Background

Gadolinium-based contrast agents (GBCAs) are chelates of gadolinium (Gd), a paramagnetic metal that induces a magnetic field to enhance tissue contrast to characterize lesions and assess perfusion and flow-related abnormalities [1, 2]. From 1988, when the first GBCA was introduced to the market, to date, there are nine GBCAs widely used in clinical practice. GBCAs can be classified into linear and macrocyclic chelates according to their chemical structure and ionic and nonionic according to the Gd ionic state [3].

As the use of GBCAs increases, understanding more specific symptomatic differences in the occurrence of HSRs across GBCAs can provide guidance for clinical use. According to an 8-year cohort study [4], the incidence of HSR was 0.4%, which is lower than iodinated contrast media, but concerns remain. However, it is mainly clinical studies [5,6,7] and case reports [8,9,10] that focus on HSRs to GBCAs. There are fewer studies of more specific responses to HSRs due to some limitations; only one pharmacovigilance study analyzed data on anaphylaxis from 1988 to 2012 [11]. In addition, it has been shown that in patients who have been treated with GBCAs, Gd may be retained sites, mainly in the bones and skin, with only a small amount present in the brain [12, 13]. The term “gadolinium deposition disease” (GDD) was first proposed by Semelka and colleagues in 2016 to describe symptoms reported by patients with normal renal function after exposure to GBCAs [14]. Although there is still much controversy regarding GDD, the Uppsala Monitoring Centre (UMC) and the FDA have been using GDD as a preferred term (PT) for the collection of reports since the term GDD emerged. However, members of the American College of Radiology’s Committee on Drugs and Contrast Agents in 2022 proposed a new term to describe symptoms reported after intravascular exposure to GBCAs—symptoms associated with gadolinium exposure (SAGE) [15]. This term was used in place of other nomenclatures that assumed causation but had not been scientifically proven, whereas according to the current study, SAGE is a combination that contains multiple PTs. Therefore, combining GDD and SAGE data may provide a more complete picture of the symptoms involved. Currently, similar studies are not comprehensive enough or out of date [16,17,18], so analyzing GBCA-associated adverse events (AEs) using the latest post-marketing data may provide the most up-to-date data to inform the safe clinical use of GBCAs.

This study aimed to compare the risk of HSRs, and GDD/SAGE of all GBCAs using disproportionality analysis, namely gadopentetate dimeglumine, gadobenate dimeglumine, gadoxetate disodium, gadoterate meglumine, gadodiamide, gadoversetamide, gadoteridol, and gadobutrol (gadofosveset was excluded because of too little data) in clinical practice by using two of the world’s largest databases of AEs, the World Health Organization database VigiBase or the FDA Adverse Event Reporting System (FAERS). However, it is important to emphasize that disproportionality analysis is a hypothesis-generating activity and represents potential statistical associations alone without causality [19]; explicit causality requires further research and verification. We additionally approximated the frequency of HSRs from 2008 to 2022 based on the VigiBase and the IQVIA Multinational Integrated Data Analysis System (IQVIA-MIDAS) to compensate for the two safety databases’ incapacity to do so without having the whole number of uses.

Methods

Data sources

This study extracted data from VigiBase (from inception to July 3, 2023) and FAERS (from inception to March 31, 2023) for GBCAs. VigiBase is the unique WHO global database of reported potential side effects of medicinal products, developed and maintained by UMC. With more than 30 million reports of suspected AEs of medications filed by the WHO Programme for International Drug Monitoring member nations since 1968, it is the largest database of its kind in the world. The FAERS is a database that contains AE reports, medication error reports and product quality complaints resulting in AEs that were submitted to the FDA. The FDA’s post-marketing safety surveillance program for drugs and therapeutic biologic products is intended to be supported by the database. We mined and cleaned pharmacovigilance data from FAERS using the open tool OpenVigil 2.1 [20]. In addition, we extracted the total sales volume of each GBCA from the IQVIA-MIDAS [21] from 2008 to 2022, with sales volume in standardized units. IQVIA-MIDAS is internally validated through other sales data sources and shows a high level of data accuracy [22].

