Real-World Disproportionality Analysis of the Food and Drug Administration Adverse Event Reporting System Database for Asciminib

Patients with chronic myeloid leukemia (CML) have the Philadelphia chromosome (Ph), which is characterized by the exchange of genetic material between chromosome 9 containing the Abelson (ABL) gene and chromosome 22 carrying the breakpoint cluster region (BCR) gene [1, 2]. In October 2021, asciminib (SCEMBLIX^®), a novel BCR-ABL1 inhibitor specifically targeting the ABL myristoyl pocket, was approved by the Food and Drug Administration (FDA) as a third-line treatment option for patients with chronic phase (CP) CML [3]. Asciminib exerts its action by specifically binding to the myristoyl pocket of ABL1, resulting in the inactivation of the BCR-ABL fusion gene and inhibition of ABL1 phosphorylation. This mechanism effectively prevents tumor cell proliferation and induces tumor cell death [4‒6]. Asciminib received accelerated approval following the achievement of its primary endpoint at week 24. Patients treated with asciminib demonstrated a significant molecular response rate of 25.5%, exceeding the 13.2% rate observed in those receiving bosutinib. Additionally, the proportion of patients achieving a complete cytogenetic response was higher in the asciminib group (40.8%) compared to the bosutinib group (24.2%) [7].

Upon its introduction to the market, the product description and initial safety evaluation of asciminib identified a spectrum of commonly observed adverse events (AEs). These include elevated pancreatic enzyme levels, hypertension, cardiovascular toxicity encompassing ischemic and thromboembolic events, and reduced counts of neutrophils and platelets [8]. The most frequently reported AEs included fatigue (18%), thrombocytopenia (17%), anemia (12%), and arthralgias (12%). No instances of cardiovascular events or occlusive arterial disease were reported. Furthermore, adjustments to the dosage, temporary suspension, or permanent treatment discontinuation were required in 26%, 25%, and 9% of patients, respectively [9].

Previous articles on asciminib were primarily based on clinical trials conducted under varying specific circumstances, and the AEs observed in these trials may not comprehensively cover all the AEs encountered in real-world clinical settings [10‒16]. Furthermore, the clinical trials conducted on asciminib thus far may not have adequate sample size or follow-up duration to comprehensively elicit all potential AEs. Moreover, the timing of AEs associated with asciminib remains uncertain. Therefore, it is imperative to investigate the potential AE signals associated with asciminib by employing data mining algorithms on extensive post-marketing dataset [6‒8, 14, 17]. The FDA Adverse Event Reporting System (FAERS) is a publicly accessible database where spontaneous reports of post-marketing AEs are submitted to the US FDA [13, 18]. The FAERS database is an invaluable resource for the surveillance of drug-related AEs, and its extensive global coverage makes it the largest pharmacovigilance database currently available. This study was aimed at mining the signals of AEs for asciminib from the FAERS database, thereby generating a reference for the safe use of asciminib in clinical practice.

The FAERS database serves as a critical system for the collection of spontaneous AEs reports worldwide, supporting the FDA’s post-market surveillance efforts for drugs and biotherapeutic products. This system plays an essential role in identifying and evaluating potential signals while quantifying the associations between specific drugs and AEs reported by individuals. The database is regularly updated every quarter, encompassing a wide range of AE records, documentation of medication errors, and instances of product quality complaints [19, 20].

This pharmacovigilance investigation evaluated the AEs associated with asciminib, as documented in the FAERS database. Data entries made between October 2021 and December 2023 were included in the analysis to coincide with the timing of FDA’s approval of asciminib. The FAERS data were obtained from the official FDA website, accessed at https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html.

The FAERS data package can be downloaded from the website as a compressed file, comprising seven distinct sections that encompass demographic and management details, drug-related information, AEs, patient outcomes, sources of reporting, treatment duration with reported drugs, indications, and removed cases [5, 21]. The intricate process of data manipulation was proficiently executed through the utilization of Python 3.10 (Python Software Foundation, Holland) and Microsoft Excel 2019, collectively contributing to the meticulous data processing endeavors. Fuzzy matching techniques were utilized to filter reports referencing the drug’s generic name, asciminib, and its product name, SCEMBLIX^®, specifically tagged as “primary suspect” drug names in the database. Given that this study hinged on a globally accessible database containing de-identified information without direct treatment intervention or the collection of human samples, the requirement for informed consent was deemed unnecessary.

