
	Version PDF

Fixed drug eruption: a disease mediated by self-inflicted responses of intraepidermal T cells

European Journal of Dermatology. Volume 17, Number 3, 201-8, May-June 2007, Review article

DOI : 10.1684/ejd.2007.0149

Summary

Author(s) : Tetsuo Shiohara, Yoshiko Mizukawa , Kyorin University School of Medicine, 6-20-2, Shinkawa, Mitaka, Tokyo 181-8611, Japan.

Summary : It is difficult to dissect disease-causing T cells and anti-inflammatory T cells in a biopsy specimen obtained at a given time, which would represent a single time point in the development of the lesions. In fixed drug eruption (FDE), the resting lesions long after clinical resolution have many clues to identify the disease-causing T cells, because they contain a large homogeneous population of CD8 ⁺ T cells that are distributed along the epidermal basal layer and have the capacity to rapidly produce large amounts of IFN-γ. These intraepidermal CD8 ⁺ T cells are likely to be a major actor in epidermal injury observed in FDE lesions. In this review, we ask how they arise and how they cause epidermal injuries, which present with a wide spectrum of clinical manifestations and are often mistaken as signs of other skin disease.

Keywords : fixed drug eruption, recall phenomenon, intraepidermal CD8 ⁺ T cells, IFN-γ, viral infection, epidermal injury

Pictures

ARTICLE

Auteur(s) : Tetsuo Shiohara, Yoshiko Mizukawa

Kyorin University School of Medicine, 6-20-2, Shinkawa, Mitaka, Tokyo 181-8611, Japan

accepté le 16 Janvier 2007

Cutaneous reactions often develop at previously inflamed or traumatized skin sites, a phenomenon referred to as “the recall response” or “the isotopic response” [1, 2]. The term “isotopic response” describes the occurrence of a new, unrelated disease that appears at precisely the same location as a previously healed disease. The interval between the initial event and the second disease arising at the sites is highly variable, ranging from days to years. Although a heterogeneous group of cutaneous diseases ranging from lichen planus to granuloma annulare shows this phenomenon, the mechanisms causing it are at present speculative. The most obvious candidate is T cells found in the lesions.A recent study in mice shows that specialized T cells found in skin, intraepidermal T cells, are induced by wounding to secrete growth factors that accelerate the healing of experimentally induced skin wounds [3]. In view of the fact that the epithelium serves as the interface between the organism and the external environment, it makes good biological sense that epidermal injury activates such intraepidermal T cells originally residing in the epidermis, which in turn promote epidermal repair. Indeed, our previous study in mice demonstrated that intraepidermal T cells expressing the second class of T-cell receptor (TCR), the γδ TCR, are essential for resistance to the induction of cutaneous graft-versus-host disease (GVHD) under conditions in which the epidermis is severely damaged in a site-restricted manner [4]. Unfortunately, however, the precise equivalent of murine intraepidermal T cells distributing densely and evenly within the epidermis has not been found in humans despite many attempts to find human equivalents. Nevertheless, the absence in normal epidermis of a precise morphologic and phenotypic equivalent does not necessarily exclude the presence of human homologues that can substitute for their function. In this regard, we found that intraepidermal T cells expressing the αβ TCR are abundantly detected between basal keratinocytes in the lesions of fixed drug eruption (FDE) over prolonged periods of time after clinical resolution [5-7]. Because a recall phenomenon can be typically seen in FDE lesions [1], this disease may serve as an important clue in understanding the clinical feature and the pathogenesis of a recall phenomenon. In this article, we will provide current insight regarding clinical and pathogenic heterogeneity of this disease and the role of intraepidermal T cells in the epidermal injury.

Clinical features

FDE usually appears as a solitary or a small number of pruritic, well circumscribed, erythematous macules that evolve into edematous plaques; these lesions typically resolve after discontinuation of the offending drug, leaving hyperpigmentation at the site of lesions (figure 1A). They recur in exactly the same sites when rechallenged with each administration of the offending drug (figure 1B). The appearance of these lesions is often preceded and accompanied by sensations of burning [8]. Although FDE may occur anywhere on the skin or mucous membrane, the most common locations are the lip, palms, soles, glans penis, and groin areas. Discrete lesions often appear in the same bilaterally symmetrical regions of the skin, particularly in the abdominal and the inner aspect of the arms and legs. Most FDE lesions occur with orally administered rather than injected drugs. In some patients, however, FDE lesions can be caused after sexual intercourse with their partners taking the offending drug [9]. Although drugs causing FDE differ in countries depending on the availability of the various drugs, the most frequent drugs associated with FDE are sulphonamides, tetracyclines, mefenamic acid, and terbinafine. Although the incidence of FDE is generally thought to be decreasing, particularly in the western world with the much lower usage of medications frequently associated with the classic form, the unusual forms described below appear to be increasing — rather than decreasing in frequency, probably due to unawareness of the unusual presentations.

