A real‐world disproportionality analysis of semaglutide: Post‐marketing pharmacovigilance data

Abstract

Aim/Introduction

The recent adverse reactions associated with semaglutide have led the Food and Drug Administration (FDA) to issue a “black box warning”, and it is necessary to analyze all reports of adverse reactions to improve the safety of its clinical use.

Materials and Methods

Statistical analyses and signal mining were performed by obtaining the adverse event reports related to semaglutide in the FAERS database from the first quarter of 2018 to the fourth quarter of 2023. We used disproportionality and Bayesian analysis to examine clinical and demographic attributes, trends reported quarterly, and contrasts between two distinct indications (obesity and type 2 diabetes).

Results

We found 10 unexpected adverse signals related to “pancreatic cancer”, “intestinal obstruction”, “cholecystitis”, and “polycystic ovary” and both the two different indications had the same serious adverse reaction events occurring.

Conclusions

This study identified many unexpected signals of serious adverse reactions, suggesting the importance of continuous post‐marketing surveillance of semaglutide to understand its potential risks.

This study identified many unexpected signals of serious adverse reactions, suggesting the importance of continuous post‐marketing surveillance of semaglutide to understand its potential risks.

INTRODUCTION

Diabetes mellitus is widely recognized as a complex cardio‐renal metabolic disease, and its prevalence and associated complications are increasing year by year. According to a global burden of disease study in 2023 ¹ , ² , it is estimated that there will be 1.31 billion diabetic patients worldwide by 2050, 95.4% of whom will have type 2 diabetes mellitus. The persistent progression of the disease process brings about metabolic and endo‐environmental homeostatic imbalances, which can affect the integrity of the vasculature and thus cause organ failure and even premature death. Diabetes mellitus is currently in an epidemic phase as a complication alongside obesity ³ . The World Health Organization reports that obesity has become the disease with the highest complication mortality rate, with about 5 million fatalities from diabetes, cancer and other diseases due to a BMI higher than optimal in 2019 ⁴ . The relevance and severity of the disease have attracted the attention of more and more scholars to the relevant therapeutic drugs.

Research has found that glucagon‐like peptide‐1 (GLP‐1) promotes insulin secretion, thereby making glucagon‐like peptide‐1 receptor agonists (GLP‐1RA) one of the important targets for the treatment of metabolism‐related diseases ⁵ . Meanwhile, semaglutide is a GLP‐1RA that inhibits gastric emptying and hepatic gluconeogenesis and has received approval from the Food and Drug Administration (FDA) in 2017 for the treatment of diabetes mellitus ⁶ . In recent years, semaglutide has also been approved for the treatment of obesity by the FDA and by the National Institute for Health and Care Excellence in England ⁷ , ⁸ . Importantly, based on the satisfactory efficacy of semaglutide in treating metabolic disorders, current research is focusing on multiple indications in the liver, brain, gut, and pancreas, which are related to complications and comorbidities associated with obesity ⁹ , ¹⁰ .

However, following the application of semaglutide in the treatment of diabetes mellitus, there is an increased risk of adverse events (AEs) occurring in patients, such as pancreatitis and medullary thyroid carcinoma, which have been labeled as safety alerts by the FDA. It is unknown whether the use of semaglutide for obesity and its related complications and comorbidities will also cause AEs. There may be more potential signals of adverse events. This raises concerns about the adverse reactions associated with semaglutide in this study.

Given that these medications have had a short time to market, there is currently a lack of international analysis of real‐world and comprehensive data related to semaglutide. ¹¹ Most of the available drug reports are sponsored by pharmaceutical companies, which have a limited ability to detect adverse reactions.

Pharmacovigilance is based on the monitoring of drug safety in the real world and the identification of new AEs that are unknown at the time of marketing approval. ¹² The main components of signal detection are data extraction, data collection, data selection, and data analysis and interpretation. Ultimately, the focus is on the quality of the data collected and making the best decision based on scientific evaluation. ¹³ The FDA Adverse Event Reporting System (FAERS) is a database that provides public access to postmarketing information about drugs and therapeutic biological products, with a spontaneous reporting system that allows for broad surveillance and detection of suspected drug AEs. ¹⁴ Therefore, we aimed to assess quantitatively the adverse events of semaglutide in FAERS from the first quarter of 2018 to the second quarter of 2023 by detecting signals through scientific algorithms to gain an accurate insight into potential adverse reaction signals and to quantify potential medication risks.

MATERIALS AND METHODS

Source and cleaning

This study obtained real data on semaglutide‐associated adverse reactions from the first quarter of 2018 to the second quarter of 2023, including seven comprehensive datasets that provide information on different aspects of drug use and adverse events. These datasets include demographic and administrative information (DEMO), drug‐related details (DRUG), reports on adverse drug reactions (REAC), patient outcomes (OUTC), sources of the reports (RPSR), drug therapy duration (THER), and indications for use or diagnosis (INDI). In compliance with FDA guidelines, our study implemented a rigorous process to identify and eliminate duplicate reports to ensure the accuracy and reliability of a comprehensive assessment of the safety of semaglutide. We selected the most recent FDA_DT entries with the same CASEID. Where CASEID and FDA_DT were the same, we prioritized the higher PRIMARYID to ensure that redundant data were excluded.

