Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis After Vaccination: Reports to the Vaccine Adverse Event Reporting System

We conducted a telephone survey of reports of Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) to the Vaccine Adverse Event Reporting System. We identified six cases of SJS or TEN after vaccination without other obvious triggers, suggesting that SJS and TEN might very rarely be caused by vaccination. Confirmation of this hypothesis will likely require controlled studies. Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are severe acute skin disorders characterized by macular rashes, often with atypical target lesions, involvement of more than one mucosal site and in TEN coalescence of the rash to widespread erythema, necrosis and bullous detachment of the epidermis. 1 The annual incidence of SJS and TEN is about one to three cases per million persons. 1 Drugs are believed to be the cause of 80 to 90% of the cases of TEN and at least 50% of the cases of SJS. The remaining cases are linked to infections and a variety of other etiologies, with a small fraction without obvious triggers. 1 SJS and TEN are sometimes considered to be part of a spectrum of skin disorders that includes erythema multiforme (EM). EM has been reported after several vaccines, although a causal relationship has not been established. 2 Very few reports of SJS after vaccination have been published, 3–5 and SJS and TEN are not usually considered among vaccine risks. We became interested in SJS and TEN after vaccination when we received a report of SJS after varicella vaccination. EM and SJS have been described as a result of varicella infection. 6 Live-attenuated varicella vaccine might very rarely cause SJS and TEN via the same mechanism as the wild virus. The pathophysiology of SJS and TEN is not certain but is believed to be immune-mediated. 1 The primary mechanism hypothesized for these disorders involves keratinocyte killing by cytotoxic T cells. Whether this occurs by direct necrosis or via release of cytokines that initiate apoptosis remains uncertain. 1 interleukin 2, interleukin 6 and tumor necrosis factor alpha have been implicated in SJS and TEN. 7–9 Vaccines induce these and other cytokines. 10–12 Individuals prone to develop SJS and TEN may have unusually high cytokine responses to immunization or may be exquisitely sensitive to a normal cytokine response. Given the complex makeup of vaccines and their potential for immune stimulation, a causal relationship between immunization and SJS and TEN is worth investigating. Case report. A previously healthy boy, age 27 months, had no history of illness, including no symptoms or signs of infection, or medication use in the 2 weeks before receiving varicella vaccine. One day after vaccination he developed a maculopapular exanthem over his face and torso. Three days after vaccination he was diagnosed with an “allergic reaction” and treated with colloidal oatmeal powder bath and diphenhydramine for pruritus. The following day the rash progressed to target lesions with sporadic bullous lesions on the arms and vesicular lesions in his mouth, associated with low grade fever. On the fifth day after vaccination he was hospitalized because of mucositis (involving the lips, eyes, anus and penile meatus) consistent with SJS. During hospitalization he developed cholestasis with jaundice and abnormal liver function tests. Two ultrasound examinations of his abdomen were normal. A skin biopsy was reported by the pathologist as consistent with Stevens-Johnson syndrome. Blood and urine cultures were sterile, as were lesional cultures for cytomegalovirus, varicella and herpes simplex virus. Polymerase chain reaction testing of the biopsy, blood and urine specimens was not conducted. Serologic testing for Mycoplasma pneumoniae was not reported. Treatment in the hospital consisted of intravenous fluids, parenteral nutrition, bacitracin ointment locally on exposed surfaces, hydrogel sheet dressings for open wounds, Atarax for itching, erythromycin ointment and artificial tears in the eyes and chlorhexidine swabs and mouth rinse containing diphenhydramine elixir, aluminum and magnesium hydroxide, simethicone and viscous lidocaine for dental care. He had previously developed erythema multiforme after amoxicillin/clavulanate and hives after trimethoprim-sulfamethoxazole. He recovered completely and has not received any further vaccines 22 months after this event. Methods. We used the national Vaccine Adverse Event Reporting System (VAERS) to identify and describe reports of SJS or TEN after immunization. VAERS is a national surveillance system for adverse events temporally, but not necessarily causally, associated with vaccination and is administered by the Food and Drug Administration and Centers for Disease Control and Prevention. VAERS receives spontaneous reports from the public, health care providers and vaccine manufacturers. VAERS provides a means of detecting possible rare adverse events that can be evaluated by further study, as recently illustrated by detection of intussusception after rotavirus vaccination. 