Lymphomas Involving the Breast: A Study of 106 Cases Comparing Localized and Disseminated Neoplasms : The American Journal of Surgical Pathology

Secondary Logo

Journal Logo

Original Articles

Lymphomas Involving the Breast

A Study of 106 Cases Comparing Localized and Disseminated Neoplasms

Talwalkar, Sameer S. MD*; Miranda, Roberto N. MD*; Valbuena, Jose R. MD*; Routbort, Mark J. MD, PhD*; Martin, Alvin W. MD; Medeiros, L. Jeffrey MD*

Author Information
The American Journal of Surgical Pathology 32(9):p 1299-1309, September 2008. | DOI: 10.1097/PAS.0b013e318165eb50
  • Free

Abstract

Nonepithelial neoplasms involving the breast are uncommon.32 In a recent study, the most common nonepithelial breast neoplasms were of hematologic origin, with non-Hodgkin lymphomas (NHLs) being most frequent.20 NHLs involving the breast account for 1.7% to 2.2% of extranodal NHL, and 0.38% to 0.7% of all NHLs.4,32,35 These neoplasms either may be localized, presumably arising in the breast, or a part of disseminated disease, presumably representing secondary involvement of the breast.

Although previous studies have reported the histologic findings of lymphomas involving the breast, many of these studies were limited by their small sample size or by the use of older lymphoma classification systems. The present study includes 106 patients with breast lymphomas seen at a single institution in the past 21 years. The specific aims of this study were 3. First, we reviewed the clinical, histologic, and immunophenotypic findings and classified each neoplasm using the current World Health Organization (WHO) classification system. Second, we subdivided these neoplasms into localized and disseminated groups and assessed the histologic and radiologic findings as well as the overall and disease-free survival (DFS) for the most common lymphoma types. Finally, we used immunohistochemical methods to subdivide a subset of localized diffuse large B-cell lymphomas (DLBCLs) into 2 groups that are associated with prognostic significance: germinal center (GC) B cell, and non-GC B cell.

MATERIALS AND METHODS

Clinical Information

The files of The University of Texas M.D. Anderson Cancer Center were searched for patients with lymphoid neoplasms involving the breast that were biopsied or excised during the course of their disease. One hundred and six patients were identified from January 1986 to August 2007. These patients were either referred to our institution for treatment or their slides were reviewed in consultation. All cases were reviewed and classified using the criteria of the WHO classification for hematolymphoid neoplasms.17 Clinical data, which included age, sex, side of involvement, clinical stage, and clinical follow-up were available for all 106 patients, whereas radiologic findings and treatment information were available for a subset of patients.

Patients were divided into 2 groups and their clinicopathologic features were assessed separately. These groups were (1) patients with localized (presumably primary) lymphomas; and (2) patients with disseminated lymphoma, involving breast and extramammary sites (presumably secondary involvement). We also included a small number of patients with a history of lymphoma or lymphoid leukemia that relapsed in the breast (often with other sites of relapse) in the group with disseminated disease.

We used the criteria proposed by Wiseman and Liao35 for designating a breast lymphoma as primary. These criteria included presence of breast parenchyma and lymphoma that are closely related, and no evidence of systemic disease after staging. Involvement of ipsilateral axillary lymph nodes is permitted in the definition of primary breast lymphoma. We also modified these criteria, as some needle biopsy specimens were small and breast tissue could not be identified. If such tumors were designated to involve the breast on the basis of radiologic studies, they were accepted as involving the breast as described previously.31

Histologic and Immunohistochemical Techniques

Hematoxylin and eosin-stained slides and unstained slides for immunohistochemical analysis were prepared from fixed, paraffin-embedded tissue sections. Immunohistochemical stains were performed using heat-induced epitope retrieval, an avidin-biotin complex method, and an automated immunostainer (Ventana Medical System, Tucson, AZ), as previously described.18

The antibody panel used to assess these cases was variable over the years and included the following antibodies (and recent dilutions): CD3 (1:150), CD10 (1:70), CD20 (1:700), CD45 (1:300), CD45RO (1:100), CD79a (1:50), terminal deoxynucleotidyl transferase (1:20), Ki-67 (1:100), BCL-2 (1:10), BCL-6 (1:10), κ (1:20), λ (1:10), myeloperoxidase (1:1000), AE1/AE3 (1:500), and cytokeratin MNF116 (1:50) (DAKO, Carpinteria, CA); CD15 (1:20), CD34 (1:20), CD43 (1:120), and CAM 5.2 (1:50) (Becton-Dickinson Biosciences, San Jose, CA); cyclin D1 (LabVision, Fremont, CA); CD5 (1:20), CD23 (1:15), and CD38 (1:75) (Novocastra, Newcastle-upon-Tyne, UK); CD2 (1:100) (Beckman Coulter, Westbrook, ME), CD30 (1:20) (Signet, Dedham, MA); MUM-1/IRF4 (1:200) (Santa Cruz Biotechnology, Santa Cruz, CA), and PAX5/BSAP (1:35) (Transduction Labs, Lexington, KY). Some immunostains were performed at referring institutions at the time of diagnosis.

