Discharges and Secretions of the Nipple

Nipple Aspiration Fluid: Characterization and Significance

Many investigators have an interest in intraductal physiology and cytology as an indicator of risk of breast cancer. It is thought that breast cancer results from a cascade of sequential molecular and morphologic events that occur in the ductal epithelial cells. If that process can be detected before transition to a malignant phenotype, there is the opportunity for prevention strategies ( Fig. 6.3 ). The methodologies for obtaining breast secretions and classifying cellular patterns, as well as the biochemical makeup of nipple aspiration fluid (NAF), are discussed in this section.

NAF is a simple, minimally invasive method of sampling histologic alterations within the breast. After cleaning of the nipple orifice to remove keratin plugs, the breast is massaged from the base to the nipple. Using a modified breast pump ( Fig. 6.4 ), suction is applied to elicit the fluid. NAF appears as droplets on the nipple and is collected with small capillary tubes. Samples are usually pooled for examination to improve cellular yield.

Cytologic examination of the breast fluid has long been considered a potential aid in the detection of breast disease. Saphir (1950) and Ringrose (1966) reported the cytologic criteria for the diagnosis of chronic mastitis, intraductal papillomas, and carcinoma based on stained smears of spontaneous nipple discharge.

After the successful application of the cervical smear (Pap smear) for the diagnosis of cervical cancer, George Papanicolaou developed the concept of a “Pap smear” for breast cancer. In 1950 Papanicolaou began a study to investigate normal, atypical, and malignant breast epithelia. In this study, he examined breast aspirates from 917 asymptomatic individuals. Using a breast pump, secretions were obtained unilaterally in 171 (18.5%) and bilaterally in 74 (8.1%). Premenopausal women yielded a higher percentage of fluid than postmenopausal women, with the highest percentage found in women 20 to 39 years of age. Four occult carcinomas and one ductal carcinoma in situ (DCIS) were found. Papanicolaou also sought to define “normal” cytologic findings in nipple fluid. The two cell types most often encountered were foam cells and duct epithelial cells. Scant histiocytes and lymphocytes were also present.

In the 1970s, using a modified breast pump, Sartorius and colleagues attempted nipple aspiration in 1503 women with suspected breast disease and 203 asymptomatic volunteers. They developed the histologic classification to define cytologic changes; it required a minimum of 50 to 100 cells for diagnosis. Diagnostic categories included the following:

• Normal: all duct lining cells uniform in size and staining characteristics

• Hyperplasia: excessive number of ductal groups with multilayering and slight variations in size and shape, but without significant nuclear abnormality and a constant nucleocytoplasmic ratio

• Atypical hyperplasia: criteria similar to that used for hyperplasia but with greater variation in nuclear size and shape, abnormal distribution of chromatin and increased nuclear:cytoplasmic ratio, and prominent nucleoli

• Suspected carcinoma: criteria similar to atypical hyperplasia but with marked nuclear abnormality, chromatin clearing, nuclear membrane irregularity, and nucleoli

See Fig. 6.5 for examples of nipple fluid cytology and corresponding histology.

Of the 203 asymptomatic volunteers, NAF was obtained in 163 (80%), with adequate cellular samples obtained in 48.7%. Of the 1503 women with breast disease, NAF with adequate cellularity was obtained in 825 (54.9%); 11.2% yielded fluid that was paucicellular. Women 31 to 50 years of age (65%) gave cellular samples, and those older than 60 years of age yielded less fluid. Of the 825 women who yielded fluid, 237 were classified as high risk on the basis of factors such as family history and history of prior breast surgery. A significantly larger percentage of high-risk women had abnormal cytology compared with the normal-risk group. This association was seen more frequently in women older than 40 years.

Petrakis and King further validated the association of atypical epithelial cells in NAF of asymptomatic women with histologically confirmed proliferative disease and atypical hyperplasia. This finding was particularly evident among women at increased risk for breast cancer. These authors first proposed the finding of atypical cells in NAF as a possible marker to identify women at risk who would need close surveillance.

King and coworkers conducted a cytologic-histologic correlation study of NAF among 82 women with cancer and 237 women subsequently diagnosed with benign breast disease. In this analysis of 134 samples (34 cancer; 100 noncancer), patients met the criteria for the presence of a minimum of 10 cells. Atypical hyperplasia in NAF was present in 80% of cancers and 39% of benign breast diseases. The diagnosis of cancer was made in only 21% of the cases. NAF was proposed as a novel approach to study breast cancer precursors but was not considered a diagnostic tool. Factors that affect NAF production in nonlactating women include age, existing proliferative breast disease, previous breast biopsy, and family history of breast cancer.

