How mammillary fistulas should be managed

Mammillary fistulas are uncommon, but when they occur they cause prolonged morbidity. The etiology and management strategies are less well established. The purpose of this study is to evaluate the etiologic factors and assess the results of surgical treatment. It is a retrospective study of all patients treated for mammillary fistula from 1990 to 2001. The clinical data, including complications of surgical treatment, were collected from medical records. Fistulas were segregated into simple and complex fistulas before analyzing the results of surgical treatment. Thirty-five patients were treated during this period. A history of either drainage of a subareolar abscess or spontaneous rupture of an inflammatory mass preceded the development of mammillary fistula in the majority of patients. Previous Hadfield's procedure for duct ectasia contributed to the development of fistula in seven patients. Seventeen patients presented with simple fistula. A large proportion of them were treated by total duct excision in recent years, with a higher rate of recurrence (4/6). Eighteen patients presented with complex fistulas; two of them had recurrences following surgical treatment. The overall recurrence rate was 23%. The majority of the patients showed features of periductal mastitis on histologic examination. Postoperative wound infection was positively associated with fistula recurrence. The best management of mammillary fistula remains a problem. Simple fistulas should be treated by fistulectomy and primary closure. Total duct excision should be reserved for complex fistulas. Postoperative wound infection is also a major factor in fistula recurrence. All patients should receive antibiotics. Surgery for duct ectasia has caused fistulas in 20% of cases in our study, raising the issue of restricting total duct excision to more severe forms of the disease. Mammillary fistulas should be treated more appropriately in a specialized breast unit with particular interest in benign breast disease.