A real-world pharmacovigilance study of severe cutaneous adverse reactions associated with antiepileptic drug combination therapy: data mining of FDA adverse event reporting system

ABSTRACT Background The aim of this study was to evaluate the association between antiepileptic drug combination regimens and severe cutaneous adverse reactions (SCAR). Research design and methods We gathered cases indication with epilepsy based on the US Food and Drug Administration Adverse Event Reporting System (FAERS) database from 2004 to 2021. Disproportionality analyses were conducted by estimating the reporting odds ratio (ROR) and the information component (IC). Results Out of 128,262 reports were collected from the FAERS database, 104,278 cases were in the antiepileptic drugs group, and 23,984 cases were in the other primary suspected drugs group. A total of 20 combination regimens were associated with increased association of SCAR, top five of them were topiramate-phenytoin (ROR 57.62, 95% CI 30.93–107.34), lamotrigine-valproic acid (ROR 52.93, 95% CI 47.09–59.49), diazepam-phenobarbital (ROR 39.61, 95% CI 20.01–78.38), zonisamide-valproic acid (ROR 36.57, 95% CI 19.16–69.80), lamotrigine-diazepam (ROR 35.22, 95% CI 15.70–79.00). Conclusion The antiepileptic agent combinations may increase the incidence of SCAR and should be carefully evaluated in clinical practice. It is recommended to choose the combination regimens which have lower SCAR reporting rate for patients.


Introduction
Severe cutaneous adverse reactions (SCAR) are rare and lifethreatening hypersensitivity reactions [1][2][3]. Acute generalized exanthematous pustulosis (AGEP), drug reaction with eosinophilia and systemic symptoms (DRESS), Steven-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are included in SCAR [4], all are non-immunoglobulin E (IgE)-mediated, delayed type IV hypersensitivity reactions [5]. The global incidence of SCAR is 0.4 to 1.2 cases per million per year [6], but the mortality of all SCAR ranges from 4% for AGEP, 2% to 6% for DRESS, and up to 48% for TEN [4]. The SJS and TEN are thought to be variants of the same condition [7], and the distinction between them is determined by the affected body surface area, with TEN affecting more [8]. However, the incidence of SCAR from the same drug varies in different populations, and may be more likely to occur in people aged from 40 to 56 [9,10].
Epilepsy is one of the most common neurologic disorders worldwide, and a number of antiepileptic drugs (AEDs) are associated with SCAR, including carbamazepine, lamotrigine, valproic acid and phenytoin [1,4,[11][12][13]. Aromatic AEDs seem to be more likely to cause SCAR than non-aromatic AEDs [14]. Through published data, the adverse cutaneous reactions caused by aromatic AEDs affect 50.000 people a year in the United Kingdom. The U.S. Food and Drug Administration (FDA) had issued black box warnings regarding SJS/TEN reactions for lamotrigine and carbamazepine [15]. In addition to pharmacological and clinical factors, genetics also play a critical role in adaptive immunity and the human leukocyte antigen (HLA) class I alleles are the most strongly associated risk genes of SCAR to AEDs [16][17][18]. The HLA-B*1502 has been found to increase susceptibility to lamotrigine and carbamazepine induced SJS/TEN in the Chinese and others of Southeast Asian origin [19]. Recent studies have shown that HLA-A*31:01 and HLA-B*13:01 could also be used as risk alleles for antiepileptic drugs-induced cutaneous adverse reactions [20,21].
When the maximum tolerated dose of a treatment regimen fails to control epilepsy, a second antiepileptic drug may be considered. Drug-resistant epilepsy was defined as seizure free failure after adequate attempts with 2 tolerated, appropriately selected antiepileptic drug regimens [22]. Combination therapy is inevitable for patients with drug-resistant epilepsy [23]. Nearly a quarter of patients with epilepsy have drug-resistant epilepsy [24]. It is worth exploring whether the combination of multiple AEDs is associated with an increased incidence of SCAR. Therefore, we aimed to analyze the association between antiepileptic combination therapy and increased rates of SCAR based on real-world data from the FDA adverse event reporting system (FAERS) database.

Study design and data source
This was a retrospective, observational pharmacovigilance study designed to analyze the SCAR events associated with the combination use of AEDs, which were reported in FAERS databases from the first quarter 2004 to the fourth quarter 2021. The FAERS database contains public and anonymous patient information, no ethical approval was required, and informed consent could not be obtained.
The FAERS consists of seven data tables, including patient demographic and administrative information (DEMO), adverse events (REAC), patient outcomes (OUTC), drug information (DRUG), report sources (RPSR), therapy start states and end dates for reported drugs (THER) and indications for drug administration (INDI).

