The association of pathological nipple discharge with malignancy, in a small proportion of patients, results in numerous women undergoing exploratory surgery to rule out evidence of serious pathology. Studies assessing pathological breast discharge report associated malignancy rates of up to 20% [1
]. Those studies with high rates of malignancy usually include patients with obvious radiological malignancy, whose initial operative procedure was a diagnostic excision of a definite breast abnormality. These studies were often conducted prior to an era of widespread preoperative diagnosis with core biopsy. In contrast, in this study, we were concerned with the outcome of patients, who in the absence of an obvious cause of discharge, underwent major duct excision primarily as a diagnostic procedure. These patients present a particular challenge to surgeons and may have a different risk profile than those patients presenting with a more clinically and radiologically apparent breast carcinoma, and who also present with associated discharge.
In the present study of 211 patients, only 4.3% of patients were diagnosed with breast carcinoma following major duct excision. This includes a cohort of patients with pathological and non-pathological discharge, both of which were associated with a low risk of carcinoma (6% and 2.3% respectively). Overall, these results emphasise that the traditionally reported 10–20% rate of breast malignancy [4
] associated with pathological discharge is not representative of those patients who present with pathological discharge alone. In a similar patient population to this study, King et al found only one case of malignancy out of 39 patients undergoing duct excisions [16
], and in a study by Chaudary et al. [3
] on patients undergoing microdochectomy, the carcinoma rate was 5.9%. Other studies have produced similar results [8
]. Justification of diagnostic excision where there is such a low yield of malignancy becomes increasingly difficult, and emphasises the need for more discriminatory guidelines for this particular cohort of patients. In addition, it must be considered that excision of the ducts may not yield a diagnosis. In this study, a specific pathological cause of discharge was not identified in 25% of patients and in a further 25% 'duct ectasia/duct dilatation' was the only pathological finding. Our findings indicate that duct excision, used as a diagnostic test, has a sensitivity in the region of 50–75%. There is increasing evidence that a considerable number of pathological lesions are not located in close enough proximity to the nipple for a surgical duct excision to adequately locate them [5
], which may further explain this low diagnostic yield.
The low rate of malignancy underpins the difficulty in identifying those patients most at risk of carcinoma. However, an important finding of this 14-year study was that none of the 59 patients who were operated on for a non-bloody discharge had malignancy on excision. The results of this study suggest that patients with a non-bloody discharge, who have no obvious clinical or radiological signs, are at low risk of malignancy. While there is certainly an association between non-bloody discharge and breast carcinoma [5
], this association is more readily apparent in studies of patients with obvious breast abnormalities. In studies focusing on patients undergoing microdochectomies, whose indication for operation is for discharge alone, the clinical presentation appears to be different. A study by Chaudary et al. [3
] on patients undergoing microdochectomy, with no associated lump, revealed that 5.9% were diagnosed with carcinoma. In that study, only two of the 16 patients who were diagnosed with carcinoma had serous discharge, but even these were hemoccult positive. In the study by Welch et al. [22
], 14 of 16 patients diagnosed with carcinoma following microdochectomy had a bloody discharge. In other studies with this type of patient population, there is either no association [17
], or a very low association [2
] with carcinoma and non-bloody discharge. Occasionally, there are exceptions to these findings [5
] such as in the study by Hou et al. [5
], reporting on a Taiwanese population with a very high rate of carcinoma (20%) and an association with serous discharge, or the study by Sharma et al [13
] demonstrating the association of carcinoma even with brown or purulent discharge. Essentially, however, it may be that small, sub-clinical cancers are more commonly associated with bloody discharge. In a patient population of this type, whose risk of carcinoma is typically low (approximately 5%), the evidence suggests that those with a non-bloody discharge are at particularly low risk, with a probable overall risk of malignancy of less than 1 or 2%. To put this in perspective, in the U.S. patients with mammographic lesions do not undergo (pre-operative) biopsy if their risk of malignancy, radiologically, is considered to be less than 2% [25
] (BI-RADS 3). Therefore, it could be argued that this low risk sub-group with non-bloody discharge may also be monitored safely.
