Assessment of Treatment Approaches and Outcomes in Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis: Insights From a Pan-European Multicenter Study

Key Points

Question  What are the treatment approaches for Stevens-Johnson syndrome and toxic epidermal necrolysis across Europe and their association with the disease course in the acute phase, as well as prognostic factors and culprit drugs?

Findings  In this cohort study, among 212 patients, treatment approaches ranged from best supportive care only to systemic glucocorticoids, intravenous immunoglobulins, cyclosporine, and antitumor necrosis factor agents; the 6-week mortality rate was 20.8% and antibiotics were the most frequent culprit drugs.

Meaning  The findings of this study suggest differences in treatment strategies for Stevens-Johnson syndrome and toxic epidermal necrolysis in Europe and highlight the importance of prospective therapeutic studies and registries.

Importance  Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are severe drug reactions associated with a high rate of mortality and morbidity. There is no consensus on the treatment strategy.

Objective  To explore treatment approaches across Europe and outcomes associated with the SJS/TEN disease course, as well as risk factors and culprit drugs.

Design, Setting, and Participants  A retrospective pan-European multicenter cohort study including 13 referral centers belonging to the ToxiTEN ERN-skin subgroup was conducted. A total of 212 adults with SJS/TEN were included between January 1, 2015, and December 31, 2019, and data were collected from a follow-up period of 6 weeks.

Main Outcomes and Measures  Risk factors for severe acute-phase complications (acute kidney failure, septicemia, and need for mechanical ventilation) and mortality 6 weeks following admission were evaluated using a multivariable-adjusted logistic regression model. One tool used in evaluation of severity was the Score of Toxic Epidermal Necrolysis (SCORTEN), which ranges from 0 to 7, with 7 the highest level of severity.

Results  Of 212 patients (134 of 211 [63.7%] women; mean [SD] age, 51.0 [19.3] years), the mean (SD) body surface area detachment was 27% (32.8%). In 176 (83.0%) patients, a culprit drug was identified. Antibiotics (21.2%), followed by anticonvulsants (18.9%), nonsteroidal anti-inflammatory drugs (11.8%), allopurinol (11.3%), and sulfonamides (10.4%), were the most common suspected agents. Treatment approaches ranged from best supportive care only (38.2%) to systemic glucocorticoids (35.4%), intravenous immunoglobulins (23.6%), cyclosporine (10.4%), and antitumor necrosis factor agents (3.3%). Most patients (63.7%) developed severe acute-phase complications. The 6-week mortality rate was 20.8%. Maximal body surface area detachment (≥30%) was found to be independently associated with severe acute-phase complications (fully adjusted odds ratio [OR], 2.49; 95% CI, 1.21-5.12; P = .01) and SCORTEN greater than or equal to 2 was significantly associated with mortality (fully adjusted OR, 10.30; 95% CI, 3.82-27.78; P < .001). Cyclosporine was associated with a higher frequency of greater than or equal to 20% increase in body surface area detachment in the acute phase (adjusted OR, 3.44; 95% CI, 1.12-10.52; P = .03) and an increased risk of infections (adjusted OR, 7.16; 95% CI, 1.52-33.74; P = .01). Systemic glucocorticoids and intravenous immunoglobulins were associated with a decreased risk of infections (adjusted OR, 0.40; 95% CI, 0.18-0.88; P = .02). No significant difference in 6-week mortality was found between treatment groups.

Conclusions and Relevance  This cohort study noted differences in treatment strategies for SJS/TEN in Europe; the findings suggest the need for prospective therapeutic studies to be conducted and registries to be developed.

Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN), or Lyell syndrome, are rare, delayed-type hypersensitivity reactions associated with high levels of morbidity and mortality.^1,2 The clinical hallmark of SJS and TEN is epidermal and mucosal detachment and necrosis. The 2 entities are defined by the body surface area (BSA) of detachable or detached skin: less than 10% of the BSA affected in SJS, 10% to 30% in overlap syndrome, and more than 30% in TEN.^1,3,4 Disease prognosis is determined by various clinical and biological factors. Some of these parameters, such as patient age and more than 10% BSA involvement, have been integrated in risk prediction models, such as the Score of Toxic Epidermal Necrolysis (SCORTEN)⁵ and, more recently, the ABCD-10.⁶ SCORTEN, which was used in our analysis, includes 7 clinical and biological parameters collected at baseline and is a predictive score for acute-phase mortality. The severity levels range from 0 to 7, with 7 the highest level.