The diagnosis of FDE is not difficult for dermatologists even after clinical resolution, when there are single or several round or oval, demarcated hyperpigmentations (figure 1A). The number of involved sites and the size of the lesions often increase with repeated exposures. Systemic provocation is safe and is still the most reliable method for establishing the etiologic agent in FDE [10]; however, convincing results can be also achieved with topical provocation of certain drugs when applied to sites of previous FDE lesions. Positive reactions are usually obtained at the sites of previous FDE but no reaction occurs on sites of previously unaffected skin. However, false negative results have been reported, and have been attributed to either inadequate transepidermal penetration of the drug or the requirement of metabolites of the drug not present in the test material: vehicles and drugs that can be used for topical provocation are still quite limited. Thus, a negative skin reaction gives no reliable information.

Although a single drug is usually responsible for FDE, in some patients FDE lesions develop following the ingestion of multiple drugs. This usually occurs when the offending drugs have common chemical structures; however, in some cases, drugs or other agents with totally different chemical structures can cause exacerbations precisely similar to those caused by the inducing drug, a phenomenon known as polysensitivity [11]. This phenomenon is much more frequent than is actually reported and may have been overlooked in the past, because the lesions flared by these agents could not generally be recognized as FDE and as a consequence, these cases would be underdiagnosed. Indeed, evidence is accumulating to indicate that FDE lesions can be flared not only by the original offending drug but also by nonspecific stimuli, including a particular food [12] and stroking with a pencil [13]. Following each exacerbation, some patients with FDE demonstrate a refractory period, during which the offending drug fails to activate the lesions [8]. The duration of this period is variable, lasting from a few weeks to several months. The eruption seems to wonder when sites involved during one flare are not involved during a subsequent flare because of a prolonged refractory period. Interestingly, despite the continued administration of the offending drug some lesions may show a gradual decrease in the intensity of flare and may even disappear. A similar desensitization procedure may be successfully applied in the case of patients who cannot avoid the offending drug for treatment [14]. Thus, the diagnosis of FDE is made more complex by the recognition that exacerbations precisely similar to those caused by the inducing drug can occur at the same sites not only by drugs or agents of totally different chemical structures but also by other nonspecific stimuli such as a combination of cytokines.

In some patients, the eruption can only be reproduced when multiple drugs are administered in combination but not separately [15]. This indicates that false negative results may be seen when systemic provocation tests are performed separately with each constituent to identify the chemical structure responsible for FDE, instead of the combined preparation, and suggests that combinatorial interactions of each constituent with unique immunomodulatory properties might be needed for full expression of the disease.

Several less common clinical types of FDE have been reported and will be described in the following sections.

Nonpigmenting FDE

Most patients with typical pigmenting FDE lesions usually have no systemic symptoms, apart from local irritation in the skin lesions. However, multiple lesions are often associated with systemic manifestations, malaise, high fever, nausea, and arthralgia. Moreover, blister formation often occurs at an advanced stage in these patients, which resembles that of Stevens-Johnson syndrome (SJS) or toxic epidermal necrolysis (TEN). It is, therefore, difficult to distinguish between SJS/TEN and such a multiple variant of FDE, particularly when the bullous lesions become more widespread with systemic manifestations and do not leave typical hyperpigmentation. In this regard, Shelley and Shelley described this variant as nonpigmenting FDE (NPFDE) [16]. This multiple and nonpigmenting variant may be frequently misdiagnosed as SJS or TEN, unless dermatologists take special care to recognize the presence of this variant. NPFDE is characterized by multiple, large, symmetrical, well-circumscribed tender erythematous plaques that are often associated with systemic symptoms [16]. Because NPFDE lesions fade without pigmentation or any other trace, identification of the lesions as being fixed is difficult to document. Particularly in lighter-hued skin, the color of FDE lesions remains purplish grey even after repeated episodes and it would eventually fade away for long periods, thereby preventing physicians from recognizing similar recurrent eruptions occurring months to years apart as appearing at identical sites. Although NPFDE was originally defined as showing a symmetrical distribution, it has become clear that this form can also appear as solitary lesions without any systemic symptoms [17]. Because the lesions are not recognized as appearing at identical sites and because of the relatively low penetration of this entity, this form may be frequently misdiagnosed as idiopathic skin diseases, such as erythema multiforme (EM). In particular, some cases reported as EM showing isomorphic phenomenon may actually be NPFDE with skin lesions confined to previously traumatized or sunburned skin [18]. In support of this possibility, we previously demonstrated that some multiple pigmenting (PFDE) lesions initially appeared at previously traumatized skin sites, such as burn scars and insect bites [19]; and EM-like FDE lesions were reported to develop in a known user of various illicit drugs [20]. Because it may be quite difficult to recognize even the PFDE lesions after a few episodes in patients with lighter-hued skin, NPFDE may be more common than the classic pigmented form (PFDE) in Caucasians. Minor lesions could easily remain unnoticed, particularly in older patients with some cognitive disturbances, thus identification of such lesions as “fixed”, occurring on the same site, is more difficult to document. According to the original description, epidermal change and dermal macrophages responsible for the hyperpigmentation clinically seen in PFDE are absent in NPFDE, thus proposing the concept that the drug-induced immune responses in NPFDE are primarily directed against the dermis and spare the epidermis [16]. However, because these findings have not been confirmed in other reports, including our own [21], this concept is no doubt oversimplified. This point will be discussed below.

