SJS indicates Stevens-Johnson syndrome; TEN, toxic epidermal necrolysis.
Forest plot of subgroup analyses according to race/ethnicity. Width of the box indicates the precision of the estimates; diamond, the overall summary estimate for the analysis (width of the diamond represents the 95% CI).
Tangamornsuksan W, Chaiyakunapruk N, Somkrua R, Lohitnavy M, Tassaneeyakul W. Relationship Between the HLA-B*1502 Allele and Carbamazepine-Induced Stevens-Johnson Syndrome and Toxic Epidermal NecrolysisA Systematic Review and Meta-analysis. JAMA Dermatol. 2013;149(9):1025–1032. doi:10.1001/jamadermatol.2013.4114
The US Food and Drug Administration recommends screening for the HLA-B*1502 allele before initiation of carbamazepine therapy in patients of Asian ancestry, but there remains unclear evidence of a relationship between HLA-B*1502 and Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) among carbamazepine users, especially in some racial/ethnic populations.
To determine the relationship between the HLA-B*1502 allele and carbamazepine-induced SJS and TEN.
A comprehensive search of the following data sources was performed without language restriction from the inception of the database until January 8, 2013: EMBASE, PubMed, clinicaltrials.gov, Cochrane Library, IPA (International Pharmaceutical Abstracts), HuGENet (Human Genome Epidemiology Network), and CINAHL (Cumulative Index to Nursing and Allied Health Literature), and the reference lists of identified studies.
Inclusion criteria were studies that investigated the relationship between HLA-B*1502 and carbamazepine-induced SJS and TEN and that reported sufficient data for calculating the frequency of HLA-B*1502 carriers among cases and controls. The search yielded 525 articles, of which 16 met the inclusion criteria. The studies included 227 SJS or TEN cases, 602 matched control subjects, and 2949 population control subjects.
Data Extraction and Synthesis
Two reviewers independently extracted the following data: study design, eligibility criteria, diagnostic criteria, patient demographics, genotype distribution, HLA-B genotyping technique, selection of cases and controls, dosage of carbamazepine and duration of use, and results of Hardy-Weinberg equilibrium in the control group. The Newcastle-Ottawa Scale was used to assess the quality of studies. The overall odds ratios (ORs) with corresponding 95% CIs were calculated using a random-effects model. The primary analysis was based on matched control studies. Subgroup analyses by race/ethnicity were also performed.
Main Outcome and Measure
The primary outcome was carbamazepine-induced SJS and TEN. The outcome measure is given as an overall OR.
The summary OR for the relationship between HLA-B*1502 and carbamazepine-induced SJS and TEN was 79.84 (95% CI, 28.45-224.06). Racial/ethnic subgroup analyses yielded similar findings for Han-Chinese (115.32; 18.17-732.13), Thais (54.43; 16.28-181.96), and Malaysians (221.00; 3.85-12 694.65). Among individuals of white or Japanese race/ethnicity, no patients with SJS or TEN were carriers of the HLA-B*1502 allele.
Conclusions and Relevance
We found a strong relationship between the HLA-B*1502 allele and carbamazepine-induced SJS and TEN in Han-Chinese, Thai, and Malaysian populations. HLA-B*1502 screening in patients requiring carbamazepine therapy is warranted.
Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are life-threatening conditions affecting the skin and mucous membranes.1 Both SJS and TEN are characterized by blisters arising on purple macules. Lesions are widespread and usually predominate on the trunk of the body. The percentage of skin detachment body surface area differentiates SJS and TEN.2 Although the incidence of SJS and TEN is low (0.4-6 cases per million persons per year),1,3 mortality rates are as high as 5% to 12.5% for SJS, 33.3% for SJS and TEN overlap, and 30% for TEN.1,4
The primary identifiable causative agent of SJS and TEN is medication use (almost 80% of cases).1 Other factors implicated include chemicals, viral infection, immunizations, and mycoplasma pneumonia. Several classes of medication have been strongly associated with SJS and TEN, including antibiotics, antiepileptic drugs, and xanthine oxidase inhibitors.1 Carbamazepine is the most common cause of SJS or TEN.5,6 A European study7 found that the risks of SJS or TEN among carbamazepine and phenytoin users were more than 90 and 53 times greater than those of nonusers, respectively.
