infections and infestations, reproductive system and breast disorders, surgical and medical procedures, endocrine disorders, product issues, congenital, familial and genetic disorders, and social circumstances (p>0.05; Table 3). These findings align with the comparable frequencies of individual ADRs (e.g. vomiting, pneumonia, pruritus) across agents observed in Table 2, confirming that these ADRs belong to SOCs with uniform occurrence profiles amongst the four agents.
Table 3.
System organ class level distribution of adverse drug reactions.
| No | System organ class | Amivantamab | Savolitinib | Capmatinib | Tepotinib | p value# |
|---|---|---|---|---|---|---|
| 1 | General disorders and administration site conditions | 578 (14.0%) | 23 (11.3%) | 779 (25.5%) | 195 (15.8%) | <0.001*** |
| 2 | Skin and subcutaneous tissue disorders | 474 (11.5%) | 5 (2.5%) | 128 (4.2%) | 79 (6.4%) | <0.001*** |
| 3 | Injury, poisoning and procedural complications | 462 (11.2%) | 16 (7.8%) | 144 (4.7%) | 61 (4.9%) | <0.001*** |
| 4 | Immune system disorders | 381 (9.2%) | 10 (4.9%) | 58 (1.9%) | 37 (3.0%) | <0.001*** |
| 5 | Vascular disorders | 331 (8.0%) | 13 (6.4%) | 115 (3.8%) | 78 (6.3%) | <0.001*** |
| 6 | Respiratory, thoracic and mediastinal disorders | 274 (6.7%) | 23 (11.3%) | 214 (7.0%) | 92 (7.4%) | 0.538 |
| 7 | Gastrointestinal disorders | 217 (5.3%) | 8 (3.9%) | 214 (7.0%) | 96 (7.8%) | 0.011* |
| 8 | Nervous system disorders | 200 (4.9%) | 7 (3.4%) | 152 (5.0%) | 78 (6.3%) | 0.834 |
| 9 | Infections and infestations | 194 (4.7%) | 4 (2.0%) | 106 (3.5%) | 46 (3.7%) | 0.172 |
| 10 | Cardiac disorders | 128 (3.1%) | 21 (10.3%) | 191 (6.2%) | 61 (4.9%) | <0.001*** |
| 11 | Investigations | 117 (2.8%) | 12 (5.9%) | 156 (5.1%) | 67 (5.4%) | <0.001*** |
| 12 | Blood and lymphatic system disorders | 114 (2.8%) | 9 (4.4%) | 38 (1.2%) | 25 (2.0%) | 0.009** |
| 13 | Metabolism and nutrition disorders | 107 (2.6%) | 6 (2.9%) | 187 (6.1%) | 53 (4.3%) | <0.001*** |
| 14 | Neoplasms benign, malignant and unspecified | 93 (2.3%) | 12 (5.9%) | 122 (4.0%) | 35 (2.8%) | <0.001*** |
| 15 | Eye disorders | 81 (2.0%) | 4 (2.0%) | 26 (0.8%) | 18 (1.5%) | 0.009** |
| 16 | Renal and urinary disorders | 71 (1.7%) | 2 (1.0%) | 56 (1.8%) | 53 (4.3%) | 0.009** |
| 17 | Musculoskeletal and connective tissue disorders | 64 (1.6%) | 6 (2.9%) | 81 (2.6%) | 38 (3.1%) | 0.011* |
| 18 | Hepatobiliary disorders | 55 (1.3%) | 20 (9.8%) | 84 (2.7%) | 40 (3.2%) | 0.009** |
| 19 | Psychiatric disorders | 52 (1.3%) | 0 | 82 (2.7%) | 29 (2.3%) | 0.009** |
| 20 | Reproductive system and breast disorders | 36 (0.9%) | 0 | 19 (0.6%) | 10 (0.8%) | 1 |
| 21 | Surgical and medical procedures | 34 (0.8%) | 0 | 31 (1.0%) | 15 (1.2%) | 1 |
| 22 | Ear and labyrinth disorders | 18 (0.4%) | 0 | 39 (1.3%) | 11 (0.9%) | 0.009** |
| 23 | Endocrine disorders | 18 (0.4%) | 0 | 14 (0.5%) | 7 (0.6%) | 1 |
| 24 | Product issues | 9 (0.2%) | 0 | 14 (0.5%) | 3 (0.2%) | 1 |
| 25 | Congenital, familial and genetic disorders | 5 (0.1%) | 3 (1.5%) | 7 (0.2%) | 6 (0.5%) | 0.055 |
| 26 | Social circumstances | 5 (0.1%) | 0 | 3 (0.1%) | 2 (0.2%) | 1 |
| 27 | Pregnancy, puerperium and perinatal conditions | 1 (0.0%) | 0 | 0 | 3 (0.2%) | 0.264 |
Data are presented as number of cases (percentage of total adverse drug reaction reports). Adjusted p values were calculated via the Holm method for multiple comparisons.
