In 2001, imatinib, the first kinase inhibitor, was approved by the U.S. Food and Drug Administration (FDA) [1]. Imatinib inhibits the Abelson (ABL) tyrosine kinase, which is expressed as a dysregulated fusion protein, BCR-ABL, in nearly all chronic myeloid leukemia (CML) cases [2]. This oncogenic fusion protein arises from a chromosomal rearrangement that juxtaposes the breakpoint cluster region protein (BCR) gene with the ABL gene, forming the Philadelphia chromosome.

The phase 3 International Randomized Study of Interferon and STI571 (IRIS) demonstrated superior efficacy and tolerability of imatinib (400 mg once daily) compared to interferon alfa plus cytarabine in newly diagnosed chronic-phase CML. In the treatment of chronic myeloid leukemia (CML) and certain gastrointestinal stromal tumors (GIST), imatinib, as the first-generation tyrosine kinase inhibitor (TKI), ushered in a new era of targeted therapy. Although the second- and third-generation TKIs (such as dasatinib, nilotinib, and ponatinib) have made significant progress in overcoming drug resistance and improving treatment efficacy, imatinib, due to its lower toxicity, good tolerance, and outstanding efficacy in first-line treatment, remains the preferred drug for many patients. At the 18-month follow-up, the imatinib arm achieved a significantly higher complete cytogenetic response rate (0% Philadelphia chromosome-positive metaphases: 76.2% (95% CI, 72.5–79.9) vs. 14.5% (95% CI, 10.5–18.5); P < 0.001) and a greater probability of freedom from disease progression to accelerated phase or blast crisis (96.7% vs. 91.5%; P < 0.001). These results established imatinib as a first-line therapy with enhanced therapeutic activity and reduced toxicity in early CML management [3].

Imatinib has fundamentally transformed the treatment of CML, resulting in significant improvements in patient prognosis [4]. In the USA, the age-adjusted annual mortality rate for CML patients decreased from 0.9 per 100,000 individuals in 1996 to 0.4 per 100,000 in 2006 [5]. Similar trends of reduced mortality have been observed in CML patients across other global regions [6, 7].

Although imatinib has undoubtedly provided substantial benefits for many cancer patients, these agents are not curative. Most merely delay tumor progression, as advanced malignancies develop escape pathways to evade targeted inhibition, leading to adverse effects and drug resistance. Thus, investigating and analyzing the real-world safety profile of imatinib remain critically important.

The FAERS, a prototypical public spontaneous reporting system, collates post-marketing safety reports and clinical study data related to FDA-approved drugs and therapeutic biologics, both within and outside the USA. It has been extensively employed for pharmacovigilance potential signal detection [8]. Our study leveraged comprehensive disproportionality analyses of imatinib-associated AEs extracted from FAERS data. By rigorously evaluating potential signal strengths within real-world evidence, this approach provides a robust framework for assessing the efficacy and safety of imatinib therapy, facilitating effective AE detection and management. Through this analysis, our research aims to elucidate imatinib’s safety profile and contribute evidence on its adverse events. This endeavor not only advances our understanding of imatinib-related risks but also underscores the necessity of continuous post-marketing drug safety surveillance. Ultimately, this pharmacovigilance study seeks to inform clinical practice, guide therapeutic decision-making, and enhance patient safety.

Study design and data sources

This study analyzed AEs associated with imatinib using data from the FAERS. FAERS is selected for its unparalleled scale of real-world post-marketing safety data, enabling global surveillance of adverse events across drugs and medical products. As FDA’s primary pharmacovigilance database, it serves as the gold standard for detecting emerging safety signals and supporting regulatory decisions. The raw data spanned 84 quarters (the first quarter of 2014 to the four quarter of 2024), encompassing 18,613,992 background reports (55,357,463 AE occurrences), with 56,364 reports (170,659 AE occurrences) specifically linked to imatinib. To address inherent limitations of spontaneous reporting systems, such as duplicate or withdrawn reports, rigorous data cleaning was performed following FDA guidelines (https://www.fda.gov/drugsatfda). Duplicate reports were removed by sorting the DEMO table fields (PRIMARYID, CASEID, FDA_DT) and retaining the entry with the highest FDA_DT for identical CASEIDs; if CASEID and FDA_DT matched, the entry with the largest PRIMARYID was retained. Post-2019 quarterly deletion lists were further utilized to exclude invalid reports. Statistical methods were applied to identify adverse drug reactions (ADRs). AEs related to imatinib were categorized into preferred terms (PTs) and system organ classes (SOCs) based on the hierarchical structure of the Medical Dictionary for Regulatory Activities (MedDRA20.0). The data screening workflow is illustrated in Fig. 1.

