Hematolymphoid Tumors of the Breast

Clinical Presentation and Imaging

Similar to carcinoma, most patients present with a clinically palpable mass. Non-mass-like enhancement with diffuse heterogeneous parenchymal involvement has also been described, as has skin thickening in a subset of DLBCL cases, best highlighted by magnetic resonance imaging (MRI). Appearance by mammography is highly variable. Lesions may appear of high density and circumscribed, microlobulated, or oval by mammography. By sonography, lesions are more typically hypoechoic but may occasionally be hyperechoic. By MRI, breast DLBCL may have irregular, smooth, or spiculated edges with variable enhancement patterns (homogeneous or heterogeneous, rim shaped), and precontrast T1 appearance may be isointense or hyperintense. By PET, the F-fluorodeoxyglucose (F-FDG ¹⁸ ) avidity is high (97%) with higher overall sensitivity than for MALT lymphoma.

Histology, Phenotype, Genotype, and Cytogenetic Findings

The histological appearance of DLBCL is that of an overtly malignant neoplasm with a diffuse infiltrate of large, transformed lymphoid cells effacing the underlying normal breast architecture ( Fig. 34.1 ). The nuclei should be equal to or greater in size than a macrophage nucleus, or twice the size of a normal lymphocyte, with occasional marked nuclear membrane irregularity. The two most common morphological variants are (1) centroblastic, with oval to round vesicular nuclei, fine chromatin, and several nucleoli; and (2) immunoblastic, characterized by a single prominent nucleolus and sometimes plasmacytoid features. Other variants are described in DLBCL-NOS including (3) anaplastic characterized by very large cells with pleomorphic nuclei, which may mimic Hodgkin’s and Reed-Sternberg cells; and (4) rare variants including small centroblastic, signet ring, or even spindle shaped. The spindle-shaped variant is described in one case report as a primary breast lesion, which could potentially be misdiagnosed as a metaplastic or other spindle cell lesion of the breast; the rare signet ring variant is described as a primary lesion in the breast of one patient in a series of seven cases of DLBCL with signet ring morphology. Centroblastic morphology appears to predominate in both localized and disseminated breast DLBCL, but this distinction is not considered an important one. Areas of background sclerosis may also be seen, especially in disseminated cases of DLBCL ( Fig. 34.2 ). Lymphoepithelial lesions have been reported in DLBCL without a coexisting MALT component. Distinction from carcinoma can be difficult on morphological grounds alone. Although the lack of cellular cohesion and absence of an in situ component may be useful histological features favoring lymphoma, the presence of in situ carcinoma does not exclude the possibility of a lymphoma because epithelial and hematopoietic neoplasms may rarely coexist.

The large lymphoid cells are immunoreactive for B-cell markers such as CD20, CD79a, PAX5, and CD19 ( Fig. 34.3 ), have variable expression of germinal center–associated and other B-cell subset markers, are negative for T-cell markers such as CD3, and are negative for epithelial markers such as cytokeratin (CK). If fresh tissue is available, flow cytometric studies can be used to look for monotypic light chain restriction ( Fig. 34.3D ), although a subset of DLBCLs are surface light chain negative (24% in one study). In addition, a negative flow cytometric evaluation would not preclude involvement by lymphoma if the involved area was not sampled or if insufficient viable tumor cells are present for evaluation, issues which may arise particularly in small biopsies or with large cell neoplasms. The nuclear proliferation marker Ki-67 can be used to assess the proliferation fraction as well as nuclear size and morphology ( Fig. 34.4 ) and may be particularly useful in distorted or small biopsies.

DLBCL-NOS is divided into two important subtypes (germinal center B-cell [GCB]–like and activated B-cell [ABC]–like) based on gene expression profiling studies which also include some unclassifiable cases. Although current techniques used for gene expression profiling use paraffin-embedded tissue, the Hans IHC algorithm which divides DLBCL into GCB and non-GCB groups based on the use of three IHC stains is considered an acceptable surrogate. The GCB group includes cases with greater than 30% CD10+ cells or, if CD10 is negative, cases with greater than 30% BCL6 ++ cells and less than 30% IRF4/MUM1 ++ cells. All other cases are considered non-GCB type. Identification of the so-called cell of origin has been required by the 2016 WHO classification because, although the literature is not completely consistent, the GCB type has a better prognosis than the non-GCB type in a moderate number of studies. It was also required because of the potential that there might be therapies which were under investigation that would help overcome the worse prognosis of the non-GCB DLBCL; however, after initial enthusiasm, recent results have not been as promising, and the need for more specific classifiers has been emphasized.