Study design

This study included all HSR-related AEs categorized by group queries according to the Medical Dictionary for Drug Regulatory Activities (MedDRA) 26.0. The narrow version of the standardized MedDRA query (SMQ) was used to define HSRs because it provides better predictability while retaining comparable sensitivity to the broad version [23]. Based on allergic symptoms, we defined HSRs in the following three SMQ levels: angioedema, severe cutaneous adverse reactions (SCARs), and anaphylactic/anaphylactoid shock conditions (ASCs). PTs in hypersensitivity or anaphylactic reactions but not included in the three SMQ levels were regarded as anaphylactic reactions, others. In addition, GDD was retrieved at the PT level in the FAERS and VigiBase. Based on a study [24], we categorized the symptoms included in SAGE into 29 PTs under 6 system organ classes and further analyzed SAGE using the FAERS database.

Four patterns of drugs were identified: primary suspect drug, secondary suspect drug, interacting drug, and concomitant drug. According to the procedure used by the UMC, concomitant-related records were eliminated to improve signal strength [25]. Since GDD and SAGE are rare and still controversial, we only estimate the frequency of HSRs of GBCAs from 2008 to 2022 based on VigiBase and IQVIA-MIDAS. The United Nations list of geographical divisions was used for the division of regions or countries. Active ingredient names were used when searching in the VigiBase, FAERS, and IQVIA-MIDAS databases.

Statistical analysis

Reporting odds ratio (ROR) was employed to show whether risk signals were present. The signal of disproportionate reporting was defined as ROR025 greater than one and at least three cases. We used Microsoft Office Excel 2016 (Los Angeles, CA, USA) to calculate demographic information and disproportionality analysis results. Calculations of ROR and 95%CI were based on 2 × 2 contingency tables. The formulas are:

Drug category

Event of interest

All other events

Target drugs

a

b

All other drugs in the database

c

d

$$\text{ROR}\hspace{0.17em}=\hspace{0.17em}\text{ad}/\text{bc}; 95\text{\%CI}={e}^{\text{ln}\left(\text{ROR}\right)\pm 1.96\sqrt{\frac{1}{a}+\frac{1}{b}+\frac{1}{c}+\frac{1}{d}}}$$

The frequency of HSRs was obtained by dividing the number of reports in VigiBase by the total sales volume (measured in millions) over 2008–2022 in IQVIA-MIDAS.

Results

Descriptive analysis of HSRs in VigiBase

In VigiBase, regarding HSRs, the number of cases of different GBCAs in descending order was gadobutrol (n = 9454), gadoterate meglumine (n = 6489), gadopentetate dimeglumine (n = 5160), gadobenate dimeglumine (n = 3653), gadoteridol (n = 2997), gadoxetate disodium (n = 1250), gadodiamide (n = 1064), and gadoversetamide (n = 323). Except for gadoxetate disodium, all GBCAs were reported more by females than males. The reports were mainly concentrated among patients aged 18–64 years. The primary reporting regions for the different GBCAs varied, with gadopentetate dimeglumine, gadobenate dimeglumine, gadodiamide, and gadoversetamide reported predominantly from the Americas; gadoxetate disodium and gadobutrol reported mainly from Asia; and gadoterate meglumine and gadoteridol reported primarily from Europe. The annual reporting rates for individual GBCAs were unstable from 2008 to 2022. The lowest reported rate of serious events (including death, life-threatening, disability, and hospitalization) was gadoxetate disodium (4.0%), while the highest was gadobenate dimeglumine (22.91%) (Table 1).

Table 1 Characteristics of reports with suspected HSRs sourced from VigiBase
Full size table

Disproportionality analysis results of suspected AEs in FAERS and VigiBase

As shown in Tables 2 and 3, all GBCAs showed significant signals of HSRs in both FAERS and VigiBase. For nonionic GBCAs, the results of both databases indicated that the two linear GBCAs had the lowest probability of HSRs among all GBCAs, with gadodiamide (ROR025 = 2.24 and 1.44 (in VigiBase and FAERS, respectively)) having a lower ROR025 than gadoversetamide (ROR025 = 3.56 and 1.76). For ionic GBCAs, the FAERS results showed that all linear GBCAs had lower ROR025 than the macrocyclic type, whereas the VigiBase results showed that only the other two linear GBCAs, except for gadobenate dimeglumine, had lower ROR025 than the macrocyclic GBCAs.