Identification of the drug product and adverse reactions

Data from the FAERS database were preprocessed using SAS and MySQL, and duplicate reports were removed to collect clean and standardized data. Initial screening of AEs was performed after data acquisition. We used semaglutide as the primary suspect (PS) drug and drug names were standardized according to Drugbank and Drugs@FDA. Corresponding data were screened in the Drug and Role code entries of the CASE_REAC and CASE_DRUG datasets. Adverse events reported in FAERS were categorized according to the Medical Dictionary for Regulatory Activities (MedDRA) terminology using the following terms Preferred Terms (PTs). To ensure a comprehensive assessment of the toxicity profile, we carefully screened relevant individual AE reports at the PT level with strict reference to MedDRA and mapped them to the corresponding system organ class (SOC).

Signal mining and statistical analysis

We used the proportional disequilibrium method, which is commonly used in pharmacovigilance studies, to uncover potential positive signals between semaglutide and AEs (Table 1). We specifically focused on analyzing AEs attributed to the studied drugs rather than those caused by the underlying disease state. In this study, we mainly used the proportional reporting ratio (PRR), reporting odds ratio (ROR), Bayesian confidence propagation neural network (BCPNN), and empirical Bayesian geometric mean (EBGM). The formulas and screening criteria of the above four signal detection methods are expressed as shown in Table 2. Valid adverse drug reaction results should simultaneously meet the positive signal selection criteria of the above four methods: PRR >2, ROR >2, IC025 >0, and EBGM05 >2.

Table 1.

Fourfold table of disproportionality method

Medicine	Target adverse events reported	Other adverse events reported	Summation
Target drugs	a	b	a + b
Other drugs	c	d	c + d
Summation	a + c	b + d	a + b + c + d

Table 2.

Four major algorithms used to assess potential associations between semaglutide and adverse events

Algorithms	Equation	Criteria
ROR	ROR = (ad)/(bc) 95% CI = eln(ROR) ± 1.96(1/a + 1/b + 1/c + 1/d)0.5, a ≥ 3	ROR >2; lower limit of 95% CI >1; N ≥ 3
PRR	PRR = a (c + d)/c/(a + b) χ2 = [(ad − bc)2](a + b + c + d)/[(a + b)(c + d)(a + c)(b + d)]	PRR >2; χ2 ≥ 4; N ≥ 3
BPCNN	IC = log2a(a + b + c + d)/((a + c)(a + b)) 95% CI = E(IC) ± 2 V(IC)0.5	IC025 >0
EBGM	EBGM = a(a + b + c + d)/(a + c)/(a + b) 95% CI = eln(EBGM) ± 1.96(1/a + 1/b + 1/c + 1/d)0.5	EBGM05 >2

95% CI, 95% confidence interval; E(IC), the IC expectations; EBGM, empirical Bayesian geometric mean; EBGM05, the lower limit of 95% CI of EBGM; IC, information component; IC025, the lower limit of 95% CI of the IC; N, the number of reports; V(IC), the variance of IC; χ², chi‐squared.

We also counted data on demographic and clinical characteristics obtained from CASE_DEMO, CASE_OUTC, CASE_RPSR, CASE_THER, and CASE_INDI concerning gender, age, and adverse outcomes concerning the region of origin of the report. In particular, we focused on the indications for use or diagnosis of semaglutide and analyzed the associated AEs generated under the two main indications of obesity and type 2 diabetes mellitus in comparison. In addition, we visualized the number of reports per quarter since the introduction of semaglutide by gender. All data related to semaglutide were processed and statistically analyzed using software such as SAS, MySQL, EXCEL, and R language. The approximate flow chart of this study is shown in Figure 1.

Figure 1.

The flow diagram of selecting semaglutide‐related adverse events from the FAERS database.

RESULTS

Basic information and demographic characteristics of adverse events

We pooled the characteristics of the semaglutide adverse events (Table 3). There were 14,512 AEs related to semaglutide, excluding the 625 reports with unknown or missing gender (4.31%), the remaining reports had 8,667 reports of patients who were female (59.72%), and 5,220 (35.97%) were male. Excluding the 7,413 reports with unknown or missing age, the median age of the patients was 50 years, and 17 reports of patients less than 20 years old (0.12%) were obtained. There were 2087 reports (14.38%) of AEs occurring in patients aged 60–69 years accounting for the largest proportion of patients. 2,687 reports of AEs were related to serious outcomes accompanied by semaglutide. Serious outcomes due to semaglutide were predominantly hospitalized (14.68%). The main reporting country was the United States (90.54%), followed by Canada (2.16%), the United Kingdom (2.15%), Japan (1.93%), and France (0.99%). In addition, the number of reported adverse events has increased each year since the introduction of semaglutide, and the number of reported adverse reactions is generally higher in women than in men (Figure 2).

Table 3.

Clinical characteristics of reports with semaglutide from the Food and Drug Administration Adverse Event Reporting System (FAERS) database (2018 Q1–2023 Q2)

Characteristics	Case number	Case proportion (%)
Number of events	14,512
Gender
Male	5,220	35.97
Female	8,667	59.72
Unknown	625	4.31
Age
<19	17	0.12
20–29	131	0.90
30–39	378	2.60
40–49	853	5.88
50–59	1,571	10.83
60–69	2087	14.38
70–79	1,646	11.34
>80	416	2.87
Unknown	7,413	51.08
Serious outcome
Hospitalization	2,130	14.68
Disability	221	1.52
Life‐threatening	179	1.23
Death	157	1.08
Unknown	11,825	81.48
Reported countries
US (United States)	12,443	85.74
CA (Canada)	302	2.08
GB (Great Britain)	283	1.95
JP (Japan)	256	1.76
FR (France)	135	0.93
Others	1,093	7.53
Dosage form
Oral	1,807	12.45
Subtance	10,461	72.09
Unknown	2,244	15.46

Figure 2.

