Adverse events associated with gepants: a pharmacovigilance analysis based on the FDA adverse event reporting system

Abstract

Background

Gepants have demonstrated notable benefits in migraine therapy, yet their safety profiles are not thoroughly investigated. This study comprehensively analyzed the adverse event (AE) risk signals of the currently approved gepants using the U.S. Food and Drug Administration Adverse Event Reporting System database, aiming to gain better understanding of their post-marketing safety features and potential risks.

Methods

All data of the gepants (rimegepant, atogepant, ubrogepant, and zavegepant) from January 1st 2020 to December 31st 2024 were retrieved from the database. Descriptive analysis was conducted to characterize the features of gepant-associated AEs. Disproportionality analysis and subsequent sensitivity analysis were employed to evaluate the risk signals of the gepants utilizing the algorithms of reporting odds ratio (ROR), proportional reporting ratio (PRR), and information component (IC).

Results

A total of 7766 reports of rimegepant, 3672 reports of atogepant, 1958 reports of ubrogepant, and 463 reports of zavegepant were identified after data processing. Most AEs were occurred within 30 days after gepant administration. The integration of disproportionality analysis and sensitivity analysis indicated that “feeling abnormal” was the most reported AE of rimegepant (n = 185, 6.81%, ROR₀₂₅ = 6.46, IC₀₂₅ = 2.59, PRR = 7.24, χ² = 998.58), while “constipation” was the most common AE of atogepant (n = 288, 16.09%, ROR₀₂₅ = 19.99, IC₀₂₅ = 4.10, PRR = 20.72, χ² = 5418.12). The most prevalent AE of ubrogepant was “fatigue” (n = 60, 7.19%, ROR₀₂₅ = 1.88, IC₀₂₅ = 0.84, PRR = 2.38, χ² = 48.82), whereas “dysgeusia” was the most frequently observed AE of zavegepant (n = 150, 45.18%, ROR₀₂₅ = 212.07, IC₀₂₅ = 6.10, PRR = 181.96, χ² = 26,975.74). Comparative analysis of AEs revealed that two AEs were shared among all gepants and zavegepant had the largest collection of unique AEs (n = 15).

Conclusions

The present pharmacovigilance study systematically revealed the significant risk signals of gepants. The common AEs and unique AEs of the four gepants were also identified and explored. Our results would provide valuable reference for the safe use of gepants, guiding personalized drug selection in clinical practice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10194-025-02091-3.

Introduction

Migraine is a prevalent neurological disease affecting more than one billion people worldwide [1]. As a complex neurovascular event, migraine is manifested as recurrent attacks of pulsatile unilateral headache lasting from hours to days, accompanied by symptoms of vomiting, nausea, vertigo, dizziness, phonophobia, and photophobia [2, 3]. There are currently multiple treatment options for migraine, including antiepileptics, analgesics, antidepressants, triptans, and ergots [4]. However, these drugs are still insufficient to fully meet clinical needs [5].

Calcitonin gene-related peptide (CGRP) is a peptide neurotransmitter that is thought to be positively correlated with headache severity [6]. Elevated CGRP level in the trigeminal vascular system is vital for the occurrence and development of migraine [7]. Research on the links between CGRP and migraine has promoted the development of CGRP-based therapies, including anti-CGRP monoclonal antibodies and small molecule CGRP receptor antagonists, unofficially known as gepants [8]. As relatively small CGRP receptor antagonists, gepants have shown great efficacy and advantages in migraine treatment given their short half-lives and ease of frequent administration in acute therapy [9]. Currently, there are four approved gepants, namely rimegepant, atogepant, ubrogepant, and zavegepant. Rimegepant, ubrogepant, and zavegepant are approved by the U.S. Food and Drug Administration (FDA) for acute treatment, while atogepant and rimegepant are approved for migraine prevention [10].

In previous clinical trials, the significant hepatotoxicity of first-generation gepant has led to its discontinuation in development [11]. As a result, the second-generation gepants were developed, including rimegepant, ubrogepant, and atogepant. Zavegepant is the only third-generation gepant approved for acute treatment of migraine, and its nasal spray form further improves patient compliance [12].

