Erythema Multiforme

Peter Fritsch Ramon Ruiz-Maldonado

The erythema multiforme group of diseases encompasses a number of acute self-limited
exanthematic intolerance reactions that share at least two characteristic features: target
lesions (stable circular erythemas or urticarial plaques with areas of blistering, necrosis,
and/or resolution in a concentric array) and, histologically, satellite-cell (or more
widespread) necrosis of the epidermis. These features are the expression of an archetypic
polyetiologic reaction pattern of the skin (i.e., a cytotoxic immunologic attack on
keratinocytes expressing non-self antigens). Antigens involved are predominantly
microbes (viruses) or drugs.
At present, two main subsets are recognized: (1) erythema multiformea fairly common,
usually mild and relapsing eruption that is most often triggered by recurrent herpes
simplex virus (HSV) infection; and (2) the Stevens-Johnson syndrometoxic epidermal
necrolysis complex (SJSTEN)an infrequent severe mucocutaneous intolerance
reaction that is most often elicited by drugs. Entities that may be grouped very close to
the erythema multiforme spectrum are the acute graft-versus-host disease, fixed drug
eruption, and erythema multiformelike disseminated allergic contact dermatitis. The
exact relationship among these entities is still a matter of controversy, as are aspects of
therapy and management. The pathogenesis is only incompletely understood.
CLASSIFICATION AND HISTORY
No generally accepted classification of the erythema multiforme spectrum exists.
Morphology remains the predominant basis for disease definitions, which have been the
arena of terminologic disputes for more than a century. Lumping and splitting of entities,
introduction of new terms, continued use of discarded ones, and too liberal usage of the
term erythema multiforme for ill-defined cutaneous eruptions have led to terminologic
and conceptual confusion that leaves most of the traditional terms without an unequivocal
meaning and jeopardizes the accessibility of the erythema multiforme literature. The
confusion can be disentangled only through a review of the historical development.
Descriptions of erythema multiforme date back to antiquity (Celsus). Bateman first
described the target lesion in 1814, and Bazin first recognized mucosal involvement and
constitutional symptoms in 1862. It was von Hebra who noted the multiform character of
erythema multiforme and grouped several then-recognized disorders into a single entity,
which he christened erythema exudativum multiforme. Von Hebra's description exactly
fits the appearance of what is now known as HSV-associated erythema multiforme,
except that he made no reference to oral mucosal involvement. Stevens and Johnson
described a severe mucosal disease with purplish cutaneous macules and necrotic
centers in 1922. In 1950, Thomas proposed that erythema multiforme and the StevensJohnson syndrome (SJS) were variants of the same pathologic process, differing only in
severity, and that they should be termed erythema multiforme minor and erythema
multiforme major, respectively. Although this suggestion was broadly accepted, the term

Stevens-Johnson syndrome continued to be used as a synonym for erythema multiforme

major.
In 1956, Lyell described TEN, again as a separate clinical entity (Lyell's syndrome), as a
life-threatening, rapidly evolving mucocutaneous reaction characterized by erythemas,
necrosis, and bullous detachment of the epidermis resembling scalding. 1 Lyell's series
included not only patients suffering from what would still be called TEN today, but also
others with staphylococcal scalded-skin syndrome and generalized fixed drug eruption; 2
thus, the series fueled terminologic disputes for many years. In the following decades, the
notion gained ground that, in most or even in all instances, TEN was equivalent to SJS of
maximal severity. Lyell himself suggested that the term toxic epidermal necrolysis be
abandoned, 3 but its use has been retained.
Clearly, the erythema multiforme spectrum has for decades been considered to include
the mild minor and the less common, but more severe, major types of the disorder
the distinctive mark being the presence of mucous membrane involvementsurmounted
by the rare devastating TEN. All these categories were interpreted as polyetiologic, with
an almost unlimited number of potential causative factors held responsible. It was
thought that the milder forms were caused more often by infections; the more severe
ones, by drugs. These prototypes were theoretically well defined, but their distinction
often proved difficult in practice because of their overlapping features.
Important recent advances have, paradoxically, added to the confusion. It has become
clear that mild erythema multiforme is strongly linked to HSV infection. 4 However,
HSV-related erythema multiforme cannot simply be equated with erythema multiforme
minor, because it frequently involves mucosal sites (most often oral) that have
previously been attributed to erythema multiforme major. Labeling such cases
erythema multiforme major is again confusing, 5 , 6 because erythema multiforme
major is firmly engrained as a synonym for SJS.
Clinical and histopathologic observations, mostly by a team of French investigators, have
suggested that there are two main morphologic patterns within the erythema multiforme
spectrum. The investigators proposed that these patterns be considered different disorders
with distinct causes 5 , 6 and 7 ( Table 58-1): HSV-related erythema multiforme,
characterized by target lesions and a predominantly inflammatory histopathologic pattern,
and the primarily drug-related SJS-TEN complex, characterized by atypical or no target
lesions at all and a predominantly necrotic histopathologic pattern. Stevens-Johnson
syndrome and TEN were taken to differ only in the body surface area (BSA) involved
and the severity.
Consensus papers have provided clinical criteria for the entities of the erythema
multiforme spectrum that have been widely used in subsequent publications. The basis of
the proposed definitions for SJS-TEN is quantitative: 5 , 7 SJS represents cases of less
than 10 percent BSA involvement; TEN indicates more than 30 percent; and those in
between are labeled SJS-TEN overlap. Although somewhat artificial and not
particularly simple, this classification is definitely useful for epidemiologic purposes