In SJS/TEN treatment, the first and most important measure is the immediate discontinuation of the culprit drug.⁷ In addition, best supportive care has been recognized as a cornerstone of SJS/TEN treatment.^8-10 In contrast, no consensus exists on the best adjuvant treatment. The rarity and severity of SJS/TEN have been obstacles to performing controlled clinical trials. Current clinical practices, including the use of systemic agents, such as cyclosporine, systemic glucocorticoids, intravenous immunoglobulins (IVIGs), and antitumor necrosis factor (TNF) agents, are based only on retrospective case reports of series and empirical experience. The association of these agents with the outcome, however, has not been proven in prospective trials.^11,12

In March 2017, the European Union created the European Reference Network for rare skin diseases (ERN-skin) to offer European Union citizens with rare skin diseases the benefits of specialized care anywhere in Europe. ERN-skin includes 56 health care settings (ie, reference centers) located in 18 European countries. ToxiTEN is the subgroup of ERN-skin dedicated to SJS/TEN. In this multicenter, retrospective cohort study, we aimed to explore and compare the clinical profile, culprit drugs, prognostic factors, and treatment approaches for SJS/TEN across members of the ToxiTEN group.

The present study was designed as a pan-European multicenter, longitudinal retrospective cohort study, performed by the ToxiTEN group (eFigure in the Supplement). Patients were recruited from 13 health care settings across 10 European countries: Czech Republic, Denmark, Finland, France, Germany, Italy, Lithuania, Romania, Switzerland, and the UK (eTable 1 in the Supplement). All participating institutes were reference centers with expertise in the management of SJS/TEN. All adult (age ≥18 years) patients diagnosed with SJS/TEN between January 1, 2015, and December 31, 2019, in these centers were included in the study. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. The study was conducted according to the ethical guidelines at the respective institutions and ethical approval was obtained by all participating centers in accordance with the Declaration of Helsinki.¹³ The participants provided informed consent and the data were deidentified.

The diagnosis of SJS/TEN was established by expert consultants specialized in inpatient dermatology on the basis of compatible clinical and histologic features.^1,2 Before enrollment in the study, all cases of SJS/TEN were validated by the principal investigator at each site through comprehensive medical records review, based on predefined clinical and histologic criteria (eg, presence of epidermal detachment, involvement of >2 mucosae, atypical target lesions, histologic evidence of epidermal necrosis, and exclusion of differential diagnoses).

Medical records were manually reviewed for each participant. The following variables, which were routinely collected during the patients’ hospital stays, were systematically retrieved and entered in a standardized form: demographic characteristics, SCORTEN at admission,⁵ initial and maximal BSA extent of epidermal detachment/detachable skin, affected mucosal surfaces, comorbidities, culprit drugs or other putative predisposing factors, hospitalization features, and the therapeutic regimens used.

Eligible patients were followed up at regular intervals for 6 weeks after the initial presentation to estimate survival status. In addition, study participants were longitudinally evaluated for the occurrence of severe acute complications. The latter variables were defined as the development of at least 1 of the following conditions during the disease course: (1) septicemia, (2) pulmonary infection, (3) acute kidney injury, and (4) respiratory distress necessitating mechanical ventilation. Subsequently, data on this longitudinal follow-up were retrospectively collected by all participating centers.

Baseline characteristics were described as means (SDs) for continuous variables and percentages for categorical values. The comparison between different variables was performed using the χ² test for categorical variables and t test for continuous variables. The fully adjusted model included all variables that were found significant in the univariate analysis.

The independent associations between demographic, disease-specific, and comorbidity variables with the risk of mortality and severe complications were evaluated using multivariable-adjusted logistic regression and are reported as odds ratios (ORs) and 95% CIs. Entry and removal limits were set at P < .05 in the univariate analysis. The covariates included in the regression model were treated as dichotomous and encompassed, among others, age, with a cutoff set at 40 years in accordance with age limit in SCORTEN, and a SCORTEN of 2 (the median value of this severity score in our cohort). In the analysis evaluating outcome measures in the frequent treatment regimens (systemic glucocorticoids, IVIG, and cyclosporine), patients receiving supportive therapy served as a reference group. Only patients treated with a single agent were included, and outcome measures were adjusted for age and SCORTEN. Findings were considered significant at unpaired and 2-sided P < .05. Statistical analyses were performed using SPSS, version 25 (IBM Corp).

