Volume 262, Issue 3 p. 255-270
INVITED REVIEW
Open Access

The fifth edition of the WHO classification of mature B-cell neoplasms: open questions for research

Sarah E Coupland

Sarah E Coupland

Liverpool Clinical Laboratories, Liverpool University Hospitals Foundation Trust, Liverpool, UK

Search for more papers by this author
Ming-Qing Du

Corresponding Author

Ming-Qing Du

Department of Pathology, University of Cambridge, Cambridge, UK

Correspondence to: M-Q Du, Department of Pathology, University of Cambridge, Box 231, Level 3, Lab Block, Addenbrooke's Hospital, Hills Road, Cambridge, CB2 2QQ, UK.

E-mail: [email protected]

Search for more papers by this author
Judith A Ferry

Judith A Ferry

Department of Pathology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

Search for more papers by this author
Daphne de Jong

Daphne de Jong

The Netherlands Cancer Institute, Amsterdam, The Netherlands

Search for more papers by this author
Joseph D Khoury

Joseph D Khoury

Department of Pathology, Microbiology and Immunology, University of Nebraska Medical Center, Omaha, NE, USA

Search for more papers by this author
Lorenzo Leoncini

Lorenzo Leoncini

Department of Medical Biotechnology, University of Siena, Siena, Italy

Search for more papers by this author
Kikkeri N Naresh

Kikkeri N Naresh

Department of Laboratory Medicine and Pathology, University of Washington, Seattle, WA, USA

Search for more papers by this author
German Ott

German Ott

Department of Clinical Pathology, Robert-Bosch-Krankenhaus, and Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany

Search for more papers by this author
Reiner Siebert

Reiner Siebert

Institute of Human Genetics, Ulm University and Ulm University Medical Center, Ulm, Germany

Search for more papers by this author
Luc Xerri

Luc Xerri

Institut Paoli-Calmettes, CRCM and Aix-Marseille University, Marseille, France

Search for more papers by this author
on behalf of the WHO 5th Edition Classification Project

the WHO 5th Edition Classification Project

Search for more papers by this author
First published: 05 January 2024

Conflict of interest statement: M-QD is an Associate Editor of The Journal of Pathology and SEC is the General Secretary of The Pathological Society of Great Britain and Ireland. No other conflicts of interest were declared.

Abstract

The fifth edition of the World Health Organization Classification of Haematolymphoid Tumours (WHO-HAEM5) is the product of an evidence-based evolution of the revised fourth edition with wide multidisciplinary consultation. Nonetheless, while every classification incorporates scientific advances and aims to improve upon the prior version, medical knowledge remains incomplete and individual neoplasms may not be easily subclassified in a given scheme. Thus, optimal classification requires ongoing study, and there are certain aspects of some entities and subtypes that require further refinements. In this review, we highlight a selection of these challenging areas to prompt more research investigations. These include (1) a ‘placeholder term’ of splenic B-cell lymphoma/leukaemia with prominent nucleoli (SBLPN) to accommodate many of the splenic lymphomas previously classified as hairy cell leukaemia variant and B-prolymphocytic leukaemia, a clear new start to define their pathobiology; (2) how best to classify BCL2 rearrangement negative follicular lymphoma including those with BCL6 rearrangement, integrating the emerging new knowledge on various germinal centre B-cell subsets; (3) what is the spectrum of non-IG gene partners of MYC translocation in diffuse large B-cell lymphoma/high-grade B-cell lymphoma and how they impact MYC expression and clinical outcome; how best to investigate this in a routine clinical setting; and (4) how best to define high-grade B-cell lymphoma not otherwise specified and high-grade B-cell lymphoma with 11q aberrations to distinguish them from their mimics and characterise their molecular pathogenetic mechanism. Addressing these questions would provide more robust evidence to better define these entities/subtypes, improve their diagnosis and/or prognostic stratification, leading to better patient care. © 2024 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Introduction

The conceptual framework and major developments of the fifth edition of the World Health Organization Classification of Haematolymphoid Tumours (WHO-HAEM5) were outlined in June 2022 [1]. That article provided an overview of the most significant changes made in WHO-HAEM5 compared with the revised fourth edition of the Classification (WHO-HAEM4R), with the changes being a product of evidence-based medicine and worldwide and multidisciplinary consultation. WHO-HAEM5 incorporates the most up-to-date medical data available. Those topics that remain challenging due to lack of sufficient evidence represent areas of future research. Hence, the WHO Classification will require regular revision as more knowledge and evidence emerge through the application of new techniques and concepts on clinically well-annotated and (preferably) large case cohorts. The current review aims to highlight some areas within WHO-HAEM5 that could benefit from focused fundamental and/or translational studies, which may lead to even more robust future frameworks for the understanding of lymphoma biology. Here, we discuss a selection of the entities and subtypes of mature B-cell neoplasms considered key ‘next research questions’ of general and significant interest.

