The major burden of illness in bipolar disorder (BD) is in the depressive pole. Lamotrigine has been shown to be useful in the long-term prophylaxis of depressive episodes in BD. Current guidelines recommend discontinuing lamotrigine in patients who develop rash. Our objective in this paper is to review literature to identify possible predictors of serious vs. benign rash that might help guide clinical decision-making and recommend titration strategy for re-introduction of lamotrigine, if indicated. We performed a search of the literature between 1966 and July 2008 to investigate the phenomenon of lamotrigine-induced rash and rechallenge procedures. The search identified six reports, and we were able to identify another case series from reviewing the bibliography of all of the above papers. We reviewed all the papers of lamotrigine rash rechallenge that resulted from the literature search. These papers describe 44 cases of lamotrigine rechallenge. Currently, there are 39 reported cases in the literature of successful lamotrigine rechallenge after a rash and five cases with rash recurrence. There are some characteristics of the rash that can help identify serious cases from benign ones. In addition, very slow titration of lamotrigine is crucial to the reduction of rash recurrence rate. Several cases that develop benign rash on lamotrigine can be rechallenged without adverse consequences. We believe that lamotrigine rechallenge in bipolar depression is an under-utilized option in our clinical armamentarium, and further studies are needed to guide us in this area.
In the World Health Organization Global Burden of Disease Study, bipolar disorder (BD) is ranked seventh among all medical disorders in years of life lost to death or disability (Chisholm et al., 2005). While manic episodes can be disruptive, episodes of depression in BD are chronic (Judd et al., 2002), highly recurrent, and carry a significant risk of suicide that ranges from 8% to 20% (Bostwick and Pankratz, 2000; Calabrese et al., 2002; Post et al., 2005). Prospective follow-up of BD patients over 1 yr (Post et al., 2003) and 12.8 yr (Judd et al., 2002) showed that depressive symptoms predominated over manic/hypomanic symptoms (>3:1).
There are significant gaps in the identification and treatment of BD (Bauer and Mitchner, 2004; Coyle et al., 2003; Frye et al., 2004; Sachs, 2003), and this is even more true in the case of bipolar depression, i.e. patients with BD suffering with a major depressive episode. Antidepressant medications are routinely used to treat unipolar depression and, despite emerging evidence of a lack of efficacy, are also used to treat bipolar depression (Sachs et al., 2007). However, the clinical long-term use of these drugs in bipolar depressed patients is limited due to the possible association with manic relapse (Prien et al., 1973), rapid cycling (Wehr and Goodwin, 1979), and treatment resistance (Ghaemi, 2008; Ghaemi et al., 2008).
Lamotrigine (LTG) has been shown to be efficacious in the long-term prophylaxis of depressive episodes in randomized controlled trials (Bowden et al., 2003; Calabrese et al., 2003; Goodwin et al., 2004). LTG is a structurally novel anticonvulsant that was approved by the FDA for the treatment of epilepsy in 1994. It is indicated for adjunctive treatment of partial seizures, primary generalized tonic-clonic seizures, and the generalized seizures of Lennox–Gastaut syndrome in adults and children. It is also indicated as monotherapy in adults with partial seizures. In 2003, the FDA approved LTG for the maintenance treatment of BD I to delay the time to occurrence of mood episodes (depression, mania, hypomania, mixed episodes) in patients treated for acute mood episodes with standard therapy. This was the first medication since lithium that was granted FDA approval for the maintenance treatment of BD I. Treatment recommendations specify that LTG should be discontinued at the first sign of a rash, regardless of its type and severity, unless the rash is clearly not drug related (PDR, 2007). Because of the fact that initial dosing and dose titration play a crucial role in the occurrence of a rash, LTG re-introduction with a very low initial dose and very slow titration schedule may result in a lack of recurrence of skin rash in some cases (Gelisse and Crespel, 2006).
The emergence of rash in LTG-treated patients with BD has potentially serious consequences both in respect to potential dermatological and systemic effects and also with regard to withdrawing a potentially useful medication and precipitating relapse. In the absence of clear clinical protocols to guide clinicians in this situation, we performed a systematic literature review of the cases and case series associated with LTG-induced rash (Table 1). We also discuss the drug interactions that could increase the chances of rash and the clinical characteristics that could differentiate benign from serious rash and provide recommendations based on the review of the literature.