13 Limitations include lack of consistent diagnostic criteria, underreporting, incomplete data, lack of denominator data and difficulty in evaluating whether a vaccine actually caused the adverse event reported. 14 Case definition. We categorized all cases according to the likelihood of SJS or TEN diagnosis. Patients with 30% skin involvement are termed TEN. 1 However, because most VAERS reports did not provide sufficient descriptive information, even after obtaining follow-up, to clearly distinguish SJS from TEN, cases were considered together as SJS/TEN. “Definite SJS/TEN” included adverse events with rapidly expanding macular rashes or atypical target lesions, involving 2 or more mucous membranes, epidermal detachment or skin necrosis and skin biopsy consistent with SJS or TEN. Cases with the same clinical features as “definite” cases but no skin biopsy, or cases diagnosed by a dermatologist, were labeled “probable SJS/TEN.” Cases reported as SJS or TEN that did not provide substantiating details or confirmation by a dermatologist were classified as “possible SJS/TEN.” Reports that gave a clear indication of a diagnosis other than SJS or TEN were classified as “other disease.” Case identification and follow-up. We searched 96 951 distinct adverse event reports to VAERS from the system’s inception in July, 1990, through September 15, 1999, for reports coded as SJS, TEN, dermal necrosis or “serious” EM after vaccination. (Serious events were defined as life-threatening or fatal, requiring hospitalization, prolonging hospitalization or causing permanent disability.) After preliminary screening VAERS reports meeting one of the case definitions were selected for follow-up by telephone by one of the authors. We attempted to obtain additional information from the physician who cared for the patient or from medical records provided by the attending physician using a questionnaire to obtain medical history, to classify the case and to assist with our causality assessment. Causality assessment. For definite and probable cases we evaluated the likelihood that the vaccine caused SJS/TEN by carefully reviewing information on alternative etiologies such as concomitant illness or medication use, searching for the presence of rechallenge with the vaccine followed by recurrence of SJS/TEN, comparing the time to onset with that reported for SJS/TEN (mean, 14 days; range, 1 to 45 days) 1 and reviewing the biologic plausibility of the hypothesis. Before concluding that alternative etiologies were not present, we required a follow-up interview with the reporter. Results. A flow diagram summarizing the case identification, classification and causality assessment process is shown in Figure 1. Among 131 reports identified by our initial search (including the case described above), there were 33 reports (25%) coded as SJS, 8 reports (6%) coded for TEN, 58 reports (44%) of serious erythema multiforme and the remaining 32 reports (24%) describing a variety of conditions with skin necrosis (e.g. injection site reactions). Thirty-five reports, all coded SJS or TEN, met the case definition for possible, probable or definite SJS/TEN before follow-up. Thirty-one of these cases were reported from the U.S., and 4 were foreign reports.Fig. 1: Case identification, classification and causality assessment.We were able to contact the reporter for 19 of the 35 cases. We obtained useful additional information for 14 reports (40%), but the reporter had no additional information for 5 cases. Two reports included complete medical records with the original report. Follow-up information changed the classification of 4 “possible” reports to “probable” and another 4 “possible” reports to “other disease.” Of the 35 reports meeting the case definition for possible, probable or definite SJS/TEN in our initial screening, 1 case was finally classified as definite SJS/TEN (described above), 14 cases as probable SJS/TEN, 13 cases as possible SJS/TEN and 7 as other disease. The cases classified as “other disease” included EM (4), varicella (1) and other conditions (2). Of the 1 definite and 14 probable reports, 1 case of probable SJS/TEN was attributed to a medication. In the reported case the individual developed SJS/TEN after simultaneously receiving a medication and a vaccination. On follow-up we learned the individual developed a recurrence of SJS/TEN when the individual took the same medication without revaccination (positive rechallenge). Eight patients with probable cases of SJS/TEN had acute illness or received various medications at the time of vaccination or after vaccination, but before the onset of rash, making it difficult to determine whether the vaccine initiated the rash. Two of these 8 patients received medications (ibuprofen, diclofenac) or had illness (Mycoplasma infection) that have been associated with SJS and TEN. 1 In the one definite and