(BSA is a prognostic factor in SJSTEN) and permits the determination of medication

risks. Also, it corresponds to a classification proposed earlier by one of us (R.R.M.). 8
Several problems remain to be clarified: (1) the issue of overlapping features has not been
resolved; 7 (2) the question of whether all instances of erythema multiforme that are not
SJS-TEN by clinical criteria are HSV-related has not been properly addressed; (3) it has
not been proven that all instances of extensive TEN, in fact, represent variants of SJS
(e.g., what has been called TEN without spots 5 ); and (4) the conditions characterized
by mucosal lesions only (formerly ectodermosis pluriorificialis) find no place in this
classification.
We deem it unwise to coin new terms before more is known about the molecular
pathophysiology of the conditions concerned. Therefore, we will use the term erythema
multiforme for all instances of the erythema multiforme spectrum that are not SJS-TEN,
and the term SJS-TEN complex for both SJS and TEN.
ERYTHEMA MULTIFORME

Definition
Erythema multiforme is a self-limited, usually mild and relapsing exanthematic reaction
of the skin that is etiologically most often related to recurrent HSV infection. It is
clinically characterized by target-shaped, urticarial plaques, and it frequently presents
with mucous membrane lesions.

Incidence and Epidemiology

Erythema multiforme is relatively common; it may account for up to 1 percent of
dermatologic outpatient visits. The exact incidence is difficult to assess because
practically all published series contain cases of SJS and/or erythema multiformelike
cases. Based on our own records, we estimate that erythema multiforme is at least 10
times as frequent as the SJS-TEN complex, for which an incidence of 1.9 cases/million
population/year has been reported in Germany. 9 Erythema multiforme may occur in
patients of all ages, but it occurs predominantly in adolescents and young adults; it is rare
both under the age of 3 years and over the age of 50 years; 75 percent of patients are
younger than 40 years. A slight preponderance has been reported both for females and
males. 10 , 11 There is no predominance for ethnic groups or geographic locations.
Seasonal clustering in spring is common (typus annuus), presumably caused by
ultraviolet light provocation of recurrent HSV infection. Erythema multiforme is often
recurrent, probably in the majority of the cases. 11 It is impossible at present to determine
how often erythema multiforme occurs as a truly single event. Recurrences may occur at
short intervals and reappear for many years.

Etiology
Although suspected since the early nineteenth century, it has been recognized only in
recent years that HSV infection is the dominant causative factor of erythema multiforme,
both in adults and in children. 4 , 10 , 12 Both HSV-1 and HSV-2 may trigger erythema
multiforme, whereas the closely related varicella-zoster virus shows no such association.