Splenic B-cell lymphoma/leukaemia with prominent nucleoli: a newly introduced placeholder

Splenic lymphomas constitute a heterogeneous family/class of lymphoid neoplasms that share overlapping clinical features, laboratory findings, and pathologic characteristics. The clinical behaviour ranges from very indolent to aggressive. Of the entities that typically present with splenomegaly, minimal or absent lymphadenopathy, and a variable leukaemic component and concomitant bone marrow involvement, some are well-defined while the nature of others as distinct entities has become increasingly debatable. To encompass the latter group, the WHO-HAEM5 has coined the term splenic B-cell lymphoma/leukaemia with prominent nucleoli (SBLPN) (Figure 1). As such, SBLPN encompasses many of the splenic lymphomas previously classified as hairy cell leukaemia variant (HCLv) and B-prolymphocytic leukaemia (B-PLL) and possibly some cases of splenic marginal zone lymphoma and splenic diffuse red pulp small B-cell lymphoma (SDRPL) with histologic progression. SBLPN is defined based on cytomorphological features of medium-size to large atypical lymphoid cells with a large prominent single nucleolus reminiscent of that of prolymphocytes and poorly defined cytoplasmic projections reminiscent of hairy cells (Figure 1). Further subclassification is difficult based on the evaluation of bone marrow and peripheral blood alone, and it is unclear whether SBLPN represents one or more distinct entities. The boundaries within this ‘placeholder entity’, if existing, are unclear. SBLPN behaves relatively indolently except for those carrying ‘transformation’ type alterations, generally including TP53 dysregulation (17p deletion, TP53 mutation). This placeholder provides the opportunity to study cases that cannot be placed in the morphologically, immunophenotypically, and genetically well-defined entities, so that evidence-based, biologically meaningful entities can be defined in future classifications; at present, such evidence is not available.

Details are in the caption following the image
Splenic B-cell lymphoma/leukaemia with prominent nucleoli (SBLPN). (A) In peripheral blood smears, SBLPN presents with atypical medium to large lymphocytes with abundant cytoplasm with typically a single prominent nucleolus and fine or poorly defined projections. In a bone marrow trephine, the infiltrate is interstitial and of varying density. The lymphoma cells are positive for CD20 and IgD but, in contrast to hairy cell leukaemia (HCL), negative for cyclin D1 and Annexin A1. (B) Comparison of major recurrent mutations among SBLPN, HCL, splenic marginal zone lymphoma (SMZL), and splenic diffuse red pulp small B-cell lymphoma (SDRPL). *Since SBLPN is a newly introduced placeholder category, formal series are not available and molecular characteristics are critically extrapolated from published series on HCLv and B-prolymphocytic leukaemia (B-PLL) [2-14] but should be interpreted with caution.** In contrast to cases of HCL with BRAF pV600E mutation, cases lacking this mutation are reported to carry a high percentage of MAP2K1 mutations (22%) and predominant IGHV4-34 usage (50%). (C) Based on currently available diagnostic information, the majority of cases formerly classified as B-PLL can currently be recognised as blastoid mantle cell lymphomas or as prolymphocytic progression of CLL. The newly defined SBLPN has absorbed any remaining cases of B-PLL and those cases formerly classified as HCLv. Rare cases of SMZL and SDRPL with similar morphological features cannot always be distinguished from SBLPN in the absence of a splenectomy specimen or supporting molecular evidence since the immunophenotype of flow cytometry is insufficiently characteristic.

Hairy cell leukaemia variant: time for renaming and redefining

Splenic B-cell neoplasms formerly classified as HCLv have some clinical and pathological similarities to HCL. It has become clear, however, that there are fundamental differences between HCL and HCLv, and that HCLv has some features in common with other splenic B-cell lymphomas. Like HCL, HCLv has cytoplasmic protrusions, but, in contrast to HCL, it usually has prominent nucleoli [3]. Bone marrow fibrosis is typically absent, and laboratory findings differ by absence of monocytopenia and higher white blood cell counts. CD103 and CD11c are expressed, but CD25, CD200, and Annexin A1 are negative and CD123 expression is weak or absent. Most importantly, the genetic profile is distinct from HCL. BRAF p.V600E is absent, as are KLF2 and CDKN1B mutations. MAP2K1 mutations may be more frequent (38–42%) [2-4]. In addition, gene expression profiling data, which are available for a limited number of cases, distinguish HCLv from HCL [5]. The clinical course of HCLv is more aggressive than HCL, and patients are generally resistant to conventional HCL treatment, which may be related to the higher frequency of TP53 alterations in HCLv. The genetic profile may provide a rationale for MAPK pathway-targeted treatment approaches as in HCL. This should not be regarded as an argument to consider HCLv as a subtype of HCL since the underlying genetic alterations leading to MAPK pathway activation are different.