Procedure
Duplicate records were excluded according to the FDA recommendations [25]. If the CASEIDs (a number used to identify a FAERS case) were the same, the latest FDA_DT (date FDA received the case) was selected. If the CASEID and FDA_DT were the same, the higher PRIMARYID (a unique number for identifying a FAERS report) was selected. Then, we used the MedEx 1.3.8 software to standardize different names of the same drug into the 'generic name' [26].
The INDI table and REAC table are both coded by Medical Dictionary for Regulatory Activities (MedDRA) preferred terms (PTs) [27]. We gathered all the epilepsy cases reported in FAERS according to the High Level Group Term (HLGT, coded by MedDRA), including 84 PTs (Table S1). We attempted to identify 55 single component AEDs according to the WHO Anatomical Therapeutic Chemical (ATC) classification from the local FAERS database. The specific drug classification and ATC codes are shown in Supplementary Table S2. We finally identified SCAR cases out of all epilepsy cases based on the standard MedDRA query (SMQ) narrow search, including 18 PTs (Table S3).
After indication identification, we eliminated cases with SCAR complications reported in the INDI table. Then, we divided the included epilepsy cases into two groups: 1) the AEDs group (AED group), which AEDs were reported as primary suspected drugs, 2) the other primary suspected drugs group (OPS group), which other drugs except AEDs were reported as primary suspected drugs.

Statistical analysis
We managed the FAERS dataset in local use through Microsoft SQL Server 2017 software. The characteristics of SCAR cases with target drugs were collected, including sex, age, reporter (health professionals or non-health professionals), report country, report year. Algorithms of reporting odds ratio (ROR) and the information component (IC) were used to detect the association between SCAR events and target drugs [28]. We used EXCEL software to calculate the value of ROR and IC. For ROR method, the significant association is detected when the case number is ≥3 and the lower limit of the two-sided 95% confidence interval (95% CI) is >1. The ROR value was proportional to the correlation between the suspected drugs and SCAR. For IC method, a significant association is detected if IC >0 and the lower limit of 95% is >0. The SCAR events were considered drug-associated only when both the ROR and the IC methods met their threshold. The calculation method of ROR, IC and 95% CI is shown in Supplemental  Table S4.

General characteristics
We finally included 128,262 cases with the indication of epilepsy from the FAERS database, 104,278 cases in the AED group, and 23,984 cases in the other primary suspected drugs group (OPS group). The flow chart of identifying SCAR cases out of epilepsy patients was showed in Figure 1. The characteristics of events were described in Table 1. The AED group and OPS group had similar demographic characteristics, and the incidence of SCAR in AED group (3.26%) was higher than that in OPS group (2.51%). In both group the SCAR events in females (51.44% vs. 52.25%) were slightly more than in males, and the highest proportion of SCAR events were reported among 18-64 year olds (46.29% vs. 47.09%). The vast majority of SCAR reports came from health professionals (80.77% vs. 84.19%) in two groups. Europe (37.22% vs. 64.39%), Asia (31.30% vs. 15.97%) and North America (21.41% vs. 12.48%) reported the most events. The number of SCAR reports was increasing almost year by year. Similar demographic characteristics were observed in the antiepileptic drug monotherapy group and multidrug therapy groups. SCAR reporting rates were significantly higher in Asia and Africa.

SCAR signal detection
Our analysis was firstly based on the primary suspect drug that means when SCAR events were reported, one drug was mostly suspected to be associated with the adverse event, but other combinations were not excluded. As we show in Table 2 Subsequently, we analyzed the impact of antiepileptic drug combination on the incidence of SCAR. Firstly, we found that 10 drugs were associated with SCAR events when monotherapy (