Evidence linking HSV to erythema multiforme is best for recurrent erythema multiforme,
as clinical lesions of recurrent HSV infection precede an outbreak of erythema
multiforme in about 80 percent of patients. 12 In situ hybridization and the polymerase
chain reaction have demonstrated HSV-DNA in the epidermal cells (not in the dermis) of
erythema multiforme lesions in up to 90 percent of patients, 13 , 14 even in individuals
without obvious preceding HSV infection. 15 In addition, immunofluorescence and
immune histochemistry have demonstrated HSV-specific antigens in lesional skin. Most
patients with recurrent erythema multiforme are seropositive for HSV, and it is
occasionally possible to recover HSV from circulating immune complexes. 16 Peripheral
blood mononuclear cells from patients with HSV infection and those with HSV-related
erythema multiforme exhibit a similarly skewed T cell receptor response on stimulation
with HSV antigen, indicating a specific immune response against HSV in both
conditions. 17 Proof for the pathogenetic relevance of HSV infection derives from the
fact that the prophylactic administration of acyclovir can effectively prevent recurrent
HSV-associated erythema multiforme; 18 this is also true for many patients with no
clinical evidence of HSV infection. There are no data at present, however, to prove that
nonrecurring erythema multiforme is as strongly related to HSV as recurring erythema
multiforme.
It is a matter of terminology (and therefore controversy) whether cases should be
classified as erythema multiforme if they fulfill the clinical criteria, but do not show
evidence of preceding HSV infection. There is little doubt that such eruptions exist, but it
is unclear how frequent they are; they may occur more often in nonrecurrent erythema
multiforme. The evidence for causative factors other than HSV infection is only
circumstantial. Published reports implicate hepatitis B and C virus, as well as other viral
infections. 19 , 20 Progesterone may elicit chronic recurrent erythema multiforme that
responds to tamoxifen treatment 21 and oophorectomy. Drugs are a rare cause of
erythema multiforme with mucous membrane lesions. 7 It may, of course, be argued
whether these eruptions were truly erythema multiforme or mere imitators; moreover,
subclinical HSV infection generally cannot be ruled out. Other problems are idiopathic
cases, in which neither HSV infection nor any other cause can be uncovered. Such cases
are fairly common under routine circumstances, but are even found in studies that
specifically looked for HSV infection. 21 Many such cases respond to prophylactic
acyclovir treatment and are thus likely to have been triggered by subclinical HSV
infection; some, however, are resistent. 21
Erythema multiformelike dermatitic eruptions result from contact sensitization to
sulfonamides, antihistamines, dinitrochlorobenzene (DNCB), diphencyprone (DPCP),
rosewood, Rhus, primula, tea tree oil, cutting oils, cinnamon, and other substances; these
rashes should be viewed as only imitators of erythema multiforme, despite clinical and
histopathologic similarities. 22

Pathogenesis
The current belief is that erythema multiforme is a cell-mediated immune reaction
leading to the destruction of keratinocytes expressing HSV antigens. However, many
aspects of the underlying pathomechanisms are not known.