NPFDE could be also misdiagnosed as bullous pemphigoid, when large bullae appear on the involved areas. Some cases reported as bullous pemphigoid appearing shortly after influenza vaccination might indeed be bullous NPFDE triggered by influenza vaccination [22, 23]. Consistent with this possibility, Garcia-Doval et al. reported generalized bullous FDE appearing 12h after influenza vaccination, which simulated bullous pemphigoid [24]. Although the authors did not describe whether these lesions were elicited with influenza vaccination at the pigmented sites, clinical manifestations of this patient were more likely to represent NPFDE than the classic PFDE. Although this diagnosis was primarily based on negative fluorescence and the spontaneous resolution in 2 weeks without recurrence, deposition of C3 and IgM along the basement membrane zone was also reported to occur during the acute phase of FDE [25], thus making the distinction between the two diseases more difficult. In addition, Hern et al. also reported generalized bullous FDE due to paracetamol where direct and indirect immunofluorescence (IF) findings were more characteristic of bullous pemphigoid: direct IF showed liner staining with IgG and C3; indirect IF demonstrated circulating antibodies reacted to a 230 kDa BP antigen, but neither to the 180 kDa antigen (BPAG2) nor to NC 16a domain fusion protein [26]. These cases allow the concept of NPFDE to be extended to include FDE showing IF findings typical of bullous pemphigoid. These cases indicate the importance of recognizing the presence of generalized bullous NPFDE often mimicking SJS/TEN or bullous pemphigoid.

Other types of FDE

FDE often presents with a wide spectrum of clinical manifestations indistinguishable from those of other skin diseases and these manifestations should not be mistaken as signs of these skin diseases.

FDE appearing in a striking linear pattern on the lower limbs has been reported [27], although it was unclear whether the patient had preceding herpes zoster before onset of the FDE. Recently, the occurrence of drug eruptions consisting of confluent linear erythema along the dermatomes previously involved by a herpes zoster episode has been described on the chest and on abdominal skin 7 days after starting treatment with aciclovir [28]. Because in this case clinical challenge with aciclovir was not performed after clinical resolution, it remains unknown whether the linear erythema was reproduced by administration of the drug at the same site, thereby making the diagnosis of FDE impossible. Nevertheless, the authors interpreted these findings as indicating that aciclovir-induced recall reactions occurred at the site of previous involvement by herpes zoster.

FDE may appear as an acute paronychia, when drug reactions occur in the distal phalanx region [29]. This case is reminiscent of our case, reported as “trauma-localized FDE” [19]. An unusual cellulitis-like FDE due to topotecan has also been reported: an erythematous and edematous plaque with undetermined borders mimicking cellulitis was elicited by the subsequent administration of the offending drug at the same site, confirming the diagnosis of FDE [29]. Interestingly, the intensity of the flare gradually decreased upon repeated administration of the offending drug. These observations suggest that FDE must be included even in the differential diagnosis of paronychia and cellulitis.

Because FDE lesions recur shortly after re-challenging with the offending drug in exactly the same site, FDE has been generally regarded as an acutely inflamed skin disease. Indeed, FDE with chronically inflamed lesions is rarely seen. However, there are reports of chronic FDE cases with unusual clinical findings: Guin and Baker reported a case of chronic FDE resembling parapsoriasis en plaques which presented persistent, stable lesions that were present for several months [30]. In general, most FDE lesions repeatedly recur at the same site when the offending drug is administered at certain intervals, such as weeks to months apart, while some previously involved sites are not affected because of an apparent refractory state in these locations when the offending drug is continuously administered. Therefore, it is theoretically unlikely that multiple, chronic, superficial lesions persist unchanged for several months despite continued use of the offending drug. However, this was the case. Guin and Baker reported that parapsoriasis-like FDE lesions had persisted unchanged for seven months until the offending drug was discontinued and that the lesions flared promptly upon administration of the drug. Thus, in spite of the atypical appearance of the lesions, the reproducibility of eruption by administration of the drug and prompt resolution upon withdrawal were all indicative of the diagnosis of FDE. These lesions may represent the chronic type of FDE and this type should be included in the spectrum of FDE. Unfortunately, however, this observation appeared to fail to achieve any significant degree of recognition.

Histology

Because most biopsy specimens are obtained from the lesions on day 1 to 2 after the onset, they show in common an interface dermatitis manifested by hydropic degeneration of basal keratinocytes associated with lymphocytic invasion of the epidermis involving mainly the interfollicular epidermis; but in some cases, the hair follicles and acrosyringia are also affected. The histologic picture of FDE is typical of a lichenoid tissue reaction with melanin incontinence. Occasionally exocytosis of lymphocytes can be marked, mimicking Pautrier’s microabscesses of mycosis fungoides. The upper dermis is edematous and contains a mixed inflammatory cellular infiltrates composed predominantly of lymphocytes, neutrophils, eosinophils, and nuclear dust. However, it should be kept in mind that biopsy material obtained for diagnostic purposes represents a single time point in the development of the lesion and the dynamics of FDE lesions have rarely been adequately described in the literature. In this regard, we sequentially obtained biopsy specimens from virtually the same site after re-challenging with the offending drug, to follow the evolution of individual lesions [21]. Marked differences depending on the time after re-challenge can be detected although they arose from virtually the same sites.