B-PLL: sorting out the wastebasket

B-PLL was initially described as a clinically aggressive lymphoma with massive splenomegaly, minimal or absent lymphadenopathy, and a typically high-level atypical peripheral lymphocytosis with >55% prolymphocytes and concomitant bone marrow involvement. Typical B-PLL morphology consists of intermediate-size cells with central nucleoli and indistinct cytoplasm with varying protrusions. In daily practice, subjective interpretation results in large interobserver variation with wide variations in reported prolymphocyte counts. Over the past 20 years, cases of de novo B-PLL have diminished substantially by excluding cases reclassified as cyclin D1-positive (blastoid) mantle cell lymphoma (MCL) [15, 16]. Cases that are now recognised as transformation from underlying or concomitant small B-cell lymphoma, most frequently chronic lymphocytic leukaemia (CLL), are also excluded. Evidence is emerging for cases of transformed marginal zone lymphoma and cyclin D1-negative MCL with upregulation of CCND2 and CCND3 to assume a picture resembling that described for B-PLL [17]. In view of these evolving insights, especially concerning B-PLL as a transformed lymphoma, reported studies on the molecular landscape are difficult to interpret. Reported high incidences of TP53 mutations and MYC alterations (translocations and gains) that are associated with high-risk clinical features at presentation and poor prognosis may rather be indicative of transformation. Other reported genetic alterations are reminiscent of the genetic landscapes of splenic marginal zone lymphoma (SMZL) and CLL and are generally heterogeneous. In aggregate, the data argue against retaining B-PLL as a diagnostic entity, hence its removal from this edition of the WHO classification. Cases with features of B-PLL that cannot be recognised as one of the aforementioned well-defined entities are accommodated in SBLPN, awaiting further study on their precise nature (Figure 1) [18].

Splenic diffuse red pulp small B-cell lymphoma

The infiltration pattern of SDRPL in the spleen is fairly typical, with diffuse infiltrations of splenic cords and sinuses. Architectural destruction by blood lakes lined by neoplastic cells is often present, resulting in pseudo sinuses. The white pulp is atrophic. Clinical and immunophenotypic features and cytomorphology (presence of unevenly distributed cytoplasmic or villous projections) are similar to HCLv. The neoplastic cells of SDRPL have small or absent nucleoli, in contrast to HCLv [8-10]. Without a splenectomy specimen, the reliability of the diagnosis of SDRPL in reference to SMZL and SBLPN is highly debatable. The mutational landscape has major overlap with cases that are now included in SBLPN but lack features characteristic of splenic marginal zone lymphoma (SMZL) and HCL [6, 7]. In SDRPL, CCND3 mutations have been reported as the most frequent finding (21–24%) [6, 11, 12]. The precise biological relation between SDRPL and SBLPN remains to be elucidated.

Splenic marginal zone lymphoma

The definition of SMZL has not changed in recent years since mounting evidence from studies adding to biological characterisation has largely confirmed existing insights concerning cell of origin (COO) and common oncogenetic features. The marginal zone B cell remains the presumed COO. The characteristic patterns of infiltration in the spleen seen in most cases for which a splenectomy specimen is available further support a separate nature of SMZL. Epidemiological and clinical evidence supports a role for hepatitis C virus and autoimmunity in the pathogenesis of subsets of SMZL [19]. The latter is supported by the preferential usage of the IGHV1-2*04 gene. Two genetic profiles can be distinguished. The first is the so-called NNK-SMZL, which is dominated by NF-κB activation (TNFAIP3, TRAF3, BIRC3 mutations), NOTCH, and KLF2 mutations and is enriched for IGHV1-2*04 usage [6]. This profile is distinct from other small B-cell lymphomas. The second, DMT-SMZL, shows more overlap with other small B-cell lymphomas and includes mutations in DNA damage response genes (e.g. TP53, ATM), characteristic MAPK pathway activation (BRAF), and Toll-like receptor signalling genes (MYD88). Currently, clinical, histomorphologic, and immunophenotypic features argue for considering SMZL as a single entity, while the interpretation of genetic subtyping awaits further data.