Although erythema multiforme lesions can exhibit viral antigens and DNA, attempts at
viral cultures almost invariably fail, and electron microscopy cannot detect intact virions.
Therefore, erythema multiforme lesions are not sites of regular HSV replication, as can
be suspected from their histopathologic appearance. This paradox has been partially
resolved: 14 no viable HSV particles are present in epidermal keratinocytes; there are
only DNA fragments, most often encoding viral DNA polymerase. Fragments of DNA
persisted for 1 to 3 months in the epidermis after healing (as in acute HSV lesions); viral
polymerase RNA, however, was identified only in fresh, not in healed erythema
multiforme lesions. These data suggest that transcription and translation are limited, and
that the development of erythema multiforme lesions may be associated with the
expression of viral polymerase (and/or additional antigens), stimulating a specific
immune response.
Obviously, HSV-DNA reaches distant cutaneous locations from a site of active replication
via the bloodstream. Immune complexes may mediate hematogenous transport; 16 more
importantly, however, peripheral blood mononuclear cells, which contained HSV-DNA in
more than 60 percent of patients during acute erythema multiforme episodes, may be the
transport vehicle. 13 The use of this vehicle may explain viral DNA fragmentation:
monocytes are nonpermissive for HSV replication, which may lead to DNA damage. 14
Upregulation of adhesion molecules greatly increases binding of HSV-containing
mononuclear cells to endothelial cells; HLA-class I and adhesion molecule upregulation
in endothelial cells may account for the dermal inflammatory response. 23 The reason
that viral DNA fragments home to the specific predilection sites of erythema multiforme
or elicit the formation of lesions there (HSV-DNA has been recovered from apparently
healthy control skin 13 )is a matter of speculation. Precipitating factors such as
ultraviolet light may play a role: sun exposure is known to trigger erythema multiforme
(light-sensitive erythema multiforme), and the lesions are often confined to sunexposed areas. At times, erythema multiforme may even assume the clinical appearance
of polymorphous light eruption and juvenile spring eruption. 24
Cytotoxic effector cells (CD8+ T lymphocytes), which predominate in the rather scarce
inflammatory infiltrate within the epidermis, carry out the immunologic attack on HSVexpressing epidermal keratinocytes. These induce apoptosis of individual scattered
keratinocytes, leading to satellite-cell necrosis in early lesions or to more widespread
necrosis in older ones. 25 Exocytosis of these cells into the epidermis is facilitated by
strong expression of intercellular adhesion molecule 1 (ICAM 1) in the basal layer and
pockets of the spinous layer, which may be caused by interferon-? released from dermal
CD4+ T lymphocytes. 26 Neighboring epidermal cells are HLA-DR-positive. 4
In comparison to SJS-TEN, erythema multiforme displays much less striking epidermal
necrosis, but more intense dermal inflammation, which is mediated chiefly by CD4+ T
lymphocytes and monocytes, 27 and is responsible for the wheal-like clinical appearance
of the typical target lesions. Inflammation is associated with microvascular damage that is
likely to be caused by HSV-containing mononuclear cells binding to the endothelium. 23
The explanation for these obvious differences in pathology may lie in the observation that
skin lesions of HSV-related erythema multiforme express interferon-?, but not tumor

necrosis factora (TNF-a), whereas the reverse is true for drug-induced SJS-TEN. 28 The
two subsets thus appear to be mechanistically different, and erythema multiforme may
have a delayed type hypersensitivity component that is absent from SJS-TEN.
As recurrent HSV infection is common and erythema multiforme episodes are much less
so, it is obvious that HSV-related erythema multiforme must be linked to a specific
predisposition. Clinically, HSV lesions associated with erythema multiforme are not
different from those without such an association. 11 One report indicated not differences
between the specific immune responses of patients suffering from recurrent HSV
infection with associated erythema multiforme and those of patients without such an
association. 29 Increased frequencies of certain HLA antigens were described in
association with HSV-related erythema multiforme, particularly HLA-Bw62 (B15), -B35,
and -DR53. 30 Kmpgen and colleagues demonstrated significant associations with the
DR4 and DQw3 alleles; the relative risk for DQw3 was 44.2. 31 There was no such
correlation in patients with recurrent HSV infection without associated erythema
multiforme.

Clinical Manifestations
As noted in Table 58-1, prodromal symptoms are absent in most cases of erythema
multiforme. If present, they are mild, variable, and nonspecific, merely suggesting an
upper respiratory infection (e.g., cough, rhinitis, low-grade fever).
The skin rash arises abruptly. In most patients, all lesions appear within 3 days; but in
some, several crops follow each other during one episode of erythema multiforme. Up to
hundreds of lesions may form, most in a symmetric, acral distribution on the extensor
surfaces of the extremities (dorsa of hands and feet, elbows, and knees) and the face; they
appear less often on palms and soles, thighs, buttocks, and trunk ( Fig. 58-1). Lesions
often first appear acrally and then spread in a centripetal manner. Mechanical (Koebner
phenomenon) and actinic (predilection of sun-exposed sites) factors appear to influence
the distribution of lesions. Zosteriform hemicorporeal and linear distribution along
Blaschko's lines of erythema multiforme lesions may occur. 32 Although patients
occasionally report burning and itching, the eruption is usually symptomless.
As the name implies, the clinical picture of erythema multiforme is variable, but the rash
is usually monomorphous in a given patient. Variability results from the variable
morphologic expression along a scale of severity of a single prototypic lesion. Typically,
this lesion is a highly regular, circular, wheal-like erythematous papule or plaque that is
stable (i.e., of fixed position and persistent for 1 week or more); it measures from a few
millimeters to about 2 cm and may expand slightly over 24 to 48 hours. While the
periphery remains erythematous and edematous, the center becomes violaceous and dark;
inflammatory activity may regress or relapse in the center, thus giving rise to concentric
rings of color change ( Fig. 58-1A). Often, the center turns purpuric and/or necrotic or
transforms into a tense vesicle or bulla. The result is the classic target or iris lesion ( Fig.
58-1A).

