Recent revisions of the classification of hematopoietic tumors

The World Health Organization (WHO) classification was revised in 2001, 2008, and 2017, respectively, and has served as a standard guideline for lymphoma diagnosis worldwide. Between 2022 and 2024, as with other organs, the classification of diseases of hematopoietic tumors is being updated to the fifth edition (WHO-5) [1] from the revised fourth edition of the WHO classification (WHO-4R) [2]. However, another international classification for hematopoietic neoplasms, the International Consensus Classification (ICC) [3], is now also being published. Although having two classifications in contemporaneous use is a difficult situation in terms of application to daily practice, understanding the similarities and differences between the two classifications will not only lead to accurate diagnosis of hematopoietic tumors, including lymphoid tumors, but will also be important for understanding the pathophysiology of tumors, leading to an improvement in treatment.

WHO-5 and ICC share the same fundamental concept, namely that a disease entity is defined based on the integration of various findings, including clinicopathological and genetic abnormalities, and no single gold standard is set. This concept has been inherited since the REAL classification [4]. WHO-5 has deleted the classification of provisional entity, and previous provisional entities in WHO-4R have either been promoted to a formal disease entity or removed [1]. Instead of describing a comprehensive list of each disease entity as already described elsewhere [1, 3, 5, 6], this review focused on entities in which significant conceptual changes are recognized in both WHO-5 and ICC from WHO-4R (four issues, Table 1) and those in which there are critical differences in the interpretation between WHO-5 and ICC in B-cell neoplasms (seven issues, Table 2).

Table 1 Commonly revised issues in WHO-5 and ICC from WHO-4R in B cell neoplasms
Full size table
Table 2 Discrepant issues between ICC and WHO-5 in B cell neoplasms
Full size table

Issues commonly revised in ICC and WHO-5 compared to WHO-4R

These changes were reached via strong consensus among current hematopathologists and hematologists and should be introduced in routine diagnosis as early as possible.

Primary cutaneous marginal zone lymphoma (pcMZL): A new organ-specific entity

Whereas WHO-5 and ICC have grouped most non-nodal marginal zone lymphomas (MZLs) together as extranodal MZLs [7], similar to WHO-4R, they both recognize primary cutaneous MZL (pcMZL) as an independent entity, namely the third organ-specific MZL following nodal and splenic MZL (Fig. 1(a)–(f)) [1, 3]. This revision is based on the following findings: (1) pcMZL expresses IgG, a “class-switched type”, especially IgG4 (Fig. 1(d) [8] unlike MZL in other organs that usually express IgM [9], a “non-class switched type”; (2) FAS mutations are found in about 60% of cases [10]; (3) lymphoepithelial lesion, a representative histological feature of MZLs in other organs, is hardly seen in pcMZL [11]; and (4) systemic dissemination or local recurrence is very rare [11]. In terms of mutational profile, primary MZL of the thyroid gland, in which TET2 mutations are present in 65% of cases [7], also appears to be unique, but is not distinguished from MZLs of other organs. pcMZL was recognized as an independent entity because of its clinicopathological features and the previously mentioned specificity of the skin as an immune organ. The recognition of pcMZL as a separate entity may have been encouraged by its clinicopathological features as well as by the concept that skin provides a unique immune environment, described as “skin-associated lymphoid tissue”. [12]

Fig. 1
figure 1

Histological findings of primary cutaneous marginal zone lymphoma. a Dense dermal infiltration of lymphoid and plasmacytoid cells is observed. Lymphoid cells form nodular structures, indicated by an arrow. b Aggregation of plasmacytoid cells. c Lymphoid cells, especially in nodular structures, are positive for CD20. Plasmacytoid cells adjacent to the nodular structures are positive for IgG d and light chain restriction e, f