Due to the inherent spontaneity of the reports, duplication was likely to occur. Therefore, a deduplication process was performed before the analysis, per the FDA guidelines [22]. We manually reviewed the reports to exclude instances where the PRIMARYID was lower when the CASEID matched. Additionally, we excluded any CASEID listed in the deleted cases file. AEs in the FAERS are classified based on the preferred terms (PTs) from the standardized Medical Dictionary for Regulatory Activities 26.0 (MedDRA 26.0). The classification is made into five levels: system organ class (SOC), high-level group term, high-level term, PT, and lowest level term [22, 23]. Figure 1 depicts a flowchart to illustrate the sequential procedures involved in extracting, processing, and analyzing data. The comprehensive methodology, which includes FAERS fields, is provided in the online supplementary materials (for all online suppl. material, see https://doi.org/10.1159/000540542).

In the context of AEs in drug-related reports, it is the responsibility of the individuals reporting these events to assign appropriate codes. The available codes for assignment were as follows: 1 = suspect, 2 = concomitant, and 3 = interacting. To ensure utmost precision and accuracy in our analysis, we made a strategic decision to designate code “1” as “PS” (primary suspected) within the DRUG files with the aim of enhancing analytical accuracy and overall conclusions [19].

Disproportionality analysis is a vital technique employed in pharmacovigilance studies, playing a pivotal role in the identification of potential signals indicating AEs associated with a drug. This approach entails comparing the frequency of AEs linked to a specific medication against the occurrence of AEs related to all other drugs. Essentially, it relies on the concept that a signal emerges during data extraction when the incidence rate of a particular AE for a given drug significantly surpasses the background occurrence rate observed across the entire database. This deviation from normality must exceed a predetermined threshold or set of criteria to be considered statistically significant [19]. To explore the association between asciminib and AEs, we applied both Bayesian and frequentist approaches, employing various statistical measures such as the reporting odds ratio (ROR), proportional reporting ratio (PRR), information component (IC), and empirical Bayes geometric mean (EBGM) [24‒26]. The equations and criteria for these four algorithms are detailed in online supplementary Tables S1 and S2.

During the research, 3,913,574 reports of AEs were recorded during the study period. After eliminating duplicate entries, we identified 966 reports specifically associated with asciminib. Table 1 provides an overview of the key characteristics observed in patients who experienced asciminib-related AEs. Most of the submitted reports (over 65%) were made by healthcare professionals, with CML being the reported indication in most cases (81.54%). Males and females comprised 49.82% and 50.06% of the cases reported, respectively. The prevalence of AEs was higher among patients aged above 65 years as compared to those aged between 18 and 65 years (51.28% vs. 48.21%), with the median age being 71 years and an interquartile range spanning from 60 to 87 years. The highest incidence of AEs associated with asciminib was observed for combinations with nilotinib, imatinib, aspirin, dasatinib, and allopurinol. Hospitalization was the most frequently observed severe outcome, accounting for 30.56% of cases, while mortality occurred in 21.94% of cases. The majority of the AE reports primarily originated from the USA (52.28%), followed by Japan (9.57%) and Great Britain (5.01%); the remaining reports were from other countries (33.15%). In terms of reporting year, the highest number of AEs occurred in 2023 (63.98%), followed by 2022 (34.89%) and Q4 of 2021 (1.19%).

Table 2 details the signal reports for asciminib at the SOC level, demonstrating a notable distribution of asciminib-related AEs across 27 organ systems. Key SOCs meeting at least one of the four analysis criteria included general disorders and administration site conditions, investigations, gastrointestinal disorders, nervous system disorders, blood and lymphatic system disorders, musculoskeletal and connective tissue disorders, and cardiac disorders. The analysis identified 61 AE signals associated with asciminib spanning 17 organ system classes. Table 3 outlines 82 PTs alongside their corresponding 13 SOCs. Notable AEs detected encompassed decreases in platelet, neutrophil, and lymphocyte counts, increases in amylase and aspartate aminotransferase levels, and occurrences of musculoskeletal pain and mental fatigue. These findings are consistent with the side effects listed on the asciminib label. Importantly, this investigation also uncovered 26 previously unrecognized and significant AEs linked to off-label usage, such as pleural effusion, pulmonary hypertension, gynecomastia, nephrotic syndrome, orchitis, pyelonephritis, hepatotoxicity, and pancreatitis.

