Spontaneous reporting of Stevens-Johnson syndrome and toxic epidermal necrolysis

associated with antiepileptic drugs

Abstract. OBJECTIVE: To assess the association between Stevens-Johnson syndrome
(SJS)/toxic epidermal necrolysis (TEN) and antiepileptics including the most recently
authorized drugs.
METHODS: In the Spanish Pharmacovigilance database, we searched for spontaneous
reports of SJS or TEN associated with antiepileptic drugs and analysed: a) reporting odds
ratio (ROR), b) age and gender of the patient, c) evolution, d) latency and recovery periods
and e) presence or absence of other suspected drugs.
RESULTS: A total of 84 reports of SJS and 80 of TEN related to 9 antiepileptic drugs were
studied. Reports were mainly associated with phenytoin (SJS: 28; TEN: 43), lamotrigine
(SJS: 37; TEN: 20) and carbamazepine (SJS: 14; TEN: 16). Other antiepileptic drugs
involved were: valproate, phenobarbital, oxcarbazepine, levetiracetam, primidone and
gabapentin. Patients were of a median age of 40 [1-87] and 57.3% of them were women.
Cases related to phenytoin were
more common in older men and to lamotrigine in younger women. The latency period of SJS
and
TEN did not exceed the first month of treatment and, in most of the analysed reports, the
outcome was recovery.
CONCLUSIONS: Our observations support the association of SJS or TEN with phenytoin,
carbamazepine, valproate or phenobarbital and enlighten the role of lamotrigine and others
such as oxcarbazepine or levetiracetam.
Key Words: Stevens-Johnson syndrome, Toxic epidermal necrolysis, Antiepileptic drugs,
Spontaneous reporting.
Introduction
Nearly 3% of hospital inpatients develop cutaneous adverse drug reactions (ADR)1, Stevens
Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) being the most severe
syndromes. These ADR have a very low annual occurence, 1-6 cases per 1 million of
population, and are clinically characterized by warning signs such as fever and malaise,
followed by the rapid onset of erythematous lesions that progress to extensive blistering and
epidermal detachment, accompanied by mucosal erosions, slight elevation of hepatic
enzymes and digestive and respiratory disorders2-4. SJS and TEN are mainly differentiated

by the severity and the percentage of body surface area damaged which is < 10% in SJS, >
30% in TEN and between 10-30% in SJS/TEN overlap5,6.
It is estimated that medications cause 30-50% of cases of SJS and up to 80% of TEN
cases in adults7. Drugs associated with the development of SJS/TEN include sulfonamides,
antiepileptics, non-steroidal antiinflammatory drugs (NSAIDs) and allopurino l8-11. Among
these pharmacological groups, antiepileptic drugs are authorized for several therapeutic
indications and are highly prescribed. The aim of our study was to assess the association
between SJS/TEN and antiepileptics, including the most recently authorized drugs, based on
the information provided by the spontaneous reporting of suspected ADR.
Methods
Using the Spanish Pharmacovigilance database (FEDRA, Farmacovigilancia Espaola
Datos de Reacciones Adversas) we selected the spontaneous reports of SJS or TEN associated
with drugs that fulfil all the following criteria: (1) Reporting date: from 01/01/1980 to
30/09/2009, (2) Type of report: spontaneous reporting from health professionals or drug
companies, (3) ADR defined by the preferred terms of MedDRA, Medical Dictionary for
Drug Regulatory Activities, a standardized dictionary of medical terminology, developed
under the auspices of the International Conference of Harmonisation (MedDRA MSSO,
Brussels, Belgium):
Stevens Johnson syndrome or toxic epidermal necrolysis, (4) Suspected drugs, by
themselves or by interaction with other drugs, included in the N03 Group (antiepileptic
drugs) of the Anatomical Therapeutical Chemical (ATC) Classification (Norwegian Institute
of Public Health, WHO Collaborating Centre For Drug Statistics Methodology, Oslo,
Norway).
In the obtained reports, we calculated the frequency of each ADR (SJS or TEN), for
each
antiepileptic drug, and analysed: (1) reporting odds ratio (ROR) and the 95% confidence
interval
(95% CI), by constructing a 2 2 contingency table for each antiepileptic drug, (2) age and
gender of the patient, (3) evolution, classified as recovered, not recovered, recovered with
consequences, in recovery, unknown and exitus, (4) latency (interval of time between the
beginning of the treatment and the onset of the adverse reaction) and recovery (interval of
time between the withdrawal of the treatment and the end of the adverse reaction) periods of

