Nipple-Invasive Primary Carcinomas: Clinical, Imaging, and Pathologic Features of Breast Carcinomas Originating in the Nipple

Eight of the authors prospectively identified cases in their practices in 2015 and 2016 at Brigham and Women's Hospital (Boston, Massachusetts; J.E.B., B.T.H., D.A.D., S.C.L.), Brigham and Women's Faulkner Hospital (Boston, Massachusetts T.J.W., X.H.), the University of Louisville Hospital (Louisville, Kentucky; M.A.S.), and Newton-Wellesley Hospital (Newton, Massachusetts; A.J.G.). More than 8000 cases of invasive breast cancer were diagnosed at the 4 hospitals during that period. Cases included surgical specimens as well as cases seen in consultation. Carcinomas were defined as originating in the nipple if most of the invasive carcinoma was located in the nipple papilla and adjacent dermis, above the smooth muscle of the areola. Focal invasion below the smooth muscle was allowed if less than 1 cm and smaller in size than the carcinoma in the nipple. Epidermal cancers of the skin or skin appendages, such as squamous cell carcinoma or basal cell carcinoma, were not included. If there was a history of breast cancer, the nipple cancer had to be considered a new primary carcinoma based on clinical features, location, and pathologic findings. Clinical notes, breast imaging reports, pathology reports, and glass slides were reviewed. The study received institutional review board approval.

The papilla of the nipple has a median height of 0.6 cm (0 to 1.9 cm) and approximately 23 lactiferous sinuses traverse this area to open onto the overlying epidermis at duct orifices (Figure 1).¹¹  Glass slides were reviewed to document that the invasive carcinoma was primarily present within the papilla.

Twenty-four invasive carcinomas originating in the nipple were identified. All occurred in women. The mean age was 63 years (range, 37–86 years). The carcinomas could be separated into 2 groups based on clinical presentation and the site of origin of the invasive carcinoma.

Ten of the 24 patients (42%) presented with skin changes from ductal carcinoma in situ (DCIS), involving the nipple epidermis (Paget disease), and associated with direct invasion into the dermis from the epidermis (Table 1). Nine of the 10 patients (90%) presented with exudative crust, erythema, pruritus, and/or bleeding due to disruption of the epidermis by the tumor cells. One of the 10 patients (10%) presented with pigmentation related to melanocytic proliferation associated with the tumor cells. The areas of invasion arose from the tumor cells in the epidermis (Figure 2). The authors previously reported a series of 7 similar patients.¹²  The 10 carcinomas reported in the current study represent new cases, identified after the conclusion of the prior study.

Of the 10 cases, mammography was negative in 8 cases (80%) and showed calcifications associated with the DCIS in 2 cases (20%). Ultrasound was performed in 1 of the 10 cases (10%) and showed skin thickening. Magnetic resonance imaging (MRI) was performed in 4 of the 10 cases (40%) and showed skin enhancement in 3 of those 4 cases (75%) and was negative in 1 case (25%).

Of the 10 patients, 8 (80%) had microinvasion (<0.1 cm) and 2 (20%) had invasive carcinomas measuring 0.2 cm and 0.36 cm, extending from the epidermis into the dermis. Six of the 10 cancers (60%) were positive for HER2 by immunohistochemistry (3⁺), and the remaining 4 cases (40%) had equivocal results. Fluorescence in situ hybridization to evaluate HER2 amplification was not performed. Four of the 10 carcinomas (40%) were negative for estrogen receptor (ER) and progesterone receptor (PR). No lymphovascular invasion was observed. Five of the 10 patients (50%) had lymph node sampling with negative results. At a mean follow-up of 15 months (range, 6–27 months), no recurrences were reported.

Fourteen of the 24 patients (58%) had carcinomas arising within the nipple dermis, but not specifically from tumor cells in the nipple epidermis (Table 2). Four of the 14 patients (29%) had symptoms for several months up to a year before diagnosis. A subtle nipple mass, thickening, or retraction was the only indication of the cancer for 7 of the 14 patients (50%; Figure 3). One of the 14 patients (7%) had a mass with distortion of the nipple (Figure 4), and the remaining 6 patients (43%) had skin changes without a mass. Although the skin changes over the underlying invasive carcinoma were similar to Paget disease (ie, exudative crust, erythema, and/or pruritus), only 2 of those 6 patients (33%) actually had Paget disease. Those 2 patients had invasive cancers measuring 2.7 cm and 4 cm. Whether the first site of invasion originated in the epidermis could not be determined, as was seen in the previously discussed cancers. One of the 14 patients (7%) had Paget disease that was focal and clinically occult.