We also specifically assessed the immunophenotype of a subset of localized DLBCL cases using antibodies reactive with CD10, BCL-6, and MUM-1/IRF-4. Gene expression profiling studies have shown that DLBCL is heterogeneous and consists of 2 major subgroups: GC B cell and non-GC B cell.29 Subsequent studies have suggested that immunohistochemical assessment for CD10, BCL-6, and MUM1/IRF4 can be used to identify these 2 groups, thereby approximating the gene expression data.6,15 A result was considered positive when >30% of the cells showed strong and diffuse staining, as previously suggested.15

Median overall survival (OS) and DFS were calculated for patients with the most common histologic types of localized and disseminated breast lymphomas, using the Kaplan-Meier method and comparisons made with the log-rank test.

RESULTS

Clinical Findings

There were 105 women and 1 man with a median age of 61 years (mean, 57 y; range, 8 to 88 y). The left breast was involved in 54 patients, the right in 45 patients, bilateral involvement was seen in 4 patients, and side of involvement was not specified in 3 patients. The specimens were obtained either by excisional biopsy (n=56) or by needle-core biopsy (n=50).

Localized Breast Lymphomas

Fifty cases of breast lymphoma were localized and presumably arose in the breast (primary). Their clinical features are summarized in Table 1. There were 49 women and 1 man with a median age of 61 years (mean, 59 y; range, 23 to 88 y). Forty-seven patients presented with a palpable breast mass and 3 had neoplasms that were detected incidentally during radiologic screening. These 3 neoplasms were 1 mucosa-associated lymphoid tissue (MALT) lymphoma, 1 classical Hodgkin lymphoma (CHL), and 1 peripheral T-cell lymphoma (PTCL). Forty-four (90%) patients had clinical stage I E disease and 5 (10%) patients had clinical stage II E disease. One patient in the localized group had bilateral involvement of the breasts for which we did not assign a stage. We recognize that others may consider this patient to have stage IV disease. However, as the lymphoma only involved the breasts, it is included in the localized group. Five patients who had stage II E disease had DLBCL (n=3), MALT lymphoma (n=1), and CHL (n=1).

T1-3
TABLE 1:
Clinical Features of Localized and Disseminated Breast Lymphomas at the Time of Diagnosis

The histologic types included 32 DLBCL, 14 extranodal marginal zone B cell lymphoma of MALT (MALT lymphoma), and 1 case each of CHL, precursor B-cell lymphoblastic lymphoma/leukemia (pre-B LBL/ALL), Burkitt lymphoma, and PTCL, unspecified. Clinical follow-up for these patients ranged from 4 to 192 months. Thirty-one patients were alive with disease, 14 patients were in remission with no evidence of disease, and 5 patients died. One patient with DLBCL was pregnant at the time of diagnosis. She had a normal postpartum period and was in remission at 27 months, with no evidence of disease. The man in this group, who had MALT lymphoma, was in clinical remission after local excision 31 months after diagnosis.

Information regarding therapy was known for 32 patients (Table 1). Most patients with DLBCL were treated with chemotherapy. In contrast, most patients with MALT lymphoma were treated by either lumpectomy or radiation therapy. Six patients had relapse after clinical remission (1 pre-B LBL/ALL, 1 PTCL, and 4 DLBCL). The patient with pre-B LBL/ALL relapsed in the contralateral breast 15 months after initial diagnosis. The patient with PTCL relapsed in the bone marrow 12 months after initial diagnosis. One of four patients with DLBCL relapsed in the same breast and also in the lymph node and the abdominal soft tissue 17 months after initial diagnosis. The other 3 DLBCL patients relapsed with involvement of the central nervous system at 64, 120, and 188 months after initial diagnosis. They were treated with intrathecal and systemic chemotherapy and local radiation. All 6 patients were alive with disease at the time of last follow-up.

Five patients in the localized group died of lymphoma, 3 with DLBCL (all non-GC immunophenotype) and 2 with MALT lymphoma, with relapse occurring at extramammary sites. All 3 patients with DLBCL had an involvement of the bone marrow. Recurrent breast involvement also occurred in 1 patient with MALT lymphoma.

Four patients had bilateral breast involvement: 3 patients had DLBCL and 1 patient had MALT lymphoma. One patient with bilateral DLBCL did not respond to the treatment and died of the disease at 58 months. Two other patients with DLBCL were alive with the disease and were on therapy at 6 and 68 months after diagnosis. The patient with bilateral MALT lymphoma was alive after 15 months of clinical follow-up.

