Spontaneous cutaneous adverse drug reaction reports—An analysis of a 10‐year dataset in Singapore

Abstract We analyzed the spontaneous adverse event database in Singapore to determine the types of cutaneous adverse drug reactions (CADRs) and causative drugs reported. We selected 10 CADRs‐of‐interest, and identified the suspected drugs and the characteristics of the at‐risk population. ADR reports received from 2006 to 2015 of the system organ class “Skin and Appendages Disorders” were analyzed based on patient demographics, the types of CADRs, suspected drugs, outcome, and latency period. Of the 104 372 reports analyzed, 56.2% involved females and 72.5% involved Chinese patients. The mean age was 41.1 years old. The top CADRs reported were rash (including nonspecified rash, follicular rash, maculopapular rash, and vesicular rash) (67.2%) and angioedema (13.9%). The drugs frequently associated with the CADRs‐of‐interest include nonsteroidal antiinflammatory drugs and antibiotics with angioedema, iohexol with urticaria, and antiepileptics and allopurinol with Stevens‐Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN). A subgroup analysis based on age, sex, and race on the 10 CADRs‐of‐interest showed the following trends in reporting: Alopecia (reported more in females), drug hypersensitivity syndrome (more in males), angioedema (more in younger patients), and photosensitivity (more in older patients). In general, the racial distribution across each CADR‐of‐interest was consistent with that of Singapore's population, with slight deviations observed for SJS/TEN, photosensitivity and skin discoloration. We analyzed CADR reports from Singapore over 10 years, and identified the types of CADRs reported, and their associated drugs, latency periods and patient characteristics. Such information could add value to healthcare professionals as they assess CADR cases and evaluate suspected drugs.


| INTRODUC TI ON
According to the World Health Organisation (WHO), an adverse drug reaction (ADR) is a response to a medicinal product which is noxious and unintended and which occurs at doses normally used in man. 1 Cutaneous ADRs (CADRs) are one of the most common ADRs, [2][3][4] with an overall incidence rate of 2%-3% in hospitalized patients. 5 In the WHO global ADR database, VigiBase, skin and appendages disorders account for 18.3% of over 13 million ADR reports received from more than 100 countries, making it the third most frequently reported system organ class (SOC). 6,7 As the national regulatory agency in Singapore, the Health Sciences Authority (HSA) receives around 20 000 ADR reports annually in the adverse event (AE) database, of which 60% were related to skin reactions.
The manifestation of CADRs can be very varied, ranging from mild, self-limiting reactions to severe cutaneous adverse reactions (SCARs) associated with significant morbidity and mortality, such as acute generalized exanthematous pustulosis (AGEP), Stevens-Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), and drug hypersensitivity syndrome (DHS). CADRs are also associated with a wide range of drugs, with antimicrobials, NonSteroidal Anti-Inflammatory Drugs (NSAIDs), antiepileptics, and analgesics as the most frequently implicated drug classes. [8][9][10][11][12] While much is known about CADRs, information on the types of CADRs reported through the spontaneous ADR reporting system is limited. We seek to fill this knowledge gap with an analysis of a large dataset with over 100 000 CADR reports from a 10-year period from the HSA AE database.
Our objectives are to determine the types of CADRs and associated drugs reported to the HSA AE database, to identify characteristics of the at-risk population, and to identify associations between CADRs and drugs.

| Data source
In Singapore, spontaneous AE reports are submitted to HSA and captured into the national AE database. These reports come primar- For each report, AEs were coded using the WHO Adverse Reaction Terminology (WHO-ART), drugs were classified using the Anatomical Therapeutic Chemical (ATC) Classification System, and causality was assessed based on the WHO Uppsala Monitoring Centre (WHO-UMC) causality assessment system.

| Inclusion criteria
Spontaneous AE reports which met the following criteria were included in our study: (1) report was received between 2006 and 2015, (2) reported AE belongs to the SOC of "Skin and Appendages Disorders" and (3) causality was assessed as certain, probable or possible.

| Data extraction, collation, and analysis
Anonymized data with basic demographic information, reporters' profession, AE description, patient outcome, and suspected drug(s) were extracted from the AE database.
To facilitate analysis, WHO-ART preferred terms used for coding the CADRs were grouped into 21 reaction types. From these, 10 CADRs-of-interest were selected for in-depth analysis based on clinical relevance.   Table 2. Rash (including nonspecified rash, follicular rash, maculopapular rash, and vesicular rash) was most frequently reported (67.2%), followed by angioedema (13.9%) and pruritus (7.4%). SCARs such as SJS/TEN, AGEP and pustular rash, and DHS were reported less frequently, in 0.7%, 0.5%, and 0.2% of the reports, respectively. Table 3 shows the top 10 drug classes, and the respective top five drugs in each class. Systemic antibacterials were most commonly implicated (43.5%) followed by antiinflammatory and antirheumatic products (16.2%) and analgesics (9.0%). Antiepileptics made up 1.6% of the suspected drugs, and was ranked seventh.  Table 4 lists the 10 CADRs-of-interest and their top 10 suspected drugs. Antibacterials and NSAIDS were frequently associated with all types of CADRs, with the exception of alopecia. 13,14 Antiepileptics and allopurinol were commonly implicated in SJS/TEN and DHS.

