Gender Differences in Adverse Drug Events of Hydroxychloroquine: Analysis of Spontaneous Reports Submitted to FAERS


 Background Several studies have investigated gender as a risk factor for the occurrence of adverse drug events (ADEs) and found that females are more likely to experience ADEs than male. Today, there is a poor knowledge about gender differences in safety profile of ADEs to hydroxychloroquine (HCQ). Identifying those gender differences in ADEs could reduce the experience of ADEs for patients with HCQ. Therefore, the aim of this explorative study was to investigate whether differences exist in reported ADEs of HCQ for male and female in the database of FDA Adverse Event Reporting System (FAERS). Methods We performed a descriptive gender-related analysis and disproportionality analysis of HCQ safety data, obtained from the FAERS. Reporting odds ratio (ROR) and 95% confidence interval (CI) were calculated to quantify the signals of gender differences for specific drug-event combinations at system organ class (SOC) and preferred term (PT) level. Results Disproportionality analysis indicated that 8 SOCs with 12 ADEs were statistically significantly more reported in female than male, including electrocardiogram Qt prolonged, retinal toxicity, musculoskeletal disorder, hypersensitivity, anaphylactic reaction, among others, and 5 SOCs with 11 ADEs were reported more in male than female, including cardiac failure, renal failure, completed suicidal, photosensitivity reaction. Common adverse events are similar between female and male. However, serious ADEs were more frequently reported in males. Conclusions Therefore, the recognition of gender differences in ADEs may be helpful in prescribing medications, e.g. greater caution should be taken when prescribing HCQ to female with conduction disorder.


Introduction
HCQ approved by the FDA that have been used for many years to prevent and treat malaria and certain in ammatory conditions. Although HCQ well-established, they possess the potential to cause numerous side effects and should be used with caution in those with cardiovascular, neurological or vision disorders [1]. However, the HCQ FDA-approved labeling does not include su cient information regarding safety issues in gender factor. So far, increasing evidence has shown that female is more likely to experience adverse drug events (ADEs) than male [2], but it is unknown whether gender differences exist in the ADEs pattern of HCQ. Identifying those gender differences in ADEs could reduce the experience of ADEs for patients and could be conducive to the development of personalized medicine for both male and female. In order to provide additional guidance to HCQ prescription, we analyzed the ADEs in HCQ treatment aiming to nd some gender-based differences using the FDA Adverse Event Reporting System (FAERS) database.
Data-mining techniques, such as signal detection algorithms, are increasingly being used to explore medical databases and analyze large volumes of accumulated data to identify potential associations between drugs and ADEs that may have escaped detection in clinical trials [3]. Data from FAERS are publicly available and routinely used by the FDA, health systems, clinical scientists, and pharmaceutical manufacturers to identify potential safety signals. Thus, provided us a unique data source to conduct research for ADEs gender differences of HCQ.

Data source
The fastest and easiest approach (though not the best one) for the assessing of the safety pro le of drugs is the Adverse Drug Reaction Spontaneous Reporting System. Despite its limitations, spontaneous reporting could represent actually a valuable tool to assess gender differences in the onset of suspected ADEs. Pharmacovigilance databases can be used to assess a drug's safety by evaluating disproportional associations between the drug and one or more ADEs. The data for this study were retrieved from the public release of the FAERS database. The FAERS is a database that contains adverse event reports, medication error reports and product quality complaints resulting in adverse events that were submitted to FDA from all over the world. The database is designed to support the FDA's post-marketing safety surveillance program for drug and therapeutic biologic products.
In FAERS database, each report is coded using preferred term (PT) in the Medical Dictionary for Regulatory Activities (MedDRA) terminology. To identify drug-event pairs that did not reach the signal threshold with single-term analysis, and to correct for underestimation of drug-event pair signi cance due to variability in the terms chosen to describe essentially the same ADE, disproportionality analysis was performed rstly using SOC, which combine similar clinically relevant PT. After gathering this overall picture, further disproportionality analysis in PT level was based on target ADEs from HCQ label be approved by FDA.