Multistep process of data extraction, processing, and analysis from the FAERS database

Full size image

Statistical analysis

A disproportionality analysis was conducted to identify potential drug-AE associations in pharmacovigilance. Four algorithms were applied: the reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN) [9, 10], and multi-item gamma Poisson shrinker (MGPS). These methods compare the proportional reporting rates of AEs between imatinib and all other drugs [11]. Four algorithms were employed to quantify imatinib-associated AE potential signals, with their equations and decision criteria detailed in Supplementary Table 1. Signal detection rules were applied at both the system organ class (SOC) and preferred term (PT) levels. A signal was flagged at the SOC level if at least one of the four indices met the predefined criteria, while PT-level signals required concurrent satisfaction of all four criteria [12, 13]. Higher scores across these parameters generally indicated stronger disproportionality [14]. Statistical analyses were performed using SAS software (version 9.4), as recommended by the FDA. Raw ASCII data downloaded from the FDA website were imported into SAS, deduplicated following FDA-recommended protocols, and subsequently analyzed.

General characteristics

In the FAERS database, 56,364 reports were identified with 170,659 AEs attributed to imatinib, averaging 3.02 AEs per individual. Females accounted for 24,179 cases (42.9%), males 27,808 (49.34%), and 4377 (7.77%) had unspecified gender. Age data were available for 28,962 reports (mean age = 59 ± 14 years). Reports aged < 18 years (n = 1180, 2.09%), 18–44 years (n = 5721, 10.15%), 45–64 years (n = 10,954, 19.43%), and ≥ 65 years (n = 11,107, 19.71%) comprised the cohorts. The highest reporting year was 2014 (n = 5131, 9.10%). Most reports originated from consumers (39.98%), with the USA contributing the majority (n = 16,962, 30.09%). The primary indication was chronic myeloid leukemia (n = 25,664, 45.53%). Severe outcomes constituted 84.24% of reports, predominantly death (n = 19,409, 34.44%) and other serious outcomes (n = 24,861, 44.11%) (Table 1). AE occurrence time (excluding unspecified medication dates) peaked at 0–30 days (24.22%) and > 360 days (38.08%) (Fig. 2).

Time to event report distribution of AE reports

Full size image

Imatinib-associated AEs affected multiple organ systems

The top four involved categories were general disorders and administration site conditions (40,355, 23.65%), gastrointestinal disorders (17,125, 10.03%), neoplasms benign/malignant/unspecified (15,433, 9.04%), and investigations (14,428, 8.45%) (Table 2).