The number of reported adverse events and gender differences in semaglutide during 2018 Q1–2023 Q2.

Detection of AE signals based on SOC and PT level

The AEs of semaglutide involved 25 organ system classifications, and after excluding SOCs not related to the pharmacological effects of semaglutide itself (product problems, various types of injuries, poisonings, and operational complications, various surgical and medical procedures, and the social environment), a total of 21 SOCs were obtained (Table 4 and Figure 3). Among them, the highest number of reports was for gastrointestinal disorders (n = 10,214, IC025 = 2.34), followed by various types of tests (n = 2,124, IC025 = 2.46), metabolic and nutritional disorders (n = 1981, IC025 = 2.75), nervous system disorders (n = 568, IC025 = 1.68), general disorders and administration site conditions (n = 547, IC025 = 2.69), endocrine system diseases (n = 458, IC025 = 2.41), and hepatobiliary system diseases (n = 424, IC025 = 2.84). There were 195 PTs that simultaneously met the four algorithms, and 159 PTs were obtained after excluding those associated with the four excluded SOCs, and the screened PTs were ranked in descending order according to the signal intensity (IC025 value) and the number of reports, respectively (Tables 5 and 6). Among them, the top six PTs were poor weight loss (IC025 = 5.51), injection site extravasation (IC025 = 4.84), eructation (IC025 = 4.77), medullary thyroid carcinoma (IC025 = 4.45), diabetic retinopathy (IC025 = 3.88), and pancreatitis (IC025 = 3.85). In addition, we compared the detected AEs with the drug insert for semaglutide and found that the signals that were not mentioned in the insert included: starvation ketoacidosis (IC025 = 3.17), chronic cholecystitis (IC025 = 3.13), obstructive pancreatitis (IC025 = 3.72), allodynia (IC025 = 2.67), pancreatic enlargement (IC025 = 1.78), breath odor (IC025 = 1.55), pancreatitis chronic (IC025 = 1.33), and polycystic ovaries (IC025 = 1.3) (Figure 4).

Table 4.

Signal strength of AEs of semaglutide at the system organ class (SOC) level in FDA Adverse Event Reporting System (FAERS) source

SOC code	System organ class (SOC)	Case reports	ROR (95% CI)	PRR (95% CI)	χ2	IC (IC025)	EBGM (EBGM05)
10017947	Gastrointestinal disorders	10,214	15.97 (15.41–16.55)	5.43 (5.38–5.49)	41962.40461	2.43 (2.34)	5.38 (5.19)
10022891	Investigations	2,124	7.08 (6.76–7.41)	6.19 (5.95–6.44)	9325.687523	2.61 (2.46)	6.11 (5.83)
10027433	Metabolism and nutrition disorders	1981	8.69 (8.29–9.12)	7.64 (7.33–7.97)	11436.66175	2.91 (2.75)	7.52 (7.17)
10029205	Nervous system disorders	568	4.04 (3.71–4.39)	3.92 (3.61–4.25)	1234.818899	1.96 (1.68)	3.89 (3.58)
10018065	General disorders and administration site conditions	547	8.29 (7.6–9.03)	8.01 (7.37–8.7)	3308.055526	2.98 (2.69)	7.88 (7.23)
10014698	Endocrine disorders	458	6.9 (6.28–7.58)	6.72 (6.13–7.35)	2202.727151	2.73 (2.41)	6.62 (6.03)
10019805	Hepatobiliary disorders	424	9.41 (8.54–10.38)	9.17 (8.34–10.08)	3027.623248	3.17 (2.84)	8.99 (8.15)
10047065	Vascular disorders	321	3.13 (2.8–3.5)	3.08 (2.77–3.44)	451.9703378	1.62 (1.25)	3.07 (2.75)
10015919	Eye disorders	252	2.5 (2.2–2.83)	2.47 (2.19–2.79)	221.0982582	1.3 (0.88)	2.46 (2.17)
10040785	Skin and subcutaneous tissue disorders	165	4.22 (3.61–4.92)	4.18 (3.59–4.87)	396.4235006	2.05 (1.54)	4.15 (3.56)
10037175	Psychiatric disorders	141	6.58 (5.56–7.77)	6.52 (5.53–7.7)	649.8346127	2.69 (2.13)	6.44 (5.44)
10038738	Respiratory, thoracic and mediastinal disorders	109	5.11 (4.22–6.17)	5.07 (4.2–6.13)	352.7812019	2.33 (1.7)	5.02 (4.16)
10038359	Renal and urinary disorders	96	3.06 (2.5–3.74)	3.05 (2.49–3.72)	131.2667476	1.6 (0.94)	3.03 (2.48)
10028395	Musculoskeletal and connective tissue disorders	44	6.46 (4.79–8.7)	6.44 (4.79–8.67)	199.2909381	2.67 (1.7)	6.36 (4.72)
10007541	Cardiac disorders	42	2.49 (1.84–3.37)	2.48 (1.83–3.36)	37.05881175	1.31 (0.32)	2.48 (1.83)
10038604	Reproductive system and breast disorders	40	3.18 (2.33–4.34)	3.17 (2.32–4.33)	59.08769951	1.66 (0.65)	3.16 (2.31)
10029104	Neoplasms benign, malignant and unspecified (incl cysts and polyps)	25	7.48 (5.03–11.11)	7.47 (5.03–11.08)	137.511341	2.88 (1.62)	7.35 (4.95)
10021881	Infections and infestations	19	3.14 (2–4.93)	3.14 (2–4.93)	27.4699026	1.64 (0.22)	3.12 (1.99)
10013993	Ear and labyrinth disorders	10	5.87 (3.14–10.96)	5.87 (3.14–10.95)	39.79740277	2.54 (0.65)	5.8 (3.1)
10021428	Immune system disorders	10	4.59 (2.46–8.56)	4.59 (2.46–8.56)	27.76355254	2.19 (0.31)	4.55 (2.44)
10036585	Pregnancy, puerperium and perinatal conditions	4	8.7 (3.23–23.44)	8.7 (3.23–23.42)	26.70382278	3.09 (0.4)	8.54 (3.17)

