Proliferative epithelial disease identified in nipple aspirate fluid and risk of developing breast cancer: a systematic review

Abstract

Background:

Guideline panels recognize the need to increase the accuracy of identifying women at high risk of developing breast cancer who would benefit from prevention strategies. The characterization of proliferative epithelial disease found in nipple aspirate fluid (PED-NAF) may be a relevant risk factor.

Objective:

To comprehensively review the published literature to characterize and summarize abnormal cytology detected by NAF and the association of PED-NAF with subsequent risk of developing breast cancer.

Research design and methods:

Literature identified by systematic searches in MEDLINE PubMed and the Cochrane Library was screened for articles containing primary data on NAF cytology based on predefined inclusion and exclusion criteria.

Main outcome measures:

Study characteristics, cytological group distribution, and incidence of breast cancer.

Results:

Thirty articles were included after full-text review, of which 16 were analyzed, containing data on 20,808 unique aspirations from over 17,378 subjects. Seven (44%) of the studies used the King cytological classification system. Among aspirations from women free of breast cancer, 51.5% contained fluid, in which over 27.7% had PED on cytology. In the two prospective studies of 7850 cancer-free women, abnormal cytology by NAF carried a 2.1-fold higher risk (95% CI, 1.6–2.6; p < 0.001) of developing breast cancer, compared with women from whom no fluid could be obtained.

Conclusions:

PED-NAF among women free of breast cancer, compared with no fluid being obtained, has an independent risk of developing breast cancer comparable to the risk of a woman with a positive family history of breast cancer. These findings have implications for augmenting risk prediction and clinical decisions concerning breast cancer surveillance and chemoprevention. As with all reviews, heterogeneity across studies may have influenced the results. The limited literature calls for prospective studies on asymptomatic women with long-term follow-up.

Keywords: :

• Atypical ductal hyperplasia
• Breast cancer
• Cytopathology
• Early detection
• Nipple aspirate fluid
• Proliferative epithelial disease

Introduction

According to the World Health Organization (WHO), breast cancer ranks as the most frequently diagnosed cancer in women, comprising 16% of all cancers in women1. The American Cancer Society (ACS) estimated that there would be 232,670 new invasive breast cancer cases diagnosed among US women in 2014, not including an estimated 62,570 new cases of in situ breast cancer, of which approximately 83% will be ductal carcinoma in situ (DCIS)2. The number of deaths from breast cancer among US women was estimated to be 40,000 in 2014, accounting for 15% of all cancer-related deaths2. The 5-year relative survival for women with localized breast cancer is 99%, declining to 24% for those with distant stage cancer2. Cancer care costs are a financial burden to patients, their families, and society as a whole3^,4.

The National Comprehensive Cancer Network (NCCN) guidelines recommend regular screening and risk-factor assessment based on complete medical history and clinical breast examination for women without any signs or symptoms of breast cancer from the age of 25⁵. Risk factors for developing breast cancer include older age, younger age at menarche, older age at first live birth or nulliparity, later onset of menopause, prior history of breast cancer, thoracic irradiation, or lobular carcinoma in situ, increasing number of first-degree relatives with breast cancer, and presence of atypical hyperplasia in a previous breast biopsy5. The risk of developing breast cancer is higher in women with dense breasts, but there is insufficient evidence to support routine supplemental screening in women who have this risk factor but no other risk factors5. Proliferative epithelial disease (PED), defined as ductal hyperplasia and atypical ductal hyperplasia, is a group of conditions marked by an increase in the growth of epithelial cells in the breast6. Mammary epithelial cells line the ductal/lobular system of the breast milk ducts and constantly secrete and absorb nipple aspirate fluid (NAF)7. Increased risk of cancer is associated with excessive growth and exfoliation of the epithelium, and as a result NAF will contain exfoliated epithelial cells from the ductal/lobular system. Nearly all breast cancers arise from mammary ductal and lobular system epithelial cells8. PED identified via histopathological examination of specimens obtained by needle or surgical biopsy has been shown to be associated with risk of developing breast cancer9–11. Papanicolaou et al. discussed the potential implications of exfoliative cytopathology in NAF specimens of the mammary gland in 195812. Since then King, Petrakis, and other researchers have introduced nipple aspiration as an alternative specimen collection method to assess the presence of PED by cytology13–15. Collection of NAF specimens involves use of devices that create negative pressure, usually with a suction cup or a breast pump16.