Full size image

Cold agglutinin disease: recognition as monoclonal lymphoid proliferations

Cold agglutinin disease (CAD), previously considered non-neoplastic lesions, are now recognized as a distinct disease entity as lymphoproliferative disorders (LPD) by both WHO-5 and ICC, as recent studies have revealed their nature as lymphoid proliferation with genetic abnormalities. Histologically, CAD is characterized by the infiltration of lymphocytes and monotypic IgM-positive plasma cells without morphological atypia (Fig. 2(a)–(d)). [13] Although the phenotype of infiltrating lymphocytes/plasma cells is similar to Waldenstrom macroglobulinemia (WM)/lymphoplasmacytic lymphoma (LPL), the infiltration pattern of CAD is usually interstitial in the bone marrow, contrasting with the paratrabecular pattern of WM/LPL. Furthermore, CAD lacks MYD88 mutations and is associated with 3, 12, and 18 trisomy and CARD11 mutations, showing a genetic profile rather more similar to MZL than LPL/WM. [14, 15]

Fig. 2
figure 2

Histological findings of cold agglutinin disease (CAD). a Nodular infiltration of lymphocytes and plasma cells, forming interstitial patterns in the bone marrow. Scattered plasma cells are positive for IgM b and kappa c and negative for lambda d, showing light chain restriction

Full size image

Non-PEL type effusion-based lymphoma: occurrence in older patients but favorable prognosis

Primary effusion lymphoma (PEL) is a disease entity characterized by effusion of body cavities, usually occurring in carriers of human immunodeficiency virus (HIV) [16]. Large lymphoma cells proliferate, confined to effusions and without forming solid masses (Fig. 3(a)). Although these lymphoma cells are of B-cell origin, B-cell markers are frequently diminished (Fig. 3(b)). Partial plasma cell differentiation, such as MUM1 expression and cytoplasmic immunoglobulin production, and co-infection with Epstein-Barr virus (EBV) and human herpesvirus type 8 (HHV-8) (Fig. 3(c)) is frequently identified. On the other hand, similar clinical presentations were also reported in HIV-uninfected patients. Recently, the clinicopathological findings of these HIV-negative cases were summarized: lymphoma cells are characterized as a mature B-cell- rather than plasma cell-phenotype (Fig. 3(d),(e)) with no association with HHV-8 (Fig. 3(f)), and occurring in older patients but showing better prognosis than PEL [17]. These cases are now recognized as a distinct disease entity named fluid overload-associated large B-cell lymphoma (WHO-5) or HHV-8 and EBV-negative primary effusion-based lymphoma (ICC). One discrepant point is that ICC excludes EBV-positive cases while WHO-5 includes them. Whether EBV positivity in the tumor cells is associated with clinicopathological or biological findings remains to be clarified.

Fig. 3
figure 3

Cytological and histological findings of primary effusion lymphoma (PEL) a–c, and fluid-overload associated large B-cell lymphoma (FOLBCL) (d–f). a Large atypical lymphoid cells are observed; some show eccentric nuclei with perinuclear halo, suggesting differentiation towards plasma cells. CD20 b is negative for tumor cells showing positivity for latency-associated nuclear antigen (LANA), a viral product of HHV-8 c In FOLBCL, atypical large lymphoid cells with pleomorphic nuclei (d, Giemsa stain) are positive for CD20 (e) and negative for HHV-8 (f).

Full size image

Gray zone lymphoma: limited to mediastinal lesions

Gray-zone lymphoma (GZL) has been recognized as a tumor intermediate between classic Hodgkin lymphoma (CHL) and large B-cell lymphomas, particularly primary mediastinal large B-cell lymphoma (PMBL). However, it was possible that the gray-zone lymphoma concept would be inappropriately applied to all cases which pathologists found challenging to diagnose. The original concept of GZL was that CHL and PMBL, both occurring in the mediastinum, share a common normal counterpart, namely asteroid B cells usually located in the thymus, and thus some cases could show borderline phenotype or trans-differentiation of CHL and PMBL. [18, 19]

Sarkozy [20] analyzed genetic abnormalities of GZL by dividing them into mediastinal and non-mediastinal lesions and found that non-mediastinal lesions had more TP53 and BCL2 mutations, showing greater similarity to DLBCL. In contrast, mediastinal lesions had a mutation profile similar to PMBL/CHL, such as SOCS1, B2M, and TNFAIP3 mutations. Based on these results, the original concept of GZL was confirmed, namely that the diagnosis of GZL requires a mediastinal lesion and that cases with non-mediastinal lesions are considered DLBCL. Tumor cells of EBV-positive LPD often show a phenotype and morphology intermediate between CHL and DLBCL, but have a distinct mutation profile from GZL, such as frequent STAT3 mutations. [20]

Different concepts between ICC and WHO-5

Several disease entities or subtypes are recognized in either one but not both of ICC or WHO-5. Each entity is based on profound discussion among experts from all over the world. Precise criteria by which to set these entities or subtypes do not exist; rather, they depend on how strictly the authors in charge judge the findings as unique to the particular entity or subtype.