Table 2 Disproportionality analysis results of all suspected HSRs associated with GBCAs in VigiBase
Full size table
Table 3 Disproportionality analysis results of all suspected HSRs associated with GBCAs in FAERS
Full size table

Figure 1a and 1b display the results of disproportionality analysis for three SMQ levels in VigiBase and FAERS. For ASCs, in both databases, it was shown that the ROR025 was greater for gadobenate dimeglumine (ROR025 = 10.45 and 18.35 (in VigiBase and FAERS, respectively)) and gadoteridol (ROR025 = 6.92 and 20.17), with no significant signal for gadoversetamide. For SCARs, both databases indicated that gadoversetamide (ROR025 = 3.80 and 3.06) had the most significant signal; gadopentetate dimeglumine, gadobenate dimeglumine, gadoxetate disodium, and gadobutrol had no significant signals. For angioedema, gadobutrol (ROR025 = 12.45 and 20.64) had the largest ROR025, gadodiamide had the smallest ROR025 in VigiBase (ROR025 = 2.23), and no significant signal in FAERS, while gadoversetamide had no significant signal in VigiBase and the smallest ROR025 (ROR025 = 1.36) in FAERS. Overall, the risk of angioedema with nonionic linear GBCAs was minimal.

Fig. 1
Fig. 1
Full size image

a Disproportionality analysis results of hypersensitivity-associated SMQ levels in VigiBase. b Disproportionality analysis results of hypersensitivity-associated SMQ levels in FAERS. SMQ, standardized MedDRA query

We further analyzed the PTs under SMQ levels comparatively, and the differences are shown in Fig. 2a and 2b, with specific values shown in the additional file (Additional file 1: Table S1-S2). Based on the results of the two databases, the following results can be summarized: gadopentetate dimeglumine had the highest likelihood of eye swelling (ROR025 = 6.06 and 11.16 (in VigiBase and FAERS, respectively)), swollen tongue (ROR025 = 3.93 and 6.35), and tracheal oedema (ROR025 = 18.94 and 27.70); gadobenate dimeglumine had the strongest signal regarding anaphylactic reaction (ROR025 = 9.56 and 16.53) and anaphylactoid reaction (ROR025 = 10.63 and 163.81); for gadoterate meglumine, swelling of eyelid (ROR025 = 6.92 and 7.26) and type I hypersensitivity (ROR025 = 77.72 and 24.82) were the highest signal; regarding gadoversetamide, the ROR025 of exfoliative rash (ROR025 = 110.98 and 125.14) and rash macular (ROR025 = 9.3 and 10.95) were the most significant; for gadoteridol, anaphylactoid shock (ROR025 = 56.22 and 58.05), circulatory collapse (ROR025 = 5.99 and 3.86), and shock (ROR025 = 7.40 and 18.55) were the strongest signal; gadobutrol had the highest signal about lip swelling (ROR025 = 3.59 and 5.94), mouth swelling (ROR025 = 1.84 and 1.52), urticaria (ROR025 = 18.91 and 41.76), and skin test positive (ROR025 = 2.79 and 23.07).

Fig. 2
Fig. 2
Full size image

a Disproportionality analysis results of hypersensitivity-associated PTs in VigiBase. b Disproportionality analysis results of hypersensitivity-associated PTs in FAERS. PT, preferred term

The disproportionality analysis results of GDD are shown in Table 4. According to the results of VigiBase, linear GBCAs had a higher risk than macrocyclic GBCAs. Gadobutrol and gadobenate dimeglumine had higher ROR025 based on FAERS results. For SAGE, no GBCAs showed signals in psychiatric disorders; all GBCAs prompted signals in ≥ 10 PTs except gadoxetate disodium, which did not prompt signals at any of the PTs; gadopentetate dimeglumine had a minimum number of signals at the PT level of only 10; gadodiamide and gadoversetamide prompted signals in 15 PTs, both prompted the highest number of signals, and gadoversetamide generated the strongest signal in 14 PTs. What is more, except for gadoversetamide, the risk of gadopentetate dimeglumine, gadobenate dimeglumine, gadodiamide, and gadoteridol regarding skin induration requires extra attention. Gadodiamide and gadoteridol have a higher risk of skin tightness than other GBCAs. The specific values of the signals are shown in Table 5.

