- Email: [email protected]
Prognostic value of histologic features of toxic epidermal necrolysis
DERMATOPATHOLOGY
Prognostic value of histologic features of toxic epidermal necrolysis Laurence Valeyrie-Allanore, MD,a Sylvie Bastuji-Garin, MD, PhD,b Sarah Guegan, MD, PhD,c Nicolas Ortonne, MD, PhD,d Martine Bagot, MD, PhD,e Jean-Claude Roujeau, MD,a Jean E. Revuz, MD,a Janine Wechsler, MD,d and Pierre Wolkenstein, MD, PhDa Creteil and Paris, France Background: The prognosis of toxic epidermal necrolysis (TEN), Stevens-Johnson syndrome (SJS), and SJS/TEN overlap syndrome has been assessed using a disease-specific severity score (SCORTEN) based on clinical and laboratory data. Histologic data may improve outcome prediction. Objective: We sought to evaluate whether dermal mononuclear infiltration and epidermal necrosis predict survival of patients with TEN, SJS, or SJS/TEN. Methods: We conducted a retrospective review of clinical records and skin biopsy specimens read without knowledge of clinical data. Results: We identified 108 patients (SJS, n = 42; SJS/TEN, n = 36; TEN, n = 30). Overall mortality was 21.3%. Dermal infiltration and epidermal necrosis were not associated with time from disease onset to biopsy. Extensive dermal infiltrates were seen in 19 (18.5%) patients and full-thickness epidermal necrosis in 56 (52%) patients. Dermal infiltrate severity was not associated with day-1 (D1) SCORTEN or hospital death. Epidermal necrosis severity showed trends toward associations with D1 SCORTEN (P = .11) and hospital death (P = .06). In univariate analyses, full-thickness epidermal necrosis was significantly associated with hospital death (32.1% vs 11.4%, P = .017) and worse D1 SCORTEN values (1.98 6 1.29 vs 1.55 6 1.21; P = .04). In the bivariate analysis, however, D1 SCORTEN remained significantly associated with hospital death (odds ratio = 3.07, 95% confidence interval 1.83-5.16) but the association with full-thickness epidermal necrosis was no longer significant (odds ratio = 2.02, 95% confidence interval 0.65-7.12). Limitations: Retrospective study design and indirect assessment of progression are limitations. Conclusion: Full-thickness epidermal necrosis was associated with mortality but did not independently predict hospital death after adjustment based on the SCORTEN value. Dermal infiltrate severity was not associated with hospital death. ( J Am Acad Dermatol 10.1016/j.jaad.2011.10.007.) Key words: full-thickness necrosis; mortality; Stevens-Johnson syndrome; toxic epidermal necrolysis.
oxic epidermal necrolysis (TEN), StevensJohnson syndrome (SJS), and SJS/TEN overlap syndrome are rare and severe mucocutaneous diseases located along a continuous spectrum.
T
Drugs are the most common causes. The histologic hallmark of these diseases is necrosis with detachment of the epidermis.1-5 The overall annual incidence is 2/106 population.6 Patients are classified
From the Departments of Dermatology,a Clinical Research and Public Health,b and Pathology,d Assistance Publique - H^ opitaux de Paris, H^ opital Henri-Mondor, Universite Paris Est, LIC EA4393 (Drs Valeyrie-Allanore, Bastuji-Garin, and Wolkenstein), F-94010, Cr eteil; and the Dermatology Departments at H^ opital Tenonc and H^ opital Saint Louis, Paris, France.e The first two authors contributed equally to the study. Funding sources: None. Conflicts of interest: None declared. Study approved by ethics committee.
Presented in part at the Journees Dermatologiques de Paris in Paris, France in December 2006 and at the European Society of Dermatological Research in Paris, France in September 2006. Accepted for publication October 4, 2011. Reprint requests: Laurence Valeyrie-Allanore, MD, Service de Dermatologie, H^ opital Henri-Mondor, F-94010 Creteil, Cedex, France. E-mail: [email protected]. Published online November 15, 2011. 0190-9622/$36.00 Ó 2011 by the American Academy of Dermatology, Inc. doi:10.1016/j.jaad.2011.10.007
e1
J AM ACAD DERMATOL
e2 Valeyrie-Allanore et al
according to the percentage of the body surface area Histopathological examination exhibiting epidermal detachment (BSA%), which is Hematoxylin-eosinestained sections were rouless than 10% in SJS, more than 30% in TEN, and 10% tinely reviewed by two of us (L. V-A. and J. W.) to 30% in SJS/TEN overlap syndrome.7 Overall morworking together to achieve a consensus, without tality is about 22%.6 The definitive diagnosis relies on knowledge of the clinical data or outcome. A stanexamination of a skin biopsy specimen, which shows dardized form was used to collect data on adnexal epidermal necrosis and a very slight to absent infilnecrosis (including necrosis of the follicles, piloertrate in the papillary dermis ector muscles, sebaceous composed chiefly of lymphoglands, and eccrine sweat CAPSULE SUMMARY cytes and macrophages. glands); dermal mononuA severity-of-illness score clear infiltration graded In toxic epidermal necrolysis, dermal for SJS, TEN, and SJS/TEN, semiquantitatively as sparse, infiltrate density has been reported to called the SCORTEN, has moderate, or extensive predict death. been validated and con(Fig 1); and epidermal necrofirmed as a tool for predicting In 108 database patients, dermal sis assessed semiquantitasurvival.8-10 The SCORTEN is infiltrate density was associated neither tively as mild (confined to based on 7 clinical and labowith the day-1 SCORTEN value nor the deep epidermal layers), ratory risk factors and can hospital death. Full-thickness epidermal moderate (prominent but not range from 0 to 7, with higher necrosis was associated with death in confluent), or severe (fullscores predicting a higher the univariate analysis but did not thickness epidermal necrorisk of death. independently predict hospital death sis) (Fig 2). In addition to these 7 facafter adjustment of the day-1 SCORTEN tors, the severity of the dervalue. Clinical data mal mononuclear infiltrate as For each patient, we abDermal infiltrate density and severity of assessed histologically has stracted the following from epidermal necrosis should not be routinely been suggested to predict the database: BSA% involved taken into account when treating survival in patients with on day 1 (D1) (first 24 hours patients with toxic epidermal necrolysis. TEN.11 after admission), D1 The aim of this study was SCORTEN, final diagnosis to evaluate whether skin biopsy specimen findings (SJS, TEN, or SJS/TEN overlap based on greatest helped to predict survival in patients with SJS, TEN, BSA% involved), dates of disease onset and hospital or SJS/TEN independently from the SCORTEN value. admission, date of skin biopsy, and vital status at The patients were those in a database established discharge. The times from disease onset to admission prospectively for a previous study.9 We retrospecand to skin