Statistical methods
A ratio of reports concerning female versus male was calculated for all of the SOCs or PTs included in FDA label for HCQ. Instead of testing whether the ratio of the occurrences of an adverse event for male and female patients was the same as the baseline ratio (i.e., ratio of drug use), which was unavailable from the database, we tested whether the proportion of a target adverse event, among all reported events, was the same between the males and females. The effects of disproportionality analysis were evaluated using the established pharmacovigilance index reporting odds ratio (ROR). Then, a 2*2 contingency table was used to detect the gender-differential drug-event combinations (Tab 1). ROR value was used to quantify the difference (i.e., effect size) between the female and male gender for a speci c SOC or adverse event.
We rst tested the gender difference in drug-event combination frequency distributions summarized at SOC level. Thus, if the lower limit of the ROR 95% CI greater than 1, we considered the overall ADEs in this SOC to be reported statistically signi cantly more often in female than male. When the upper limit of the ROR 95% CI less than 1, we considered the overall ADEs in this SOC to be reported statistically signi cantly more often in male than female.
A similar ROR test for a 2*2 contingency table was used to detect the gender differences at PT level for target ADEs listed in label. To correct for multiple testing and provide a strong control of the family-wise error rate, the P values were adjusted using Bonferroni correction. Drug-event combinations with at least 50 overall occurrences altogether in both genders [4] and the number of female or male related ADE should be greater than 3 [5]. ROR value greater than 2 and the lower limit of the 95% CI greater than 1 indicates that the female patients are more likely to report a given adverse event than male patients. On the other hand, when the ROR value is less than 0.5 and the upper limit of the 95% CI less than 1, male patients tend to report the adverse event more frequently.
At adverse events level, we manually selected target ADEs from HCQ label be approved by FDA [1]. These adverse reactions have been identi ed during post-approval use of hydroxychloroquine or other 4aminoqunoline compounds. Because these reactions are reported voluntarily from a population of uncertain size, it is not always possible to reliably estimate their frequency or establish a causal relationship to drug exposure. These target ADEs involved 15 different SOCs with a total of 89 preferred terms according to the MedDRA Terminology.

Descriptive analysis
Up to 2020, FAERS received 25619 reports in which HCQ was reported as a suspect or interacting drug. Table 2 lists the demographic data of HCQ reports. More than three-quarters ratio of the reports concerned female (76.61%) exclude the unknown gender. The overall female to male ratio reported in these reports is 4.97:1 and that for adults (18-64 years) is 5.25:1. Patients age distribution was similar for male and female in different age groups and mainly concentrated in the 18-64 years age group. There are peaks in the 40-50 and 60-70 age groups regardless of gender or not (Fig. 1). One possible explanation for this concentration of ADEs in patients may be related to the age-speci c incidence rates of most common therapeutic indications (e.g. rheumatoid arthritis and lupus erythematosus) [6]. The number of reports was consistently low before 2010, whereas a remarkable rise was observed from 2015 to 2019, with a difference in the male and the female groups in particular from 2015 (Fig. 2). This can be explained by the use of HCQ is increased due to a growing list of clinical indications, and subsequently growing side effects being under recognized and reported. Almost two-thirds of the cases reporter region derived from foreign area and there is basically no difference in the proportion considered each area for both genders. In the United States, the male to female ratio is 5.23:1. Healthcare professionals were responsible for 64.92% of the reports, while 24.57% were reported by consumer regardless of gender.
Reporter distribution was similar for male and female.
The distribution of ADEs outcomes between male and female seemed to be similar. Both genders more frequently reported hospitalized and other outcomes. However, in terms of seriousness, the proportion of serious cases is higher in male than in female including death and life-threatening cases (5.40%, 4.76% in males and 2.91%, 2.31% in females). This nding can be partially explained by the fact that female report more physical symptoms and use more medical services compared to male [7].
At PT level, the most frequently reported ADEs, regardless of gender or not, were drug ineffective, rheumatoid arthritis, drug intolerance, and pain. Table 3 lists the rst 30 most-reported ADEs for each gender, which account for 41.37% of all ADEs reported by female and 38.83% of those reported by male.
Other events had a very similar distribution between genders. Obviously, some ADEs reports are drug indications or original diseases related symptoms for HCQ, such as rheumatoid arthritis, joint swelling and arthragia. FDA does not require that a causal relationship between a product and event be proven, so these ADEs have also been reported. Percentage calculated based on the total number of reported cases for the female gender (n = 19627) and male gender (n = 3946) respectively. Percentage calculated based on the total number of reported cases for the female gender (n = 19627) and male gender (n = 3946) respectively.