Plasma cell neoplasms: Should classification be based on genomic or clinical findings?

Multiple myeloma, a tumor of plasma cells, is a solid disease entity characterized by unique clinical manifestations such as osteolysis and renal disorder and by the appearance of M protein in laboratory findings [2]. Recent advances in genomic analysis have revealed that translocations, including IGH, account for 40–50% of MM cases, and have a significant impact on clinical findings and gene expression profiles [21, 22]. Translocation partners with IGH include CCND1 and other CCND family genes, MAF family genes, and NSD2. [3] Most of the remaining cases show hyperdiploidy, characterized by trisomy of multiple chromosomes, including chr 3, 5, and 7 [23]. ICC has introduced new subtypes of MM based on these genomic abnormalities and considers cases that do not fit into any of these categories as MM, NOS [3]. In contrast, WHO-5 introduces these genomic abnormalities but does not mention genomic classification (Fig. 4).

Fig. 4
figure 4

Scheme of the difference between WHO-5 and ICC in classifying plasma cell neoplasms

Full size image

Monoclonal gammopathy of undetermined significance (MGUS) has long been recognized as a precancerous lesion of MM. It is defined by a low number of plasma cells (< 10%) and M protein (< 3 g/dl) in the bone marrow, usually showing the absence of CRAB (hypercalcemia, renal dysfunction, anemia, bone lesions), which are typical clinical manifestations of MM. WHO-5 defined monoclonal gammopathy of renal significance as cases showing kidney injury otherwise compatible with MGUS. On the other hand, ICC claimed that such cases could be described as “MGUS accompanied with renal disorders”, and accordingly does not consider it a disease subtype [3]. This discrepancy highlights the difference in stance on the classification between ICC and WHO-5: the former tends to emphasize genomic findings, and the latter tends to favor essential clinicopathological findings (Fig. 4).

High-grade B-cell lymphomas, including double hit lymphoma: significance of BCL6 alterations

Although lymphoma cells in these entities are not always "large" cells, they are included in the umbrella category of large B-cell lymphoma in WHO-5 in the sense that they represent aggressive B-cell lymphoma. Entities associated with high-grade B-cell lymphoma comprise those meeting both morphology- and genetics-based classifications. In terms of morphologic classification, borderline cases between Burkitt lymphoma (BL) and DLBCL were proved to show clinically more aggressive behavior than DLBCL, leading to the definition of a new disease entity of "B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and BL (BL/DLBCL)" in WHO-4 [24]. On the other hand, some experimental mouse models inoculated with cells in which co-expression of BCL2 and MYC—representative oncoproteins in lymphomagenesis—were introduced, showed very aggressive tumor growth and a fatal course [25, 26]. Subsequently, actual human cases with simultaneous rearrangements of MYC and BCL2 were also identified and showed a very aggressive clinical course. These were recognized as double-hit lymphomas. [27, 28] About half of double-hit lymphomas have a BL/DLBCL-like morphology, about 40% are "blastoid types" that resemble lymphoblastic lymphoma, and about 10% resemble DLBCL. [29, 30] Furthermore, as cases with rearrangements of MYC and BCL6 were also shown to have a poor prognosis [31], the genetic concept of "High-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements" was defined in WHO-4R. In contrast, non-DHL lymphomas with BL/DLBCL or blastoid morphology also showed a poor prognosis, like DHL [29, 31], leading to the definition of high-grade B-cell lymphoma, NOS.

Subsequent studies reconfirmed that BCL2/MYC DHL is associated with poor prognosis, while the prognostic significance of BCL6/MYC DHL varied among studies, raising debate about this disease entity [32]. Furthermore, Cucco [33] analyzed the gene expression profile of high-grade B-cell lymphomas and showed that BCL2/MYC DHL was a homogenous group with germinal center B-cell (GCB) or BL-like profile, whereas BCL6/MYC DHL comprised cases with heterogenous profiles, including activated B-cell-like phenotypes indistinguishable from those of DLBCL, NOS. Based on these data, BCL6/MYC DHL is no longer recognized as a uniform disease entity and is included in WHO-5 in DLBCL, NOS or high-grade B-cell lymphoma, NOS. In contrast, ICC retains BCL6/MYC DHL as a disease entity, like WHO-4R.