biopsy were computed. The day of tively reviewed the skin biopsy specimens to assess disease onset was defined as the first day with the severity of dermal mononuclear infiltration, epcutaneous or mucous membrane lesions that were idermal necrosis, and adnexal necrosis. not ascribable to other conditions and that were followed within 3 days by ulcers or blisters as defined elsewhere.3 Patients were classified accordMETHODS ing to vital status and D1 BSA% (\10%, 10%-30%, or Patients [30%). Disease progression during the hospital stay We identified 108 patients via a database that had was indirectly evaluated as a change in the abovebeen established prospectively to evaluate the described categories. SCORTEN measured during the first 5 hospital days.9 Inclusion and exclusion criteria have been Statistical analysis described elsewhere.7 Briefly, patients were eligible Qualitative data were described as number (perfor inclusion in the database if the discharge diagcent) and analyzed using the Fisher exact test. nosis was SJS, TEN, or SJS-TEN overlap and if they Quantitative data were described as mean (6SD) received symptomatic treatment only. Patients inor median (interquartile range) as appropriate and cluded in therapeutic trials and/or those who rewere analyzed using the nonparametric Kruskalceived potential curative treatment were excluded Wallis test. To assess whether dermal infiltrate severfrom the database.12-14 Of the 144 patients in the ity on D1 was associated with D1 BSA%, disease database, 108 had skin biopsy specimens available at progression, D1 SCORTEN, and/or hospital death, the pathology department. In the current study, we we used the Fisher exact test or nonparametric reviewed all 108 slides and looked for correlations Kruskal-Wallis test as appropriate. Similar analyses with the prognosis. d
d
d
J AM ACAD DERMATOL
Valeyrie-Allanore et al e3
Abbreviations used: BSA%: CI: D1: SJS: TEN:
percentage of body surface area exhibiting epidermal detachment confidence interval day 1 Stevens-Johnson syndrome toxic epidermal necrolysis
were performed for the extent of epidermal necrosis. To compare our results with those of Quinn et al,11 we also performed the same analyses in the subgroups of patients whose skin biopsy specimen was obtained within 48 hours after admission or whose discharge diagnosis was TEN. The SCORTEN has been validated as a severity-ofillness score that accurately predicts survival in patients with SJS, TEN, or SJS/TEN. Therefore, for histologic variables associated with hospital death (P \ .15) in the univariate analysis, we planned to conduct bivariate analyses adjusting for the SCORTEN value. Odds ratios were estimated with their 95% confidence intervals (CI). These analyses were restricted to patients who had a skin biopsy within 48 hours after admission. All tests were two-tailed. P values less than or equal to .05 were considered statistically significant. Data were analyzed using statistical software (Stata, Release 8.0, StataCorp, College Station, TX). This study was approved by an ethics committee on May 10, 2011.
RESULTS Study population Mean age of the 108 database patients with available skin biopsy specimens was 47.5 6 20.1 years (range, 13-92). Among these patients, 58 (53.7%) were male and 50 female. Median time from disease onset to admission was 4 days (3-6) and median time from disease onset to skin biopsy was 6 days (5-9). The skin biopsy was performed within 48 hours after admission in 99 (91.7%) patients. Mean D1 SCORTEN was 1.75 6 1.24 (Table I). Of the 100 patients whose D1 BSA% was no greater than 30%, 22 experienced enough disease progression to move to another category of body surface area involvement. The discharge diagnosis was SJS in 42 patients, SJS/TEN in 36, and TEN in 30. In all, 23 patients died, yielding a hospital mortality of 21.3% (95% CI 13.6-29.0). Histologic findings Table I reports the main histologic findings. The dermal infiltrate was sparse or moderate in 84
Fig 1. Dermal infiltrate in patients with Stevens-Johnson syndrome, toxic epidermal necrolysis, or overlap syndrome evaluated as sparse (A), moderate (B), or extensive (C). (A to C, Hematoxylin-eosin stain; original magnifications: 3100.)
(81.5%) patients. Full-thickness epidermal necrosis was observed in 56 (56%) patients. Neither the severity of the dermal infiltrate nor the severity of the epidermal necrosis was significantly influenced by the time from disease onset to skin biopsy (P = .7 and P = .42, respectively). Table II reports the associations between infiltrate severity and the clinical and outcome data in the univariate analysis. Dermal infiltrate severity showed a significant negative association with D1 BSA% (P = .04): D1 BSA% values lower than 10% were more common among patients having moderate or extensive infiltrates. Nevertheless, dermal infiltrate severity was not significantly associated with D1 SCORTEN, disease progression, or hospital death (Table II). Similar results were obtained in analyses restricted to the 93 patients whose skin biopsy was performed within 48 hours after admission or to the 29 patients with a discharge diagnosis of TEN (data not shown). Table III reports the associations between epidermal necrosis severity and the clinical and outcome data. Epidermal necrosis severity was significantly associated with disease progression (P = .002) and nonsignificantly associated with the D1 SCORTEN (P = .11) and hospital death (P = .06). The results were similar when the analysis was restricted to the 93 patients whose skin biopsy was performed within 48 hours after admission. Hospital mortality was
J AM ACAD DERMATOL
e4 Valeyrie-Allanore et al
Table I. Clinical and histologic characteristics of 108 study patients
Fig 2. Epidermal necrosis in patients with StevensJohnson syndrome, toxic epidermal necrolysis, or overlap syndrome evaluated as minor (A), moderate (B), or full thickness (C). (A to C, Hematoxylin-eosin stain; original magnifications: 340.)
significantly higher among patients with fullthickness epidermal necrosis compared with those with mild or moderate epidermal necrosis, both in the overall population (18/56, 32.1% vs 5/44, 11.4%; P = .017) and in the 93 patients with early biopsies (17/53, 32.1% vs 5/40, 12.5; P = .047). D1 SCORTEN was significantly higher (indicating greater severity) among patients with full-thickness epidermal necrosis compared with those with mild or moderate epidermal necrosis (2.0 6 1.3 vs 1.55 6 1.21; P = .04 and 2.0 6 1.3 vs 1.5 6 1.24; P = .02). In the bivariate analyses, the SCORTEN value remained significantly associated with hospital death (odds ratio 3.07, 95% CI 1.83-5.16) but full-thickness epidermal necrosis was not significantly associated with hospital death (odds ratio 2.02, 95% CI 0.65-7.12).