According to the proposed classification, typical target lesions consist of at least three
concentric components: (1) a dusky central disk, or blister; (2) more peripherally, a ring
of pale edema; and (3) an erythematous halo. 5 , 7 Not all lesions of erythema multiforme
are typical: some display two rings only (raised atypical targets), but none are flat,
which is the typical lesion of SJS-TEN. In some patients with erythema multiforme, most
lesions are livid vesicles overlying a just slightly darker central portion, encircled by an
erythematous margin ( Fig. 58-1C). Larger lesions may have a central bulla and a
marginal ring of vesicles (herpes iris of Bateman; Fig. 58-1B).
Mucosal lesions are present in up to 70 percent of patients, 11 almost exclusively limited
to the oral cavity and most often only a few. Predilection sites are the lips, palate, and
gingivae. On the lips, identifiable target lesions may form; otherwise, vesicles, erosions,
and crusting are evident. Extensive, painful oral involvement with only few skin lesions
occasionally occurs. Cervical lymphadenopathy is usually present in these patients.
Ectodermosis pluriorificialis is a rare occurrence characterized by severe involvement
of two or three mucosal sites in the absence of skin lesions. Its place within the erythema
multiforme spectrum is not quite clear, but its often relapsing nature suggests that it is
HSV-related.
In most cases, erythema multiforme affects well under 10 percent of the BSA. Rare
instances of extensive skin lesions and prominent involvement of the oral mucosa (at
times called bullous erythema multiforme or erythema multiforme majorsee
Classification and History, above) may be difficult to distinguish from SJS-TEN.
Patients with erythema multiforme are afebrile, physical examination is normal, and there
are no reports of joint or muscular complaints or lymphadenopathy, except in patients
with mucosal erosions.

Course and Prognosis

Erythema multiforme runs a mild course in most cases, and each individual attack
subsides within 1 to 4 weeks. Recovery is complete, and there are no sequelae (not even
in mucosal lesions), except for transient hypo- or hyperpigmentation in some cases. The
condition does not progress to SJS-TEN. Recurrences are common and may occur in the
majority of cases. In one report on a large series of patients with recurrent erythema
multiforme, the mean number of attacks was 6 per year (range, 2 to 36), and the mean
total duration of disease was 9.5 years; in 33 percent, the condition persisted for more
than 10 years. 10 Up to 50 recurrences have been described in a single patient. 21 The
severity of episodes in patients with recurrent erythema multiforme is highly variable and
unpredictable. The frequency of episodes and cumulative duration of disease are not
correlated with the severity of attacks. The frequency and severity of recurrent erythema
multiforme tend to improve spontaneously over time (after 2 years or more), parallel to
the improvement of recurring HSV infection.
RELATIONSHIP TO RECURRENT HSV INFECTION In more than 70 percent of
patients with recurrent erythema multiforme, an episode of recurrent HSV infection
precedes the rash; the association with herpes labialis predominates (9:1) over that with