At present, the post-marketing safety data of gepants remains limited, with most of the existing data originating from clinical trials and subsequent meta-analysis [13–17]. Previous pharmacovigilance studies have also investigated the adverse events (AEs) of gepants [18–24]. However, these studies were limited to single or partial list of gepants, lacking comprehensive comparisons between gepants. Therefore, it is essential to systematically examine the potential safety profiles of gepants through pharmacovigilance analysis of real-world data. Here, we updated the safety features of all approved gepants by conducting disproportionality analysis based on the U.S. FDA Adverse Event Reporting System (FAERS) database. The FAERS database is a global spontaneous AE reporting system serving as a fundamental tool for drug safety monitoring [25, 26]. By mining and comparing the common and distinctive characteristics of AEs among gepants, our results would provide comprehensive insights into the safe use of gepants, guiding personalized drug selection in clinic.

Materials and methods

Data source and data cleaning

All existing pharmacovigilance data of the four gepants were retrieved from the FAERS database, encompassing the timeframe from January 1 st 2020 to December 31 st 2024. Data extraction was conducted on a single day of February 1 st 2025. The target drugs were restricted to the generic names of gepants, including “rimegepant”, “atogepant”, “ubrogepant”, and “zavegepant”, and reports were included if the gepants were marked as “prime suspect” in the database. We utilized the Medical Dictionary for Regulatory Activities (MedDRA, Version 27.0) to classify and describe AEs using system organ class (SOC) and preferred terminology (PT), respectively.

According to the FAERS database guidelines (https://fis.fda.gov/extensions/FPD-FAQ/FPD-FAQ.html), reports owning the same identifier values (e.g., case ID, date, and drug) were recognized as duplicate reports and were excluded from the analysis. The time-to-onset (TTO) of AE was defined as the interval from the first time of gepant treatment until the onset of AE. In TTO analysis and descriptive analysis, data entries with missing values or incorrectly formatted values were omitted. In addition, since FAERS allows the input of free text entries which results in a large amount of invalid data and abnormal data [27]. Hence, inappropriate PT, such as “drug ineffective”, “off label use”, “no adverse event”, “product quality issue”, and “adverse drug reaction”, were carefully removed to ensure data quality.

In disproportionality analysis, we noticed that the top ranked PTs, such as “migraine”, “headache”, “vomiting”, “nausea”, “vertigo”, “dizziness”, “photophobia”, and"phonophobia", actually belong to the medication indication and symptoms associated with migraine. Since different AE signals are interdependent in disproportionality analysis [28], the strong signals of the above PTs might mask the positive signals of other PTs, resulting in potential reporting bias. Therefore, we performed a subsequent sensitivity analysis in which the above PTs were excluded.

Data mining and analysis

To assess the risk signals of the four gepants, three distinct algorithms, namely reporting odds ratio (ROR), proportional reporting ratio (PRR), and Bayesian confidence propagation neural network (BCPNN), were employed in the disproportionality analysis and sensitivity analysis. Information component (IC) is the measure of BCPNN algorithm, and confidence interval (CI) is a range of upper and lower limits which describes the possible mean of a sample [29–31]. The ROR method is computationally simple and highly sensitive, whereas the BCPNN and PRR methods provide more stable and specific results through Bayesian logic and neural network structure [31, 32].

In this study, the three algorithms were integrated to minimize the potential bias of single-algorithm approach. ROR, PRR, and IC were calculated based on the 2 × 2 contingency table (Table 1). According to the formulas and significant criteria listed in Table 2, a significant signal was determined if the lower CI limits of ROR (ROR₀₂₅) and the lower CI limits of IC (IC₀₂₅) were greater than 1 and 0, respectively [30]. A significant PRR signal was considered if the value of PRR ≥ 2, and the value of chi-squared (χ²) ≥ 4 [30]. Data analysis was performed using SAS software (Version 9.4), and data visualization was conducted with GraphPad Prism (Version 9.3.1) and online plotting platform (https://www.chiplot.online/, https://bioincloud.tech/).

Table 1.

The 2 × 2 contingency table of disproportionality analysis

	Target adverse events	Other adverse events	Total
Target drugs	a	b	a + b
Other drugs	c	d	c + d
Total	a + c	b + d	a + b + c + d

Table 2.

