Bethesda proposals for classification of lymphoid neoplasms in mice

A systematic classification of mouse hematopoietic neoplasms was first put forward by Dunn in 1954.1 In this assemblage, tumor types were related to presumed cells of origin, including undifferentiated cells, lymphocytes, granulocytes, “reticulum cells” Types A and B, plasmacytes, and tissue mast cells. Similar names for tumor types can be found in the 1966 Rappaport nomenclature for human hematopoietic tumors.2 Following the identification of T- and B-cell lymphocyte subsets in the 1970s, Pattengale and Taylor3revised the mouse classification scheme to generally parallel the 1974 Lukes-Collins proposal for human hematopoietic neoplasms.4 Within these frameworks, most tumors previously defined as of reticulum cell origin were recognized to correspond to B-cell lineage lymphomas. In 1994, Fredrickson et al5 used the human Kiel classification of 1981⁶as modified in 1988⁷ as the basis for a nomenclature. Among other features, it encompassed previously unrecognized subtypes of mouse B-cell lineage lymphoma, including splenic marginal zone lymphoma (SMZL).8 Most recently, the consensus nomenclatures for human hematopoietic tumors established by the 1995 Revised European-American Lymphoma9 and descendant 2001 World Health Organization (WHO) schemes10,11 were adopted as a model by investigators from the National Institutes of Health in an effort to develop a contemporaneous scheme for classification of mouse lymphoma and leukemia.12-14 

The iterative nature of the lists and definitions of disease entities in the mouse derives from the increasingly well-supported presumption that carefully defined and validated model systems can provide fundamental insights into human disorders with attendant implications for prevention and intervention. The opportunities to extend this concept provided by manipulation of the mouse genome were recognized in the formation of the Mouse Models of Human Cancers Consortium (MMHCC) by the United States National Cancer Institute (http://emice.nci.nih.gov). The Hematopathology Subcommittee of the MMHCC was given the charge of developing a consensus list of hematopoietic neoplasms with descriptions and criteria for diagnosis. The goal was to define disease entities that could be recognized by pathologists and related to human disorders where possible. To meet this challenge, gatherings of international experts in human and mouse hematopathology were convened. The result of their deliberations is a proposed classification of hematopoietic diseases that stratifies disorders according to cell lineage. The resulting major subgroups are, accordingly, lymphoid and nonlymphoid. The lymphoid disorders will be discussed here and the nonlymphoid in a separate article.48 

Most of our present knowledge of mouse hematopoietic neoplasms is based on studies of such diseases in inbred and outbred mice occurring either spontaneously or following induction with irradiation, chemicals, or exogenous infection with murine leukemia viruses (MuLVs).3,5,8,15,16 There is considerable strain dependence of disease types and incidence, with much of the differences being genetically determined. Some of these variations are associated with expression of endogenous MuLV and virus-controlling genes,17,18 although infectious MuLVs are not required for induction or progression of many disorders. These considerations are not obviated for diseases developing in genetically engineered mice (GEM)19 but are likely of less importance. Although the range of lymphoid neoplasms seen in conventional mice is broad, it has been substantially extended through genetic engineering. Some of the newly recognized disorders in GEM recapitulate those of humans with greater or lesser levels of fidelity, whereas others appear not to. Indeed, some of the lymphomas and leukemias seen in GEM have never been seen previously as spontaneous lymphomas in mice. In this regard, it is important to remember that only a limited number of strains have been evaluated in a rigorous manner for the characteristics of spontaneous lymphomas. The information at hand that forms much of the basis for our proposals may thus not be representative of the full range of diseases that develop among unmanipulated mice.