The primary objective of this study was to comprehensively review the published literature to characterize and summarize: (1) the epidemiology of abnormal cytology detected by NAF, and (2) the association of PED detected by NAF cytology with subsequent risk of developing breast cancer.

Methods

Search strategy and study selection

This systematic search was conducted in MEDLINE PubMed (National Center for Biotechnology Information, US National Library of Medicine) and the Cochrane Library. The search terms were based on the MeSH terms of key phrases from the research question, and were specific to three categories: site/specimen, method, and outcomes. The search was limited to articles published in English language on human subjects, until October 2013 (Table 1).

Table 1. Search terms.

No.	Search terms^a	Category	Hits
1	breast OR mammary OR nipple	Site/Specimen	376,482
2	aspirate* OR (aspirate fluid) OR (aspirate specimen)		23,002
3	1 AND 2		1716
4	cytolog* OR cytopatholog*	Method	863,772
5	epitheli* OR exfoliat*		375,257
6	4 OR 5		1,154,066
7	risk factor* OR epidemiolog*	Outcomes	1,695,774
8	hyperplasia OR ductal OR patholog* OR disease* OR cancer OR neoplasms OR precancerous		6,269,714
9	mortality OR death OR die		1,204,252
10	8 OR 9		6,817,466
11	inciden* OR prevalen* OR rate OR risk		3,385,485
12	10 AND 11		1,620,320
13	7 OR 12		2,634,741
14	3 AND 6 AND 13	Results	262
	Limitations: Humans, English, published by October 2013		241

^aAsterisks represent truncation in PubMed search. For example, a search with “aspirate*” will capture all terms that have the root “aspirate”.

The title/abstract review of the retrieved publications was conducted by two analysts independently, based on predefined inclusion and exclusion criteria. Studies that reported primary data on the cytopathology of NAF and/or risk of developing breast cancer were included. To comprehensively review the literature, studies of all designs and on subjects with all disease status (asymptomatic, free of cancer, with cancer) were included.

Studies were excluded if they had any of the following characteristics: (1) irrelevant research question; (2) review articles or letters to editor without primary data or with only qualitative data; (3) collection of cytological specimens that are not nipple aspirate fluid, e.g., breast masses/lumps, breast cyst fluid, nipple discharge, or from sites other than breast; (4) use of fine-needle aspiration (FNA), ductal lavage, etc.; (5) absence of cytological analysis. Citation search and reference review were conducted to identify additional studies not captured in the original search. Full-text articles were then retrieved and assessed for eligibility, with a few publications further excluded from data extraction.

Data extraction

Data from the included studies were extracted using the Data Extraction Sheet (DES) built in Microsoft Excel 2007 version. The data fields were developed and then tested with 10 included publications for assessment of validity and comprehensiveness. The DES consisted of four data categories: study overview, subject characteristics, specimen description and cytology, and cancer outcomes by cytology, each of which included multiple data fields.

Double data extraction was conducted by two independent trained analyst groups. The principle investigator performed data verification to ensure data extraction quality and consistency in reporting methods. Information on study characteristics such as study design, number of subjects, subject characteristics, methods used to collect NAF, and methods used to establish breast cancer status was extracted from full-text articles. Study outcomes were assessed, including distribution of cytological groups (overall and by risk factors) and the number of breast cancer cases in each group, if reported. The commonly reported risk factors included age of study participants at enrollment, menopausal status, age at menarche, and family history.

Statistical methods

The primary endpoints of this analysis were the incidence of the NAF cytological groups, the risk of breast cancer for each group, and the relative risk (RR) of developing breast cancer between the cytological groups. To evaluate the association between cytological abnormalities in NAF and breast cancer, invasive carcinoma and DCIS were both considered as cancer cases. Association measures and the related statistics (95% confidence intervals and p-value) were recorded from each included study. To generate a meaningful estimate of RR of developing breast cancer between different NAF cytological groups, results were summarized by cytological classification system cited by study authors and subject disease status. The pooled analysis of RR was conducted with STATA 9.2 version (Stata Corp., College Station, TX, USA), which combined results from studies of similar design. The fixed-effect model was used and the studies were weighted using inverse-variance methodology.