Algorithms, formulas, and positive signal criteria of disproportionality analysis

Algorithms	Formulas	Criteria
ROR	ROR = =	ROR025 > 1, ≥ 3
	95%CI =
BCPNN	IC =	IC025 > 0, > 0
	IC025 =
PRR	PRR =	χ2 ≥ 4, PRR ≥ 2, ≥ 3
	χ2 =  O = , E =

Abbreviations: ROR Reporting odds ratio, ROR ₀₂₅ the lower limit of the 95% confidence interval of the ROR, 95%CI 95% confidence interval, BCPNN Bayesian confidence propagation neural network, IC Information component, IC₀₂₅ the lower limit of the 95% confidence interval of the IC, PRR Proportional reporting ratio, χ² chi-squared

Results

Descriptive analysis

After data processing, a total of 13,859 gepants-related AEs were obtained from the FAERS database. Rimegepant had the highest number of AEs reports (n = 7766), followed by atogepant (n = 3672), ubrogepant (n = 1958), and zavegepant (n = 463). The most common age group was 45–64 for rimegepant (n = 1735, 22.34%), atogepant (n = 521, 14.19%), and ubrogepant (n = 201, 10.27%), whereas the most prevalent age range was 18–44 for zavegepant (n = 71, 15.33%). The majority of reports were originated from North America (n = 7538, 97.06% for rimegepant; n = 3434, 93.52% for atogepant; n = 1906, 97.34% for ubrogepant; n = 459, 99.14% for zavegepant), and were mostly reported by consumers (n = 6498, 83.67% for rimegepant; n = 2997, 81.62% for atogepant; n = 1574, 80.39% for ubrogepant; n = 170, 36.72% for zavegepant). The AEs of rimegepant (n = 2965, 38.18%) and ubrogepant (n = 604, 30.85%) were mostly reported in 2022. The majority of reports were observed in 2023 for atogepant (n = 1517, 41.31%), while most reports were submitted in 2024 for zavegepant (n = 236, 50.97%). Regarding reaction outcome, most of the cases were non-serious (n = 7150, 92.07% for rimegepant; n = 2900, 78.98% for atogepant; n = 1627, 83.09% for ubrogepant; n = 445, 96.11% for zavegepant). The detailed demographic characteristics of the reported cases are depicted in Table 3.

Table 3.

Demographic characteristics of cases with gepant-related adverse events

Characteristics	Rimegepant	Atogepant	Ubrogepant	Zavegepant
	N (%)	N (%)	N (%)	N (%)
Number of reports	7766	3672	1958	463
Gender
Female	4807 (61.90)	2809 (76.50)	1294 (66.09)	344 (74.30)
Male	754 (9.71)	458 (12.47)	253 (12.92)	39 (8.42)
Unknown	2205 (28.39)	405 (11.03)	411 (20.99)	80 (17.28)
Age
< 18	44 (0.57)	5 (0.14)	37 (1.89)	0 (0.00)
18–44	1297 (16.70)	385 (10.48)	157 (8.02)	71 (15.33)
45–64	1735 (22.34)	521(14.19)	201 (10.27)	56 (12.10)
≥ 65	676 (8.70)	143 (3.89)	95 (4.85)	11 (2.38)
Unknown	4014 (51.69)	2618(71.30)	1468 (74.97)	325 (70.19)
Median (Q1, Q3)	51 (40, 61)	50 (39, 60)	50 (37, 62)	44 (35, 54)
Geographical distribution
North America	7538 (97.06)	3434 (93.52)	1906 (97.34)	459 (99.14)
South America	16 (0.21)	0 (0.00)	0 (0.00)	0 (0.00)
Europe	73 (0.94)	49 (1.33)	0 (0.00)	0 (0.00)
Asia	11 (0.14)	4 (0.11)	2 (0.10)	0 (0.00)
Unknown	128 (1.65)	185 (5.04)	50 (2.55)	4 (0.86)
Reporter type
Consumer	6498 (83.67)	2997 (81.62)	1574 (80.39)	170 (36.72)
Physician	519 (6.68)	306 (8.33)	150 (7.66)	136 (29.37)
Pharmacist	734 (9.45)	361 (9.83)	230 (11.75)	157 (33.91)
Unknown	15 (0.19)	8 (0.22)	4 (0.20)	0 (0.00)
Reporting year
2020	586 (7.55)	0 (0.00)	289 (14.76)	0 (0.00)
2021	1677 (21.59)	19 (0.52)	350 (17.88)	0 (0.00)
2022	2965 (38.18)	1463 (39.84)	604 (30.85)	0 (0.00)
2023	1388 (17.87)	1517 (41.31)	388 (19.82)	227(49.03)
2024	1150 (14.81)	673 (18.33)	327 (16.70)	236 (50.97)
Reaction outcomea
Non-serious	7150 (92.07)	2900 (78.98)	1627 (83.09)	445 (96.11)
Life-Threatening	21 (0.27)	11 (0.30)	2 (0.10)	0 (0.00)
Hospitalization	111 (1.43)	159 (4.33)	87 (4.44)	5 (1.08)
Disability	31 (0.40)	39 (1.06)	28 (1.43)	0 (0.00)
Death	69 (0.89)	28 (0.76)	30 (1.53)	0 (0.00)
Congenital Anomaly	4 (0.05)	1 (0.03)	0 (0.00)	0 (0.00)
Required Intervention	11 (0.14)	21 (0.57)	6 (0.31)	0 (0.00)
Other	452 (5.82)	625 (17.02)	245 (12.51)	13 (2.81)