p<0.05,
p<0.01,
p<0.001.
Figure 5 illustrates the mean onset time of the top eight ADRs stratified by death and non-death cases, with all ADRs selected from the top events in Table 2. In death cases, the most rapidly occurring ADRs included fatigue, respiratory failure, rash and pneumonitis, all presenting with extremely short mean onset times. In contrast, amongst non-death cases, dyspnoea, increased blood creatinine and nausea were the earliest-onset events, with notably shorter mean occurrence times compared to other non-fatal ADRs. These findings highlight distinct temporal patterns of fatal and non-fatal ADRs, providing critical evidence for early clinical risk monitoring.
Figure 5.
Mean onset time of top ADRs in death and non-fatality cases for the four targeted antitumour agents.
All adverse drug reactions (ADRs) were selected from the top events in Table 2. The horizontal axis represents the mean occurring time (days) of each ADR.
Amongst these, respiratory failure in savolitinib-treated patients exhibited the highest mortality rate (75.0%), followed by pulmonary embolism in tepotinib-treated patients (66.7%). For tepotinib, additional high-mortality ADRs included pneumonitis (33.3%) and increased blood creatinine (21.1%). Notably, oedema-related events (peripheral oedema, oedema) showed variable mortality across agents, with the highest rates observed in savolitinib (25.0% for peripheral oedema, 12.5% for oedema). Gastrointestinal and constitutional symptoms (diarrhoea, nausea, fatigue or rash) generally presented with lower mortality across all four agents. These findings align with the temporal patterns observed in Figure 5, highlighting that rapidly occurring fatal ADRs (e.g. respiratory failure, pulmonary embolism) are associated with extremely high mortality, warranting urgent clinical intervention (Table 4).
Table 4.
Mortality of key ADRs.
| Preferred term# | Savolitinib | Tepotinib | Amivantamab | Capmatinib |
|---|---|---|---|---|
| Respiratory failure | 75.0% | |||
| Pulmonary embolism | 66.7% | |||
| Pneumonitis | 33.3% | |||
| Pleural effusion | 17.9% | 28.9% | 10.5% | |
| Peripheral oedema | 25.0% | 14.4% | 3.7% | 9.1% |
| Increased blood creatinine | 21.1% | |||
| Diarrhoea | 6.7% | 17.2% | 5.7% | |
| Nausea | 15.5% | 10.9% | 10.5% | |
| Dyspnoea | 12.0% | 11.8% | 14.7% | |
| Oedema | 12.5% | 14.3% | 7.1% | 9.0% |
| Pyrexia | 10.3% | |||
| Rash | 3.1% | 7.6% | ||
| Fatigue | 1.6% | 6.1% | 5.1% |
All adverse drugs reactions (ADRs) were selected from the top events in Table 2. Data are presented as mortality percentage (%).
