Epilepsy Research 135 (2017) 19–28

Contents lists available at ScienceDirect

Epilepsy Research
journal homepage: www.elsevier.com/locate/epilepsyres

Association between the HLA-B alleles and carbamazepine-induced SJS/ MARK

TEN: A meta-analysis
⁎ ⁎
Qianyu Wanga, Shusen Sunb, Meng Xiea, Kun Zhaoa,b, Xingang Lia, , Zhigang Zhaoa,
a
Department of Pharmacy, Beijing Tiantan Hospital, Capital Medical University, Beijing, China
b
College of Pharmacy, Western New England University, Springfield, MA, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Purpose: From our current understanding, the association between the human leukocyte antigen (HLA), HLA-
Carbamazepine B*1502, and carbamazepine(CBZ)-induced Stevens-Jonson syndrome and toxic epidermal necrolysis (SJS/TEN)
Stevens-Johnson syndrome in the Asian population is quite clear. However the relationship between other HLA-B alleles and CBZ-induced
Toxic epidermal necrolysis severe cutaneous adverse drug reactions (SCADRs) remains unclear. We aimed to identify other non-HLA-
Human leukocyte antigen
B*1502 alleles in patients with CBZ-induced SCADRs through a meta-analysis.
Meta-analysis
Materials and methods: A thorough literature search was performed using Embase, PubMed, Web of Knowledge
and Cochrane databases. A meta-analysis was performed from their inceptions to May 31, 2016. Studies in-
vestigating the association of HLA-B alleles and CBZ-induced SJS/TEN were retrieved. Two reviewers in-
dependently extracted the data. Overall odds ratios (ORs) with corresponding 95% confidence intervals (CIs)
were calculated using the RevMan 5.3 software.
Results: A total of 11 studies met the inclusion criteria, totaling 343 CBZ-induced SJS/TEN cases, 838 CBZ
tolerant controls, and 978 population controls. We observed HLA-B*1511 as a risk marker, and HLA-B*4001 and
HLA-B*4601 as protective markers for the development of SJS/TEN in patients taking CBZ. SJS/TEN cases were
found to be significantly associated with HLA-B*1511 in both the tolerant group
(OR = 17.43;95%CI = 3.12–97.41;P = 0.001) and the population-control group (OR = 11.11;
95%CI = 2.62–47.09; P = 0.001). The sensitivity analysis found that HLA-B*5801 was a protective marker in
the Southeast Asian population (OR = 0.23; 95%CI = 0.09-0.58; P = 0.002).
Conclusion: Our study demonstrated that in the Asian population, HLA-B*4001, HLA-B*4601, HLA-B*5801 were
strong protective factors in the development of CBZ-induced SJS/TEN whereas HLA-B*1511 was a risk factor.
While more studies may be needed in order to confirm these findings, consideration should be taken into testing
Asian patients for at-risk alleles prior to CBZ therapy initiation.

1. Introduction drug-induced SJS/TEN still remains unclear, the pathogenesis of the

Carbamazepine (CBZ) is a widely prescribed medication, not only
effective as a first-line antiepileptic drug, but as an effective treatment
in trigeminal neuralgia and bipolar disorder (Wiffen et al., 2011).
However, about 10% of patients taking CBZ will develop cutaneous
adverse drug reactions (cADRs) (Marson et al., 2007). Reactions can
range from mild cADRs to severe cADRs (SCADRs). Mild cADRs such as
mild maculopapular exanthema (MPE) are usually self-limited, whereas
severe cADRs could be life-threatening, like Stevens-Johnson syndrome
(SJS)/Toxic Epidermal Necrolysis (TEN). Although the incidence of
SJS/TEN is extremely low (Chan et al., 1990), the mortality rate is up to
30% (Svensson et al., 2001). CBZ is the main causal drug that leads to
the SJS/TEN (Roujeau and Stern, 1994). Although the mechanism of
life-threatening cutaneous ADRs is believed to be immune mediated.
The current hypothesis is that the human leukocyte antigen (HLA) is
involved in SJS/TEN by its capacity to present CBZ to T-cells and in-
itiate an immune response to CBZ (Chung et al., 2008). Protein mole-
cules such as cellular drug metabolizing enzymes and T-cell receptors,
as well as major histocompatibility complex (MHC), can contribute to
the ensuing immunological responses and hapten formation (Yun et al.,
2012). SJS/TEN is considered to be a single disease called a “bullous”
cADR, while MPE is a “non-bullous” cADR. Historically, the patho-
genesis of CBZ-induced SJS/TEN and MPE were thought to be different.
Therefore, researchers usually discuss the relationship between SJS/
TEN and MPE with gene alleles separately (Yang et al., 2015).
Many researchers have reported the association between human

⁎
Corresponding authors.
E-mail addresses: lxg198320022003@163.com (X. Li), 1022zzg@sina.com (Z. Zhao).

http://dx.doi.org/10.1016/j.eplepsyres.2017.05.015
Received 23 December 2016; Received in revised form 15 May 2017; Accepted 27 May 2017
Available online 03 June 2017
0920-1211/ © 2017 Elsevier B.V. All rights reserved.
 Table 1
Characteristics of studies included in the meta-analysis.

References Population Phenotypes Case/control Comparable Patients Patients Drug- Normal Method of Reported HLA-B allele/s included NOS
Q. Wang et al.

(study studied groups on CBZ with SJS/ tolerant controls genotyping in meta-analysis score
location) (n) TEN patients
(n)

Hung et al. Han Chinese SJS/TEN and case: CBZ was regarded as the Nontolerant 235 60 144 N MALDI-TOF mass 1502, 4001 6
(2006) (Taiwan) HSS/MEP offending drug if the onset of vs. spectrometry
cADRs symptomsoccurred tolerant
within the first 2 months of
exposure and the symptoms
resolved upon withdrawal of the
drug.
Control: patients who received
CBZ for at least 3 months
without evidence of adverse
reactions.
Tassaneeyakul Thai SJS/TEN case: CBZ was identified as the Nontolerant 84 42 42 N PCR-SBT 1502, 1521,1535,1301,1801,1802,2706,3505,3802,3909, 5
et al. (2010) (Thailand) culprit drug if the symptoms vs. 4001,4601,5101,5102,5502, 5601,5801
occurred within the first 3 tolerant
months of CBZ exposure and the
symptoms resolved upon
withdrawal of this drug.
control: patients who had used
CBZ for 6 monthswithout
evidence of anycutaneous
reactions were recruited as
controls.

20
Kaniwa et al. Japanese SJS/TEN NR Nontolerant 28 4 N 493 PCR-SBT 1511 5
(2010) (Japan) vs.
normal
controls
Kim et al. (2011) Korean SJS/TEN and case:CBZ was regarded as the Nontolerant 74 7 50 485 PCR locus-specific 1502,1511,5101 6
(Korea) HSS/MEP causative drug if the onset of the vs. primer; PCR group-
adverse reaction occurred tolerant specific primer
within 2 months of exposure and
without another suspected high- Nontolerant
risk drug medication in the vs.
period, and the symptoms normal
resolved after discontinuing the controls
drug. control：patients who
hadreceived CBZ for at least 3
months without evidence of an
adversereaction.