Hairy cell leukaemia

HCL is another well-defined and distinct entity among splenic leukaemias/lymphomas. The immunophenotype is characterised by bright expression of CD200, CD103, CD123, CD25, CD11c, and Annexin A1. Expression of CD10, CD5, CD23, and SOX11 is typically absent. Rare CD5- and CD10-positive cases have been described. More than 95% of cases carry a BRAF p.V600E mutation, with KLF2, CDKN1B,and KMT2C also frequently mutated [13, 14]. BRAF p.V600E-negative cases frequently carry MAP2K1 mutations as an alternative route for MAPK pathway activation (22%). Integration of clinical, pathological, and molecular parameters can generally distinguish HCL from other entities, and the underlying biology serves as a basis for targeted treatment directed at the MAPK pathway [20].

Questions to be answered

In summary, acknowledging a major evidence gap for meaningful classification into biologically distinct entities, the descriptive name of splenic B-cell lymphoma/leukaemia with prominent nucleoli was established. SBLPN absorbs cases previously classified as HCLv and CD5-negative B-PLL as well as rare (progressed) cases of SMZL or SDRPL that cannot be recognised as such in the absence of a splenectomy specimen or in the absence of documented prior low-grade splenic B-cell lymphoma (Figure 1, Table 1). With this new nomenclature, a clear break is made with the past to avoid confusion with acknowledged entities, to stimulate open-minded research, promote better description of relevant cases, interpret emerging patterns, and address deficiencies in data to inform a more meaningful classification of splenic B-cell lymphomas/leukaemias in the future.

Table 1. Summary of major uncertainties and questions to be investigated.
Splenic B-cell lymphoma/leukaemia with prominent nucleoli (SBLPN)
○ What are the ‘true’ entities within the placeholder term of SBLPN?
○ What are the biological boundaries between SBLPN and existing splenic lymphoma entities, and what is the best way to establish a comprehensive set of defining criteria for their diagnosis and differential diagnosis?
Follicular lymphoma (FL)
○ What is the best way to define entities and subtypes within the family of FL and their boundaries to other mature B-cell lymphomas, specifically marginal zone lymphoma?
○ What are the molecular mechanisms involved in the pathogenesis of BCL2-uR FL, including those with BCL6-R?
○ Is it possible to establish a subclassification of FL (including BCL2-uR FL) integrating the emerging new knowledge on various normal germinal centre B-cell subsets?
DLBCL/HGBL with MYC and BCL2 rearrangements (DLBCL/HGBL-MYC/BCL2)
○ Why does the MYC translocation partner have significant impact on clinical outcome, with IG conferring significantly inferior survival than non-IG partner?
○ What is the spectrum of the non-IG gene partners of MYC translocation, their effects on MYC expression and clinical outcome?
○ What is the best way to discriminate MYC translocation with different biological potential and clinical impact and translate these into routine clinical application?
DLBCL with MYC and BCL6 rearrangements (DLBCL-MYC/BCL6)
○ How frequent are MYC/BCL6 double hits due to a direct MYC/BCL6 genomic fusion, and what is their biological consequence? If causing transcriptional activation, which gene is transactivated?
○ What is the best way to assess the genomic configuration and biological consequence of MYC and BCL6 rearrangements and classify this heterogeneous group?
High-grade B-cell lymphoma, not otherwise specified (HGBL, NOS)
○ Does ‘true’ HGBL, NOS exist as an entity and, if so, how are the boundaries defined from its mimics?
High-grade B-cell lymphoma with 11q aberrations (HGBL-11q)
○ What are the morphological and immunophenotypic spectrum and boundary of HGBL-11q?
○ How should cases with concurrent MYC-R and 11q aberrations be classified?
○ What are the molecular and genetic bases of HGBL-11q?

Follicular lymphoma and its many faces

Follicular lymphoma (FL) has been well defined since the Kiel classification [21]. The core of the definition has remained largely unchanged and consists of a description of morphology with at least in part a follicular architecture, a mixture of centrocytes and centroblasts, a germinal centre (GC) immunophenotype, a typical clinical presentation of predominantly nodal disease, and a generally indolent clinical course. Nonetheless, variations in histological presentations exist, and several subtypes have been introduced in WHO-HAEM5 to recognise these differences. Classic FL (cFL) is the major subtype, comprising cases with typical follicular growth patterns of a mixture of centrocytes and centroblasts, which are traditionally graded as 1, 2, and 3A. Such histological grading is considered optional in WHO-HAEM5 due to its lack of reproducibility and insufficient prognostic value under current immunochemotherapy [22-26]. Robust investigations using highly reproducible technology (i.e. artificial intelligence-based morphological prognostic algorithms) are required to accurately quantify and systematically explore various morphometric features to establish an optimal grading system for prognosis in the context of current immunochemotherapy. The other histologic subtypes include FL with unusual cytological features (resembling blastoid cells or large centrocytes), FL with predominantly diffuse growth pattern, and follicular large B-cell lymphoma (FLBCL) that is introduced to replace the former FL grade 3B. Although the majority of cFL exhibits a relatively uniform genetic landscape, including t(14;18)/IGH::BCL2, a sizable minority lacks this genotype, as is also the case for other entities and subtypes within the family of FL (Figure 2).