Table 5 presents the detailed outcome distribution of the seven ADRs with no statistically significant inter-drug differences in overall incidence (from Table 2). Whilst the overall occurrence rates of these ADRs were comparable across agents, significant differences in outcome profiles were observed for hypotension, constipation and acute kidney injury (p<0.05). Specifically, amivantamab and capmatinib were associated with higher proportions of death and life-threatening events for hypotension, whilst capmatinib had the highest death rate for constipation. For acute kidney injury, significant inter-drug outcome differences were driven by higher proportions of hospitalization and life-threatening events in capmatinib-treated patients, alongside death events in savolitinib. In contrast, vomiting, pneumonia, pruritus and general physical health deterioration showed no significant inter-drug differences in outcome distribution (p>0.05), indicating consistent clinical severity profiles across the four agents.
Table 5.
Outcome comparison of non-significant preferred terms.
| No | Preferred term# | p value (χ2) | p value (Fisher) |
|---|---|---|---|
| 1 | Acute kidney injury | 0.004** | 0.006** |
| 2 | Constipation | 0.022* | 0.018* |
| 3 | General physical health deterioration | 0.624 | 0.644 |
| 4 | Hypotension | 0.024* | 0.008** |
| 5 | Pneumonia | 0.225 | 0.205 |
| 6 | Pruritus | 0.694 | 0.631 |
| 7 | Vomiting | 0.472 | 0.359 |
Preferred term was from Table 2. Outcome included: death, disability, hospitalization, life-threatening, intervention, congenital anomaly, required intervention and others.
p<0.05,
p<0.01.
Discussion
Cancer ranks as the second leading cause of mortality in the USA, whereas it constitutes the foremost cause of death in China.8 This comprehensive pharmacovigilance study analysed real-world AE reports using the FAERS database (2016–2025) to characterize and compare the safety profiles of four clinically important MET-targeted agents: amivantamab, capmatinib, tepotinib and savolitinib in patients with NSCLC. To our knowledge, this is the first retrospective disproportionality analysis directly comparing an EGFR-MET bispecific antibody with three MET-TKIs in a post-marketing setting.3 Our findings demonstrate highly distinct baseline demographics, treatment patterns, ADR spectra, onset kinetics, severity and mortality risks across the four agents, with substantial implications for making personalized clinical decisions and toxicity management.
Baseline comparisons revealed striking and statistically significant differences in patient characteristics, treatment contexts and reporting patterns (all p<0.001). Patients receiving tepotinib were significantly older (mean 74.7 years), with a notably high proportion of individuals aged ≥85 years, whereas those receiving amivantamab were younger and more commonly middle-aged. In contrast, recipients of amivantamab tended to be younger and were more frequently middle-aged, which reflected its wider utilization amongst relatively healthy patients. Savolitinib was associated with a higher female proportion and was disproportionately reported in South Korea and China, consistent with its regional clinical development and application. These imbalances must be considered when analysing and interpreting safety comparisons across different agents. Amongst reporters, the proportions of AE reports submitted by doctors and consumers for capmatinib, tepotinib and amivantamab are higher than those for savolitinib. The disparity in the sources of these reports (consumers versus doctors) also reflects differing levels of awareness and reporting practices. The higher mortality rates associated with capmatinib and tepotinib may be attributable to this factor, in addition to the requirement for intravenous administration of amivantamab in hospital settings. Differences in administration routes and reporting parties lead to inconsistent AE submission timelines. Amivantamab is infused intravenously in hospitals. Routine lab tests and full clinical monitoring are arranged before and after each infusion. Medical staff detect mild oedema and renal abnormalities early and submit FAERS reports in a timely manner. Capmatinib, tepotinib and savolitinib are oral drugs for home use. No regular in-person follow-up is provided. Cumulative toxicities like peripheral oedema and renal impairment progress slowly. Patients overlook mild, asymptomatic biochemical changes at first. They only visit hospitals after symptoms worsen, which delays safety reports. Some AE reports for oral TKIs that come from patients themselves, and they cannot recognize hidden sub-clinical organ damage. As a result, early AE signals are collected late.