D1-BSA%, n (%) \10% 57 (52.8) 10%-30% 43 (39.8) [30% 8 (7.4) D1-SCORTEN, mean 6 SD 1.75 6 1.24 Discharge diagnosis, n (%) SJS 42 (38.9) SJS/TEN 36 (33.3) TEN 30 (27.8) Infiltrate severity (n = 103), n (%) Sparse 33 (32.0) Moderate 51 (49.5) Extensive 19 (18.5) Epidermal necrosis severity (n = 100), n (%) Minor 25 (25.0) Moderate 19 (19.0) Severe 56 (56.0) Adnexal necrosis, n (%) Follicles No 36 (34.0) Yes 27 (25.5) No follicles seen 43 (40.6) Eccrine sweat glands No 95 (89.6) Yes 3 (2.8) No glands seen 8 (7.6) Sebaceous glands No 28 (26.4) Yes 8 (7.6) No glands seen 70 (66.0) Piloerector muscle No 52 (49.1) Yes 16 (15.1) No muscle seen 38 (35.9) D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on first hospital day; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; SJS, Stevens-Johnson syndrome; SJS/TEN, StevensJohnson syndrome/toxic epidermal necrolysis overlap syndrome; TEN, toxic epidermal necrolysis.
In the subgroup of 29 patients with a discharge diagnosis of TEN, epidermal necrosis severity was associated neither with hospital death nor with the D1 SCORTEN. Finally, in patients with adnexal necrosis, neither eccrine sweat gland necrosis nor follicular necrosis was significantly associated with death (data not shown).
DISCUSSION In this population of 108 patients with SJS, TEN, or SJS/TEN, none of the histologic criteria studied was significantly associated with hospital mortality after
J AM ACAD DERMATOL
Valeyrie-Allanore et al e5
Table II. Dermal infiltrate severity according to percentage of body surface area with epidermal detachment on day 1, day-1 SCORTEN value, disease progression, and mortality Dermal infiltrate severity
Overall population (n = 103) D1-BSA%, n (%) \10% 10%-30% [30% D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Skin biopsy within 48 h (n = 95) D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Discharge diagnosis of TEN (n = 29) D1-SCORTEN, mean 6 SD Hospital death, n (%)
Sparse
Moderate
Extensive
(n = 33)
(n = 51)
(n = 19)
11 (33.3) 18 (54.6) 4 (12.1) 1.97 (1.21) 10 (30.3) 9 (27.3) (n = 28) 1.96 (1.26) 9 (32.1) 8 (28.6) (n = 13) 2.38 (1.12) 4 (30.8)
29 (56.9) 20 (39.2) 2 (3.9) 1.63 (1.3) 14 (27.5) 8 (15.7) (n = 50) 1.62 (1.31) 14 (28.0) 8 (16.0) (n = 10) 2.6 (1.58) 4 (40.0)
13 (68.4) 4 (21.1) 2 (10.5) 1.63 (1.16) 6 (31.6) 4 (21.0) (n = 17) 1.65 (1.22) 6 (35.3) 4 (23.5) (n = 6) 1.83 (0.75) 3 (50.0)
P value
.04* .33y .93* .44* .40y .83* .39* .61y .70*
D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on day of admission; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; TEN, toxic epidermal necrolysis. *P value by Fisher exact test. y P value by nonparametric Kruskal-Wallis test.
adjustment on the SCORTEN values determined on the day of admission. Overall mortality of SJS and TEN was recently reported to be approximately 22%.6 The BSA% involved at admission is one of the main prognostic factors and is among the 7 factors included in the SCORTEN severity-of-illness score.8,9 In addition to these 7 factors, dermal infiltrate severity was reported by Quinn et al11 to be associated with hospital death. In our population, we found no significant association between dermal infiltrate severity and hospital death. This discrepancy may be related to a difference in the timing of the skin biopsy. In the previous study,11 mean time from disease onset to skin biopsy was 9.4 to 10.2 days (depending on histologic grade). Thus, no further skin detachment was occurring at the time of skin biopsy in most of the patients. Furthermore, dermal infiltrate severity may have been influenced by events that occurred during the interval since disease onset, such as infection, local care, or specific immunosuppressive treatment. In our study, 92% of skin biopsies were performed within 48 hours after admission and median time from disease onset to skin biopsy was 6 days, and consequently potential effects of confounding factors were minimized. For a prognostic tool, a short time to result availability is crucial. Therefore, only early skin biopsy specimen results are relevant here. Quinn et al11 included only patients with TEN (BSA% [ 30). Mortality was 27%, 53%, and 71% in patients with sparse, moderate, and
severe dermal infiltrates, respectively, leading Quinn et al11 to suggest a role for immunocytes in the severity and prognosis of TEN. We included consecutive patients regardless of BSA%, and our overall mortality was only half that recorded by Quinn et al.11 However, dermal infiltrate severity was not associated with hospital mortality in our subgroup of patients with TEN (n = 29), regardless of time to biopsy. Epidermal necrosis severity was associated with hospital mortality in our study. Thus, the presence of full-thickness epidermal necrosis was significantly associated with worse D1 SCORTEN values and with higher hospital mortality, regardless of time to biopsy. Furthermore, full-thickness epidermal necrosis was associated with disease progression as indirectly assessed by a change in disease category. Epidermal necrosis severity may be directly related to the magnitude of the immunologic and chemokine storm that characterizes SJS and TEN. Greater epidermal necrosis on biopsy specimens, which are usually taken at the junction between involved and normal-appearing skin, may therefore indicate a higher risk of progression with delayed healing and subsequent complications such as local and systemic infections. However, given the retrospective study design, we had no control over the biopsy specimen sites. This point is among the limitations of our study. We also studied possible associations between adnexal necrosis and hospital mortality. Follicular
J AM ACAD DERMATOL
e6 Valeyrie-Allanore et al
Table III. Epidermal necrosis severity according to percentage of body surface area with epidermal detachment on day 1, SCORTEN value on day 1, disease progression, and mortality Epidermal necrosis
Overall population (n = 100) D1-BSA%, n (%) \10% 10%-30% [30% D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Skin biopsy within 48 h (n = 93) D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Discharge diagnosis of TEN (n = 29) D1-SCORTEN, mean 6 SD Hospital death, n (%)
Minor
Moderate
Severe
(n = 25)
(n = 19)
(n = 56)
16 (64.0) 8 (32.0) 1 (4.0) 1.56 (1.16) 2 (8.0) 3 (12.0) (n = 23) 1.48 (0.71) 2 (8.7) 3 (13.0) (n = 2) 2.5 (0.71) 1 (50.0)
8 (42.1) 8 (42.1) 3 (15.8) 1.53 (1.31) 3 (18.8) 2 (10.5) (n = 17) 1.53 (1.37) 3 (17.7) 2 (11.8) (n = 6) 2.67 (1.86) 1 (16.7)
27 (48.2) 26 (46.4) 3 (5.4) 1.98 (1.29) 24 (42.9) 18 (32.1) (n = 53) 2.0 (1.3) 23 (43.4) 17 (32.1) (n = 21) 2.33 (1.15) 10 (47.6)
P value*
.34* .11y .002* .06* .07y .005* .095* .90y .46*
D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on first hospital day; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; TEN, toxic epidermal necrolysis. *P value by Fisher exact test. y P value by nonparametric Kruskal-Wallis test.