CI, confidence interval; EBGM, empirical Bayesian geometric mean; IC, information component; PRR, proportional reporting ratio; ROR, reporting odds ratio; χ², chi‐squared.

Figure 3.

The bar plot shows a statistical graph of the 21 SOCs associated with case reports of semaglutide adverse events. Percentage values labeled in the figure represent the proportion of cases in which semaglutide had an adverse event in each SOC. The different colors represent the signal strength (IC025), the redder the color, the higher the signal strength.

Table 5.

The top 20 signal strength of AEs of semaglutide ranked by IC025 at the preferred terms level in FDA Adverse Event Reporting System

System organ class	Preferred term	Case reports	ROR (95% CI)	PRR (95% CI)	χ2	IC (IC025)	EBGM (EBGM05)
Metabolism and nutrition disorders	Weight loss poor	138	82.96 (69.14–99.53)	83.74 (69.69–100.62)	9299.046561	6.11 (5.51)	69.2 (57.59)
General disorders and administration site conditions	Injection site extravasation	339	40.84 (36.57–45.61)	41.79 (37.33–46.78)	11992.31176	5.22 (4.84)	37.24 (33.27)
Gastrointestinal disorders	Eructation	315	38.9 (34.69–43.61)	39.74 (35.36–44.66)	10631.41235	5.15 (4.77)	35.62 (31.7)
Endocrine disorders	Medullary thyroid cancer	15	88.11 (50.46–153.85)	88.2 (50.49–154.08)	1063.853192	6.18 (4.45)	72.74 (41.64)
Endocrine disorders	Diabetic retinopathy	50	30.28 (22.73–40.33)	30.38 (22.78–40.51)	1318.62183	4.82 (3.88)	28.27 (21.2)
Gastrointestinal disorders	Pancreatitis	422	18.91 (17.17–20.82)	19.44 (17.61–21.46)	6852.902595	4.18 (3.85)	18.12 (16.41)
Psychiatric disorders	Food aversion	26	35.05 (23.49–52.3)	35.11 (23.52–52.42)	792.5525416	5.02 (3.74)	32.38 (21.69)
Hepatobiliary disorders	Obstructive pancreatitis	18	41.58 (25.61–67.5)	41.63 (25.63–67.62)	647.4175251	5.24 (3.72)	37.85 (23.3)
Metabolism and nutrition disorders	Hunger	113	21.24 (17.59–25.64)	21.39 (17.7–25.86)	2072.698904	4.34 (3.71)	20.24 (16.74)
Metabolism and nutrition disorders	Lack of satiety	7	68.53 (30.79–152.51)	68.56 (30.8–152.64)	399.2655767	5.88 (3.51)	58.88 (26.45)
Hepatobiliary disorders	Bile duct stone	35	24.81 (17.65–34.89)	24.87 (17.67–34.99)	754.3982396	4.55 (3.45)	23.46 (16.67)
Hepatobiliary disorders	Cholecystitis acute	49	20.52 (15.41–27.32)	20.58 (15.44–27.43)	866.6283964	4.29 (3.36)	19.59 (14.7)
Musculoskeletal and connective tissue disorders	Neck mass	33	22.69 (15.99–32.2)	22.74 (16.01–32.3)	648.5169362	4.43 (3.3)	21.56 (15.18)
Vascular disorders	Retinopathy	47	19.56 (14.6–26.2)	19.62 (14.64–26.3)	790.2691723	4.23 (3.27)	18.72 (13.96)
Metabolism and nutrition disorders	Food craving	38	21.03 (15.19–29.12)	21.08 (15.21–29.22)	689.7327673	4.33 (3.27)	20.05 (14.47)
Metabolism and nutrition disorders	Starvation ketoacidosis	5	68.53 (26.59–176.6)	68.55 (26.59–176.71)	285.1890301	5.88 (3.17)	58.88 (22.84)
Hepatobiliary disorders	Cholecystitis chronic	16	28 (16.88–46.44)	28.02 (16.89–46.51)	389.9567891	4.72 (3.13)	26.27 (15.83)
General disorders and administration site conditions	Injection site discharge	32	19.79 (13.88–28.2)	19.83 (13.9–28.28)	544.6236666	4.24 (3.1)	18.92 (13.27)
Gastrointestinal disorders	Vomiting projectile	27	21.19 (14.4–31.18)	21.22 (14.41–31.26)	493.9078903	4.34 (3.1)	20.2 (13.71)
Metabolism and nutrition disorders	Increased appetite	114	13.32 (11.06–16.04)	13.42 (11.13–16.18)	1259.471531	3.69 (3.07)	12.94 (10.73)

Table 6.