Notably, at the preferred term level, the four drugs demonstrated target-specific safety profiles with non-overlapping characteristics. Amivantamab, an EGFR-MET bispecific antibody, displayed a unique toxicity profile dominated by infusion-related reactions, cutaneous toxicities (rash, paronychia, acneiform dermatitis) and mucosal inflammation, findings that are consistent with the known class effects of EGFR inhibition and monoclonal antibody administration.9 Disproportionality analysis confirmed strong reporting ROR signals for these events, with infusion-related reactions occurring early and at the highest frequency. Amivantamab also demonstrated significant signals for pulmonary embolism, thrombocytopenia and pneumonitis, supporting the need for vigilant monitoring of thromboembolic and pulmonary risks during treatment.10,11 Analysis at the SOC level further underscored distinct organ-specific toxicity patterns across the four agents. Amivantamab was associated with a high proportion of skin, general, immune-mediated and vascular disorders. Capmatinib and tepotinib predominantly induced general disorders (notably oedema) as well as gastrointestinal, metabolic and renal disorders. By contrast, savolitinib was closely linked to a higher incidence of hepatic, cardiac and haematological disorders.
Amongst the MET-TKIs, capmatinib and tepotinib shared a core safety phenotype characterized by peripheral oedema, generalized oedema and fluid retention, which are well established class effects of MET inhibition. Substantial evidence has established that HGF and VEGF drive oncogenic progression through their respective receptors, MET and VEGFR2. VEGFR2 negatively regulates MET activity by promoting its dephosphorylation,12 whereas MET conversely downregulates VEGFR2 expression through enhancing proteasome-mediated receptor degradation.13 This reciprocal interplay between the MET and VEGF signalling axes may compromise vascular integrity and increase permeability, offering a credible pathophysiological basis for treatment-associated oedema. Capmatinib showed the highest reporting rates for peripheral swelling and oedema, whilst tepotinib displayed stronger signals for renal impairment, increased blood creatinine and interstitial lung disease, indicating a higher risk of renal and pulmonary toxicities. These observations align with previous clinical trial data and extend them to a large unselected real-world population.14–16
Savolitinib is a highly selective MET-TKI that has received regulatory approval for the treatment of NSCLC harbouring MET exon 14 skipping mutations or MET gene amplification.17–21 Savolitinib demonstrated a distinct safety profile marked by prominent hepatotoxicity, drug-induced liver injury and elevated transaminases, as well as rare but high-risk markers for severe haematological and coagulation disorders, including disseminated intravascular coagulation and haemolysis.3 Although these events were uncommon, their high ROR and life-threatening potential warrant rigorous monitoring of liver function and coagulation parameters. Patients taking tepotinib are significantly older (mean 74.7 years) and experience a higher mortality (21%). Thus, the higher mortality of certain drugs is highly confounded by baseline patient age and comorbidities rather than purely indicating intrinsic drug toxicity.
Temporal and severity analyses further differentiated the agents. ADRs related to amivantamab occurred early, most notably infusion-related reactions, which are typically acute and manageable with premedication. Capmatinib and tepotinib exhibited longer median ADR durations, consistent with chronic fluid retention and organ-specific toxicities. Tepotinib was associated with the highest proportion of fatal outcomes (21.0%), followed by capmatinib (15.9%), likely reflecting the older age and greater comorbidity burden of treated patients. Mortality analysis of key ADRs identified respiratory failure (savolitinib), pulmonary embolism (tepotinib) and renal impairment (tepotinib) as events with the highest fatality rates, underscoring the need for prompt clinical intervention. Tepotinib may elevate serum creatinine levels by inhibiting the function of these renal tubular transporters for creatinine. Elevated serum creatinine represents another well-documented and common ADR, as confirmed by multiple studies.22–24 Severe respiratory failure associated with savolitinib and pulmonary embolism related to tepotinib have been rarely documented in previous post-marketing safety evaluations. These findings underscore the necessity for clinicians to closely monitor respiratory symptoms, oxygen saturation and coagulation parameters during treatment, thereby minimizing the risk of life-threatening or fatal AEs.