Before re-challenge, a small number of lymphocytes tagging to the epidermis, minimal perivascular lymphocytic infiltrates, and a normal-appearing epidermis are seen in the lesional skin; but the basal later is found to contain increased numbers of lymphoid cells, when compared with that in the adjacent uninvolved perilesional skin (figures 2A and B). There is considerable variability among patients in the extent of the inflammatory responses after clinical challenge. Second biopsy specimens obtained upon the appearance of irritation or slight erythema in the previously involved sites (usually 2-3 h after challenge) show lymphoid cells reaching the lower half of the epidermis while relatively preserving the epidermal architectures. No significant changes in the numbers and distribution of dermal infiltrates can be seen: the upper dermis shows lymphoid cells in a perivascular distribution. Hydropic degeneration of the basal layer and the extensive exocytosis of dermal lymphoid cells into the epidermis referred to as “epidermotropism”, are typically seen at 24 h after challenge. These findings correspond well to those known as the histology of established FDE lesions. Very few polymorphonuclear cells and only a few eosinophils can be found in the vast majority of FDE lesions. In certain cases, however, some FDE lesions show a marked perivascular and interstitial mixed inflammatory infiltrate with a predominance of neutrophils, similar to another previously reported case [31]. Because no interface changes were seen despite the presence of a bulla in the case, Agnew and Oliver proposed a new entity “neutrophilic FDE” which differs from a classic-type FDE in that the basal epidermis is intact without epidermotropic infiltration of lymphocytes and the histologic features are consistent with those of a neutrophilic dermatosis [31]. However, our FDE case series show that mixed inflammatory infiltrates with a predominance of neutrophils, said to be specifically present in neutrophilic FDE but not in a classic-type of FDE, are largely a function of both the age and site of lesions biopsied: inflammatory infiltrates predominantly composed of neutrophils would be expected in either the earliest phase of the immune reactions in the lesions or in FDE of the flexural areas. Thus, it remains unknown whether neutrophilic FDE represent a real entity or a stage in the histopathological evolution of FDE.

When biopsy specimens are taken from the same lesions on day 3 or 4 after challenge or the onset, histopathology often reveals a psoriasiform lichenoid pattern with slight focal spongiosis: the epidermis is irregularly thickened and individual necrotic keratinocytes are occasionally seen. This represents a healing process from the epidermal damage caused by epidermotropic infiltration of lymphocytes. Thus, there is need for sequential biopsies from the same lesions to elucidate the dynamic histopathologic changes that occur in the evolving FDE lesions before proposing a new entity.

Immunohistochemical findings

These considerations led us to thoroughly examine the resting FD lesions before challenge. Resting FDE lesions were characterized by the presence of large numbers of CD3⁺CD8⁺ T cells aligned along the epidermal side of the dermoepidermal junction (figure 3) with occasional CD68⁺ macrophages distributing the upper dermis. In the epidermis, CD8⁺ T cells overwhelmingly predominated over CD4⁺ T cells. In the perivascular and interstitial dermis, CD4⁺ T cells generally outnumbered CD8⁺ T cells by approximately 2: 1. Our recent studies have shown that intraepidermal T cells, abundantly identified in the resting FDE lesions, are composed of a phenotypically homogeneous population that expresses αβ TCR, CD45RA, and CD11b, but not CD27 and CD56 [21]. This phenotype of T cells most closely resembles that of effector memory T cells [32, 33]. Recently, memory CD4⁺ and CD8⁺ T cells have been shown to consist of at least two functionally distinct subsets: non-polarized ‘central-memory’ T cells that preferentially home to secondary lymphoid organs, and polarized ‘effector memory’ T cells that have the capacity to migrate to non-lymphoid tissues such as the skin [32-35]. However, intraepidermal T cells are not necessarily identical to effector memory T cells in that the overwhelming majority of these intraepidermal T cells constitutively express both CLA and α_Eβ₇[21], whereas effector memory T cells are supposed to be negative for CLA. Because TGF-β is the only cytokine that can upregulate the cell surface expression of both CLA and α_Eβ₇ on T cells, TGF-β produced by epidermal keratinocytes may serve to retain these T cells in the epidermal compartment. With regard to the local retention of antigen-specific T cells at sites previously elicited by allergic contact dermatitis (ACD), Moed et al. demonstrated the role of CCR10- CCL27 interactions [36]: expression of the CCR10 targeting ligand CCL27 remained increased after clinical resolution of ACD and could play a role in retaining CCR10⁺CD4⁺ T cells within a previously affected area of the skin. Nevertheless, they also showed that CCR10⁺CD8⁺ T cells, unlike CCR10⁺CD4⁺ T cells, were not detected in the previously inflamed skin, indicating that mechanisms responsible for the local retention of CD8⁺ T cells would be different from those responsible for CD4⁺ T cell retention.