A prominent goal of studying mouse tumors is to develop an understanding of pathogenesis that would provide opportunities for preventing or treating similar disorders of humans. It has, therefore, been important to determine whether diseases in the 2 species are true homologs and deserve identical names. We have found this case is often difficult to make. The reasons are multiple. Prominent among these reasons are fundamental differences between the species in the characteristics of primary and secondary lymphoid organs. Uniquely in mice, extramedullary hematopoiesis continues in the red pulp of the spleen throughout life, and the thymus persists well into adulthood. The splenic marginal zone of mice has also been shown to differ significantly from that in humans.20 

An additional critical difference is that many modalities used to classify human lymphomas have been applied much less routinely or rigorously to studies of mouse tumors. Without these direct comparisons at hand, guesswork and wishful thinking provide insufficient grounds for identifying true parallels. Consequently, although the committee developed a set of recommendations to be used by pathologists and investigators diagnosing these diseases, these proposals will be altered and updated as additional information becomes available. The fundamental elements contributing to classification of lymphomas in the WHO classification are clinical features, morphology, immunophenotype, and genetic abnormalities. The committee concluded that the same types of information should be developed, when possible, in studies of mouse hematopoietic neoplasms.

Committee recommendations for approaches to diagnosing lymphoid neoplasms of mice are not presented here because of space constraints but are provided in supplementary material available on theBlood website; see the Supplemental Data link at the top of the online article. These recommendations should be reviewed in detail by those new to the field of mouse hematopathology studying either spontaneous or induced models of disease. The guidelines may also provide a helpful review for more established investigators. We anticipate that more uniform application of these approaches to diagnosis and classification by the hematopathology community will facilitate the characterization of lymphomas and leukemias and yield descriptions that can be more readily understood by all. The supplementary material provides approaches to necropsies, sample collection and storage, fixation and staining, as well as flow cytometric and molecular evaluations of tumor samples. Critical parameters for distinguishing reactive from malignant processes complete the online package. The terminology used in the textual material of the print version is based on the definitions provided in the supplementary information.

The consensus recommendations of the Hematopathology Subcommittee of the MMHCC for classification of mouse lymphoid neoplasms follow the WHO classification in many respects but use distinct terminology when appropriate. Differences in the schemes have several origins that will be dealt with in describing each of the diseases.

Just as the classification of human lymphomas is a work in progress with the WHO scheme being just the most recent version, we view the proposed classification for mouse lymphoid neoplasms as a statement of where we are at present. The scheme presented in Figure1 will be revised as new data on established diseases, both human and mouse, become available and as new disorders are described.

Table 1 compares the features of the lymphoma/leukemia types and their relation to diseases in humans. Several conventions will be followed in the table. First, the phenotype of mature B-cell lymphomas will be given as sIg⁺ B220 [CD45R(B220)]⁺ CD19⁺ with the recognition that other markers may be useful in distinguishing distinct types. Second, the molecular characteristics of mature B-cell lymphomas, regardless of type, will be presented in Table 1 as if both heavy chain and both kappa light chain alleles have undergone rearrangements [IgH R/R, IgK R/R] with the recognition that only one allele of each may be rearranged and that lambda light chain may be rearranged and used in a subset of the neoplasms. In these diagnostic categories, the T-cell locus will be listed as unrearranged, indicating only the T-cell receptor locus (TCRβ G/G), with the recognition that the locus is rearranged in some B-lineage tumors. An expanded description of the occurrence of lymphomas of each diagnostic category is given in Table 5 of the supplementary material, and a more complete listing will be given at the MMHCC Web site (http://mmr.afs.apelon.com/heme/). Finally, although cytologic details are described, they pertain to the appearance of cells in hematoxylin and eosin (H&E)–stained tissue sections of formalin-fixed, paraffin-embedded tissues. Representative cases of lymphoma types are shown in Figures 2-4.

Hypermutation of immunoglobulin variable region sequences indicates that human lymphomas of this type are of germinal center B-cell origin or have evidence of having passaged the germinal center. Careful studies among medical hematopathologists led to the conclusion that it is difficult to reproducibly distinguish morphologic variants designated centroblastic, immunoblastic, T-cell/histiocyte rich, and other subsets of this disorder defined in pre-WHO nomenclatures.9 The prospect that morphologic variants may in time be associated with specific genetic features defined by using complementary DNA microarray or other technologies led to the suggestion that use of the terminology for the variants remains optional.10 The mouse lymphomas may provide a stronger case for continuing with subset designations in view of highly distinctive morphologic and histologic features of several proposed types (Figure 3A-D). Until such time as these designations can be buttressed by criteria other than histologic appearance, it would seem prudent to keep the use of these morphologic subgroups optional.

