Toxic epidermal necrolysis and Stevens-Johnson syndrome

Initial symptoms of toxic epidermal necrolysis (TEN) and Stevens Johnson Syndrome (SJS) can be unspecific and include symptoms such as fever, stinging eyes and discomfort upon swallowing. Typically, these symptoms precede cutaneous manifestations by a few days. Early sites of cutaneous involvement are the presternal region of the trunk and the face, but also the palms and soles. Involvement (erythema and erosions) of the buccal, genital and/or ocular mucosa occurs in more than 90% of patients, and in some cases the respiratory and gastrointestinal tracts are also affected [20, 21]. Ocular involvement at the onset of disease is frequent, and can range from acute conjunctivitis, eyelid edema, erythema, crusts, and ocular discharge, to conjunctival membrane or pseduomembrane formation or corneal erosion, and, in severe cases, to cicatrizing lesions, symblepharon, fornix foreshortening, and corneal ulceration [22, 23]. The severity of acute ocular manifestation is not however predictive of late complications [24]. The morphology of early skin lesions includes erythematous and livid macules, which may or may not be slightly infiltrated, and have a tendency to rapid coalescence (Table 1). The above mentioned skin signs associated with mucosal involvement are clear danger signs and warrant the initiation of rapid diagnostic confirmation with immediate cryosections of a skin biopsy. Histological examination including direct immunofluorescence analysis of the skin biopsy is also important in order to rule out differential diagnoses such as autoimmune blistering diseases, bullous fixed drug eruption, acute generalized exanthematic pustulosis, and due to its rarity in adults, to a lower extend staphyloccocal scalded skin syndrome.

In a second phase, large areas of epidermal detachment develop. In the absence of epidermal detachment, more detailed skin examination should be performed by exerting tangential mechanical pressure on several erythematous zones (Nikolsky sign). The Nikolsky sign is positive if mechanical pressure induces epidermal detachment, but is not specific for TEN or SJS, as it can also be positive in, for example, autoimmune bullous skin diseases.

The extent of skin involvement is a major prognostic factor. It should be emphasized that only necrotic skin, which is already detached (e.g. blisters, erosions) or detachable skin (Nikolsky positive) should be included in the evaluation of the extent of skin involvement. Bastuji-Garin et al. proposed classifying patients into three groups according to the degree of skin detachment (Table 1, Figure 1) [1].

Sequelae are common features of late phase TEN. According to the study of Magina et al [25] following symptoms are found: hyper- and hypopigmentation of the skin (62.5%), nail dystrophies (37.5%), and ocular complications. According to a study of Yip et al. 50% of patients with TEN develop late ocular complications including, by order of decreasing frequency, severe dry eyes (46% of cases), trichiasis (16%), symblepharon (14%), distichiasis (14%), visual loss (5%), entropion (5%), ankyloblepharon (2%), lagophthalmos (2%), and corneal ulceration (2%) [24]. Hypertrophic scars are only seen in very few patients [26]. Long-term complications of mucosal involvement occur in 73% of patients who present mucosal involvement in the acute phase, and the mucosal sequelae involve mainly the oral and oesophageal mucosa, and to a lesser extent lung and genital mucosa [27]. In a small post SJS/TEN study seven out of nine patients had either xerostomia or keratoconjunctivitis or both, resembling Sjögren-like syndrome [28]. Additionally, another group reported a patient with Sjögren-like pluriglandular exocrine insufficiency including exocrine pancreatic impairment [29].

Genetic factors associated with drug hypersensitivity are a complex issue that has been studied in different populations and a variety of ethnic backgrounds. A unique and strong association between HLA, drug hypersensitivity and ethnic background was discovered by Chung et al. who showed a strong association in Han Chinese between the HLA-B*1502, SJS and carbamazepine [30]. This high association with an odds ratio of 2504 led to further studies in a similar ethnical group of Hong Kong Han Chinese with severe adverse reactions to antiepileptic drugs [31]. Another study confirmed the susceptibility of individuals with HLA-B*1502 to carbamazepine in a Thai population [32]. A smaller Indian based study however showed only a weak correlation between HLA-B*1502 and carbamazepine induced severe drug allergy. A genetic correlation could not however be shown in Japanese or Europeans [33‐35]. Indeed, in a large European study (RegiSCAR), HLA-B genotyping was performed in patients with severe cutaneous adverse reactions caused by the two previously mentioned drugs (carbamazepine, allopurinol) and another three high risk drugs (sulfamethoxazole, lamotrigine, NSAID's of oxicam-type). This RegiSCAR study revealed that HLA-B*1502 is neither a marker for carbamazepine, sulfamethoxazole, lamotrigine, or NSAID's of oxicam-type induced SJS/TEN nor a sufficient explanation for the cause of the disease in Europeans [35, 36]. This leads to the conclusion that this genetic constellation (HLA-B*1502) is not a population independent marker for SJS/TEN in carbamazepine exposed individuals. Severe cutaneous reactions in HLA-B*1502 subjects were not only associated with the drug carbamazepine, but also, to a lesser extent (lower odds ratio), with phenytoin and lamotrigine [31].

