In the current series, 5/8 patients had concurrent metastatic disease to sites other than the stomach. Similar pattern of metastases was reported by Taal et al. (10). In the three cases with gastric involvement alone, the diagnosis was confirmed by comparison of the gastric biopsies with the primary breast carcinoma.
A high index of suspicion for metastatic breast cancer should be maintained when a patient has a history of breast cancer and develops gastric symptoms. Endoscopic, radiological and histological evaluation is essential to discriminate primary gastric cancer from breast cancer metastasis to the stomach. Radiological and endoscopic findings are nonspecific, and may be hard to distinguish from primary gastric cancer and non-Hodgkin’s lymphoma. The most common pattern of breast cancer metastasis to the stomach is linitis plastica with diffuse infiltration of the submucosa and muscularis propria, while the pattern of discrete nodules or external compression is less common (5, 10, 15). Radiological evidence of linitis plastica has been reported to be consistent with the metastatic involvement of the stomach by lobular breast cancer. Taal et al. (10) reported that 83% of patients with gastric involvement had lobular breast cancer as a primary histological subtype. In the current series 5/8 patients had the lobular type of breast carcinoma. Because the lesion is often limited to the submucosal and seromuscular layers of the stomach, the endoscopic evaluation may be normal in 50% of cases or may show only discrete mucosal abnormalities indistinguishable from other tumors or benign disease (16). Radiological findings on CT scan or barium meal include encasement of the whole stomach as seen in linitis plastica, multiple lesions of the stomach or extrinsic lesions of the gastric wall. Deep biopsies are needed to obtain representative material and the histological features should be compared with those of the primary breast tumor in order to confirm the diagnosis. Occasionally, lobular breast carcinoma may produce a signet ring morphology which can be confused with a primary gastric adenocarcinoma (10). The large number of signet-ring cells combined with a gastric mucosal spreading pattern can mean that metastatic disease to the stomach is almost indistinguishable from primary gastric linitis plastica (9). However, breast signet-ring cell carcinoma may show some morphological differences from gastric and colonic signet-ring cell carcinoma, such as a single, well-circumscribed univacuolated intracytoplasmic lumina, with a central eosinophilic inclusion, whereas other signet-ring cell carcinomas have the extended, globoid, and optically clear cytoplasmic acid mucin that pushes the nuclei against the cell membrane (17).
Detailed immunohistochemical analysis may be the only reliable method to differentiate between metastatic and primary gastric carcinoma. Metastatic breast carcinoma is usually positive for CK7, GCDFP-15, carcinoembryonic antigen (CEA), ER and progesterone receptor (PgR), and negative for CK20. CK20 proves to be particularly positive in gastric, colorectal, pancreatic and in transitional cell carcinomas, while it is not observed in any carcinomas of the breast (18). CK7 in contrast is extensively expressed in 90% of breast carcinomas and its expression was also observed in 50-64% of primary gastric adenocarcinoma (19). Although ER and PgR positivity in gastric biopsies suggest breast cancer metastasis to the stomach, it is worth noting that ER and PgR positivity with weak to moderate staining intensity has been reported in 32% and 12%, respectively, of patients with gastric cancer (5). ERα can be used to diagnose gastric metastasis from breast cancer as van Velthuysen et al. (20) reported that no primary gastric tumor expressed ERα. They also observed that the absence of E-cadherin staining was significantly related to metastatic breast carcinoma (20). Furthermore, cytoplasmic positivity for GCDFP-15 may also function to confirm mammary origin. Positive staining with GCDFP-15 has been found to be a sensitive (55-76%) and specific (95-100%) marker for correctly identifying a malignant lesion as metastatic breast carcinoma (21). An excellent correlation between GCDFP-15 positivity and the origin of a metastatic breast adenocarcinoma has been demonstrated (5).The positivity for CK7 and GCDFP-15 including hormone receptor expression, and in contrast, the negativity for CK20 and CA19-9 were of great value in differentiating an unsuspected lobular breast carcinoma from a gastric carcinoma in the present cases.
Data on treatment are scarce. Breast cancer metastasis to the stomach represents evidence of systemic disease and systemic therapy, such as chemotherapy and/or hormonal therapy, rather than surgical resection is indicated (5, 10, 22, 23). Only after a correct diagnosis can the treatment of generalized breast cancer be initiated. Surgical treatment should be reserved only for patients who develop complications such as obstruction or bleeding (7, 8, 10, 15, 22, 23). In most cases, surgical resection is not possible due to local invasion (10). McLemore et al. (22) reported that surgical intervention did not have a significant effect on survival (28 vs. 26 months). However, patients with metastasis only to the gastrointestinal tract who underwent palliative surgical resection tended to have a more prolonged median survival (44 vs. 9 months). This difference was not statistically significant. The decision-making process for surgical intervention should be based on the clinical presentation and symptoms, the availability of chemotherapeutic options and a quality of life discussion (22). In the current study, surgery was of benefit in two patients in whom gastric involvement was the only site of metastases. Both these patients were ER and/or PgR positive and were treated with hormonal therapy post-surgically. Another patient will proceed to surgical intervention as the stomach is the only metastatic site. Interestingly, these three patients were alive after 9-44 months of follow-up, whereas all the rest succumbed after a median time of 9 months. This difference was statistically significant (p=0.024), which would justify the use of surgical intervention in cases of operable solitary breast cancer metastasis to the gastrointestinal tract.
The choice of systemic treatment is based on age, presenting symptoms, performance status, ER status and previous treatment. All the present patients had received prior chemotherapy and two of them were treated with hormonal therapy (tamoxifen, letrozole). The response rate in this small series was 50%, similar to that reported by Taal et al. (10) (46%) but higher than that reported by Schwarz et al. (7) (30%). The median survival of the present patients from the time of diagnosis of gastric metastases was 11 months, similar to that reported by Taal et al. (10) (10 months) but shorter than that reported by McLemore et al. (22) (28 months) and by Ayantunde et al. (13) (20 months). In three patients, survival exceeded 2 years, while in the series of Taal et al. (10), 22% of the patients survived for more than 2 years.
In conclusion, primary gastric cancer can be distinguished from metastasis from a breast carcinoma by means of immunohistochemical analysis. Complete histopathological and immunohistochemical analysis of the gastric biopsies in comparison with the original breast cancer histology is essential to support the diagnosis of metastatic breast cancer. Surgical intervention should be reserved for palliation or may be a reasonable choice in cases of solitary resectable gastrointestinal tract metastases. Appropriate systemic treatment for metastatic breast carcinoma is the preferred treatment.