Results

The systematic search identified 241 articles in PubMed, of which 25 studies met the inclusion criteria for full-text review through the title/abstract screening process. Seven additional articles were identified through citation search and reference review. The independent search in the Cochrane Library using the same search terms identified no new articles. Two articles were excluded after full-text review due to their irrelevance to the objective of this analysis. Subgroup analyses and updates on the previously studied populations were identified through original study method descriptions, provided references, or direct confirmation by the study author. One most informative article was selected to represent the cohorts that were studied more than once to avoid double-counting the study subjects, leaving 16 articles for analysis of incidence of cytological groups (Figure 1)12^,17–31. Eight of the 16 articles reported information on cancer outcomes (Table 2)12^,17^,18^,23^,29–32.

Figure 1. Flow diagram of the systematic search and screening. Updating reports or subgroup analyses were excluded from analysis but discussed along with the results from the representative publication of the corresponding populations. The 7 articles identified by citation search and reference review included Buehring38, Filassi20, King (Part II)22, Krishnamurthy23, Papanicolaou12, Petrakis (Environ Health Perspect)36, and Proctor25. FNA, fine-needle aspiration.

Table 2. Association of cytological finding and carcinoma outcomes.

Study	Study description	Cytological Subgroup	Number of subjects (% of total)	% with breast carcinoma	Association measure^a and 95% CI	p Value
King classification
Wrensch et al.30	Long-term follow-up of subjects free of cancer at study entry and within 6 months of study entry (Group 1 from breast screening center, variety of clinics and health fairs; Group 2 from breast and other clinic patients and UCSF employees)	No NAF	2775 (40.2)	3.7	RR^b (referent)
		Scant cells	453 (6.6)	7.5	1.2 (0.8–1.7)	–
		Normal cells	2627 (38.1)	6.6	1.3 (1.0–1.7)	–
		Hyperplasia	940 (13.6)	8.2	2.0 (1.5–2.8)	–
		Atypia	109 (1.6)	11.0	2.1 (1.1–3.9)	–
Baltzell et al.17		Long-term follow-up of cohort seen by Sartorius at SB breast clinic (free of cancer at initial visit and within 6 months of initial visit)	No NAF	714 (75)	9	OR^c (referent)
	Scant cells		19 (2)	11	1.4 (0.32–6.4)	0.64
	Normal cells		89 (9)	12	1.7 (0.87–3.5)	0.12
			Hyperplasia or atypia	124 (13)	14	2.0^d (1.1–3.6)	0.02
Sauter et al.32^e	Cross-sectional comparison of NAF from mastectomy against histological results (breasts having undergone excisional biopsy demonstrating DCIS or IBC; mastectomy performed after excisional biopsy)	Scant cells or no atypia	57 (79.2)	43.9	–	<0.001
		Atypia or malignant	15 (20.8)	93.3	–
Papanicolaou classification
Buehring et al.18	Long-term follow-up of subjects that were mostly asymptomatic, self-selected by responding to appeals through newspapers and television, posted announcements, and oral presentations	No NAF	517 (53.2)	11.6	RR^f (referent)	≤0.005
		No cells	316 (32.5)	11.0	1.05 (0.71–4.57)
		Any cells	139 (14.3)	18.3	1.92 (1.22–3.01)
Papanicolaou et al.12		Cross-sectional comparison of NAF and clinical & pathological findings. Three groups: (1) secretion smears from women without breast complaints; (2) secretion smears obtained at the breast clinics from patients with symptoms of breast disease; (3) secretion smears from one breast in patients who either had carcinoma present in the other breast or had had unilateral mastectomy for carcinoma	No NAF	1703 (55.0)	0.4	–	–
	Class I		534 (17.3)	0.6	–	–
	Class I–II		506 (16.4)	1.8	–	–
	Class II		278 (9.0)	2.9	–	–
	Class CD		39 (1.3)	10.3	–	–
	Class III		15 (0.5)	66.7	–	–
	Class IV		4 (0.1)	100.0	–	–
	Class V		15 (0.5)	100.0	–	–
The NCI FNA criteria
Krishnamurthy et al.23	Cross-sectional comparison of NAF from affected breasts and histopathologic findings in women with biopsy confirmed intraductal or invasive carcinoma with or without a history of preoperative neoadjuvant chemotherapy	Scant cells	30 (40.5)	0.0	–	–
		Normal cells	34 (45.9)	85.2	–	–
		Mild atypia	5 (6.8)	20.0	–	–
		Marked atypia	1 (1.4)	100.0	–	–
		Malignancy	4 (5.4)	100.0	–	–
Other cytological classification system
Zimmerman et al.31	Cross-sectional comparison of NAF and histological diagnosis	No epithelial groups	1552 (41.2)	0.5	–	–
		Epithelial groups	2218 (58.8)	1.2	–	–
Cytological classification system not reported
Sauter et al.29	Cross-sectional comparison of NAF and histopathologic results in women with a suspicious breast lesion identified on an imaging study and/or a solid palpable breast mass, prior to scheduled excisional breast biopsy	Scant cells	107 (70.4)	31.8	–	–
		No atypia	39 (25.7)	48.7	–	–
		Atypia	4 (2.6)	50.0	–	–
		Malignant	2 (1.3)	100.0	–	–

