Ductal Lavage: What We Know and What We Don’t

Ductal lavage is a technology available to women with clinical evidence of increased breast cancer risk. It offers the potential for detecting abnormal proliferative activity in the breast via cytologic evaluation of ductal fluid retrieved from one or more nipple orifices. The primary motivation for developing this category of technology is based upon two issues: (1) the magnitude of the female breast cancer burden; and (2) the availability of strategies with documented efficacy in reducing breast cancer risk. Background Approximately 200,000 American women are diagnosed with breast cancer annually,[1] and each will face the subsequent need for disfiguring surgery and/or the potential morbidity associated with radiation therapy and/or chemotherapy. Most distressing however, is that despite multimodality treatment, all of these women must nonetheless face the lifetime possibility of disease relapse and breast cancer mortality. Through the combined strengths of screening mammography and systemic therapies we have accomplished earlier detection of disease and more effective eradication of micrometastases, yet more than 40,000 American women continue to die from breast cancer every year.[1] It is therefore appropriate that breast cancer prevention strategies have been aggressively pursued. Surgical maneuvers include prophylactic mastectomy and prophylactic oophorectomy, which offer 90% and 50% risk reduction, respectively.[2,3] Selective estrogen-receptor modulator (SERM) therapy with tamoxifen(Drug information on tamoxifen) can decrease breast cancer risk by nearly 50%,[4] and future studies are likely to define the chemoprevention effectiveness of other hormonally active medications such as raloxifene(Drug information on raloxifene) (Evista) and aromatase inhibitors.[5,6] Unfortunately, the risk-reduction benefits of these interventions are gained at the cost of facing radical surgery or premature menopause with mastectomy or oophorectomy; and uterine cancer/thromboembolic phenomena or complications of osteoporosis with SERMs or aromatase inhibitors, respectively. The potentially lifethreatening adverse sequelae of riskreducing strategies mandate that they be reserved for consideration only in women facing the greatest risk. Risk-Assessment Strategies Currently, the established, conventional risk-assessment strategy involves individualized risk estimates generated by the Gail model.[7] The Gail model is a logistic regression equation that computes the likelihood of a woman developing breast cancer within a finite period of time. This risk-assessment calculation is based on four different breast cancer risk factors derived from the study of a case-control subset of women participating in the American Cancer Society's Breast Cancer Detection Demonstration Project (BCDDP). The model has been validated as accurately predicting the number of breast cancers likely to be detected in three different cohorts of mammographically screened white American women. Among participants of the Texas breast cancer screening program,[ 8] among the Nurses' Health Study,[9,10] and among the placebocontrolled participants of the National Surgical Adjuvant Breast and Bowel Project (NSABP), the predictive accuracy rates were more than 90%. The Gail model accounts for age at menarche, parity, first-degree family history of breast cancer, and biopsy history in projecting risk of disease. Lobular carcinoma in situ is not included in this risk assessment. Some patients with hereditary predisposition for breast cancer may have their risk underestimated, because the model does not consider the extended family history of breast cancer. Furthermore, statistical evaluation of the discriminatory accuracy of the model reveals only modest strength,[10] indicating that while the model will reliably identify groups of high-risk women, it is less stable for individualized risk assessment. Other models used for breast cancer risk assessment are tailored more for the prediction of whether a woman is at risk for breast cancer associated with an inherited mutation in one of the breast cancer susceptibility genes.[11] These models typically rely on more detailed family history information, age at disease onset, and ethnicity (because of the influence of founder effects in genetic transmission of risk). Since only 5% to 10% of breast cancer is related to expression of hereditary risk, these types of models are not likely to be helpful in evaluating risk among large, unselected patient populations. Alternative measures of individualized risk assessment are therefore needed as we expand chemoprevention programs and as novel riskreduction agents are developed, requiring evaluation in future clinical trials. A pure individualized predictor would be some feature that is reliably measurable and consistently associated with increased risk. Potential candidates would include mammographic density, serologic markers, or tissue characteristics. Mammographic density has received increased attention over the past decade as a marker of abnormal proliferative changes in the breast; studies have demonstrated that certain patterns of density are indeed associated with breast cancer risk. Wolfe and coworkers[12,13] pioneered the initial studies involving risk stratification by breast density. In more recent investigations, Ziv et al[14] and Ursin et al[15] have demonstrated that mammographic density correlates with breast cancer risk among patients with familial risk, and among those of different ethnic backgrounds; Boyd et al[16,17] found that mammographic density varied with risk in association with circulating hormone levels, and that breast density is at least partially determined by inherited