Delayed reepithelialization and scarring deregulation following drug-induced toxic epidermal necrolysis

1. Introduction 

Toxic epidermal necrolysis (TEN) is a rare potentially lethal disease characterized by sudden necrosis involving the epidermis as well as the genital and oropharyngeal mucosae [1]. The disease corresponds to an adverse drug reaction, especially to sulfonamides, phenytoin, antibiotics, non-steroidal anti-inflammatory drugs and allopurinol. Fatal outcome may occur in about one-third of the patients. Various long-term sequels can affect the lungs, heart, joints, kidneys and gut. However, the most common persistent damage involves the eyes, nails, genital mucosae and skin.

In survivors under supportive care, reepithelialization of the eroded lesions is notoriously rapid [1]. The dermis remains uninvolved by the disease except an early increased number of Factor XIIIa positive type I-dendrocytes (DD-1) in the adventitial sector [2]. As a rule, there is no or little scarring of the involved skin. In particular, and in contrast with burns, there is apparently no increased risk for hypertrophic or keloidal scarring [1].

We report a patient who developed severe TEN with unfortunate delayed withdrawal of the culprit drug. Unusual protracted skin erosions were followed by extensive hypertrophic and keloidal scarring despite coverage of the eroded areas by cultured keratinocyte allografts. The clinical impact of the delayed withdrawal of the culprit drug is discussed in relation to the retarded epidermal repair and the abnormal scarring. The usefulness of keratinocyte allografts in TEN treatment is also questioned.

2. Case report 

The patient is a 51-year-old woman who suffered from systemic lupus erythematous responsible for a moderate alteration of the renal function. She received allopurinol 200mg daily for controlling elevated serum levels of uric acid. The patient developed a cutaneous rash corresponding to TEN 14 days after initiation of this oral treatment. The erythematous rash was studded with discrete bullae which progressively spread during the next 10 days over the back and limbs. She complained of general malaise with fever, headache and conjunctivitis, but she did not discontinue the allopurinol intake. On the 11th day after TEN onset, she consulted a dermatologist who biopsied a bullous lesion. The diagnosis of TEN was confirmed histologically on the basis of the presence of a subepidermal blister with rare inflammatory mononuclear cells scattered between confluent necrotic keratinocytes. The patient was hospitalized in a burn unit where she was placed on a fluidized bed. Lesions were treated with topical antiseptics. The patient also benefited from supportive care including daily baths. Intravenous immunoglobulins [IVIg] were perfused at a 1g/kg daily dosage for three consecutive days. Corticosteroids were not administered.

Despite IVIg treatment, blisters and epidermal sloughing worsened to reach more than 80% of the body surface 7 days after admission. Erosions of the ocular, genital and oral mucosae also developed. Three cutaneous biopsies were performed at 2-day intervals in both taut blisters and clinically uninvolved skin searching for histological clues explaining the severe course of the disease and predicting its outcome compared to the TEN biopsies retrieved from our files. The only singularity found in the present patient was the small number of Factor XIIIa+ DD-1 cells dispersed in a lymphoid infiltrate. The patient's general condition progressively improved during the following weeks. Partial reepidermization took place, but large eroded areas of the trunk, thighs and upper limbs were not yet closed 1 month after admission. Daily cutaneous swabs performed over several sites of the eroded areas always remained sterile. No septicemia occurred. One month after admission, i.e. more than 40 days after the onset of the disease, biopsies were performed again in TEN involved and uninvolved skin. Immunohistochemistry revealed the presence of sparse Factor XIIIa+ DD-1 expression cells. There was a moderate amount of CD45R0+ T lymphocytes admixed with numerous CD68+ macrophages. Myofibroblasts enriched in α-actin were almost absent. The α1 (IV) collagen chain was not expressed in the eroded dermis.

A first allograft (600cm2) of frozen cultured keratinocytes (Cryoceal®, Xcellentis, Ghent, Belgium) was applied on randomized eroded areas of the body, followed by two other allografts performed 8 days (900cm2) and 15 days (800cm2) later. Before each allograft procedure, skin biopsies were taken from two TEN eroded lesions located on the grafted site of the abdomen and on the ungrafted site of the left thigh, respectively (Fig. 1a and b). Repeated skin swabs remained sterile during this treatment phase. No clinical difference was noted between the treated and the control eroded areas. Immunohistochemical examinations did not show any difference between the two sites with regard to the numerical densities in Factor XIIIa+ DD-1, α-actin+ myofibroblasts, CD68+, macrophages and CD45RO+ T cells, respectively. Considering the evolution in time, Factor XIIIa+ DD-1 increased in number at D8 and D15 post-grafting. No other change was disclosed in the dermis during that period. At both sites, no deposits of α1 (IV) collagen chain were disclosed on top of the dermis. There was no sign of vasculitis or infection. The retarded epithelialisation delayed the hospital discharge until 2 months after admission. At that time, some small eroded areas were still present on the arms.

View Large Image Download to PowerPoint

Fig. 1. (a and b) Erosive scars on the abdomen (a) and left thigh (b) 2 months after TEN onset. (c and d) Similar thickness and redness of the hypertrophic scars on the abdomen (treated with cultured keratinocytes) and the left thigh (control) 1 year after TEN onset.