genital herpes, or that with herpes in other locations. The average interval is 8 days
(range, 2 to 17); 10 the duration of the lag period appears to be specific for individual
patients. In a small number of patients, HSV recrudescence and erythema multiforme
may occur simultaneously. Not all episodes of erythema multiforme are preceded by
clinically evident HSV infection, and not all HSV episodes are followed by erythema
multiforme. Episodes of recurrent HSV infection may precede the development of HSVrelated erythema multiforme for many years.
CONTINUOUS AND PERSISTENT ERYTHEMA MULTIFORME A small fraction of
patients experience a prolonged series of overlapping attacks of erythema multiforme;
this constellation has been labeled continuous erythema multiforme 21 and has been
connected to the systemic administration of glucocorticoids. Similar rare cases with
widespread and therapy-resistant lesions have been described as persistent erythema
multiforme. 33 It is doubtful that all these cases truly represent erythema multiforme.
Pathology
Early lesions of erythema multiforme exhibit lymphocyte accumulations at the dermalepidermal interface, with exocytosis into the epidermis, scattered keratinocyte necrosis
with lymphocytes attached to the necrotic keratinocyte (satellite-cell necrosis),
spongiosis, vacuolar degeneration of the basal layer, and focal junctional and
subepidermal cleft formation ( Fig. 58-2). Epidermal necrosis is more extensive in central
portions and older lesions. The papillary dermis is often highly edematous with a dense
mononuclear cell infiltrate, which is more abundant in older lesions than in fresh lesions.
The vessels are ectatic with swollen endothelial cells; there may be extravasated
erythrocytes and eosinophils. Advanced lesions show subepidermal blisters and frank
epidermal necrosis ( Fig. 58-3). Immunofluorescence findings are nonspecific. In late
lesions, melanophages may be prominent.
The histopathologic appearance of erythema multiforme is different from that of SJSTEN, in which dermal inflammation is moderate to absent and epidermal necrosis is
much more pronounced. 27 , 34 , 35 Formerly interpreted as expressions of the variability
of the erythema multiforme spectrum 36 or as consecutive stages in the dynamic
evolution of target lesions, 37 these differences have been demonstrated in very early
lesions of the respective entities and thus appear to be inherent features. Still, the
histopathologic appearances are somewhat overlapping and do not allow the distinction
of erythema multiforme from SJS-TEN in all instances. 38

Laboratory Investigations
Laboratory findings are usually normal in patients with erythema multiforme. In more
severe cases, an elevated erythrocyte sedimentation rate, moderate leukocytosis, acute
phase proteins, and mildly elevated liver transaminase levels may occur.

Differential Diagnosis
In typical cases, erythema multiforme is easily identifiable because of its characteristic
appearance, stability, and symmetric acral distribution of the target lesions. At times,
however, atypical cases display features of a variety of other skin conditions ( Table 58-

2). A disorder frequently mistaken for erythema multiforme is acute annular urticaria,
which is often drug-induced; this disorder is characterized by annular polycyclic,
concentric, or arcuate urticarial lesions that extend peripherally, leaving central
ecchymotic violaceous areas, thus mimicking target lesions. Histologically, only slight
edema and minimal perivascular infiltrates are found. Target purpura (acute
hemorrhagic edema of infancy) is a manifestation of leukocytoclastic vasculitis in infants
and small children; it is characterized by symmetric erythematous and purpuric
concentric rings in acral locations, similar to target lesions, that spontaneously resolve in
about a week. Erythema multiformelike lesions are sometimes found in lupus
erythematosus (Rowell's syndrome). 39 It may sometimes be difficult to distinguish
mucosal lesions from pemphigus vulgaris and herpetic gingivostomatitis. Disseminated
contact dermatitis may exhibit lesions that mimic target lesions closely, both clinically
and histologically. 22

Treatment
In most cases, erythema multiforme causes little discomfort and regresses spontaneously
within about 2 weeks. Symptomatic treatment with shake lotions, topical steroids,
analgesics, and antihistamines has little impact on the course, but may reduce subjective
symptoms. Liquid antacids, topical glucocorticoids, and local anesthetics relieve
symptoms of painful mouth erosions. The systemic administration of glucocorticoids is
unnecessary and may even have worsened some cases.
In recurrent erythema multiforme, early treatment of HSV infection with oral acyclovir
(200 mg five times per day for 5 days) or its derivatives with better bioavailability (e.g.,
valacyclovir or pencyclovir) may prevent erythema multiforme, but often it comes too
late. In such cases, continuous administration of low-dose acyclovir (400 to 800 mg per
day) or valacyclovir (500 mg per day) for a 6-month period is indicated. This regimen is
highly effective in preventing episodes of both HSV infection and erythema multiforme
as long as it is continued, even in patients in whom HSV is not the obvious precipitating
factor. 10 Attacks recur after withdrawal, but generally tend to be less frequent and less
severe. As alternatives for patients who do not respond to this treatment, dapsone,
antimalarials, and even azathioprine and thalidomide have been advocated.