the five remaining probable cases (outlined in Table 1), onset of rash occurred after vaccination but before the use of other medications. For these cases the median time to onset of first symptom from vaccination was 1 day (mean, 5 days; range, 0 to 22 days). In these six cases singly administered vaccines included hepatitis B (2), influenza (1), and varicella (1). Two children received more than one vaccine simultaneously. One child received Haemophilus influenzae type b and measles, mumps and rubella vaccines. The other child received diphtheria, tetanus, whole cell pertussis, H. influenzae type b, measles, mumps and rubella and oral polio vaccines. In all six cases at least one antiviral vaccine was administered. None of these individuals reported a prior history of drug allergies. All of these cases occurred in the US.Table 1: Probable SJS/TEN reports with no medication use between vaccination and onset of rashDiscussion. For the one definite and five probable cases, two features support a causal relationship between vaccination and SJS/TEN. First the patients did not have evidence of an acute illness and were not taking other medications at the time of vaccination, reducing the chance the rash was caused by another etiology. Second the timing of the onset of symptoms and the diagnosis of SJS/TEN after vaccination is consistent with the published literature on SJS and TEN after medication and infection. 1 A causal link between SJS/TEN and vaccination cannot be established by this study because we cannot be sure the medication given after the onset of rash or clinically inapparent illness at the time of vaccination did not contribute to the development of SJS/TEN. In addition we were limited by the absence of reports demonstrating recurrence of SJS/TEN with revaccination and the nature of passive surveillance which precludes us from estimating the relative risk of SJS and TEN from vaccination. SJS and TEN may also occur idiopathically. However, by limiting our analysis to definite or probable SJS/TEN cases, we reduced the likelihood that these events were actually other disorders. Requiring a follow-up interview with the reporter before concluding alternative etiologies were not present minimized the chance of concluding that a case of SJS/TEN is plausibly linked to vaccination simply because of a reporting error. Another limitation of this study is that a uniform diagnostic approach was not applied to each patient to investigate known associations with SJS/TEN, and we do not know whether complete evaluations were conducted to rule out other causes. When a case of SJS or TEN occurs after vaccination, a complete diagnostic evaluation should be undertaken to determine the etiology. In addition to the antigen(s) vaccines often contain other components, such as adjuvant and preservative. When evaluating whether a vaccine caused SJS or TEN, these components must be considered. These other components can also have important clinical implications. For example if a person has a previous episode of SJS or TEN after an antibiotic, the antibiotic content of all vaccines should be reviewed before administration. Varicella vaccine contains the antibiotic neomycin. Neomycin has cross-sensitivity with other aminoglycoside antibiotics. If a child is allergic to an aminoglycoside antibiotic, the varicella vaccine might be contraindicated. Despite the plausibility of a causal relationship between vaccination and SJS and TEN, our review of VAERS is reassuring in view of the small number of reports and the large number of vaccines administered in the US during the surveillance period. If vaccination causes SJS and TEN, it happens very rarely and under most circumstances is unlikely to outweigh the benefits of vaccination. Although immunization has not been evaluated as an etiology of SJS and TEN, the possible role of immunization should be considered in future studies because of the severe morbidity and mortality associated with SJS and TEN and the plausibility of this hypothesis. Acknowledgments. We thank Dr. Lois LaGrenade for helpful suggestions, the health care providers and staff who graciously gave of their time to be interviewed and send medical records and all individuals who reported cases to VAERS. Appendix. The VAERS Working Group consists of individuals from the Food and Drug Administration (FDA), CDC and Health Resources and Services Administration as follows: FDA, Susan Ellenberg, Suresh Rastogi, Carol Krueger, Robert Wise, Manette Niu, David Davis, Frederick Varricchio, Phil Perucci, Tracy DuVernoy, Chris Bechtel; CDC, Robert Chen, Penina Haber, Vitali Pool, Wendy Wattigney, Tara Strine, Robert Pless; Health Resources and Services Administration, Vito Caserta, Geoffrey Evans. Information on VAERS, including the reporting form, is available on the internet at http://www.fda.gov/cber/vaers/vaers.htm. We encourage reports of SJS, TEN and any other adverse events after vaccination to VAERS using this form or by calling 1-800-822-7967.