Cytology of Spontaneous Nipple Discharge—Is It Worth It?: Performance of Nipple Discharge Preparations in the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytopathology

Data from the CAP Interlaboratory Comparison Program in NGC were analyzed. This educational program consists of a quarterly glass slide mailing with 5 slides and a brief history. Before acceptance into the NGC program, each slide in the program is reviewed by 2 or 3 board-certified anatomic pathologists, who agree with the submitted diagnosis and conclude that these are good examples of the entity. Malignant diagnoses are usually histologically confirmed. Each slide has a general category (benign, suspicious, or malignant) and each slide has a specific reference diagnosis assigned to a body site. The program participants are able to choose a general category in which to place the case (benign, suspicious, malignant, unsatisfactory), and then a more specific reference diagnosis. After circulation, each slide has a unique performance profile. The performance of each slide can be compared against that of the others in the same reference category.

Participant responses for NDP between 2005 and 2009 were evaluated (N = 2506). General diagnostic category (benign, suspicious, malignant), participant type (pathologist, cytotechnologist), stain (Papanicolaou, modified Giemsa), and year in the program (2005–2009) were analyzed using χ²  and a nonlinear mixed model for slide factor correlation structure. By design, all papillary lesions were categorized as suspicious, and were the only entity in the suspicious general category. Differentiating benign papillomas from well-differentiated papillary cancers is difficult, and surgical extirpation is usually required.¹⁰  Therefore, the option of suspicious was afforded participants in those cases without single malignant-appearing cells in a papillary process.

Of the 2506 responses, the general category was malignant in 1280 (51%), suspicious/papillary in 171 (7%), and benign in 1055 (42%). There was a statistical difference between the general categories (P = .01); benign challenges were more often concordant to the correct general category than malignant or suspicious challenges. Cytotechnologists performed statistically better than pathologists in arriving at the correct general category (P = .03). There were no differences between stains or program years (Table 1). Two hundred twenty-two responses were discordant to the general diagnostic category, with a false–positive/suspicious rate of 12.8% and a false-negative rate of 3.4%. Cytotechnologists performed better than pathologists, with a higher false-negative rate for pathologists (15.3% versus 7.9%, P < .001). The most common false-negative diagnosis of adenocarcinoma was mastitis/abscess (125 of 1272 responses; 9.8%) (Table 2). The most common false-positive response for a benign case was papillary lesion (26 of 457 responses; 5.7%) (Table 3). Responses for papillary lesions are seen in Table 4.

Images of cases that performed well and poorly are seen in the Figure. The adenocarcinoma pictured in the Figure, A, is an example of adenocarcinoma with 100% agreement. There was a necrotic diathesis and numerous 3-dimensional fragments of cells, single cells, and degenerating cells with anisonucleosis, nuclear enlargement, and irregular nuclear membranes. Nucleoli were prominent and there was a contrast between the background macrophages and the malignant cells. The Figure, B, is an example of a benign challenge with 97% agreement by participants of benign. These cases usually had a granular background and scattered macrophages. The discordant diagnoses of the Figure, B, were suspicious diagnoses of papillary lesion. The Figure, C, is an example of a benign challenge that was not categorized as benign in 35% of responses. There were rare clusters of ductal cells scattered on the preparations with loose aggregates of macrophages. The history of the Figure, C, was that of a bilateral nipple secretion in a woman taking oral contraceptives. The Figure, D, is an example of a malignant NDP that was called benign 22% of the time. The malignant cells were scattered in a thin proteinaceous background, rarely clustered, and often found singly.

