Julie Gueudry, Jean-Claude Roujeau, Michel Binaghi, Gisèle Soubrane, Marc Muraine. Risk Factors for the Development of Ocular Complications of Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis. Arch Dermatol. 2009;145(2):157–162. doi:10.1001/archdermatol.2009.540
To describe the acute and late ocular manifestations of toxic epidermal necrosis (TEN), Stevens-Johnson syndrome (SJS), and overlap syndrome and to identify predictors for the development of ocular complications.
Retrospective cohort study.
A single referral unit in a university hospital.
The study included 159 patients (mean [SD] age, 49.9 [19.8] years) with TEN and SJS during an 8-year period. Forty-nine patients were contacted at least 15 months after hospital discharge.
Main Outcome Measures
Records were reviewed for demographics, cause of the condition, and severity of ocular involvement. The patients were contacted to assess late ocular complications.
A total of 117 patients (74%) had acute ocular involvement, which was mild in 58%, moderate in 8%, and severe in 8%. Patients with TEN had more frequent (odds ratio [OR], 2.7; 95% confidence interval [CI], 1.06-6.90; P = .05) but not more severe (OR, 0.95; 95% CI, 0.20-4.5; P = .99) acute ocular involvement. Forty-nine patients were contacted at least 15 months after hospital discharge, and 63% had late ocular complications. Dry eye syndrome was the most common. The mean (SD) Ocular Surface Disease Index score was 32.9 (30.3) (range, 0-97.5). The severity of the acute ocular disease was found to be the only significant risk factor of late complications (P = .002), even though 5 patients without acute ocular involvement developed dry eye syndrome.
Ocular involvement is common in patients with SJS and TEN. Late complications are more frequent in patients with severe initial eye involvement but may also develop in patients without patent initial ocular symptoms.
Toxic epidermal necrolysis (TEN) and Stevens-Johnson syndrome (SJS) are severe, rare skin reactions, usually to drugs, and are associated with widespread destruction of the epidermis. The 2 diseases are closely related: SJS is a limited form characterized by mucous membrane erosions and blisters on limited areas of the skin (<10% of the total body surface area), and TEN resembles superficial burns because of the confluence of blisters and erosions on more than 30% of the total body surface area. Overlapping cases are defined by an intermediate extent of the skin lesions.1 The incidence of SJS and TEN is approximately 2 cases per million persons per year,2 and the overall mortality rate is 20% to 25%.3 Ocular complications in SJS and TEN may occur acutely along with or after the onset of skin involvement. Severe ocular complications may even result in permanent visual loss due to corneal scarring or vascularization.4- 6 Little is currently known about the ocular symptoms of these diseases, particularly the long-term manifestations. Stevens-Johnson syndrome and TEN are viewed as a spectrum of systemic disease with increasing severity and mortality. A scoring system estimates severity and predicts mortality in patients with SJS or TEN.7 However, it remains unclear whether there are factors predictive of the frequency and severity of ocular complications, both acutely and over the long term. This study aimed to describe acute and late ocular manifestations and complications in SJS and TEN and to identify possible clinical predictors for the development of ocular complications.
We retrospectively reviewed the medical records of all patients admitted to Henri Mondor Hospital, Créteil, France, between 1994 and 2002 for SJS, TEN, or overlap syndrome (SJS/TEN). The classification criteria of Bastuji-Garin et al1 were used to define SJS, overlap syndrome, and TEN. We collected demographic and clinical data on each patient, information about the pathogenesis of the SJS or TEN, the family of drugs involved, and treatments. Patients with clinical evidence of acute ocular complications were examined by an ophthalmologist during the initial hospitalization to determine the type, extent, and severity of the ocular involvement. Acute ocular involvement was defined as previously described by Power et al.8 Mild ocular involvement consisted of eyelid edema, and/or mild conjunctival injection, and/or chemosis only. Moderate involvement consisted of membranous conjunctivitis, and/or corneal epithelial defects, more than 30% healing with medical treatment, and/or corneal ulceration, and/or corneal infiltrates. Severe involvement consisted of symblepharon formation, and/or nonhealing corneal epithelial defects, and/or visual loss, and/or conjunctival fornix foreshortening. All surviving patients were contacted by mail or telephone and offered an ophthalmologic follow-up examination from January 11, 2004, to January 5, 2005.