Niihara et al. Japanese SJS/TEN and Case: patients who showed Nontolerant 48 3 33 N PCR-rSSO 1507,5502,1511,4002,4001,5601 6
(2011) (Japan) HSS/MEP onset of cADRs within the first 3 vs.
months after starting CBZ tolerant
treatment and skin rash subsided
on discontinuation of
treatment.
Control: patients who did not
suffer from cADRs for at least 3
months on CBZ treatment.
Shi et al. (2012) Han Chinese SJS/TEN Case: patients who developed Nontolerant 111 18 93 0 PCR-SSP; PCR-SBT 1301,1502, 1511,3802,4001,4601,5401,5601 6
(southern SJS or TEN within 8 weeks after vs.
China) commencing CBZ therapy, for tolerant
(continued on next page)
Epilepsy Research 135 (2017) 19–28
 Table 1 (continued)

References Population Phenotypes Case/control Comparable Patients Patients Drug- Normal Method of Reported HLA-B allele/s included NOS
(study studied groups on CBZ with SJS/ tolerant controls genotyping in meta-analysis score
Q. Wang et al.

location) (n) TEN patients

(n)

which no other etiologic causes and

were found. Nontolerant
Control:patientswho did not vs.
show any cutaneous reaction normal
after 3 months taking CBZ. controls
Hsiao et al. Han Chinese SJS/TEN and Case: NR Nontolerant 356 112 152 N Sequence-specific 1501,1502,2704,4001,4801,5101,5401,5502 6
(2014) (Taiwan) HSS/MEP Control:patients who didnot vs. oligonucleotide
suffer from cADRs for at least 3 tolerant reverse line blots
months on CBZ treatment.
XiaoJ He(2013) Chinese SJS/TEN Case:CBZ was regarded as the Nontolerant 145 20 125 N PCR–SBT 5801 6
(Northeast offending drug when the onset vs.
China) of cADRs symptoms occurred tolerant
within the first 3 months of
exposure and the symptoms
resolved upon withdrawal of the
drug.
Control:patients were classified
as CBZ-tolerant controls if they
underwent CBZ therapy for
more than 3 months without
evidence of adverse reactions.
Kwan et al. Han Chinese SJS/TEN Case:patients who presented Nontolerant 161 26 135 N PCR–SBT 1301,1501,1502,1525,3501,3802,4001,4601,5101,5102, 5
(2014) (Hong Kong) with SJS/TEN within 12 weeks vs. 5401,5502,5601,5801

21
after commencing CBZ. tolerant
Control:patients who were
tolerant to AEDs after taking
them for at least 3 months
without developing
skin rash were identified.
Nguyen et al. Vietnamese SCADRs Case:symptoms commenced Nontolerant 63 35 25 N polymerase chain 1301,1512,1513,1518,1525,3503,3505,3801,3802,3901, 6
(2015) (Vietnam) include SJS/ within the first 2 months of vs. reaction and 4001,4002,4601,5101,5102,5401,5502,5604,5701,5801
TEN exposure to the drug and when tolerant sequence-specific
those symptoms resolved upon oligonucleotide
withdrawal of the drug. probes
Control：NR
Wang et al. Han Chinese SJS/TEN Case:patients who developed Nontolerant 47 16 39 N PCR–SBT 5801 5
(2014) (West China) AED-induced SJS/TEN after vs.
taking AEDs for less than eight tolerant
weeks;
Control:patients who had been
on an AED for more than three
months without experiencing an
allergic reaction.

All patients included were Asian population.

SJS: Stevens-Johnson syndrome; TEN: toxic epidermal necrolysis.
NOS: Newcastle-Ottawa Scale.
MALDI-TOF:Matrix-Assisted Laser Desorption/Ionization Time of Flight.
Epilepsy Research 135 (2017) 19–28
Q. Wang et al. Epilepsy Research 135 (2017) 19–28

Table 2
Pooled odds ratios for studies.

HLA-B types number population patients patients showing SJS/ CBZ-tolerant normal control OR(95%CI) P I2% fail-safe
of on CBZ TEN with CBZ(n) patients (n) group(n) number
studies
HLA total HLA total HLA total
positive positive positive

*1301 3 Thai(Tas), 305 13(12.6%) 103(33.8%) 22 202 NR NR 1.30 0.5 32 −1.08816

Chinese(Kwan) (0.60,2.80)
Viet (Ngu)
*1511 Nt 2 Korean(Kim), 105 4 10 3 88 NR NR 17.43 0.001 0 6.04125
vs. Japanese(NIIH) (3.12,97.41)
Tc
Nt 2 Japanese(Kan) 40 3 40 NR NR 6 757 11.11 0.001 0 9.28458
vs. Chinese(Shi) (2.62,47.09)
Pc
*1525 2 Chinese(Kwan) 227 2 61 4 160 NR NR 1.08 0.94 0 −1.59406
Viet(Ngu) (0.18,6.60)
*3802 3 Thai(Tas), 330 6 103 19 202 NR NR 0.75 0.58 0 −2.76266
Chinese(Kwan) (0.26,2.11)
Viet(Ngu)
*3505 2 Chinese(Kwan), 152 2 80 3 67 NR NR 0.67 0.8 55 0.99581
Viet (Ngu) (0.03,16.5)
*4001 6 Japanese 1005 26 278 170 531 NR NR 0.22 < 0.00001 36 12.00659
(Kan,NIIH),Chinese (0.14,0.35)
(Hung,Hsi,Kwan),Viet
(Ngu)
*4601 3 Thai(Tas), 374 17 103 52 202 NR NR 0.49 0.03 0 −0.34147
Chinese(Kwan) (0.25,0.95)
Viet(Ngu)
*5101 5 Thai(Tas),Korean 665 9 222 30 404 NR NR 0.76 [0.18, 0.71 64 −2.88354
(Kim),Chinese 3.25]
(Hsi,Kwan),
Viet (Ngu)
*5102 3 Thai(Tas), 313 2 103 6 202 NR NR 0.75 [0.18, 0.69 0 −1.42626
Chinese(Kwan), 3.12]
Viet(Ngu)
*5401 3 Chinese(Hsi,Kwan), 433 4 162 16 251 NR NR 0.35 [0.12, 0.07 0 −0.26683
Viet (Ngu), 1.07]
*5502 5 Thai(Tas) 643 5 230 21 387 NR NR 0.48 [0.18, 0.13 13 −0.99955
Japanese(NIIH), 1.24]
Chinese(Hsi,
Kwan),
Viet (Ngu)
*5601 3 Thai(Tas), 290 4 71 5 210 NR NR 2.53 [0.61, 0.2 3 −0.43288
Japanese(NIIH), 10.59]
Chinese(Kwan),
*5801 5 Thai(Tas) 567 9 139 53 366 NR NR 0.43 [0.09, 0.29 73 −0.42618
Chinese 2.05]
(He,Kwan,Wang)
Viet(Ngu)]

Nt: Nontolerantgroup;Tc: Toletant control group; Pc: Population control group.