In a large prospective trial to determine breast cancer risk in relation to NAF, Wrensch and coworkers enrolled 2701 nonlactating women between 1973 and 1980 and reported their long-term follow-up (average, 12.7 years) in 1991. In this study, women who did not yield any fluid were described as the referent group because they represent the lowest risk of developing breast cancer. With follow-up at 87%, the study confirmed that women with atypical hyperplasia were 4.9 times more likely to develop breast cancer compared with non–NAF-producing patients; this risk was 2.8 times more likely compared with women with normal cytology ( Fig. 6.6 ). The increase in the risk of breast cancer with abnormal cytology was more pronounced in younger (25–54 years) versus older (>55 years) women. Women with atypical NAF and family history of breast cancer were six times more likely to develop breast cancer than women with atypical hyperplasia but without a family history of breast cancer. The study concluded that women with cytologic diagnosis of atypical hyperplasia experienced a significant increase in risk of developing breast cancer.

In 2001, Wrensch and colleagues reported their findings on extended follow-up of 3627 volunteers. This study demonstrated that 3.5% of women developed breast cancer (median of 12 years’ follow-up), and 7.8% developed cancer in the median of 21 years’ follow-up. Women with NAF production were more likely to develop breast cancer than the non–NAF-producing group. These long-term follow-up data upheld their previous findings of increased risk for developing breast cancer related to abnormal cytology.

Sauter and colleagues used a modified breast pump to collect NAF in 177 patients, with a success rate of 94%; evaluable cytology was obtained in 95%. NAF cytology correlated with increased risk of breast cancer. Birth control pills, hormone replacement therapy, or phase of menstrual cycle had no influence on NAF production. These authors were successful in obtaining NAF in postmenopausal women and advocated the use of NAF as a screening tool and as a marker for chemoprevention.

Minimally Invasive Techniques for Determining Risk of Breast Cancer

Ductal Lavage

Although quite promising, NAF cytology has major limitations in clinical screening because of the inconsistency of adequate cellular yield and the presumption that nipple aspiration fails to harvest cells from the distal portion of the duct lobular system. To overcome these limitations, a new approach was developed in 1999 that allows collection of higher numbers of exfoliated cells from the TDLU. The DL procedure involves nipple aspiration using a modified suction cup to localize the NAF-yielding ducts. NAF-yielding ducts can subsequently be localized and cannulated using a microcatheter (Hologic, Bedford, MA). Fluid-yielding ducts are infused with normal saline. Ductal effluent collected through the microcatheter is then analyzed cytologically ( Fig. 6.7 ).

The DL technique was first attempted during a breast duct endoscopy study in which dilators were introduced into the ducts. Thereafter, a cannula (outer diameter, 0.4 mm) was inserted into the duct, and saline was injected to wash the ductal lumen. The washings were collected and studied cytologically. The technique was later refined with special development of a double-lumen catheter. Dooley and colleagues designed a prospective multicenter study to compare DL and nipple aspiration with regard to safety, tolerability, and outcomes for detection of abnormal cells. The effluent was collected, and cytologic preparations were performed using the Millipore technique (also used in early NAF studies). The study enrolled 507 women, 291 (57%) of whom had a history of breast cancer and 199 (39%) of whom had a Gail risk of breast cancer of 1.7%. The minimum age was 52 years. Overall, the procedure was tolerated well. NAF was obtained in 417 women, and DL was performed in 383. On average, the NAF had 120 epithelial cells compared with 13,200 epithelial cells in the DL samples. Greater cellularity and high diagnostic yield were also noted in DL samples. Importantly, atypia was diagnosed in 24% of the DL samples versus 10% of the NAF samples. Only ducts yielding NAF were targeted for DL (average 1.5 ducts/breast were lavaged).

Ductal lavage procedure

Preparation for the DL procedure is labor- and time-intensive; the technique requires breast massage for 30 minutes. A topical anesthetic (2.5% lidocaine and 2.5% prilocaine [EMLA]) is applied. Keratin plugs are removed from the nipple. The nipple suction device is used to obtain NAF and localize the ducts that can be cannulated. A grid can be used to record the location of the ducts for future reference. Usually 5 to 10 mL of saline are infused, collected, and submitted to a cytology laboratory in CytoLyt (Cytyc) solution. In the laboratory, a ThinPrep slide can be prepared and stained using the routine Papanicolaou stain. The diagnostic categories used for reporting DL cytology were developed by Wrensch and coworkers and were further refined using the National Cancer Institute Consensus Conference Statement on breast FNAs. The categories included the following:

• Insufficient cellular material for diagnosis: A minimum of 10 epithelial cells should be present per slide. The presence of macrophages alone does not render a specimen diagnostic.