necrosis is a rarely described but common feature of SJS and TEN.15 In our study, 43% of patients had follicular necrosis, which was not significantly associated with hospital mortality. The specific sequence of molecular and cellular events leading to SJS or TEN has been partially elucidated. The lesions contain CD81 T lymphocytes that exhibit drug-specific, major histocompatibility complexerestricted cytotoxicity against keratinocytes. Immunopathological studies have demonstrated that the cytotoxic effects of these CD81 T cells are mediated by a variety of soluble mediators, not only perforin and granzyme B, but also granulysin and perhaps other compounds.16-18 The paucicellularity of the dermal infiltrate compared with other inflammatory diseases is difficult to understand but suggests a role for mechanisms such as indirect keratinocyte apoptosis via specific molecular contact. In sum, epidermal necrosis severity was significantly associated with hospital death in our study. However, the association was no longer significant after adjustment for the SCORTEN value on the day of admission. REFERENCES 1. Stevens AM, Johnson FC. A new eruptive fever associated with stomatitis and ophthalmia: report of two cases in children. Am J Dis Child 1922;24:526-33. 2. Valeyrie-Allanore L, Sassolas B, Roujeau JC. Drug-induced skin, nail and hair disorders. Drug Saf 2007;30:1011-30.
3. Roujeau J-C, Stern RS. Severe cutaneous adverse reactions to drugs. N Engl J Med 1994;331:1272-85. 4. Auquier-Dunant A, Mockenhaupt M, Naldi L, Correia O, Schroder W, Roujeau JC, SCAR Study Group. Correlations between clinical patterns and causes of erythema multiforme majus, Stevens-Johnson syndrome, and toxic epidermal necrolysis: results of an international prospective study. Arch Dermatol 2002;138:1019-24. 5. Roujeau JC, Kelly JP, Naldi L, Rzany B, Stern RS, Anderson T, et al. Medication use and the risk of Stevens-Johnson syndrome or toxic epidermal necrolysis. N Engl J Med 1995;333: 1600-7. 6. Mockenhaupt M, Viboud C, Dunant A, Halevy S, Bouwes Bavinck JN, Sidoroff A, et al. Stevens Johnson syndrome and toxic epidermal necrolysis: assessment of medication risks with emphasis on recently marketed drugs; the EuroSCAR study. J Invest Dermatol 2008;128:35-44. 7. Bastuji-Garin S, Rzany B, Stern RS, Shear NH, Naldi L, Roujeau JC. Clinical classification of cases of toxic epidermal necrolysis, Stevens-Johnson syndrome, and erythema multiforme. Arch Dermatol 1993;129:92-6. 8. Bastuji-Garin S, Fouchard N, Bertocchi M, Roujeau JC, Revuz J, Wolkenstein P. SCORTEN: a severity-of-illness score for toxic epidermal necrolysis. J Invest Dermatol 2000;115:149-53. 9. Guegan S, Bastuji-Garin S, Poszepczynska-Guigne E, Roujeau JC, Revuz J. Performance of the SCORTEN during the first five days of hospitalization to predict the prognosis of epidermal necrolysis. J Invest Dermatol 2006;126:272-6. 10. Cartotto R, Mayich M, Nickerson D, Gomez M. SCORTEN accurately predicts mortality among toxic epidermal necrolysis patients treated in a burn center. J Burn Care Res 2008;29: 141-6. 11. Quinn A, Brown K, Bonish B, Curry J, Gordon KB, Sinacore J, et al. Uncovering histologic criteria with prognostic significance in toxic epidermal necrolysis. Arch Dermatol 2005;141:683-7.
J AM ACAD DERMATOL
12. Wolkenstein P, Latarjet J, Roujeau JC, Duguet C, Boudeau S, Vaillant L, et al. Randomized comparison of thalidomide versus placebo in toxic epidermal necrolysis. Lancet 1998;352:1586-9. 13. Bachot N, Revuz J, Roujeau JC. Intravenous immunoglobulin treatment for Stevens-Johnson and toxic epidermal necrolysis. Arch Dermatol 2003;139:33-6. 14. Valeyrie-Allanore L, Wolkenstein P, Brochard L, Ortonne N, Ma^ıtre B, Revuz J, et al. Open trial of cyclosporin treatment for Stevens-Johnson syndrome and toxic epidermal necrolysis. Br J Dermatol 2010;163:847-53. 15. Braun-Falco O. Das Lyell syndrome. Bern (Switzerland): Haus Haker Verlog, 197061-70.
Valeyrie-Allanore et al e7
16. Nassif A, Bensussan A, Dorothee G, Mamai-Chouaib F, Bachot N, Bagot M, et al. Drug specific cytotoxic T-cells in the skin lesions of a patient with toxic epidermal necrolysis. J Invest Dermatol 2002;118:728-33. 17. Chung WH, Hung SI, Yang JY, Huang SP, Wei CY, Chin SW, et al. Granulysin is a key mediator for disseminated keratinocytes death in Stevens-Johnson syndrome and toxic epidermal necrolysis. Nat Med 2008;14:1343-50. 18. Viard I, Werhli P, Bullani R, Schneider P, Holler N, Salomon D, et al. Inhibition of toxic epidermal necrolysis by blockade of CD95 with human intravenous immunoglobulin. Science 1998; 282:490-3.