The top 25 of AEs of semaglutide that meet the four algorithms ranked by case reports at the preferred terms level in FDA Adverse Event Reporting System

System organ class	Preferred term	Case reports	ROR (95% CI)	PRR (95% CI)	χ2	IC (IC025)	EBGM (EBGM05)
Gastrointestinal disorders	Nausea	2,675	6.24 (6.02–6.46)	7.42 (7.11–7.74)	11943.2113	2.62 (2.48)	6.16 (5.9)
Gastrointestinal disorders	Vomiting	1,621	6.73 (6.43–7.05)	7.45 (7.07–7.85)	7915.308768	2.73 (2.56)	6.64 (6.3)
Gastrointestinal disorders	Diarrhea	1,256	3.32 (3.15–3.5)	3.54 (3.34–3.75)	2074.835605	1.72 (1.53)	3.3 (3.12)
Metabolism and nutrition disorders	Decreased appetite	978	7.77 (7.31–8.26)	8.26 (7.74–8.82)	5714.259659	2.93 (2.71)	7.65 (7.16)
Investigations	Weight decreased	835	5.59 (5.23–5.97)	5.87 (5.47–6.29)	3134.063379	2.47 (2.23)	5.52 (5.15)
Gastrointestinal disorders	Constipation	697	5.97 (5.55–6.43)	6.23 (5.77–6.72)	2868.938803	2.56 (2.3)	5.9 (5.47)
Investigations	Blood glucose increased	635	5.97 (5.53–6.45)	6.2 (5.72–6.72)	2609.781903	2.56 (2.29)	5.9 (5.45)
Gastrointestinal disorders	Abdominal pain upper	530	4.74 (4.36–5.15)	4.88 (4.47–5.32)	1557.701408	2.23 (1.94)	4.7 (4.3)
Gastrointestinal disorders	Pancreatitis	422	18.91 (17.17–20.82)	19.44 (17.61–21.46)	6852.902595	4.18 (3.85)	18.12 (16.41)
Gastrointestinal disorders	Abdominal pain	391	3.41 (3.09–3.76)	3.48 (3.14–3.85)	665.5157142	1.76 (1.42)	3.39 (3.06)
Gastrointestinal disorders	Abdominal distension	382	7.42 (6.72–8.2)	7.6 (6.86–8.42)	2093.542229	2.87 (2.53)	7.31 (6.6)
Gastrointestinal disorders	Abdominal discomfort	347	3.23 (2.91–3.58)	3.28 (2.95–3.65)	532.9070724	1.68 (1.32)	3.21 (2.88)
General disorders and administration site conditions	Injection site extravasation	339	40.84 (36.57–45.61)	41.79 (37.33–46.78)	11992.31176	5.22 (4.84)	37.24 (33.27)
Metabolism and nutrition disorders	Dehydration	317	5.4 (4.84–6.03)	5.5 (4.92–6.15)	1127.348408	2.42 (2.04)	5.35 (4.78)
Gastrointestinal disorders	Eructation	315	38.9 (34.69–43.61)	39.74 (35.36–44.66)	10631.41235	5.15 (4.77)	35.62 (31.7)
Gastrointestinal disorders	Dyspepsia	269	5.39 (4.79–6.08)	5.47 (4.85–6.18)	953.3247476	2.42 (2.01)	5.34 (4.73)
Investigations	Blood glucose decreased	265	11.36 (10.06–12.82)	11.55 (10.21–13.06)	2439.743515	3.47 (3.06)	11.08 (9.8)
Nervous system disorders	Vision blurred	258	3.55 (3.14–4)	3.59 (3.17–4.07)	469.916055	1.82 (1.4)	3.52 (3.11)
Gastrointestinal disorders	Flatulence	255	9.36 (8.28–10.59)	9.51 (8.39–10.78)	1866.021778	3.2 (2.78)	9.18 (8.1)
Gastrointestinal disorders	Gastrointestinal disorder	187	3.78 (3.28–4.36)	3.82 (3.3–4.41)	380.1267489	1.91 (1.43)	3.75 (3.25)

Figure 4.

The radial plot shows 36 common and new preferred terms in the top 6 signal strength of SOCs of semaglutide ranked by IC025. Different colors represent different SOCs. The number marked in the graph represents the signal strength (IC025), and the higher the number, the higher the signal strength.

