Clinicopathologic Correlation in Erythema Multiforme and Stevens-Johnson Syndrome

Abstract

Background and Design: To confirm the recent hypothesis that the spectrum of severe erythema multiforme (EM) is actually composed of two different disorders, we retrospectively studied 38 such cases, particularly in regard to their histopathologic features. Based on photographs and a recent proposal, the cases were classified as EM major when the eruption consisted of typical or raised atypical target lesions located on the extremities and/or the face or as Stevens-Johnson syndrome when the eruption consisted of flat atypical target lesions or purpuric macules that were widespread or distributed on the trunk. The cases were also assessed for causal agent. A biopsy specimen was obtained in each case. Several histologic parameters were analyzed (and scored) without clinical data and correlated to the clinical pattern. These parameters were first studied in a global assessment and then in a detailed evaluation. Results: The global assessment showed two different histologic patterns: (1) a predominantly inflammatory pattern characterized by a lichenoid infiltrate and epidermal necrosis that mainly affected the basal layer; and (2) a predominantly necrotic pattern in which major epidermal necrosis and minimal inflammatory infiltration were found. The former pattern was associated with EM major, the latter with Stevens-Johnson syndrome (P<.001) and with drug-related cause (P<.001). The detailed evaluation showed also less epidermal necrosis, more dermal inflammation, and more exocytosis in EM major. Conversely, there was more epidermal necrosis, less dermal inflammation, and less exocytosis in Stevens-Johnson syndrome. The difference was statistically significant for the inflammation and exocytosis. Conclusions: This study suggests that the two different symptomatologies in the spectrum of severe EM correlate with two different patterns of histopathologic changes. A prospective multicentered study should be conducted to definitively characterize these entities. (Arch Dermatol. 1995;131:1268-1272)