Although the numbers of intraepidermal and dermal T cells were not significantly increased at 2-3h after challenge, intraepidermal T cells initially aligned along the dermoepidermal junction showed a change in distribution: they appeared to reach the lower half of the epidermis. In some areas, they formed small clusters within the epidermis associated with focal spongiosis. Because at this time dermal T cells largely remained localized to perivascular adventitial zones, the change in the distribution of intraepidermal T cells within the epidermis would precede the influx of dermal T cells into the epidermis. Biopsy specimens taken from the 1-day-old lesions showed a dense diffuse infiltrate of CD4⁺ T cells obscuring the dermoepidermal junction. Epidermal changes ranging from extensive epidermal damage to individual or grouped apoptotic keratinocytes represented as colloid bodies were observed in association with the intraepidermal CD8⁺ T cells and infiltrating CD4⁺ T cells. The ratio of CD4⁺ T cells to CD8⁺ T cells within the epidermis was increased to 2: 1 at this time, although the number of intraepidermal CD8⁺ T cells was not significantly decreased as compared with these in the resting lesions. At this time, the number of CD4⁺ dermal T cells also increased significantly and they showed significant extension into the interstitial dermis. There was a concomitant increase in the number of evenly distributed dermal CD8⁺ T cells. The lowest increase in the frequency of intraepidermal CD4⁺ T cells was noted in the lesions which showed the most severe epidermal damage. These results suggest that the epidermotropic migration of CD4⁺ T cells into the epidermal compartment may serve to ameliorate the tissue damaging effect of intraepidermal CD8⁺ T cells originally residing in the resting lesions although the possibility cannot be excluded that other populations of CD4⁺ T cells may preserve their helper function for the CD8⁺ T cells. Because the most predominant influx of CD4⁺CD25⁺FoxP3⁺ T cells into the epidermis was found in the 1-day-old lesion of a case in which epidermal damage was mild; in contrast, this subset was rarely detected in the corresponding lesion of another case which showed the most severe damage. These results suggest that this subset would represent regulatory T cells. More work will be necessary to determine whether this subset could have the capacity to inhibit the function of intraepidermal CD8⁺ T cells.

Interestingly, in the resting lesions, these intraepidermal CD8⁺ T cells constitutively expressed CD69, an early activation marker, but not CD25, another activation marker. The constitutive expression of CD69 on intraepidermal T cells in resting FDE lesions is reminiscent of unstimulated NK cells that have been shown to constitutively express low levels of CD69 [37], suggesting that intraepidermal T cells share several features with NK cells. In contrast, CD69 was rarely detected in the dermal T cells before challenge. Upon challenge, a moderate increase in CD69 and CD25 expression was detected on both CD4⁺ and CD8⁺ T cells in the dermis at 3h and 24h.

Because long-term memory is maintained in the lesional skin of FDE, one may reasonably ask how long these intraepidermal CD8⁺ T cells would persist and maintain their effector memory phenotype in the absence of overt antigenic stimulation. To this end, we examined the lesional skin of a patient with FDE, in which the causative drug had not been given for at least 4 years. Surprisingly, CD8⁺ T cells were abundantly detected along the dermoepidermal junction, although the number was significantly decreased as compared with 4 years before. Virtually all of the intraepidermal CD8⁺ T cells detected in the basal layer expressed CD45RA, CLA, α_Eβ₇ and CD69, just as those 4 years before had done. In contrast, CD4⁺ T cells, abundantly detected in the 1-day-old lesions, could hardly be found in any part of the epidermis in the lesions 4 years after the final flare. These results indicate that intraepidermal CD8⁺ T cells persist in the lesional epidermis for long periods (at least 4 years) as a stable population of effector memory T cells, even in the absence of an encounter with cognate antigen and that CD4⁺ T cells which migrated from the circulation later on in the evolving FDE lesions have no propensity to lodge in the basal layer and may migrate out of the epidermis or undergo programmed cell death upon clinical resolution.