We performed a search on PubMed, the ISI Web of Knowledge, EBSCO (PsycINFO included), Scopus, and CINAHL databases for reports between 1966 to July 2008 using the key words ‘lamotrigine rechallenge’ (‘lamotrigine’ [Substance Name] OR lamotrigine [Text Word] AND rechallenge [All Fields]). Other search terms used were ‘rash rechallenge’ (‘exanthema’ [TIAB] NOT Medline [SB] OR ‘exanthema’ [MeSH Terms] OR rash [Text Word] AND rechallenge [All Fields]). This search resulted in seven papers, six of those were relevant to the focus of this paper – four case series of epilepsy patients (Gelisse and Crespel, 2006; Huang et al., 2002; P-Codrea Tigaran et al., 2005; Tavernor et al., 1995) and two case reports of BD patients (Buzan and Dubovsky, 1998; Manfredi et al., 2004). We were also able to identify one more case series on reviewing the bibliography of all of the above papers (Besag et al., 2000) (see Table 1).
We also searched the ISI Web of Knowledge database from 1992 to July of 2008 for the term ‘lamotrigine rash’ (214 papers) then refined results by ‘rechallenge,’ which gave seven results. Four of those were relevant and were already included in the PubMed search above. The other three were out of the scope of the focus of this paper. We searched all the databases mentioned thereafter for all available manuscripts until July of 2008. EBSCO databases were searched (PsycINFO included) for a combination of ‘lamotrigine’ and ‘rechallenge’ which revealed four papers. Two were out of scope and two were already included in the above PubMed search. Search of the Scopus database for a combination of ‘lamotrigine’ and ‘rechallenge’ revealed 16 papers. Eleven were out of scope and five relevant ones were already included in our previous PubMed search. Search of CINAHL database using ‘lamotrigine rechallenge’ gave no results for this topic.
There are more cases in the literature on LTG rechallenge (Table 1) in epilepsy (42 cases) than BD patients (two cases).
Rechallenge with LTG was successful in one case of BD (Manfredi et al., 2004). LTG was re-started at 12.5 mg/d for 2 wk, and then increased to 25 mg/d for 2 wk, 50 mg/d for 2 wk, 75 mg/d for 2 wk, and finally 100 mg/d. The initial maculopapular pruritic rash had developed after 2 wk of 25 mg/d LTG as a starting dose. The other case of BD rechallenge was described by Buzan and Dubovsky (1998). The drug was initially started at 25 mg/d and increased by 25 mg every 3 d for 2 wk, then by 50-mg increments every 3 d to a dose of 300 mg/d. A maculopapular rash appeared at 300 mg/d. The medication was titrated faster than recommended by the manufacturer and to a higher dose than the dose recommended for BD (200 mg/d). Subsequently, the patient was rechallenged with LTG by an allergist at a dose of 5 mg/d for 3 d with the dose increasing by 5–10 mg/d every 3 d for 3 wk; and then in 25 mg/d increments up to 300 mg/d. The maculopapular pruritic rash re-appeared. The rash did not improve after decreasing LTG down to 150 mg/d and the addition of 20 mg/d prednisone.
In the largest reported case series of epilepsy patients, 52/688 patients (8%) developed a rash (P-Codrea Tigaran et al., 2005). Twelve patients developed a rash that was deemed coincidentally associated with the initiation of LTG therapy. In these cases, LTG was continued with success without intermission. The other 40 patients had a rash that was deemed a ‘genuine rash’ which was induced by LTG. Nineteen of the 40 patients with LTG-induced rash were rechallenged with LTG, with a success rate of 84%. The dosage schedule was: 5 mg every day or every second day for 14 d, and increased by 5 mg every 14 d. After achieving the daily dosage of 25 mg/d, the up-titration was completed following the PDR recommendations (PDR, 2007).
In another case series of adult epilepsy patients, 3/13 patients with allergic skin responses to LTG were maintained non-stop on LTG with the concomitant use of anti-allergic medications, and their rash gradually disappeared (Huang et al., 2002). In the same series, three other patients agreed to be rechallenged with a dose of 12.5 mg/d LTG. There was no recurrence of the allergic reactions.