Multiple investigators have reported a relationship between HLA genes and SJS and TEN induced by carbamazepine. Specifically, the HLA-B*1502 allele (OMIM 142830.0002) has been associated with carbamazepine-induced SJS and TEN.8- 11 In addition to carbamazepine, allopurinol can be a causative agent of SJS and TEN, particularly among those carrying the HLA-B*5801 allele.12 Based on these observational studies, the US Food and Drug Administration published an alert to health care professionals that severe allergic skin reactions can be caused by carbamazepine use in patients with the HLA-B*1502 allele and recommended screening for the allele before initiation of carbamazepine therapy in patients of Asian ancestry.13
Despite the alert from the US Food and Drug Administration, some studies9,11,14- 16 have demonstrated no such association in the Japanese population. This difference likely reflects race/ethnicity variation, as well as nonreporting of less severe cutaneous drug reactions.9,11,14- 16 To fully understand the relationship between HLA-B*1502 and SJS and TEN among carbamazepine users, a systematic review identifying all studies assessing such association is needed. A transparent and reproducible process may be provided using a systematic review approach. This method can elucidate potential similarities and differences that may explain the findings. Therefore, this study was undertaken to systematically review all relevant studies and to quantitatively synthesize the magnitude of the relationship between the HLA-B*1502 allele and carbamazepine-induced SJS and TEN.
We searched EMBASE, PubMed, clinicaltrials.gov, The Cochrane Library, IPA (International Pharmaceutical Abstracts), HuGENet (Human Genome Epidemiology Network), and CINAHL (Cumulative Index to Nursing and Allied Health Literature). All databases were searched from their inception until January 8, 2013. Searches were performed using keywords and synonyms for HLA-B and carbamazepine and relevant terms for SJS and TEN. For PubMed and The Cochrane Library, Medical Subject Headings were searched for HLA-B and carbamazepine and for SJS and TEN. There was no language or study design restriction, but only human studies were included. Additional studies were retrieved from the reference lists of the selected articles.
Two reviewers (W. Tangamornsuksan and R.S.) independently assessed abstracts and titles retrieved from the comprehensive searches for study inclusion. The inclusion criteria were studies that (1) investigated the relationship between HLA-B*1502 and carbamazepine-induced SJS and TEN and (2) reported sufficient data for calculating the frequency of HLA-B*1502 carriers among cases and controls. Any disagreements were discussed until consensus between the 2 reviewers could be reached.
All articles were extracted independently by the reviewers; discrepancies were resolved by discussion. The following data were extracted from each study: study design, eligibility criteria, diagnostic criteria, patient demographics, genotype distribution, HLA-B genotyping technique, selection of cases and controls, dosage of carbamazepine and duration of use, and results of Hardy-Weinberg equilibrium (a state of equilibrium under the mendelian law of genetic inheritance) in the control group. Hardy-Weinberg equilibrium was tested to check if the included individuals were in equilibrium for the frequencies of genotypes.17,18 Equilibrium implies that the included individuals are likely representative of the population.19,20 We used the Newcastle-Ottawa Scale21 to assess the quality of the studies included in the review. This scale is an 8-item instrument, categorized into the following 3 domains: selection of participants, comparability between 2 groups, and the assessment of exposures and outcomes. A system of stars is used to provide quality ratings for studies.