Prognostic value of histologic features of toxic epidermal necrolysis Laurence Valeyrie-Allanore, MD,a Sylvie Bastuji-Garin, MD, PhD,b Sarah Guegan, MD, PhD,c Nicolas Ortonne, MD, PhD,d Martine Bagot, MD, PhD,e Jean-Claude Roujeau, MD,a Jean E. Revuz, MD,a Janine Wechsler, MD,d and Pierre Wolkenstein, MD, PhDa Creteil and Paris, France Background: The prognosis of toxic epidermal necrolysis (TEN), Stevens-Johnson syndrome (SJS), and SJS/TEN overlap syndrome has been assessed using a disease-specific severity score (SCORTEN) based on clinical and laboratory data. Histologic data may improve outcome prediction. Objective: We sought to evaluate whether dermal mononuclear infiltration and epidermal necrosis predict survival of patients with TEN, SJS, or SJS/TEN. Methods: We conducted a retrospective review of clinical records and skin biopsy specimens read without knowledge of clinical data. Results: We identified 108 patients (SJS, n = 42; SJS/TEN, n = 36; TEN, n = 30). Overall mortality was 21.3%. Dermal infiltration and epidermal necrosis were not associated with time from disease onset to biopsy. Extensive dermal infiltrates were seen in 19 (18.5%) patients and full-thickness epidermal necrosis in 56 (52%) patients. Dermal infiltrate severity was not associated with day-1 (D1) SCORTEN or hospital death. Epidermal necrosis severity showed trends toward associations with D1 SCORTEN (P = .11) and hospital death (P = .06). In univariate analyses, full-thickness epidermal necrosis was significantly associated with hospital death (32.1% vs 11.4%, P = .017) and worse D1 SCORTEN values (1.98 6 1.29 vs 1.55 6 1.21; P = .04). In the bivariate analysis, however, D1 SCORTEN remained significantly associated with hospital death (odds ratio = 3.07, 95% confidence interval 1.83-5.16) but the association with full-thickness epidermal necrosis was no longer significant (odds ratio = 2.02, 95% confidence interval 0.65-7.12). Limitations: Retrospective study design and indirect assessment of progression are limitations. Conclusion: Full-thickness epidermal necrosis was associated with mortality but did not independently predict hospital death after adjustment based on the SCORTEN value. Dermal infiltrate severity was not associated with hospital death. ( J Am Acad Dermatol 10.1016/j.jaad.2011.10.007.) Key words: full-thickness necrosis; mortality; Stevens-Johnson syndrome; toxic epidermal necrolysis.
oxic epidermal necrolysis (TEN), StevensJohnson syndrome (SJS), and SJS/TEN overlap syndrome are rare and severe mucocutaneous diseases located along a continuous spectrum.
T
Drugs are the most common causes. The histologic hallmark of these diseases is necrosis with detachment of the epidermis.1-5 The overall annual incidence is 2/106 population.6 Patients are classified
From the Departments of Dermatology,a Clinical Research and Public Health,b and Pathology,d Assistance Publique - H^ opitaux de Paris, H^ opital Henri-Mondor, Universite Paris Est, LIC EA4393 (Drs Valeyrie-Allanore, Bastuji-Garin, and Wolkenstein), F-94010, Cr eteil; and the Dermatology Departments at H^ opital Tenonc and H^ opital Saint Louis, Paris, France.e The first two authors contributed equally to the study. Funding sources: None. Conflicts of interest: None declared. Study approved by ethics committee.
Presented in part at the Journees Dermatologiques de Paris in Paris, France in December 2006 and at the European Society of Dermatological Research in Paris, France in September 2006. Accepted for publication October 4, 2011. Reprint requests: Laurence Valeyrie-Allanore, MD, Service de Dermatologie, H^ opital Henri-Mondor, F-94010 Creteil, Cedex, France. E-mail: [email protected]. Published online November 15, 2011. 0190-9622/$36.00 Ó 2011 by the American Academy of Dermatology, Inc. doi:10.1016/j.jaad.2011.10.007
e1
J AM ACAD DERMATOL
e2 Valeyrie-Allanore et al
according to the percentage of the body surface area Histopathological examination exhibiting epidermal detachment (BSA%), which is Hematoxylin-eosinestained sections were rouless than 10% in SJS, more than 30% in TEN, and 10% tinely reviewed by two of us (L. V-A. and J. W.) to 30% in SJS/TEN overlap syndrome.7 Overall morworking together to achieve a consensus, without tality is about 22%.6 The definitive diagnosis relies on knowledge of the clinical data or outcome. A stanexamination of a skin biopsy specimen, which shows dardized form was used to collect data on adnexal epidermal necrosis and a very slight to absent infilnecrosis (including necrosis of the follicles, piloertrate in the papillary dermis ector muscles, sebaceous composed chiefly of lymphoglands, and eccrine sweat CAPSULE SUMMARY cytes and macrophages. glands); dermal mononuA severity-of-illness score clear infiltration graded In toxic epidermal necrolysis, dermal for SJS, TEN, and SJS/TEN, semiquantitatively as sparse, infiltrate density has been reported to called the SCORTEN, has moderate, or extensive predict death. been validated and con(Fig 1); and epidermal necrofirmed as a tool for predicting In 108 database patients, dermal sis assessed semiquantitasurvival.8-10 The SCORTEN is infiltrate density was associated neither tively as mild (confined to based on 7 clinical and labowith the day-1 SCORTEN value nor the deep epidermal layers), ratory risk factors and can hospital death. Full-thickness epidermal moderate (prominent but not range from 0 to 7, with higher necrosis was associated with death in confluent), or severe (fullscores predicting a higher the univariate analysis but did not thickness epidermal necrorisk of death. independently predict hospital death sis) (Fig 2). In addition to these 7 facafter adjustment of the day-1 SCORTEN tors, the severity of the dervalue. Clinical data mal mononuclear infiltrate as For each patient, we abDermal infiltrate density and severity of assessed histologically has stracted the following from epidermal necrosis should not be routinely been suggested to predict the database: BSA% involved taken into account when treating survival in patients with on day 1 (D1) (first 24 hours patients with toxic epidermal necrolysis. TEN.11 after admission), D1 The aim of this study was SCORTEN, final diagnosis to evaluate whether skin biopsy specimen findings (SJS, TEN, or SJS/TEN overlap based on greatest helped to predict survival in patients with SJS, TEN, BSA% involved), dates of disease onset and hospital or SJS/TEN independently from the SCORTEN value. admission, date of skin biopsy, and vital status at The patients were those in a database established discharge. The times from disease onset to admission prospectively for a previous study.9 We retrospecand to skin biopsy were computed. The day of tively reviewed the skin