Sarah Tang and Gerald Gui

Introduction

After breast lumps and mastalgia, spontaneous nipple discharge forms the next most common symptom, comprising 3–8% of referrals to symptomatic breast clinics.^1,2 The majority of nipple discharge is benign, with up to 20% being associated with an underlying malignancy.³ Conventional methods of investigating nipple discharge include mammography, ultrasound and smear cytology, each of which have recognised limitations. Standard operations such as microdochectomy and major duct excision are undirected procedures that carry a risk of leaving undiagnosed pathology in the breast.

Mammary ductoscopy allows direct visualisation of the duct epithelium to locate the lesion precisely and to map the three-dimensional anatomy. Endoscopic instruments for diagnostic biopsy and therapeutic excision are available. The ability to visualise normal or benign ductal structures may facilitate conservative management of symptomatic nipple discharge and enable targeted excision of visualised lesions or indeterminate areas.

Causes of nipple discharge

The causes of nipple discharge are wide ranging. The majority of discharge is physiological, hormone related or results from benign breast change. A recent meta-analysis that included over 3000 women with nipple discharge demonstrated an overall incidence of underlying breast malignancy of 18.7%,³ which is higher than the figure reported in many individual series.

Bilateral multiduct discharge can be stimulated by nipple manipulation in the majority of premenopausal women. The production of such physiological fluid is exploited by researchers of the intraductal approach for cytological, molecular and protein-based studies. This fluid is more likely to be released from the ducts if the nipples are repeatedly stimulated by chafing against clothing and during physical activities such as jogging. Squeezing of the nipples to elicit such discharge perpetuates symptoms. This is because the nipple ducts are plugged by keratin and repeated stimulation or squeezing removes the keratin plugs and allows the fluid that is normally present in the ductal tree to leak on to the surface of the nipple. The fluid in physiological discharge ranges from clear to white to yellow to green to black.

The most common physiological nipple discharge is lactation. Ongoing milky discharge may occur up to 2 years following a pregnancy and is a normal phenomenon. Galactorrhoea can also result from prolactin-secreting pituitary adenomas, medication that influences the oestrogen, progesterone or prolactin pathways, hypothyroidism and recreational drugs such as marijuana. Commonly prescribed medical drugs that may cause nipple discharge are summarised in Table 3.1.

Duct ectasia and benign breast change are common causes of multiduct discharge. In duct ectasia, the breast ducts become tortuous and dilated, predisposing to fluid accumulation. This fluid may discharge spontaneously or upon manipulation when the sphincters distal to the lactiferous sinuses relax and the keratin plugs are displaced. Common situations that may trigger loss of these keratin plugs and sphincter relaxation include massage and warm comfortable environments such as in a bath or under bedsheets. Cysts do not usually communicate directly with breast ducts. It has been suggested that inflammation related to a cyst may result in erosion into a duct system but this is unproven. The colour of nipple discharge associated with duct ectasia and fibrocystic change can vary from clear to white, yellow, grey, green, brown or black. This is the same colour range seen in physiological discharge. Nipple discharge arising in relation to inflammation and irritation of the lining of the ducts may be blood stained and contain acute inflammatory cells.

Persistent spontaneous discharge from a single nipple orifice is usually indicative of specific pathology involving that duct. Benign intraductal papillomas account for about 80% of single-duct bloodstained nipple discharges.^2,4,5 Papillomas give rise to nipple discharge of varying colour but as these structures can bleed intermittently, papillomas are often associated with blood staining. Recent bleeding may manifest as frank blood, but stagnant bloodstained fluid can be dark red, brown or black.

Periductal mastitis is a chronic and recurrent inflammatory condition associated with smoking in younger women. Purulent nipple discharge can be seen in later stages when the duct ruptures and an abscess or mammary duct fistula has developed.

Invasive breast cancer is an uncommon cause of spontaneous nipple discharge as the proliferating mass usually obliterates the duct lumen. Ductal carcinoma in situ or an extensive area of in situ disease in association with an invasive breast cancer may predispose to bloodstained nipple discharge. Bloodstained nipple discharge should be distinguished from bleeding from the nipple surface that may occur with Paget’s disease.

Assessment

A thorough breast problem orientated history is taken including a drug history, and a complete clinical examination should also be performed.

Bilateral physiological multiduct discharge aggravated by manipulation in younger patients usually resolves when the triggering factors are removed. Bilateral and profuse milky discharge in the younger population should be investigated by measuring serum prolactin and if elevated, supplemented by magnetic resonance imaging (MRI) of the pituitary gland. Thyroid function tests may be indicated if there are appropriate clinical features.

The colour of nipple discharge is not a reliable method of distinguishing between physiological, benign or malignant aetiologies.