Each patient underwent an ophthalmologic evaluation and completed the Ocular Surface Disease Index (OSDI) questionnaire to estimate his or her handicap. This score, which was designed to assess ocular irritation symptoms consistent with dry eye syndrome, as well as their impact on visual function, was chosen because in prior reports dry eye syndrome appeared as a frequent disability after SJS/TEN.9- 11 The OSDI contains 25 questions on the frequency of the symptoms and the difficulties of performing a specific task because of the symptoms. Responses were rated on a scale of 0 to 4, where 0 indicates none of the time, 1 indicates some of the time, 2 indicates half of the time, 3 indicates most of the time, and 4 indicates all of the time. The maximum score was 100, with higher scores indicating increasing symptom severity and vision-related disability. This research was designed and carried out in accordance with the Helsinki Declaration. Informed oral consent was obtained from each patient.
The data were analyzed using XLSTAT software (version 2007.6). The Fisher exact test was used to compare the severity and frequency of acute ocular involvement between SJS and TEN. Associations of late complications and categorical data were assessed using the Fisher exact test or the Cochran-Armitage trend test; the Mann-Whitney U test was performed for continuous data. P values of less than .05 were considered statistically significant.
There were 159 consecutive patients (76 males and 83 females) with SJS, TEN, or overlap syndrome during the study period. Their mean (SD) age was 49.9 (19.8) years (range, 11-91 years). Thirty deaths (19%) occurred during the acute phase of disease, 3 in patients diagnosed as having SJS, 14 in patients with overlap syndrome, and 13 in patients with TEN. Regarding a possible cause, antibacterial agents were the most commonly implicated group of drugs in this series (26%), with sulfamethoxazole-trimethoprim being the most common (10%). Antiepileptic drugs were associated with 14% of cases, especially carbamazepine (9%). Allopurinol was implicated in 4% of cases, and nonsteroidal anti-inflammatory drugs were implicated in 9% of cases.
The extent of acute ocular involvement is listed in Table 1. One hundred seventeen of 159 patients (74%) had ocular involvement; 35 patients (22%) had no ocular involvement; and the ocular involvement was unknown in 7 patients (4%). The acute ocular involvement was mild in 58% of the patients, moderate in 8%, and severe in 8%. The patients with TEN (odds ratio [OR], 2.7; 95% confidence interval [CI], 1.06-6.90; P = .05) or overlap syndrome (OR, 4.46; 95% CI, 1.72-11.61; P = .002) had a significantly higher frequency of acute ocular involvement than those with SJS. The patients with TEN (OR, 0.95; 95% CI, 0.20-4.5; P = .99) or overlap syndrome (OR, 0.79; 95% CI, 0.17-6.61; P = .37) did not have significantly more severe acute ocular involvement than patients with SJS.
Sixty-one of 129 surviving patients (47%) were successfully contacted during the study and completed the OSDI questionnaire. They had a mean (SD) OSDI score of 31.2 (30.0) and a median score of 22.7 (range, 0-97.5). Ophthalmologic examination data were available for 49 of these patients (38%). The mean duration between hospital discharge and most recent ophthalmologic examination was 72 months (range, 15-131 months). The mean duration between hospital discharge and completion of the OSDI questionnaire was 82 months (range, 25-130 months). Dry eye syndrome and subconjunctival fibrous scarring were the 2 most common late complications. Four patients lost vision owing to SJS/TEN ocular complications: 3 developed bilateral conjunctivalization of the cornea, and 1 developed a severe keratoconjunctivitis sicca in 1 eye, leading to corneal scarring. The details of late ocular complications are given in Table 2. No patient developed ankyloblepharon. Table 3 shows the demographic data, disease groups, severity of acute ocular involvement, and general treatments used during the acute presentation of the systemic disease in 49 patients contacted at least 15 months after discharge from the hospital. The only factor that appeared predictive of late ocular manifestations was an initial greater severity of eye involvement, but it is noteworthy that 5 patients (aged 29, 35, 41, 46, and 77 years) without initial ocular involvement also developed dry eye syndrome. No patient received systemic steroids during acute presentation of the systemic disease. High-dose intravenous immunoglobulin therapy was administered to 5 patients. We recorded the presence or absence of ocular sequelae in the patients as a function of the local ocular treatment received during initial hospitalization in a dermatology unit (Table 4). Precise data concerning local ocular treatment were collected for 38 patients. Only 1 patient received local corticosteroid treatment (dexamethasone and tobramycin eye drops [1 drop 4 times a day]), for active ocular toxoplasmosis. Ocular treatment during the acute presentation included antiseptic eye drops and lubricant eye drops containing preservatives, which were used in 89% and 63% of patients, respectively. The late ocular complication rates for patients treated with lubricant eye drops containing preservatives were not higher than those for patients not treated with eye drops of this type. The 49 patients had a mean OSDI score of 32.9 (30.3) and a median score of 25.0 (range, 0-97.5). Nineteen of these 49 patients (39%) used lubricant eye drops at the time the OSDI questionnaire was administered. Table 5 presents the OSDI scores and characteristics of these patients. The symptom identified as the most troublesome varied among the patients. In all cases, the most troublesome symptoms involved irritation rather than vision loss. Twenty-seven patients (55%) found it difficult to read, to drive at night, or to use a computer for at least half the time. The mean OSDI score was 42.1 (31.7), and the median OSDI score was 41.6 (range, 0-97.5) for the 31 patients who presented with ocular sequelae.