NR: Not Reported.
Bold characters highlight significantly associated alleles with respective P values.
CBZ, carbamazepine;SJS: Stevens-Johnson syndrome; TEN: toxic epidermal necrolysis OR, odds ratio; CI, confidence interval.

leukocyte antigen (HLA) genes and CBZ induced-SJS/TEN (Grover and
Kukreti, 2014; Yip et al., 2012). HLA plays an important role in the
immune response (Chung et al., 2007). The HLA-B gene is a member of
the MHC, a region of the human genome located on chromosome 6. The
HLA-B protein, a cell-surface molecule, is responsible for the pre-
sentation of endogenous peptides to CD8+ T-cells. Due to the highly
polymorphic nature, a large number of HLA-B alleles have been iden-
tified. Information on the frequencies of over 2800 HLA-B alleles in
populations worldwide can be found at The Allele Frequency Net Da-
tabase (http://www.allelefrequencies.net/). Patients who carry HLA-
B*1502 allele have a significantly higher risk of developing SJS/TEN
(Bloch et al., 2014). The US Food and Drug Administration strongly
recommends that Asian patients should be screened for the HLA-
B*1502 allele before taking CBZ (FDA, 2013). Screening can help to
predict whether it is safe for a patient to take CBZ, and can therefore
help preventing the occurrence of SJS/TEN (Hakimian and Miller,
2011). CBZ-induced SJS/TEN and HLA-B*1502 has the strongest
association based on current data. However, HLA-B*1502 cannot ex-
plain all of the SJS/TEN cases. We believe that additional risk alleles
are important for SJS/TEN (Bloch et al., 2014; Yip et al., 2012). The
occurrence of SJS/TEN is not simply caused by just one gene allele,
many other genes may participate in the process of immune-response-
induced SJS/TEN (Chung et al., 2007).
We aimed to identify the associations between HLA-B alleles with
CBZ-induced SCADRs other than the HLA-B*1502 allele. The results
could provide recommendations for genetic testing prior to patients
receiving CBZ therapy.
2. Materials and methods
2.1. Search strategy and studies identification
We systematically searched Embase, PubMed, the Cochrane Library,
Web of Knowledge databases. All databases were searched from its