The patient with localized CHL was referred with left axillary lymphadenopathy and complained of a left breast swelling. She had these symptoms for the past 4 years, during which time symptomatic therapy was provided while the patient had the provisional and clinical diagnosis of reactive lymphadenopathy. Excisional biopsies from the breast and axillary lymph nodes supported the diagnosis of CHL, nodular sclerosis type. She was considered to have localized disease, clinical stage II E, according to the inclusion criteria for this group used in this study. However, most likely this neoplasm arose in lymph node and secondarily involved the breast. She was alive with the disease 14 months after pathologic diagnosis.

Disseminated Breast Lymphomas

This group consisted of a total of 56 patients. All patients had evidence of lymphoma involving extramammary sites. It is recognized that a small subset of these cases could have arisen in the breast initially with subsequent dissemination. However, it is more likely that most of these patients had systemic lymphoma with secondary involvement of the breast. Their clinical features are summarized in Table 1. There were 55 women and 1 girl with a median age of 62 years (mean 57 y, range, 8 to 82 y). The 8-year-old girl had pre-B LBL/ALL. The neoplasms involved the left breast in 28 patients, right breast in 26 patients, and the side was not specified in 2 patients. Each patient had one or more palpable breast masses at the time of diagnosis. Clinical stage was available for 52 patients; 32 were stage III and 20 were stage IV.

Four patients had a history of lymphoma with relapse in the breast and were not staged. Two of these patients had pre-B LBL/ALL involving the breast and both had a history of chronic myelogenous leukemia. The breast neoplasms were considered lymphoid blast crisis, and these patients died 13 months and 31 months after diagnosis of pre-B LBL/ALL, respectively. Two patients had anaplastic lymphoma kinase (ALK)-negative anaplastic large cell lymphoma (ALCL). One patient had a history of lymphomatoid papulosis and CHL. At the time ALK-ALCL developed, the tumor involved the breast, skin, lung, lymph node, and bone marrow. The patient died at 36 months after the diagnosis of ALK-ALCL. The second patient had a history of cutaneous ALCL before developing ALK-ALCL of the breast. This patient was alive with no evidence of the disease with 5 months of follow-up after breast involvement was diagnosed.

The histologic types of lymphomas in the disseminated group included 15 follicular lymphoma, 12 DLBCL, 10 MALT lymphoma, 6 pre-B LBL/ALL, 6 ALCL, 4 mantle cell lymphoma, 3 CHL, and 1 case of pre-T LBL/ALL. Clinical follow-up ranged from 4 to 252 months. Thirty-eight patients were alive with the disease, 10 patients were in remission with no residual lymphoma, and 8 died. Two patients, 1 with DLBCL and 1 with pre-T LBL/ALL were pregnant at the time of diagnosis. One patient with MALT lymphoma had a coexistent ductal carcinoma of the breast, with involvement of multiple axillary lymph nodes. This woman was alive with disease at the time of last follow-up at 36 months. In the subset of 8 patients who died, 3 had pre-B LBL/ALL, 3 had MALT lymphoma, and 1 each had DLBCL and ALCL. Of the 3 patients with MALT lymphoma who died, 2 transformed to large B-cell lymphoma and 1 died because of comorbid condition complicated by chemotherapy.

Histologic Features

The histologic features of all cases of breast lymphoma are summarized in Table 2.

T2-3
TABLE 2:
Histologic Features of Localized and Disseminated Breast Lymphomas*

DLBCL

DLBCL was the most common histologic type (n=44), which included 32 localized and 12 disseminated cases. The most common cytologic variant in both groups was centroblastic (n=32) (Fig. 1). Localized cases were usually relatively well-circumscribed, whereas disseminated lymphomas were more often diffusely infiltrative and were more often associated with sclerosis. Disseminated lymphomas also tended to have a higher proliferation index as assessed by immunohistochemical staining for Ki-67. One case had a very high mitotic rate, a diffuse “starry-sky” pattern, an increased number of apoptotic cells, was BCL-2 negative, and had a Ki-67 rate of approximately 95%. The histologic features and immunophenotype were suggestive of atypical Burkitt/Burkitt-like lymphoma. However, fluorescence in situ hybridization analysis did not reveal c-myc rearrangement. In both localized and disseminated cases, the lymphoid infiltrate involving the breast parenchyma spared the ductal and acinar elements. Lymphoepithelial lesions were not identified.

F1-3
FIGURE 1.:
DLBCL localized to breast. A, The neoplasm replaced breast parenchyma in a diffuse pattern. Residual breast epithelium is present at the upper left of the field. B, The neoplastic cells are large with centroblastic cytologic features. A and B, Hematoxylin-eosin, A, 100×; B, 400×.