Ten of the 14 patients (71%) underwent mammographic evaluation that showed no abnormality in 7 of the 10 cases (70%), skin thickening in 1 case (10%), distortion in 1 case (10%), and calcifications in the remaining case (10%). Ultrasound was performed for 5 of the 14 patients (36%) and showed a mass in 3 cases (60%) and no changes in 2 cases (40%). An MRI was performed for 3 of the 14 patients (21%) and showed a mass in 1 of the 3 patients (33.3%), skin thickening in 1 patient (33.3%), and no changes in 1 patient (33.3%).

Three of the 14 carcinomas (21%) invaded into and ulcerated the epidermis and would be classified as American Joint Committee on Cancer (AJCC) T4b.¹³  Five of the 14 carcinomas (36%) invaded into the epidermis but did not cause ulceration. The remaining 6 carcinomas (43%) did not involve the epidermis. Of the 14 cancers, 8 (57%) were only present in the papilla and smooth muscle, 2 (14%) focally invaded below smooth muscle (<0.1 cm), and 4 (29%) invaded below the smooth muscle for 0.5 to 0.9 cm. In all cases, most of the invasive carcinoma was present in the papilla of the nipple.

The histologic type for these 14 cases was ductal in 7 cases (50%), lobular in 3 cases (21%), and with features of both in 4 cases (29%). Size ranged from 1.4 cm to 4 cm (mean, 2.1 cm). Very large carcinomas were excluded from the study because the site of origin within the nipple could not be determined. The larger carcinomas in the study invaded laterally in the dermis, above and into the smooth muscle of the areola (Figure 5). Well (5 of 14 cases; 35.7%), moderate (5 of 14 cases; 35.7%), and poorly (4 of 14 cases; 28.6%) differentiated carcinomas were observed. Twelve of the 14 carcinomas (86%) were positive for ER and negative for HER2 (“luminal-like”), and the remaining 2 carcinomas (14%) were positive for both ER and HER2. Of the 14 carcinomas, 7 (50%) were associated with DCIS (3 of the 7 [43%] with Paget disease) and 3 of the 14 (21%) with lobular carcinoma in situ (LCIS). The in situ carcinomas were present in lactiferous sinuses. An in situ component was not identified in 4 of the 14 cases (29%).

Dermal lymphovascular invasion was identified in 4 of the 14 cases (29%). None of those patients had a clinical presentation of inflammatory carcinoma. Ten of the 14 patients (71%) (mean size of invasive carcinomas, 2 cm) underwent lymph node sampling, and 3 of the 10 (30%) had a metastasis in a single node (2 macrometastases and 1 micrometastasis) (Table 2). All of the lymph node metastases arose from poorly differentiated carcinomas, and 2 of those 3 carcinomas (67%) were HER2 positive. One of the 14 patients (7%) had isolated tumor cells in a node. No recurrences have been reported at a mean follow-up of 2 years (range, 1–6 years).

The first, and only previous, large series of carcinomas originating in the nipple was published in 1956.¹⁴  Of approximately 10 000 cases of malignant disease of the breast in patients undergoing surgery at the Mayo Clinic in 46 years, only 25 (0.25%) originated in the nipple (excluding cases of skin cancer and carcinomas associated with more than focal Paget disease). We also found that those cancers are quite unusual, only encountering them rarely in surgical and consult cases (24 cases in >8000 cases [0.3%]). A major difficulty in identifying these cancers in databases or in the literature is the absence of a specific term to describe them. We propose nipple-invasive primary carcinoma as a term which, if used, could assist in future studies.

The cases we identified fell into 2 main groups. The first group presented with extensive skin changes secondary to Paget disease, with invasion arising directly from tumor cells in the epidermis (Table 1). We and others^12,15–21  have previously reported on 22 cases of “invasive Paget disease” (Table 3). The additional 10 cases reported in this study confirm the previous findings. The areas of invasion are small because larger carcinomas cannot be identified as specifically arising from tumor cells involving the skin.²²  Because of the association with Paget disease, most overexpress HER2. Of the 22 patients undergoing lymph node sampling, only 1 patient (4.5%) has had a lymph node macrometastasis (Table 3). That patient was unusual in that there were multiple areas of dermal invasion (the largest, 3.5 mm).²¹  No cases of distant metastasis or recurrence have been reported.

Carcinomas that did not qualify for the first group were larger and presented as a nipple mass or thickening and/or were due to skin changes from the underlying mass (Table 2). Twenty published studies were identified that reported 48 patients with cancers that were described as being similar to the second group of nipple-invasive carcinomas in the current study (Table 4).^14,22–40  Except for the first series from the Mayo Clinic, the rest of the studies were reports of 1 to 2 patients. As in the cases in the current study, most patients presented with nipple masses and/or with skin changes related to the carcinoma. Paget disease was only reported in 4 of the 48 cases (8%). Imaging studies were reported for 12 of the 48 patients (25%) and showed a mass in 6 of the 12 patients (50%), calcifications in 2 patients (17%), and had negative findings in 4 patients (33%). Carcinomas not observed in the current study were those associated with a benign nipple adenoma or papilloma.^23,25,28,34 

Ductal was the most common histologic type, followed by lobular, “neuroendocrine,” and “basaloid.” Twelve of the 48 published cancers (25%) reported a modified Bloom Richardson grade. Of those 12, 3 (25%) were well differentiated, 5 (42%) were well to moderately or moderately differentiated, and 4 (33%) were poorly differentiated. The mean size was 1.6 cm (range, <0.1–3.5 cm).