Several ADRs, encompassing vomiting, pneumonia and pruritus, exhibited no statistically significant differences in incidence amongst the drugs, implying that these events are more likely attributable to general cancer or treatment-related phenomena rather than being specific toxicities associated with particular drugs. However, even amongst ADRs with similar overall frequencies, significant differences in clinical outcomes were observed for hypotension, constipation and acute kidney injury, indicating that clinical severity may vary independently of reporting rate. The severity of AEs attributable to hypotension was significantly greater in patients receiving amivantamab and capmatinib compared with those receiving savolitinib and tepotinib. To our knowledge, this observation has not been documented in prior research and requires further validation. These data highlight the need for monitoring rigorous blood pressure and optimal control in patients treated with amivantamab and capmatinib. With respect to acute kidney injury, the incidence of elevated serum creatinine was lower in patients receiving capmatinib than in those administered with tepotinib. Nevertheless, capmatinib was linked to a greater incidence of life-threatening events, underscoring the necessity for clinicians to closely monitor serum creatinine dynamics and remain vigilant for the development of acute renal impairment during capmatinib treatment.
This study exhibits several notable strengths. First, the utilization of a large sample size and an extended study duration facilitate robust identification of rare and severe ADRs. Second, the incorporation of disproportionality analysis, temporal stratification, clustering and mortality-stratified onset time analysis offers a comprehensive and reproducible approach to safety characterization. Third, the direct comparative evaluation of four contemporary MET-targeted agents provides clinically relevant and actionable insights.
Several limitations merit consideration. First, the FAERS database is inherently subject to underreporting, reporting biases and incomplete data, particularly for weight, cumulative dose and onset time. Second, as a retrospective observational study, causality cannot be definitively established. Third, unbalanced baseline characteristics, including age, ethnicity and treatment line, may confound safety comparisons. Fourth, the dataset lacks information on performance status, tumour stage and concurrent medications, all of which may influence ADR development and outcomes.
Conclusion
Overall, our real-world pharmacovigilance study identifies distinct and clinically meaningful safety profiles amongst amivantamab, capmatinib, tepotinib and savolitinib. Amivantamab displays prominent acute skin, mucosal and infusion-related toxicities; capmatinib is associated with substantial oedema and fluid retention; tepotinib confers heightened risks of renal and pulmonary injury, with fatal events closely linked to older age; and savolitinib presents unique hepatotoxic and haematological risks. These observations support the implementation of individualized monitoring strategies and offer evidence-based guidance for treatment selection in patients who are older, have comorbidities or have high-risk MET-altered NSCLC.
Acknowledgements
None.
Footnotes
Contributions: NL performed the design of this study and JY performed analysis. All authors revised the manuscript. All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published. The authors decline the use of artificial intelligence, language models, machine learning, or similar technologies to create content or assist with writing or editing of the manuscript.
Disclosure and potential conflicts of interest: The authors declare that they have no conflicts of interest relevant to this manuscript. The International Committee of Medical Journal Editors (ICMJE) Potential Conflicts of Interests form for the author is available for download at: https://www.drugsincontext.com/wp-content/uploads/2026/08/dic.2026-4-7-COI.pdf
Funding declaration: This research is supported by National High Level Hospital Clinical Research Funding (80102022624).
Correct attribution: Copyright © 2026 Li N, Cong M, Yang J. https://doi.org/10.7573/dic.2026-4-7. Published by Drugs in Context under Creative Commons License Deed CC BY NC ND 4.0.
