To determine the diagnostic yield of skin biopsy in patients with leukemia, new skin lesions, and suspected infection.

We reviewed the medical records of all patients with leukemia who underwent skin biopsy for new lesions and clinical suspicion of infection over 4 years. Biopsy was considered to have changed the diagnosis, if the results differed from the prior leading clinical impression.

Seventy-six (39%) of 195 patients had infections identified via skin biopsy. Among the remaining patients, the most common diagnoses were leukemia cutis, drug reactions and Sweet's syndrome. Ulcerated or necrotic lesions and bacteremia or fungemia were the only independent predictors of infection. 55% of patients with severe neutropenia had biopsy-proven infectious causes of their skin lesions. One third of all afebrile patients had skin manifestations owing to infection. Skin biopsy results differed from the initial clinical impression in 34% of all patients. In 45% of infected patients, pathogens were identified by skin biopsy alone.

Noninfectious causes accounted for a large proportion of new skin lesions in leukemia patients with suspected infections. Absence of neutropenia or fever did not rule out infection. Ulcerated or necrotic lesions and bacteremia or fungemia were independent predictors of infection. In the evaluation of patients with leukemia and new skin lesions, skin biopsy remains an important procedure to rule out infection, and is particularly useful for pathogen identification.