For 9 of the 48 published cancers (19%), both ER and HER2 were determined. Five cases (56%) were ER⁺ and HER2⁻ (“luminal-like”), 1 (11%) was ER⁺ and HER2 equivocal, 1 (11%) was ER⁻ and HER2⁺, and 2 (22%) were ER⁻ and HER2⁻ (“triple negative” or “basal-like”) (Table 4). Two of the 4 cancers (50%) associated with Paget disease were tested, and were both HER2⁺.

A total of 58 of the published cases had lymph node sampling, and of those, 18 (31%) had positive nodes, including 1 of 17 patients (6%) in the first group (Table 3) and 17 of 41 patients (41%) in the second group (Table 4). The mean size of carcinomas with both size and lymph node status reported in the second group was 1.8 cm (Table 4). Only 5 of the 18 cancers (28%) with lymph node metastases had both grade and tumor markers reported. Three of those 5 cancers (60%) were poorly differentiated and either positive or equivocal for HER2. Two of these 3 cases (67%) were negative for ER. The remaining 2 cancers (40%) were moderately differentiated, ER⁺, and HER2⁻. The number of positive nodes was reported in 8 of the 18 cases (44%). Seven of the 8 cases (88%) had only 1 positive node, and the remaining cancer (12%), associated with 8 lymph node metastases, was ER⁻ and HER2⁺.²² 

The published cases of carcinomas arising in the nipple and the additional cases identified for this study demonstrate that, apart from the invasive carcinomas arising in association with Paget disease, which are predominantly HER2⁺, carcinomas arising in the nipple are similar to carcinomas arising in the remainder of the breast. They include different histologic types, grades, and biologic types. A specific type of carcinoma arising from the special structures of the nipple was not identified. Therefore, these cancers may arise from similar structures as those found in the remainder of the breast. Lobules are present within the papilla in up to 26% of nipples (17 of 65 nipples [26%] in 1 study⁴¹) and could be the site of origin. Another possibility is that carcinoma in situ arose deeper in the breast and extended into the lactiferous ducts with invasion occurring in the nipple. Five of the 24 carcinomas (21%) in this study were associated with DCIS deeper in the breast.

Carcinomas involving the skin, superficially located, and/or close to the nipple are associated with a higher risk of lymph node metastases.^7–10  This may be due to those carcinomas passing through the dense, subareolar plexus of lymphatics as they invade from deeper in the breast into the dermis (Figure 1).⁴²  In the cases in this study and in the published cases, the number of carcinomas with lymph node metastases was comparable to cancers of similar size and was not markedly increased.⁴³  This may be due to those carcinomas being located above the lymphatic plexus. Only the larger carcinomas may have invaded deep enough to encounter the areolar lymphatics. Three of the 24 carcinomas (12.5%) reported in this study had lymph node metastases. Two of the 3 carcinomas (67%) with lymph node metastases invaded below the smooth muscle of the areola (both with macrometastases) and 1 (33%) did not (the carcinoma with a micrometastasis). The association of skin involvement with an increased likelihood of lymph node metastasis may be a surrogate for the presence of invasion through the lymphatic plexus rather than being due to skin involvement itself.⁷ 

Women opting to undergo prophylactic nipple-sparing mastectomies rather than total mastectomies accept the risk of a carcinoma arising in the nipple. Although some surgeons advocate removing the central core of tissue from the nipple, it is unclear how successful that technique is in practice and the procedure has not been standardized.⁴⁴  The observed risk of subsequent cancer after nipple-sparing prophylactic mastectomy is quite low—only 9 cases in 776 patients (1.2%) in 3 studies.^30,45,46  Only 1 of the 9 cancers (11%) occurred at the nipple. Therefore, residual breast tissue in the skin flaps or in far lateral areas of the breast appears to provide a greater risk than the nipple itself. The current study has shown that the rare cancers that originate in the nipple are similar to other breast cancers and do not carry an extremely high risk of nodal metastasis. Patients undergoing this procedure should be aware that cancers arising in the nipple can have a subtle presentation and can reach a substantial size before being detected by imaging. In addition, they should bring any skin changes (erythema, exudative crust, pigmentation, or pruritus) or nipple thickening to the immediate attention of their health care providers.