22
Q. Wang et al.
inception until May 31, 2016. The search strategy used was “HLA” OR
“human leukocyte antigen” AND “carbamazepine” OR “CBZ” AND
“Stevens Johnson Syndrome” OR “SJS” OR “Toxic Epidermal
Necrolysis” OR “TEN”. The Medical Subject Headings (MeSH) were
employed when searching a database when MeSH terms were available
(“HLA-B Antigens” AND “carbamazepine” AND “Stevens-Johnson
Syndrome” OR “Toxic Epidermal Necrolysis”). There was no restriction
on the language and the year of publication. We further searched the
bibliographies of the included articles to identify other pertinent stu-
dies.
2.2. Study selection
Two reviewers (QW & MX) independently evaluated the titles and
abstracts for study inclusion. The included studies must meet the fol-
lowing criteria: 1) the article investigated the association between HLA-
B and CBZ-induced SJS/TEN; 2) the study method should be either a
case-control study or a cohort study; 3) the study provided sufficient
data to calculate the carrier frequency of a variant HLA-B allele among
CBZ-tolerant patients, CBZ-induced SCADRs group or healthy control
and 4) cases were defined according to the detached body surface area
as SJS (< 10%), SJS/TEN overlap (10–30%) and TEN (> 30%) (Grover
and Kukreti, 2014). Case reports or case series, animal studies, review
articles, duplicate studies, and those studies with only HLA-B*1502
data which could be extracted were excluded. Besides SJS/TEN and
MPE should be analyzed separately, such studies can’t meet this re-
quirement should also be excluded.
2.3. Data extraction and quality assessment
All articles were extracted independently by two reviewers
(QW & MX). Discrepancies were compared and discussed until a con-
sensus among the investigators was reached. The following data were
extracted from each article: first author, population, ethnicity, study
design, diagnostic criteria of SJS or TEN, selection of cases and controls,
patients’ demographics, and the HLA-typing method. We used the
Newcastle-Ottawa quality assessment scale (Wells et al., 2000) for as-
sessing the quality of the included studies in this review. The New-
castle-Ottawa Scale (NOS) scores are displayed in Table 1. The HLA-B
variants data were extracted if the frequency data were mentioned in at
least two different studies (Table 2). Finally, we analyzed the potential
internal correlations between patients with SCADRs and patients tol-
erant to CBZ with the following 13 spanning HLA-B loci:*1301, *1511,
*1525, *3505, *3802, *4001, *4601, *5101, *5102, *5401, *5502,
*5601, *5801. Since the research results on HLA-B*1502 are clear, we
did not extract data related to this allele. The allele frequency data were
available in some studies. Similarly to Tanaka et al (Tanaka et al.,
1996), we converted the frequency data into the number of carrier
patients with special HLA-B locus present in at least one allele.
2.4. Statistical analysis
The overall odds ratios (ORs) with corresponding 95% confidence
intervals (CIs) were calculated to determine the association between the
presence of the HLA-B loci and CBZ-induced SJS/TEN. Carriers (n) are
individuals tested positive for at least one copy of a particular allele. P
value is two-tailed and P < 0.05 indicates a statistical significance.
The meta-analysis was performed with RevMan (version 5.3, the Nordic
Cochrane Centre, the Cochrane Collaboration, Copenhagen, Denmark).
Forest plots were used to display the visual representation of the meta-
analysis results (Lewis and Clarke, 2001). Sensitivity analyses were also
performed to determine the robustness of the findings by regional dif-
ferences. The statistical heterogeneity was assessed via the Q-statistic
and I-squared tests (Begg and Berlin, 1989).If there was heterogeneity
(P < 0.05 or I2 > 50%) among the studies, the random-effects models
were used (DerSimonian and Laird, 1986). If heterogeneity was not
23
Epilepsy Research 135 (2017) 19–28
present, the fix-effects models were used(Mantel and Haenszel, 1959).
The sensitivity analysis was carried out by singly removing each
study in order to understand the impact of the exclusion by calculating
the pooled ORs against the remaining studies. Considering there were
only limited studies, the funnel plots hardly show publication bias.
Therefore, the fail-safe number with significance set at 0.05 (Nfs0.05)
(Rosenberg, 2005) for each meta-analysis was applied to assess the
publication bias.
3. Results
A total of 931 articles were identified after searching different da-
tabases. A total of 266 duplicated studies and 577 unrelated articles
were excluded. The full texts of 54 records including studies based on
human population, HLA-B variants and the CBZ-induced cADRs were
then reviewed. A study which just simply divided patients into severe
cADRs group and mild cADRs group was excluded (Cheung et al., 2013;
Alfirevic et al., 2006). A total of 11 studies (Hung et al., 2006; He et al.,
2013; Hsiao et al., 2014; Kaniwa et al., 2010; Kim et al., 2011; Kwan
et al., 2014; Niihara et al., 2012; Nguyen et al., 2015; Shi et al., 2012;
Tassaneeyakul et al., 2010; Wang et al., 2014) met the inclusion and
exclusion criteria (Fig. 1). All the studies we included were fromAsian
population.No study based on white or black populations met our in-
clusion criteria. The study by Alfirevic (Alfirevic et al., 2006) did not
classify the severe hypersensitivity reaction which contained SJS/TEN.
Table 1 lists all the information in each study we extracted. Besides the
CBZ-induced SCADRs group and the CBZ-tolerant group, some studies
had normal control groups where patients in these groups were com-
bined with patients who were never prescribed to CBZ. All of the studies
confirmed with SJS/TEN diagnosis criteria followed Roujeau's criteria
(Cozzani et al., 2011). No additional articles were identified via a re-
view of the bibliographies of the included studies.
The analyzed studies included 343 CBZ-induced SJS/TEN cases, 838
CBZ tolerant-controls, and two studies (Kaniwa et al., 2010; Kim et al.,
2011) with 978 normal-controls. All of the SJS/TEN patients were di-
agnosed based on the clinical morphology defined by Roujeau
(Roujeau, 1994). In this study, we used NOS to evaluate the quality of
each study, with limit of > 5.
3.1. Meta-analysis results
Of the 11 studies we analyzed, 10 studies had detailed data re-
garding CBZ non-tolerant group and the tolerant group, and 3 studies
(Kaniwa et al., 2010; Kim et al., 2011; Shi et al., 2012) provided data
about CBZ non-tolerant group versus the normal control group. Only
one study did not provide the data of the CBZ non-tolerant group
(Kaniwa et al., 2010). We collected 13 variants of HLA-B genetic in-
formation which is displayed in Table 1. Among these HLA-B variants
data, HLA-B*1511, *4001, *4601, *5801 showed a significant dis-
tribution in the process of CBZ-induced SJS/TEN. In these included
SJS/TEN patients, the carrier rates of HLA-B*1511, *4001, *4601,
*5801 were 40.0%, 9.3%, 16.5% and 6.5%, respectively. HLA-B*1502
is the most relevant allele related to CBZ-induced SJS/TEN, and a
systematic review (Yip et al., 2012) reported that the HLA-B*1502
among patients with CBZ-induced SJS/TEN was 0.96 as compared with
0.11 in tolerant controls. The summary estimates for sensitivity and
specificity values were 0.96 and 0.88, respectively. A study (Hsiao
et al., 2014) included 112 SJS/TEN patients compared with 152 CBZ-
tolerant controls found that the positive predictive value and the ne-
gative predictive value were 90% and 92%, respectively.
3.1.1. Association between HLA-B*4001 and CBZ-induced SJS/TEN
The forest plots are shown in Fig. 2. There were six studies included
278 cases and 531 tolerant-controls. There was no significant hetero-
geneity found based on I2 (36%) and Q-statistics (P = 0.17). Therefore
the fixed-effects model was utilized to analyze the data. In the tolerant-
Q. Wang et al.
control group, 170patients were carriers of the allele. A larger pro-
portion of the controls (170/531 vs. 26/278) were carriers of the allele.
SJS/TEN cases were significantly less likely to carry the HLA-B*4001
allele compared with the tolerant-control (OR = 0.22,
95%CI = 0.14–0.35, P < 0.00001). As a protective factor, the sensi-
tivity and the specificity values of the HLA-B*4001 allele for prediction
of CBZ-induced SJS/TEN were 32.0% and 90.6%, respectively. The
positive predictive value and the negative predictive value were esti-
mated to be 86.7% and 41.1%, respectively.
3.1.2. Association between HLA-B*1511 and CBZ-induced SJS/TEN
Two studies (Niihara et al., 2012; Kim et al., 2011) included 10
cases and 88 tolerant-controls, and another two studies (Kaniwa et al.,
2010; Shi et al., 2012) included 40 cases and 757 population-controls.
The pooled ORs for the association between HLA-B*1511 and CBZ-in-
duced SJS/TEN are displayed in Table 3. No significant heterogeneity
was found among these studies based on I2 = 0. In the comparison of
cases and tolerant-controls, the carrier frequency of the cases (4/10)
was much greater than the tolerant-controls (3/88), (pooled OR 17.43;
95%CI3.12–97.41; P = 0.001) (Fig. 3). This indicated a strong asso-
ciation between HLA-B*1511 and CBZ-induced SJS/TEN. Two studies
provided data of cases (3/40) versus population-controls (6/757) for
the analysis. Based on the minimal I2 (I2 = 0) and the data above, we
could draw the same conclusion: compared with the general
24
Epilepsy Research 135 (2017) 19–28
Fig. 1. Flow diagram for study identification, screening ex-
clusion, and inclusion.
population, HLA-B*1511 also played an important role in the process of
CBZ-induced SJS/TEN (OR = 11.11; 95%CI = 2.62–47.09;
P = 0.001). The risks of SJS/TEN among CBZ users carrying the HLA-
B*1511 allele are 17.43-fold compared with tolerant-controls and are
11.11-fold compared with the general population. The sensitivity and
the specificity of the HLA-B*1511 allele for prediction of CBZ-induced
SJS/TEN were 14.0% and 98.9%, respectively. The positive predictive
value and the negative predictive value were estimated to be 43.8% and
95.1%, respectively.
3.1.3. Association between HLA-B*4601 and CBZ-induced SJS/TEN
The standard forest plots of HLA-B*4601 are shown in Fig. 4. Three
studies (Kaniwa et al., 2010; Nguyen et al., 2015; Tassaneeyakul et al.,
2010) met inclusion criteria, and there were 103 SJS/TEN cases and
202 tolerant-controls. Of the cases, 17 (16.5%) patients were carriers of
this allele, whereas in the tolerant-control group, the number of carriers
was 52 (25.7%). There was no significant heterogeneity (I2 = 0%) and
it was not significant given the Q-statistic (P = 0.49). The pooled OR
was 0.49 with 95%CI of 0.25–0.95 (P = 0.03). The sensitivity and the
specificity of the HLA-B*4601 allele for prediction of CBZ-induced SJS/
TEN were 25.7% and 83.5%, respectively. The positive predictive value
and the negative predictive value were estimated to be 75.4% and
36.4%, respectively. The result indicated that HLA-B*4601 may be
another protective factor in the process of CBZ-induced SJS/TEN. The
Fig. 2. Fixed-effects meta-analyses of CBZ-Induced Stevens-
Johnson Syndrome and Toxic Epidermal Necrolysis among
carriers of the HLA-B*4001 Allele.
Q. Wang et al. Epilepsy Research 135 (2017) 19–28

Table 3
Sensitivity analysis with pooled odds ratios for studies exploring association of HLA-B variants in patients showing SJS/TEN and drug-tolerant patients treated with carbamazepine
therapy by removing one study or studies belonging to specific region each time.