Follicular lymphoma: Is grading optional or mandatory?

Follicular lymphoma (FL) includes aggressive B-cell lymphomas with large cell morphology like DLBCL and indolent lymphomas like CLL/SLL and MZL. The grading aimed to distinguish them [2, 34]. In WHO-5, grading of FL is no longer mandatory (Fig. 5). The majority of cases with FL1-3A are recognized as indolent lymphoma, named classic follicular lymphoma (cFL), and those with FL3B as follicular large cell lymphoma (FLBCL). This revision is based on criticism of the poor concordance among pathologists in the conventional four-level grading system and the fact that most recent clinical trials of FL set eligibility criteria of Grades 1-3A (Clinicaltrial.gov). [35] As an exception to cFL, WHO-5 defined FL with uncommon features (uFL), which is characterized by large centrocytes and blastoid cells (Fig. 5). Although the large centroblasts frequently observed in FLBCL are not common in cases with uFL, they are characterized by a high frequency of mitosis and high Ki-67 (MIB-1) indexes, associated with a poor prognosis [36]. However, this clinical significance of uFL does not appear conclusive and should be confirmed by further studies. Furthermore, the frequency of uFL varies among studies, namely 2.7% versus 12.3% of all FL cases [36, 37], indicating interobserver discrepancies in the morphological evaluation of uFL. Acknowledgment of this morphological variation and need for interobserver standardization are required to establish the clinicopathological significance of uFL. On the other hand, ICC maintains grading. Regarding ambiguous cases, additional immunohistochemistry and cytogenetic analysis are recommended. Ambiguous cases with t(14;18) (BCL2 rearrangement) and CD10-positivity, namely “typical FL type” are included in Grade 3A, and the ‘‘atypical type’’ into aggressive B-cell lymphoma like grade 3B [3]. Although the ICC method appears more objective than WHO-5, only a limited number of facilities can actually perform the ICC method, which remains a critical concern. [3]

Fig. 5
figure 5

Scheme of the difference between WHO-5 and ICC in the classification of follicular lymphoma. rBCL2, rearrangement of BCL2

Full size image

Lymphoproliferative disorders associated with immune deficiency and dysregulation: Introduction of integrated nomenclature in WHO-5

Although only post-transplant lymphoproliferative disorders (PTLD) and other iatrogenic immunodeficiency-associated LPD (Oii-LPD) were listed as disease entities in WHO-4R [2], LPD associated with immune deficiency and dysregulation (IDD-LPD) is now considered to form a wider spectrum, which includes cases in association with HIV infection, primary immune disorders, and immune senescence by aging [38, 39]. Therefore, WHO-5 recommends an integrated diagnostic nomenclature for IDD-LPD based on workshop discussion [40]. Each IDD-LPD is described in terms of the combination of the three contexts of histology, viral involvement, and cause of immunodeficiency under the umbrella disease category of IDD-LPD (Fig. 6) [1]. Inborn-error immunity is also associated with LPD. Nevertheless, its background genetic abnormalities and clinical manifestations are markedly heterogenous, differentiating it from the IDD-LPD umbrella [40]. EBV-positive diffuse large B-cell lymphoma (EBV + DLBCL), lymphomatoid granulomatosis, and HHV8-related LPDs are also not fully included under the IDD-LPD umbrella and continue to be recognized as separate entities. EBV + DLBCL tends to occur more frequently in the elderly, suggesting that age-related immune senescence could affect lymphomagenesis, which can be considered IDD-LPD [41]. However, since there are presently no objective quantitative methods able to prove immune senescent conditions, it seems challenging to abandon the recognition of EBV + DLBCL as a separate entity and include it under the IDD-LPD umbrella. In contrast, the ICC classification continues etiology-based classification, the same as WHO-4R, defining PTLD and Oii-LPD as representative disease entities of IDD-LPDs. [3]