biopsy specimens to assess disease onset was defined as the first day with the severity of dermal mononuclear infiltration, epcutaneous or mucous membrane lesions that were idermal necrosis, and adnexal necrosis. not ascribable to other conditions and that were followed within 3 days by ulcers or blisters as defined elsewhere.3 Patients were classified accordMETHODS ing to vital status and D1 BSA% (\10%, 10%-30%, or Patients [30%). Disease progression during the hospital stay We identified 108 patients via a database that had was indirectly evaluated as a change in the abovebeen established prospectively to evaluate the described categories. SCORTEN measured during the first 5 hospital days.9 Inclusion and exclusion criteria have been Statistical analysis described elsewhere.7 Briefly, patients were eligible Qualitative data were described as number (perfor inclusion in the database if the discharge diagcent) and analyzed using the Fisher exact test. nosis was SJS, TEN, or SJS-TEN overlap and if they Quantitative data were described as mean (6SD) received symptomatic treatment only. Patients inor median (interquartile range) as appropriate and cluded in therapeutic trials and/or those who rewere analyzed using the nonparametric Kruskalceived potential curative treatment were excluded Wallis test. To assess whether dermal infiltrate severfrom the database.12-14 Of the 144 patients in the ity on D1 was associated with D1 BSA%, disease database, 108 had skin biopsy specimens available at progression, D1 SCORTEN, and/or hospital death, the pathology department. In the current study, we we used the Fisher exact test or nonparametric reviewed all 108 slides and looked for correlations Kruskal-Wallis test as appropriate. Similar analyses with the prognosis. d
d
d
J AM ACAD DERMATOL
Valeyrie-Allanore et al e3
Abbreviations used: BSA%: CI: D1: SJS: TEN:
percentage of body surface area exhibiting epidermal detachment confidence interval day 1 Stevens-Johnson syndrome toxic epidermal necrolysis
were performed for the extent of epidermal necrosis. To compare our results with those of Quinn et al,11 we also performed the same analyses in the subgroups of patients whose skin biopsy specimen was obtained within 48 hours after admission or whose discharge diagnosis was TEN. The SCORTEN has been validated as a severity-ofillness score that accurately predicts survival in patients with SJS, TEN, or SJS/TEN. Therefore, for histologic variables associated with hospital death (P \ .15) in the univariate analysis, we planned to conduct bivariate analyses adjusting for the SCORTEN value. Odds ratios were estimated with their 95% confidence intervals (CI). These analyses were restricted to patients who had a skin biopsy within 48 hours after admission. All tests were two-tailed. P values less than or equal to .05 were considered statistically significant. Data were analyzed using statistical software (Stata, Release 8.0, StataCorp, College Station, TX). This study was approved by an ethics committee on May 10, 2011.
RESULTS Study population Mean age of the 108 database patients with available skin biopsy specimens was 47.5 6 20.1 years (range, 13-92). Among these patients, 58 (53.7%) were male and 50 female. Median time from disease onset to admission was 4 days (3-6) and median time from disease onset to skin biopsy was 6 days (5-9). The skin biopsy was performed within 48 hours after admission in 99 (91.7%) patients. Mean D1 SCORTEN was 1.75 6 1.24 (Table I). Of the 100 patients whose D1 BSA% was no greater than 30%, 22 experienced enough disease progression to move to another category of body surface area involvement. The discharge diagnosis was SJS in 42 patients, SJS/TEN in 36, and TEN in 30. In all, 23 patients died, yielding a hospital mortality of 21.3% (95% CI 13.6-29.0). Histologic findings Table I reports the main histologic findings. The dermal infiltrate was sparse or moderate in 84
Fig 1. Dermal infiltrate in patients with Stevens-Johnson syndrome, toxic epidermal necrolysis, or overlap syndrome evaluated as sparse (A), moderate (B), or extensive (C). (A to C, Hematoxylin-eosin stain; original magnifications: 3100.)
(81.5%) patients. Full-thickness epidermal necrosis was observed in 56 (56%) patients. Neither the severity of the dermal infiltrate nor the severity of the epidermal necrosis was significantly influenced by the time from disease onset to skin biopsy (P = .7 and P = .42, respectively). Table II reports the associations between infiltrate severity and the clinical and outcome data in the univariate analysis. Dermal infiltrate severity showed a significant negative association with D1 BSA% (P = .04): D1 BSA% values lower than 10% were more common among patients having moderate or extensive infiltrates. Nevertheless, dermal infiltrate severity was not significantly associated with D1 SCORTEN, disease progression, or hospital death (Table II). Similar results were obtained in analyses restricted to the 93 patients whose skin biopsy was performed within 48 hours after admission or to the 29 patients with a discharge diagnosis of TEN (data not shown). Table III reports the associations between epidermal necrosis severity and the clinical and outcome data. Epidermal necrosis severity was significantly associated with disease progression (P = .002) and nonsignificantly associated with the D1 SCORTEN (P = .11) and hospital death (P = .06). The results were similar when the analysis was restricted to the 93 patients whose skin biopsy was performed within 48 hours after admission. Hospital mortality was
J AM ACAD DERMATOL
e4 Valeyrie-Allanore et al
Table I. Clinical and histologic characteristics of 108 study patients
Fig 2. Epidermal necrosis in patients with StevensJohnson syndrome, toxic epidermal necrolysis, or overlap syndrome evaluated as minor (A), moderate (B), or full thickness (C). (A to C, Hematoxylin-eosin stain; original magnifications: 340.)
significantly higher among patients with fullthickness epidermal necrosis compared with those with mild or moderate epidermal necrosis, both in the overall population (18/56, 32.1% vs 5/44, 11.4%; P = .017) and in the 93 patients with early biopsies (17/53, 32.1% vs 5/40, 12.5; P = .047). D1 SCORTEN was significantly higher (indicating greater severity) among patients with full-thickness epidermal necrosis compared with those with mild or moderate epidermal necrosis (2.0 6 1.3 vs 1.55 6 1.21; P = .04 and 2.0 6 1.3 vs 1.5 6 1.24; P = .02). In the bivariate analyses, the SCORTEN value remained significantly associated with hospital death (odds ratio 3.07, 95% CI 1.83-5.16) but full-thickness epidermal necrosis was not significantly associated with hospital death (odds ratio 2.02, 95% CI 0.65-7.12).