The study population comprised 212 patients; with data on sex available on 211 individuals, 77 (36.3%) were men and 134 (63.7%) were women. The mean (SD) age at diagnosis of SJS/TEN was 51.0 (19.3) years, and the median age was 49.2 years (range, 18.1-94.7 years). Most patients were of White race (142 [67.0%]), whereas 38 (17.9%) were of African ancestry and 25 (11.8%) were of Asian ancestry. Table 1 further delineates the demographic features of study participants.

The mean (SD) SCORTEN was estimated at 2.2 (1.5); the mean (SD) percentage of BSA epidermal detachment/detachable skin at presentation was 27.0% (23.8%) and, at the maximal activity of the disease, was 39.5% (29.2%). The oral mucosa was the most frequently affected mucosal surface (184 of 207 [88.9%]), followed by the ocular (138 of 207 [66.7%]), genital (129 of 207 [62.3%]), laryngeal/pharyngeal (75 of 207 [36.2%]), and anal (44 of 203 [21.7%]) mucosae. The mean (SD) number of involved mucosal surfaces was 2.8 (1.3), and 66 (31.1%) patients had 4 or more affected mucosae.

A culprit drug was identified in 176 (83.0%) patients. Nonsulfonamide antibiotics (21.2%), anticonvulsants (18.9%), nonsteroidal anti-inflammatory drugs (11.8%), allopurinol (11.3%), and sulfonamides (including sulfonamide antibiotics, 10.4%) were the most common triggering agents (Figure, A). Regarding antibiotics, β-lactams (44.4%), fluoroquinolones (35.6%), and vancomycin (11.1%) were the most common classes (Figure, B). Among patients with anticonvulsant-induced SJS/TEN, exposure to aromatic anticonvulsants (eg, carbamazepine, phenytoin, and phenobarbital), was documented in 87.5% of the patients (Figure, C).

In 36 patients (17.0%), a detailed medical history did not reveal exposure to potential culprit medications. In these patients, the development of SJS/TEN was attributed to Mycoplasma pneumoniae infection (4 [11.1%]), other infectious agents (6 [16.7%]), and autoimmune comorbidities (5 [13.9%]). No trigger was found in the remaining 21 (58.3%) patients who had no culprit drug identified.

Our analysis showed that treatment for most patients was managed in dermatology inpatient wards (144 [69.6%]), whereas treatment for 100 patients (48.3%) was managed in intensive care units and, for 40 patients (19.3%), in burn units; 72 patients (34.0%) received treatment in both dermatology and intensive care units during the acute phase. One hundred thirty-two patients (62.3%) were referred from other health care facilities, and 80 individuals (37.7%) were directly admitted to the participating centers (eTable 2 in the Supplement). The mean (SD) duration of inpatient hospital stays was 25.0 (28.3) days.

Treatment for 81 (38.2%) patients was managed with supportive treatment only. Systemic glucocorticoids were the most frequently used therapy, being administered to 75 patients (35.4%). Other agents used included IVIG (50 [23.6%]), cyclosporine (22 [10.4%]), and antitumor necrosis factor (7 [3.3%]) (Figure, D; eTable 3 in the Supplement).

Forty-four patients (20.8%) died during the initial 6 weeks of the study. Table 2 reports univariate and multivariable analyses to identify risk factors for mortality within the first 6 weeks. SCORTEN greater than or equal to 2 was significantly associated with susceptibility to mortality (fully adjusted OR, 10.30; 95% CI, 3.82-27.78; P < .001). Other demographic variables, comorbid conditions, disease-specific features, and treatment modalities were not independently associated with the risk of mortality (Table 2).

We assessed the frequency and risk factors of severe acute-phase complications. A total of 135 patients (63.7%) experienced severe complications; septicemia occurred in 59 patients (27.8%) and pulmonary infections occurred in 58 patients (27.3%). Acute kidney injury developed in 43 patients (20.3%), and 67 patients (31.6%) required mechanical ventilation (eTable 2 in the Supplement).