Provenance: Submitted; externally peer reviewed.
Drugs in Context is published by BioExcel Publishing Ltd. Registered office: 6 Green Lane Business Park, 238 Green Lane, New Eltham, London, SE9 3TL, UK.
BioExcel Publishing Limited is registered in England Number 10038393. VAT GB 252 7720 07.
For all manuscript and submissions enquiries, contact the Editorial office editorial@drugsincontext.com
For all permissions, rights, and reprints, contact David Hughes david.hughes@bioexcelpublishing.com
References
- 1.Siegel RL, Kratzer TB, Wagle NS, Sung H, Jemal A. Cancer statistics, 2026. CA Cancer J Clin. 2026;76(1):e70043. doi: 10.3322/caac.70043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Su PL, Furuya N, Asrar A, et al. Recent advances in therapeutic strategies for non-small cell lung cancer. J Hematol Oncol. 2025;18(1):35. doi: 10.1186/s13045-025-01679-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wang J, Ma R, Qu B, Li X. Adverse event signal analysis of type Ib MET tyrosine kinase inhibitors based on food and drug administration adverse event reporting system. Expert Opin Drug Saf . 2025:1–10. doi: 10.1080/14740338.2025.2487158. [DOI] [PubMed] [Google Scholar]
- 4.Remon J, Hendriks LEL, Mountzios G, et al. MET alterations in NSCLC-current perspectives and future challenges. J Thoracic Oncol. 2023;18(4):419–435. doi: 10.1016/j.jtho.2022.10.015. [DOI] [PubMed] [Google Scholar]
- 5.Lee JB, Shim JS, Cho BC. Evolving roles of MET as a therapeutic target in NSCLC and beyond. Nat Rev Clin Oncol. 2025;22(9):640–666. doi: 10.1038/s41571-025-01051-9. [DOI] [PubMed] [Google Scholar]
- 6.Drusbosky LM, Dawar R, Rodriguez E, Ikpeazu CV. Therapeutic strategies in METex14 skipping mutated non-small cell lung cancer. J Hematol Oncol. 2021;14(1):129. doi: 10.1186/s13045-021-01138-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yang J, Li N, Lin W, et al. Machine learning for predicting hyperglycemic cases induced by PD-1/PD-L1 inhibitors. J Healthcare Eng. 2022;2022:6278854. doi: 10.1155/2022/6278854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yang J, Li N, Lin W, et al. Comprehensive analysis of diabetes mellitus-related gene expression and associated prognoses in human lung cancer. Curr Cancer Drug Targets. 2023;23(11):889–899. doi: 10.2174/1568009623666230529154306. [DOI] [PubMed] [Google Scholar]
- 9.Zeng Y, Li Y, Du Z, Shu R, Fu C, Zeng Q. Cutaneous adverse events associated with Rybrevant based on FDA adverse event reporting system (FAERS) database from 2021 to 2024. Naunyn-Schmiedeberg’s Arch Pharmacol. 2026;399:12157–12168. doi: 10.1007/s00210-026-05168-1. [DOI] [PubMed] [Google Scholar]
- 10.Fu X, Zeng D, Li M, et al. Post-marketing safety surveillance of Amivantamab: a real world study based on the FDA adverse event reporting system. Expert Opin Drug Saf. 2025;26:1–9. doi: 10.1080/14740338.2025.2471512. [DOI] [PubMed] [Google Scholar]