Table 4 Disproportionality analysis results for suspected GDD cases of GBCAs in FAERS and VigiBase
Full size table
Table 5 ROR025 of suspected SAGE-related PTs
Full size table

Time to onset analysis of suspected HSRs

Excluding reports with unknown onset times, the probability of HSRs within 1 day for individual GBCAs ranged from 93.73 to 98.89%, with those greater than 1 day being predominantly associated with SCARs.

Estimation of the frequency of suspected HSRs

Based on data from the IQVIA-MIDAS database from 2008 to 2022, the usage frequency of the different GBCAs varied by region, with gadobutrol (24.92%) having the highest rate of use in the Americas, gadopentetate dimeglumine (48.01%) in Asia, and gadoterate meglumine (48.98%) in Europe (Additional file 1: Fig. S1).

The total sales volumes for 2008–2022 are shown in the additional file (Additional file 1: Table S3), and the highest total frequency of the three SMQs was gadoxetate disodium (Fig. 3a). In addition, we estimated the frequency of each SMQ level for the Americas, Asia, and Europe, and the results are shown in Fig. 3b–d, with sales in the three regions for the years 2008–2022 also displayed (Additional file 1: Table S4). Concerning ASCs, in all three continents, the frequency of ASCs was highest for gadobenate dimeglumine and lowest for gadoversetamide. In the Americas, Asia, and Europe, the highest frequency of angioedema was gadopentetate dimeglumine, gadobutrol, and gadoterate meglumine, respectively; for SCARs, the highest frequency was gadoversetamide, gadoteridol, and gadobenate dimeglumine, respectively.

Fig. 3
Fig. 3
Full size image

a Total frequency of the three SMQ levels in three regions from 2008 to 2022. b Total frequency of the three SMQ levels in the Americas from 2008 to 2022. c Total frequency of the three SMQ levels in Asia from 2008 to 2022. d Total frequency of the three SMQ levels in Europe from 2008 to 2022. SMQ, standardized MedDRA query; ASCs, anaphylactic/anaphylactoid shock conditions; SCARs, severe cutaneous adverse reactions

Discussion

GBCAs have been used internationally for more than 35 years in hundreds of millions of patients. Although GBCAs are relatively safe, several studies have reported some AEs associated with GBCAs, including HSRs and GDD/SAGE. This study compared the safety of all GBCAs regarding HSRs and GDD/SAGE by analyzing the global post-marketing data, providing data to support a more rational clinical management of GBCA-associated AEs.

Based on demographic information, AE reports were concentrated in the 18–64 age group, with a predominance of those ≥ 45 age group, which may be because middle-aged and older adults are the main population using GBCAs. AE reports on females make up the majority of HSR reports. This may be related to the existence of genetic differences between the sexes. Regarding regions, the subject countries of reported AEs varied among GBCAs, which in combination with sales data is mainly, but not absolutely, related to the usage preference of each region. For example, Asia, which has the third highest sales of gadobutrol among the three regions, reported more AEs regarding HSRs than the Americas and Europe, which may be related to genetic and ethnic factors. In our study, the frequency was lower than in previous studies, which may be due to the fact that VigiBase and FAERS are spontaneous reporting systems (SRSs) that do not cover all AEs during the course of the study as in the clinical randomized trials.

Allergy is one of the triggers of angioedema [26]. In the general population, the incidence of severe allergic reactions is 0.01–0.03% [27]. However, 46% of allergic reactions are accompanied by angioedema [28]. Angioedema caused by GBCAs can occur in the face, eyes, lips, throat, etc., and has also been reported in the small intestine [29]. The results of this study showed that gadobutrol had a greater ROR025 for angioedema. However, when considering the amount of use, gadobutrol had the highest frequency of angioedema in Asia, but its percentage of use in Asia was only in the fourth place, which may be related to reporting bias or racial differences. The site of occurrence of angioedema differs among drugs; therefore, when using a particular GBCA, special attention needs to be paid to its high-risk site.