factors. Identification of high-risk women on the basis of mammographic density, however, is hampered by the fact that risk conferred by this feature is strongest when analyzed as a comparison feature among groups of patients: In this setting, breast cancer incidence is consistently greater in the highest-density breasts compared to the lowest-density breasts. Establishing absolute, objective values for mammographic density that are uniformly defined and reproducibly interpreted for the individual woman is more challenging. Potential molecular markers of breast cancer risk would include p53, HER2/neu, and circulating hormone levels. Unfortunately, there are inconsistencies in correlating these features with malignant potential within mammary tissue, and substantial interlaboratory variability exists in their measurement. Histopathologic Indicators of Risk Reliable, individualized, and measurable features of breast cancer risk are lacking, and this deficiency has motivated interest in the detection of histopathologic indicators of risk. Candidate markers in this category include lobular carcinoma in situ, radial scar, papillomatosis, and atypical hyperplasia of both the ductal and lobular varieties. The first three lesions are relatively uncommon; they can only be detected via open biopsy, which yields a wedge of tissue for microscopic evaluation. This leaves atypical hyperplasia as the most promising feature, as atypia can be identified cytologically as well as on histopathologic tissue analysis. Atypical Hyperplasia
Hutchinson et al[18] in 1980, and Dupont and Page[19] 5 years later, provided some of the initial data documenting the association between atypical hyperplasia and breast cancer risk. These retrospective analyses of benign breast biopsies demonstrated relative risks of 2.85 and 5.3, respectively. Wrensch et al[20] reported that women found to have atypia detected in nipple aspirate fluid had a breast cancer relative risk of 4.9. Numerous other studies looking at atypical hyperplasia have confirmed this correlation-with a relative risk averaging from 3 to 5-and have furthermore demonstrated that atypia is a risk factor for breast cancer regardless of whether it is detected in a nipple aspirate, needle biopsy, or open surgical biopsy specimen. One of the unique aspects of atypia as a risk indicator is that the subsequent breast cancer risk seems to be predominantly expressed in the 5 years following detection[21]; atypia may therefore provide some temporal measure of breast cancer risk. Atypia is present in 2% to 19% of benign breast biopsy specimens, and 15% to 25% of cancerous biopsies (Table 1). Reliance on atypia as an indicator of breast cancer risk is even more compelling in light of subset analysis from the NSABP's initial chemoprevention trial of tamoxifen vs placebo in high-risk women. Approximately 600 women in each arm of this study had a history of atypical hyperplasia, and tamoxifen exerted the most dramatic risk-reduction effects in this category; participants with atypia who were randomized to tamoxifen had an 86% lower breast cancer incidence compared to the placebo subset. Atypical hyperplasia also seems to provide more concrete evidence of breast cancer risk to chemoprevention candidates, whereas conventional measures of risk such as the Gail model do not necessarily empower high-risk women to make decisions about committing to tamoxifen therapy. Port et al[22] reported the experience of 43 high-risk women seen at the Memorial Sloan-Kettering Cancer Center. All 43 women were counseled about the benefits of chemoprevention, yet only 2 (4.7%) decided definitively to accept tamoxifen therapy. Similarly, Vogel et al[23] reported that of risk-eligible women evaluated for participation in the NSABP's current chemoprevention trial comparing tamoxifen and raloxifene, only 21% have agreed to randomization. In contrast, approximately one-third of risk-eligible women who also have a history of atypia agreed to randomization. Atypia is apparently a more compelling motivation that enables high-risk women to make difficult decisions regarding risk reduction strategies. Ductal Lavage The technology of ductal lavage was developed in the effort to easily identify high-risk women with a relatively noninvasive procedure, as it would clearly be inconvenient and of questionable efficacy for a woman to undergo surgery purely for risk assessment. Ductal lavage (performed using a topical anesthetic agent only) permits cannulation of a fluid-yielding ductal orifice with a specially designed catheter, followed by lavage with approximately 20 mL saline, and aspiration for cytology evaluation. The landmark report by Dooley et al in 2001[24] involved a comparison of direct nipple aspirates to ductal lavage specimens in series of more than 500 high-risk women. This study revealed that ductal lavage was 3.2 times more likely to produce cytologically evaluable fluid compared to nipple aspirates; a more robust cellular specimen was also retrieved with the lavage specimens. The median number of cells retrieved per lavage specimen was 13,500, compared to 120 per breast in the nipple aspirates. The Dooley et al[24] study also demonstrated that ductal lavage is unlikely to detect cancer; even among this high-risk cohort of patients, fewer than 1% had frankly malignant cells on cytology. At best, the procedure will only evaluate cells from a portion of the ductal system that has been cannulated. It is therefore unlikely to provide sufficient information regarding the entire breast for it to be of much value as a screening modality. As a risk-assessment adjunct, however, it is conceptually much more promising. One goal of a risk-assessment program is to reliably identify women who have evidence of abnormal proliferative activity; intervention with a chemoprevention agent at this stage theoretically will either stabilize or reverse this sequence and avoid eventual carcinogenesis. Assuming that the production of nipple fluid in a nonlactating woman is one indicator of abnormal proliferative activity, then it is biologically plausible that analysis of cells shed into the fluid-yielding duct would be the most likely site for the detection of atypia. Limitations