Additionally, certain AEs unrelated to asciminib at the PT level were identified (online suppl. Table S3), primarily including benign, malignant, and unspecified neoplasms (including cysts and polyps) (SOC: 10029104). However, the occurrence of benign, malignant, and unspecified neoplasms (including cysts and polyps) may potentially be more closely associated with cancer progression.

The initiation period of asciminib-related AEs was determined through meticulous data collection from the database. After eliminating duplicate and inaccurate reports, 136 AEs were analyzed for details regarding their onset time. The median initiation period was calculated to be 52.5 days (interquartile range: 17–122.75 days).

Asciminib is being recognized as a novel standard for CML [5, 6]. The ongoing Phase III asciminib trials, such as the ASC4FIRST trial, aim to explore the application of asciminib as a first-line treatment option for newly diagnosed Ph+ CML-CP patients [27]. Previous research on asciminib has primarily concentrated on clinical trials, mechanisms of action, and literature reviews, leaving a gap in real-world evidence concerning the drug’s safety profile. This study is the first pharmacovigilance effort to investigate the potential links between asciminib and AEs by utilizing the FAERS database for post-marketing safety evaluation. Our findings indicate that of the total AEs analyzed, 404 instances (49.82% of the cohort) were reported in males, and 406 cases (50.06% of the cohort) were in females, suggesting a comparable occurrence rate of AEs between genders. Additionally, our results underscore the necessity of vigilant AE monitoring in elderly male patients, particularly those aged over 65 years, due to the observed higher incidence rate in this demographic. Promptly identifying these events is critical, considering their potential for severe life-threatening risks and their impact on disease management and progression.

When administered in patients with Ph-positive CML-CP previously treated with two or more tyrosine kinase inhibitors (TKIs), asciminib commonly induces AEs such as hematological toxicity, pancreatic complications, hypertension, hypersensitivity reactions, and cardiovascular issues. These AEs are explicitly stated in the drug description and have been substantiated by our research findings.

Based on the disproportionality analysis, the prevalent and significant indications observed at the level of SOC were blood and lymphatic system disorders, cardiac disorders, and gastrointestinal disorders. Our findings are consistent with the AEs reported during both phase 1 and phase 3 clinical trials [8, 14, 16, 18, 28‒30]. In the phase 3 trial, approximately 30% of patients receiving asciminib exhibited thrombocytopenia, with 21.8% of cases categorized as grade 3 or 4 events within 24 weeks [14]. Hematologic AEs were the most frequently observed across all treatment groups, as expected with TKI therapy. However, emerging evidence suggests that asciminib may be associated with a lower incidence of grade 3 or higher neutropenia, leukopenia, and anemia compared to comparator treatments. Further investigation is warranted to assess potential delayed AEs [5, 6, 31, 32]. Our research findings validate the high prevalence of cytopenia as the most common blood and lymphatic system disorder, with signal intensities recorded at ROR, PRR, IC, and EBGM of 26.29 (17.09–40.44), 26.08 (503.45), 4.70 (3.05), and 25.92 (16.85), respectively. Therefore, to ensure safety, clinicians should perform complete blood counts every 2 weeks for the first 3 months of treatment and monthly after that or as clinically indicated. The severity of the AEs of thrombocytopenia and/or neutropenia should further determine whether treatment needs to be continued, decreased, or terminated.

Furthermore, asciminib has been associated with significant signals of AEs, including cardiac disorders, specifically cardiac failure, acute myocardial infarction, and pericardial effusion [14]. A study demonstrated that the incidence of arterial occlusive events remained low across all groups involved in the trial [6]. However, given that prolonged exposure may increase the risk of such events, further follow-up is necessary to better evaluate long-term safety concerns. Notably, in the ASCEMBL trial, patients with preexisting cardiovascular risk factors and who had been heavily pretreated with TKIs were included. Those who received asciminib experienced a decrease in the incidence of arterial occlusive events adjusted for exposure over time [33]. In our analysis, although the number of angina pectoris cases was low, at only 5, we observed a significant signal intensity with ROR of 6.16 (2.56–14.82), PRR of 6.15 (21.52), IC of 2.62 (1.09), and EBGM of 6.14 (2.55).

