Reducing severe cutaneous adverse and type B adverse drug reactions using pre‐stored human leukocyte antigen genotypes

Abstract Background Several type B adverse drug reactions (ADRs), especially severe cutaneous adverse reactions (SCARs), are associated with particular human leukocyte antigen (HLA) genotypes. However, pre‐stored HLA information obtained from other clinical workups has not been used to prevent ADRs. We aimed to simulate the preemptive use of pre‐stored HLA information in electronic medical records to evaluate whether this information can prevent ADRs. Methods We analyzed the incidence and the risk of ADRs for selected HLA alleles (HLA‐B*57:01, HLA‐B*58:01, HLA‐A*31:01, HLA‐B*15:02, HLA‐B*15:11, HLA‐B*13:01, HLA‐B*59:01, and HLA‐A*32:01) and seven drugs (abacavir, allopurinol, carbamazepine, oxcarbazepine, dapsone, methazolamide, and vancomycin) using pre‐stored HLA information of transplant patients based on the Pharmacogenomics Knowledge Base guidelines and experts' consensus. Results Among 11,988 HLA‐tested transplant patients, 4092 (34.1%) had high‐risk HLA alleles, 4583 (38.2%) were prescribed risk drugs, and 580 (4.8%) experienced type B ADRs. Patients with HLA‐B*58:01 had a significantly higher incidence of type B ADR and SCARs associated with allopurinol use than that of patients without HLA‐B*58:01 (17.2% vs. 11.9%, odds ratio [OR] 1.53 [95% confidence interval {CI} 1.09–2.13], p = 0.001, 2.3% versus 0.3%, OR 7.13 [95% CI 2.19–22.69], p < 0.001). Higher risks of type B ADR and SCARs were observed in patients taking carbamazepine or oxcarbazepine if they had one of HLA‐A*31:01, HLA‐B*15:02, or HLA‐B*15:11 alleles. Vancomycin and dapsone use in HLA‐A*32:01 and HLA‐B*13:01 carriers, respectively, showed trends toward increased risk of type B ADRs. Conclusion Utilization of pre‐stored HLA data can prevent type B ADRs including SCARs by screening high‐risk patients.

B ADR and SCARs associated with allopurinol use than that of patients without HLA-B*58:01 ( Vancomycin and dapsone use in HLA-A*32:01 and HLA-B*13:01 carriers, respectively, showed trends toward increased risk of type B ADRs. Conclusion: Utilization of pre-stored HLA data can prevent type B ADRs including SCARs by screening high-risk patients.  3,4 Individual genetic variability results in susceptibility to different ADRs; therefore, it is crucial to utilize the genomic data of patients for drug prescription. 3 Currently, various genetic tests are performed in hospitals, and a vast amount of genetic information is already pre-stored in electronic medical records (EMRs). However, this information is rarely used for indications outside its primary purpose.
Lack of integration between the genetic information of the patient and the EMRs is an obstacle in patient-specific drug prescription at the point-of-care. 4 Data on major pharmacogenomic (PGx) variants pre-stored in the EMR should be used when prescribing high-risk drugs to patients. 3,5,6 The clinical validity of the drug-gene relationship used in this approach is mainly based on the Clinical Pharmacogenetics Implementation Consortium guidelines. 7 Preemptive genotyping has many advantages compared to reactive genotyping. For example, the genotype information of patients can be used without delay in the prescription process. The genotype information can also be used to build a system to support physicians in making personalized prescription decisions. Furthermore, preemptive genotyping is a cost-effective approach as many drug-related variants can be obtained using a single panel. 8 In reality, preemptive genotyping is not widely used in clinical practice, and PGx genes and variants found in a majority of PGx panels mainly focus on the pharmacokinetic/pharmacodynamic genes, including cytochrome P450 enzyme families. 4 Therefore, these PGx genes are not tested for purposes other than their use in drug prescription.
In recent decades, particular human leukocyte antigen (HLA) alleles have been found to be strongly associated with the development of certain drug-related SCARs. 9,10 We hypothesize that use of HLA PGx alleles can prevent SCARs. Despite the strong associations between some HLAs and drug-related SCARs, pre-stored HLA data obtained from transplant workup tests are not being utilized to screen individuals at a risk of developing SCARs when high-risk drugs are prescribed. Storing HLA data in a structured, standardized format in EMRs is challenging as different testing methods have been used to determine HLAs over the years. Nonetheless, if pre-existing HLA data can be successfully retrieved and re-used based on the PGx indications, it would reduce the costs of testing and effort required to obtain the same HLA information. A clinical decision support system using the pre-stored genetic data can also be utilized as a part of the point-of-care if successfully integrated.
In a previous study, we extracted, parsed, and saved the HLA data of transplant patients in a structured, standard format from prestored unstructured HLA data. 11 This study investigated the potential clinical benefits of using the extracted HLA genotypes as a risk prediction marker for ADR.