In this study, we also analysed those who had a specific cause for discharge identified following excision, and compared them to those with non-specific or normal findings. It was of interest, that patients on hormone therapy were found to be more at risk of non-specific findings on excision. Not surprisingly, discharge of a colour other than bloody or serous, was associated with non-specific disease, whereas serous discharge was associated with specific lesions. All the specific lesions in patients with serous discharge were papillomas including one case of papilloma and ADH. Currently, there is no consensus as to whether papillomas pose a significant risk of carcinoma or whether they should be removed [26
]. Certainly, the finding that uniductal, as well as serous discharge, was associated with specific lesions supports the use of the term 'pathological' discharge in this category.
In the present study, the efficacy of various clinical parameters and tests were also analysed. While clinical history was the most useful indicator of risk with regard to the nature of discharge, age and menopausal status were of limited value. Forty four percent of patients diagnosed with carcinoma were under the age of 50, and were premenopausal. This contrasts with the findings of Lau et al. [23
], who reported that 10 of 11 patients with cancer were postmenopausal and who subsequently recommended excision in all postmenopausal women with pathological discharge. Clinical examination was also found not to be discriminatory, most likely because patients with suspicious palpable abnormalities would have undergone a different procedure or would have had a preoperative histological diagnosis. All the patients with carcinoma in this study, even those with an associated palpable abnormality, were not considered suspicious on clinical examination.
Imaging in this study was also found to be unreliable. Only 9 mammograms demonstrated any abnormality and there were 8 false-negative mammograms. Ultrasound performed on 33 patients identified a cause of discharge in only 3 of the patients. The 2 patients with carcinoma who underwent an ultrasound examination were not diagnosed pre-operatively. The poor sensitivity of radiology, in this patient setting, is a reflection of higher pre-operative diagnosis rates. The more widespread use of core biopsies means that there are considerably less patients undergoing exploratory surgery for lesions visible on mammography. This change in practice is reflected by our high rate of false-negative mammography in diagnostic surgery and our very low overall malignancy rate. However, these results do suggest that routine ultrasonography of the retroareolar region, in the absence of mammographic abnormalities is of limited value.
Other potential pre-operative tests such as ductography (galactography) or ductoscopy are not performed in our unit. Ductography was introduced over 60 years ago [29
], but has not become standard practice in many parts of the world [6
]. It has not proved to be very effective at discriminating between benign and malignant lesions, and a negative test does not exclude malignancy [1
]. Likewise, ductoscopy is also limited by its low specificity [18
]. Despite these disadvantages, ductoscopy and ductography appear to play an important role in guiding a more conservative excision or in localising lesions at some distance from the nipple [18
]. Several studies have indicated that lesions are frequently found at some distance from the nipple orifice [5
]. More disturbingly, Hou et al. [5
] reported that 70% of patients with carcinoma presenting with discharge had their lesions located > 2 cm from the nipple [5
These studies imply the need for localisation techniques. Localisation of lesions appears to be most effective if done at the time of operation. In a study by Cabioglu et al [7
] most patients had preoperative ductography performed. However it was patients who underwent a ductography guided operation or that had a localising study at the time of operation, who where significantly more likely to have a specific underlying lesion identified, as compared to patients who underwent a central duct excision alone. Similarly, Van Zee et al. [33
] demonstrated the value of immediate preoperative ductography in identifying specific lesions, though a more recent study suggested that localisation through means of ductoscopy guided ductal excision may not be as effective [19
]. Our study had a high rate of non-specific or benign findings on excision, indicating that a surgical procedure alone may not be adequate as a diagnostic tool. Increasing interest into more accurate techniques to identify lesions, including the use of thinner more flexible fiberoptics ductoscopy [34
] may enhance the value of surgical excision.