HLA-B type number of Population (references) OR (95% CI) P I2

studies (%)

*4001 5 All except Taiwan Han Chinese(Hsi) 0.23 [0.12, 0.42] < 0.00001 49
5 All except Taiwan Han Chinese(Hung) 0.25 [0.15, 0.43] < 0.00001 41
5 All except HongKong Han Chinese(Kwan) 0.24 [0.15, 0.39] < 0.00001 41
3 All Chinese(Hung,Hsi,Kwan) 0.46 [0.11, 1.98] < 0.00001 0
3 All except Chinese(Hung,His,Kwan) 0.70 [0.24, 2.03] 0.51 0
5 All except Vietnamese(Ngu) 0.21 [0.13, 0.34] < 0.00001 40
5 All except Japanese(NIIH) 0.21 [0.13, 0.33] < 0.00001 0
5 All except Thai(Tas) 0.21 [0.13, 0.34] < 0.00001 43
5101 4 All except Taiwan Han Chinese(Kwan) 0.90 [0.13, 6.29] 0.92 69
4 All except Korean(Kim) 0.51 [0.10, 2.67] 0.42 62
4 All except HongKong Han Chinese(Kwan) 0.48 [0.11, 2.06] 0.33 56
4 All except Vietnamese(Ngu) 0.99 [0.18, 5.38] 0.99 70
4 All except Thai(Tas) 1.10 [0.25, 4.81] 0.9 64
5502 4 All except Taiwan Han Chinese(Hsi) 0.87 [0.28, 2.75] 0.82 0
4 All except HongKong Han Chinese(Kwan) 0.52 [0.19, 1.45] 0.21 31
4 All except Vietnamese(Ngu) 0.40 [0.14, 1.15] 0.09 23
4 All except Japanese(NIIH) 0.32 [0.10, 1.00] 0.05 0
4 All except Thai(Tas) 0.47 [0.17, 1.34] 0.16 35
2 Chinese(Hsi,Kwan) 0.18 [0.03, 0.98] 0.05 0
5801 4 All except northeastern Han Chinese(He) 0.23 [0.09, 0.58] 0.002 0
3 All except HongKong Han Chinese(Kwan) and northeastern Han Chinese(He) 0.21 [0.07, 0.68] 0.009 0
3 All except Vietnamese(Ngu) and northeastern Han Chinese(He) 0.29 [0.11, 0.77] 0.01 0
3 All except Thai(Tas) and northeastern Han Chinese(He) 0.21 [0.06, 0.72] 0.01 0
3 All except Han Chinese(Wang) and northeastern Han Chinese(He) 0.28 [0.11, 0.75] 0.01 0
1301 2 All except HongKong Han Chinese(Kwan) 0.72 [0.24, 2.12] 0.55 0
2 All except Vietnamese(Ngu) 1.34 [0.30, 5.89] 0.38 62
2 All except Thai(Tas) 1.82 [0.73, 4.50] 0.2 7
3802 2 All except HongKong Han Chinese(Kwan) 0.84 [0.20, 3.49] 0.81 0
2 All except Vietnamese(Ngu) 0.77 [0.23, 2.56] 0.66 0
2 All except Thai(Tas) 0.68 [0.20, 2.31] 0.53 0
4601 2 All except HongKong Han Chinese(Kwan) 0.53 [0.24, 1.15] 0.11 26
2 All except Vietnamese(Ngu) 0.59 [0.27, 1.28] 0.18 0
2 All except Thai(Tas) 0.36 [0.15, 0.89] 0.03 0
5102 2 All except HongKong Han Chinese(Kwan) 0.84 [0.16, 4.44] 0.84 45
2 All except Vietnamese(Ngu) 0.32 [0.04, 2.75] 0.3 0
2 All except Thai(Tas) 1.43 [0.25, 8.33] 0.69 0
5401 2 All except Taiwan Han Chinese(Hsi) 0.68 [0.13, 3.68] 0.66 0
2 All except HongKong Han Chinese(Kwan) 0.32 [0.10, 1.05] 0.06 0
2 All except Vietnamese(Ngu) 0.28 [0.07, 1.10] 0.07 0
5601 2 All except HongKong Han Chinese(Kwan) 4.53 [0.64, 32.01] 0.13 0
2 All except Japanese(NIIH) 1.81 [0.35, 9.25] 0.48 0
2 All except Thai(Tas) 3.28 [0.14, 77.18] 0.46 53

Bold characters highlight significantly associated alleles with respective P values.

SJS: Stevens-Johnson syndrome; TEN: toxic epidermal necrolysis OR, odds ratio; CI, confidence interval.

Fig. 3. Fixed-effects meta-analyses of CBZ-Induced Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis among carriers of the HLA-B*1511 Allele.