With the Hans algorithm, primary breast DLBCL predominantly shows a non-GCB phenotype ( Fig. 34.5 ). This finding is consistent with the lack of ongoing somatic hypermutation in DLBCLs of the breast. The non-GCB type also more commonly shows a double expressor phenotype with IHC overexpression of MYC and BCL2, which is correlated with an unfavorable prognosis. These patients account for approximately 25% of DLBCL cases. This double expressor phenotype should not be confused with high-grade B-cell lymphoma (HGBCL), with MYC and BCL2 and/or BCL6 rearrangements (“double hit” or “triple hit”), recognized by the 2016 WHO classification as a separate, highly aggressive and uncommon type of lymphoma. HGBCL characterized by concurrent rearrangements of MYC and BCL2 and/or BCL6 genes comprises approximately 5% of what previously would have been considered DLBCL. Although HGBCL often shows IHC overexpression of MYC and BCL2, overlapping with the double expressor phenotype, it is predominantly of the GCB type, and usually exhibits an extremely poor prognosis in most, but not all, cases. The 5-year overall survival (OS) of 28.6%, described in one study specifically for primary breast HGBCL (PB-HGBCL), was lower than the 75.9% 5-year OS for primary breast DLBCL (PB-DLBCL), which was also influenced by the type of treatment received. Whether the cases with BCL2 versus BCL6 rearrangements should be segregated from each other is currently controversial. Limited data are available regarding PB-HGBCL given the rarity of this entity.

More recent molecular/cytogenetic studies have further subdivided DLBCL into four to six molecular/ genetic groups with varying proportions of unclassifiable cases. The different groups vary in terms of their prognosis, clinical associations, and potentially their therapeutic implications. The majority of breast DLBCLs fall into the MYD88 cluster of Lacy et al, which is similar to the MCD cluster of Schmitz et al and the C5 cluster of Chapuy et al. This cluster is associated with an ABC cell of origin and an adverse prognosis in general, and includes the majority of primary CNS and testicular lymphomas. These findings suggest that breast DLBCLs may be grouped with DLBCLs that arise at immune-privileged sites and also may relate to their propensity for CNS involvement. This molecular/cytogenetic classification is not, however, a part of routine clinical practice at this time.

Some cases of DLBCL may also express CD5, a T-cell–associated antigen, uncommonly seen in PB-DLBCL (13% in one series of 15 cases). The two patients with CD5+ breast DLBCLs in this study were both still alive without relapses after 35 and 171 months of follow-up, although CD5 expression has been reported to be associated with an unfavorable prognosis in de novo DLBCL. The absence of cyclin D1 and SOX11 expression helps distinguish DLBCL from the vast majority of aggressive variants of mantle cell lymphomas (MCLs), which are also CD5+ lymphoid neoplasms. It is important to use one of the more specific antibodies for SOX11.

Treatment and Prognosis

Treatment includes limited surgery, with no increased benefit for mastectomy, together with anthracycline-based chemotherapy and an anti-CD20 monoclonal antibody such as rituximab, in the combination R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone), and radiation therapy (RT). A more recent study including 36 cases of PB-DLBCL and HGBCL found an increased benefit from R-EPOCH (dose adjusted rituximab, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin), resulting in statistically significant better 5-year OS for patients with PB-DLBCL (78.9% vs. 62.5%, p =.024), though R-EPOCH resulted in more severe side effects, such as myelosuppression. In addition, cases that fall under the non-GCB phenotype and MYD88 cluster may also benefit from drugs that target B-cell receptor-dependent NF-kB activation. Compared with nodal DLBCL, PB-DLBCL has a significantly worse disease-free survival (DFS) and OS, and the sites of progression tend to be mainly extranodal. One contributing factor could be the propensity for CNS involvement, although the role of CNS prophylactic treatment is still controversial. Patients with extramammary involvement have a shorter DFS than those with localized disease.

Prognostic Factors

The International Extranodal Lymphoma Study Group found that in a multivariate analysis of patients not receiving rituximab, a favorable international prognostic index (IPI) score, anthracycline-containing chemotherapy, and radiotherapy (RT) were significantly associated with longer OS. Other studies have confirmed the prognostic significance of IPI, Ann Arbor stage, and age.