Abbreviations: Q1 Lower Quartile, Q3 Upper Quartile

^aSome cases may contain more than one outcome

Disproportionality analysis of AE signals

Figure 1 exhibits the top 30 most common PTs of gepants, and the ROR₀₂₅ values were transformed to logarithms to ensure readability of the graph. We found that “migraine” was the most frequently occurred PT of atogepant (n = 763, 20.78%, ROR₀₂₅ = 64.50, IC₀₂₅ = 5.74, PRR = 63.07, χ² = 46,252.62) and ubrogepant (n = 286, 14.61%, ROR₀₂₅ = 46.49, IC₀₂₅ = 5.20, PRR = 48.69, χ² = 13,333.40). The PT of “nausea” was most prevalent in rimegepant (n = 765, 9.85%, ROR₀₂₅ = 4.20, IC₀₂₅ = 2.00, PRR = 4.33, χ² = 1981.19), while “dysgeusia” was the most common PT of zavegepant (n = 153, 33.05%, ROR₀₂₅ = 131.45, IC₀₂₅ = 5.90, PRR = 130.83, χ² = 19,693.09). Moreover, we found consistent results of pharmacovigilance metrics across the three algorithms (Table S1).

Fig. 1.

The top 30 preferred terminologies and reporting odds ratio (ROR) of the four gepants in disproportionality analysis

Sensitivity analysis of AE signals

In order to eliminate reporting bias mediated by the indication and symptoms of migraine, a subsequent sensitivity analysis was conducted. Figure 2 demonstrates the top 30 most common PTs of gepants. We found that “feeling abnormal” was the most reported PT of rimegepant (n = 185, 6.81%, ROR₀₂₅ = 6.46, IC₀₂₅ = 2.59, PRR = 7.24, χ² = 998.58), while “constipation” was the most common PT of atogepant (n = 288, 16.09%, ROR₀₂₅ = 19.99, IC₀₂₅ = 4.10, PRR = 20.72, χ² = 5418.12). The most prevalent PT of ubrogepant was “fatigue” (n = 60, 7.19%, ROR₀₂₅ = 1.88, IC₀₂₅ = 0.84, PRR = 2.38, χ² = 48.82), whereas “dysgeusia” was the most frequently observed PT of zavegepant (n = 150, 45.18%, ROR₀₂₅ = 212.07, IC₀₂₅ = 6.10, PRR = 181.96, χ² = 26,975.74). Furthermore, we found similar pharmacovigilance data among all three algorithms (Table S2).

Fig. 2.

The top 30 preferred terminologies and reporting odds ratio (ROR) of the four gepants in sensitivity analysis

The AE signals in SOC level indicated that “gastrointestinal disorders” was most common in rimegepant (n = 968, ROR₀₂₅ = 1.99) and atogepant (n = 633, ROR₀₂₅ = 1.83). The “nervous system disorders” was the most frequently detected SOC in ubrogepant (n = 314, ROR₀₂₅ = 1.92) and zavegepant (n = 217, ROR₀₂₅ = 5.48) (Table 4).

Table 4.