both ADR, and (5) presence or absence of other suspected drugs in the treatment received by
the patient.
Statistical Analysis
We studied the main features of the cases of SJS or TEN separately and carried out
comparisons between them by Chi-squared test for qualitative variables and t-test for
quantitative variables. Data mining and statistical analysis of the reports were performed with
SPSS for Windows (IBM SPSS Statistics 19.0, Armonk, NY, USA). p < 0.05 was considered
statistically significant.
Results
A total of 84 reports of SJS and 80 of TEN associated with 9 antiepileptic drugs were
found. Table I shows the frequency and the disproportionality analysis of the most frequently
involved antiepileptic drugs: phenytoin, lamotrigine, carbamazepine, valproate and
phenobarbital. Other antiepileptic drugs less frequently reported were: oxcarbazepine (SJS, 1;
TEN, 2), levetiracetam (SJS, 1; TEN, 1), gabapentin (SJS, 1) and primidone (TEN, 1).
The age and gender of the patients affected by SJS versus TEN associated with
antiepileptic drugs, considered as a group, are summarized in Table II. There were no
differences in the median age of patients. However, a different distribution of the age groups
was observed by comparing SJS versus TEN; SJS was less frequent than TEN in adult and
elderly patients. By studying the reports on all the antiepileptic drugs together, no difference
(p=0.5) was found in the distribution by gender between SJS and TEN (Table II). On the
other hand, the comparison between antiepileptic drugs (Table III) showed that patients with
SJS associated with phenytoin were significantly (p < 0.01) older than patients with SJS
associated with other antiepileptic drugs. In addition, TEN related to phenytoin was
significantly (p < 0.01) less frequent in women. However, TEN cases associated with
lamotrigine were significantly (p < 0.01) more frequent in women and in younger patients
than TEN cases associated with other antiepileptic drugs. The latency period of SJS and TEN
did not exceed the first month of treatment (Table II). In most of the analysed reports the
outcome was recovery although some cases of SJS or TEN evolved to exitus. Furthermore,
some of the analysed reports included more than one suspected drug; the mean of suspected
drugs included in
the TEN reports being higher (p < 0.05) than in SJS reports. The percentages of the
reports of SJS or TEN associated with one antiepileptic drug alone are presented in Figure 1.
In all cases associated with valproate there were other concomitant suspected drugs.

Table II. Main features of cases of SJS or TEN associated with antiepileptic drugs.

Discussion
In our study, the three antiepileptic drugs most frequently associated with reports of
SJS or TEN were phenytoin, lamotrigine and carbamazepine; our results for phenytoin and
carbamazepine being in agreement with some previous studies8,9,12. Lamotrigine was the
second most frequent antiepileptic drug involved in our observations whereas in the SCAR
(Severe Cutaneous Adverse Reaction) study8, carried out from 1989 to 1993, there were no
cases associated with lamotrigine. In the EuroSCAR (European Severe Cutaneous Adverse
Reaction) study9, conducted from 1997 to 2001, only fourteen cases were associated with
lamotrigine. Some cases of SJS or TEN associated with lamotrigine have been published1315 and one previous Spanish study in pediatric patients found that lamotrigine was frequently
related to SJS or TEN16. In Spain, lamotrigine was authorized in 1993 and our work,
although only including reports from the Spanish population, covers the period of time from
that year to 2009. This, and perhaps a wider prescription of the drug, could explain this high
number of cases related to lamotrigine.
In addition, we found reports of SJS or TEN associated with more recently authorized
antiepileptic drugs such as, oxcarbazepine, levetiracetam, primidone and gabapentin that
were
Table III. Age and gender differences between antiepileptic drugs in the analysed reports.