A second strong association between HLA genotype and SJS/TEN has been reported for allopurinol. Indeed, 100% of Han Chinese patients with a severe adverse drug reaction to allopurinol were HLA-B*5801 positive [37]. Subsequently, a strong association between SJS/TEN and HLA-B*5801 was found in Japanese patients [34], Thai patients [32], and also, to a lesser extent (55% of cases), in patients of European origin [36].

The pathogenesis of SJS/TEN is not fully understood but is believed to be immune-mediated, as re-challenging an individual with the same drug can result in rapid recurrence of SJS/TEN [38, 39]. The histopathology of SJS/TEN lesions show that keratinocyte apoptosis followed by necrosis is the pathogenic basis of the widespread epidermal detachment observed in SJS/TEN. The clinical, histopathological and immunological findings in SJS/TEN support the currently prevalent concept, that SJS and TEN are specific drug hypersensitivity reactions in which cytotoxic T lymphocytes (CTL) play a role in the initiation phase. Indeed, in the early phase of disease, blister fluid contains mainly cytotoxic CD8+T lymphocytes [40, 41], suggesting that a major histocompatibility (MHC) class-I restricted drug presentation leads to clonal expansion of CD8+ CTLs, and the subsequent - to date only incompletely understood - immune reaction that causes SJS/TEN. These CD8+ T cells express common cutaneous leukocyte antigen (CLA) and are negative for CD45RA and CD28. Nassif et al. were able to demonstrate that blister T cells from patients exert drug specific cytototoxic activity against both autologous B-lymphocyte cell lines and keratinocytes [42], and furthermore demonstrated that this cell-mediated cytotoxicity was mediated by granzyme B. The discrepancy between the paucity of the infiltration of immune cells (including CTLs) in the skin of patients with SJS/TEN and the overwhelming keratinocyte apoptosis has however lead to the search for cytotoxic proteins and/or cytokines that may "amplify" the extent of keratinocyte apoptosis that CTLs alone could induce upon cell-cell contact. To date, the strongest evidence suggests a key contribution of the cytotoxic molecules FasL and granulysin as molecules responsible for the disseminated keratinocyte apoptosis in SJS/TEN [43, 44].

The role of the membrane form of the death ligand FasL and its cognate death receptor Fas in the signalling that triggers keratinocyte apoptosis is supported by research performed using an ex-vivo experimental set up with TEN lesional skin biopsy cryostat section overlays with Fas-expressing lymphoid target cells [44]. However, the functional relevance of up-regulated keratinocyte membrane FasL, and thus its ability to induce keratinocyte cell death, has been questioned by some as the above ex-vivo demonstration of the lytic ability of keratinocyte FasL in TEN was limited in its effect on lymphoid target cells and not demonstrated with keratinocytes as target cells. It is well known that primary keratinocytes are sensitive to the cytolytic effect of FasL in vitro, and this sensitivity can be further enhanced by interferon gamma, a cytokine known to be present in the skin during TEN [45‐47]. However, it is still not fully understood what causes the up-regulation of FasL/Fas on keratinocytes, and how the immune system, including T cells found in blister fluid at the onset of disease may regulate this.

The role of soluble FasL (sFasL) in SJS/TEN remains controversial. It appears clear now that increased levels of sFasL can be found in the serum of patients with SJS/TEN, and levels of sFasL are consistently elevated when analysis is performed preceding skin detachment [48]. Soluble FasL as opoposed to membrane-bound FasL is, however, very poorly cytolytic, and it is therefore unlikely to be a cause of keratinocyte apoptosis in TEN [49]. Nevertheless, one study showed that sera of SJS/TEN were able to induce abundant keratinocyte apoptosis and furthermore that peripheral blood monuclear cells of patients stimulated by the causative drug excreted high levels of sFasL [50], but it should be noted that sera can contain small membrane vesicles with membarne bound FasL that can account for the observed activity.