Fig. 6
figure 6

Scheme of the difference between WHO-5 and ICC in the classification of lymphoproliferative disorder (LPD)-associated immune deficiency and dysregulation. EBVMCU, EBV-associated mucocutaneous ulcer. Oii, Other iatrogenic immunodeficiency

Full size image

Extranodal large B-cell lymphomas: primary large cell lymphoma of immune-privileged sites as an umbrella category

Most cases of DLBCL occurring in an extranodal organ as a primary site were included in DLBCL, NOS in WHO-4R, except for primary large B-cell lymphoma of the central nervous system (PCNSL). Although PCNSL is difficult to distinguish from other large B-cell lymphomas morphologically, it is defined as a distinct disease entity due to the unique characteristics of infrequent involvement of organs other than the brain and specific treatment including MTX, an agent with good migration across the blood–brain barrier to the CNS [24]. In the current revision, ICC newly defines primary testicular lymphoma (PTL) as a unique extranodal large B-cell lymphoma [3]. PTL shows mostly large B-cell lymphoma histology with a non-GCB or ABC profile and high frequency of MYD88/CD79B mutation, classified as the so-called MCD genotype, similar to PCNSL [42, 43]. Furthermore, PTL rarely involves lymph nodes but rather more often the brain. These features separate PTL from other extranodal large B-cell lymphomas, most of which are classified as DLBCL, NOS [44]. On the other hand, WHO-5 further expanded the concept of primary large cell lymphoma of immune-privileged sites as an umbrella disease category comprising brain, testicular, and intravitreal lymphomas (Table 3). [1] Although the detailed mechanisms of immune-privileged organs are not fully understood, several reports support the concept of immune privilege, such as the high transplantation success rate of corneas regardless of HLA. [45]

Table 3 Representative umbrella categaries of WHO-5 in B cell neoplasms
Full size table

The biological background of the association between MCD genotype and immune privilege has yet to be well explained. However, the frequent alterations of several genes involved in antigen presentation, such as the HLA-related gene cluster and B2M, suggest their contribution to immune escape [1, 46]. ICC pointed out that intravascular lymphomas, lymphomas of the adrenal glands, and some extranodal DLBCLs have characteristics similar to PCNSL and PTL, and argued that immune-privileged sites cannot be limited to the central nervous system, testis, and vitreous body. [47]

Transformations of indolent B-cell lymphomas: definition and cutoff issues

The histological transformation of low-grade (well-differentiated) tumors into high-grade (poorly-differentiated) tumors during the clinical course is a common finding as a general characteristic of tumors. Among lymphoid neoplasms, representative examples include Richter transformation of CLL/SLL and histological transformation of FL [48,49,50]. In the current revision, “transformation of indolent B-cell lymphomas” is introduced as a new disease concept in WHO-5. Although clonal commonality should in principle be confirmed between lesions before and after transformation, this is not practical to apply for all cases in the clinical setting. WHO-5 requires this clonal confirmation for CLL only as the prognosis differs between clonally identical Richter’s syndrome and clonally distinct de novo aggressive lymphoma [51]. An integrated score system of clinical findings has also been proposed, such as rapid tumor growth, elevation of tumor markers, e.g. LDH, and the appearance of systemic symptoms [52]. Although not incorporated into the ICC as a disease entity, histological transformation is described in each section of indolent B-cell lymphoma.

Regarding WHO-5, the electronic version published in October 2023 has a chapter titled “Extranodal marginal zone lymphoma” which defines a quantitative criterion for the histological transformation of EMZL as a large cell proportion of over 20% of tumor cells. In contrast, a criterion in the chapter titled “Transformations of indolent B-cell lymphomas” require the recognition of sheet-like proliferation of large cells. Maeshima [53] proposed a large cell proportion of over 30% of tumor cells as the cutoff criterion. Considering interobserver discrepancy between pathologists, as recognized in the grading of FLs, the 20% cutoff value does not appear desirable. The authors agree with the description of “transformations of indolent B-cell lymphomas”, which focuses on sheet growth of large cells and emphasizes the pathologist's judgment (Fig. 7). A correction of this discrepant description has been requested, but the outcome is unclear due to conflicts with the publication schedule (personal communication). This indicates the difficulty of maintaining consistency in situations where the same pathological phenomenon require addressing in different chapters.