Differential Diagnosis

The major differential diagnosis of DLBCL includes carcinoma, nonhematopoietic neoplasms, other types of malignant lymphoma, and rarely, lymphoid hyperplasia. One must always keep the latter possibility in mind when evaluating small or extremely disrupted CNB specimens in which the tissue architecture is not evident, because on rare occasions, large sheets of confluent, naked reactive germinal centers without clear-cut mantle zones may simulate DLBCL by forming apparent sheets of transformed B cells. Follicular dendritic cell stains (CD21, CD23, CD35) may be useful in highlighting an underlying follicular dendritic cell meshwork.

If confronted with a malignant neoplasm composed of large B cells, before a diagnosis of DLBCL-NOS is rendered other types of malignant lymphoma should be excluded, including the previously mentioned category of HGBCL with rearrangements of Fig. 34.6 MYC and BCL2 and/or BCL6 or HGBCL-NOS. Cases that show high-grade features but lack MYC and BCL2 and/or BCL6 rearrangements and do not fall into the category of DLBCL or Burkitt lymphoma (BL) are currently classified as HGBCL-NOS. These cases may show an increased proliferation index as indicated by prominent mitotic activity, apoptosis and/or tingible body macrophages, marked nuclear pleomorphism or a more monomorphic appearance reminiscent of BL, and/or blastoid morphology. Aggressive variants of MCL may also be composed of large lymphoid cells with nuclear pleomorphism and/or blastoid features; however, the tumor cells are usually cyclin D1+ and harbor the translocation t(11;14)(q13;q32) corresponding to the IGH/CCND1 rearrangement. The much less frequent cyclin D1-negative MCL may be identified with the help of a SOX11 stain. Although breast DLBCL has been reported to have a high proliferation index, ranging from 60% to 95%, a proliferation index of nearly 100% would be unusual and could raise the possibility of HGBCL or BL with a greater degree of nuclear pleomorphism than typically seen ( Fig. 34.6 ⁶⁷ ). A high proliferation fraction, however, by itself, is insufficient to make the diagnosis of a HGBCL. In addition, poor tissue preservation may obscure the classic morphological features of BL. If the phenotype is typical for BL (CD10+, BCL6 + , BCL2−), FISH studies for MYC , BCL2 , BCL6 , and immunoglobulin heavy and light chains could be performed on paraffin-embedded tissues for further evaluation. BLs are characteristically associated with a sole MYC translocation to IG (mostly the IGH locus at 14q32) and lack concurrent BCL2 and BCL6 rearrangements. It is important to separate BL from DLBCL because BLs receive more aggressive chemotherapeutic regimens, and R-CHOP is not considered adequate. Concurrent MYC and BCL2 and/or BCL6 rearrangements signify HGBCL with a “double hit” or “triple hit” as described previously.

Primary mediastinal (thymic) large B-cell lymphoma (PMBCL) is another type of large B-cell lymphoma that rarely involves the breast and is distinguished from DLBCL-NOS. PMBCL usually occurs in young adults with a female predominance, who may present with shortness of breath caused by a large mediastinal mass. A characteristic feature of extramediastinal disease is the involvement of unusual extranodal locations, including the lung, pleura, pericardium, and breast, which may undergo biopsy before the mediastinal mass is discovered. Histologically, sheets of large lymphoid cells that frequently have pale cytoplasm are separated by delicate strands of compartmentalizing fibrosis. The cells are typically immunoglobulin-negative, CD20+ B cells, although surface immunoglobulin has been reported to be present in a significant subset of cases using flow cytometry. There is some morphological, immunophenotypic, and molecular overlap with Hodgkin’s lymphoma (HL). Some of the neoplastic cells may resemble Reed-Sternberg cells and variants, most cases demonstrate at least some CD30 expression, although it is typically less uniform and less intense than in HL, and there is an overlap in the gene expression profiles. Other helpful but not completely specific IHC stains include CD23, MAL, and CD200, which are much more often positive in PMBCL than in DLBCL. Gains of chromosome 9p have been described as characteristic of PMBCL, occurring in 75% of PMBCLs as detected by FISH studies. The OS of PMBCL is more favorable than DLBCL-NOS.