25
Q. Wang et al.
Fig. 4. Fixed-effects meta-analyses of CBZ-Induced Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis among carriers of the HLA-B*4601 Allele.
Fig. 5. (1): Random-effects meta-analyses of Carbamazepine-Induced Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis among Carriers of the HLA-B*5801 Allele before
removing the northeast Chinese population. (2): Fixed-effects meta-analyses of Carbamazepine-Induced Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis among Carriers of the
HLA-B*5801 Allele after removing the northeast Chinese population.
SJS/TEN cases were significantly less likely to carry this allele.
3.1.4. Association between HLA-B*5801 and CBZ-induced SJS/TEN
To analyze the association of HLA-B*5801 and CBZ-induced SJS/
TEN, we included 5 studies (He et al., 2013; Kwan et al., 2014; Nguyen
et al., 2015; Tassaneeyakul et al., 2010; Wang et al., 2014) with re-
spective forest plots of HLA-B*5801 displayed in Fig. 5(1). We found
that there was a significant heterogeneity among these studies with an
I2heterogeneity of 73%, and a significant Q statistic (P = 0.006). A
sensitivity analysis was then performed to examine the heterogeneity.
The study was identified and excluded from the analysis(He et al.,
2013). Additionally, the study by He et al. was based on the Chinese
population in the northeast region of China. After exclusion of the
study, heterogeneity (I2% = 0) was absent, and proper analysis criteria
was achieved (OR = 0.23; 95%CI = 0.09–0.58; P = 0.002). The mod-
ified forest plots are displayed in Fig. 5(2). The sensitivity and the
specificity of the HLA-B*5801 allele for prediction of CBZ-induced SJS/
TEN were 19.5% and 95.8%, respectively. The positive predictive value
and the negative predictive value were estimated to be 90.4% and
37.0%, respectively.
3.1.5. Sensitivity analysis
To verify the robustness of the meta-analysis results, we carried out
a sensitivity analysis (Table 3) by single removal of studies or by se-
paration of all Han Chinese population-based studies to calculate the
pooled ORs against the remaining studies. In the analysis of the HLA-
B*4001 genotype, the summary OR for patients except all Han Chinese
was 0.70 (95%CI = 0.24–2.03), with an I2 heterogeneity of 0%. Such
26
Epilepsy Research 135 (2017) 19–28
contradictory results may be due to the study on Japanese population
(Niihara et al., 2012).
As for HLA-B*5801, such heterogeneity probably arose from a non-
homogeneous patient population from the different location of
Northern Hemisphere. In fact a study by He(He et al., 2013) including
patients from northeastern China found a low frequency of HLA-B*1502
in CBZ-induced SJS individuals. However, a strong association existed
between the south/southwestern and the central Chinese population.
Considering the regional difference, the remaining cases all came from
Southeast Asia.
3.1.6. Test for publication bias
Funnel plots were not used to evaluate publication bias since the
number of included reports were insufficient, instead we used Nfs0.05 to
figure out the publication bias. The elevated Nfs0.05 (Table 2) values
suggested that the associations based on the current data was reliable.
4. Discussion
Our findings reveal the association between HLA-B alleles (B*4001,
B*1511, B*4601 and B*5801) and CBZ-inducedSJS/TEN in Asian po-
pulation. Carriers of HLA-B*4001, *4601 and *5801 alleles are sig-
nificantly less likely to develop SJS/TEN, and the HLA-B*1511 allele is
associated with a significantly increased risk of developing SJS/TEN in
patients taking CBZ. The odds ratio for CBZ-induced SJS/TEN in HLA-
B*1511 carriers was 17.43 times compared with CBZ-tolerant controls,
and was 11.11 times greater when compared to normal population-
controls. Of all the 11 includued studies, the quality score represented a
Q. Wang et al.
reliable quality of overall evidence.
The CBZ-induced SCADRs are not just restricted to those HLA-B
molecule mentioned above. Based on the included studies, the potential
protective markers for developing CBZ-induced SCADRsmay include:
HLA-B*5101 (Hsiao et al., 2014; Kwan et al., 2014; Kim et al., 2011;
Nguyen et al., 2015; Tassaneeyakul et al., 2010), HLA-B*5102 (Kwan
et al., 2014; Nguyen et al., 2015; Tassaneeyakul et al., 2010), HLA-
B*5401 (Hsiao et al., 2014; Kwan et al., 2014; Nguyen et al., 2015),
HLA-B*5502 (Hsiao et al., 2014; Kwan et al., 2014; Niihara et al., 2012;
Nguyen et al., 2015; Tassaneeyakul et al., 2010), HLA-B*4601, and
HLA-B*3802 (Kwan et al., 2014; Nguyen et al., 2015; Tassaneeyakul
et al., 2010), where the potential riskmarker may be HLA-B*5601
(Kwan et al., 2014; Niihara et al., 2012; Tassaneeyakul et al., 2010).
Because of the relatively small mumber studies and the total sample
size, it is hard to quantitatively synthesize the magnitude of the asso-
ciation for the alleles mentioned above.A larger number of studies and
sample sizes are needed.
In patients with HLA-B polymorphisms, several mechanisms in-
cluding immunologic response have been proposed to explain the oc-
currence of SJS/TEN (Chung et al., 2007). However, the pathogenesis
of CBZ-induced SCADRsis not fully understood yet. The hapten hy-
pothesis indicates that the reactive drug could covalently bind to an
endogenous peptide through the peptide binding groove of the HLA
molecule which then is processedand presented to T-cells (Pichler,
2003). A specific HLA-B allele is therfore required for the activation of
drug-specific T cells by the culprit drug (Tangamornsuksan et al.,
2013). Several HLA alleles have been associated with immune-mediated
SCADRs to CBZ. As far as we know, the HLA-B*1502 remains to be the
strongest association (Bloch et al., 2014). The developing process of
SJS/TEN is the result of the interaction by multiple genes, not only just
by HLA-B*1502. There were a large number of patients carrying HLA-
B*1502 but didn’t show any adverse reactions, which indicated that
only a combination of certain risk factors and the lack of some pro-
tective factors can trigger the SCADRs. Identifying these factors could
improve the predictive power of genetic testing (Amstutz et al., 2014).
In 2012, Chen (Chen et al., 2012) conducted a prospective study that
suggested screening for HLA-B*1502 allele before taking CBZ would
decrease the incidence of CBZ SJS/TEN. Such screeningis already
mandatory in Hong Kong and Taiwan based on the FDA warning. Be-
sides HLA-B*1502, the related HLA-B alleles we found in this research
should also be involved in the genetic blood test. By testing all these
related alleles, we can more reliably evaluate the risks. But how to
define the weighting of each allele needs to be studied further.
Alleles in the HLA-B*15 group encode molecules belonging to many