^aOdds ratio (OR) or risk ratio/relative risk (RR) reported by the studies. If no direct report, the cell was left with a hyphen.

^bAdjusted by Cox regression for age, age squared, and year of specimen collection.

^cAdjusted for age at the time of the NAF visit.

^dUsing the equation RR = OR/(1 − ACR × (1 − OR)), the corresponding RRs are 1.4 for ‘scant cells’, 1.6 for ‘normal cells’, and 1.8 for ‘hyperplasia or atypia’. ACR, assumed control risk.

^eThis study analyzed residual cancer in mastectomy specimens with histological review.

^fAdjusted for age and length of time in study as continuous variables.

NAF, nipple aspirate fluid; DCIS, ductal carcinoma in situ; IBC, invasive breast cancer; NCI, National Cancer Institute; FNA, fine-needle aspiration.

Cytological classification systems

The cytopathological assessment of NAF classifies the specimens based on the type of cells present and their morphological characteristics. The first cytological classification system was developed by King, Petrakis, and colleagues in 1975 and it was further modified in 198314^,17^,18. Other classification systems include the classification adapted from the NCI criteria for breast fine-needle aspiration biopsy (NCI FNA criteria)33, the classification defined by Papanicolaou and colleagues12^,34, and others35. Classification systems vary in how they define their cutoff points for each cytological group. Table 3 shows the most commonly used King classification system in which cytology is classified as benign, hyperplasia, atypical hyperplasia, or malignant.

Table 3. Cytological classification system (King)^a.

Cytological group	Description
No NAF	Nipple aspiration attempted, but fluid not obtained.
NAF/scant cells	Specimen obtained, but unsatisfactory for cytological diagnosis, e.g., less than 10 cells in specimen or unacceptable technical quality.
NAF/benign	Duct epithelial cells within normal limits; foam cells; apocrine metaplastic cells.
NAF/hyperplasia	Minimal changes, slight cell and nuclear enlargement; cell distribution predominantly in groups; papillary grouping – a subcategory; apocrine type – a subcategory.
NAF/atypia	Moderate-to-severe abnormality, distinct nuclear enlargement, increasing N:C alteration and chromatin change; cell distribution in groups and papillary formations; single cells increase in numbers with increasing degree of abnormality; apocrine type – a subcategory.
NAF/malignancy	Single cells and groups of cells with unequivocal nuclear features of cancer.

^aTable adapted from King et al. Nipple aspirate cytology for the study of breast cancer precursors. J Natl Cancer Inst 1983;71:1115-2140, with permission. Content in JNCI published prior to 1997 is in the public domain for reuse.

NAF, nipple aspirate fluid; N:C, nuclear-to-cytoplasmic ratio.

Study characteristics

Seven of the 16 articles (44%) based their classification of NAF on the King system17^,22^,25–28^,30, among which two articles used a modified version of King27^,28 that combined two of the original groups to form an aggregated group. The next most cited systems were the NCI FNA criteria and the Papanicolaou classification, cited by three19^,23^,24 and two12^,18 articles, respectively. The rest of the articles either used other classification systems20^,31, or did not provide any specific information about the source of the classification system21^,29.

The baseline characteristics of the included studies are shown in Supplemental Table 1. The studies included were observational in nature, with the majority being cross-sectional studies that compared cytological findings to histopathological findings at one specific point in time. The sample sizes ranged from 12 to 6904, totaling 20,808 specimens from over 17,378 subjects. The average follow-up time for prospective studies was 17 years (range 9–25). Most of the studies obtained NAF using the method or device similar to that first introduced by Sartorius et al.26. Studies varied in subject inclusion criteria and methods to establish and confirm cancer status. None of the studies provided quantitative information on the number of epithelial cells required to support PED.

