Recurrence of lymphoedema-associated cellulitis (erysipelas) under prophylactic antibiotherapy: a retrospective cohort study


*Corresponding author, Department of Lymphology, Hôpital Cognacq-Jay, Site Broussais, 102 rue Didot, 75014 Paris, France, tel. +33 1 56 56 42 57; fax +33 1 56 56 42 58; E-mail:


Objective  To identify the predictors of successful antibiotic prophylactic treatment using benzathin-penicillin G to prevent recurrence of erysipelas in patients with secondary upper limb lymphoedema.

Design  A retrospective cohort study.

Setting and patients  Patients with secondary arm lymphoedema were recruited in a single lymphology unit between 1990 and 2003. All patients had had at least three recurrences of erysipelas. Patients were given 2.4 MU benzathin-penicillin G intramuscularly at 14-day intervals. For each patient, the following data were recorded: characteristics of breast cancer treatment (type of surgery, radiotherapy, hormone therapy), number of erysipelas recurrences before inclusion, patient characteristics including body mass index (BMI) and lymphoedema volume at inclusion.

Main outcome measures  The outcome studied was the occurrence of erysipelas on the affected arm under antibiotic prophylactic treatment.

Results  With a 4.2-year median follow-up from the onset of antibiotic prophylactic treatment, 23 of 48 women experienced recurrence of erysipelas. The median duration of erysipelas recurrence-free period under this treatment was 2.7 years. The estimated rate of recurrence was 26%[95% confidence interval (CI) 13–38%] at 1 year and 36% (95% CI 22–50%) at 2 years. No patient stopped the treatment because of side-effects. No predictive factor of erysipelas recurrence under antibiotic prophylactic treatment was identified.

Conclusions  Antibiotic prophylaxis using benzathin-penicillin is a well-tolerated treatment of erysipelas recurrence in patients with upper limb lymphoedema secondary to breast cancer. The rate of erysipelas recurrence was 26% at 1 year in patients who had a history of at least one erysipelas. We did not find any predictor of erysipelas recurrence.