MALT Lymphoma

MALT lymphoma was the second most common histologic type (n=24), with 14 localized (Fig. 2) and 10 disseminated cases. Unlike DLBCL, both localized and disseminated MALT lymphomas were characterized by a diffuse pattern of infiltration. In most cases, the cells were small-to-medium sized with irregular nuclear contours, condensed chromatin, and moderately abundant pale cytoplasm. The latter cases had a monocytoid appearance. Reactive follicles were seen more commonly in localized compared with disseminated neoplasms. Two localized cases showed prominent plasmacytic differentiation. Mitotic activity was low and no discernible areas of necrosis were seen. Lymphoepithelial lesions were identified in a subset of localized and disseminated neoplasms. In the case associated with ductal carcinoma, the neoplastic lymphocytes partially surrounded the epithelial tumor, almost simulating a reactive lymphoid process when seen at lower magnification. The solitary case associated with sclerosis involved the breast as a part of disseminated disease.

F2-3
FIGURE 2.:
MALT lymphoma localized to breast. A, A needle biopsy specimen is extensively replaced by lymphoma. B, The neoplastic cells have abundant pale (monocytoid) cytoplasm. C to E, The monocytoid cells were positive for CD20 (C) and the plasmacytoid cells expressed cytoplasmic λ (D) and were negative for κ (E). A reactive germinal center is also present in the field. A and B, Hematoxylin-eosin. C to E, Immunohistochemistry with hematoxylin counterstain. A, 20×; B to E, 400×.

Follicular Lymphoma

Follicular lymphoma was the third most common subtype of lymphoma involving the breast (n=15); all cases occurred as a part of disseminated disease. Histologically, these neoplasms resembled their nodal counterparts. All of the grade 3 cases had a follicular and diffuse pattern, with the latter often predominant. No lymphoepithelial lesions were identified.

Lymphoblastic Lymphoma/Leukemia

This group included 6 pre-B cases and 1 pre-T case of LBL/ALL. Each neoplasm diffusely involved the breast stroma, usually in a single cell pattern dissecting through the stroma simulating, in part, lobular carcinoma of the breast (Fig. 3). Unlike lobular carcinoma, a concentric pattern of neoplastic cells around normal ducts was not observed. The individual cells were small to medium-sized, round to oval, with irregular nuclear contours, finely stippled chromatin, inconspicuous nucleoli, scant cytoplasm, and high mitotic rate. Each of these neoplasms was positive for terminal deoxynucleotidyl transferase.

F3-3
FIGURE 3.:
Pre-B LBL/ALL involving the breast in a patient with disseminated disease. A, The neoplasm replaces breast parenchyma but spares breast epithelium in some areas. B, The neoplastic cells in many areas infiltrated the breast parenchyma in a single cell pattern, and were small with immature chromatin. A and B, Hematoxylin-eosin. A, 100×; B, 1000×.

ALCL

There were 6 cases of ALCL, 4 ALK+ and 2 ALK−. The patients with ALK+ ALCL had lymphoma involving other sites at the time of breast involvement. These sites included lymph nodes, and some patients also had other extranodal sites of disease, but the bone marrow was not involved. The ALK+ ALCL cases were composed of large pleomorphic cells with irregular nuclear contours and multiple nucleoli, including some hallmark cells, consistent with the common variant. Mitotic figures were numerous. All 4 neoplasms had a nuclear and cytoplasmic pattern of ALK immunostaining consistent with the t(2;5)(p23;q35).

The 2 patients with ALK− ALCL were mentioned previously because both had a history of lymphoma before the breast involvement, which occurred at time of relapse. Both ALK−ALCL were associated with breast implants and were composed of large, anaplastic cells and a high mitotic rate (Fig. 4). Hallmark cells were identified but were less frequent than in ALK+ neoplasms. One of the patients with ALK− ALCL had bone marrow involvement at the same time the breast was involved.

F4-3
FIGURE 4.:
ALCL ALK-negative, involving breast and associated with an implant. The neoplastic cells are large and anaplastic. Inflammatory cells are also present in the background. Hematoxylin-eosin, 400×.

Other Lymphomas

There were 4 cases of CHL involving the breast; each neoplasm was of the nodular sclerosis type. Two cases had numerous Reed-Sternberg and Hodgkin cells associated with many eosinophils. Each patient also had involvement of lymph nodes.

There were 3 cases of mantle cell lymphoma that involved the breast in a diffuse pattern. Each patient had a history of lymph node involvement. The findings in the breast resembled those in lymph nodes with hyalinized blood vessels, many histiocytes with eosinophilic cytoplasm, a monotonous population of small neoplastic cells. Each neoplasm was positive for cyclin D1.