Incidence of cytological groups

The distribution of cytological groups from individual studies is shown in Supplemental Table 2. In general, the studies reported the presence or absence of NAF based on a simple visual inspection of the nipple following attempted aspiration. This may have prevented detection of smaller NAF specimens that were too small to visualize. Across studies that reported data on NAF availability, considerable heterogeneity existed in the percentage of women from whom fluid was obtained (mean 43.5%, range 25.0%–94.1%). Half (51.5%) of the aspiration attempts in cancer-free breasts (asymptomatic, breast complaints, or benign breast diseases prior to NAF collection) got fluid specimens. The incidence of cytological PED within the fluid specimens was 27.7%. The asymptomatic subgroup obtained fluid specimens in 36.0% of breasts and a 20.5% incidence rate of PED. In contrast, almost all studies that enrolled subjects having a current diagnosis or personal history of breast cancer did not provide data on subjects from whom NAF was not obtained. Presence of abnormal cytology increased to 34.1%. The four most common risk factors for breast cancer were examined for their association with cytological findings in breast fluid, among which family history (number of first-degree relatives with breast cancer) was shown to be associated with the degree of cytological abnormalities by three studies25^,36^,37. When age was considered, abnormalities or just the presence of NAF seemed to be relatively more frequent in adult women under the age of 54, a trend seen in four different studies12^,18^,26^,37. Similarly, percentage generating NAF was higher in pre-menopausal than in post-menopausal women12^,38. Age at menarche was not associated with cytological group distribution (Supplemental Table 3).

Association with breast cancer

Seven (87.5%) of the eight studies reported information on cancer outcomes. As the severity of cytological abnormality increased, so did the risk of having breast cancer at the beginning of the study (if women with breast cancer were enrolled) or developing cancer later (if asymptomatic women or women with benign breast diseases were enrolled) (Table 2).

Cross-sectional (3) and prospective studies (5) were analyzed separately. The cross-sectional comparisons confirmed that cytological abnormalities in NAF were associated with histopathological findings12^,23^,29^,31^,32. Atypical and malignant cytology was especially predictive of histopathologically confirmed carcinoma. In two of the prospective studies that included women free of cancer and used the King classification system, the RR of developing breast cancer associated with cytological PED, compared with no NAF, was 2.1 ([95% CI, 1.6–2.6], adjusted for age, age squared, and year of specimen collection) in a cohort of 6904 women30 and odds ratio (OR) of 2.0 ([95% CI, 1.1–3.6], adjusted for age) in a cohort of 946 women17. The pooled RR generated by STATA for these two prospective studies involving 7850 women was 2.1 ([95% CI, 1.6–2.6; p < 0.001], unadjusted), within an average follow-up of 17 years. The pooled RR between PED-NAF and normal cytology was 1.27 ([95% CI, 1.01–1.60; p < 0.05], unadjusted), and that between PED-NAF and NAF without PED (including scant cells and normal cytology) was 1.25 ([95% CI, 1.00–1.56; p = 0.053], unadjusted). Increased risk of developing breast cancer was also observed when comparing PED-NAF to NAF with scant cells, but the results were not significant. In another prospective study of 1681 women that were mostly asymptomatic with a 25.5-year follow-up that used the Papanicolaou classification system, the RR of developing cancer when NAF was obtained with epithelial cells versus no NAF was reported to be 1.92 ([95% CI, 1.22–3.01; p ≤ 0.005], adjusted for age and length of time in study)18. Assessment of the association among asymptomatic women would be more informative in clinical practice, but none of the three prospective studies separated this subgroup from the subjects who were free of breast cancer at enrollment.

It is also note-worthy that the subgroup analyses of the Wrensch et al. 2001 study cohort (n = 6904, enrolled between 1972 and 1991) showed an association between abnormal cytology in NAF with various breast cancer risk factors36^,37^,39–46. In 2005, Tice et al. conducted a post-hoc analysis which used proportional hazards modeling to recalculate the coefficients for the predictor variables used in the Gail model, and showed that including NAF cytology improved the predictive ability of the Gail model47. NAF cytology was proven to improve the risk stratification among women with increased risk of developing breast cancer by traditional risk factors (e.g., aged 25–54 years, having first-degree relative with cancer, having prior biopsy)37^,46.