Disproportionality analysis of imatinib AEs

Using the ROR method, the top 50 PTs by potential signal frequency and strength were ranked (Fig. 3). Frequent potential events signals included death (n = 15,667; ROR (95% CI) = 7.34 (7.22–7.46)), nausea (n = 3000; ROR = 1.39 (1.34–1.44)), diarrhea (n = 2532; ROR = 1.46 (1.40–1.52)), malignant neoplasm progression (n = 2177; ROR = 8.33 (7.98–8.69)), and vomiting (n = 2000; ROR = 1.57 (1.50–1.64)). The strongest potential signals (highest ROR) were blast crisis in myelogenous leukemia (n = 762; ROR = 612.66 (542.99–691.27)), pubertal failure (n = 7; ROR = 452.74 (143.69–1426.52)), large intestine fibroma (n = 4; ROR = 431.18 (96.50–1926.58)), slit-lamp tests abnormal (n = 5; ROR = 404.23 (108.55–1505.38)), and chromosome analysis abnormal (n = 124; ROR = 326.24 (254.21–418.68)). A four-algorithm integrated analysis (Table 3) ranked the frequent potential events signals as [1] death (n = 15,667; ROR = 7.34 (7.22–7.46); PRR = 6.76; IC = 2.73; EBGM = 6.64), [2] nausea (n = 3000; ROR = 1.39 (1.34–1.44); PRR = 1.38; IC = 0.47; EBGM = 1.38), [3] diarrhea (n = 2532; ROR = 1.46 (1.40–1.52); PRR = 1.45; IC = 0.54; EBGM = 1.45), [4] drug ineffective (n = 2436; ROR = 0.67 (0.64–0.69); PRR = 0.67; IC =  − 0.57; EBGM = 0.67), and [5] malignant neoplasm progression (n = 1519; ROR = 8.33 (7.98–8.69); PRR = 8.23; IC = 3.01; EBGM = 8.05). By potential signal strength (ROR lower limit, Table 4), the top potential signals were [1] blast crisis in myelogenous leukemia (ROR = 612.66 (542.99–691.27); PRR = 609.93; IC = 7.73; EBGM = 211.98), [2] pubertal failure (ROR = 452.74 (143.69–1426.52); PRR = 452.72; IC = 7.56; EBGM = 189.22), [3] large intestine fibroma (ROR = 431.18 (96.50–1926.58); PRR = 431.17; IC = 7.53; EBGM = 185.36), [4] slit-lamp tests abnormal (ROR = 404.23 (108.55–1505.38); PRR = 404.22; IC = 7.49; EBGM = 180.21), and [5] chromosome analysis abnormal (ROR = 326.24 (254.21–418.68); PRR = 326.00; IC = 7.35; EBGM = 162.84).

Signal strength of AEs at the PT level in ROR. A Ranked by frequency of positive signals. B Ranked by intensity of positive signals

Full size image

This study is the first comprehensive and systematic pharmacovigilance investigation, using the FDA Adverse Event Reporting System (FAERS) database to analyze the related adverse events of imatinib after its market launch. Among them, signal detection is the most important part of the study and also the first step.

This study revealed that AEs associated with imatinib occurred slightly more frequently in males (49.34%) than females (42.9%), aligning with epidemiological trends [15]. The majority of affected reports (39.14%) were aged ≥ 45 years, reflecting the predominant demographic of CML and GIST populations [16]. Consumers submitted 39.98% of reports, consistent with the spontaneous reporting nature of the FAERS database [17]. Severe AEs accounted for 84.24% of imatinib-related reports, likely attributable to its mechanism as a multi-target Abl kinase inhibitor, which modulates critical biological pathways [18, 19]. The high mortality rate (34.44% of outcomes) may stem from life-threatening complications such as tumor progression or organ failure. Notably, 38.08% of AEs occurred > 360 days post-treatment, underscoring the necessity for long-term safety monitoring. At the SOC level, the frequent potential events categories included general disorders and administration site conditions, gastrointestinal disorders, and neoplasms benign, consistent with SPCs’ safety profiles. Common PTs such as nausea, diarrhea, vomiting, and malignant neoplasm progression were frequently reported, corroborating clinical trial findings [20, 21].

While most AEs in our analysis were largely consistent with safety data from drug SPCs and clinical trials, we identified previously unreported significant AE potential signals not explicitly documented in regulatory trials. These included new-onset benign tumors (e.g., lipomas, fibromas) or neoplasms of uncertain behavior. The FDA SPCs for imatinib primarily focus on its therapeutic efficacy against existing malignancies (e.g., GIST, CML) and known complications (e.g., tumor hemorrhage) but do not explicitly address the potential induction of or association with new benign or unclassified neoplasms during treatment [22]. The insidious toxicities associated with long-term medication require particular attention, such as sciatica, joint ankylosis, and even ischemic necrosis of the femoral head observed in the musculoskeletal system, while the FDA SPCs merely mentions “musculoskeletal pain” without specifying severity or chronic consequences [23]. Severe dermatological reactions like exfoliative dermatitis and acute generalized exanthematous pustulosis are inadequately characterized in product SPCs despite documented progression risks in clinical cases [24, 25]. We advocate for enhanced long-term monitoring, particularly targeting musculoskeletal toxicity, metabolic syndrome, and other insidious risks, with recommendations including regular bone density assessments and systematic skin reaction evaluations. This call emphasizes the necessity to bridge SPCs-practice discrepancies through proactive surveillance mechanisms.