Disproportionality Analyses
Detect gender difference for speci c drug-event combinations summarized at SOC level The data on suspected ADEs classi ed by SOC showed some gender-based differences (Table 4). In the absolute number of reports, the ADEs number of female in each SOC was higher than that of male, but the disproportional outcome is different between male and female. Of the total 27 SOCs, ROR analysis show that the female patients were more likely to report these drug-event combinations in 10 different SOCs, of which target ADEs from HCQ label involved 8 SOCs, including eye disorders, musculoskeletal and connective tissue disorders, skin and subcutaneous tissue disorders, gastrointestinal disorders, immune system disorders, nervous system disorders, endocrine disorders, ear and labyrinth disorders. On the contrary, male patients were more susceptible to reporting these drug-event combinations in 11 different SOCs, and 5 SOCs involve target ADEs in HCQ label, including cardiac disorder, hepatobiliary disorders, renal and urinary disorders, psychiatric disorders, blood and lymphatic system disorders. No gender differences were observed in 6 SOCs, except for general disorders and administration site conditions or metabolism and nutrition disorders, which also not involve target ADEs in HCQ label, including investigations, pregnancy, puerperium and perinatal conditions, reproductive system and breast disorders, product issues.

Discussion
Despite the obvious physical and physiological gender differences, gender differences in ADEs are lack of awareness rarely considered in clinical treatment. More speci cally, gender differences in ADEs are still widely unknown and poor data is available in the literature in this topic. To the best of our knowledge, this is the rst comparative safety study on FAERS aiming at assessing gender differences for HCQ related ADEs. Overall, three main ndings emerged: a) reports of ADEs are more likely to concern female than male regardless of SOC or PT level, especially the ADEs that are common, such as fatigue, alopecia, vomiting, malaise, dizziness, and so on; b) we found disproportionality signal for female in ocular, musculoskeletal, skin, gastrointestinal, ear, neurological, immune systems and endocrine systems at SOC level, for male in cardiac, hematologic, renal, liver and psychiatric; c) disproportionality analysis for female showed a higher reporting in electrocardiogram Qt prolonged, retinal toxicity, musculoskeletal disorder, muscle spasms, hypersensitivity, anaphylactic reaction, among others, while cardiac failure, completed suicidal, photosensitivity reaction, kidney injury include renal failure have been more commonly reported for male. To provide instructional advice on gender difference, we focused our analysis on label warning related ADEs of HCQ provided by FDA, as shown below.
Cardiotoxicity The cardiac effects induced by HCQ predominantly including conduction disorders and cardiomyopathy. The mechanism is poorly understood but likely involves direct lysosomal dysfunction via inhibition of lysosomal enzymes, leading to accumulation of metabolic products [8][9]. We found that conduction disorders associated with HCQ, especially QT interval prolongation, were more commonly reported in female, while cardiomyopathy -though rare -which may result in cardiac failure was more frequent in male.
Conduction disorders Conduction disorders complications from HCQ therapy are rare but clinically important events manifest as QT interval prolongation, torsade de pointes, ventricular arrhythmias, bundle branch block/atrio-ventricular block, ventricular tachycardia and ventricular brillation were reported in FAERS. Susceptibility to HCQ-induced QT interval prolongation was greatest for female in our study. This nding was con rmed by clinical observations and in vitro experimental data that female gender is an independent risk factor for drug-induced Long QT syndrome [10][11][12] which in females is in uenced by other factors, including age, menstrual cycle, and hormone replacement therapy in previous studies [13]. The mechanism remains uncertain, the estrogen-mediated reduced repolarization reserve in female is believed to be responsible for their higher susceptibility to drug-induced Long QT syndrome [14,15]. Medications that cause QT prolongation generally act by blocking Ikr, a potassium channel protein that regulates an important rapid delayed repolarizing current in phase 3 of the cardiac action potential [14]. Underlying genetic defects of ion channels that may even be asymptomatic in normal conditions may be unmasked by QT prolonging drugs in female more frequently than in male [11]. Additionally, overdose of HCQ was more likely to concern female than male [116/7, ROR = 2.80, 95%CI (1.30; 6.00)], which may resulted in conduction disorders including QT prolongation.