The study included single cases of Burkitt lymphoma and PTCL. The case of Burkitt lymphoma was composed of intermediately sized neoplastic cells arranged in a diffuse, starry-sky pattern. The neoplastic cells were positive for B-cell antigens including CD10 and fluorescence in situ hybridization performed on a fixed, paraffin-embedded tissue section showed split myc signals in most of the cells examined. The case of PTCL involved lobules imparting a nodular pattern, with individual nodules being separated by fibrous stroma (Fig. 5). Ductal structures were pushed to the periphery and lymphoepithelial lesions were identified. At scanning magnification, the pattern simulated lymphocytic mastitis. However, the infiltrate was dense, with numerous small lymphocytes admixed with scattered epithelioid histiocytes. Immunophenotypic studies showed T-cell lineage and polymerase chain reaction analysis of the T-cell receptor γ chain gene showed monoclonal rearrangements.

F5-3
FIGURE 5.:
Peripheral T-cell lymphoma localized to breast. A, The neoplasm selectively involved lobules. B, The neoplastic cells were mostly small lymphocytes with scattered large cells. A and B, Hematoxylin-eosin. A, 20×; B, 1000×.

Correlation of Histologic Classification With Presentation

DLBCL was more common in the localized group (32 of 50; 64%) than in the disseminated (12 of 60; 20%) and this difference was statistically significant (P<0.01). Certain histologic types of lymphoma were restricted to patients with disseminated lymphoma. These included follicular lymphoma, ALCL, mantle cell lymphoma, and pre-T LBL/ALL. The difference for follicular lymphoma between the localized and disseminated groups was statistically significant (P=0.0004). MALT lymphomas occurred at similar frequency in the localized (28%) and disseminated (18%) groups.

Immunohistochemical Findings

In each case, the immunohistochemical findings supported the histologic diagnosis in accord with the criteria of the WHO classification. All B-cell and T-cell neoplasms expressed pan-B-cell (CD20, CD79a, or PAX5/BSAP) or pan-T–cell antigens (CD3, CD5, or CD45RO), respectively. B-cell and T-cell lymphomas of specific types expressed the appropriate markers. One case of MALT lymphoma in the disseminated group was CD5+.

A subset of 26 cases of localized DLBCL, with paraffin blocks or unstained slides available for testing, was evaluated for CD10, BCL-6, and MUM-1/IRF-4. Seventeen cases had a non-GC immunophenotype (Fig. 6). All were strongly positive for MUM-1/IRF-4 and negative for CD10; 12 were also positive for BCL-6. Eight cases had a GC immunophenotype; 5 were positive for CD10, 7 were positive for BCL-6, and 1 was positive for MUM-1/IRF-4. The MUM-1/IRF-4 positive case was also positive for CD10 and BCL-6. One neoplasm could not be classified because it was negative for all the 3 markers. All 3 patients with DLBCL who died had neoplasms with a non-GC immunophenotype.

F6-3
FIGURE 6.:
DLBCL localized to breast with a non-GC cell immunophenotype (same case as shown in Fig. 1). A, CD10 was negative. B, MUM-1 was positive. A and B, Immunohistochemistry with hematoxylin counterstain.

Fifteen cases of follicular lymphoma were further assessed using immunophenotypic methods. Twelve (80%) cases were positive for CD10, 11 of 13 (85%) were positive for BCL-2, and all 15 cases were positive for BCL-6. This profile is similar to that of nodal follicular lymphomas.

Radiologic Findings

Radiologic findings, including the results of mammography, ultrasonography, or PET/CT, or more than one of these modalities, were available in a subset of 32 cases and are summarized in Table 1. Of these, 21 patients had localized and 11 patients had disseminated lymphomas. The histologic subtypes were as follows: DLBCL (n=24), follicular lymphoma (n=4), ALCL (n=2), mantle cell lymphoma (n=1), and MALT lymphoma (n=1). The predominant finding in all cases was a solitary mass. All patients with localized lymphomas had unifocal and solitary lesions. In contrast, 5 of 11 patients with disseminated lymphomas were multicentric. One patient with DLBCL had multiple circumscribed masses shown by mammography, and 4 patients (2 DLBCL, 1 follicular lymphoma, and 1 MALT lymphoma) had multifocal and multicentric masses shown by ultrasonography.

Treatment

Treatment data were available on 56 patients, 32 localized and 24 disseminated cases, and are summarized in Table 1. The treatment regimens were used variably, but all patients received stage-adapted modern therapy according to multimodality protocols at our institution. Forty-five patients received chemotherapy: 38 only chemotherapy, 4 with radiation therapy, and 3 with lumpectomy and radiation therapy. Five patients received only radiation therapy and 4 patients only lumpectomy. Two patients were observed; both had localized MALT lymphomas, were followed at regular intervals, and they eventually received chemotherapy when disseminated disease was detected.

Survival Analysis

The median OS and DFS of patients with localized or disseminated lymphomas were calculated (Table 3). There was no significant difference in OS between patients with localized and disseminated DLBCL (P=0.46) or between patients with localized and disseminated MALT lymphoma (P=0.24). There was also no difference in OS between patients with disseminated DLBCL, MALT lymphoma, and follicular lymphoma (P=0.31).

