Introduction

Steven–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN) and drug rash with eosinophilia and systemic symptoms (DRESS) are considered Severe Cutaneous Adverse Reactions (SCARs). SJS and TEN are characterized by destruction of the epidermis and mucosal epithelium. These conditions often presented with internal organs involvement. Mortality rates were 1–5% for SJS and 20–30% for TEN.1, 2, 3 Moreover, DRESS presents as a group of symptoms including extensive mucocutaneous rash, fever, lymphadenopathy, hepatitis, eosinophilic infiltration and multiple organ damages. Onset usually begins after 2 weeks and may occur at any time within 3 months. Incidence of DRESS is 1 case in 1000 to 10 000 drug exposures, whereas incidence of SJS and TENs is 1.2 to 6 cases per million person-years approximately.4, 5

Pharmacogenetic study is becoming a key component for SCARs prevention. Moreover, aromatic antiepileptic drugs (AEDs), such as carbamazepine, phenobarbital and phenytoin, are the most common causes of SCARs.5 In 2007, the US Food Drug Administration (USFDA) recommended to perform genetic screening on HLA-B*15:02, which has been strongly associated with serious skin reactions, in ancestry across broad areas of Asians prior to starting carbamazepine treatment. The patients who carry HLA-B*15:02 should not use this medicine unless benefits are outweigh. As aromatic AEDs share similar aromatic structure, cross-reactivity among aromatic AEDs were regularly reported.6 Therefore, USFDA states that phenytoin should be avoided as an alternative for carbamazepine in patients who are positive for HLA-B*15:02 because of the increased risk of SJS/TEN in patients of Asian ancestry.7 Recent study of HLA-B*15:02 provided evidence of association between phenytoin and SJS in Thai epilepsy patients (odds ratio (OR)=18.5, 95%CI 1.82–188.40) although it is less strong than the similar evidence of carbamazepine-induced SCARs in Han Chinese (OR=2504, 95% CI=126–49 522).8, 9 In addition, Thai drug label indicates not to use carbamazepine in patients who carry HLA-B*15:02 because there are strong evidences in developing SJS/TEN in such patients. In case of phenytoin, limited evidences suggest that phenytoin may be a risk factor for SJS/TEN, however, consideration to avoid using this drug is given in HLA-B*15:02 positive patients. These Thai label warnings are as same as contents in the US package insert.

The recommendation to screen HLA-B*15:02 in phenytoin-treated patient before starting treatment has been recently proposed in Clinical Pharmacogenetics Implementation Consortium (CPIC) Guidelines for CYP2C9 and HLA-B Genotype and Phenytoin Dosing.7

Genetic polymorphisms also contribute to variability of cytochrome P450 (CYP) activities. Interestingly, CYP enzymes, which are the keystones in metabolism of AEDs, are recently suggested as a determinant factor for SCARs induced by phenytoin in addition to HLA-B*15:02.10 Both CYP2C9 and CYP2C19 are substantial enzymes in phenobarbital and phenytoin metabolism.11, 12 These certain enzymes transform phenobarbital and phenytoin to inactive substances, p-hydroxyphenobarbital and p-hydroxyphenytoin, respectively.13, 14 Thus, loss of function of these enzymes owing to genetic polymorphisms may result in delay clearance and excess toxic metabolites accumulation.10 The association of CYP2C9*3 variant was formerly reported in Korean that done on phenytoin-treated neurological patients.15 Later, CYP2C9*3 was also determined from Taiwanese genome-wide association study, which was followed by direct sequencing of the associated loci in phenytoin-treated epilepsy patient.10 Though CYP2C9*3 association study was done in epilepsy Korean and Taiwanese patients, it has never been investigated in Thais.

The variability of hepatic CYP2Cs expression and their catalytic activity may lead to the skin reaction development in children population. The CYP2C9 activity of children reaches adult level between 5 months to 2 years old.7, 16 Therefore, loss of function CYP2C9 might cause more serious problem to this special population. The goal for this research was to examine the association of CYP2C9*3 and phenytoin-induced SCARs in Thai epilepsy children.