Clinical Presentation and Imaging

The clinical presentation of MALT lymphoma does not appear to differ significantly from carcinoma because patients often present with asymptomatic mass lesions. There is no evidence to support that screening mammography increases the detection of breast lymphoma; however, it may identify at least a rare MALT lymphoma.

On mammogram, the masses may have irregular, partly defined or well-defined borders. Doppler sonography reveals heterogeneity and the strong vascularization of MALT lymphomas. By contrast-enhanced breast MRI, MALT lymphomas are hyperintense on T2-weighted images and isointense on T1-weighted images with strong and rapid contrast enhancement. Computed tomography (CT) has been reported to show homogeneous attenuation and moderate enhancement of well-marginated mass-like lesions. Although PET is less sensitive in indolent lymphoma, 54% of MALT lymphomas were reported to be 18F-FDG avid compared to 97% of DLBCLs, and 100% of HLs, BLs, MCLs, nodal marginal zone lymphomas, and lymphoblastic lymphomas. Interestingly, 18F-FDG/PET may be more sensitive in MALT lymphomas with plasmacytic differentiation owing to significantly increased uptake.

Histology, Phenotype, Genotype, and Cytogenetic Findings

MALT lymphomas usually form a discrete mass, with a well-circumscribed lymphoid infiltrate and occasional bands of sclerosis ( Fig. 34.7 ). The low-magnification appearance may resemble an intramammary lymph node; however, entrapped epithelium is present and a well-defined nodal architecture is lacking ( Fig. 34.8 ). In other cases, the infiltrate is less well circumscribed. The lymphoid cells are predominantly small with clumped chromatin. Although a monocytoid appearance is typical with a moderate amount of pale cytoplasm ( Fig. 34.9 ), some MALT lymphomas have only scant cytoplasm. Numerous reactive germinal centers may be present within or at the periphery of the mass, with or without infiltration by the MALT lymphoma (follicular colonization) ( Fig. 34.10 ). Difficulties may arise in distinguishing a MALT lymphoma with extensive follicular colonization from an unusual follicular lymphoma. In this situation, cytogenetic FISH studies for the follicular lymphoma-associated IGH/BCL2 rearrangement could be useful (see “Treatment and Prognosis” for a more detailed discussion).

Plasmacytic differentiation may also be present ( Fig. 34.11 ). In one series, a monotypic plasma cell component was identified in 72% of breast MALT lymphomas, with 36% of breast MALT lymphomas showing greater than 20% plasma cells within the infiltrate. In some cases, plasmacytic differentiation may be so marked that plasma cell myeloma or plasmacytoma could be considered in the differential diagnosis. Lymphoepithelial lesions may be present ( Fig. 34.12 ), but unlike gastric or salivary gland MALT lymphomas, they are infrequent. Although scattered larger transformed-appearing cells may be seen, sheet-like proliferations of large lymphoid cells are not, and if identified, a separate diagnosis of DLBCL should be rendered.

The neoplastic cells are monoclonal B cells that express pan-B-cell markers such as CD20, PAX5, and CD79a, and often demonstrate coexpression of BCL2. Most, but not all, cases are CD5−, and CD10 expression is not seen. Expression of CD43 has been reported in 30% to 50% of breast MALT lymphomas. Unlike most MCLs, MALT lymphomas lack cyclin D1 expression.

When reactive follicles accompany the neoplastic infiltrate, immunostains can be helpful in delineating the BCL6+, usually CD10+, and BCL2− reactive germinal center cells from the BCL6−, CD10−, usually BCL2 ++ MALT lymphoma cells that surround and infiltrate follicles ( Fig. 34.13 ). Follicular dendritic cell (FDC) markers (CD21, CD23, CD35) may also be useful in highlighting expanded FDC meshworks ( Fig. 34.14 ), which can be histologically subtle and not easily recognized on the hematoxylin-eosin (H&E)–stained sections. Although a Ki-67 immunostain will typically show a low proliferation index among the neoplastic B-cell population, this stain will highlight numerous proliferating germinal center cells in the reactive follicles that should not be interpreted as areas of transformation ( Fig. 34.15 ). Particularly in cases with plasmacytic differentiation, IHC and/or in situ hybridization (ISH) studies for kappa and lambda can establish light chain restriction ( Fig. 34.16 ). If eosinophilic amorphous extracellular material is noted, a Congo red stain may be useful to evaluate for amyloid.