T3-3
TABLE 3:
Comparison of Staging and Outcome of DLBCL, MALT Lymphoma, and Follicular Lymphoma of Breast

There was a significant difference in DFS between patients with localized and disseminated DLBCL (P=0.003). In other words, patients with localized DLBCL had a lower risk of subsequent disease compared with patients who had disseminated DLBCL. In patients with disseminated lymphoma, those with DLBCL had a worse DFS than patients with MALT lymphoma and follicular lymphoma (P=0.01). The DFS of patients with localized and disseminated MALT lymphoma were not significantly different.

In the subset of patients with localized DLBCL in which their neoplasms were assessed for CD10, BCL-6, and MUM-1/IRF-4, we did not show a significant difference in OS or DFS between patients with GC versus non-GC tumors (P=0.42).

DISCUSSION

The present study is the largest clinicopathologic analysis of breast lymphomas, and unlike earlier reports, all of the neoplasms are classified using the criteria specified in the WHO classification. Our results show that histologic classification of lymphomas involving the breast correlates with clinical presentation. In patients with localized breast lymphoma, DLBCL and MALT lymphoma are the most common types. In contrast, patients with disseminated lymphomas have a wider range of lymphoma types involving the breast with the most frequent being follicular lymphoma, followed by DLBCL and MALT lymphoma. In this study, DLBCL was significantly more frequent in the localized group (64% localized vs. 20% disseminated; P<0.01) and follicular lymphoma only occurred in the disseminated group (P=0.0004).

Histologic classification also correlated with radiologic findings and survival. Patients with localized lymphomas presented with solitary masses in the breast. In contrast, approximately half of patients with disseminated lymphomas had multifocal involvement or masses in the breast. Regarding prognosis, there was no significant difference in OS between patients with localized versus disseminated DLBCL or MALT lymphoma. However, there were significant differences in DFS. Patients with disseminated DLBCL had a worse DFS than patients with localized DLBCL, and also a worse DFS than patients with follicular lymphoma and disseminated MALT lymphoma. These correlations with histologic classification may be useful in directing the staging work-up and assessing clinical stage, and also for predicting histologic type based on radiologic studies.

Others have suggested that NHLs of the breast fit within 1 of 2 clinicopathologic groups. The first group is composed of young women who frequently had bilateral and diffuse disease. Histologically, these lymphomas are commonly high-grade and they commonly disseminate rapidly to extranodal and extramammary sites.1,19 Some of the tumors reported were associated with pregnancy or detected in the postpartum period. The second group, which is more common, was composed of older women with a unilateral breast mass that clinically often mimicked breast carcinoma.35 However, in other studies in the literature and in the present study, these 2 clinicopathologic groups were not apparent. The reasons for the discrepancy are uncertain, but one likely possibility is that clinical practices have changed with the marked emphasis on screening methods (ie, mammography) and also the use of needle biopsy, both probably resulting in earlier diagnosis. For example, in the literature, localized DLBCL of the breast has been described as an aggressive neoplasm with a high risk of progression and involvement of the central nervous system, usually associated with high mortality.1,8,19 In the present study, 3 of 32 (9%) patients with localized DLBCL had involvement of the central nervous system, and all survived. Recent reports of DLBCL involving the breast are similar to our results; they reported a 5 to 10% frequency of central nervous system involvement.27,28

Yoshida and colleagues36 have reported 15 cases of localized DLBCL of the breast that they immunophenotyped into GC and non-GC types, and then correlated the results with survival. In the present study, we assessed 26 localized DLBCL and found that 18 (69%) were of non-GC immunophenotype and 8 (31%) were of GC immunophenotype. The explanation for DLBCL with a GC immunophenotype arising at an extranodal site such as the breast is uncertain. Although patients with DLBCL of non-GC immunophenotype have a work prognosis, we did not find a significant difference in the OS between localized DLBCL with a GC or non-GC immunophenotype. However, this may be attributable to the limited number of cases and relatively short follow-up intervals in this study.

In this study, the relative frequency of MALT lymphomas was similar in patients with localized and disseminated lymphomas, 28% versus 18%, respectively. Initially this seemed surprising to us, as MALT lymphomas are well known for remaining localized to their extranodal sites of origin for prolonged periods of time before disseminating. However, when MALT lymphomas disseminate, they are known to have a propensity to spread to extranodal sites.17 Infrequent CD5 expression in MALT lymphomas has been reported, and it has been suggested that this immunophenotypic finding may be associated with more widespread disease and a worse prognosis.3,10 We cannot confirm this suggestion as only 1 of 10 disseminated MALT lymphomas was CD5+ in this study.