• Benign: Samples categorized as benign contain 10 or more epithelial cells and variable number of other cellular components such as squamous cells, lymphocytes, and foamy macrophages. The benign duct epithelial cells are 10 to 15 µm in diameter and have a cyanophilic cytoplasm that may contain fine vacuoles. Ductal cells are more easily recognized in groups than in single cells, which are difficult to differentiate from lymphocytes. Cells may be arranged in monolayer sheets or strands.

• Mild atypia: In mild atypia, there is nuclear enlargement 1.5 to 3 times the normal ductal cell, and higher nuclear:cytoplasmic ratio. The nucleus may be hyperchromatic or hypochromatic, and there may be nucleoli. Multilayering or complex and papillary architectural groups may be present and usually indicate hyperplastic changes.

• Marked atypia: The preceding abnormalities are more pronounced, nucleoli are more prominent, and nuclear membrane irregularities are easily appreciable.

• Malignant cells: Malignant cells on DL have obvious features of malignancy with high nuclear:cytoplasmic ratio, hyperchromasia, macronucleoli, and dyshesion ( Figs. 6.8 and 6.9 ).

  
  

The diagnosis of malignancy on DL is uncommon, because the large majority of women who undergo the procedure have no palpable or mammographic abnormality. Occasionally, however, a high-grade DCIS may be diagnosed in this manner. In Dooley’s study, less than 1% of cases were diagnosed as malignant.

Limitations of ductal lavage

The procedure is not widely available. The cost of disposables is high. Special expertise and training is required for the performance of DL. Cytologic interpretation may be variable and training is required for pathologists to become familiar with DL cytology. DL detects milder atypia than NAF (17% vs. 6%). It remains to be seen, however, as in NAF, whether mild atypia poses an increased risk. The negative predictive value (NPV) of the procedure is not known (i.e., number of women at increased risk with negative DL who will develop breast cancer).

There are also important limitations to the nipple DL (NDL) methodology. Maddux and colleagues evaluated the atypia rate by NDL fluid-producing ducts compared with non–fluid-producing ducts and the atypia rate in high-risk versus low-risk patients. Fifty-five women with 226 ducts were lavaged, resulting in 136 ducts producing fluid versus 90 ducts producing no fluid. Of these, 44% had a Gail risk index greater than or equal to 1.7, and 56% had a Gail risk index less than 1.7%. Cytologic atypia was diagnosed in 34% of patients. The cytologic atypia rate was not statistically different in low-risk versus high-risk women based on a Gail index (33% vs. 35%; P = 1) or fluid-producing versus non–fluid-producing ducts (19% vs. 15%; P = 0.61), respectively. In this study, there was a higher atypia rate for fluid-producing ducts versus non–fluid-producing ducts (32% vs. 11%) in women at high risk (Gail risk ≥1.7).

Another limitation of NDL is the lack of studies related to the performance characteristics of the procedure. It is possible that some atypical lavages may reflect underlying atypical hyperplasia, whereas others may reflect reversible physiologic changes. Johnson-Maddux and associates proposed that persistent lavage atypia may be related to underlying pathology such as atypical duct hyperplasia, whereas reversible atypia may be associated with physiologic changes. In a reproducibility study for NDL, Johnson-Maddux and associates found marked atypia in 36% of breasts with incident carcinoma and in 24% of benign breasts ( P = 0.19). However, marked atypia was diagnosed more frequently in breasts with an incident carcinoma (22%) than in unaffected breasts (7%; P = 0.01). Interestingly, the insufficient sample rate was higher for ducts in breasts with an incident carcinoma (40%) than for ducts in breasts unaffected with carcinoma (27%; P = 0.06). Thirty-two patients with atypical lavage from the unaffected breast underwent repeat lavage at median 8.3 months, with atypia occurring in only 48%. The number of duct orifices presents a challenge in reidentification with repeat lavage. Reproducibility is also affected by variables such as hormonal changes and number of cells and cytologic interpretation. The insufficient rate on repeat lavage was 29%. The higher prevalence of lavage atypia, along with low reproducibility, limits the use of a single NDL in predicting high risk of breast carcinoma.

Ductal lavage and molecular markers

In addition to routine cytology, DL may permit analysis of molecular markers associated with breast carcinoma. Cytologic preparations of DL can be used for immunocytochemical analysis. Molecular cytogenetics analysis by King and colleagues on 39 paired cases of DL and surgically excised breast lesions revealed interphase fluorescent in situ hybridization (FISH) cytogenetic changes on chromosomes 1, 8, 11, and/or 17 in 10 of 14 (71%) malignant tumors versus 2 of 18 (11%) from benign neoplasms ( Fig. 6.10 ). This study confirms the potential utility of FISH as a future adjunctive technique. In another study by Yamamoto and associates, DL was performed in women with nipple discharge and abnormal ductography. None of the samples collected from 54 benign cases showed aneusomy for chromosomes 1, 11, and 17, giving a specificity of 100%, whereas aneusomy of at least three of these chromosomes was seen in all six malignant cases. In the future, the adjunctive use of FISH may be helpful in improving the discriminatory value of cytology.