serologic subgroups including B75 (Lee et al., 2000). Patients with the
presence of HLA-B*1511weresignificantly more likely to develop SJS/
TEN. Because HLA-B*1511 and HLA-B*1502 belong to the same sub-
family of B75 serotypeand therefore share the same structural similarity
(Grover and Kukreti, 2014), they may cause many serologic cross-re-
activity results (Lee et al., 2000). Since the strong association of HLA-
B*1502 has already been confirmed from widely separated geographic
areas (Chung et al., 2007) (India, Malaysia, Singapore, TaiWan, Hon-
gKong), both HLA-B*1511 and HLA-B*1502 could be associated with
CBZ induced SJS/TEN given the similarities in haplotype or serotype
(Bloch et al., 2014). The HLA-B*1511 associations were found mainly in
Southern Han Chinese, Japanese, and Korean populations.
The sensitivity analysis on HLA-B*4001 and HLA-B*5801 (excluding
the study (He et al., 2013)) suggested that the summary ORs remained
significant regardless of regional differences, indicating that the results
were robust. However, in analysis of HLA-B*4601, the result was not
significant when we excluded HongKong Han Chinese (Kwan et al.,
2014) or Vietnamese populations (Nguyen et al., 2015). Consideringthe
limited number of studies and the sample size, the result was still ac-
ceptable. In previous studies, there was a strong association between
the HLA-B*5801 allele and allopurinol-induced SJS/TEN which differed
among different ethnic populations (Kaniwa et al., 2008; Lonjou et al.,
27
Epilepsy Research 135 (2017) 19–28
2006). However, in populations of Southeast and Northeast China,
HLA-B*5801 was a protective and a risk factor, respectively. It's worth
mentioning that the *1502 allele frequency is very low in northeastern
China where the absence of *1502 could also be an important con-
founding variable. The opposite effect caused by HLA-B*5801 in dif-
ferent populations and in different drugs requires further research.
Our study has several limitations: 1) this study only focused on the
relationship between the HLA-B alleles but not HLA-A. In fact, theHLA-
A alleles also play an important role in the process of CBZ-induced SJS/
TEN, but related studies were not included in this research, 2) the
sample size of some specific HLA-B alleles was relatively small and the
statistical power needs to be strengthened, 3) many studies did not
provide data on the tolerant-control group and we were unable to add
them into this meta-analysis, and 4) we only considered severe cuta-
neous ADRs (SJS/TEN) but not mild maculopapular exanthema and
drug reactions with eosinophilia and systemic symptoms.
5. Conclusion
We found strong relationships between HLA-B*4001, *4601, *5801,
*1511 and CBZ-induced SJS/TEN in the Southeast Asian population. Of
which,HLA-B*4001, *4601, *5801 were protective biomarkers,
whereas HLA-B*1511 was a risk biomarker in the process of developing
SJS/TEN for patients taking CBZ. These findings may need to be con-
firmed in future with the inclusion of more studies and large sample
sizes. Screening more HLA-B genes may enable us to get a better pre-
diction of the possibility for developing SJS/TEN in patients prior to
taking carbamazapine.
References
Alfirevic, A., Jorgensen, A.L., Williamson, P.R., Chadwick, D.W., Park, B.K., Pirmohamed,
M., 2006. HLA-B locus in Caucasian patients with carbamazepine hypersensitivity.
Pharmacogenomics 7, 813–818.
Amstutz, U., Shear, N.H., Rieder, M.J., Hwang, S., Fung, V., Nakamura, H., Connolly,
M.B., Ito, S., Carleton, B.C., 2014. Recommendations for HLA-B*15:02 and HLA-
A*31:01 genetic testing to reduce the risk of carbamazepine-induced hypersensitivity
reactions. Epilepsia 55, 496–506.
Begg, C., Berlin, J., 1989. Publication bias and dissemination of clinical research. J. Natl.
Cancer Inst. 81, 107–115.
Bloch, K.M., Sills, G.J., Pirmohamed, M., Alfirevic, A., 2014. Pharmacogenetics of anti-
epileptic drug-induced hypersensitivity. Pharmacogenomics 15, 857–868.
Chan, H.L., Stern, R.S., Arndt, K.A., Langlois, J., Jick, S.S., Jick, H., Walker, A.M., 1990.
The incidence of erythema multiforme, Stevens-Johnson syndrome, and toxic epi-
dermal necrolysis. A population-based study with particular reference to reactions
caused by drugs among outpatients. Arch. Dermatol. 126, 43–47.
Chen, P., Lin, J.J., Lu, C.S., Ong, C.T., Hsieh, P.F., Yang, C.C., Tai, C.T., Wu, S.L., Lu, C.H.,
Hsu, Y.C., Yu, H.Y., Ro, L.S., Lu, C.T., Chu, C.C., Tsai, J.J., Su, Y.H., Lan, S.H., Sung,
S.F., Lin, S.Y., Chuang, H.P., Huang, L.C., Chen, Y.J., Tsai, P.J., Liao, H.T., Lin, Y.H.,
Chen, C.H., Chung, W.H., Hung, S.I., Wu, J.Y., Chang, C.F., Chen, L., Chen, Y.T.,
Shen, C.Y., 2012. Carbamazepineinducedtoxic effects and HLA-B*1502screening in
Taiwan. N. Engl. J. Med. 364, 1126–1133.
Cheung, Y.K., Cheng, S.H., Chan, E.J., Lo, S.V., Ng, M.H., Kwan, P., 2013. HLA-B alleles
associated with severe cutaneous reactions to antiepileptic drugs in Han Chinese.
Epilepsia 54, 1307–1314.
Chung, W.H., Hung, S.I., Chen, Y.T., 2007. Human leukocyte antigens and drug hy-
persensitivity. Curr. Opin. Allergy Clin. Immunol. 7, 317–323.
Chung, W.H., Hung, S.I., Yang, J.Y., Su, S.C., Huang, S.P., Wei, C.Y., Chin, S.W., Chiou,
C.C., Chu, S.C., Ho, H.C., Yang, C.H., Lu, C.F., Wu, J.Y., Liao, Y.D., Chen, Y.T., 2008.
Granulysin is a key mediator for disseminated keratinocyte death in Stevens-Johnson
syndrome and toxic epidermal necrolysis. Nat. Med. 14, 1343–1350.
Cozzani, E., Di Zenzo, G., Calabresi, V., Caproni, M., Schena, D., Quaglino, P., Marzano,
A.V., Fabbri, P., Rebora, A., Parodi, A., 2011. Anti-desmoplakin antibodies in er-
ythema multiforme and Stevens-Johnson syndrome sera: pathogenic or epipheno-
menon? Eur. J. Dermatol. 21, 32–36.
DerSimonian, R., Laird, N., 1986. Meta-analysis in clinical trials. Control. Clin. Trials 7,
177–188.
Food and Drug Administration Alert, 2013. Information for Healthcare Professionals:
Dangerous or Even Fatal Skin Reactions: Carbamazepine (marketed as Carbatrol,
Equetro, Tegretol, and Generics). (Accessed 13.10). http://www.fda.gov/Drugs/
DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/
ucm124718</tp:e-path>.htm.
Grover, S., Kukreti, R., 2014. HLA alleles and hypersensitivity to carbamazepine: an
updated systematic review with meta-analysis. Pharmacogenet. Genomics 24,
94–112.
Hakimian, S., Miller, J.W., 2011. The genetics of antiepileptic drug-induced skin
Q. Wang et al.
reactions. Clin. Pharmacol. Ther. 89, 908–910.
He, X.J., Jian, L.Y., He, X.L., Wu, Y., Xu, Y.Y., Sun, X.J., Miao, L.Y., Zhao, L.M., 2013.
Association between the HLA-B*15:02 allele and carbamazepine-induced Stevens-
Johnson syndrome/toxic epidermal necrolysis in Han individuals of northeastern