In the phase I study, the latest follow-up (median duration: 14 months) revealed a low incidence of pleural effusion in patients undergoing asciminib therapy, accounting for only 6% of the cases [13]. Although the phase 3 ASCEMBL trial has not reported any documented case of pleural effusion or pulmonary hypertension associated with the use of asciminib as a standalone treatment [14], we observed signals for respiratory, thoracic, and mediastinal disorders (ROR: 0.97 [0.77–1.22]; PRR: 0.97 [0.06]; IC: −0.04 [−0.05]; and EBGM: 0.97 [0.77]) in our study.

Asciminib demonstrates affinity toward BCRP and P-gp, acting as a substrate for these transporters. Additionally, asciminib inhibits BCRP, P-gp, OATP1B1, OATP1B3, and OCT1 [34‒36]. Therefore, concomitant administration of asciminib with substrates for OATP1B or BCRP could lead to increased exposure in a dosage-dependent manner, resulting in an increased risk of associated AEs. Our study revealed that asciminib-related musculoskeletal and connective tissue disorders had a signal strength, with ROR of 1.31 (1.08–1.59), PRR of 1.31 (1.08–1.59), IC of 0.37 (0.30), and EBGM of 1.29 (1.06). Hence, patients receiving asciminib at the recommended doses while concurrently using other OATP1B or BCRP substrates should be closely monitored for any negative responses [22].

As per the asciminib label, gastrointestinal issues were observed in more than 20% of patients, with diarrhea being a commonly reported ADR. However, diarrhea appears to be a rare occurrence with asciminib since no cases were reported during the phase 3 trial up to the 24-week data cutoff point [14]. On the other hand, our study identified additional associated AEs such as pancreatitis, melena, and gastric perforation. Among these gastrointestinal complications, gastric perforation is notable since it exhibited notable signal intensity.

The phase I study of asciminib therapy revealed that lipase elevation was the most common non-hematological AE, affecting almost 27% of patients. However, only 10% of these cases were classified as grade 3/4 events [13]. Crucially, during the 24-week phase 3 trial, no cases of pancreatitis were reported among patients treated with a 40 mg twice daily [14] dose of asciminib. However, our analysis identified a significant incidence of elevated lipase levels as a noteworthy AE associated with asciminib use. The strength of the signal for this event is reflected in a ROR of 93.71 (CI: 60.06–146.20), a PRR of 92.98 (with a corresponding number of reports at 1,780.77), an IC of 6.51 (4.17), and an EBGM of 91.00 (CI: 58.33). Elevated lipase levels in patients undergoing asciminib treatment highlights a potential link to the drug’s role as a TKI, which is known to be associated with increased levels of pancreatic enzymes and the risk of subsequent pancreatitis. Present theories suggest that pancreatic dysfunction may result from the unintended inhibition of the c-ABL tyrosine kinase by asciminib [37].

There have been no documented cases of pleural effusion or pulmonary hypertension linked to asciminib monotherapy in the phase 3 ASCEMBL trial [14]. However, our findings indicate a correlation between asciminib and respiratory, thoracic, and mediastinal disorders, with a signal strength represented by ROR of 8.90 (5.52–14.35), PRR of 8.85 (118.18), IC of 3.14 (1.95), and EBGM of 8.83 (5.48). Additionally, we found a signal linking asciminib to vascular disorders, with the correlation represented by ROR of 3.36 (2.31–4.87), PRR of 3.33 (45.82), IC of 1.74 (1.20), and EBGM of 3.33 (2.29). The potential occurrence of AEs, such as pleural effusion or pulmonary hypertension, tends to manifest after prolonged administration. This indicates the need for careful consideration when prescribing asciminib for long-term administration.

In both trials, asciminib demonstrated good tolerability in patients with impaired kidney and liver function. The safety profile of asciminib in these patient populations was consistent in both studies, with no new safety concerns identified [13, 14, 38, 39]. Additionally, in both trials, most of the AEs reported were of mild severity (grade 1 or 2), although one instance of grade 3 neutropenia occurred in a participant with severe renal impairment on day 2; however, this AE resolved spontaneously, without any need for intervention, and did not have any clinical impact [40]. Here, it is important to highlight that our study suggests a potential association between asciminib and renal failure, as indicated by the ROR of 7.85 (5.62–10.97), PRR of 7.76 (206.04), IC of 2.95 (2.11), and EBGM of 7.75 (5.55). Healthcare professionals should consider these points when prescribing asciminib in these patient populations. Thus, healthcare providers prescribing asciminib should closely monitor patients for elevated pancreatic enzyme levels, high blood pressure, adverse cardiovascular effects such as ischemia and thromboembolism, and reduced neutrophil and platelet counts [41].