Mammography is indicated in women over the age of 35 years and may demonstrate a mass or microcalcification in association with nipple discharge of malignant aetiology. Even in younger women with bloodstained discharge, mammography can demonstrate calcification in women with widespread ductal carcinoma in situ (DCIS). Other mammographic features may also be evident in patients with benign nipple discharge. Duct ectasia may occasionally be visible on mammography. The sensitivity of mammography in detecting pathology in nipple discharge is 57.1% (positive predictive value (PPV) of 16.7% and negative predictive value (NPV) of 91.4%).⁷ In a study of 306 patients with nipple discharge who had normal mammography and ultrasound, 10% were subsequently found to have underlying malignancy.⁷

Ultrasound can reliably diagnose duct ectasia and can identify discrete intraductal lesions. Papillomas below a threshold size of 1–2 mm may not be visible on ultrasound imaging. Ultrasound-guided core biopsy can be used to obtain a tissue diagnosis of any lesions visualised but papillary lesions are often assessed as an indeterminate B3 lesion and so usually still require excision for a definitive tissue diagnosis. A vacuum-assisted ultrasound-guided mammotome biopsy can be both diagnostic and therapeutic. Fine-needle aspiration of papillary lesions is usually unhelpful as cytology rarely resolves diagnostic uncertainty.

Ductography is less widely used but is an investigation that gives clear delineation of the breast anatomy. A small amount of radio-opaque contrast is instilled into a cannulated nipple duct. Accurate detection of small filling defect(s) caused by papillomas and duct narrowing or obstruction from malignant change can be recognised. However, ductography does not provide a tissue diagnosis and the precise position of the area of interest within the breast is not usually available to the surgeon, although it is possible to target the area of abnormality with a wire localisation procedure. The use of MRI to evaluate the ductal tree is gaining interest but is not a standard investigation of nipple discharge. In a comparative study of 163 patients with nipple discharge who had normal mammography and ultrasound, ductography was found to have a PPV of 19% and an NPV of 63%. MRI was performed in 52 patients and found to have a PPV of 56% and an NPV of 87%.⁷

Smear cytology of nipple discharge may show the presence of malignant cells, papillary cells, red blood cells, benign duct epithelium or foam cells of macrophage origin. The available literature demonstrates sensitivities of smear cytology ranging from 11.1% to 16.7%^8,9 and specificities ranging from 66.1% to 96.3%. In Simmons et al.’s study of 108 pathological specimens for nipple discharge, the PPV of smear cytology was 50% and the NPV was 76.5%.⁸ These data arise from an era before mammary ductoscopy when surgery providing the pathology used as the gold standard could not be targeted, so some of the pathologies present may have been missed at surgery, hence lowering the PPV of cytology. A further limitation arises when insufficient epithelial cells are available for a reliable diagnosis. This may be overcome by enhanced techniques of obtaining nipple fluid such as ductal lavage to increase the cell yield, which can improve the performance of cytological assessment.

Conventional surgery

The indications for surgical intervention are persistent (defined as >2 per week) single-duct discharge in a woman over the age of 50 years, guided further by the results of abnormal investigations. Pressure over a discrete lesion such as a symptomatic papilloma often results in duct-specific discharge. The target lesion may be skin marked or wire localised if impalpable, to ensure excision of the target lesion.

Current surgery for pathological nipple discharge consists of major duct excision and microdochectomy. Major duct excision in the absence of an imaging abnormality is an undirected operation where a disc or wedge of retroareolar tissue is excised. The amount of breast tissue that is excised is at the discretion of the surgeon’s clinical judgment but often includes a considerable amount of tissue that does not need to be resected for a benign process. The greater the amount of tissue resected, the higher the risk of surgical complications including altered sensation, haemoseromas, poor aesthetic outcome from volume loss and nipple ischaemia. Microdochectomy aims to resect a single duct guided by passing a lacrimal probe down the offending duct system. The branching system of the ductal tree makes this procedure imprecise and the surgeon is blind throughout the operation as to the location, depth and number of lesions that form the target of the resection. The risk of perforating the cannulated duct with the probe is high and if the surgeon performs a wide resection around the probe then it in effect becomes a much bigger procedure. Proximal lesions further than 3 cm from the nipple base are likely to be missed by these undirected techniques.

Despite the limitations, these operations do form the current standard surgical practice. Surgery to the breast ducts should be avoided in women of childbearing age who wish to breast feed, as surgical injury or scarring are likely to lead to restricted or failed lactation.

Mammary ductoscopy and nipple discharge

Spontaneous nipple discharge lends itself well to mammary ductoscopy (MD) as the ductal orifice is readily identified and coexisting duct ectasia can aid scope passage. Positive endoscopic findings can be used to guide surgery resulting in targeted excision. Visualisation of remaining ducts can rule out coexisting pathology and avoid unnecessary resection of normal parenchyma. MD is not widely available and its role as an established procedure remains to be defined.

MD by an experienced endoscopist can be more accurate than mammography in identifying extensive in situ disease (65% vs. 50%).¹⁰ Ling et al. found that MD-guided intraductal biopsy had a significantly higher papilloma detection rate than undirected miocrodochectomy (92.6% vs. 40.7%, P < 0.05).¹¹ Normal findings at MD may endorse a conservative approach or alternatively facilitate a recommendation for duct transaction alone to manage symptomatic nipple discharge.