The distribution and risk signals in system organ class of the four gepants

	No	SOC	N	ROR025
Rimegepant	1	Gastrointestinal disorders	968	1.99
	2	Nervous system disorders	675	1.32
	3	General disorders and administration site conditions	640	0.72
	4	Skin and subcutaneous tissue disorders	586	2.01
	5	Psychiatric disorders	471	1.54
	6	Respiratory, thoracic and mediastinal disorders	237	0.78
	7	Immune system disorders	230	3.20
	8	Musculoskeletal and connective tissue disorders	215	0.69
	9	Investigations	185	0.43
	10	Eye disorders	146	1.03
Atogepant	1	Gastrointestinal disorders	633	1.83
	2	Nervous system disorders	450	1.24
	3	General disorders and administration site conditions	409	0.65
	4	Psychiatric disorders	392	1.88
	5	Skin and subcutaneous tissue disorders	211	0.93
	6	Investigations	202	0.71
	7	Musculoskeletal and connective tissue disorders	168	0.78
	8	Metabolism and nutrition disorders	125	1.17
	9	Eye disorders	109	1.09
	10	Respiratory, thoracic and mediastinal disorders	105	0.46
Ubrogepant	1	Nervous system disorders	314	1.92
	2	General disorders and administration site conditions	242	0.80
	3	Gastrointestinal disorders	211	1.13
	4	Psychiatric disorders	128	1.15
	5	Skin and subcutaneous tissue disorders	106	0.93
	6	Musculoskeletal and connective tissue disorders	102	0.96
	7	Respiratory, thoracic and mediastinal disorders	80	0.73
	8	Eye disorders	70	1.41
	9	Investigations	69	0.45
	10	Cardiac disorders	43	0.62
Zavegepant	1	Nervous system disorders	217	5.48
	2	Respiratory, thoracic and mediastinal disorders	157	6.37
	3	Gastrointestinal disorders	33	0.42
	4	Eye disorders	24	1.28
	5	Skin and subcutaneous tissue disorders	21	0.44
	6	General disorders and administration site conditions	17	0.11
	7	Immune system disorders	10	0.79
	8	Musculoskeletal and connective tissue disorders	8	0.14
	9	Psychiatric disorders	8	0.13
	10	Ear and labyrinth disorders	7	1.18

Abbreviations: SOC System organ class, ROR₀₂₅ the lower limit of the 95% two-sided reporting odds ratio of the information component

TTO analysis

After data processing, 449 reports of rimegepant, 354 reports of atogepant, 150 reports of ubrogepant, and 43 reports of zavegepant were included in the TTO analysis. We found that the majority of AEs occurred within 30 days after gepant administration (n = 389, 86.64% for rimegepant; n = 219, 61.86% for atogepant; n = 121, 80.67% for ubrogepant; n = 43, 100.00% for zavegepant). The median TTO of the four gepants were as follows: rimegepant (median, [Lower Quartile (Q1), Upper Quartile (Q3)]) = 0 [0, 6]; atogepant (median, [Q1, Q3]) = 13 [0, 84]; ubrogepant (median, [Q1, Q3]) = 0 [0, 15]; zavegepant (median, [Q1, Q3]) = 0 [0, 0] (Fig. 3).

Fig. 3.

The onset time of adverse events of the four gepants

Comparisons of AE signals at the PT level

A comparative analysis of PTs was performed to evaluate the overlapping and distinctive features of AEs among gepants. Only the top 20 positive PTs of each gepant were shown considering readability of the graph. As depicted in Fig. 4, “abdominal discomfort” and “abdominal pain upper” were shared among all four gepants, and zavegepant had the greatest number of unique positive PTs (n = 15), including “lacrimation increased”, “oropharyngeal pain”, “oropharyngeal discomfort”, “nasal congestion”, “eye swelling”, “rhinorrhea”, “upper-airway cough syndrome”, “sinus pain”, “taste disorder”, “nasal discomfort”, “epistaxis”, “rhinalgia”, “eye irritation”, “dysgeusia”, and “burning sensation”.

Fig. 4.

The Venn network of the top 20 positive preferred terminologies of the four gepants

Discussion

To the best of our understanding, this is the first real-world pharmacovigilance study investigating the AEs of all the approved gepants. In descriptive analysis, the higher number of AEs in female patients echoed the epidemiological evidence that migraine predominantly affects women, and the mechanism behind it is likely to be related to the fluctuations in estrogen levels [33–35]. Our data regarding age of patients also supported the view that migraine shows higher prevalence in young and middle-aged women [12]. Of note, we found that the vast majority of reports were from North America and were mainly reported by consumers rather than healthcare professionals, which might affect the generalizability and reliability of the data. Therefore, the inclusion of AE data from other databases (e.g., VigiBase and EudraVigilance) and education of AE reporting among healthcare professionals would be important for future studies.