The occurrence of drug-induced torsade de pointes (TdP), which is much more frequent in female. During AV block, longer QT interval and female gender strongly correlate with the risk of TdP [14]. Female gender increases the risk of TdP in the Long QT syndrome and it is assumed that this increased risk is due to their longer QT interval. However, each of these variables is an independent predictor of TdP [16]. There was a trend of female signal in TdP, but the gender difference was not statistically signi cant. The fact may be that few patients reported to suffered from abnormal conduction such as TdP and ventricular arrhythmias owe to relatively unrecognized and many cases may go undiagnosed and serious underreported.
Cardiomyopathy Epidemiological studies showed that male sex is an important risk factor for developing cardiac failure [17], which consistent with the disproportionality signal for female of our study. Cardiac failure associated with HCQ can be the consequence of a number of cardiovascular conditions including cardiomyopathy, myocarditis [9]. The gender difference been con rmed in clinical studies that myocarditis is considered as a male predominant disease. For myocarditis and cardiomyopathy female gender is protective, in sharp contrast to its more adverse impact in more clearly de ned autoimmune disorders such as lupus and rheumatoid arthritis [18]. And the better myocardial recovery in females than males was observed in murine models of myocarditis and in clinical studies of recent onset cardiomyopathy. The mechanism remains uncertain, but differences in immune function by gender almost certainly play an important role [18]. It is crudely noted in our results that cardiomyopathy associated with HCQ may be more probable with male compared with female even though the gender difference was not statistically signi cant.
Ocular Retinopathy, the most well-known complication of HCQ, may manifest as a spectrum of changes from asymptomatic and reversible pigment changes to visual loss persisting or progressing after drug discontinuation. HCQ affect lysosomal function of retinal pigment epithelium with an additional role of oxidative stress [19]. A retrospective case-control study involving 2361 people who had used HCQ continuously for at least 5 years indicates that the risk factors of HCQ retinopathy include high daily intake and with long durations of use or in the presence of kidney disease or concurrent tamoxifen therapy, excluding gender [20]. Here, interestingly, ocular toxicity appears to be stronger associated with the use of HCQ in female patients, both at the SOC and PT level. Considering that few male patients suffered from HCQ retinopathy especially macular degeneration, photophobia and so on owe to the epidemiology of autoimmune diseases which predominantly affect female and consequently to the gender distribution. We cannot conclude if female gender is a risk factor for retinal damage due to HCQ. We will continue to focus on the gender difference of retinal toxicity.
Musculoskeletal Our study observed a slightly higher frequency for female gender in musculoskeletal and connective tissue disorders at the SOC level. The most commonly reported musculoskeletal ADEs with HCQ were predominantly skeletal muscle myopathy and neuromyopathy, which usually leading to progressive weakness and atrophy of proximal muscle groups, depression of tendon re exes and abnormal nerve conduction. Clinically, most of the symptomatic patients exhibited muscle weakness with proximal greater than distal distribution.  [21]. Female gender is one of the risk factors for development of curvilinear bodies [22] and myopathy [21] related to HCQ, which could partially explain our nding. This is an occasionally and interesting nding of this study that respiratory failure maybe associated with proximal myop-nathy secondary to HCQ [23] was more often reported in male.
Skin In our study, female is at greater risk of anaphylactic reaction and hypersensitivity related HCQ than male, and this was proved by the conclusion female is a risk factor for allergic reactions to certain drugs. Additionally, in gender-based medicine, a particularly signi cant, emerged from several studies, is the higher risk of female developing dermatologic events (e.g., rash) than male [2,12,[24][25]. This can be explained by the disproportionality signal in SOC "skin and subcutaneous tissue disorders" for female when viewing the skin disorders as a whole. Besides, we found a disproportionality signal for female in alopecia, whereas, for male, the ROR became signi cant in photosensitivity reaction, and there was no difference in the proportion of male or female in severe cutaneous drug reactions, which including Stevens-Johnson syndrome, toxic epidermal necrolysis, erythema multiforme, dermatitis bullous, acute generalized exanthematous pustulosis and drug reactions with eosinophilia and systemic symptoms due to few data. Drug-induced photosensitivity reactions can be subdivided into photoallergy and phototoxicity, which clinically resemble contact dermatitis and sunburn, respectively [26,26]. Interestingly, HCQ, often used for photosensitive diseases such as lupus erythematosus (up to 69% of patients with SLE have photosensitivity), can induce photoallergy [27]. Furthermore, we need to pay attention to skin hyperpigmentation which may serve as a marker for patients at risk for developing more severe side effects [27].