The overall odds ratios (ORs) with corresponding 95% CIs were calculated to determine the relationship between the presence of HLA-B*1502 in at least 1 allele and carbamazepine-induced SJS and TEN. All analyses were performed with the method by DerSimonian and Laird22 using a random-effects model. The analyses were also performed separately on studies using different types of control groups (eg, control subjects obtained from the study [matched control study] or control subjects obtained from the population database [population study]). The primary outcome measure was analyzed using the matched control study. Subgroup analyses by race/ethnicity were also performed to determine the robustness of the findings. Statistical heterogeneity was assessed via the Q statistic and I2 tests.23P ≤ .10 indicated heterogeneity between studies. I2 values of 25% and 50% denoted low heterogeneity and moderate heterogeneity, respectively, across studies.24 Funnel plot, Begg test, and Egger test were used to evaluate publication bias.25,26
In total, 525 articles were identified by the search. Sixteen articles14,16,27- 40 met the inclusion criteria. Some SJS and TEN cases included in the studies by Chung et al,9 Ikeda et al,15 Wang et al,41and Kulkantrakorn et al42 were also in the studies by Hung et al,28 Kaniwa et al,29 Shi et al,36 and Tassaneeyakul et al,37 respectively. In addition, 1 study43 and 2 abstracts44,45 potentially met the inclusion criteria but lacked key information. Of the original 525 articles, 16 remaining studies were included in the meta-analysis (Figure 1). No additional articles were identified via a review of the bibliographies of the included studies.
Characteristics of the included studies are summarized in Table 1. The studies included 227 SJS or TEN cases14,16,27- 40; in addition, there were 602 matched control subjects14,28,30- 32,35- 40 and 2949 population control subjects.16,27,29,30,33,34 Eleven studies14,28,30- 32,35- 40 were conducted using matched controls, while 6 studies16,27,29,30,33,34 used the population as controls. Fifteen studies were conducted among Asian populations, including 1 study30 in Koreans, 1 study34 in Indians, 2 studies32,37 in Thais, 2 studies27,38 in Malaysians, 3 studies16,29,35 in Japanese, and 6 studies28,31,36,39- 41 in Han-Chinese. Two studies14,34 examined individuals of white race/ethnicity, while 3 studies27,31,33 evaluated multiethnic populations (Table 1).
Among studies having matched controls, the mean ages of included patients were 54 years28,30,32,35- 40and 47 years28,30,36- 40 in cases and controls, respectively; 42.6% (75 of 176) of cases28,30,32,35- 40 and 47.0% (206 of 438) of controls28,30,36- 38,40 were male. Among studies29,30,33,34 using the population as controls, the mean age of included patients was 51 years in cases; 40.4% (21 of 52) of cases were male,29,30,33,34 while age and sex data were not reported for controls.
Four studies28,30,36,39 and 2 studies30,36 reported the mean carbamazepine dosages in the case and control groups, which were 559 mg/d and 514 mg/d, respectively. One population control study30 reported this information, which was 243 mg/d for the case group.
Most studies (8 of 11 [72.7%]) using matched controls specified the use of criteria by Roujeau and Stern1 and by Roujeau et al7 as SJS and TEN diagnostic criteria for cases, while 4 studies14,16,31,38 provided no information on the criteria used. Among studies using population controls, 3 studies27,33,38 reported use of the criteria by Roujeau and colleagues.1,7 In all studies with matched controls, patients without SJS and TEN who had used carbamazepine for at least 3 to 6 months were included in the control group, while patients who developed SJS and TEN within 2 to 12 weeks after the start of carbamazepine therapy were included in the case group.
Only 2 studies28,33 reported Hardy-Weinberg equilibrium, and these confirmed no departure from equilibrium. Characteristics of the alleles (heterozygous vs homozygous) in cases and controls were reported in 2 studies.32,34 All 6 patients in the study by Mehta et al34 had a heterozygous HLA-B*1502 allele, while 20 of 21 patients in the study by Locharernkul et al32 had a heterozygous allele. Eleven studies16,27,30- 34,36- 38,40 identified HLA-B*1502 using polymerase chain reaction sequence-specific primers, 3 studies28,29,39 used polymerase chain reaction sequence-based typing, and 2 studies14,32 used polymerase chain reaction sequence-specific oligonucleotides. None of the included studies described a blinding procedure for personnel performing the genotyping.