There are aspects to distinguishing diffuse large B-cell lymphomas (DLBCLs) from other entities in both humans and mice that deserve comment. First, human follicular lymphomas are defined by a combination of what, for purposes of discussion, can be termed cytologic and architectural features within lymph nodes. The cytologic range of these neoplasms comprises primarily centrocytes and centroblasts. The representation of centroblasts is variable, ranging from less than 5 per high-power field to solid sheets of blasts. The architectural hallmark of this disease is poorly defined follicular structures. The proportion of lymph nodes with follicular structures can vary from less than 25% to complete, the remaining areas being diffusely involved. Retention of follicular architecture may thus be said to be the signature distinction between human DLBCL of centroblastic type and follicular lymphomas comprising exclusively centroblasts.

In mice, the cytologic and architectural features that distinguish follicular from DLBCLs are distinct. First, mouse follicular tumors composed almost uniformly of centrocytes are extremely rare, and more evenly balanced populations of centrocytes and centroblasts tend to be the distinguishing feature (Figure 2E). Second, follicular lymphomas in mice usually arise in spleen rather than lymph nodes, meaning that the architectural-distinguishing features of the diseases are fundamentally different for the 2 species. The appearance of involved spleen is nodular at both the gross and microscopic levels with progressive expansion of multiple or all follicles in the white pulp leading to compression of the red pulp as well as the T-cell periarteriolar lymphocytic sheath. Progression may lead to lymphomas with an almost pure population of centroblasts that remain restricted to the white pulp or spread into the red pulp but with the follicular origin still evident. These are centroblastic lymphomas of follicular origin. For historical reasons, it is worth noting that Pattengale and Taylor3 defined follicular center cell lymphomas as small cell, large cell, and mixed. The large cell type corresponds to what we are here calling DLBCL.

Two other subtypes of centroblastic lymphoma have been observed. The first represents a progression of SMZL to a higher grade with clear anatomic evidence of its origins.8 The other type comprises centroblastic tumors for which an origin from the follicle or marginal zone cannot be inferred.

Recent studies have described clonal B-cell lymphomas with high proportions of nonclonal T cells.21 It has been suggested that these mouse tumors may be the equivalent of human T-cell–rich DLBCL, but the frequencies of T cells have not been determined by flow cytometry and may not reach the levels seen in the human disease.

Second, there are mouse lymphomas that are histologically indistinguishable from precursor B-cell and precursor T-cell lymphomas but comprise mature sIg⁺ B cells.22,23 In the veterinary literature, these lymphomas have been classified together with precursor neoplasms as lymphoblastic lymphoma. The fact that some lymphomas of this type derived from 2 different populations of mice exhibited structural alterations in BCL-6, the most common genetic abnormality of human DLBCL,24 originally was used to argue for their classification as DLBCL, despite their lymphoblastic appearance and the lack of human tumors with this combination of histologic and immunophenotypic features.22,23 With the more recent opportunity to study large numbers of cases of mouse Burkitt lymphoma (Figure 3E),25 it became apparent that a proportion of the sIg⁺ lymphoblastic tumors resembled mouse Burkitt lymphoma histologically while sharing a proliferative fraction approaching 100%. This finding suggests atypical Burkitt or Burkittlike as perhaps more appropriate terminology for these lymphomas (Figure 3F). In the WHO classification, however, this term is reserved for cases with nonclassical cytology despite proven or likely translocations activating MYC.10 MYC is not rearranged in a significant proportion of the mouse lymphomas of this type, weakening an argument for a change in classification terminology from DLBCL. Nevertheless, the common morphologic and behavioral features of mouse Burkitt lymphoma and mouse sIg⁺ “lymphoblastic” tumors suggest that they may share analogous pathogenesis. Therefore, we recommend that the term Burkittlike be applied to mature B-cell neoplasms with proliferative fractions approaching 100% even in the absence of MYC translocation. The term Burkitt lymphoma will be reserved for tumors with compatible morphology, phenotype, and MYC dysregulation because of translocation or genetic manipulation. This area is clearly one in which it can be anticipated that molecular profiling will be of great use in sorting out histologically similar mature B-cell lymphomas.