China. Pharmacol. Rep. 65, 1256–1262.
Hsiao, Y.H., Hui, R.C., Wu, T., Chang, W.C., Hsih, M.S., Yang, C.H., Ho, H.C., Chang, Y.G.,
Chen, M.J., Lin, J.Y., Chen, D.P., Chang, P.Y., Wu, T.L., Hung, S.I., Chung, W.H.,
2014. Genotype-phenotype association between HLA and carbamazepine-induced
hypersensitivity reactions: strength and clinical correlations. J. Dermatol. Sci. 73,
101–109.
Hung, S.I., Chung, W.H., Jee, S.H., Chen, W.C., Chang, Y.T., Lee, W.R., Hu, S.L., Wu, M.T.,
ChenmG.S. Wong, T.W., Hsiao, P.F., Chen, W.H., Shih, H.Y., Fang, W.H., Wei, C.Y.,
Lou, Y.H., Huang, Y.L., Lin, J.J., Chen, Y.T., 2006. Genetic susceptibility to carba-
mazepine-induced cutaneous adverse drug reactions. Pharmacogenet. Genomics 16,
297–306.
Kaniwa, N., Saito, Y., Aihara, M., Matsunaga, K., Tohkin, M., Kurose, K., Sawada, J.,
Furuya, H., Takahashi, Y., Muramatsu, M., Kinoshita, S., Abe, M., Ikeda, H.,
Kashiwagi, M., Song, Y., Ueta, M., Sotozono, C., Ikezawa, Z., Hasegawa, R., 2008.
HLA-B locus in Japanese patients with anti-epileptics and allopurinol-related Stevens-
Johnson syndrome and toxic epidermal necrolysis. Pharmacogenomics 9, 1617–1622.
Kaniwa, N., Saito, Y., Aihara, M., Matsunaga, K., Tohkin, M., Kurose, K., Furuya, H.,
Takahashi, Y., Muramatsu, M., Kinoshita, S., Abe, M., Ikeda, H., Kashiwagi, M., Song,
Y., Ueta, M., Sotozono, C., Ikezawa, Z., Hasegawa, R., 2010. HLA-B*1511 is a risk
factor for carbamazepine-induced Stevens-Johnson syndrome and toxic epidermal
necrolysis in Japanese patients. Epilepsia 51, 2461–2465.
Kim, S.H., Lee, K.W., Song, W.J., Kim, S.H., Jee, Y.K., Lee, S.M., Kang, H.R., Park, H.W.,
Cho, S.H., Park, S.H., Min, K.U., Chang, Y.S., 2011. Carbamazepine-induced severe
cutaneous adverse reactions and HLA genotypes in Koreans. Epilepsy Res. 97,
190–197.
Kwan, P.K., Ng, M.H., Lo, S.V., 2014. Association between HLA-B*15:02 allele and an-
tiepileptic drug-induced severe cutaneous reactions in Hong Kong Chinese: a popu-
lation-based study. Hong Kong Med. J. 20, 16–18.
Lee, K.W., Jeon, H., Park, J.Y., 2000. HLA-B*15 diversity in the Korean population. Tissue
Antigens 56, 428–435.
Lewis, S., Clarke, M., 2001. Forest plots: trying to see the wood and the trees. BMJ 322,
1479–1480.
Lonjou, C., Thomas, L., Borot, N., Ledger, N., de Toma, C., LeLouet, H., Graf, E.,
Schumacher, M., Hovnanian, A., Mockenhaupt, M., Roujeau, J.C., 2006. A marker for
Stevens-Johnson syndrome.: ethnicity matters. Pharmacogenomics J. 6, 265–268.
Mantel, N., Haenszel, W., 1959. Statistical aspects of the analysis of data from retro-
spective studies of disease. J. Natl. Cancer Inst. 22, 719–748.
Marson, A.G., Al-Kharusi, A.M., Alwaidh, M., Appleton, R., Baker, G.A., Chadwick, D.W.,
Cramp, C., Cockerell, O.C., Cooper, P.N., Doughty, J., Eaton, B., Gamble, C.,
Goulding, P.J., Howell, S.J., Hughes, A., Jackson, M., Jacoby, A., Kellett, M., Lawson,
G.R., Leach, J.P., Nicolaides, P., Roberts, R., Shackley, P., Shen, J., Smith, D.F., Smith,
P.E., Smith, C.T., Vanoli, A., Williamson, P.R., SANAD Study group, 2007. The
SANAD study of effectiveness of carbamazepine, gabapentin, lamotrigine, oxcarba-
zepine, or topiramate for treatment of partial epilepsy: an unblinded randomised
controlled trial. Lancet 369, 1000–1015.
Nguyen, D.V., Chu, H.C., Nguyen, D.V., Phan, M.H., Craig, T., Baumgart, K., van Nunen,
28
Epilepsy Research 135 (2017) 19–28
S., 2015. HLA-B*1502 and carbamazepine-induced severe cutaneous adverse drug
reactions in Vietnamese. Asia Pac Allergy 5, 68–77.
Niihara, H., Kakamu, T., Fujita, Y., Kaneko, S., Morita, E., 2012. HLA-A31 strongly as-
sociates with carbamazepine-induced adverse drug reactions but not with carbama-
zepine-induced lymphocyte proliferation in a Japanese population. J. Dermatol. 39,
594–601.
Pichler, W.J., 2003. Delayed drug hypersensitivity reactions. Ann. Intern. Med. 139,
683–693.
Rosenberg, M.S., 2005. The file-drawer problem revisited: a general weighted method for
calculating fail-safe numbers in meta-analysis. Evolution 59, 464–468.
Roujeau, J.C., Stern, R.S., 1994. Severe adverse cutaneous reactions to drugs. N. Engl. J.
Med. 331, 1272–1285.
Roujeau, J.C., 1994. The spectrum of Stevens-Johnson syndrome and toxic epidermal
necrolysis: a clinical classification. J. Invest. Dermatol. 102, 28S–30S.
Shi, Y.W., Min, F.L., Qin, B., Zou, X., Liu, X.R., Gao, M.M., Wang, Q., Zhou, J.Q., Liao,
W.P., 2012. Association between HLA and Stevens-Johnson syndrome induced by
carbamazepine in Southern Han Chinese: genetic markers besides B*1502? Basic
Clin. Pharmacol. Toxicol. 111, 58–64.
Svensson, C.K., Cowen, E.W., Gaspari, A.A., 2001. Cutaneous drug reactions. Pharmacol.
Rev. 53, 357–379.
Tanaka, H., Akaza, T., Juji, T., 1996. Report of the japanese central bone marrow data
center. Clin. Transpl. 139–144.
Tangamornsuksan, W., Chaiyakunapruk, N., Somkrua, R., Lohitnavy, M., Tassaneeyakul,
W., 2013. Relationship between the HLA-B*1502 allele and carbamazepine-induced
Stevens-Johnson syndrome and toxic epidermal necrolysis: a systematic review and
meta-analysis. AMA Dermatol. 49, 1025–1032.
Tassaneeyakul, W., Tiamkao, S., Jantararoungtong, T., Chen, P., Lin, S.Y., Chen, W.H.,
Konyoung, P., Khunarkornsiri, U., Auvichayapat, N., Pavakul, K., Kulkantrakorn, K.,
Choonhakarn, C., Phonhiamhan, S., Piyatrakul, N., Aungaree, T., Pongpakdee, S.,
Yodnopaglaw, P., 2010. HLA-B*1511 is a risk factor for carbamazepine-induced
Stevens-Johnson syndrome and toxic epidermal necrolysis in Japanese patients.
Epilepsia 51, 2461–2465.
Wang, W., Hu, F.Y., Wu, X.T., An, D.M., Yan, B., Zhou, D., 2014. Genetic predictors of
Stevens-Johnson syndrome and toxic epidermal necrolysis induced by aromatic an-
tiepileptic drugs among the Chinese Han population. Epilepsy Behav. 37, 16–19.
Wells, G.A., Shea, B., O'Connell, D., Peterson Welch, V., Losos, M., Tugwell, P., 2000. The
Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in
meta-analyses. Appl. Eng. Agric. 18, 727–734.
Wiffen, P.J., Derry, S., Moore, R.A., McQuay, H.J., 2011. Carbamazepine for acute and
chronic pain in adults. Cochrane Database Syst. Rev. 19, CD005451.
Yang, F., Yang, Y., Zhu, Q., Chen, S.A., Fu, X., Yan, S., Meng, C., Ma, L., Sun, X., Xu, J.,
Luo, X., Xing, Q., 2015. Research on susceptible genes and immunological patho-
genesis of cutaneous adverse drug reactions in chinese hans. J. Investig. Dermatol.
Symp. Proc. 17, 29–31.
Yip, V.L., Marson, A.G., Jorgensen, A.L., Pirmohamed, M., Alfirevic, A., 2012. HLA
genotype and carbamazepine-induced cutaneous adverse drug reactions: a systematic
review. Clin. Pharmacol. Ther. 92, 757–765.
Yun, J., Adam, J., Yerly, D., Pichler, W.J., 2012. Human leukocyte antigens (HLA) as-
sociated drug hypersensitivity: consequences of drug binding to HLA. Allergy 67,
1338–1346.