| Pruning human leukocyte antigen alleles and drugs associated with adverse drug reactions
The gene/variant-drug relationship data were downloaded from the Pharmacogenomics Knowledge Base (PharmGKB) (accessed on January 25, 2019). 12 (Table S1). Table 1 presents the list of drugs, HLA alleles, and related ADRs included in this study. The level of evidence of the relationship in Table 1 was verified from PharmGKB, and the references were based on the PharmaGKB database or experts' review.

| Identifying patients with human leukocyte antigen-related adverse drug reactions
To determine the number of patients who experienced HLA-related ADRs due to the seven drugs included in this study, the prescription data of participants, HLA allele information, diagnostic codes of type B ADRs such as "toxic maculopapular eruption," "acute generalized exanthematous pustulosis (AGEP)," "SJS," "TEN," "DRESS syndrome," and "drug eruption" were reviewed in SUPREME ® . The ADRs that we used were classified and represented in Figure S1. We utilized two different data sources to determine the number of patients who experienced HLA-related ADR due to the seven drugs included in this study. The first was a database of all adverse drug events, ICSR diagnostic code of EHR extracted from SUPREME ® . In addition to the diagnostic code, HLA prescription information and HLA allele information were also obtained from SUPREME ® . To target only drug side effects that are likely to be related to HLA, we first limited the types of ADRs to Type B reactions. We divided the Type B reactions into mild/moderate and severe according to the severity because it is challenging to define a causal relationship between the various ADRs and the drugs, especially when the ADRs are mild or subtle such as a slight skin rash. We classified "toxic maculopapular eruption," "acute generalized exanthematous pustulosis (AGEP)," "SJS," "TEN," and "DRESS syndrome," as severe Type B reactions, and if a patient ever had this diagnosis in EHR or had been reported in the ICSR with one of those diagnoses, this patient was defined as showing a severe Type B reaction. Considering that if the patient's ADRs are very certain and severe, the attending physician is likely to register for the diagnosis immediately, we used both diagnosis codes and ADRs report. This is because if a ADR is very clear and causal relationship is certain, the physician might not consult the investigation for ADRs to Drug Safety Center. On the other hand, in the case of mild/moderate Type B, it was difficult to confirm a causal relationship by the dianosis code with retrospective record review, so we only used the reports of the ICSR, which clearly reported a causal relationship (Table S2).

| Statistical analysis
Statistical analysis focused on the ADR events according to the risk drug prescription and HLA genotype status. Categorical variables were compared between the two groups, with and without HLA risk alleles. A 2 � 2 table was made for each prescribed drug to compare the frequencies of patients with HLA PGx alleles and ADR occurrence. The p-values, odds ratios (OR), and 95% confidence intervals (CI) were calculated using a two-tailed Fisher's exact test. A p-value less than 0.05 was considered statistically significant. Haldane's correction was used by adding one to all cells if a SCAR or type B ADR was not reported in patients without the HLA PGx alleles. 13 Haldane's correction was not used if a SCAR or type B ADR was not reported in patients with the HLA PGx alleles. All analyses were conducted using the R statistical software version 3.0.2. 14 3 | RESULTS

| Characteristics of human leukocyte antigentested patients and allele frequency
The clinical characteristics of 11,988 patients with HLA testing are summarized in Table 2 There were changes to the HLA testing methods over the 10-year study duration, and different testing methods were used depending on the transplanted organ. In total, 11,929 (99.5%) patients underwent HLA-B typing, whereas HLA-DQB1 typing was performed least with the frequency of 20.5% (Table 2).