Despite the apparent and presumed benefits of ductal lavage in refining the evaluation of high-risk women, several limitations must be acknowledged. First, the very premise upon which ductal lavage bases its riskassessment value-the detection of atypical cells in the lavage fluid-has yet to be proven by prospectively collected data. As noted previously, atypia has consistently been associated with a three- to fivefold relative risk for breast cancer when detected in any of the various interventions-nipple aspirates, needle biopsy, and open biopsy. Although it seems reasonable to assume that atypia identified on lavage cytology would confer a similar magnitude of risk, this has not been confirmed. As cohorts of women undergoing the lavage procedure are prospectively followed, a quantifiable appraisal of this risk should become available. A second issue regarding ductal lavage that will require further investigation is that of the reproducibility of its cytologic analyses. Bonnie King participated in the cytopathology review for both the Dooley et al[24] study of ductal lavage and the Wrensch et al[20] study of nipple aspirates, providing inferential evidence that atypical cells retrieved by both interventions were well-standardized. Nonetheless, there is established precedent confirming the subjectivity and variation that can exist in histopathologic assessment of borderline breast lesions,[25] and one might reasonably infer that differences in the cytologic evaluation of ductal lavage specimens might also exist. It would be a reasonable endeavor to perform studies assessing the interlaboratory reproducibility of ductal lavage analyses. There are many potential areas where ductal lavage might be a valuable risk-assessment adjunct, as women belonging to several categories of risk might be interested in a procedure that could potentially document their immediate level of risk. Breast cancer risk associated with exogenous hormones and therapeutic chest wall irradiation are two examples. The Women's Health Initiative[26] has documented the increase in breast cancer risk conferred by prolonged postmenopausal hormone replacement therapy. However, postmenopausal women also face increased risk of the thromboembolic- and uterine tumorigenesis- related morbidity from chemoprevention with tamoxifen. Irradiation of breast tissue during adolescence and early adulthood, as experienced by many young women treated for Hodgkin's disease, is followed by a substantially increased breast cancer risk.[27] These radiation- related breast tumors typically occur approximately 15 to 20 years later, when these women are often still premenopausal. While tamoxifen has a more favorable risk/benefit profile in women under the age of 50 years, it is not approved during pregnancy and therefore cannot be used by women who are contemplating childbearing. Although the detection of atypia via ductal lavage in these scenarios may facilitate decisions regarding chemoprevention, the yield and efficacy of the ductal lavage procedure have not been evaluated in these specific categories of patients. The correlation of ductal lavage findings with tumors within a known cancerous breast is also uncertain and warrants further study. Khan et al[28] reported ductal lavage findings when the procedure was topographically mapped in relation to a breast cancer by injecting a gelatinous substance into the lavaged ductal system of mastectomy specimens. This investigation found that the tumor was located in the lavaged segment for only twothirds of cases. Dooley et al[29] performed ductal lavage in the contralateral breasts of women undergoing breast cancer surgery and detected atypia in 32%, 22%, and 6.7% of T1a/b, T1c, and T2 lesions, respectively. Collectively, these data confirm that ductal lavage is unlikely to be a worthwhile screening test for breast cancer. Early pathology studies of atypia demonstrated that breast cancer risk tends to return to baseline after approximately 5 years if no other riskrelated events intervene. If a cancer arises within a field of high-risk breast tissue, then surrounding (including contralateral) breast tissue that previously harbored atypical hyperplasia may continue to undergo regression in accordance with the data, indicating a return to baseline risk 5 years after the atypia has been detected. The cancerous lesion may progressively lose an association with atypia over time, and will not necessarily remain associated with a fluid-yielding ductal system. Furthermore, breast cancer pathogenesis is likely to be heterogeneous; not every area of ductal atypia is committed to progressing to invasive cancer if left untreated, and conversely, some cancers may arise without having passed through an atypical hyperplasia precursor phase.[30] Treatment Options for Abnormal Results