Although previous studies1,2,5,7–10,16,19–23,26,33 have reported follicular lymphomas localized to the breast, this was not our experience. All the patients with follicular lymphoma in this study had evidence of disseminated disease. It is possible that some of these cases could have arisen in the breast, before dissemination elsewhere, but this seems unlikely for most of the patients in this study. In support of this argument, the immunophenotypic data of all these neoplasms were consistent with a nodal origin with secondary dissemination, as most cases in this study were BCL-2 positive, and primary extranodal follicular lymphomas are often BCL-2 negative.14 The reason for this discrepancy between our study and the results of others is unknown. In a recent study, follicular lymphomas involving the breast were usually stage III or IV, in accord with our findings.12

Two cases of ALCL in this study, both ALK−, were associated with breast implants. Others have previously reported an association between breast implants and lymphoma involving the breast, including 5 cases of ALCL.11,13,25,30 The possible pathogenesis of this occurrence is currently unknown. The possibility of chance alone is difficult to exclude, especially when one considers the very few cases of breast lymphoma reported, and the large number of women who have undergone breast implants. However, it is intriguing that ALCL, a relatively uncommon type of lymphoma representing approximately 2% of all NHLs, seems to be associated with implants. If chance alone were the explanation, one would see more cases of DLBCL or MALT lymphoma associated with implants reported in the literature.

One unusual aspect of this study was the relatively high frequency of precursor LBL/ALL. There were 7 cases in this study, including 6 of pre-B and 1 pre-T lineage. The explanation for this high frequency is uncertain, but may be related to the referral patterns of patients who come to this institution. In support of this hypothesis, 2 patients with pre-B LBL/ALL in this study had a history of chronic myelogenous leukemia and this institution has a very large chronic myelogenous leukemia program. Most of the patients with LBL/ALL, 6 of 7 (83%), had disseminated lymphoma associated with breast involvement. Others have shown that pre-B LBL/ALL may uncommonly present as an extranodal mass although the breast is an unusual site.26 Precursor T-cell LBL/ALL of the breast is rare and almost all of the reported cases had evidence of systemic disease,24,34 as was the case for the patient in our study.

In summary, we have presented 106 cases of lymphomas involving the breast, the largest study to date, and we have shown that histologic classification correlates, in part, with clinical presentation, radiologic findings, and prognosis. Our results also differ in some ways from older studies of breast lymphoma. We suggest that patterns of clinical practice in the workup of patients with breast lymphoma have changed, leading to earlier detection of disease, possibly as a result of increased emphasis on screening and the use of needle biopsy for diagnosis.