D1-BSA%, n (%) \10% 57 (52.8) 10%-30% 43 (39.8) [30% 8 (7.4) D1-SCORTEN, mean 6 SD 1.75 6 1.24 Discharge diagnosis, n (%) SJS 42 (38.9) SJS/TEN 36 (33.3) TEN 30 (27.8) Infiltrate severity (n = 103), n (%) Sparse 33 (32.0) Moderate 51 (49.5) Extensive 19 (18.5) Epidermal necrosis severity (n = 100), n (%) Minor 25 (25.0) Moderate 19 (19.0) Severe 56 (56.0) Adnexal necrosis, n (%) Follicles No 36 (34.0) Yes 27 (25.5) No follicles seen 43 (40.6) Eccrine sweat glands No 95 (89.6) Yes 3 (2.8) No glands seen 8 (7.6) Sebaceous glands No 28 (26.4) Yes 8 (7.6) No glands seen 70 (66.0) Piloerector muscle No 52 (49.1) Yes 16 (15.1) No muscle seen 38 (35.9) D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on first hospital day; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; SJS, Stevens-Johnson syndrome; SJS/TEN, StevensJohnson syndrome/toxic epidermal necrolysis overlap syndrome; TEN, toxic epidermal necrolysis.
In the subgroup of 29 patients with a discharge diagnosis of TEN, epidermal necrosis severity was associated neither with hospital death nor with the D1 SCORTEN. Finally, in patients with adnexal necrosis, neither eccrine sweat gland necrosis nor follicular necrosis was significantly associated with death (data not shown).
DISCUSSION In this population of 108 patients with SJS, TEN, or SJS/TEN, none of the histologic criteria studied was significantly associated with hospital mortality after
J AM ACAD DERMATOL
Valeyrie-Allanore et al e5
Table II. Dermal infiltrate severity according to percentage of body surface area with epidermal detachment on day 1, day-1 SCORTEN value, disease progression, and mortality Dermal infiltrate severity
Overall population (n = 103) D1-BSA%, n (%) \10% 10%-30% [30% D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Skin biopsy within 48 h (n = 95) D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Discharge diagnosis of TEN (n = 29) D1-SCORTEN, mean 6 SD Hospital death, n (%)
Sparse
Moderate
Extensive
(n = 33)
(n = 51)
(n = 19)
11 (33.3) 18 (54.6) 4 (12.1) 1.97 (1.21) 10 (30.3) 9 (27.3) (n = 28) 1.96 (1.26) 9 (32.1) 8 (28.6) (n = 13) 2.38 (1.12) 4 (30.8)
29 (56.9) 20 (39.2) 2 (3.9) 1.63 (1.3) 14 (27.5) 8 (15.7) (n = 50) 1.62 (1.31) 14 (28.0) 8 (16.0) (n = 10) 2.6 (1.58) 4 (40.0)
13 (68.4) 4 (21.1) 2 (10.5) 1.63 (1.16) 6 (31.6) 4 (21.0) (n = 17) 1.65 (1.22) 6 (35.3) 4 (23.5) (n = 6) 1.83 (0.75) 3 (50.0)
P value
.04* .33y .93* .44* .40y .83* .39* .61y .70*
D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on day of admission; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; TEN, toxic epidermal necrolysis. *P value by Fisher exact test. y P value by nonparametric Kruskal-Wallis test.
adjustment on the SCORTEN values determined on the day of admission. Overall mortality of SJS and TEN was recently reported to be approximately 22%.6 The BSA% involved at admission is one of the main prognostic factors and is among the 7 factors included in the SCORTEN severity-of-illness score.8,9 In addition to these 7 factors, dermal infiltrate severity was reported by Quinn et al11 to be associated with hospital death. In our population, we found no significant association between dermal infiltrate severity and hospital death. This discrepancy may be related to a difference in the timing of the skin biopsy. In the previous study,11 mean time from disease onset to skin biopsy was 9.4 to 10.2 days (depending on histologic grade). Thus, no further skin detachment was occurring at the time of skin biopsy in most of the patients. Furthermore, dermal infiltrate severity may have been influenced by events that occurred during the interval since disease onset, such as infection, local care, or specific immunosuppressive treatment. In our study, 92% of skin biopsies were performed within 48 hours after admission and median time from disease onset to skin biopsy was 6 days, and consequently potential effects of confounding factors were minimized. For a prognostic tool, a short time to result availability is crucial. Therefore, only early skin biopsy specimen results are relevant here. Quinn et al11 included only patients with TEN (BSA% [ 30). Mortality was 27%, 53%, and 71% in patients with sparse, moderate, and
severe dermal infiltrates, respectively, leading Quinn et al11 to suggest a role for immunocytes in the severity and prognosis of TEN. We included consecutive patients regardless of BSA%, and our overall mortality was only half that recorded by Quinn et al.11 However, dermal infiltrate severity was not associated with hospital mortality in our subgroup of patients with TEN (n = 29), regardless of time to biopsy. Epidermal necrosis severity was associated with hospital mortality in our study. Thus, the presence of full-thickness epidermal necrosis was significantly associated with worse D1 SCORTEN values and with higher hospital mortality, regardless of time to biopsy. Furthermore, full-thickness epidermal necrosis was associated with disease progression as indirectly assessed by a change in disease category. Epidermal necrosis severity may be directly related to the magnitude of the immunologic and chemokine storm that characterizes SJS and TEN. Greater epidermal necrosis on biopsy specimens, which are usually taken at the junction between involved and normal-appearing skin, may therefore indicate a higher risk of progression with delayed healing and subsequent complications such as local and systemic infections. However, given the retrospective study design, we had no control over the biopsy specimen sites. This point is among the limitations of our study. We also studied possible associations between adnexal necrosis and hospital mortality. Follicular
J AM ACAD DERMATOL
e6 Valeyrie-Allanore et al
Table III. Epidermal necrosis severity according to percentage of body surface area with epidermal detachment on day 1, SCORTEN value on day 1, disease progression, and mortality Epidermal necrosis
Overall population (n = 100) D1-BSA%, n (%) \10% 10%-30% [30% D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Skin biopsy within 48 h (n = 93) D1-SCORTEN, mean 6 SD Disease progression, n (%) Hospital death, n (%) Discharge diagnosis of TEN (n = 29) D1-SCORTEN, mean 6 SD Hospital death, n (%)
Minor
Moderate
Severe
(n = 25)
(n = 19)
(n = 56)
16 (64.0) 8 (32.0) 1 (4.0) 1.56 (1.16) 2 (8.0) 3 (12.0) (n = 23) 1.48 (0.71) 2 (8.7) 3 (13.0) (n = 2) 2.5 (0.71) 1 (50.0)
8 (42.1) 8 (42.1) 3 (15.8) 1.53 (1.31) 3 (18.8) 2 (10.5) (n = 17) 1.53 (1.37) 3 (17.7) 2 (11.8) (n = 6) 2.67 (1.86) 1 (16.7)
27 (48.2) 26 (46.4) 3 (5.4) 1.98 (1.29) 24 (42.9) 18 (32.1) (n = 53) 2.0 (1.3) 23 (43.4) 17 (32.1) (n = 21) 2.33 (1.15) 10 (47.6)
P value*
.34* .11y .002* .06* .07y .005* .095* .90y .46*
D1-BSA%, Percentage of total body surface area exhibiting epidermal detachment on first hospital day; D1-SCORTEN, disease-specific severity score (SCORTEN) measured on day of admission; TEN, toxic epidermal necrolysis. *P value by Fisher exact test. y P value by nonparametric Kruskal-Wallis test.