Finally, although the term plasmablastic lymphoma is used for a variant of human DLBCL, it is subsumed under plasma cell neoplasms in mice (Figure 4). The major reason for this decision is that most mouse plasmacytic lymphomas cover a range of cytologic features, ranging from mature monoclonal plasma cells to plasmablasts in varying proportions. The plasmablastic variant represents one extreme. Human plasmablastic lymphoma, by comparison, typically lacks mature plasma cells and can have a morphologic spectrum that ranges from immunoblastic to Burkitt lymphoma.

Several diseases are recognized as occurring in unusual circumstances. One is found in association with 2 different mutations in a common signaling pathway, and a second is found in response to infection with a unique mixture of MuLVs. These are the diseases of mice homozygous for mutations of Fas or Fasl and mice with the retrovirus-induced immunodeficiency syndrome, MAIDS. These diseases are characterized in their early stages by greatly expanded polyclonal proliferations of lymphocytes that are ultimately replaced by monoclonal populations of B cells. Cells in the early stages of these chronic proliferative disorders do not transplant to immunocompetent or immunodeficient hosts, whereas those cells from late-stage animals can be transferred most readily to mice with impaired immunity. The cells that proliferate in the adoptive hosts are clearly lymphomas, but their presence in the original animal is not often apparent, as the expanded populations of nontransformed cells obscure their presence. The committee was hesitant to classify these lymphomas with those readily diagnosed in the primary host because there was no way to determine the stage of the tumor in the primary host or to know how passage in vivo may have influenced histologic appearance and phenotype. The difficulties associated with defining the MAIDS tumors as true lymphoma have been discussed.26 These disorders and their associations with lymphoma are important areas for further study.

Mice with mutations of Fas or Fasl develop massive lymphadenopathy and splenomegaly due to the polyclonal accumulation of TCRα/β⁺CD3⁺CD4⁻CD8⁻B220^dull“double-negative” T cells.27 Later, up to 60% of mutants from 6 to 15 months of age have clonal populations of B cells (in spleen or lymph nodes) that are not discernible among the mass of double-negative T cells but can be uncovered by transplantation to immunodeficient hosts.28 The tumors are immunoglobulin class switched, but there are no MYC translocations. They also secrete immunoglobulin in the primary animal, on transplantation, and after adaption to tissue culture. The histologic features on transplantation were described as plasmacytoid lymphoma.28 Some humans with mutations in FAS have developed B-cell lymphomas of diverse types, suggesting analogous processes in mice and humans.29,30 

Mice infected with the LP-BM5 mixture of MuLV develop a degree of splenomegaly and lymphadenopathy rivaled only by mice with mutations in Fas and Fasl.31,32Initially, germinal centers are very prominent with large numbers of plasma cells intermixed with centroblasts and increasing numbers of immunoblasts. The disease is initially polyclonal for both B and T cells,33,34 followed by the appearance of oligoclonal populations of B cells more often than of T cells around 8 weeks after infection.33 By 12 weeks, all mice have monoclonal populations of T or B cells that can be transplanted to immunodeficient hosts.35,36 In occasional mice, lymphomas can result in paralysis and death from invasion of the central nervous system.33 Under the proposed nomenclature, the transplants can be described as DLBCL. B-cell lineage lymphomas occur at high frequencies in patients with genetically determined, iatrogenic, or acquired immunodeficiencies such as AIDS,37 providing possible human parallels.

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does the mention of trade names, commercial products, or organizations imply endorsement by the United States government.