The PT distribution for the two different indications

Signal counts and comparisons were performed for the two primary indications (type 2 diabetes mellitus and obesity) (Table 7). Among the AEs with type 2 diabetes as the primary indication, the top three reported numbers were nausea (n = 955), vomiting (n = 651), and diarrhea (n = 516). Whereas, among the AEs with obesity as the primary indication, the top three reports were overdosing (n = 353), nausea (n = 267), and product use for unapproved indications (n = 195). The AEs related to the pancreas and gallbladder were tapped in both indications, including pancreatitis, pancreatic cancer, acute pancreatitis, obstructive pancreatitis, metastatic pancreatitis, pancreatic adenocarcinoma, acute cholecystitis, chronic cholecystitis, infectious cholecystitis, and gallstone disease. In addition, we found a new adverse reaction (intestinal obstruction) and a specific adverse reaction signal (suicidal thoughts) that had not yet been mentioned in the specification to co‐occur in both indications.

Table 7.

Distribution of PT in the top 10 ranked patients for the two main indications

Type 2 diabetes mellitus (n = 4,228)	Case number	Percentage	Obesity (n = 1,325)	Case number	Percentage
Nausea	955	7.4	Off label use	353	8.0
Vomiting	651	5.1	Nausea	267	6.1
Diarrhea	516	4.0	Product use in unapproved indication	195	4.4
Decreased appetite	405	3.2	Vomiting	156	3.5
Weight decreased	298	2.3	Diarrhea	100	2.3
Constipation	242	1.9	Decreased appetite	75	1.7
Blood glucose increased	240	1.9	Headache	74	1.7
Headache	224	1.7	Injection site extravasation	70	1.6
Off label use	216	1.7	Fatigue	69	1.6
Abdominal pain upper	211	1.6	Device malfunction	68	1.5

Multiple PTs can appear in a single case report, so the sum of the number of reports for each PT can be greater than n.

DISCUSSION

In this study, among the 14,512 adverse events after the marketing of semaglutide, we focused on gender differences, important adverse signals related to the gastrointestinal, hepatobiliary, and endocrine systems, and seven new serious adverse signals distributed in different systems, including metastatic pancreatic cancer, pancreatic adenocarcinoma, small bowel obstruction, functional bowel obstruction, paralytic bowel obstruction, paralytic bowel obstruction, and metastatic pancreatic cancer, pancreatic adenocarcinoma, small bowel obstruction, functional bowel obstruction, paralytic bowel obstruction, acute cholecystitis, chronic cholecystitis, and polycystic ovaries. A comparative analysis of AEs in two different indications (type 2 diabetes mellitus and obesity) was also performed to provide pharmacovigilance to improve drug safety.

Sex difference

The statistical results of our study showed that there was a more significant gender difference in the incidence of adverse reactions to semaglutide, with a greater number of adverse reactions reported in females (n = 8,667, pct = 59.72%) than in males (n = 5,220, pct = 35.97%). Since the approval of semaglutide for the treatment of obesity in 2021, the number of AEs in women has been approximately 1–2 times higher, whereas previously there was no gender difference in the number of reports. Studies have shown that women appear to carry a greater burden of risk factors when diagnosed with type 2 diabetes, particularly obesity. ¹⁵ However, data from the World Health Organization 2021 suggest that the global prevalence of type 2 diabetes is instead male‐dominated. ¹⁵ Based on this, the reversal in the sex ratio may be partly due to greater drug exposure in female patients following the approval of weight‐loss drugs; thus, further research on the extent to which medication side effects in obese patients due to sex steroid hormones is needed to improve the safety of drug use in women. ¹⁶

Adverse reactions related to the gastrointestinal system

SOC‐based data mining showed that most reported AEs cases with semaglutide focused on gastrointestinal disorders (n = 10,214, IC025 = 2.34), accounting for 70.38% of total AEs. Nausea (n = 2,675), vomiting (n = 1,621), and diarrhea (n = 1,256) had the highest reported frequencies, as described in the package insert as very common adverse events. In addition, the incidence of acute pancreatitis was described in the specification as 0.3% and 0.2% in the semaglutide group versus the control group, described as occasional, which is consistent with the results of this study showing a low number of reports of acute pancreatitis (n = 64, IC025 = 1.67). This confirms the accuracy and reliability of the four data algorithms used in this study.

Surprisingly, many reports of pancreatitis (n = 422, IC025 = 3.85) were found, ranking 6th among all semaglutide PT signals and showing a very strong signal association. In addition, new adverse pancreas signals were identified, including pancreatic carcinoma metastatic and adenocarcinoma pancreas.

In the first few years after GLP‐1RA was introduced, numerous reports of pancreatitis associated with pancreatic cancer were recorded. Related pharmacovigilance found a strong association between the use of GLP‐1RAs and the incidence of precancerous pancreatic lesions ¹⁷ . Pancreatic cancer is now the third leading cause of cancer‐related deaths in the United States. Most patients are in the advanced stages of the disease and are difficult to cure, and the onset of pancreatitis may be one of the causes of the pancreatic cancer process. ¹⁸ Some epidemiologic data also suggest an increased risk of acute pancreatitis and pancreatic cancer following exposure to GLP‐1RA ¹⁹ . GLP‐1 RAs increase the risk of pancreatitis in patients by prolonging gastric emptying, increasing the contractility of the sphincter of oddi, impeding the excretion of bile acids and pancreatic enzymes, thereby increasing the risk of pancreatitis. ²⁰ Pancreatitis stimulates an increase in inflammatory cytokines, reactive oxygen species, and affects cell proliferation, which is a risk factor for pancreatic cancer. Several studies have shown that GLP‐1 regulates the cAMP, PKA, and PI3K/AKT/FOXO1 signaling pathways, causing pancreatic cell hyperplasia and increasing the risk of pancreatic intraepithelial neoplasia. ²¹ , ²² , ²³ Interestingly, GLP‐1 RAs can activate cAMP in a GLP‐1R‐dependent manner, inhibit the AKT and ERK1/2 pathways, induce apoptosis, and inhibit proliferation of human pancreatic cancer cell lines in vitro. ²⁴ Therefore, the results of studies on the association between GLP‐1 RAs and pancreatic cancer are still inconsistent.