Table 3 presents the risk factors for severe acute-phase complications during the course of SJS/TEN. Maximal BSA detachment greater than or equal to 30% was found to be independently associated with the occurrence of severe acute-phase complications (fully adjusted OR, 2.49; 95% CI, 1.21-5.12; P = .01). Although ocular involvement (age- and severity-adjusted OR, 2.25; 95% CI, 1.17-4.32; P = .02) and cyclosporine treatment (age- and severity-adjusted OR, 4.17; 95% CI, 1.17-14.89; P = .03) were associated with complications after adjusting for age and severity, they fell short of significance in the fully adjusted model.

Table 4 delineates SJS/TEN-related outcomes and complications among patients who received monotherapy with systemic glucocorticoids (n = 52), IVIG (n = 28), or cyclosporine (n = 20) compared with those treated supportively (n = 81). Following the adjustment for age and SCORTEN, systemic glucocorticoids were associated with a lower risk of extensive disease (defined as maximal BSA detachment ≥30%; adjusted OR, 0.16; 95% CI, 0.05-0.51; P = .002), and cyclosporine was associated with a greater than or equal to 20% increase in BSA detachment during the disease course (adjusted OR, 3.44; 95% CI, 1.12-10.52; P = .03).

Cyclosporine use was associated with an increased risk for any type of infection (adjusted OR, 7.16; 95% CI, 1.52-33.74; P = .01). Systemic glucocorticoids were associated with a reduced risk of any infection (adjusted OR, 0.40; 95% CI, 0.18-0.88; P = .02) and mechanical ventilation (adjusted OR, 0.34; 95% CI, 0.12-0.96; P = .05). Treatment with IVIG was associated with a decreased risk of septicemia (adjusted OR, 0.17; 95% CI, 0.05-0.63; P = .008). The baseline characteristics of patients in the different treatment groups are outlined in eTable 4 in the Supplement.

The aim of this multicenter retrospective study was to explore the profile of patients, culprit drugs, treatment approaches, and prognostic factors of SJS/TEN across expert centers within European countries. In line with previous data,¹ among the 212 adults with SJS/TEN included in our cohort, no culprit drug could be identified in approximately 20% of the patients with SJS/TEN. Antibiotics (not including sulfonamides), including β-lactams, were the most common identified culprit drugs in these patients, followed by allopurinol and anticonvulsants. Although these are the overall most common causative agents of SJS/TEN, our results differ from other studies. A study of a cohort of 187 patients from South Korea identified allopurinol as the most common culprit drug, followed by carbamazepine and lamotrigine.¹⁴ In a multicenter retrospective study from the US, sulfamethoxazole with trimethoprim was identified as the most common culprit drug.¹⁵ In the EuroSCAR study, antibiotics (except cotrimoxazole) were considered as lower risk drugs.¹⁶ The divergences between these observations can, at least partially, be associated with the genetic background, ie, differences in risk human leukocyte antigen allele expression, which varies among races/ethnicities¹: most of our patients were of White race. Moreover, regional differences in prescription practices, such as the use of cotrimoxazole for urinary tract infections or soft tissue infections, could further contribute to this observation.¹⁷

There is an overall consensus on the importance of best supportive care, which is recognized in various national guidelines as the cornerstone of SJS/TEN treatment to reduce mortality.^9,10,18-20 In contrast, the efficacy of adjuvant therapies for SJS/TEN is still debated. To our knowledge, there are neither guidelines nor comparative studies on the use of adjuvant agents in SJS/TEN, and their use mostly relies on case series, case reports, and clinical experience. Our study, thus, provides a comparative report on the wide range of these adjuvant therapies and our results mirror the center-dependent diversity of therapeutic approaches in SJS/TEN, ranging from IVIG,^21,22 systemic glucocorticoids,²³ antitumor necrosis factor agents,^24-26 and cyclosporine^27-29 to supportive care only. Although our study was not designed or able to compare the efficacy of different treatment approaches, it yielded some interesting results: cyclosporine was associated with an increase of BSA affected and an overall higher risk of infections, and IVIG was associated with a decreased risk of sepsis. Systemic glucocorticoids were associated with a lower rate of patients with a maximal affected BSA greater than or equal to 30% over the disease course. Although our results are not sufficient for conclusions to be made as to the superiority of any adjuvant treatment, they suggest the need to investigate the associations of different treatments not only with mortality, but also with the occurrence of life-threatening complications and epidermal healing during the acute phase. Given that our study only covered the acute-phase observation period, further studies are warranted to explore whether patients may exhibit different clinical outcomes, including long-term complications and mortality.