| Combined results of human leukocyte antigen testing and prescription records
We investigated the status of the eight ADR-related HLA-alleles, seven risk drugs, and target ADRs reported in 11,988 HLA-tested patients, as shown in Figure 2. In total, 4092 patients (34.1%) had at least one of the risk HLA alleles. The HLA-B*58:01 allele had the highest frequency at 11.0% (Figure 2). In addition, 4583 patients (38.2%) were prescribed at least one of the seven risk drugs. The drug with the highest number of prescriptions was allopurinol (n = 2782, 23.2%).
Some discrepancies in the reports of adverse events were identified using SUPREME ® and the ICSR. To resolve the discrepancies, we regarded all cases of the reported drug-related side effects in either of the two databases (SUPREME ® and ICSR) as true. Additionally, when the severity of ADRs reported in the two databases did not match, we assumed that more severe events took place. Data of a total of 580 patients with ADRs related to the seven PGx drugs were evaluated. was not statistically significant. There were no cases with vancomycin-related SCARs among nine patients with the HLA-A*32:01 allele. Overall, the risk of developing SCARs due to allopurinol and carbamazepine use and type B ADRs due to allopurinol, carbamazepine, and oxcarbazepine use were significantly higher in patients with the risk alleles ( Figure 3). It is assumed that if patients with the risk alleles had not been prescribed the high-risk drugs, the following SCARs would have been prevented: 7/15 (46.7%) for allopurinol, 1/1 (100%) for carbamazepine, 1/2 (50%) for oxcarbazepine, and 2/2 (100%) for dapsone ( Our findings support the claim that PGx information should be integrated into EMRs using a clinical decision support system. 5,15 In case of organ transplantation patients who have already undergone tests for HLA genotyping, physicians could use this HLA PGx information to reduce ADR risks. Although our study did not find statistically significant differences in the rate of ADRs for some drugs, possibly owing to the small number of the study population who took them, the available HLA data could be useful in preventing ADRs.

| Hypersensitivity risk according to human leukocyte antigen alleles
In addition to the data on the HLA PGx alleles, other readily available HLA allele data can be used to diagnose and identify individuals at a risk of various autoimmune diseases. 16 The exact incidence of SCARs is unknown; however, the incidence of SJS/TEN is estimated at 1-2 cases per 1,000,000 people per year. 18 Although the incidence of SCARs is very low, it carries significant morbidity, and the mortality rates are 10% for SJS, 30% for SJS/TEN overlap, 50% for TEN, and 5% for DRESS. 18 Furthermore, SCARs may severely damage the affected mucosa or skin and leave permanent sequelae. Therefore, although the absolute risk reduction is small owing to its rare occurrence, the potential benefits of preventing SCARs are substantial. Considering patient data for the HLA PGx alleles is already available, it is reasonable to integrate this data and ensure its availability for clinicians at the point-of-care to help prevent SCARs. To achieve this, a clinical decision support system that instantly informs the personalized estimated risk to physicians by automatically linking the existing genetic information with the prescription should be implemented.
Pharmacogenomic studies showed that certain HLA genotypes induce T cell activation to a specific drug, resulting in the develop- proposed as an additional allele type associated with carbamazepineinduced SJS in Koreans. 23 In patients with carbamazepine-induced DRESS, the HLA-A*31:01 allele was reported as a risk marker in Europeans, Japanese, and Koreans. 23 Oxcarbazepine, a 10-keto analog of carbamazepine, has also been associated with the HLA-B This means that some results may not be accurate in some patients.
However, this is unlikely to result in significant error based on the allele frequency data. 28 For example, the HLA-B58 status was also shown to be strongly associated with allopurinol hypersensitivity and HLA-B*58:01 is the only HLA-B58 allele in the Korean population. 31 Second, the causal relationships in this study were not assessed by confirmative tests, and therefore, misdiagnosis and overestimation are possible. In our analysis, we assumed that the