In a third case series of adult epilepsy patients, eight patients were rechallenged with LTG following the initial development of skin rash (Tavernor et al., 1995); (seven had eventual positive results on rechallenge). After re-introduction of LTG, six of the eight patients (75%) had no recurrence of the rash. Of the remaining two patients, one developed a mild fluctuating and qualitatively different skin rash on rechallenge and was not rechallenged again. On the other hand, the other patient developed a dose-related rash necessitating dose reduction from 150 mg/d to 100 mg/d, and the rash disappeared. The first attempt after 10 months to re-increase the LTG dose to 125 mg/d resulted in reappearance of the rash. The rash, again, disappeared on reduction to 100 mg/d. Nonetheless, 4 months later, the dose was again increased to 125 mg/d with success and without further development of rash, and then was increased over 4 months to 250 mg/d without adverse effects. The authors suggested that patients who experience a rash, but have a good therapeutic response to LTG, might be considered for re-dosing (Tavernor et al., 1995).
Another paper reported case reports of two adult epilepsy patients on LTG and valproate that were both successfully rechallenged with LTG without untoward consequences (Gelisse and Crespel, 2006). LTG was re-introduced at a dose of 5 mg/d with 5 mg/d increment every 2 wk to a dose of 50 mg/d.
Finally, a series of seven children and adolescents who developed mild rash on LTG (out of 150 on the medicine) were successfully rechallenged with LTG with 100% rate of non-recurrence of rash (Besag et al., 2000). This group discontinued LTG immediately upon rash occurrence, waited for at least 6 wk, and re-started LTG at a very slow titration schedule of 0.1 mg/d dose (see Table 2). LTG was then titrated to 12.5 mg/d in 12 wk, doubled every 2 wk to 50 mg/d, and increased in increments of 50 mg every 2 wk thereafter.
While LTG-induced rash is a potentially serious condition, clinicians should carefully evaluate the nature of the rash, and the risks and benefits of discontinuing potentially useful medication in patients with a chronic, severe and disabling condition. Currently, there are 39 reported cases in the literature of successful LTG rechallenge after a rash and five cases with rash recurrence. In spite of the limitations that there are a small number of cases found in the literature, and the fact that we cannot exclude the possibility of a publication bias, there are some characteristics of the rash that can help identify serious cases from benign ones. It is also clear that a high target dose and concomitant anticonvulsants play a major role in the outcome of rash recurrence. We suggest that a conservative and slower up-titration of LTG on rechallenge seems to be an optimal course for patients where the decision to rechallenge is made. The slow up-titration regimen, implemented by Besag et al. (2000), mentioned in Table 2 had a 100% success rate in children. Other authors (Eames, 1989; Smith and Newton, 1985) have also suggested that such slow titration desensitization regimes might be used to prevent rash developing with carbamazepine.
Pharmacokinetics and drug interactions
LTG is completely absorbed with minimal first-pass metabolism and its bioavailability is 98%. The half-life of LTG as monotherapy is around 25 h in healthy volunteers. In-vitro studies show that LTG is ∼55% bound to plasma proteins but as it is not highly protein bound, drug interaction at that site is unlikely (PDR, 2007).
The clearance of LTG is affected by the co-administration of certain medications. Because LTG is metabolized predominantly by glucuronic acid conjugation, medications that induce or inhibit glucuronidation may affect the clearance of LTG. Therefore, carbamazepine, phenytoin, phenobarbital, primidone, oestrogen-containing oral contraceptives and rifampin increase the clearance of LTG, and the mean half-life of LTG is reduced to 14 h when given with carbamazepine or phenytoin. On the other hand, the mean half-life is increased to 70 h when co-administered with valproate. Accordingly, when LTG is co-administered with valproate, LTG must be given at a reduced dosage, or the risk of rash could be increased by co-administration of LTG and valproate (PDR, 2007). Therapeutic plasma concentration of LTG is unknown, and according to the manufacturer, dosing should be based on therapeutic response.
The complicated pharmacokinetics mentioned above make the up-titration of LTG among the most complex of all psychiatric medications, espcially when co-adminstrated with the above-mentioned medications that affect the half-life. Clinical trials have suggested that 200 mg/d is the target dose of LTG when used without medication that considerably affects the half-life (Bowden et al., 1999; Calabrese et al., 1999). LTG is considered to be toxic at ∼15 mg/l (Devulder, 2006). In studies of epilepsy and neuropathic pain, a daily oral dose has ranged from 300 to 400 mg, although doses up to 800 mg were administered resulting in a plasma level below 15 mg/l (Wolf et al., 2006). Surprisingly, LTG-associated dermatological events (DEs) that can be life-threatening do not correlate with plasma drug levels. However, exceeding the recommended initial dosage or dosage escalation of LTG may, although yet to be proven, increase the risk of rash (PDR, 2007).