The Buehring et al. study evaluated the usefulness of breast exfoliative cytology in screening asymptomatic women for breast atypias in a manner that might ultimately be applicable to routine physical examinations in a physician’s office38 and a follow-up study determined the incidence of breast cancer during the 25 years since enrollment18. Both these studies collectively suggest that the use of NAF for cytological examination may be a useful tool for assessing subsequent breast cancer risk.

Discussion

This is the first systematic review of the published literature on the epidemiology of cytological proliferative epithelial disease identified by nipple aspirate fluid, and its associated risk of developing breast cancer. Among the eight studies that examined the association of cytological PED and breast cancer risk, three were prospective cohort studies from which meaningful inferences about risk could be drawn. The other eight studies provided useful information on the proportion of women with specific types of cytological abnormalities at time of collection and evaluation of NAF.

The three prospective studies consistently identified increased risk of developing breast cancer related to presence of cytological abnormalities in NAF17^,18^,30. In the two studies that cited the King classification system, the detection of either hyperplasia or atypical hyperplasia was associated with a two-fold risk of developing breast cancer compared with no NAF. Unsatisfactory specimens with scant cells correlated with a slightly increased risk (range 1.2–1.4) versus no NAF, and normal cytology indicated a relative risk of 1.3–1.6. The cytology–cancer association, when cytology is assessed using the Papanicolaou classification system, was only available for comparison of cellular specimen versus no NAF and non-cellular specimen. The presence of cells versus no cells also was associated with an almost two-fold risk of developing breast cancer.

The literature on NAF provides limited evidence on the difference in risk between typical and atypical hyperplasia, which contrasts the findings from studies of surgical excision and/or biopsies, where the relative risks were <2 for typical hyperplasia and ≥3.7 for atypical hyperplasia10^,11^,48. The main reason for the difference is the characteristics of the study populations. Studies of NAF collection targeted women free of breast cancer, most of whom were likely to be asymptomatic, while women enrolled in studies of invasive biopsy had breast signs or symptoms in which biopsy was deemed appropriate to their medical care. Although difference exists between PED on NAF cytology and on surgical biopsies, the correlation between them has been well documented12^,14^,22.

The relative risks of developing breast cancer associated with PED cytologically identified in NAF were comparable to or higher than those of several known risk factors included in the Gail model. Wrensch et al. reported that hyperplasia (RR = 2.0) and atypical hyperplasia (RR = 2.1) were stronger independent predictors for future breast cancer than age at first pregnancy ≥25 (RR = 1.3), nulliparity (RR = 1.6), age at menarche ≤12 (RR = 1.4), history of biopsy (RR = 1.9), and being white (RR = 1.2). The relative risks associated with PED were also higher than having one first-degree relative with breast cancer (RR = 1.6), and were only lower than that associated with having two or more first-degree relatives with breast cancer (RR = 2.8)30^,47. Similarly, in the 25-year prospective study by Buehring et al., presence of epithelial cells in NAF featured a comparable relative risk to age and family history, stronger than that associated with race, parity, or age at menarche18. These authors suggested that “a possible clinical application of this is routine nipple aspiration and NAF cytology during general physicals and gynecologic examinations of premenopausal women.”

Most studies assessed NAF specimens only once per enrolled woman and did not indicate whether the remaining specimens were stored for re-analysis. One study explored within-subject longitudinal changes of NAF cytology. Petrakis et al. assessed 24 women pre- and post-consumption of soy protein isolate on breast secretion, showing a 25.0% increase in women with hyperplasia in NAF (pre, 4.2%; post, 29.2%)43. Petrakis et al. also found a significantly higher incidence of cytological PED among US-born Asian women compared with contemporaneously studied immigrant Asian women45. Together, these studies suggest PED incidence may change over time and may be associated with dietary intake.