Gene expression analysis of blister fluid cells, and analysis of blister fluid from patients with SJS/TEN has also recently identified secretory granulysin (a cationic cytolytic protein secreted by CTL's, NK cells and NKT cells) as a key molecule responsible for the induction of keratinocyte death in TEN [43]. Blister fluid cells express high levels of granulysin mRNA, the protein is found in increased concentrations in blister fluid, and most importantly, recombinant granulysin mimicks features of SJS/TEN when injected intradermally in mice. The finding that elevated serum granulysin levels apparently discriminate between serious and non-blistering adverse drug reactions, serum granulysin levels being normal in the latter, lends further support an important role of granulysin in SJS/TEN [51].

In conclusion, and based on our knowledge to date, CD8 T-cells as well as the cytolytic molecules FasL and granulysin are key players in the pathogenesis of SJS/TEN. How a culprit drug in a given patient who will develop SJS/TEN regulates the function of these key players is the subject of ongoing research.

Drug exposure and a resulting hypersensitivity reaction is the cause of the very large majority of cases of SJS/TEN. In absolute case numbers, allopurinol is the most common cause of SJS/TEN in Europe and Israel [52], and mostly in patients receiving daily doses of at least 200 mg.

In a case control study of patients hospitalized for SJS/TEN in selected hospitals in France, Germany, Italy and Portugal between 1989 and 1993, Roujeau et al. reported that the following drugs are at increased risk of inducing SJS/TEN when used over a short period of time: trimetroprim-sulfamethoxazole and other sulfonamide-antibiotics, aminopenicillins, cephalosporins, quinolones and chlormezanone. Among drugs usually taken for longer periods of time (carbamazepine, phenytoin, phenobarbital, valproic acid, non-steroidal antinflammatory drugs of the oxicam-type, allopurinol and corticosteroids), the highest risk of induction of SJS/TEN occurs during the first 2 months of treatment with a sharp drop of incidence thereafter [8]. However, although these drugs have a high relative risk compared to other drugs, the actual risk remained low with 5 cases or less per million users per week. A similar population was studied between 1997 and 2001 by Mockenhaupt et al. in a multinational case-control study in Europe covering more than 100 million inhabitants in which special attention was given to newly marketed drugs [53]. This study identified nevirapine, lamotrigine, and sertraline, as drugs with a significantly increased risk of inducing SJS/TEN. Older drugs identified as having a high risk of inducing SJS/TEN were sulfamethoxazol/trimethoprim (SMX/TMP), sulfonamides (sulfasalazine, sulfadiazine, sulfadoxine, sulfafurazole), allopurinol, carbamazepine, phenytoin, phenobarbital, and NSAID's of the oxicam-type (meloxicam, piroxicam, tenoxicam). However the incidence of SJS/TEN under treatment with valproic acid is confounded by the concomitant use of other drugs, such as lamotrigine [5]. Mockenhaupt et al. were able to show that almost all cases of SJS/TEN developed within 63 days of starting use of antiepileptic drugs, and that the risk of developing SJS/TEN per 10 000 new users was significantly increased for carbamazepine (1.4 cases per 10 000 users), lamotrigine (2.5), phenobarbital (8.1) and phenytoine (8.3). The incidence for valproic acid was low compared to other antiepileptic drugs with 0.4 cases per 10 000 users [54]. Furthermore, studies in different populations indicate that the risk of developing SJS/TEN is highest when the drug has been recently initiated, and subsequently declines within 8 weeks or more of administration [5, 55]. Interestingly the long term use of glucocorticosteroids for a variety of diseases does not change the incidence of the occurrence of SJS/TEN for the incriminated drugs, but it appears that glucocorticoids lengthen the interval between the beginning of the intake of the drug and onset of SJS/TEN [56]. A recent survey of TEN in children identified similar drugs to those in adults, as well as a possibly increased susceptibility to acetaminophen (paracetamol) [57].