Characteristics of benign vs. serious rash
Of all the psychotropic medications currently available, the mood-stabilizing agents, including LTG, have the highest incidence of severe and life-threatening adverse DEs (Warnock and Morris, 2003). However, more patients taking carbamazepine were discontinued from studies because of rash than were those taking LTG (Brodie et al., 1995, 1999). The clinical presentation is highly variable and ranges from the most common transient and benign erythema that occurs 6–9 d after the introduction of a new drug, to the rarer and most severe forms that affect 1/1000 to 1/10 000 users (Mockenhaupt et al., 2005; Roujeau, 2005). Specifically, the DEs for mood-stabilizing agents can be divided into four groups (modified from Warnock and Morris, 2003):
The most common of all the DEs is a hypersensitivity reaction that can occur with most medications. This can present with fixed eruptions, pruritis, exanthematous, urticaria, and oedema.
Non-life-threatening DEs are more commonly associated with mood stabilizers, e.g. photosensitivity, alopecia, and drug-induced pigmentation. This is a heterogeneous group that has different pathophysiologies, the details of which are beyond the scope of this paper.
Severe and life-threatening DEs, namely erythema multiforme, Stevens–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), severe hypersensitivity reactions, and exfoliative dermatitis.
Exacerbation of common skin disorders such as acne, psoriasis, seborrhoea, and hyperhydrosis.
Even though there are some overlapping or unclassifiable presentations, it is important for clinicians to recognize and differentiate the different DEs as their prognosis varies. For instance, bullous fixed drug eruptions are extremely benign. On the other hand, the mortality rate is 5% for acute generalized exanthematous pustulosis, 10% for SJS, and up to 45% for TEN (Roujeau, 2005; Wolf et al., 2006).
Typically, the rash associated with LTG occurs between day 5 and week 8 of administration (Calabrese et al., 2002). Most of these are simple benign morbilliform ones (both pruritic and non-pruritic) and occur at a rate of 5–15%, which is a slightly higher rate than placebo-associated rash (5–10%) (Calabrese et al., 2002). While LTG-associated rash is not dose-dependent, it does seem that it is intimately related to the titration rate (Steiner et al., 1997). A minority of patients may develop severe DEs to LTG, e.g. erythema multiforme, SJS, TEN, or severe drug hypersensitivity syndrome with estimates varying from 0.1% to 1% (Calabrese et al., 2002; Warnock and Morris, 2003). It is important to note that early systemic manifestations (e.g. fever, malaise, lymphadenopathy) may be present even though a rash is not evident (Calabrese et al., 2002). Serious rash associated with hospitalization occurred in 0.3% (11/3348) of adult patients in the pre-marketing clinical trials of LTG in treatment of epilepsy. These trials were in the 1980s before precautions in the titration rate were being used (Calabrese et al., 2002). Risk factors associated with the development of the rash were: co-therapy with cytochrome P450 inhibitors (e.g. valproate), rapid dose escalation, lack of patient education regarding potential DEs, young age (<12 yr), and prior rash with other anticonvulsants. When modifiable risk factors were corrected and non-modifiable ones heeded, the incidence of the rash decreased in the 1990s (Huang et al., 2002).
It is important to recognize the fact that there is a discrepancy in the incidence of rash in BD patients compared with patients treated for epilepsy. This may partly reflect the changes in prescribing and dosing in BD patients which were informed following the initial reports of rash in patients with epilepsy (Calabrese et al., 2002; PDR, 2007). In BD clinical trials conducted in the 1990s and 2000s, the rate of serious rash in adult patients receiving LTG monotherapy was 0.08% (1/1233) (Calabrese et al., 2002) compared to 0.3% (11/3348) of adult patients in the clinical trials of LTG in treatment of epilepsy mentioned above. In the BD clinical trials where LTG was used as an adjunctive therapy, the rate of serious rash was 0.13% (2/1538) (Ketter et al., 2006). No fatalities occurred among these individuals. This lower incidence probably reflects improved clinical practice, but might also be a different pathophysiology in epilepsy than BD patients.