After integrating disproportionality analysis with sensitivity analysis, we found more diverse AEs compared to the FDA instructions (Table S3). In neurological and psychiatric signals, we found significant PTs of “insomnia”, “somnolence”, and “anxiety” in rimegepant, atogepant, and ubrogepant. The AEs associated with sleep alteration might be linked to the neuromodulatory effects of CGRP on glutamatergic signaling which affecting central sensitization [36]. In addition, CGRP-related anxiety-like behaviors in stress and pain states are also thought to be closely associated with sleep function [37]. The PTs of “feeling abnormal”, “paresthesia”, “burning sensation”, “hypoesthesia”, “memory impairment”, and “brain fog” might be related to the widespread expression of CGRP in brain regions associated with memory and sensory function [38]. Indeed, there is a strong relationship between sleep quality, sensory input, and migraine [39]. Hence, future studies collecting detailed records of relevant medical history would contribute to the deeper understanding of these AEs.

In the SOC of “gastrointestinal disorders”, the PT of “constipation” was frequently reported in second-generation gepants. CGRP serves as a regulatory factor of gastrointestinal activity and is widely discovered in the gastrointestinal system [36, 40]. Results from phase III clinical trials revealed that 6.9–7.7% of patients had atogepant-related constipation [41]. Another post-marketing investigation showed that 4.7% of patients developed ubrogepant-induced constipation [42]. Importantly, FDA instruction has noted a warning on the risk of atogepant-induced constipation [43]. Consistent with these findings, we found that atogepant (ROR₀₂₅ = 19.99) displayed the strongest signal of constipation, followed by rimegepant (ROR₀₂₅ = 3.09) and ubrogepant (ROR₀₂₅ = 2.45). Although these results could be attributed to the notoriety effect that severe or labeled AEs tend to be overreported [44], gepant-related constipation remains a significant AE that warrants attention.

As for skin and subcutaneous tissue disorders, “pruritus” was the most common PT among second-generation gepants. This PT might be related to the modulating role of CGRP in pruritus that CGRP can affect itch sensation by regulating interleukin-31 and thymic stromal lymphopoietin [45, 46]. Another new PT “alopecia” was detected in rimegepant and atogepant. A large body of evidence revealed the mechanism of microvascular circulation disorders in alopecia, that is, decreased cutaneous CGRP levels can cause vasoconstriction and hair growth dysfunctions [47–49]. Thus, close collaboration between clinicians and dermatologists would be highly beneficial for patients receiving second-generation gepants.

In terms of cardiac disorders, the PT of “palpitations” was reported in all second-generation gepants. CGRP receptors are considered to regulate smooth muscle cells of the cardiovascular system, though no severe cardiovascular AE associated with anti-CGRP drugs have been reported so far [50, 51]. Clinical evidence suggested that cardiovascular AEs caused by anti-CGRP drugs are mainly palpitations, flushing, hypotension, and warm sensation [52]. Considering the complex confounding factors of this category of AE, the background characteristics of patients, including age, basal metabolism, vasomotor, cardiovascular parameters, and menopausal status, should be comprehensively weighed when receiving second-generation gepants [53].

Moreover, given the hepatotoxicity of first-generation gepant, we additionally examined the hepatic safety profiles of the four gepants. Previous study using physiologically based pharmacokinetic modeling has reported no significant hepatotoxicity of the four gepants [54]. Our results aligned with previous studies that hepatotoxicity-related AEs were rarely observed (rimegepant, n = 14; atogepant, n = 27; ubrogepant, n = 12; zavegepant, n = 0). However, although there is no FDA warning for hepatotoxicity of gepants, since hepatobiliary elimination is their primary excretion route [55], clinically based evidence suggested that gepants are not recommended for patients with severe hepatic impairment [15, 56].