- 11.Sun R, Ning Z, Qin H, et al. A real-world pharmacovigilance study of amivantamab-related cardiovascular adverse events based on the FDA adverse event reporting system (FAERS) database. Sci Rep. 2024;14(1):9552. doi: 10.1038/s41598-024-55829-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lu KV, Chang JP, Parachoniak CA, et al. VEGF inhibits tumor cell invasion and mesenchymal transition through a MET/VEGFR2 complex. Cancer Cell. 2012;22(1):21–35. doi: 10.1016/j.ccr.2012.05.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chen TT, Filvaroff E, Peng J, et al. MET Suppresses epithelial VEGFR2 via intracrine VEGF-induced endoplasmic reticulum-associated degradation. EBioMedicine. 2015;2(5):406–420. doi: 10.1016/j.ebiom.2015.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lin Y, Xu S, Deng M, Cao Y, Ding J, Lin T. Exploring ototoxicity associated with capmatinib: insights from a real-world data analysis of the FDA adverse event reporting system (FAERS) database. Clin Epidemiol. 2025;17:513–521. doi: 10.2147/clep.S528454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Chang X, Shao D, Bao J, Liu Z, Ma J. Comparative toxicovigilance of capmatinib and tepotinib in NSCLC: respiratory signal detection and adverse event profiling via FAERS. J Biochem Mol Toxicol. 2026;40(1):e70657. doi: 10.1002/jbt.70657. [DOI] [PubMed] [Google Scholar]
- 16.Wei S, Zhang L, Liu C, Qi X. Real-world safety of tepotinib: insights from the food and drug administration adverse event reporting system. PLoS One. 2025;20(12):e0339005. doi: 10.1371/journal.pone.0339005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lu S, Fang J, Li X, et al. Once-daily savolitinib in Chinese patients with pulmonary sarcomatoid carcinomas and other non-small-cell lung cancers harbouring MET exon 14 skipping alterations: a multicentre, single-arm, open-label, phase 2 study. Lancet Respir Med. 2021;9(10):1154–1164. doi: 10.1016/s2213-2600(21)00084-9. [DOI] [PubMed] [Google Scholar]
- 18.Choueiri TK, Heng DYC, Lee JL, et al. Efficacy of savolitinib vs sunitinib in patients with MET-driven papillary renal cell carcinoma: the SAVOIR phase 3 randomized clinical trial. JAMA Oncol. 2020;6(8):1247–1255. doi: 10.1001/jamaoncol.2020.2218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee J, Kim ST, Kim K, et al. Tumor genomic profiling guides patients with metastatic gastric cancer to targeted treatment: the VIKTORY umbrella trial. Cancer Discovery. 2019;9(10):1388–1405. doi: 10.1158/2159-8290.Cd-19-0442. [DOI] [PubMed] [Google Scholar]
- 20.Yoh K, Hirashima T, Saka H, et al. Savolitinib ± osimertinib in Japanese patients with advanced solid malignancies or EGFRm NSCLC: Ph1b TATTON part C. Target Oncol. 2021;16(3):339–355. doi: 10.1007/s11523-021-00806-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Choueiri TK, Plimack E, Arkenau HT, et al. Biomarker-based phase II trial of savolitinib in patients with advanced papillary renal cell cancer. J Clin Oncol. 2017;35(26):2993–3001. doi: 10.1200/jco.2017.72.2967. [DOI] [PubMed] [Google Scholar]
- 22.Zheng R, Chen J, Han J, et al. A tri-scale in silico framework integrating pharmacovigilance and mechanistic modeling suggests tepotinib-associated acute kidney injury risk. Ren Fail. 2026;48(1):2610895. doi: 10.1080/0886022x.2025.2610895. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chen MF, Harada G, Liu D, et al. Brief report: tyrosine kinase inhibitors for lung cancers that inhibit MATE-1 can lead to “false” decreases in renal function. J Thoracic Oncol. 2024;19(1):153–159. doi: 10.1016/j.jtho.2023.09.1444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wijtvliet V, Roosens L, De Bondt C, Janssens A, Abramowicz D. Pseudo-acute kidney injury secondary to tepotinib. Clin Kidney J. 2023;16(4):760–761. doi: 10.1093/ckj/sfac180. [DOI] [PMC free article] [PubMed] [Google Scholar]