Differential diagnosis of LTG-related severe DEs
One way to conceptualize severe DEs associated with LTG and other anticonvulsants is by considering two dimensions – the degree of internal organ involvement as one dimension and the degree of cutaneous involvement as another. Anticonvulsant hypersensitivity syndrome (AHS) involves both, while erythema multiforme involves predominantly the skin (with <10% epidermal detachment). Both SJS and TEN involve both dimensions, with greater epidermal detachment in SJS than erythema multiforme, and even more in TEN (Warnock and Morris, 2003).
Anticonvulsant hypersensitivity syndrome
AHS is defined by Kaur and colleagues as an acute, life-threatening, idiosyncratic drug reaction seen with certain anti-epileptic drugs (Kaur et al., 2002). It is a type of drug hypersensitivity reaction and is more likely to occur with anticonvulsants. Carbamazepine, oxcarbazepine, valproate, and LTG are some of the medications implicated and cross-reactivity is about 50–80% (Karande et al., 2006; Kaur et al., 2002). It is more common in anticonvulsants that have an aromatic benzene ring (e.g. carbamazepine, oxcarbazepine) and its incidence is about 1/1000 to 1/10 000 exposure (Kaur et al., 2002).
AHS usually presents with fever, rash, and internal organ involvement (lymphadenopathy, hepatitis, and haematological abnormalities). Typically, there is no mucous membrane involvement (Vittorio and Muglia, 1995). Of note, liver failure due to fulminant hepatitis is the most common cause of death from AHS with a mortality rate of 10–40% in untreated cases (Vittorio and Muglia, 1995; Warnock and Morris, 2003).
Erythema multiforme is a cutaneous eruption of pathognomonic target lesions and gradually increasing mucocutaneous and systemic involvement (epidermal detachment up to 10%). It is associated with carbamazepine, oxcarbazepine, valproate, gabapentin, and LTG use.
SJS has also been associated with carbamazepine, oxcarbazepine, valproate, gabapentin, and LTG use. It occurs in ∼2/10 000 adults and 4/10 000 children (Sadock and Sadock, 2005). SJS is characterized by the onset of high fever and ‘flu-like symptoms followed by skin erythema, tenderness, and most importantly, cutaneous and mucosal exfoliation (incidence of 10–30%). SJS is progressive and associated with up to 10% mortality rate, mainly due to sepsis (Warnock and Morris, 2003).
Toxic epidermal necrolysis
TEN is an extreme variant of SJS that is associated with epidermal detachment in more than 30% of cases. It presents similarly to SJS, but is more rapidly progressive (over 1–5 d). Common offenders include carbamazepine, oxcarbazepine, valproate, and LTG. TEN is associated with up to 45% mortality rate, mainly due to sepsis originating in denuded skin and lungs (Warnock and Morris, 2003).
Management of LTG rash
The manufacturer's recommendations specify that LTG should be discontinued at the first sign of a rash, regardless of its type and severity, unless the rash is clearly not drug related (PDR, 2007). This practice ensures that LTG is discontinued in instances of serious rash, but may lead to unnecessary discontinuation of LTG in cases of non-serious rash. Such dermatological precautions recommend against starting LTG within 2 wk of a rash, vaccination, or viral illness. There are recommendations to avoid new cosmetics, foods, fabric softeners, poison ivy, poison oak, or sunburn for 3 months after starting LTG.
A rash occurring in the first 5 d of therapy is probably non-drug related (except for hypersensitivity reaction). The patient should stop the next dose and contact the physician. The rash is most likely benign if it has the following characteristics (Calabrese et al., 2002):
peaks within days and settles in 10–14 d;
spotty, non-confluent, non-tender;
absence of systemic features;
complete blood count, liver function tests, blood urea, serum creatinine, and urine analysis are within normal limits.
In such case, LTG dose should be reduced. If the rash is pruritic, antihistaminics or topical corticosteroids can be prescribed. The patient should be monitored closely, and LTG discontinued if the rash worsens or new symptoms emerge. The patient may be rechallenged at a lower dose (5–12.5 mg/d) and with slower titration (P-Codrea Tigaran et al., 2005; Tavernor et al., 1995) or even slower in children and adolescents (Besag et al., 2000).