History and technology

Microendoscopic technology has advanced and earlier limitations of poor optical resolution and problems with access because of the large calibre of scopes have been overcome. The development of working channels within microendoscopes has made it possible to biopsy lesions under direct vision using cytology brushes and microbiopsy forceps. Accessories for localisation, such as self-retaining hookwires or rhomboids that are immobilised by expansion within the duct, can be passed down the working channel to demarcate the area of interest for open therapeutic excision.

Current scopes can be flexible or rigid, with diameters that range from 0.7 to 1.2 mm (Fig. 3.1). Microendoscopes magnify up to 60 times to produce high-quality images with saline for insufflation and irrigation of the ducts. More recent developments include autofluorescence technology,¹² which may help to distinguish between benign, premalignant and malignant lesions.

Technical considerations

The duct orifices are identified by nipple fluid expression by warming, massage or gentle pressure. The orifice may be dilated and some systems require a working shaft to be inserted, through which the scope is passed.

MD, when associated with a standard surgical procedure, is often performed under general anaesthetic. Diagnostic MD can be performed as an outpatient procedure with topical local anaesthetic applied to the nipple, followed by periareolar infiltration and/or infusion down the cannulated duct, the latter facilitating relaxation of the muscle sphincters.^13,14 Methylene blue dye, marking wires, clips and transillumination¹⁵ may be used for orientation, to localise lesions, guide surgical excision and facilitate pathological assessment.^15–17

Complications of MD are uncommon and include pain, inflammation and infection. Duct perforation by the scope creates a false passage that can be recognised by the yellow cavernous honeycomb appearance of adipose tissue.

Intraduct appearances

There is general consensus on the intraductal appearances of common lesions from studies that have used histological correlation (Figs 3.2 and 3.3). Malignant lesions are more likely to display haemorrhagic characteristics, streaking, fissuring and irregularity of the ductal walls or present with a luminal mass that might obstruct the duct. In contrast, benign non-papillomatous lesions have a smooth, level surface without haemorrhagic features.^4,18–21

MD sensitivity at detecting papillomas is high at 96–97%^11,19 with a PPV of 73%.²¹ Benign duct strictures may on occasion limit MD access.²¹ Sensitivity for detecting DCIS and invasive disease is lower, with reports ranging from 41% to 81%.^19,22 The presence of an extensive non-invasive component associated with an invasive cancer increases significantly the likelihood of abnormal ductoscopic findings (71% vs. 16%).²³

Pathological considerations

The diagnostic sensitivity of MD can be improved by incorporating ductal lavage cytology,²⁴ which generates information from the proximal ducts that may be out of reach of the ductoscope. In a study of 415 women, the PPV of 80% of MD alone in detecting DCIS was improved to 100% by the addition of lavage cytology.²⁵ The cell count can be further enhanced using fine brushes passed down the working channel to exfoliate intraluminal lesions under direct vision, yielding up to 33 000 cells.^26,27 A cell-rich endoscopy specimen allows better discrimination between benign cells, mild to severe atypia and malignant cells. A normal duct on MD is infrequently associated with abnormal cytology.¹⁴

Reliable methods of intraductal biopsy are essential if MD is to achieve a histological diagnosis without surgical excision. Various techniques have been developed using vacuum-assisted biopsy systems, endobaskets and grasping clips.^11,23,28 The intraductal breast biopsy (IDBB) system pioneered by Hunerbein and Matsunaga consists of a 0.7-mm gradient index endoscope covered by an external metal sheath containing a side opening aperture near its tip.^23,28 This technique yields diagnostic material in 89–92% of cases with a sensitivity of 76.2% and a specificity of 100% for papillomas.^10,23 Using biopsy clips, Ling et al. were able to produce diagnostic samples in 90% of patients.¹¹ Biopsy techniques can be extended to perform therapeutic endoscopic papillomectomy.^29,30 In two studies, IDDB stopped symptomatic nipple discharge with therapeutic efficacies of 77.6% and 95.4%.^28,29 Carcinomas are associated with a lower diagnostic biopsy yield (58.3%).²⁸ It is postulated that carcinomas are generally located more peripherally in the terminal ducts and can be difficult to access by MD.²⁸

Limitations

In the context of nipple discharge, the majority of lesions are found within 5 cm of the nipple but up to 37% may be proximal and therefore out of the reach of a 6-cm ductoscope.^15,21 Unlike other applications for microendoscopy such as the salivary gland, where there is a sizeable single-duct orifice, breast ducts open via multiple tiny orifices at or just below the nipple surface. There is variability in the number of duct systems, with one anatomy study demonstrating 29 ducts arising from 15 orifices.³¹ This study also found that some ducts branch within the nipple itself and may not be accessed readily by lavage cannulae or MD.

Ductoscopy and breast cancer

MD as an adjunct to breast conservation surgery for cancer is technically feasible, with high rates of tumour visualisation.¹³ In a subsequent publication, Dooley successfully identified the malignant lesion on MD in 150 out of 201 patients (74.6%), found additional lesions outside the planned excision site in 83 cases (41%) and decreased the positive margin rate from 23.5% to 5%.³² Louie et al. reported lower tumour visualisation rates in 6 of 14 (43%) patients and did not show any benefit of MD in reducing positive margin rates.²² The role of duct endoscopy in breast cancer management is currently limited outside clinical trials.

References