The mortality rate of 20.8% at 6 weeks in our study was similar to some studies,^30,31 but higher than in other studies.^15,19 Our investigation suggests that the extent of BSA detachment and respiratory failure were risk factors for both mortality and life-threatening complications during the acute phase.^5,32-34

This study has limitations. First, we did not include children because triggers, prognosis, and management of SJS/TEN in children differ considerably from those in adults. Furthermore, childhood SJS/TEN may be misdiagnosed as erythema multiforme.³⁵ Second, although our multicenter study included various countries and centers, with substantially differing numbers of patients included per center, the selection of the sites may have biased our findings. Third, owing to the retrospective data collection, caution is warranted when drawing conclusions regarding treatment strategies. Prospective clinical trials are needed to evaluate the efficacy of the different adjuvant treatments in SJS/TEN.

This multicenter European retrospective study on 212 adults with SJS/TEN provides an overview on patients’ profiles, drug triggers, prognosis, and treatment regimens within the ERN-skin network. This study describes the current diversity of treatment approaches across centers and countries. Future studies and registries, such as IRTEN (the international registry of TEN),² may aim to prospectively collect comprehensive clinical and biological information on more patients with SJS/TEN to ultimately improve the treatment.

Accepted for Publication: July 5, 2021.

Published Online: August 25, 2021. doi:10.1001/jamadermatol.2021.3154

Corresponding Author: Saskia Ingen-Housz-Oro, MD, Dermatology Department, AP-HP, Henri Mondor Hospital, 51 avenue du Maréchal de Lattre de Tassigny, 94000 Créteil, France (saskia.oro@aphp.fr).

Author Contributions: Drs Kridin and Ingen-Housz-Oro had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Kridin and Brüggen contributed equally, and Drs Vorobyev and Ingen-Housz-Oro contributed equally.

Concept and design: Kridin, Brüggen, Walsh, Nägeli, Tétart, Didona, Březinová, Tiplica, de Prost, Vorobyev, Ingen-Housz-Oro.

Acquisition, analysis, or interpretation of data: Kridin, Chua, Bygum, Walsh, Nägeli, Kucinskiene, French, Milpied, Ranki, Salavastru, Březinová, Divani-Patel, Lorentzen, Loft Nagel, Valiukeviciene, Karpavičiūtė, Tiplica, Oppel, Oschmann, Vorobyev, Ingen-Housz-Oro.

Drafting of the manuscript: Kridin, Brüggen, Ranki, Salavastru, Divani-Patel, Karpavičiūtė, Tiplica, Vorobyev, Ingen-Housz-Oro.

Critical revision of the manuscript for important intellectual content: Kridin, Brüggen, Chua, Bygum, Walsh, Nägeli, Kucinskiene, French, Tétart, Didona, Milpied, Ranki, Salavastru, Březinová, Lorentzen, Loft Nagel, Valiukeviciene, Oppel, Oschmann, de Prost, Vorobyev.

Statistical analysis: Kridin, Divani-Patel, Karpavičiūtė, Vorobyev.

Obtained funding: Kridin.

Administrative, technical, or material support: Kridin, Brüggen, Bygum, Walsh, Nägeli, Kucinskiene, Ranki, Divani-Patel, Oppel, Oschmann, Vorobyev.

Supervision: Kridin, Brüggen, Bygum, Walsh, Nägeli, French, Didona, Březinová, Valiukeviciene, de Prost.

Conflict of Interest Disclosures: Dr Walsh reported receiving nonfinancial support from UCB Pharma for sponsored attendance at meeting outside the submitted work. Dr Salavastru reported receiving nonfinancial support and personal fees from Leo Pharma, and nonfinancial support from AbbVie outside the submitted work; in addition, Dr Salavastru had a patent for Springer Nature Switzerland AG with royalties paid. Dr Tiplica reported receiving personal fees from Antibiotice, nonfinancial support from A&D Pharma Marketing & Sales Services, nonfinancial support from EGIS Pharmaceutical PLC, and nonfinancial support from Sanofi Romania SRL outside the submitted work. No other disclosures were reported.