| RE SULTS
For each CADR-of-interest, a subgroup analysis was performed based on age, sex, race, and latency of reaction (Table 5). Alopecia had the lowest male/female ratio at 0.28 while DHS had the highest at 1.40. Angioedema was reported by younger patients (median age = 34 years) while photosensitivity, DHS and SJS/TEN were reported by older patients (median age = 50-58 years). In general, the racial distribution across each CADR-of-interest was consistent with that of Singapore's population, with trending towards the Malays, Chinese and Indians for SJS/TEN, photosensitivity and skin discoloration, respectively. These trends remained even after taking into account the population size 15 of each race to estimate the frequency of each CADR-of-interest across races. The latency period for CADRs also varies, from acute ones (eg angioedema, urticaria) to those which take weeks to occur (eg photosensitivity, alopecia).

| D ISCUSS I ON
In this study, we analyzed a large volume of spontaneous AE reports and highlighted the various patterns of CADRs and their implicated is postulated that radiocontrast media acts via a nonimmunologic mechanism by triggering direct mast cell degranulation, resulting in histamine release. 20,21 As such, most cases of symptomatic urticaria can be managed with an antihistamine such as diphenhydramine. 22

| Stevens-Johnson syndrome, toxic epidermal necrolysis (SJS/TEN) and drug hypersensitivity syndrome (DHS)
While SJS/TEN has multiple causes, most are drug-induced, 23  The frequency for this allele in Singaporeans is approximately 1 in  85.5% of these occurred in Chinese patients, although the Chinese make up 74.1% of the general population. 16 In contrast to CBZ, a cost effectiveness study found that genotyping all gout patients for the HLA-B*5801 allele prior to initiation of allopurinol is currently not cost-effective for Singapore's overall population from a health systems perspective. The relatively low positive predictive value of the test and limited alternative urate-lowering therapies were contributory factors to this. 30 In the EuroSCAR study, oxicam-NSAIDs (eg piroxicam) were found to be strongly associated with SJS/TEN. 29 Conversly, COX-2 inhibitors, celecoxib and rofecoxib, did not show such an association, 34 although cases of etoricoxib-induced TEN have been reported. 32,33 In our setting, systemic oxicam-NSAIDs are not widely used, and we found COX-2 inhibitor etoricoxib as the most frequently reported NSAID associated with SJS/TEN instead.
Sulphonamide antibiotics (eg cotrimoxazole) are well known to cause SJS/TEN, but for amoxicillin and amoxicillin-clavulanic acid, this is less clear. Half of the SJS/TEN reports we received for the drug class penicillins had co-suspected drugs, and in some cases, the antibiotic was started only after the prodromal symptoms of SJS/ TEN appeared.
Omeprazole was found to be a co-suspected drug in 60.9% of the omeprazole-SJS/TEN cases, suggesting it could be an innocent bystander. Similarly, in other studies, 31,34 it was found that pantoprazole was commonly taken with drugs which carry higher risks for SJS/TEN, and was often not temporally convincing. Other proton pump inhibitors (PPIs) have also been reported to carry nonsignificant risk for SJS/TEN. 31

| Fixed drug eruption (FDE)
Drugs, specifically NSAIDs and antibiotics, are the most common cause of FDEs, and our findings reflect this. Interestingly, a retrospective chart review by the Singapore National Skin Center 35 also identified etoricoxib as the most common cause of FDE, accounting for 38.7% of 62 FDE patients. In that study, three-quarters of the patients reacting to etoricoxib were Chinese patients, and this was also observed in our study. The possibility of genetic predisposition was considered, suggesting a genetic association resulting in higher incidence of etoricoxib-induced FDE in the local Chinese population. 35

| Acute generalized exanthematous pustulosis (AGEP) and pustular rash
In a EuroSCAR study, 36 aminopenicillins, sulphonamides, and quinolones were found to be highly associated with AGEP, but not paracetamol and cephalosporins. In comparison, our results showed that all these drugs were frequently reported to cause AGEP, although in the paracetamol and ceftriaxone cases, one-third were accompanied with co-suspect drugs. While our database did not feature reports of (hydroxy)chloroquine, terbinafine and diltiazem-associated AGEP prominently (≤5 reports for each drug), the EuroSCAR study did detect them as culprit drugs which have a high risk of causing AGEP.
Different prescribing patterns and under-reporting of AEs could explain this. For similar reasons, we also identified different NSAIDs as causative agents: The EuroSCAR study reported oxicam-NSAIDs, as opposed to diclofenac and ibuprofen seen in our study.