Antigen specificity of intraeidermal T cells involved in the pathogenesis of FDE

Intraepidermal CD8⁺ T cells are likely to be involved in the pathogenesis of FDE; because they are abundant in the resting and acute lesions of FDE. It is reasonable to hypothesize that these T cells could contribute to epidermal damage upon recognition of a relevant antigen. However, we were unable to detect a significant proliferation of these T cells isolated from the resting FDE lesions in response to the causative drug even in the presence of autologous antigen presenting cells. Instead, some intraepidermal T cells have been shown to recognize self-proteins [38]. Thus, the ligands for intraepidermal T cells are likely not simply drug antigens presented by self MHC molecules but rather secondary antigens induced during the inflammatory process, such as stress proteins, as suggested with murine γδ TCR⁺ intraepidermal T cells [39]. As an alternative approach, therefore, TCR α-chain and β-chain V gene usage by intraepidermal T cells was investigated by quantitative PCR. The results of PCR analyses showed that intraepidermal T cells isolated from the resting FDE lesions utilize a very limited range of Vα and Vβ gene families as compared with peripheral blood lymphocytes obtained from the same patients [38]. These results indicate that a striking preferential usage of Vα and Vβ gene families by intraepidermal T cells may represent the preferential migration or the local expansion of T cells capable of recognizing a limited set of antigens. Clinical findings that FDE lesions can often be flared by the intake of other drug with totally different chemical structures or non-specific stimuli may give an important clue as to the nature of the ligand/antigen recognized by intraepidermal T cells. In view of the similarity between intraepidermal T cells and NK cells, these findings might be explained by assuming that a down-regulation of MHC class I associated with increased expression of stress proteins on surrounding keratinocytes, which occur upon nonspecific stimuli, may be sufficient for intraepidermal T cells to set in motion a program of activation. An alternative explanation is that intraepidermal T cells could be broadly cross-reactive in nature with a variety of exogenous antigens while preserving the fine specificity for a self-MHC-bound peptide modified by the original causative drug: however, this specificity may not be reproduced by stimulating with autologous antigen presenting cells modified with the causative drug in vitro.

Considering the clinical observation that some FDE lesions initially appeared at sites of recent herpes simplex lesions and herpes zoster lesions, we could assume that virus-specific memory intraepidermal T cells that cross-react with drug antigens could play a role in FDE. In view of our findings that intraepidermal CD8⁺ T cells persist in the lesional epidermis for long periods as a stable population of effector memory T cells without stimulation with relevant drug antigens, these T cells would be retained as long-term drug-specific memory T cell population by stimulation with viruses in patients who have no longer been exposed to the relevant drug antigens. We hypothesize that effector T cells primed during viral infections may evolve into long-lived memory cells and, once cross-reacted with relevant drugs, would become effectors of epidermal damage typically seen in FDE lesions (figure 4). Alternatively, viral infections as well as drug antigens could stimulate innate immune responses, thereby up-regulating co-stimulatory molecules and self-antigen expression on antigen-presenting cells and releasing cytokines. This sequence of event may result in activation of otherwise dormant intraepidermal CD8⁺ T cells with autoreactive potential.

Epidermal injury mediated by intraepidermal T cells

Taken together, there is emerging evidence that intraepidermal CD8⁺ T cells residing in FDE lesions have the key role in mediating the localized epidermal injury. Indeed, using intraepidermal CD8⁺ T cells isolated form resting FDE lesions and subsequently expanded in vitro, we previously demonstrated that these intraepidermal CD8⁺ T cells, upon stimulation in an Ag-nonspecific fashion via the CD3/TCR complex, displayed cytolytic activity against NK-sensitive or NK-resistant tumor cells and cultured keratinocytes [38], while they were not constitutively cytolytic, not consistent with murine γδ counterparts [39]. In addition, our intracellular cytokine assay showed that the vast majority of intraepidermal CD8⁺ T cells freshly isolated from resting FDE lesions produced IFN-γ upon stimulation with PMA/ionomycin; in contrast, the proportion of these T cells producing IL-4 was less than 1% [21]. When compared with the frequencies of peripheral blood counterparts from the same patients, those of these intraepidermal CD8⁺ T cells capable of producing IFN-γ were much higher [21]. In support of the pathogenic role of intraepidermal T cells, our in situ RT-PCR studies demonstrated that a strong and exclusive induction of IFN-γ mRNA and protein was seen in intraepidermal T cells with much faster kinetics than their dermal and peripheral counterparts [21]. Because rapid production of large amounts of IFN-γ at an early time point (2 to 3h after challenge) by these T cells preceded the epidermal invasion of other T cells and was rapidly followed by localized epidermal injury, these intraepidermal CD8⁺ T cells are likely a major actor in epidermal injury observed in the fully evolved lesions of FDE. Apart from producing IFN-γ, intraepidermal CD8⁺ T cells might also contribute to the development of FDE lesions by directly interacting with other inflammatory cells [40]. Thus, severe epidermal damage observed in the late stage of FDE lesion would be caused not only by production of early burst of IFN-γ by intraepidermal T cells but also by perforin and granzyme B and other cytokines produced secondarily by rapid release of IFN-γ. Indeed, granzyme B-positive and perforin-positive cells have been identified in the same distribution pattern within the epidermis as intraepidermal CD8⁺ T cells in the evolving FDE lesions (Mizukawa Y et al. manuscript submitted). At least in FDE it appears that activation of intraepidermal T cells is necessary for initiating the cascade of events resulting in epidermal injury but that full blown epidermal injury requires additional mechanisms acting independently or together. In this regard, a recent study, Fas-Fas L interactions has been shown to be involved in the development of FDE lesions [41]. Collectively, there is communication between intraepidermal CD8⁺ T cells and other types of inflammatory cells (infiltrating CD4⁺ T cells, macrophages and keratinocytes) and these interactions might contribute to the development of FDE lesions. The nature of these interactions can also determine the final outcome of disease.