The methodological quality of all studies was summarized as a mean Newcastle-Ottawa Scale score of 5 (range, 4-6; maximum score, 9). These results are given in Table 1.
Eleven studies16,27,30- 34,36- 38,40 were included in a comparison of the HLA-B*1502 carrier frequencies in the cases and carbamazepine-tolerant controls. There were 175 SJS or TEN cases and 602 matched controls. Of the controls, 53 were carriers of the allele, while a much larger proportion of the cases (152 of 175) carried the allele. We found a clear relationship between HLA-B*1502 and carbamazepine-induced SJS and TEN (summary OR, 79.84; 95% CI, 28.45-224.06). Heterogeneity was minimal (I2 = 45.1%) and was not significant by the Q statistic (P = .07). There was no apparent publication bias by Begg test (P = .92), Egger test (P = .23), or funnel plot.
Six studies16,27,29,30,33,34 with 52 SJS or TEN cases and 2949 population controls were included in a separate analysis of carrier frequency to test the relationship between the HLA-B*1502 genotype and carbamazepine-induced SJS and TEN compared with the general population. Cumulative carrier frequencies for these 6 studies were 23 of 52 for cases and 51 of 2949 for population controls. The relationship between the HLA-B*1502 genotype and carbamazepine-induced SJS and TEN compared with the general population was statistically significant (summary OR, 57.56; 95% CI, 12.50-265.09). Heterogeneity was present (I2 = 53.1%, P = .007). There was no apparent publication bias by Begg test (P = .31) or Egger test (P = .81) and with a funnel plot.
Subgroup analysis in the selected matched control studies14,28,30- 32,35- 40 according to race/ethnicity yielded similar and significant findings. Carbamazepine-induced SJS or TEN was associated with the HLA-B*1502 allele in Han-Chinese,28,31,36,39,40 Thai,32,37 Korean,30 and Malaysian38 patients. The summary ORs for Han-Chinese, Thai, Korean, and Malaysian populations are given in Table 2 and in Figure 2. The summary OR for Han-Chinese was 115.32 (95% CI, 18.17-732.13), with I2 heterogeneity of 71.3%. Such heterogeneity probably arose from a nonhomogeneous patient population from the north and south of China.47 This was confirmed by a reanalysis that excluded this study and yielded similar results (OR, 113.65; 95% CI, 44.20-292.22) but with much less heterogeneity (I2 = 14.4%). In the sole matched control study among individuals of white14 and Japanese35 race/ethnicity, no patients carried the HLA-B*1502 allele.
Our results indicate that the HLA-B*1502 allele is strongly associated with increased risk of developing SJS and TEN in patients using carbamazepine, especially Asians. This severe adverse event could be prevented if genetic information were known a priori. Physicians and policymakers should consider our findings to support the implementation of genetic testing before initiation of carbamazepine therapy.
Data reveal a high risk of developing SJS and TEN among carbamazepine users with HLA-B*1502. For Han-Chinese, Thais, Koreans, and Malaysians, the risks of SJS and TEN among carbamazepine users carrying the HLA-B*1502 allele are approximately 115-fold, 60-fold, 25-fold, and 220-fold increases, respectively.