Zavegepant is an effective non-oral option that may benefit individuals requiring rapid symptom relief or those unable to tolerate oral medications [57]. Given the recent market entry of zavegepant, its real-world safety profiles are currently lacking. In this study, we collected the largest sample size of zavegepant-related AEs to date. Our real-world data covered broader range of populations compared to clinical trials, aiding in the discovery of unknown AE signals of zavegepant. Clinical trial-based data indicated that zavegepant exhibits good tolerability over 52 weeks of long-term treatment [58], and the common AEs were mild or moderate levels of dysgeusia, nasal discomfort, nausea, nasal congestion, throat irritation, and back pain [59, 60]. In this study, we found that zavegepant did not exhibit the common AEs of second-generation gepants. Notably, “dysgeusia” and “taste disorder” were recognized as the unique PTs of zavegepant, and the underlying mechanism could be that zavegepant affects the taste buds via nasopharyngeal route [61]. Previous clinical trial data showed that 39.1% of patients experienced zavegepant-related dysgeusia [59], which was generally consistent with our results (45.18%). Other clinical trial evidence suggested that the AEs of taste disorders are mostly mild and can be resolved without intervention [59]. Our results supported this evidence that AEs of zavegepant were mostly non-serious (96.11%). We also identified other unique PTs of zavegepant, including “nasal discomfort” “rhinorrhea”, “epistaxis”, “nasal congestion”, and “rhinalgia”, which were considered as local irritation AEs associated with nasal administration [62]. However, we should emphasize that the robustness of signal detection might be constrained due to the limited number of zavegepant-related AE reports collected. Furthermore, we should reinforce the fact that continuous monitoring of zavegepant-related AE is critical given its novel administration routine.

TTO analysis is an important method for healthcare professionals to predict and identify AEs [63]. Due to the large number of missing or invalid values in FAERS database, only 7.19% of the collected reports were included in the TTO analysis of this study. Our results indicated that most AEs occurred within 30 days following gepant administration, suggesting the necessity of close monitoring of the patients, especially during the first month. We also noticed that atogepant had the largest median TTO value among all gepants. Further subgroup analysis revealed greater TTO values of atogepant in certain PTs, including “alopecia” (108.00, [14.00,274.00]), “memory impairment” (89.00, [88.00,345.00]), and “seizure” (50.00, [29.00,61.00]). Such delayed TTO of atogepant appeared to be associated with its relatively longer elimination half-life compared to other gepants [64]. Therefore, education of potential delayed AEs and long-term periodic monitoring should be required in patients receiving atogepant treatment.

This study has several limitations that require acknowledgement. First, since FAERS relies on spontaneous reporting, reports with incomplete or incorrect data are thereby inevitable. This resulted in an insufficient number of cases in descriptive analysis and TTO analysis. Furthermore, due to the intrinsic limitation that FAERS lacks the information of detailed medical records and denominator information, all results only allowed us to determine potential drug-AE correlations rather than confirmed causality. Follow-up studies in combination with other sources, such as medical claims and electronic health records, would provide more comprehensive assessment of AE causality. Lastly, the extensive disproportionality analysis of gepants and PTs in this study might potentially increase the risk of false-positive signals. Subsequent studies applying multiple false discovery rate corrections (e.g., Benjamini–Hochberg and Bonferroni) would be helpful in limiting false-positive signals.

Conclusions

Our study comprehensively analyzed the significant safety signals of the approved gepants. By mining and comparing the common and distinctive features of AEs among gepants, our results provided valuable reference for the safe use of gepants, offering guidance for clinicians to tailor individualized treatments.

Abbreviations

AE

Adverse event

BCPNN

Bayesian confidence propagation neural network

CGRP

Calcitonin gene-related peptide

CI

Confidence interval

FAERS

FDA Adverse Event Reporting System

FDA

Food and Drug Administration

IC

Information component

PRR

Proportional reporting ratio

PT

Preferred terminology

ROR

Reporting odds ratio

SOC

System organ class

TTO

Time-to-onset

Authors’ contributions

YT contributed to the conception and design of the study. QS, SG, and YT collected the data. QS analyzed the data. QS and YT drafted the manuscript. All authors approved the final version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (82104223 and 82204372), Guangdong Basic and Applied Basic Research Foundation (2020A1515110008), Science and Technology Program of Guangzhou (202102021022, 2024A04J10001, and 2025A03J3308), Medical Scientific Research Foundation of Guangdong (A2024618), and Guangzhou Municipal Key Discipline in Medicine (2025–2027).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Since all data of our study was originated from public databases with non-identifiable datasets, no ethical approval is required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.