necrosis is a rarely described but common feature of SJS and TEN.15 In our study, 43% of patients had follicular necrosis, which was not significantly associated with hospital mortality. The specific sequence of molecular and cellular events leading to SJS or TEN has been partially elucidated. The lesions contain CD81 T lymphocytes that exhibit drug-specific, major histocompatibility complexerestricted cytotoxicity against keratinocytes. Immunopathological studies have demonstrated that the cytotoxic effects of these CD81 T cells are mediated by a variety of soluble mediators, not only perforin and granzyme B, but also granulysin and perhaps other compounds.16-18 The paucicellularity of the dermal infiltrate compared with other inflammatory diseases is difficult to understand but suggests a role for mechanisms such as indirect keratinocyte apoptosis via specific molecular contact. In sum, epidermal necrosis severity was significantly associated with hospital death in our study. However, the association was no longer significant after adjustment for the SCORTEN value on the day of admission. REFERENCES 1. Stevens AM, Johnson FC. A new eruptive fever associated with stomatitis and ophthalmia: report of two cases in children. Am J Dis Child 1922;24:526-33. 2. Valeyrie-Allanore L, Sassolas B, Roujeau JC. Drug-induced skin, nail and hair disorders. Drug Saf 2007;30:1011-30.
3. Roujeau J-C, Stern RS. Severe cutaneous adverse reactions to drugs. N Engl J Med 1994;331:1272-85. 4. Auquier-Dunant A, Mockenhaupt M, Naldi L, Correia O, Schroder W, Roujeau JC, SCAR Study Group. Correlations between clinical patterns and causes of erythema multiforme majus, Stevens-Johnson syndrome, and toxic epidermal necrolysis: results of an international prospective study. Arch Dermatol 2002;138:1019-24. 5. Roujeau JC, Kelly JP, Naldi L, Rzany B, Stern RS, Anderson T, et al. Medication use and the risk of Stevens-Johnson syndrome or toxic epidermal necrolysis. N Engl J Med 1995;333: 1600-7. 6. Mockenhaupt M, Viboud C, Dunant A, Halevy S, Bouwes Bavinck JN, Sidoroff A, et al. Stevens Johnson syndrome and toxic epidermal necrolysis: assessment of medication risks with emphasis on recently marketed drugs; the EuroSCAR study. J Invest Dermatol 2008;128:35-44. 7. Bastuji-Garin S, Rzany B, Stern RS, Shear NH, Naldi L, Roujeau JC. Clinical classification of cases of toxic epidermal necrolysis, Stevens-Johnson syndrome, and erythema multiforme. Arch Dermatol 1993;129:92-6. 8. Bastuji-Garin S, Fouchard N, Bertocchi M, Roujeau JC, Revuz J, Wolkenstein P. SCORTEN: a severity-of-illness score for toxic epidermal necrolysis. J Invest Dermatol 2000;115:149-53. 9. Guegan S, Bastuji-Garin S, Poszepczynska-Guigne E, Roujeau JC, Revuz J. Performance of the SCORTEN during the first five days of hospitalization to predict the prognosis of epidermal necrolysis. J Invest Dermatol 2006;126:272-6. 10. Cartotto R, Mayich M, Nickerson D, Gomez M. SCORTEN accurately predicts mortality among toxic epidermal necrolysis patients treated in a burn center. J Burn Care Res 2008;29: 141-6. 11. Quinn A, Brown K, Bonish B, Curry J, Gordon KB, Sinacore J, et al. Uncovering histologic criteria with prognostic significance in toxic epidermal necrolysis. Arch Dermatol 2005;141:683-7.
J AM ACAD DERMATOL
12. Wolkenstein P, Latarjet J, Roujeau JC, Duguet C, Boudeau S, Vaillant L, et al. Randomized comparison of thalidomide versus placebo in toxic epidermal necrolysis. Lancet 1998;352:1586-9. 13. Bachot N, Revuz J, Roujeau JC. Intravenous immunoglobulin treatment for Stevens-Johnson and toxic epidermal necrolysis. Arch Dermatol 2003;139:33-6. 14. Valeyrie-Allanore L, Wolkenstein P, Brochard L, Ortonne N, Ma^ıtre B, Revuz J, et al. Open trial of cyclosporin treatment for Stevens-Johnson syndrome and toxic epidermal necrolysis. Br J Dermatol 2010;163:847-53. 15. Braun-Falco O. Das Lyell syndrome. Bern (Switzerland): Haus Haker Verlog, 197061-70.
Valeyrie-Allanore et al e7
16. Nassif A, Bensussan A, Dorothee G, Mamai-Chouaib F, Bachot N, Bagot M, et al. Drug specific cytotoxic T-cells in the skin lesions of a patient with toxic epidermal necrolysis. J Invest Dermatol 2002;118:728-33. 17. Chung WH, Hung SI, Yang JY, Huang SP, Wei CY, Chin SW, et al. Granulysin is a key mediator for disseminated keratinocytes death in Stevens-Johnson syndrome and toxic epidermal necrolysis. Nat Med 2008;14:1343-50. 18. Viard I, Werhli P, Bullani R, Schneider P, Holler N, Salomon D, et al. Inhibition of toxic epidermal necrolysis by blockade of CD95 with human intravenous immunoglobulin. Science 1998; 282:490-3.