Fig. 7
figure 7

Histological transformation of marginal zone lymphomas. a Almost all cells are large with distinct nucleoli and few reactive inflammatory infiltrates, showing large cell sheets. b Even though sheet-like formation is not observed, large cells are predominant, indicating histological transformation. c Most atypical cells are medium-sized nuclei with abundant clear cytoplasm, showing monocytoid features. Histological transformation with these features is unlikely

Full size image

DLBCL, NOS: should molecular classification be introduced?

Large B-cell lymphoma is the most frequent histological pattern among lymphoid neoplasms. Distinct entities of large B-cell lymphomas, whose characteristics or etiology were clarified, have been recognized. These include PCNSL and high-grade B cell lymphoma with BCL2 and MYC translocation, which are differentiated from other large B-cell lymphomas. WHO-5 has 18 disease entities of large B-cell lymphoma. Cases not classified into these entities are regarded as DLBCL, NOS. WHO-5 and ICC have somewhat different stances in their understanding of DLBCL, NOS, a highly heterogeneous disease.

DLBCL, NOS is divided based on gene expression profile into a germinal center B-cell (GCB) type, activated B-cell (ABC) type, and an unclassifiable type. This classification reflects the maturation stage of B cells in the germinal center and is called the cell of origin (COO) classification. The COO classification correlates with prognosis [54], and both WHO-5 and ICC recommend its use in cases of DLBCL, NOS. However, ICC defines molecular subtypes of DLBCL, NOS as GCB and ABC types, whereas WHO-5 does not. Installing a comprehensive gene expression analysis in daily diagnostic practice does not appear practical in most medical institutes.

In contrast, the immunostaining-based Hans classification is a simplified version of the gene expression profile which classifies DLBCL, NOS into GCB and non-GCB types, but cannot identify unclassifiable types. The prognostic impact of Hans classification and its clinical significance is still under debate. [54, 55]

CD5 + DLBCL [56] and BCL2/MYC double expressor lymphoma [57] are also reported to correlate with prognosis, but neither is yet recognized as a unique entity as they do not fully represent the biological subtype. Based on the results of comprehensive genomic sequencing, about 50–60% of DLBCL, NOS can be classified into seven subtypes: MCD/C5, N1, A53/C2, BN2/C1, ST2/C4, MYC + EZB, and MYC-EZB. [43, 58, 59] ICC introduces this genomic classification with a positive attitude, showing several figures which address the possibility of utilizing genomic profiles for the future classification of DLBCL. On the other hand, WHO-5 states that the introduction of genomic classification is premature. As seen in the genomic classification of multiple myeloma and COO classifications, this may reflect the different attitudes between ICC, which is more proactive in classification based on molecular findings, and WHO-5, which tends to focus more on clinical findings.

Future perspective of the classification of lymphoid neoplasms

The existence of two international classifications which are somewhat discrepant might confuse pathologists and hematologists in charge of diagnostic and therapeutic practice. These classifications should be integrated in the next revision, probably in 5–10 years. On the other hand, a positive aspect of this ‘‘split’’ is that two different expert groups have addressed the current pathophysiology of malignant lymphoma from their own perspectives and criticized each other's concepts, and have thereby highlighted currently unresolved issues. The points of disagreement between WHO-5 and ICC, discussed in the latter half of this article, represent such issues and will serve as a starting point for future research.

A second issue with the current revision is that more disease entities/subtypes can be diagnosed only in laboratories and hospitals with sufficient molecular analytical resources, such as fluorescence in-situ hybridization and next-generation sequencing, particularly with regard to the ICC. This is an obstacle to the coherence of diagnosis. Kikkeri [60] reported that a cutoff value of 80% for MYC immunohistochemistry is highly predictive of translocation of the MYC gene. Tsuyama61 generated an accurate evaluation system for BCL2 expression by using reactive T cell expression levels as an internal control, and this is expected to be useful in achieving more precise prognosis estimations and drug selection. Further studies such as these—with a focus on both elevating the level of sophistication of immunohistochemistry while advancing its standardization—are warranted as a means of replacing complex and expensive molecular analyses with simpler methods which are better able to utilize the updated classification.