Given that the FDA has included acute pancreatitis as a boxed warning for the use of semaglutide, ²⁵ our study similarly collected a large number of real‐world clinical event reports of pancreatitis and pancreatic cancer caused by semaglutide, supporting research on its harmful effects on the body. Accordingly, semaglutide should be used with caution in patients with a history of pancreatitis and in patients with triglycerides ≥6 mmol/L for safety reasons. However, given the nature of observational studies, such data may be biased.

Clinical trials with GLP‐1 RAs have shown mild, asymptomatic increases in lipase and amylase levels because GLP‐1 RAs stimulate pancreatic β‐cells and exocrine duct cells, resulting in pancreatic injury due to follicular cell hypertrophy and induction of pro‐inflammatory cytokines. ²⁶ GLP‐1 signaling is now known to be mediated by PI3K, cAMP, and Ca(2+)‐CamKIIalpha. ²⁷ In light of this, it is imperative to further investigate whether GLP‐1 RAs affect pancreatic cells through PI3K/AKT, ERK1/2, or other signaling pathways leading to the progression of pancreatitis with its associated carcinomas in the future.

Importantly, this study uncovered new adverse reaction signals related to intestinal obstruction, including small intestinal obstruction (n = 19), ileus (n = 17), and ileus paralytic (n = 8), which were not mentioned in the specification. Relevant studies have shown that in severe cases, the intestinal blood supply may be compromised leading to the death of intestinal tissue, which can threaten the patient's life ²⁸ . Sodhi et al. ²⁹ found an increased risk of pancreatitis and intestinal obstruction with GLP‐1RA for weight loss compared with bupropion‐naltrexone. In disproportionate analyses ³⁰ , enteric insulin analogs reported intestinal obstruction more than 4.5 times more frequently than other diabetes medications. In 2023, the FDA disclosed that of the 8,571 reports of gastrointestinal disorders associated with the drug semaglutide, 33 were related to intestinal obstruction reactions, two of which resulted in the patient's death, and affirmed the likelihood and risk of intestinal obstruction appearing, and updated Ozempic's labeling in response ²⁹ . Despite the current low number of reported adverse reactions associated with intestinal obstruction, the concern for unexpected and serious life‐threatening AEs raises concerns in this study about the potential risk of gastrointestinal adverse events due to semaglutide. Considering this, it is of paramount importance for clinicians and patients to consider these possible serious adverse events and to consider changes in treatment regimens to safeguard patient benefit in the event of relevant symptoms.

Adverse reactions related to the hepatobiliary system

Among the most reported hepatobiliary disorders associated with semaglutide, with the highest signal intensity were cholelithiasis (n = 127, IC025 = 2.81) and bile duct stone (n = 35, IC025 = 3.45). These accounted for 38.20% of the reports of adverse reactions in the hepatobiliary system, and they are also described in the specification as common. Unexpectedly, we found other new adverse reaction signals related to the gallbladder that were not mentioned in the specification, including cholecystitis acute (IC025 = 3.36) and cholecystitis chronic (IC025 = 3.13), both of which exhibited greater signal intensity than cholelithiasis (IC025 = 2.81). GLP‐1 inhibits gastrointestinal motility and gallbladder dynamics by suppressing cholecystokinin secretion, which delays gallbladder emptying and may increase the risk of ductal obstruction. Furthermore, cholecystokinin‐expressing neurons in the brainstem are necessary for the anorexic and weight‐reducing effects of GLP‐1RA. ³¹ , ³²

A systematic evaluation and meta‐analysis based on clinical trials ³³ also indicated that the use of GLP‐1RA was associated with an increased risk of gallbladder or biliary tract disease, especially in the case of longer‐term use of larger doses and weight loss. Prolonged use of GLP‐1RA may lead to the development of gallstones, which can increase the risk of cholecystitis due to impaired gallbladder contraction. ³⁴ Therefore, the correlation between cholecystitis and gallstones is well‐established. The interplay between these two conditions creates a vicious cycle of inflammation in the gallbladder and obstruction in the bile duct. ³⁵ , ³⁶ Early studies have shown that the emergency surgical mortality rate for acute complications of bile duct stone in patients with diabetes is 15–20%, which is four to five times higher than nondiabetic patients. ³⁷ Based on this, the study concluded that medication outcomes related to the gallbladder may pose a threat to patients' lives. Patients who have undergone cholecystectomy should avoid using semaglutide. It is suggested that the FDA may supplement the specification with descriptions of AEs for complications related to cholelithiasis in order to increase physician and patient medication vigilance.