Tremendous heterogeneity across the included studies may have led to bias (Figure 2). Studies differ in design and devices used to collect NAF. The definition of NAF producer status also varies – some studies used the qualitative threshold of ‘fluid appearance’ while others applied quantitative measures requiring 1 µL of NAF or larger amounts. To avoid misinterpretation, we did not perform statistical pooling of the study findings to understand the predictive power of each risk factor. Instead, we summarized the results of each included study, and highlighted common associations and trends across similar studies to provide the most comprehensive assessment of the literature. Conclusions on cancer risk were based on the three studies that were prospectively designed and had long follow-up periods. Most studies were conducted in community based settings, and so may be more generalizable to routine practice than if studies had been conducted predominantly in specialized referral centers.

Figure 2. Assessment of risk of bias. Sartorius26 reported two studies, thus there were 17 studies from 16 articles.

Despite the relative merits of the existing longitudinal studies, future prospective research with long follow-up periods in well defined cohorts is needed. This effort would be valuable to further estimate the strength of the association demonstrated by current evidence and refine the risk assessments for hyperplasia and atypical hyperplasia separately. Of particular value are prospective cohort studies that enroll only asymptomatic women, stratify participants by disease status and traditional risk factors, and have longer follow-up periods to assess incidence of breast diseases and cancer by NAF status without combining PED subgroups.

Guidelines continue to acknowledge the need for more accurate risk stratification for developing breast cancer, especially to assist in early detection in order for chemoprevention. The US Preventive Services Task Force (USPSTF) recommends chemoprevention strategies for women with increased risk of breast cancer and low risk for adverse events49. According to the NCCN guideline, breast cancer risk-reducing agents (e.g., tamoxifen, raloxifene) are only recommended for women aged 35 years or older who have a 1.7% or greater risk of developing breast cancer as determined by the modified Gail model50. Both the USPSTF and NCCN recognize atypical hyperplasia as an independent risk factor that should be considered in risk stratification algorithms.

The studies summarized herein show an independent and significant effect of abnormal cytopathology in NAF, which may be useful for updating risk estimates as specimen collection technologies undergo review by the US Food and Drug Administration. For example, among women who have a risk score close to, but less than, the critical threshold of 1.7%, detection of PED by NAF cytology shows promise as an independent and relevant predictor of breast cancer risk. As per Buehring et al.’s suggestion, routine nipple aspiration may be incorporated in general physicals and gynecologic examinations. Women at different levels of increased risk identified through NAF cytopathology could then be managed differently than those with normal cytology.

Conclusions

Several large, long-term, longitudinal cohort studies provide evidence on the independent risk of developing breast cancer associated with cytological PED in NAF. The presence of PED – regardless of whether the specimen is obtained by biopsy, surgical excision, or nipple aspiration – has been found to be an independent predictor of developing breast cancer. The magnitude of risk of developing breast cancer with PED identified in NAF is similar to the risk of proliferative disease identified by surgical excision or core/needle biopsies, and exceeds in magnitude the risk associated with several factors currently used in guideline-recommended clinical risk models, including family history. The advent of new technologies for collection of NAF shows promise in reducing the inconveniences of routine specimen collection for breast cancer screening and in increasing the early, accurate detection of PED. The consequent effect of these enhancements, therefore, would be to improve clinical decision-making about the appropriate use of pharmacological interventions intended to prevent future breast cancer.

Transparency

Declaration of funding

The study is supported by Atossa Genetics. Publication of the study results was not contingent on sponsor’s approval or censorship of the manuscript; analysis and its reporting was directed without influence from the sponsors. Author contributions – J.H.: design, data analysis and interpretation, writing of manuscript; S.-C.C.: design, data interpretation, review of manuscript; Q.L.: design, literature search and data extraction, data analysis and interpretation, writing of manuscript; P.K.: data interpretation, review of manuscript; S.C.Q.: design, data interpretation, review of manuscript.

Declaration of financial/other relationships

S.C.Q. and S.-C.C. have disclosed that they are affiliated with Atossa Genetics. J.H., Q.L., and P.K. have disclosed that they are employees of Cedar Associates LLC, which received funding to be the research coordinating center for this study. The primary research was solely the responsibility of the authors that are not affiliated with Atossa Genetics Inc.

CMRO peer reviewers on this manuscript have received an honorarium from CMRO for their review work, but have no other relevant financial relationships to disclose.

Acknowledgments

Dr. Susan Love and Dr. Margaret Wrensch reviewed the manuscript and provided feedback. Patricia Choy contributed to literature screening and data extraction. Cole Johnson and Kevin Cheung helped with data extraction. The project was supported by Atossa Genetics Inc.