Trisomies 3, 12, and 18 may be seen in breast MALT lymphomas, although the frequency of trisomy 3 is lower than in MALT lymphomas of the stomach, parotid, and thyroid (33% vs. 60%). MALT1 gene rearrangements [t(11;18)(q21;q21), t(14;18)(q32;q21)], which are frequent among some MALT lymphomas at other sites, have generally not been identified in breast MALT lymphomas, with the exception of one group which found three cases with t(11;18)(q21;q21) and one case with t(14;18)(q32;q21) among the nine cases of primary breast MALT lymphomas assessed. Other MALT lymphoma–associated cytogenetic abnormalities not identified in localized breast lesions include BCL10 [t(1;14)(p22;q32)] and FOXP1 [t(3;14)(p14.1;q32)] translocations.

Treatment and Prognosis

In contrast to DLBCL, patients with localized MALT lymphomas may be managed with local therapy, including intensive site RT and limited surgery. Immunotherapy with rituximab has also been shown to be safe and efficacious in MALT lymphoma, and may be considered in some cases. Observation may also be considered for some patients, especially if the diagnostic biopsy was excisional, although locoregional RT should be considered in the setting of positive margins. Immunochemotherapy may be recommended for advanced stages (3 or 4) or symptomatic disease relapse. If MALT lymphoma coexists with DLBCL, the tumors should be treated as DLBCL.

Breast MALT lymphomas are typically indolent and patients have an excellent OS. In the largest series examining clinical outcomes of indolent breast lymphomas, the OS was 92% at 5 years and 65% at 10 years. However, the 5-year and 10-year progression-free survival (PFS) rates were 56% and 34%, respectively, with up to 50% of relapses occurring within the first 5 years of follow-up.

Prognostic Factors

There is a higher risk of relapse if the management includes only surgery. Most of the relapses are again responsive to treatment and do not affect OS. Late relapses are known to occur with extranodal MALT lymphomas, including breast MALT lymphomas, necessitating long follow-up periods.

Differential Diagnosis

The differential diagnosis includes both specific and nonspecific benign infiltrates as well as other small B-cell lymphomas. Reactive lymphoid proliferations may form tumor-like lesions with a dense inflammatory infiltrate and simulate lymphoma, termed pseudolymphoma in the older literature. Some of the histological features of pseudolymphoma, such as the presence of germinal centers, a polymorphous lymphoid infiltrate, and a predominance of mature lymphocytes, are now recognized as classic features of MALT lymphoma. Some specific benign clinicopathological entities, such as immunoglobulin G4 (IgG4)–related sclerosing mastitis, have also probably been included within this category. In general, reactive infiltrates are composed of a heterogeneous admixture of T and B cells, without overt destruction of the underlying architecture. Although sheets of B cells outside follicles are not usually seen in extranodal locations, they may be seen in some benign breast processes, including lymphocytic mastitis/diabetic mastopathy and cutaneous lymphoid hyperplasia of the nipple (see follicular lymphoma differential diagnosis section). Flow cytometric, cytogenetic, and/or molecular studies may be useful ancillary studies to identify and characterize clonal lymphoid populations.

Lymphocytic mastitis/diabetic mastopathy is an uncommon mass-forming lesion that most frequently occurs in the context of a long history of diabetes mellitus. However, it may also be seen in nondiabetic patients, including those with autoimmune disease, and healthy subjects. Lymphocytic mastitis/diabetic mastopathy is characterized by marked lobular and perivascular lymphoid infiltrates accompanied by dense stromal keloid-like fibrosis and variable numbers of epithelioid fibroblasts. Although the fibrosis may be marked and overlap with IgG4-related sclerosing mastitis, the lack of significant numbers of IgG4+ cells in these cases suggests a different disease process. The lymphoid cells are small in size and lack cytological atypia ( Fig. 34.17 ). Rare larger cells and scattered plasma cells are admixed in some cases. These cases may show a marked predominance of B cells, without follicular structures or germinal center formation; however, cases in which B cells exceed 60% are still more frequently found in MALT lymphomas. Lymphoepithelial lesions may also be seen and are not considered a distinguishing feature of malignancy at this site. If fresh tissue is available, flow cytometric analysis may be performed to assess surface immunoglobulin light chain expression, as a light chain–restricted B-cell population would support a neoplastic process. B-cell–rich infiltrates associated with lymphocytic mastitis/diabetic mastopathy lack clonal immunoglobulin gene rearrangements, so PCR-based molecular studies on formalin-fixed paraffin-embedded tissue (FFPE) may also be useful to exclude a B-cell clone.