Under the SOC “Investigations,” PTs encompassed non-specific enzymatic elevations (e.g., alkaline phosphatase (ALP), lactate dehydrogenase [26]) or structural organ changes identified via imaging. While the FDA SPCs lists abnormal liver function as a monitoring parameter, it does not classify “investigation abnormalities” as standalone adverse reactions, potentially underestimating their impact on clinical decision-making [27, 28].

For the SOC “Pregnancy, Puerperium, and Perinatal Conditions,” cases accounted for 0.41% of reports, yet the ROR method detected 13 potential positive signals (1.07%). Specific PTs included fetal developmental anomalies and gestational complications (e.g., miscarriage, preterm delivery). The FDA currently warns only of “embryo-fetal toxicity” and recommends contraception but lacks explicit risk data on imatinib use during pregnancy [29, 30]. This underscores the need for enhanced monitoring of pregnancy-exposed reports and research into imatinib’s potential effects on placental function or fetal organogenesis. Within the psychiatric disorders (SOC), “psychiatric symptoms” accounted for 1.58% of reports, with specific PTs including depression, anxiety, and cognitive impairment. Notably, the FDA-approved SPCs for imatinib do not list psychiatric symptoms as common or severe adverse reactions, only mentioning that impaired driving ability may be associated with fatigue [31, 32]. Further investigation is warranted to determine whether imatinib crosses the blood–brain barrier to directly alter neurotransmitter activity or synergizes with the psychological burden inherent to chronic illness.

The FDA-approved SPCs for imatinib do not currently acknowledge ototoxicity-related risks. However, our study identified potential signals in the Ear and Labyrinth Disorders System Organ Class (SOC), including preferred terms (PTs) such as tinnitus, hearing impairment, and vestibular dysfunction [33]. This finding underscores the need for proactive monitoring of vestibular function in reports undergoing long-term imatinib therapy, particularly among elderly individuals or those with comorbid renal impairment.

This study leverages the inherent advantages of large-scale real-world investigations and sophisticated data mining techniques. However, several limitations necessitate careful consideration. The FAERS database, as a spontaneous reporting system, may introduce analytical biases due to incomplete or inaccurate data collection across countries and healthcare professionals, including reporting bias and indication bias, with challenges in distinguishing adverse events caused by the drug versus disease progression. Additionally, adverse events with extremely low incidence rates associated with imatinib might lack statistical significance in disproportionality analyses, potentially leaving undetected safety potential signals. The specificity of attributing adverse events to imatinib is further limited by confounding from concomitant medications. Finally, while disproportionality analysis identifies statistical significance based on potential signal strength, it cannot fully eliminate confounding effects from polypharmacy.

This study conducted a real-world pharmacovigilance analysis of imatinib using the FDA Adverse Event Reporting System (FAERS) database. Disproportionality analysis (ROR, PRR, BCPNN, EBGM) was applied to evaluate adverse events (AEs) from 2004 to 2024. Results confirmed expected AEs (e.g., nausea, diarrhea) aligned with drug SPCs, while identifying undocumented potential signals such as pubertal failure (ROR = 452.74), large intestine fibroma (ROR = 431.18), ototoxicity, and pregnancy complications. Severe outcomes accounted for 84.24% of reports (34.44% death), with 38.08% of AEs occurring > 360 days post-medication. The study underscores the need for vigilant monitoring of long-term toxicities (e.g., bone metabolism disorders, psychiatric effects) and rare AEs to enhance patient safety.

These findings highlight the necessity of conducting rigorous monitoring of the long-term toxicity and rare AEs of imatinib, particularly in terms of potential long-term effects such as adolescent developmental disorders, colonic fibromas, ototoxicity, and pregnancy. The results of this study not only provide important references for clinicians, alerting them to pay more attention to the potential risks of long-term medication use when using imatinib, but also provide a basis for optimizing drug vigilance strategies. We suggest that in future drug vigilance work, enhanced monitoring and research of rare AEs should be conducted, and further confirmation of the authenticity and relevance of these potential signals should be achieved through multi-source data verification. Additionally, these findings may also prompt regulatory agencies to re-evaluate the safety of imatinib and consider updating relevant information in the drug instructions and regulatory policies to better ensure patient safety.