Materials and methods

DNA samples

DNA samples were obtained from Faculty of Medicine, Ramathibodi Hospital.17 A total 72 DNA samples from epileptic patients were examined in this study. These samples were divided into four groups. Seventeen subjects were in phenytoin-induced SCARs group. All of cases were 0–18 years old with SCARs diagnosed, composing of SJS, TEN and DRESS, from phenytoin after take drugs within 12 weeks. Sixteen subjects who tolerated phenytoin SCARs after taking phenytoin at least 12 weeks were allocated to phenytoin-tolerant group. Another two groups included phenobarbital-tolerant group (19 subjects) and phenobarbital-induced SCARs (20 subjects). Major characteristic of SJS and TEN were skin detachment and mucosal erosion: skin separation <10% of body surface area in SJS; and >30% of body surface area in TEN.18 The diagnostic criteria for DRESS were scored from these symptoms, fever >38.5 °C, lymph nodes enlargement, eosinophilia, atypical lymphocyte, skin rash >50% of body surface area and organ involvement (liver, kidney, lung, pancreas and so on).19 The protocol was approved by the Human Rights And Ethic Committee of Faculty of Medicine, Ramithibodi Hospital, Mahidol University. Written informed consents were obtained from each participant and their parents.

Primer design and validation

Two pairs of primers were designed for allele-specific PCR (AS-PCR), one pair for CYP2C9*3 allele detection and the other pair for non CYP2C9*3 allele detection. All designed primers were provided as follow: forward primer 1 (5′-TGC ACG AGG TCC AGA GAT ACA-3′), reverse primer 1 (5′-TAC AAA CCT TTA TAG CCC CAA AC-3′); forward 2 (5′-TGA ACG TGT GAT TGG CAG AAA C-3′); and reverse 2 (5′-CTG GTG GGG AGA AGG TCA AG-3′). PCR products from two reactions were difference in length. A forward 1 and a reverse 1 primer were designed to detect a non CYP2C9*3 allele, and they amplified 263 base pairs of DNA. Conversely, using a forward 2 primer and a reverse 2 primer to detect CYP2C9*3 allele produced 114 base pairs of DNA.

Analysis of primers was based on electronic database to prevent the formation of primer–dimers, hairpin and inappropriate melting temperature. First, the general properties of primers were examined by going to site on the sequence manipulation suite.20 Afterward, the investigation of target DNA specificity was performed by blasting novel primers with human genome sequence on nucleotide blast database.21 In order to validate designed primers, the known CYP2C9*3 carriers and the CYP2C9*3 noncarriers were used as positive and negative controls.

Allele-specific PCR and genotype interpretation

AS-PCR method and size determination by gel electrophoresis were selected for genotyping assay. PCR amplification was performed by using T100 Thermal Cycler (Biorad, Hercules, CA, USA). The compositions of PCR reaction were 20 ng of DNA templates and 1 × of KAPA2G Fast multiple PCR Kit (Kapa Biosystems, Wilmington, MA, USA), 0.2 μM of each primer. KAPA2G contains Hot start DNA polymerase (1 U per 25 μl reaction), KAPA2G Buffer A (1.5 × at 1 ×), dNTPs (0.2 mM each dNTP at 1X), MgCl2 (3.0 mM at 1X) and stabilizers. The condition for each primer was optimized by gradient temperature in accordance with primer profiles. PCR conditions were listed as following: an initial denaturation 95 °C for 3 min, 30 cycles of amplification [95 °C for 15 S, 68 °C for 30 S, 72 °C for 30 S], a final extension 72 °C for 1 min. Size of PCR product was determined in gel electrophoresis. 5 μl of each PCR products were analyzed in 2% agarose gel with 0.5X TBE buffer followed by ethidium bromide staining. The genotype data were interpreted from the length of PCR product which was compared with 100 bps ladder (Invitrogen, Waltham, CA, USA). The presence of PCR band was observed under UV light with gel documentation (Unidoc, UK). DNA samples which were detected both CYP2C9*3 allele and non CYP2C9*3 allele were interpreted as CYP2C9*3 heterozygous genotype. If DNA samples were identified only CYP2C9*3 allele or non CYP2C9*3 allele in both reaction, they were considered as homozygous genotype of that specific allele (shown in Figure 1).