Recent studies have shown that CD8⁺ T cells that share some of phenotypic features with these intraepidermal T cells residing in resting FDE lesions are abundantly identified in the blister fluid from patients with TEN [42]. At present, factors that determine whether patients have self-limited FDE or subsequently progress to TEN are unknown. In this regard, it should be kept in mind that the presence of intraepidermal CD8⁺ T cells was diluted by the prominent influx of CD4⁺ T cells (FoxP3⁺ and FoxP3^-) into the epidermis in 1 day-old-FDE lesions whereas only low numbers of corresponding CD4⁺ T cells were identified in the established lesions of TEN [42]: such overwhelming predominance of CD4⁺ T cells over CD8⁺ T cells seems to be a prominent feature in the evolving FDE lesions but not in TEN.

Conclusion

Although there is much evidence that intraepidermal CD8⁺ T cells have a major role in initiating epidermal injury observed in FDE and possibly in TEN, many issues remain unresolved. The persistence of these intraepidermal CD8⁺ T cells in FDE lesions may depend on factors that not only prevent apoptosis of these cells but also enable their subsequent reactivation upon stimulation with relevant drug antigens; however, investigation of the factors is still in its infancy. We do not know how intraepidermal CD8⁺ T cells could initially migrate from circulation to the epidermis, how the lesional epidermis could selectively retain these intraepidermal T cells for long periods of time without stimulation and how these intraepidermal T cells could interact with other types of inflammatory cells to cause extensive epidermal injury. Future studies will address these issues and characterization of mechanisms involved in activation, retention and long-term survival of intraepidermal T cells within the epidermal compartment might present new perspectives on the regulation of drug- or virus-induced inflammation and disease progression to TEN.

Acknowledgements

This work was supported in part by grants from the Ministry of Education, Sports, Science and Culture of Japan (T.S. and Y.M.), and from the Ministry of Health, Labor and Welfare of Japan (T.S.).

References

1 Shiohara T, Mizukawa Y. Recall phenomenon: Some skin-resident cells remember previous insults. Dermatol 2003; 207: 127-9.

2 Wolf R, Brenner S, Ruocco V, Grimaldi Filioli F. Isotopic response. Int J Dermatol 1995; 34: 341-8.

3 Jameson J, Ugarte K, Chen N, Yachi P, Fuchs E, Boismenu R, Havran WL. A role for skin γδ T cells in wound repair. Science 2002; 296: 747-9.

4 Shiohara T, Moriya N, Hayakawa J, Itohara S, Ishikawa H. Resistance to cutaneous graft-vs.-host disease is not induced in T cell receptor γ gene-mutant mice. J Exp Med 1996; 183: 1483-9.

5 Teraki Y, Moriya N, Shiohara T. Drug-induced expression of intracellular adhesion molecule-1 on lesional keratinocytes in fixed drug eruption. Am J Pathol 1994; 145: 550-60.

6 Shiohara T. What is new in fixed drug eruption? Dermatology 1995; 191: 185-7.

7 Shiohara T, Moriya N. Epidermal T cells: their functional role and disease relevance for dermatologists. J Invest Dermatol 1997; 109: 271-5.

8 Korkij W, Soltani K. Fixed drug eruption. Arch Dermatol 1984; 120: 520-4.

9 Gruber F, Stasic A, Lenkovic M, Brajac I. Postcoital fixed drug eruption in a man sensitive to trimethoprim-sulphamethoxazole. Clin Exp Dermatol 1997; 22: 144-5.

10 Breathnach SM. Management of drug eruptions, part II: diagnosis and treatment. Australas J Dermatol 1995; 36: 187-91.

11 Shiohara T, Kokaji T. Polysensitivity in fixed drug eruption. J Am Acad Dermatol 1997; 37: 1017-8.

12 Kelso JM. Fixed food eruption. J Am Acad Dermatol 1996; 35: 638-9.

13 Pellicano R, Camassa F, Lomuto M. Localized dermographism at the site of a fixed drug eruption. Br J Dermatol 1995; 132: 156-8.

14 Kelso JM, Keating RM. Successful desensitization for treatment of a fixed drug eruption to allopurinol. J Allergy Clin Immunol 1996; 97: 1171-2.

15 Verbov J. Fixed drug eruption due to a drug combination but not to its constituents. Dermatologica 1985; 171: 60-1.

16 Shelley WB, Shelley ED. Nonpigmenting fixed drug eruption as a distinctive reaction pattern: Examples caused by sensitivity to pseudoephedrine hydrochloride and tetrahydrozoline. J Am Acad Dermatol 1987; 17: 403-7.

17 Hindioglu U, Sahin S. Nonpigmenting solitary fixed drug eruption caused by pseudoephedrine hydrochloride. J Am Acad Dermatol 1998; 38: 499-500.

18 Huff JC, Weston WL. Isomorphic phenomenon in erythema multiforme. Clin Exp Dermatol 1983; 8: 409-13.

19 Mizukawa Y, Shiohara T. Trauma-localized fixed drug eruption: Involvement of burn scars, insect bites and venipuncture sites. Dermatology 2002; 205: 159-61.