Photo-induced TEN or SJS are only reported in very rare cases. Case reports exist for hydroxchloroquine [58], naproxene [59] and clobazam [60]. An often addressed issue is the induction of TEN or SJS after vaccination. The vaccine adverse event reporting system concludes that despite the plausibility of a relationship between vaccination and SJS/TEN, the very small number of reports compared to the large amount of vaccinations and the benefits of vaccinations outweighs the potential risk of SJS/TEN [61].

The diagnosis relies on the one hand on clinical symptoms and on the other hand on histological features. Typical clinical signs initially include areas of erythematous and livid macules on the skin, on which a positive Nikolsky sign can be induced by mechanical pressure on the skin, followed within minutes to hours by the onset of epidermal detachment characterized by the development of blisters. It should be noted, however, that the Nikolsky sign is not specific for SJS/TEN. Mucosal, including ocular, involvement develops shortly before or simultaneously with skin signs in almost all cases. To distinguish SJS, SJS-TEN and TEN the surface area of the detachment is the main discriminating factor (Figure 1). Histological work up of immediate cryosections or conventional formalin-fixed sections of the skin revealing wide spread necrotic epidermis involving all layers confirms the diagnosis. In order to rule out autoimmune blistering diseases, direct immune fluorescence staining should be additionally performed and no immunoglobulin and/or complement deposition in the epidermis and/or the epidermal-dermal zone should be detected.

Major differential diagnosis of SJS/TEN are autoimmune blistering diseases, including linear IgA dermatosis and paraneoplastic pemphigus but also pemphigus vulgaris and bullous pemphigoid, acute generalized exanthematous pustulosis (AGEP), disseminated fixed bullous drug eruption and staphyloccocal scalded skin syndrome (SSSS). SSSS was one of the most important differential diagnoses in the past, but the incidence is currently very low with 0.09 and 0.13 cases per one million inhabitants per year [54].

Management in the acute stage involves sequentially evaluating the severity and prognosis of disease, prompt identification and withdrawal of the culprit drug(s), rapidly initiating supportive care in an appropriate setting, and eventual "specific" drug therapy as described in detail below.

Due to the often combined involvement of the skin, eyes and mucous membranes (oral, gastrointestinal, pulmonary, genital, as well as urinary), the follow up and treatment of sequelae should be interdisciplinary. Special attention should be given to the prevention of ocular complications. Early referral to an ophthalmologist is mandatory for assessment of the extent of eye involvement and prompt treatment with topical steroids. Visual outcome is reported to be significantly better in patients who receive specific ophtalmological treatment during the first week of disease [23]. Some of the ocular complications have an inflammatory background and have to be treated occasionally with ophthalmic steroids and/or extensive lubrication of the eye [26] in order to prevent progression leading ultimately to the need for corneal transplantation. A small single retrospective study with IVIG showed no significant effect on ocular complications in frequency and severity, but the power of the study was weak [99]. The benefit of local antibiotic treatment (ointments) is not clear. Yip et al. have reported that the use of local antibiotic treatment leads to more late complications, including, for example, dryness of the eyes [24]. Hypopharyngeal stenosis combined with dysphagia and oesophageal strictures are long-term complications which are difficult to treat [100, 101] and may require laryngectomy.

A detailed drug history is very important when striving to identify the culprit drug in SJS/TEN. In some cases several drugs are possible candidates and allergological testing can be of help in identifying the most likely candidate. In principle, the severity of SJS and TEN does not allow re-challenge and intradermal testing with the culprit drugs due to the feared risk of re-inducing a second episode of SJS/TEN, although two case reports describe intradermal testing without triggering of a second episode of TEN [102, 103]. Induction of SJS/TEN has, however, been documented following local eye treatment [104, 105].

Patch-testing is an investigational option, but not a routine diagnostic option at the moment. Data from Wolkenstein et al. has shown that low sensitivity is a problem with patch testing in SJS/TEN, as only two of 22 tested patients had a relevant positive patch test [106].

Currently the focus of allergological testing lies more on ex vivo/in vitro tests. The lymphocyte transformation test (LTT), that measures the proliferation of T cells to a drug in vitro has shown a sensitivity of 60-70% for patients allergic to beta-lactam antibiotics [107]. Unfortunately the sensitivity of the LTT is still very low in SJS/TEN even if performed within one week after onset of the disease [108].

Another recently reported approach looks for up-regulation of CD69 on T-lymphocytes two days after lymphocyte stimulation in vitro as a sign of drug hypersensitivity [109]. Novel in vitro methods to help identify culprit drug in SJS/TEN are still needed [110].