| Bullous eruptions
Bullous eruptions encompass a range of clinical presentations such as pemphigus, bullous pemphigoid, and linear IgA bullous dermatosis. 5,37 Pemphigus can be triggered via a biochemical pathway by thiol drugs (eg d-penicillamine, captopril, lisinopril) or phenol drugs (eg rifampicin, aspirin, levodopa), or via an immune-mediated pathway by non-thiol drugs (eg cephalosporin, penicillin, enalapril), with both pathways leading to acantholysis. 37 Bullous pemphigoid, on the other hand, is most often associated with thiol drugs. 5,37 In linear IgA bullous dermatosis, vancomycin is the most common culprit drug. 37 While most of our reports of bullous eruptions did not specify the type of bullous disorder, we did receive reports of vancomycinassociated linear IgA bullous dermatosis, bullous pemphigoid secondary to enalapril, pemphigus vulgaris associated with captopril, as well as with pemphigus with rifampicin.

| Photosensitivity
Drugs often associated with photosensitivity (eg tetracyclines, thiazides, chlorpromazine, amiodarone) were also elucidated by our study. 5,38,39 Although entecavir is not known to cause photosensitivity reactions, there were two such reports in our database, one of which was confirmed with skin biopsy showing deep perivascular dermatitis with eosinophils. The elderly patient was on entecavir for 5 months, and recovered 3 months after cessation of drug. Our results also suggest that patients who are female, Chinese or of older age are more likely to develop and report photosensitivity reactions.
The protection offered by melanin in darker-skinned patients could explain the higher reporting of photosensitivity reactions by the Chinese. 40

| Alopecia
Drug-induced alopecia could be categorized into anagen effluvium (ie hair growth phase) or telogen effluvium (ie resting phase). 41 Anagen hair loss is often dose related and commonly associated with chemotherapy, but reports of chemotherapyinduced alopecia are lacking from the AE database. One reason is that expected reactions such as these tend to be under-reported.
Telogen hair loss is linked to a wider variety of endogenous and exogenous factors, such as major surgery, serious illness,  (17); Doxycycline (16); Aspirin (14); Mefenamic Acid (14) DHS 325 Allopurinol (57) The number of drugs involved ≠ number of reports for CADR as more than 1 drug could be suspected in a single report.

| Skin discoloration
Drugs often implicated in causing skin discoloration are amiodarone, tetracycline, NSAIDs, antimalarials, and psychotropic drugs.
Amiodarone is associated with slate-gray pigmentation in unprotected light exposed skin areas, especially in light-skinned individuals on prolonged therapy. This hyperpigmentation is due to the accumulation of amiodarone and its metabolites in the skin. 43,44 Long term use of tetracyclines are also known to cause abnormal pigmentation. With minocycline, especially when used long term for treatment of acne, pigments are deposited as a result of an insoluble minocycline-melanin complex. 45 Our study found that females are more likely to report drug-induced skin discoloration.
Although Indians make up about 9.2% of the local population, 16 there appears to be a greater proportion of reports of skin discoloration from Indians (20.0%).

| CON CLUS ION
While our analysis did not detect any new associations or drug signals, it is interesting to note that CADRs such as SJS/TEN, photosensitivity and skin discoloration seem to be reported more frequently in Malays, Chinese, and Indians, respectively. However, given the limitations of the spontaneous reporting system, these observations should be taken into context. To overcome these limitations, HSA is looking into leveraging on electronic medical records to detect ADRs, including SCARs. With information on race, diagnoses, and medication history, it will become possible to explore the distribution of various CADRs across different races.
In summary, we have analyzed a large dataset of over 100 000 CADR reports from a 10-year period and identified the types of CADRs reported, as well as their associated drugs, latency periods, and patient characteristics. Such information could add value to healthcare professionals as they assess CADR cases and evaluate suspected drugs. Timely and accurate identification of the causative drug can enable HCPs to take the appropriate actions and improve patient clinical outcome.

ACK N OWLED G EM ENTS
The authors are grateful for the advice on statistical analysis by   Table 5 lists the CADRs-of-interest from most to least commonly reported. The median age, sex ratio (M:F), racial distribution, and median latency of each CADR is listed. CADRs with sex ratio (M:F) greater than one indicate that the CADR is reported more in males than in females, ie FDE, bullous eruption, and DHS. The racial distribution across each CADR was generally consistent with that of the local population [Chinese (74.1%), Malays (13.4%), Indians (9.2%), Others (3.3%) 16 ], with slight deviations noted for SJS/TEN (more frequently reported in Malays), photosensitivity (more frequently reported in Chinese) and skin discoloration (more frequently reported in Indians).
(Vigilance & Compliance Branch) at the HSA, for their steadfast support to pursue this research to advance the organisation's pharmacovigilance mission.

D I SCLOS U R E
None declared.