Morrow et al[31] have presented a comprehensive schema describing the options for management of women following an abnormal ductal lavage. Many such patients will opt for chemoprevention with tamoxifen. Follow- up lavage 6 to 12 months later would appear reasonable. However, there are no prospective data available regarding the yield and/or the significance of sequential ductal lavage findings. At least one potential concern might be the possibility of variation in lavage cytology related to the menstrual cycle. While serial lavage results have not been reported on any large patient cohorts, extrapolation from the direct nipple aspirate literature offers some encouraging insight that ductal fluid findings are independent of menstrual cycle phase. Mitchell et al[32] performed weekly nipple aspirates on 15 premenopausal volunteers, and over the course of two menstrual cycles no significant differences in cellular profile were detected. A diagnostic dilemma is created in the rare circumstance of frankly cancerous cells detected within a ductal lavage specimen. Since the ductal lavage procedure should only be performed as a risk-assessment adjunct for a woman who is without any evidence of a preexisting cancer, these would presumably be cases associated with a negative mammogram and clinical exam. Nonetheless, it would be worthwhile to repeat the mammographic work-up of the affected breast in this scenario. Confirmation of the lavage findings by second opinion cytopathology review is also essential. Other maneuvers might include repeating the lavage procedure in conjunction with ductography, wholebreast ultrasound, and even breast magnetic resonance imaging. If all studies are negative for identifying a source of the cancerous cytology, then a blind terminal duct excision might also be considered. The disadvantage of this latter approach is that the tumor could easily be located peripheral to the subareolar ductal system. Once the terminal duct apparatus has been divided and resected, the option of repeat cannulation and ductal lavage will be lost. Patients who decide to undergo prophylactic mastectomy should be forewarned of the possibility that identification of a primary tumor mass in the breast may be difficult, if not impossible. Lastly, the cost-efficiency of ductal lavage is uncertain at this time. Third-party payors have not established any uniform reimbursement policy for the procedure, and patients considering this strategy for risk assessment must understand that significant out-of-pocket expenses may be incurred. Access to the ductal lavage technology is therefore clearly not universal. Translational Research
Having considered the limitations of ductal lavage, it is also worthwhile to discuss the potential for incorporating the procedure into translational research. Several lines of research are currently under way to evaluate the feasibility of these approaches. Evron et al[33] have studied the possibility of detecting cancer cells in ductal lavage fluid by methylation-specific polymerase chain reaction to study cyclin D2, RAR-beta, and Twist. Investigations of direct nipple aspirates have been successful in measuring basic fibroblast growth factor,[34] carcinoembryonic antigen,[35] and HER2/neu[36] as indicators of breast cancer. Extension of these studies to analyze ductal lavage fluid is expected, and its potential value is reviewed by Klein and Lawrence.[37] The evolution of microarray technology for the analysis of DNA content and "genetic profiling" has added another layer to the sophistication and complexity of tissue research.[38] The prospect of applying these strategies to cells retrieved from ductal lavage is provocative. It will be important to consider the influence of lavageassociated, nonepithelial cytology components on these investigations. In particular, the substantial contribution of macrophage-derived mammary foam cells to the cellular content of nipple aspirates as well as lavage fluid has been documented by Krishnamurthy et al[39] and King et al.[40] These foam cells may affect the observed microarray patterns if DNA is retrieved from total cellular content. Conclusions In summary, we can catalog the known and unknown issues related to the history and possible future of ductal lavage. The following has been documented:

• The magnitude of the breast cancer burden among American women is substantial.
• Breast cancer risk-reduction strategies are available, but are associated with potential morbidity from adverse effects.
• Existing risk-assessment methods have limitations, and alternative, individualized risk assessment strategies are needed.
• Atypia consistently and reliably identifies one category of high-risk women, and is particularly sensitive to the antiproliferative effects of chemoprevention agents such as tamoxifen.
• Atypia can be detected on fineneedle aspiration, core biopsy specimens, tissue specimens, nipple aspirates, and ductal lavage cytology.
• Ductal lavage specimens are more likely to yield cytologically evaluable fluid compared to direct nipple aspirates.
• Ductal lavage is not a breast cancer detection/screening modality. The following questions remain:
• Does atypia within ductal lavage specimens confer the same magnitude of future breast cancer risk compared to atypia detected via other sources?
• Is the cytologic interpretation of ductal lavage reproducible, or is there interlaboratory variation?
• Can ductal lavage correct the deficiencies of existing risk-assessment methods and improve the individualized risk assessment of non- white American women, and women exposed to chest wall irradiation or prolonged postmenopausal hormone replacement therapy?
• Is ductal lavage cost-effective?
• Is the lavaged ductal system likely to be the highest-risk area of the breast?
• What is the significance of cancer cells detected in lavage fluid?
• What is the significance of serial ductal lavage studies?
• Will cytology from lavage specimens change over time and in accord with risk-reducing interventions?
• Are there other applications for ductal lavage, such as the detection of risk-associated proteins, or the collection of cells that may be studied with microarray technology?