At the end of hospitalization, both the keratinocyte-treated and control areas progressively developed similar erythematous hypertrophic scarring. Similar lesions developed on other previously TEN-affected areas, especially on the upper limbs and chest. The colour of the scars as assessed by narrow band spectrophotometry (Mexameter® MX16, C+K electronic, Cologne, Germany) showed no difference in the erythema index E on the two comparative sites. Ten months after hospital discharge, i.e. 1 year after the TEN onset, the hypertrophic scars of the keratinocyte-treated and control areas began to flatten simultaneously (Fig. 1c and d). Conventional histology and immunohistochemistry performed on skin biopsies taken 3 and 6 months after grafting looked similar on both sites. At D90, they showed numerous large dendritic and globular Factor XIIIa+ DD-1 expression cells admixed with fewer and smaller α-actin+ myofibroblasts. At D180, a marked thickening of the collagen bundles was present. Moderate numbers of Factor XIIIa+ DD-1, CD68+ macrophages and α-actin+ myofibroblasts were present throughout the dermis. Eighteen months after TEN onset, the abnormal scarring on the upper arms presented as thick and erythematous keloidal tumours (Fig. 2). Histological sections showed a massive accumulation of thickened hyalinized collagen bundles. These lesions were treated by repeated triamcinolone injections and by endermology pressure therapy. A moderate improvement was observed 6 months after initiating this treatment.

View Large Image Download to PowerPoint

Fig. 2. Post-TEN keloidal scars of the right arm evolving for 1 year.

3. Discussion 

Fibroblasts and myofibroblasts play key roles in the scarring process [3], [4]. In addition, inflammatory cells, DD-1 and keratinocytes, which orchestrate the normal healing process may also influence fibroblasts via the secretion of multiple growth factors and cytokines [3], [5], [6], [7], [8]. Macrophages can possibly initiate abnormal scarring by releasing fibroblast-activating cytokines such as transforming growth factor-β (TGFB-β) and platelet-derived growth factor (PDGF). They also produce interleukin-1 (IL-1) which is important for the extracellular matrix degradation by inducing the metalloproteinase release and activity [5], [9]. In addition, many lymphocyte-mediated cytokines modulate fibroplasia, especially TGF-β, tumour necrosis factor-β and interferon β [5], [7].

Persistence of activated keratinocytes in predisposed individuals or a prolonged time to reepithelialization could lead to hypertrophic or keloidal scar formation [3]. Indeed, the epidermis of hypertrophic scars was found to exhibit reduced IL-1α expression and increased PDGF expression. IL-1α plays a role in the extracellular matrix degradation at the wound site whereas PDGF is involved in tissue fibrosis [5], [9]. Keratinocytes covering keloids also increase TGF-β1 expression in fibroblasts suggesting that besides causing an excess in production of collagen, an increased resistance to apoptosis may develop in keloid fibroblasts [10]. In our TEN patient, frozen keratinocyte allografts failed to show any effect on the Factor XIIIa+ DD-1, α-actin+ myofibroblasts, CD45RO T lymphocytes and CD68+ macrophages. They also had no clinical impact on the healing process since no difference was evidenced between the grafted and ungrafted sites.

In the presently reported TEN patient, the number of Factor XIIIa+ DD-1 was surprinsingly low in the early stage of the disease. This contrasted with the usual DD-1 hyperplasia found in TEN [2]. As discussed below, the DD-1 rarefaction in our patient might be related to the late withdrawal of the culprit drug. Later on, at the initiation of reepithelialization of the protracted erosions, an increased Factor XIIIa+ DD-1 population was disclosed, suggesting a positive effect on wound closure. It is noteworthy that the changes in the number of Factor XIIIa+ DD-1 are associated with an increase in their size. They became first more dendritic at D15 post-grafting, and they appeared globular at D90. They resumed a normal aspect at D180. These variations in cell tensegrity may be found in Factor XIIIa+ DD-1 present in some connective tissue disorders [11], [12]. The tensegrity changes in the Factor XIIIa+ DD-1 observed in the present TEN case were not seen in other dermal cells, particularly in the α-actin+ myofibroblasts. These cells are normally involved in wound healing when contraction of the granulation tissue takes place. This process, which occurs after loss of dermal tissue to ensure wound contraction [4], is not operative in TEN healing. In the present case, there was no immunohistochemical sign suggesting dermal cell mobility during the protracted healing process.

It has been postulated that the phagocytic properties of DD-1 contribute to limit any evolving fibrotic process. Indeed, DD-1 are able to remove fragmented collagen fibers [13]. However in this TEN case, the overexpression of Factor XIIIa+ DD-1 did not prevent hypertrophic scarring.

Usually TEN skin erosions heal without dermal scarring [1] and hypertrophic or keloid scars have been rarely described [14]. The keloid and hypertrophic pattern in our patient may be linked to the late withdrawal of the culprit drug with a prolonged and intense effect on keratinocytes. This situation probably led to high levels of proapoptotic and proinflammatory keratinocyte-derived cytokines production, and to a dense inflammatory infiltrate composed of T lymphocytes and reactive macrophages that can in turn influence the fibroblast function. Moreover, high concentrations of TNF-α, that is one of the major cytokines involved in TEN lesions, might be toxic for Factor XIIIa+ DD-1 thus impairing their function in the healing process. Clinically, abnormal scarring most commonly occurs after reepithelialization has been delayed [15].

Autologous and allogenic cultured keratinocyte sheets or suspensions have been previously used in burn patients to prevent or to treat abnormal scarring [16], [17]. They have also been employed to boost healing in TEN patients [18]. It has also been demonstrated that cryopreserved cultured keratinocytes retain their healing potential [19]. In the present TEN patient, no differences were found in the healing rate and in the quality of scarring between the keratinocyte-grafted and ungrafted areas. The main cause of failure of cultured keratinocytes in wound treatment is a bacterial overload in the wound bed [19], but this was not the case in our TEN patient.

In conclusion, TEN may be associated with delayed healing, especially after late withdrawal of the culprit drug. Prolonged TEN cutaneous erosions can lead to keloidal and hypertrophic scarring. In the present TEN case, allogenic cultured keratinocytes did not improve the healing rate and the quality of the scarring process.