20 Inserra DW, Camisa C. Erythema multiforme-like fixed drug eruption. Cutis 1989; 44: 223-5.

21 Mizukawa Y, Yamazaki Y, Teraki Y, Hayakawa J, Hayakawa K, Nuriya H, Kohara M, Shiohara T. Direct evidence for interferon-γ production by effector- memory-type intraepidermal T cells residing at an effector site of immunopathology in fixed drug eruption. Am J Pathol 2002; 161: 1337-47.

22 Lear JT, Tan BB, English JS. Bullous pemphigoid following influenza vaccination. Clin Exp Dermatol 1996; 21: 392.

23 Downs AM, Lear JT, Bower CP, Kennedy CT. Does influenza vaccination induce bullous pemphigoid? A report of four cases. Br J Dermatol 1998; 138: 363.

24 Garcia-Doval I, Roson E, Feal C, Dela Torre C, Rodriguej T, Cruces MJ. Generalized bullous fixed drug eruption after influenza vaccination, simulating bullous pemphigoid. Acta Derm Venereol 2001; 81: 450-1.

25 Hindsen M, Christensen OB, Gruic V, Lofberg H. Fixed drug eruption: an immunohistochemical investigation of the acute and healing phase. Br J Dermatol 1987; 116: 351-60.

26 Hern S, Harman K, Clement M, Black MM. Bullous fixed drug eruption due to paracetamol with an unusual immunofluorescence pattern. Br J Dermatol 1998; 139: 1129-31.

27 Sigal-Nahum M, Kongui A, Gaulier A, Sigal S. Linear fixed drug eruption. Br J Dermatol 1988; 118: 849-58.

28 Carrasco L, Pastor MA, Izquierdo MJ, Ferina MC, Fortes MJ, Requena L. Drug eruption secondary to aciclovir with recall phenomenon in a dermatome previously affected by herpes zoster. Clin Exp Dermatol 2002; 27: 132-4.

29 Senturk N, Yanik F, Yildiz L, Aydin F, Canturk T, Turanli AY. Topotecan-induced cellulites-like fixed drug eruption. J Eur Acad Dermatol Venereol 2002; 16: 414-6.

30 Guin JD, Bader GF. Chronic fixed drug eruption caused by acetaminophen. Cutis 1988; 41: 106-8.

31 Agnew KL, Liver GF. Neutrophilic fixed drug eruption. Australas J Dermatol 2001; 42: 200-2.

32 Mackay CR, von Adrian UH. Memory T cells – local heroes in the struggle for immunity. Science 2001; 291: 2323-4.

33 Masopust D, Vezys V, Marzo AL, Lefrançois L. Preferential localization of effector memory cells in nonlymphid tissue. Science 2001; 291: 2413-7.

34 Sallusto F, Lenig D, Förster R, Lipp M, Lanzavecchia A. Two subsets of memory T lymphocytes with distinct homing potentials and effector functions. Nature 1999; 401: 708-12.

35 Reinhardt RL, Khorufs A, Merica R, et al. Visualizing the generation of memory CD4⁺ T cells in the whole body. Nature 2001; 401: 101-5.

36 Moed H, Boorsma DM, Tensen CP, Flier J, Jonker MJ, Stoof TJ, von Blomberg BME, Bruynzeel DP, Scheper RJ, Rustemeyer T, Gibbs S. Increased CCL27-CCR10 expression in allergic contact dermatitis: implications for local skin memory. J Pathol 2004; 204: 39-46.

37 Carnaud C, Lee D, Donnars O, et al. Cross-talk between cells of the innate immune system: NK t cells rapidly activate NK cells. J Immunol 1999; 163: 4647-50.

38 Komatsu T, Moriya N, Shiohara T. T cell receptor (TCR) repertoire and function of human epidermal T cells: restricted TCR Vα-Vβ genes are utilized by T cells residing in the lesional epidermis in fixed drug eruption. Clin Exp Immunol 1996; 104: 343-50.

39 Shiohara T, Mizukawa Y, Teraki Y, Fukuda T. Gamma-delta T cells with emphasis on their functional role in the epidermis. Chem Immunol 2001; 79: 66-86.

40 Shiohara T, Mizuakwa Y, Teraki Y. Pathophysiology of fixed drug eruption: the role of skin-resident T cells. Curr Opin Allergy Clin Immunol 2002; 2: 317-23.

41 Choi HJ, Ku JK, Kim MY, Kang H, Cho SH, Kim HO, Park YM. Possible role of Fas/Fas ligand-mediated apoptosis in the pathogenesis of fixed drug eruption. Br J Dermatol 2006; 154: 419-25.

42 Nassif A, Bensussan A, Boumsell L, Deniaud A, Mosleehi H, Wolkenstein P, Bagot M, Roujeau JC. Toxic epidermal necrolysis: effector cells are drug-specific cytotoxic T cells. J Allergy Clin Immunol 2004; 114: 1209-15.