Nipple discharge is a relatively common complaint, causing patient anxiety and discomfort. Most nipple discharge symptoms are caused by benign conditions and may be physiologic (endocrine or drug related) or caused by an underlying nonmalignant ductal abnormality (duct ectasia, fibrocystic change, or papilloma). Historically, nipple discharge might have been evaluated primarily with cytologic evaluation, but recently, alternative evaluation algorithms have been used, which may include cytologic evaluation after a thorough history, physical examination, and imaging studies are completed.⁸ 

Although microscopy may have been used historically to assess nipple discharge for the presence of blood, the correlation of bloody discharge with malignancy is not absolute. Some investigators have reported higher rates of malignancy in patients with bloody nipple discharge,^11,12  whereas others have not found this association.^6,13–15  Morrogh et al⁶  reported that 55% of patients with underlying malignancy or a high-risk lesion had nonbloody nipple discharge on visual inspection, and 31% of patients with malignancy or high-risk lesion had nonbloody nipple discharge by both color and hemoccult testing.

Nipple discharge cytology has been described for many years.¹⁶  Criteria for nipple discharge cytology have been well defined, but there is morphologic overlap of features between benign and malignant entities. Usually there is a proteinaceous background in any discharge preparation. Benign cells of NDPs include normal duct cells, ductal cells exhibiting apocrine metaplasia, histiocytes, and varying numbers of inflammatory cells. Red blood cells may be seen in either benign or malignant specimens. Three-dimensional clusters of cells may be seen in NDPs in cases of papilloma or carcinoma. Nuclear atypia may be seen with intraluminal papilloma, an overlapping feature with carcinoma. It is commonly taught to avoid a malignant diagnosis unless there are many individual atypical cells. Conventional teachings also recommend histologic confirmation prior to definitive therapy with any case for which a malignant diagnosis is entertained when evaluating an NDP.¹⁰ 

In clinical studies, the sensitivity and specificity of cytology of NDPs have been cited as 46% and 95%,¹⁷  but other retrospective studies have indicated that the sensitivity may be as low as 11% and specificity only 76%,¹⁸  making it inappropriate as the only modality for evaluation of nipple discharge.¹⁷ 

In the CAP NGC program, there was a false–positive/suspicious rate of 12.8% and a false-negative rate of 3.4%, which confirmed the difficulty in accurately assessing NDPs. The most common false-negative diagnosis was mastitis when adenocarcinoma was present. The most common nonbenign diagnosis for a negative challenge was papilloma, which by program design is considered suspicious. The stain and program year did not have a significant impact on accuracy, but pathologists performed more poorly than cytotechnologists and participants were better at identification of negative cases.

It is important to note that the CAP NGC program is a purely educational program in interlaboratory comparison. The participants are given a single slide with limited clinical information. The performance of slides in this program is based upon morphologic evaluation and therefore does not reflect usual clinical practice. However, it also isolates morphology as a single parameter; thus, NDPs in the CAP NGC are parallel to clinical samples usually received in the laboratory. Indeed, NDPs received in daily practice are rarely accompanied by clinical or radiologic information. In those instances, the cytotechnologist and pathologist rely on morphologic impressions to arrive at a diagnosis, similar to the challenges in the CAP NGC.

If the findings in the CAP NGC parallel clinical practice, they illustrate that NDPs may adversely impact patient care. Although the CAP NGC findings reflect the rate of interpretive false-negative and false-positive results, they do not address the clinical false-negative results when abnormal cells may not be obtained from the patient. A benign NDP cytologic diagnosis does not exclude malignancy, and the clinical overall false-negative rate of NDPs is higher. In practice, nipple discharge cytology is only useful when positive for malignant cells, with up to 50% false-negative rates reported.^12,19 

Malignant cells may not be present in a discharge from a patient specimen, unlike the CAP NGC, which has confirmed malignant cells that were considered false-negative when unrecognized by participants. Conversely, the high false–positive/suspicious rate of the cytologic diagnosis requires corroborative evidence prior to definitive therapy.

In clinical practice, the significant risk factors associated with underlying breast carcinoma in a woman with a pathologic nipple discharge include age greater than 50 years, bloody nipple discharge, presence of a breast lump, and single-duct discharge.⁹  Cytology of NDPs would seem of little value in light of these risk factors, and direct duct excision after appropriate radiologic studies may be more prudent than using an inaccurate test such as NDP.