These observed racial/ethnic differences could be explained by the fact that HLA-B*1502 may not be the only gene associated with SJS and TEN in patients using carbamazepine. For example, another HLA allele (HLA-A*3101) is related to SJS and TEN; the ORs for carriers of this gene using carbamazepine are approximately 26-fold higher for Europeans48 and 11-fold higher for Japanese.49 Furthermore, Hung et al28 reported that HLA-A*3101 was associated with carbamazepine-induced maculopapular eruptions (OR, 17.5; 95% CI, 4.6-66.5) and hypersensitivity syndrome (OR, 7.1; 95% CI, 3.1-16.5) in a Han-Chinese population. Similar to the observed relationship between HLA-B*1502 and SJS and TEN among carbamazepine users, there are some examples of associations in which more than 1 gene has a major role.50,51
Although the mechanism by which HLA-B*1502 contributes to SJS and TEN among carbamazepine users is not fully understood, it is well recognized that HLA-B molecules supply endogenous or processed exogenous antigens to T cells, eliciting an adaptive immune response. It seems that a specific HLA allele is required for the activation of drug-specific T cells by the culprit drug. The T-cell receptor of the effector T cell is thought to recognize the drug-peptide complex bound by the specific HLA-B molecule on the antigen-presenting cell, resulting in the release of immune mediators and leading to robust adaptive immune reactions in severe cutaneous reactions.52- 54
Our systematic meta-analysis has several limitations. First, this study focused only on the relationship between the HLA-B*1502 allele and carbamazepine-induced SJS and TEN. Other genes that may be associated with SJS and TEN (such as HLA-A*3101) were excluded from our study. Second, an article43 and 2 abstracts44,45 that potentially met our study criteria were not included in the meta-analysis because they did not provide the number of control patients, which was important information for the analysis. Third, only 2 studies28,33 among 16 articles provided Hardy-Weinberg equilibrium information, preventing us from assessing the representativeness of the controls.
Our study reveals that HLA-B*1502 is strongly associated with the development of SJS and TEN. The prevalence of HLA-B*1502 among residents of southeast Asian countries was reported to be 8%.9,55 This is against the background of a high incidence of SJS and TEN among this population (8 cases per 1 million person-years).56,57 In Thailand, Singapore, and Malaysia, carbamazepine use represents up to one-third of all SJS and TEN cases.56- 58 The relationship between the HLA-B*1502 allele and carbamazepine-induced SJS and TEN was strong in Thai (OR, 54.43) and Malaysian (OR, 221.00) populations in our study, reflecting the high disease risk associated with HLA-B*1502 carrier status among patients of Asian ancestry. Because carbamazepine use for epilepsy is normally a long-term therapy, the safety requirements are high. It is important to consider genetic screening among populations at high risk, such as those from southeast Asian countries. However, there is a strong need to consider other factors such as feasibility, ethical and legal issues, and cost-effectiveness before implementation of such genetic screening.
In conclusion, we found a strong relationship between HLA-B*1502 and carbamazepine-induced SJS and TEN in Han-Chinese, Thai, and Malaysian populations. Recognition of HLA-B allele status before initiation of the drug may be beneficial to some groups of patients. Such information will assist physicians in determining the optimal drug therapy.
Corresponding Author: Nathorn Chaiyakunapruk, PharmD, PhD, Discipline of Pharmacy, Monash University Sunway Campus, Jalan Lagoon Selatan, 46150 Bandar Sunway, Selangor, Malaysia (firstname.lastname@example.org).
Accepted for Publication: April 26, 2013.
Published Online: July 24, 2013. doi:10.1001/jamadermatol.2013.4114.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Tangamornsuksan, Chaiyakunapruk, Somkrua, Lohitnavy.
Acquisition of data: Tangamornsuksan, Tassaneeyakul.
Analysis and interpretation of data: Tangamornsuksan, Chaiyakunapruk, Tassaneeyakul.
Drafting the manuscript: Tangamornsuksan, Chaiyakunapruk, Lohitnavy.
Critical revision of the manuscript for important intellectual content: All authors.
Conflict of Interest Disclosures: Dr Tassaneeyakul was supported by the Higher Education Research Promotion and National Research University Project of Thailand through the Center of Excellence in Specific Health Problems in the Greater Mekong Subregion cluster, Khon Kaen University, Kohn Kaen, Thailand.
Funding/Support: This study was supported by grants from the National Science and Technology Development Agency, Thailand.
Additional Contributions: Surasak Saokaew, PharmD, Norman Scholfield, PhD, and Piyameth Dilokthornsakul, PharmD, assisted in the preparation of the manuscript.