not related to severe cutaneous ADR in the previous above-mentioned studies7,8,

although isolated cases associated with some of these antiepileptics have been reported17-21.
Oxcarbazepine and levetiracetam are the most recently marketed, but gabapentin was
authorized the same year as lamotrigine and the number of reports related to gabapentin was
lower than that related to lamotrigine.
This difference seems not to be based on differences in consumption because,
according to the report published by the Spanish Health Agency of Medicines and Medical
Devices22, from 2000 to 2006 the consumption of gabapentin in Spain was higher than the
consumption of lamotrigine.
The mechanism action of lamotrigine to induce SJS or TEN is under debate.
Lamotrigine is predominantly metabolized in the liver by glucuronidation and the genetic
alteration of this process, with a lower clearance of lamotrigine, has been proposed as a
potential mechanism involved in the development of some TEN cases23. Furthermore,
lamotrigine has an aromatic amine structure and this type of antiepileptic drugs has been
more commonly related than others to the development of some hypersensitivity disorders,

such as SJS and TEN12,24-26. However, gabapentin is a non-aromatic antiepileptic drug and
is excreted primarily unchanged and, hence, the production of toxic metabolites with this
drug is very low. On the other hand, carbamazepine, phenytoin and phenobarbital are
metabolized in a high proportion (90-100%)23,27. These pharmacokinetic differences could
partially explain differences between antiepileptic drugs in the induction of severe cutaneous
ADR.
Valproate was the fourth most frequently antiepileptic drug related to cases of SJS or
TEN in our study, although it was always in combination with other antiepileptic drugs such
as lamotrigine, phenytoin or carbamazepine and it is difficult to establish the partial
contribution of each drug in the development of the ADR. Some potential consequences of
the interaction between valproic acid and lamotrigine have been previously described28,29.
Valproic acid interacts with the metabolism of lamotrigine by inhibiting its glucuronidation27
that secondarily induces the increase in the oxidative process and the production of toxic
metabolites, that could trigger cutaneous ADR30,31. In the analysed reports, SJS and TEN
were developed more frequently in women than men and the median age of patients affected
was nearly 40 years, in agreement with the observations of previous studies8,9,32,33. These
features were common to all antiepileptic drugs analysed apart from the cases associated with
phenytoin, which were more frequent in men, and the cases associated with lamotrigine, more
common in younger women. These age and gender differences could simply reflect
differences in the prescription and consumption of the antiepileptic drugs. Consumption
data by age and gender have not been published but lamotrigine monotherapy at doses lower
than 300 mg per day is considered one of the safest antiepileptic treatments in women of
childbearing potential34. Phenytoin has a narrow therapeutic range and exhibits saturable
kinetics with therapeutic doses35 and its prescription can be lowered in children, young
women and elderly patients. Latency period to develop SJS or TEN was under 4 weeks, this
period being within the interval of 1-8 weeks described in the EuroSCAR study9. Recovery
period was almost 2 weeks in the cases of SJS and 3 weeks in the cases of TEN, certainly in
relation with the severity in the TEN cases. Most of the reports included 1 or 2 suspected
drugs, TEN reports being those that included a higher number of concomitant drugs, perhaps
due to a tendency by the reporters to collect and enclose more information
concomitant medications in the most severe cases.

about

Conclusions
Our observations are in line with those identified in previous studies, providing
information on the features of cases and suggesting associations with more recently
authorized drugs within the group of antiepileptic drugs studied.
Conflict of Interest
The Authors declare that there are no conflicts of interest.