Adverse reactions related to the endocrine system

Among the endocrine disorders, we identified a very strong signal association with the FDA black box warning for the AE medullary thyroid cancer (IC025 = 4.45) which ranks fourth in the PT, despite the recent disclosure by the European Union that there is no causal relationship between semaglutide and medullary thyroid cancer. In animal studies, GLP‐1RAs stimulate thyroid parafollicular C‐cell proliferation and elevated calcitonin levels in rodents, which in turn increases the risk of thyroid cancer. ³⁸ Protein blot analysis revealed a significant negative correlation between the expression level of GLP‐1R and cluster of differentiation 26 (CD26) in medullary thyroid carcinoma. ³⁹ Meanwhile, CD26 in mice can promote blood glucose regulation by regulating GLP‐1. ⁴⁰ It is currently unclear how precisely human GLP‐1 regulates medullary carcinoma through CD26 regulation. Further fundamental research is required to confirm this in the future.

In addition, we found an AE – polycystic ovary which ranked 10th among hepatobiliary system PTs. Polycystic ovary syndrome (PCOS) is a highly prevalent condition that represents the most common endocrine metabolic disorder among women of reproductive age ⁴¹ . Notably, semaglutide is considered to play a potentially important role in the treatment of PCOS, including the alleviation of hyperinsulinemia and hyperandrogenemia ⁴² . A controlled study showed that semaglutide significantly delayed gastric emptying for 4 h in women with PCOS and obesity ⁴³ . Thus, the risks and benefits of medicating patients for aesthetic reasons to treat postpartum obesity have not been weighed, and ongoing attention to the safety of polycystic ovaries is still needed.

Signal analysis of adverse reactions for two different indications

Unexpectedly, in addition to type 2 diabetes mellitus, extremely severe gastrointestinal AEs associated with pancreatic and intestinal obstruction have been observed in patients with obesity. At the same time, the PT of “over‐the‐counter use” of semaglutide is surprisingly found in two of the main indications, especially in obesity, which has also been brought to our attention. Although it is not clear whether the emerging AEs are causally related to patients’ over‐the‐counter use of the drug, it is certain that clinicians should strengthen their guidance to patients on the use of the drug.

In addition, reports of suicidal ideation were seen in the case statistics for both indications. The FDA had required pharmaceutical companies to report these specific risks regarding suicidal thoughts and ideation. The European Medicines Agency is currently reviewing the GLP‐1RA and investigating this new adverse event. This particular adverse reaction has attracted the attention of regulators ⁴⁴ . While Barbara et al. ⁴⁵ also studied resting state function of brain regions to observe changes in suicidal ideation/behavior in patients with semaglutide, no significant differences were seen between the placebo group and the semaglutide group, however, the results were limited to a relatively small sample size. Therefore, suicidal ideation still needs to be taken seriously.

Given that the treatment of obesity may provide additional benefits for obesity complications as well, including non‐alcoholic steatohepatitis (NASH) and non‐alcoholic fatty liver disease (NAFLD), the future use of semaglutide will become more and more widespread in all types of indications, and some efficacy studies are under way, including in the hepatobiliary system, the nervous system, and even cancer ⁴⁶ , ⁴⁷ . A 72 week, double‐blind, phase II trial demonstrated that treatment with semaglutide resulted in a significantly higher proportion of patients with NASH regression compared with placebo ⁴⁸ . A phase II, open‐label, proof‐of‐concept trial demonstrated that semaglutide was efficacious and well tolerated in the treatment of NAFLD ⁴⁹ . It is well known that obesity‐related adipokines are risk factors for Alzheimer's disease (AD) and cancer. ⁵⁰ , ⁵¹ In AD models, GLP‐1RAs are efficacious in reducing neuroinflammation and oxidative stress, neurotrophic effects, Aβ deposition, and tau hyperphosphorylation, and are currently in clinical trials ⁵² ; Recent studies have found that semaglutide could restore the body's natural killer cells, thereby reducing the risk of cancer ⁵³ . In addition, studies from the PIONEER, SUSTAIN and STEP 3 clinical trials showed that the incidence of AEs with semaglutide in participants ≥55 years of age was similar to that observed in the overall population, supporting the studies evaluating semaglutide in an elderly population with early‐stage AD. ⁵⁴

However, several studies have now shown a significant correlation between a fatty liver and the severity of pancreatitis acute ⁵⁵ , ⁵⁶ ; malignant bowel obstruction is a common complication in cancer patients ⁵⁷ . Therefore, whether serious adverse effects such as pancreatic cancer and intestinal obstruction may also occur in the treatment of the relevant indications of semaglutide requires attention and vigilance in subsequent studies.

The results of disproportionality analysis should always be interpreted considering its limitations. First, since the FAERS database is a spontaneous reporting system, incomplete patient information may bias the comprehensiveness of the statistics. For instance, the consequences of serious adverse reactions were unknown in 81.41% of the reported patients in this study, and the age of 51.08% of the reported patients was missing. Second, semaglutide has been on the global market for a relatively short period, and there is still a shortage of relevant AEs data. Third, many AEs in the gastrointestinal system of semaglutide were reported, which may have weakened the degree of association between the new AEs and the drug, thus affecting the relevant study inference. Fourth, due to the widespread use of drugs in the present day, AE reporting of drugs may be influenced by the extent of drug use, publicity, and personal bias. ⁵⁸ Fifth, as with other pharmacovigilance, any hypotheses generated by this study still need to be validated by prospective studies.

DISCLOSURE

The authors have no relevant financial or non‐financial interests to disclose.

Approval of the research protocol: N/A.

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.