Galactography is not an obsolete investigation in the evaluation of pathological nipple discharge

Purpose To evaluate the malignancy rate and diagnostic performance of galactography in patients with pathological nipple discharge (PND) after negative clinical breast examination, mammography and ultrasound. Materials and methods We retrospectively evaluated all galactograms obtained between January 2006 and December 2014 in women with PND. Galactographic findings were classified into 6 groups according to a modified Galactogram Image Classification system (GICS) to comply with the breast imaging reporting and data system classification. Observers were blinded to the final histology and clinical outcome at the time of analysis. MRI was performed as a problem solving ancillary examination. Imaging findings, pathological diagnosis and follow-up data were evaluated. The diagnostic performance of MRI and technically successful galactography in the detection of neoplastic or risk lesions were separately calculated. Results A total of 146 patients with PND (mean age, 51.5 years; range, 17–93) were examined. Malignant lesions were detected in only 4 patients (2.7%) and risk-lesions in 5 patients (3.4%). Only one low-grade ductal carcinoma in situ was missed by galactography (GICS 1) and MRI. MRI examinations were performed in 21 (14.4%) patients; one of these patients (4.8%) had a malignant finding (GICS 0), two (9.5%) had risk-lesions (GICS 2 and 5). In the detection of neoplastic or risk lesions the sensitivity and specificity of galactography were 77.4% and 75.7% and of MRI 85.7% and 71.4%, consecutively. Conclusion The malignancy rate is negligible if clinical, mammography, ultrasound and galactography examinations are negative. Galactography remains a practical, valuable and cost-effective examination procedure. If galactography is technically unsuccessful, MRI should be considered as an additional ancillary tool to evaluate the possible etiology of symptoms, but the routine use of MRI in all patients cannot be justified.