Dense fibrosis has also been described in IgG4-related sclerosing mastitis, which is a dense, mass-forming lymphoplasmacytic infiltrate accompanied by stromal sclerosis, loss of breast lobules, and occasional phlebitis. Germinal centers may be seen but may be regressed or have thin mantle zones. IgG4+ plasma cells are present in significantly increased numbers, with more than 50 per HPF and with an IgG4+/IgG+ ratio greater than 40%. Awareness of this entity is important because the IgG4 sclerosing diseases may involve multiple sites, are responsive to steroids, and show a favorable clinical outcome.

Lupus mastitis is a rare manifestation of systemic lupus erythematosus (SLE) or discoid lupus erythematosus that presents as single or multiple subcutaneous or deep breast masses. The lymphocytic infiltrate is typically more extensive than in lymphocytic mastitis/diabetic mastopathy and includes frequent germinal center formation and hyaline fat necrosis. In contrast to lymphocytic mastitis/diabetic mastopathy, dense fibrosis and epithelioid fibroblasts are lacking. Paraffin section immunostains confirm a polytypic plasmacytosis and a mixed chronic inflammatory lymphoid infiltrate with predominantly CD3+ and CD4+ T cells admixed with CD20+ B cells. If these features are identified, particularly hyaline fat necrosis, clinical correlation is essential to establish the correct diagnosis because lupus mastitis may be the initial presentation of SLE.

Although very rare, secondary syphilis may manifest as a cutaneous nodular plaque with a dense lymphoplasmacytic infiltrate and simulate lymphoma, particularly extranodal MALT lymphoma. The organisms can be identified by silver stains or with a specific antitreponemal IHC stain. If numerous plasma cells are noted, IHC or in situ hybridization studies may be extremely useful to exclude monotypic cytoplasmic immunoglobulin expression.

Cutaneous lymphoid hyperplasia (CLH) of the nipple and areolar region may also have a marked B-cell predominance; however, in contrast to lymphocytic mastitis/diabetic mastopathy, the majority of the infiltrate is composed of lymphoid follicles with germinal centers. Interestingly, 47% of CLH cases were associated with Borrelia burgdorferi in a large European study. The B cells are usually expected to be polyclonal in CLH; however, the diagnosis of malignant lymphoma should not rest on a positive molecular study alone, because at least one study has found monoclonal immunoglobulin heavy chain ( IGH ) gene rearrangements in cases believed to represent cutaneous lymphoid hyperplasia and pseudoclonality may occur in lymphoid infiltrates, especially if only a minority of B cells is present.

Differentiation from other small B-cell lymphomas is generally not problematic on the basis of morphological findings and phenotype ( Table 34.2 ). As previously mentioned, difficulties may arise in distinguishing a MALT lymphoma with extensive follicular colonization from an unusual follicular lymphoma, in which case cytogenetic FISH studies for the follicular lymphoma-associated IGH/BCL2 rearrangement may be of value. MALT lymphoma with plasmacytic differentiation may also be difficult to distinguish from lymphoplasmacytic lymphoma (LPL). LPL is predominantly bone marrow based, sometimes with involvement of lymph nodes and spleen. Extranodal breast involvement is very unusual; however, it has rarely been reported as the manifestation of disseminated disease. Clinical correlation may be very important, as well as testing for MYD88 L265P mutation, which is present in about 90% of LPLs and in few marginal zone lymphomas.

Finally, the differential diagnostic considerations may also include distinction from plasma cell neoplasms as some MALT lymphomas may show extreme plasmacytic differentiation. If the monotypic plasma cells show strong cyclin D1 and/or CD56 expression, this would favor involvement by a plasma cell neoplasm rather than a B-cell lymphoma with extreme plasmacytic differentiation. Additional immunostains or ISH studies for IGHs may be useful. Most MALT lymphomas, including those of the breast, express IgM compared with myelomas, which mainly express IgG ( Fig. 34.18 ). However, a minority of breast MALT lymphomas are IgG+, as are most cutaneous marginal zone lymphomas; in addition, IgM+ myelomas occur, so clinical correlation is crucial here as well.