Discussion
SCAR is one of the most concerned adverse reactions in the use of AEDs [29]. Currently, several studies based on realworld databases of adverse event reports have explored the association of SCAR with AEDs [15,[30][31][32]. However, only Yoshihiro's study [31], based on the Japanese pharmacovigilance database, demonstrated that nine antiepileptic combination regimens increased the incidence of SJS or TEN compared to monotherapy. The research suggested that carbamazepine-lorazepam and fosphenytoin-lorazepam may increase the risk of SJS; clobazam-gabapentin, phenytoingabapentin, valproic acid-gabapentin, clobazam-clonazepam, clobazam-valproic acid, fosphenytoin-lamotrigine and lacosamide-levetiracetam may increase the incidence of TEN. In our study, a total of 28 combinations were detected to cause SCAR, and 20 of them had an increased reporting rates of SCAR compared with monotherapy. To the best of our knowledge, this is the largest study which involving the most antiepileptic combination regimens to analysis the correlation of SCAR based on real-world data.
It is worth noting that all categories of AEDs may cause SCAR, among which barbiturates, hydantoin and carboxamide have the highest reporting rate, and fatty acids had the lowest. We analyzed the correlation of SCAR caused by individual drugs from two aspects. The first was based on the primary suspect drug, which means when SCAR events were reported, one drug was suspected to have a primary association with the adverse event, but other combinations were not excluded. Secondly, we analyzed the correlation between monotherapy and SCAR, which could exclude the influence of other combinations. The top five drugs with the highest reporting rates of SCAR detected by both methods were zonisamide, lamotrigine, phenytoin, phenobarbital and carbamazepine, all of which belong to the aromatic AEDs [14]. Cutaneous toxicity of lamotrigine [33], carbamazepine [34], phenytoin [35] and phenobarbital [36] has been demonstrated in previous studies. Particularly, only a few cases of SCAR caused by zonisamide have been reported [37]. However, zonisamide showed the strongest signal of association with SCAR when monotherapy in current study. This result is the same as Eric's [15]. Structurally, zonisamide is a sulfonamide derivative containing an aromatic ring, and hydroxylation breaks the molecule down into toxic arene oxide, increasing risk of allergic reactions [14]. Therefore, the cutaneous toxicity of zonisamide should not be ignored.
We found an increased reporting rate of SCAR with the 20 combination regimens compared with monotherapy. We detected no SCAR correlation with topiramate when monotherapy, but the reporting rate of SCAR was highest when combined with phenytoin. Similarly, valproic acid has a lower correlation of SCAR, but when combined with lamotrigine and zonisamide, the reporting rates of SCAR were significantly increased. Of the 20 combinations found in this study, 14 included at least one of the top five highest reporting rates drugs when monotherapy (lamotrigine, zonisamide, carbamazepine, phenytoin and phenobarbital). Therefore, it should be noted in clinical practice that the incidence of SCAR may be further increased when these five drugs are used in combination. Unlike Yoshihiro's study, we did not find an increased reporting rate of SCAR when lorazepam and gabapentin were used as one of the combination drugs. As we all know, skin rashes was the most common adverse effect in antiseizure patients aged <18 years [38]. In our study, patients younger than 18 years who received two antiepileptic drugs had significantly higher rates of SCAR reporting rate than those who received one antiepileptic drug (Table 1). This suggested that combination therapy may also increase the risk of SCAR in patients under 18 years of age. However, more researches may be needed to prove it.
The mechanism by which antiepileptic drug combinations may increase SCAR risk has not been elucidated. The immunological mechanism of SCAR is the non-covalent binding of drug to HLA and CD8 + T-cell receptor (TCR) to form HLA-drug -TCR complex, which may cause a series of reactions [39]. Related HLA is a necessary condition and drug concentration is a sufficient condition. Therefore, we hypothesized that the combination of drugs may increase the chance of HLA-drug-TCR complex formation, and increase SCAR risk.
There were several limitations to this study. First, the FAERS is a case report database of adverse drug reaction, various reporting biases due to spontaneous reporting was inevitable. Second, the total number of patients using antiepileptic drugs in FAERS database is not known, thus we could not calculate the incidence and prevalence. Finally, the mechanism by which the combination therapy increases the cutaneous toxicity cannot be explained in our study. We expect large prospective or retrospective studies to validate our results and elucidate the mechanism involved.

Conclusion
This study supports previous evidence that AEDs, including lamotrigine, carbamazepine, phenytoin and phenobarbital, are strongly associated with SCAR. In addition, the cutaneous toxicity of zonisamide may have been underestimated in past studies.
Current study has demonstrated the increased reporting rates of SCAR with 20 combination regimens. The top five combinations with the highest reporting rates were topiramate-phenytoin, lamotrigine-valproic acid, diazepam-phenobarbital, zonisamidevalproic acid, and lamotrigine-diazepam. Hence, patients treated with these combinations should receive special attention.