Figure 1
figure 1

Experimental design for CYP2C9*3 genotyping. (a) Model of expected PCR products from designed primer. The presence of PCR product from forward 1(F1) and reverse 1 (R1) primers (263 bps) demonstrated non CYP2C9*3 allele. The detection of DNA (114 bps) from forward 2 (F2) and reverse 2 (R2) primers indicated CYP2C9*3 allele. The left panel pattern determined CYP2C9*3 non carrier. The pattern of CYP2C9*3 carrier were shown in middle panel (heterozygous) and right panel. (b) Examples of PCR band from the experiment. Sample A was interpreted as CYP2C9*3 non carrier, and sample B was CYP2C9*3 carrier (heterozygous).

Full size image

Statistical analysis

The protocol was designed with 80% power to detect a significant difference (P-value=0.05, two sided). Statistical analysis was performed by SPSS software, version 16.0 (SPSS Inc., Chicago, IL, USA). Demographic data (continuous variable) was analyzed and presented as mean± s.d., median or frequency. The normality of continuous data were tested by Kolmogorov–Smirnov Test. In order to compare the difference among continuous variables, student t-test (normal distribution) or Mann–Whitney U-test (if the data are not normal distribution) was used for analyses. The results of study were presented as frequencies, P-value, OR and 95% CI. In order to calculate OR, Haldane’s modification, adds 0.5 to all cells, is applied for all variables when there was an absence in genotype frequency.

χ2-test was used for categorical variables, such as gender. Fisher’s exact test was performed to find association between genotypes frequencies and the incident of SCARs. The Fisher’s exact significant level, 95% CI and ORs, were computed in R console 3.1.1 statistical software for window (http://cran.r-project.org/) using ‘Exact 2X2’ package.22 P-value<0.05 is considered as statistical significance. The exact test for Hardy–Weinberg disequilibrium was performed to exclude the genotype error that might distort the proportion of heterozygotes and homozygotes with significant departure from Hardy–Weinberg equilibrium.

The associations of CYP2C9 in AEDs-induced SCARs patients were compared with drug-tolerant patients and Thai healthy adult population from published data.23

Results

Patient characteristics

Among 72 patients, there were 17 patients of phenytoin-induced SCARs, 16 of phenytoin tolerances, 20 subjects of phenobarbital-induced SCARs and 19 of phenobarbital tolerances. The median ages and dosages were summarized in Table 1.

Table 1 Demographic data of phenytoin-treated, phenobarbital-treated patients
Full size table

The clinical characteristics of SCARs patients were presented in Table 2. Majority of phenytoin cases (88%) exhibited DRESS symptoms, whereas 12% of cases were diagnosed as SJS–TEN. Ninety percentages of phenobarbital cases (18 subjects) were diagnosed as DRESS but SJS–TEN were diagnosed in two subjects (10%). Liver enzymes from SCARs patients were slightly higher than normal range in both drug-induced SCARs groups because the majority of cases in this study belong to DRESS, in which hepatitis is a common characteristic. These SCARs symptoms from both agents prolonged hospitalization for at least 1 week.

Table 2 Clinical characteristic of SCARs patients
Full size table

Genotyping

The examined alleles did not differ from Hardy–Weinberg equilibrium in both SCARs group and control group (P-value>0.05). The data of healthy samples from published study were included in this analysis.23 The attribution of these samples was summarized below. There were 326 Thai healthy samples. The age range was between 26 and 62 years old (mean 39±15 years old).

Five cases from 33 phenytoin-treated individuals were CYP2C9*3 carriers (heterozygous). None of this allele was identified in tolerant subjects. All CYP2C9*3 carriers were heterozygous and they belong to phenytoin-induced SCARs group (Table 3). The CYP2C9*3 carrier rate was 29.4% in phenytoin-induced SCARs patients. CYP2C9*3 significantly associated with phenytoin-induced SCARs (OR=14.52; 95% CI=1.1754—infinity, P-value exact=0.044; Table 3). Genotype frequency of CYP2C9*3 in phenobarbital-induced SCARs versus tolerant group were 5 and 15.8%. However, phenobarbital-induced SCARs group was not associated with CYP2C9*3. In addition, the association between phenytoin-induced SCARs and CYP2C9*3 allele were confirmed when compared with published genotyping data of CYP2C9*3 in healthy adult Thais23 (OR=4.43; 95% CI=1.39–13.97, P-value exact=0.016) (Table 3).