Similarly, Evron and coworkers performed methylation-specific polymerase chain reaction (PCR), or MSP, for the cyclin D2, RAR B, and Twist genes on DL specimens. At least one of the genes was methylated in 96% of surgically excised primary tumors and 57% of DCIS, but not in normal breast tissue, resulting in high sensitivity and specificity. Thirty-seven women with biopsy-proven cancers underwent DL. On MSP, 17 of 20 women (85%) also had at least one gene methylated in the DL sample. All four women with negative DL were additionally observed to be negative by MSP. These data demonstrate the feasibility for the early detection of molecular markers in DL, which may play a role in the early detection of breast cancer.

Euhus and colleagues measured the prevalence of tumor suppressor gene methylation by quantitative multiplex methylation-specific PCR (QM-MSP) for cyclin D2, APC, HIN1, RASSF1A, and RAR-β2. Methylation of at least two genes correlated with marked atypia in univariate analysis but not in multivariate analysis that was adjusted for sample cellularity and risk group stratification. This study concluded that both methylation and marked atypia are independently associated with highly cellular sampling, Gail index, and personal history of breast cancer. On the basis of cytology and methylation profiles, Euhus and colleagues concluded that NDL ipsilateral to a breast cancer rarely retrieves cancer cells (9%). It is apparent that NDL does not seem to be suitable for early detection of focal lesions such as carcinoma, and addition of molecular biomarkers does not resolve this issue. However, NDL may be useful for detection of more diffuse risk-associated field changes in the breast. Cytologic assessment of QM-MSP analysis of NDL samples may provide an approach for identifying high-risk women for monitoring the effects of chemopreventive strategies to reduce the risk of cancer.

Ongoing investigations are exploring the proteome for unique signatures that can be exploited for diagnosis as well as risk determination. High throughput proteomic technologies such as surface-enhanced laser desorption and ionization time of flight (SELDI-TOF) mass spectrometry have been used in a number of studies to generate unique signatures expressed in cancerous breasts and normal breasts. Using SELDI-TOF to examine NAF in 12 women with breast cancer and 15 healthy control participants, Paweletz and associates identified unique proteomic patterns that were discriminatory between the two groups. Using gel-based proteomics, Kuerer and coworkers detected 30 to 202 qualitative protein expression differences in NAF in the breast with cancer compared with the contralateral breast without cancer in the same patient. Varnum and colleagues identified 64 proteins in NAF, including osteopontin and cathepsin D, which have been previously documented to vary with breast cancer status.

Frequency and Etiology of Nipple Discharge

Spontaneous nipple discharge is the chief complaint in 3% to 6% of women presenting to breast specialty services. In a series of 10,000 encounters, Seltzer reported that only 3% were the result of nipple discharge. Of those, one-third occurred in women older than 50 years. However, nipple discharge may not always be reported by the patient. Newman and colleagues found spontaneous discharge in 10% of 2685 women seen for routine examination. The great majority of nipple discharges are caused by benign conditions ( Table 6.1 ). Papillomas and papillomatosis are the most frequent causes of pathologic nipple discharge, which has been consistent over time from the earliest studies to the present. The frequency of cancer as the etiology of nipple discharge varies in reported series. This may be related to the nature of the institutional practice and whether it is heavily weighted to cancer patients. Also, in some series, the patients ultimately undergoing surgery were highly selected and, thus, would be expected to have a higher proportion of cancers. In the series reported by Florio and colleagues, only those patients with abnormal cytology had surgical resection of the duct. Cabioglu and coworkers studied patients presenting with nipple discharge to the University of Texas MD Anderson Cancer Center and found that 20% were diagnosed with cancer. The incidence of cancer in patients with nipple discharge is lower in other series. Benign papillomas and duct ectasia are more common in younger women, and cancer is more common in older women.

The clinical significance of nipple discharge and the appropriate decision-making for its management are most important in the absence of a palpable mass. Cancers infrequently present as an isolated discharge. Chaudary and coworkers reported that only 16 of 2476 (0.6%) cancers presented with isolated discharge. Devitt reported discharge associated with 2% of all breast cancers. Seltzer found that 9% of patients older than 50 years of age presenting with nipple discharge had breast cancer compared with only 1% of those younger than 50 years of age.