A rash occurring between 5 d and 8 wk is probably LTG related. The patient should stop taking LTG and contact the physician immediately. Features suggestive of the development of severe DEs include:
confluent and widespread rash;
purpuric and tender rash;
associated fever, malaise, pharyngitis, lymphadenopathy, or anorexia;
abnormalities in the laboratory test values mentioned above;
any involvement of eyes, lips, mouth, or other mucous membranes;
prominent involvement of neck or upper trunk.
In these circumstances, LTG should be discontinued with close monitoring for internal organ involvement; hospitalization should be seriously considered, and immediate dermatological consultation obtained. The patient should not be rechallenged with LTG. In case of TEN, the patient should wear a medical alert bracelet to ensure that the medication is never re-introduced.
Although there are rare but fatal types of rash that can develop on LTG which should be carefully teased out and closely supervised, rechallenge in bipolar depression is an under-recognized and under-utilized option in our clinical armamentarium. Given the fact there is very limited medication that can help in treatment maintenance and decreasing the incidence of episodes in BD, re-introduction of LTG can be cautiously recommended in certain cases with close observation and after thorough evaluation of the risks and benefits. These cases might be selected based on characteristics of rash. In addition, a very slow up-titration seems to be crucial, based on this literature review, for the success of the rechallenge. Moreover, a high target dose and concomitant anticonvulsants play a major role in the outcome of rash recurrence. Nonetheless, larger studies are needed to better guide decision-making in these difficult clinical dilemmas, especially given the small number of cases in the literature that are available for review.
This publication was made possible by support from NARSAD (Z.B.), the National Institutes of Health (1K23MH077914: Z.B.) and the CSTA grant number UL1 RR024139 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. We thank Mrs Susan Youssef for the grammatical review of the manuscript, and Dr Walid Rahhal for the translation of a French paper.
Statement of Interest
Dr Youssef has received honoraria for presentations and is on the speakers' panel of AstraZeneca, Bristol–Myers Squibb, and Pfizer Pharmaceuticals; and has received research support from Avanir Pharmaceutical. Dr Bhagwagar is on the speakers' panel for AstraZeneca, Bristol–Myers Squibb, and Janssen Pharmaceuticals; and has received research support from Bristol–Myers Squibb.
(2003). A placebo-controlled 18-month trial of lamotrigine and lithium maintenance treatment in recently manic or hypomanic patients with bipolar I disorder. Archives of General Psychiatry 60, 392–400.doi:10.1001/archpsyc.60.4.392pmid:12695317
(1999). Multicentre, double-blind, randomised comparison between lamotrigine and carbamazepine in elderly patients with newly diagnosed epilepsy. The UK Lamotrigine Elderly Study Group. Epilepsy Research 37, 81–87.doi:10.1016/S0920-1211(99)00039-Xpmid:10515178
(1995). Double-blind comparison of lamotrigine and carbamazepine in newly diagnosed epilepsy. UK Lamotrigine/Carbamazepine Monotherapy Trial Group. Lancet 345, 476–479.doi:10.1016/S0140-6736(95)90581-2pmid:7710545
(2003). A placebo-controlled 18-month trial of lamotrigine and lithium maintenance treatment in recently depressed patients with bipolar I disorder. Journal of Clinical Psychiatry 64, 1013–1024.pmid:14628976
(2003). Depression and bipolar support alliance consensus statement on the unmet needs in diagnosis and treatment of mood disorders in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 42, 1494–1503.doi:10.1097/00004583-200312000-00017pmid:14627885
(2006). Reintroduction of treatment with lamotrigine in combination with valproate after an initial allergic skin reaction [in French]. Revue Neurologique 162, 1122–1124.doi:10.1016/S0035-3787(06)75125-6pmid:17086149
(2006). The effect of dermatologic precautions on the incidence of rash with addition of lamotrigine in the treatment of bipolar I disorder: a randomized trial. Journal of Clinical Psychiatry 67, 400–406.pmid:16649826
(2003). A re-evaluation of the role of antidepressants in the treatment of bipolar depression: data from the Stanley Foundation Bipolar Network. Bipolar Disorders 5, 396–406.doi:10.1046/j.1399-5618.2003.00065.xpmid:14636363
(2006). Drug rash with eosinophilia and systemic symptoms versus Stevens–Johnson Syndrome – a case that indicates a stumbling block in the current classification. International Archives of Allergy and Immunology 141, 308–310.doi:10.1159/000095437pmid:16940741