One of the main limitations in assessing long-term outcomes of patients with SJS and TEN is the rarity of the disease and the relatively small number of patients treated at any 1 center each year. There have therefore been few studies focusing on the long-term ophthalmologic sequelae of surviving patients.12- 16 Just as no specific treatment for the general disease has yet been shown to be effective, there is also no specific treatment for preventing ocular damage and complications. Current management strategies based on the use of systemic and local corticosteroids seems unable to prevent late ocular complications.8,9 Intravenous immunoglobulin does not limit ocular damage.17 Amniotic membrane transplantation, which is known to have anti-inflammatory effects and to favor wound healing, is already used to treat ocular burns.18- 20 Encouraging preliminary results have been obtained for the use of this technique in isolated cases of ocular complications of SJS/TEN.21- 23 We therefore tried to identify factors that were predictive of subsequent ocular complications. In our study, only the initial severity of ocular complications constituted a risk factor for the development of ocular sequelae and greater functional impairment during the initial cutaneous disease, as described by Arstikaitis,10 but this risk was not subsequently confirmed. Indeed, Yip et al15 recently reported that the severity of initial ocular damage was not a risk factor for ocular sequelae, after statistical adjustment, whereas local antibiotic use was. In their study,15 antibiotics may have been used in patients who presented with ocular problems that were considered to be more severe. In the virtual absence of local antibiotic use, the frequency of ocular sequelae was similar to that in our study. We did not observe a deleterious effect of any local treatment in our patients, but the numbers were too low to exclude this hypothesis. The use of preservative-free products was suggested to be essential,15 because preservatives in eye drops are known to cause ocular surface damage,24,25 and iatrogenic effects of nonsteroidal anti-inflammatory eye drops have been described in patients with these conditions.26 Nowadays, preservative-free lubricant eye drops are used for each patient in our department. Isotonic sodium chloride solution (normal saline) is not used as irrigant. We identified no general risk factor, and there was no significant difference in the frequency of long-term ocular complications between patients with SJS, overlap syndrome, and TEN (P = .25). This finding may appear surprising, as detachment of the skin is a major prognostic factor in terms of general complications and death, which, in turn, implies that all patients in the acute phase of the disease must display the same ophthalmologic vigilance. Studies of other initial clinical factors of severity, such as damage to the mucous membranes or the severity scores for TEN (SCORTEN), have not shown these factors to have predictive value for ocular prognosis.15
Ocular involvement was observed during the acute phase of the disease in 74% of the patients (64% of patients with SJS, 88% of patients with overlap syndrome, and 84% of patients with TEN). The incidence of ocular involvement during the acute phase has been reported to lie between 69% and 82% for SJS and between 50% and 88% for TEN.8,16,27- 31 We have shown that damage at the acute stage is more frequent in patients with disease affecting more than 10% of the total body skin area. The presence of more diffuse mucous and cutaneous damage may be associated with a higher risk of damage to the ocular mucous membrane. Arstikaitis10 reported similar findings, but these findings were not confirmed by subsequent studies.15 This difference in results may be attributable to differences in the definitions of SJS, overlap syndrome, and TEN.32 However, there was no difference in the severity of acute ocular complications between the SJS group and the TEN group, as previously described.15,28
The most frequent late ocular complication was dry eye syndrome, which affected 59% of the patients, consistent with the findings of previous studies.16,27,33 Binaghi et al,9 Magina et al,13 and Yip et al15 reported a frequency of dry eye syndrome of 42% in 26 patients with a mean follow-up period of 3 years, 50% in 8 patients with a mean follow-up period of 3 years, and 46% in 44 patients with a mean follow-up period of 35.5 months, respectively. The frequency observed in our study is higher, probably because our patients were asked to attend consultations and were reexamined specifically to check for ocular sequelae, leading to a recruitment bias. Furthermore, we counted all cases of dry eye syndrome, whereas Yip and colleagues15 counted only cases of “severe dry eye syndrome.”