AEs associated with ASCs have been reported in the literature more frequently and sometimes fatally [30,31,32]. According to the results of FAERS and VigiBase, the ROR025 of gadobenate dimeglumine and gadoteridol is larger. Among the three regions, gadobenate dimeglumine had the highest frequency of ASCs, followed by gadoteridol. Shock is a more serious reaction, and patients who experience shock may not have a history of allergy. Therefore, preparation of resuscitation measures is necessary before administering GBCAs. In addition, skin testing may be beneficial in patients with a history of systemic reactions to Gd [33].

SCARs were mainly urticaria, exfoliative rash, etc., and GBCAs were classified as drugs that could cause acute generalized exanthematous pustulosis [8]. Preclinical research indicates that Gd compounds may activate dermal fibroblasts, encourage their proliferation, and aid in the development of myofibroblasts, hence inducing profibrotic reactions in the skin in vitro. Moreover, Gd affects the synthesis of collagen and the amount of collagen in the skin [34]. In FAERS, only gadoversetamide and gadodiamide hinted the signal for SCARs, while in VigiBase, in addition to gadoversetamide and gadodiamide, gadoteridol and gadoterate meglumine also hinted the signal. Although SCARs are rare, similar reactions may occur multiple times in sensitive patients and may be delayed, so they should be evaluated with caution if reused and observed for several days after use. The frequency of SCARs varies by region, which may be related to regional usage preferences or ethnicity.

GDD was proposed in 2016 [35]; there are several hypotheses about GDD but the mechanism of occurrence and pathophysiology is still unclear. However, some studies have suggested that GDD is more likely to occur with linear GBCAs [36], and the results of VigiBase reaffirm this conclusion. SAGE was proposed in 2018 to describe the symptoms associated with exposure to GBCAs and is used to help researchers report the associated symptoms in a more standardized way to allow for a more coherent interpretation of related studies. GDD can also be considered a subset under the symptoms associated with the gadolinium exposure rubric [35]. This study found differences in the risk of different GBCAs occurring in different symptoms by analyzing the data in FAERS, with gadoversetamide having the strongest signal in all 14 PTs, but in terms of sales volume, its use is now very low, and the risk of associated ADEs is considerably lower. In contrast, gadoxetate disodium did not suggest a suspicious signal. In addition, five GBCAs had a higher risk of skin induration; gadodiamide and gadoteridol also need more attention regarding skin tightness. The AEs of GBCAs involve multiple systems and can ultimately affect a patient’s quality of life, so it is important to prevent the associated risks. The National Institutes of Health held a meeting in 2018 to clarify the direction of future research on gadolinium retention [37]. It is hoped that the results of this study will provide data to support subsequent research.

This study, like many studies of pharmacovigilance databases, has certain limitations. Firstly, the FAERS and VigiBase are SRSs with diverse, uncertain data sources, and the number of reports for a given drug may be affected by the extent of the product’s use, publicity, and the nature of the AEs, as well as underreporting or overreporting. Therefore, even if the results of disproportionality analysis are significant, it is not possible to confirm a causal relationship between the drug and the AEs. Secondly, we recognize that the estimation of AE frequency is rough because not all patients use a single standard unit of GBCA. In addition, the extent of reporting in different areas can affect the estimation of frequency. Thirdly, the quality of relevant reports cannot be guaranteed as there is still a lot of controversy and the understanding of GDD/SAGE is still limited, so we did not estimate the frequency of GDD/SAGE, and the results of disproportionality analysis are only a preliminary probe based on the current data; more research needs to be continued. Finally, the site of use of GBCAs also affects the frequency of HSRs, and we were unable to analyze site-specific frequency further due to missing data. However, the inclusion of real-world data was as comprehensive and systematic as possible to ensure the reliability and authenticity of this study and provide data support for further research.

Conclusions

Different GBCAs exhibited diverse profiles regarding HSRs, which could be further affected by geographic area. Our study provides additional data support for the prevention of HSRs. As for the differences of GBCAs in GDD/SAGE, this study only analyzed the currently available post-marketing data as a preliminary exploration, and more unbiased studies are needed.