REFERENCES

1. Abbondanzo SL, Seidman JD, Lefkowitz M, et al. Primary diffuse large B-cell lymphoma of the breast. A clinicopathologic study of 31 cases. Pathol Res Pract. 1996;192:37–43.
2. Aguilera NS, Tavassoli FA, Chu WS, et al. T-cell lymphoma presenting in the breast: a histologic, immunophenotypic and molecular genetic study of four cases. Mod Pathol. 2000;13:599–605.
3. Ballesteros E, Osborne BM, Matsushima AY. CD5+ low-grade marginal zone B-cell lymphomas with localized presentation. Am J Surg Pathol. 1998;22:201–207.
4. Brogi E, Harris NL. Lymphomas of the breast: pathology and clinical behavior. Semin Oncol. 1999;26:357–364.
5. Cabras MG, Amichetti M, Nagliati M, et al. Primary non-Hodgkin's lymphoma of the breast: a report of 11 cases. Haematologica. 2004;89:1527–1528.
6. Chang CC, McClintock S, Cleveland RP, et al. Immunohistochemical expression patterns of germinal center and activation B-cell markers correlate with prognosis in diffuse large B-cell lymphoma. Am J Surg Pathol. 2004;28:464–470.
7. Cohen PL, Brooks JJ. Lymphomas of the breast. A clinicopathologic and immunohistochemical study of primary and secondary cases. Cancer. 1991;67:1359–1369.
8. Domchek SM, Hecht JL, Fleming MD, et al. Lymphomas of the breast: primary and secondary involvement. Cancer. 2002;94:6–13.
9. Farinha P, Andre S, Cabecadas J, et al. High frequency of MALT lymphoma in a series of 14 cases of primary breast lymphoma. Appl Immunohistochem Mol Morphol. 2002;10:115–120.
10. Ferry JA, Yang WI, Zukerberg LR, et al. CD5+ extranodal marginal zone B-cell (MALT) lymphoma: a low-grade neoplasm with a propensity for bone marrow involvement and relapse. Am J Clin Pathol. 1996;105:31–37.
11. Fritzsche FR, Pahl S, Petersen I, et al. Anaplastic large-cell non-Hodgkin's lymphoma of the breast in periprosthetic localisation 32 years after treatment for primary breast cancer—a case report. Virchows Arch. 2006;449:561–564.
12. Ganjoo K, Advani R, Mariappan MR, et al. Non-Hodgkin lymphoma of the breast. Cancer. 2007;110:25–30.
13. Gaudet G, Friedberg JW, Weng A, et al. Breast lymphoma associated with breast implants: two case-reports and a review of the literature. Leuk Lymphoma. 2002;43:115–119.
14. Goodlad JR, Batstone PJ, Hamiltin DA, et al. BCL2 gene abnormalities define distinct clinical subsets of follicular lymphoma. Histopathology. 2006;49:229–241.
15. Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103:275–282.
16. Hugh JC, Jackson FI, Hanson J, et al. Primary breast lymphoma. An immunohistologic study of 20 new cases. Cancer. 1990;66:2602–2611.
17. Jaffe ES, Harris NL, Stein H, et al, eds. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of Haematopoietic and Lymphoid Tissues. Lyon, France: IARC Press; 2001.
18. Khoury JD, Jones D, Yared MA, et al. Bone marrow involvement in patients with nodular lymphocyte predominant Hodgkin lymphoma. Am J Surg Pathol. 2004;28:489–495.
19. Kuper-Hommel MJ, Snijder S, Janssen-Heijnen ML, et al. Treatment and survival of 38 female breast lymphomas: a population-based study with clinical and pathological reviews. Ann Hematol. 2003;82:397–404.
20. Lin Y, Govindan R, Hess JL. Malignant hematopoietic breast tumors. Am J Clin Pathol. 1997;107:177–186.
21. Loughrey MB, Windrum P, Catherwood MA, et al. WHO classification of breast lymphomas. J Clin Pathol. 2004;57:1213–1214.
22. Mambo NC, Burke JS, Butler JJ. Primary malignant lymphomas of the breast. Cancer. 1977;39:2033–2040.
23. Mattia AR, Ferry JA, Harris NL. Breast lymphoma. A B-cell spectrum including the low-grade B-cell lymphoma of mucosa associated lymphoid tissue. Am J Surg Pathol. 1993;17:574–587.
24. Morel P, Lepage E, Brice P, et al. Prognosis and treatment of lymphoblastic lymphoma in adults: a report of 80 cases. J Clin Oncol. 1992;10:1078–1085.
25. Olack B, Gupta R, Brooks GS. Anaplastic large cell lymphoma arising in a saline breast implant capsule after tissue expander breast reconstruction. Ann Plast Surg. 2007;59:56–57.
26. Park YH, Kim SH, Choi SJ, et al. Primary malignant lymphoma of the breast: clinicopathological study of nine cases. Leuk Lymphoma. 2004;45:327–330.
27. Ribrag V, Bibeau F, El Weshi A, et al. Primary breast lymphoma: a report of 20 cases. Br J Haematol. 2001;115:253–256.
28. Ryan G, Martinelli G, Kuper-Hommel M, et al. Primary diffuse large B-cell lymphoma of the breast: prognostic factors and outcomes of a study by the International Extranodal Lymphoma Study Group. Ann Oncol. 2007 [Epub ahead of print].
29. Rosenwald A, Wright G, Chan WC, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med. 2002;346:1937–1947.
30. Sahoo S, Rosen PP, Feddersen RM, et al. Anaplastic large cell lymphoma arising in a silicone breast implant capsule: a case report and review of the literature. Arch Pathol Lab Med. 2003;127:e115–e118.
31. Talwalkar SS, Valbuena JR, Abruzzo LV, et al. MALT1 gene rearrangements and NF-kappaB activation involving p65 and p50 are absent or rare in primary MALT lymphomas of the breast. Mod Pathol. 2006;11:1402–1408.
32. Tavassoli FA, Devilee P, eds. Malignant Lymphoma and Metastatic Tumours. World Health Organization Classification of Tumours. Pathology and Genetics. Tumors of the Breast and Female Genital Organs. Lyon, France: IARC; 2003:107–109.
33. Topalovski M, Crisan D, Mattson JC. Lymphoma of the breast. A clinicopathologic study of primary and secondary cases. Arch Pathol Lab Med. 1999;123:1208–1218.
34. Vakiani E, Savage DG, Pile-Spellman E, et al. T-cell lymphoblastic lymphoma presenting as bilateral multinodular breast masses: a case report and review of the literature. Am J Hematol. 2005;80:216–222.
35. Wiseman C, Liao KT. Primary lymphoma of the breast. Cancer. 1972;29:1705–1712.
36. Yoshida S, Nakamura N, Sasaki Y, et al. Primary breast diffuse large B-cell lymphoma shows a non-germinal center B-cell phenotype. Mod Pathol. 2005;18:398–405.
Keywords:

breast; non-Hodgkin lymphoma; immunohistochemistry; diffuse large B-cell lymphoma; MALT-lymphoma

© 2008 Lippincott Williams & Wilkins, Inc.