Table 3 Association testing among phenytoin-induced SCARs patients, tolerant patients and healthy adult Thais
Full size table

Another analysis is determination of the additive effects from CYP2C9*3 and HLA-B*15:02, two reported alleles associated with phenytoin-induced SCARs group. Some patients carries either CYP2C9*3 or HLA-B*15:02 or both of them. As association between HLA-B*15:02 and phenytoin-induced SCARs was absence in this study, combined information from HLA-B*15:02 and CYP2C9*3 genotypes demonstrated significant association with phenytoin-induced SCARs (OR=10.5; 95% CI=1.22–247.93, P#-value=0.039) but less association than CYP2C9*3 alone.

Discussion

Both CYP2C9 and CYP2C19 enzymes are involved with metabolizing pathway of phenytoin. Previous pharmacogenetic study with CYP2C19*2, a loss of function allele of CYP2C19 and phenytoin-induced SCARs has failed to prove their association. Therefore, CYP2C9*3, an allele that encodes decrease activity enzyme, is another candidate target to examine this association. CYP2C9 is located on chromosome 10, juxtaposed by CYP2C19. CYP2C9*3 is nonsynonymous exchange of isoleucine to leucine at position 359.24 Both homozygous and heterozygous CYP2C9*3 reduce metabolic clearance of phenytoin.24 From current analysis, significant associations were identified between phenytoin-induced SCARs and CYP2C9*3 carriers, in particular heterozygous because the homozygous for CYP2C9*3 (*3/*3) was rare in Thais and not present in this study.23 The CYP2C9*3 carriers has 14.5 times higher OR for development of SCARs from phenytoin over those CYP2C9*3 noncarriers and significant difference of genotype distribution when compared with Thai healthy adults. These associations demonstrated the same trend as data from Taiwanese.10 Taiwanese data also suggested that CYP2C9*3 carriers had higher phenytoin plasma level than that of noncarriers. Moreover, their meta-analysis study showed significant association between CYP2C9*3 and SCARs from phenytoin in Japan, Malaysia and Taiwan.10 Our supplied evidence in Thais can confirm the association of CYP2C9*3 and phenytoin-induced SCARs in Southeast Asia region.

The exact mechanism explaining why reduced function of CYP2C9 contributed to phenytoin-induced SCARs was not known at the time of this study. To the best of our knowledge, intermediate toxic metabolites might be accumulated more owing to slower clearance in combination with nonlinear pharmacokinetic properties of phenytoin. These metabolites bind to cellular macromolecules resulted in hapten formation and stimulated immunological reactions.6, 25

HLA-B*15:02 was suggested as predictive marker for SCARS from phenytoin albeit with inconsistencies when compared with carbamazepine.

Negative result of HLA-B*15:02 was demonstrated in this study.17 The screening for HLA-B*15:02 or CYP2C9*3 in our study exhibited the superior results compared with that for HLA-B*15:02 alone but of those are inferior compared with CYP2C9*3 alone.

This study carried some limitations. First, sample sizes were small because SCARs were not common adverse reaction. Previous Korean study provided 10 cases of phenytoin-induced cutaneous adverse reactions, 3 out of 10 were heterozygous of CYP2C9*3.15 Second, CYP2C9*3 has lower clearance activity of enzyme, but we did not measure phenytoin plasma level of SCARs patients in this study. These DNA samples came from retrospective study, which almost all patients had recovered from those reactions, and rechallenging of phenytoin in SCARs are prohibited owing to patient safety. Therefore, it is not feasible to measure phenytoin concentration in their plasma. In addition, the case and control populations were not perfectly matched due to difficulty in enrolled control group resulted in significant difference in demographic data except gender. In pediatric practice in Thailand, phenobarbital was frequently prescribed among the first-line drug for epilepsy in children owing to its lowest cost.

Further study may need to clarify these limitations. However, given the association in other Asian populations, this association analysis is providing important information for the association of these pharmacogenetic markers in Thai population. The precise determination of these allele effect sizes will require a larger sample sizes in a well-matched cases–control study or prospective study; it is unlikely that this association is happened by spurious association. As these markers are not explaining all phenytoin-induced SCAR, other enzymes in phenytoin metabolizing pathway should also be investigated in next study.

This study is the first report demonstrated the association of CYP2C9*3 to phenytoin-induced severe cutaneous drug reaction in Thai epileptic young patients. CYP2C9*3 is a reasonable predictive genetic marker of phenytoin-induced SCARs to identify the risk of phenytoin-induced SCARs in Asian populations.