Gastrointestinal Our results come as no surprise to us that female gender tends to have a higher risk of gastrointestinal events (e.g., vomiting) in speci c SOC and PT level. This corresponds to the clinical observation that gastrointestinal disorders, which are generally considered to be a dose-related or commonly occurring ADEs, occurred in a higher proportion of female than male [2]. Many of the pharmacokinetic differences, such as higher plasma drug levels in female and higher percentage of body fat in female (e.g., drug redistribution effect), may result in female patients experiencing more doserelated effects [12]. Additionally, this nding is coincidental with higher rates of drug utilization and exposure to a larger variety of drug classes among female patients [25]. Others HCQ -uncommonly recognized as a nephrotoxin, lysosomal accumulation of globotriaosylceramide leads to chronic kidney disease. Histologically, laminated podocyte inclusions known as zebra bodies are the hallmark nding [28]. A stronger association with renal insu ciency including renal failure were emerged in male following HCQ treatment and further researches are needed to con rm this.
Completed suicidal associated with HCQ appears to be more often in male patients. This was consistent with epidemiological study shows that females have higher rates of suicidal ideation and behaviour while completed suicides occur primarily among males [29].
Some studies continue to suggest female are more likely to develop DILI [30], but others suggest that the genders share a nearly equal risk [31]. Reports of HCQ in liver injury or liver failure still insu cient to draw rm conclusions in our study.
Fewer cases of SOC "endocrine disease" were reported, but the most signi cant difference between male and female due to disease interferes such as osteoporosis, resulting in skewed results. Hypoglycaemia related to HCQ, though not common, is an ADE of clinical special interest as it may be life-threatening, and there is no signi cant difference between male and female.
We acknowledge inherent limitations of our study: by de nition, FAERS data cannot be used to calculate the incidence of an ADE, because of the under-reporting phenomenon, the lack of solid data on population exposure and the absence of certainty that the drug is actually responsible for the reported event [32]. ROR computing does not allow a quanti cation of the true risk of ADE for different gender, it only suggests a statistically signi cant disproportionality of speci c drug-event pair of interest, which should be further investigated for signal validation [33]. In addition, various biases such as over-reporting, duplicates, missing data, the possibility effect of diseases and drug interaction on signal detect should be taken into account when interpreting the results obtained from spontaneous report analysis [34]. Despite these caveats, the ROR methodology is a well-established tool for identifying drug-event pairs reported more frequently between females and males, as it is likely that many of the limitations mentioned are equalized across different gender and over time [35]. In our analysis, four steps were taken to weaken for the limitations of disproportionality analysis of FAERS data: (1) De nite gender reports were used for analysis; (2) SOC analysis was performed to control for underestimation of HCQ-event pair signi cance due to variability in describing the same ADEs with different PTs; (3) ltered out the HCQ-event with occurrences less than 50 for two genders combinations; (4) ROR ≥ 2 for female or ROR ≤ 0.5 for male thresholds and n ≥ 3 were selected to increase the certainty of recognition HCQ signals at PT level; (5) correlation of disproportionality results to data from published researches, where possible.

Conclusion
In conclusion, ADEs submitted to the FDA were reviewed to con rm whether differences exist in reported ADEs of HCQ for male and female and to clarify rank-order in terms of susceptibility. Despite the inherent limitations of the FAERS data, some of the results we obtained regarding the gender difference are in agreement with the results of previous studies, suggesting the usefulness of the FAERS database and the data mining method used herein. Greater caution should be taken when prescribing HCQ to female in electrocardiogram Qt prolonged, retinal toxicity, musculoskeletal disorder, hypersensitivity, anaphylactic reaction, while cardiac failure, renal failure, completed suicidal, photosensitivity reaction, to male, but these require further investigation to better de ne the risk. In summary, a heightened understanding of gender differences is critical for improving treatment strategies and clinical management that will lead to optimal care for both female and male with these potentially devastating. And such gender-based differences pharmacovigilance study could blaze new trails for further research in gene-gender effects. Declarations