In terms of demographic and clinical features, no clinically significant differences were observed in the pharmacokinetics of asciminib with respect to sex, age (ranging from 20 to 88 years), race (including Asian, White, Black/African American), body weight (ranging from 42 to 184 kg), or mild to severe renal impairment (estimated glomerular filtration rate ranging from 15 to 89 mL/min/1.73 m²). Similarly, no notable variations were found in individuals with hepatic impairment within the specified range of total bilirubin levels and aspartate aminotransferase activity. However, this study revealed 26 previously unidentified and significant AEs associated with off-label use of asciminib, including gynecomastia, nephrotic syndrome, orchitis, pyelonephritis, hepatotoxicity, and pancreatitis. Since asciminib has been introduced relatively recently, it is crucial to establish the drug’s safety by extensively analyzing real-world self-reported AEs using big data. Therefore, forthcoming clinical investigations should entail extended follow-up durations for asciminib-associated AEs to enable more comprehensive monitoring.

This study is subject to certain limitations. Primarily, the reliance on the FAERS, which is based on voluntary reporting, introduces potential biases, including underreporting, inaccurate reports, and delays in submissions. These uncertainties may affect the robustness of our findings. Additionally, the FAERS database’s lack of complete patient exposure information hampers our capacity to precisely calculate the incidence of AEs and to establish a definitive causal relationship between the medication and the observed AEs. The absence of case narratives within the FAERS database further restricts our ability to assess causality thoroughly. Therefore, considering the absence of controlled study conditions, these AEs may manifest in diverse scenarios. Nonetheless, this underscores the imperative necessity for more rigorous research to apprise healthcare providers regarding potential guideline efficacy.

Moreover, since the FAERS data do not incorporate AE grading, the analysis was limited to documenting outcomes as severe or nonserious without the ability to categorize each AE individually. Our analysis also did not consider unmeasured confounding factors, such as potential drug-drug interactions, combinations of medications, and patient comorbidities, which could influence AEs. Given these constraints, further research is necessary to deepen our understanding and conclusions.

Based on the FAERS database, we used the disproportionality method to unveil safety signals and potential risks associated with asciminib use. Notably, this investigation uncovered 26 significant AEs linked to off-label asciminib use, encompassing conditions such as gynecomastia, nephrotic syndrome, orchitis, pyelonephritis, hepatotoxicity, and pancreatitis. Further research can be conducted to explore the causal relationship between them and medications, while being cautious about the impact of these AEs on the underlying diseases and patients. The robust signaling AEs we discovered in our research partially offset the limited sample size of this drug in clinical studies. Due to the absence of controlled study conditions, these AEs may occur in various scenarios. However, this emphasizes the need for more rigorous research to inform healthcare providers about the potential utility of guidelines. Therefore, further prospective clinical investigations are warranted to validate and elucidate the relationship between asciminib and these AEs. Crucially, this comprehensive post-marketing safety assessment is instrumental in enhancing our understanding of asciminib’s safety characteristics, offering valuable insights that will guide future research and clinical practices in this area.

Ethical approval and consent were not required as this study was based on publicly available data.

The authors declare no competing interests.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Z.L. and D.W.: conceptualization, methodology, data curation, formal analysis, visualization, and writing – original draft; C.K.: data curation and formal analysis; Q.N. and Y.C.: formal analysis and visualization; and Y.H. and M.C.: conceptualization, methodology, formal analysis, writing – original draft, writing – review and editing, and supervision. All authors strongly contributed to the interpretation of the results and to the reviews and editing of the draft. All authors read and approved the final version. The authors declare that all data were generated in-house and that no paper mill was used.

Zhijing Liu and Dongzhi Wu should be regarded as joint first authors.

All source code is publicly available on GitHub (https://github.com/MaohuaChen/source-code-FAERS-Asciminib-). Data sources and handling of the publicly available datasets used in this study are described in the Materials and Methods. The other data that support the findings of this study are available from the corresponding author upon reasonable request.