Adult survivors of TEN had a substantially lower overall health-related quality of life than the normal population.14 Ocular complications greatly impaired their overall quality of life.14,34 It is difficult to evaluate the degree of handicap due to ocular sequelae in these patients. Indeed, description of the objective signs, ie, changes in visual acuity, or the biomicroscopic signs of damage to the ocular surface alone does not adequately reflect the functional impairment that was caused. In our study, 27 of the patients with late ocular complications (87%) found it difficult to read, to drive at night, and to use a computer all day or even half a day. These difficulties had a major impact on their professional activities. Given the high frequency of dry eye syndrome, we decided to evaluate patients with the OSDI, which was originally developed and validated for dry eye syndrome.11,35 This scale gives a score, making statistical comparisons possible. The mean OSDI score for the 31 patients with ocular sequelae was 42.1, corresponding to severe functional impairment. For comparison, the mean score of patients with Goujerot-Sjögren syndrome was 30.35
Our study has several weaknesses. During the acute phase, the ophthalmologist screened only patients with suspected ocular involvement identified by the treating physician, which may have resulted in an underestimation of ocular involvement. The comparison of severity of late ocular complications between patients with SJS or TEN is probably biased, because the mortality rate during the initial phase of the disease is higher for TEN than for SJS.
The number of patients who were unavailable for follow-up constitutes a bias. It was not possible to contact these patients, despite an extremely thorough search by telephone, through other family members, or through the patient's physician. It is not possible to reach a conclusion regarding the ophthalmologic state of these patients. The fact that the mean OSDI score of the subset of patients who filled out the OSDI questionnaire but were not examined was very close to the mean score of the examined patients suggests that there was an absence of major selection bias linked to this subgroup. To our knowledge, this study involves the largest series of patients ever described in Europe. The series that was most similar to ours in terms of size, which was carried out on an Asian population, had a similar number of patients who were unavailable for follow-up.
The OSDI represents a specific measurement for discomfort due to dry eye syndrome. It does not therefore reflect all the repercussions of SJS or TEN on quality of life, as patients may have other problems in addition to those concerning the ocular surface. Furthermore, symptom scores based on the OSDI may not necessarily be appropriate for all cases of SJS/TEN, as severe cases with keratinization tend to present with fewer symptoms than might be expected. Nonetheless, the OSDI provides an indication of the patient's disability, as dry eye syndrome is the most frequently reported sequela. Furthermore, it has been used to assess improvement in patients with ocular surface problems in cases of chronic graft-vs-host disease after allogenic hematopoietic cell transplantation.36
Furthermore, ocular disease progression may occur at variable periods after the acute disease episode.37 Patients should also be offered an ophthalmologic follow-up after being discharged from initial hospitalization. Actually, of 31 patients with long-term ocular disorder, 5 had no evidence of eye lesions during the acute phase of the disease. These 5 patients developed dry eye syndrome. Haber et al14 reported a similar case. Although we cannot exclude the coincidental occurrence of idiopathic dry eye syndrome unrelated to SJS or TEN in some of these cases, we consider the coincidence unlikely for all 5 patients.
Therefore, we conclude that all patients with SJS or TEN should undergo initial ophthalmologic screening and ophthalmologic follow-up during the acute phase of the disease. The use of local nonsteroidal anti-inflammatory agents and eye drops containing preservatives should be avoided. Further knowledge of risk factors should make it possible to improve initial management. In the future, more aggressive ophthalmologic treatment protocols may be offered to patients with severe ocular complication to prevent the occurrence of sequelae. Amniotic membrane transplantation should be evaluated by controlled studies, and new immunosuppressive or immunomodulatory molecules for administration via the general or local route should be tested for efficacy in preventing the ocular complications of SJS and TEN. We thus strongly suggest that prospective follow-up is necessary, probably for at least 1 year.
Correspondence: Julie Gueudry, MD, Department of Ophthalmology, Hôpital Charles Nicolle, 1 rue de Germont, 76031 Rouen, France (firstname.lastname@example.org).
Accepted for Publication: May 21, 2008.
Author Contributions: Dr Gueudry had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Gueudry, Roujeau, Binaghi, Soubrane, and Muraine. Acquisition of data: Gueudry, Roujeau, and Binaghi. Analysis and interpretation of data: Gueudry, Roujeau, and Muraine. Drafting ofthe manuscript: Gueudry. Critical revision of the manuscript for important intellectual content: Roujeau, Binaghi, Soubrane, and Muraine. Statistical analysis: Gueudry. Administrative, technical, and material support: Roujeau, Binaghi, and Soubrane. Study supervision: Roujeau and Muraine.
Financial Disclosure: None reported.
Previous Presentation: This study was presented in part as a poster and won the THEA award at the European Association for Vision and Eye Research Meeting; October 5-8, 2005; Vilamoura, Portugal.