Details are in the caption following the image
Follicular lymphoma and its heterogeneous features. (A) Comparison of major genetic changes between FL with and without BCL2-R [27-38]. *Approximate frequency of alterations; some alterations may be subclonal. (B) Underappreciated relationship between germinal centre B-cell subsets and FL and its heterogeneous presentations. FL develops from a stepwise acquisition of cooperating genetic events, and these genetic changes dysregulate the B-cell maturation/differentiation process in the germinal centre and, consequently, ‘imprint’ their effects on their phenotype and biological properties, together with the microenvironmental milieu. BCL2-R, BCL2 rearranged; BCL2-uR, BCL2 un-rearranged; FCRL, Fc receptor like.

BCL2 rearrangement negative follicular lymphoma: searching for alternative oncogenic mechanisms

The t(14;18)(q32;q21) results in a BCL2 rearrangement (BCL2-R) via IGH::BCL2 juxtaposition, the genetic hallmark of FL, and is found in ~85% of cases [31-34], much higher in an advanced stage (stage III/IV, 90%) than an early stage (stage I/II, 50%) FL [32, 33]. The translocation drives aberrant BCL2 expression, and this prevents B cells from undergoing apoptosis and prolongs their stay in the GC and their iterative GC reactions. As a result, translocation-bearing B cells are relentlessly exposed to somatic hypermutation (SMH) and IGH class switch recombination activities and are at high risk of acquiring genetic changes [39-41].

Around 10–15% of nodal cFL lack BCL2-R [35], and their pathobiology remains to be fully characterised (Table 1). The morphological features of most BCL2 un-rearranged (BCL2-uR) FLs are indistinguishable from BCL2-R FL. Overall, BCL2-uR FLs display copy number alterations (CNAs) and mutation profiles similar to those of BCL2-R FL [36, 37], with the exception of frequent STAT6 (and, to a lesser extent, CREBBP and TNFRSF14) mutations but infrequent KMT2D mutations in the former (Figure 2) [32, 36, 38]. There is accumulating evidence that inactivation of CREBBP and/or TNFRSF14 by mutations/deletions act in synergy with STAT6 activation in the pathogenesis of BCL2-uR FL [38].

BCL2-uR FLs in the inguinal region have been described since 2009 and often show a predominantly diffuse growth pattern, typical expression of CD23, deletions in 1p, frequent STAT6 mutations, and a low number of CNAs and are associated with localised disease and indolent evolution [36, 42, 43]. The typical morphological, genetic, and clinical characteristics justify recognition as a separate subtype, namely FL with predominantly diffuse growth pattern, under the entity of FL in WHO-HAEM5. However, the full characteristics of this subtype and its molecular pathogenesis remain to be further refined. Specifically, whether and which molecular mechanisms mediate apoptosis protection in BCL2-uR FL remains to be elucidated [31]. A mechanism involving BCL-xL/BCL2L1 upregulation was suggested on the basis of the observation of subclonal STAT6 mutations with concurrent SOCS1 loss [44]. Activation of the STAT6/ IL4 pathway may be involved to sustain the growth of FL with STAT6 mutations [45]. IL4-producing cells like TFH cells from the tumour microenvironment (TME) are indeed known to favour the proliferation of neoplastic B cells [46]. The possible influence of the TME in BCL2-uR FL growth is also supported by frequent CREBBP and CIITA co-deletion/mutation, pointing to a putative immune evasion process [44]. In addition, it is noteworthy that a role for autoimmunity as a proliferation driver in some FL cases has been suggested due to occasional self-antigen recognition by FL B-cell receptors (BCRs) [47, 48]. In conventional BCL2-R FL, new N-glycosylation sites are commonly acquired in the rearranged IGH genes, and the mannosylated BCR binds the adhesion molecule DC-SIGN expressed by macrophages, resulting in chronic BCR signalling [49]. In contrast, advanced-stage BCL2-uR FL showed a lower frequency of acquired N-glycosylation sites in their rearranged IGH genes, whereas a certain proportion displayed a biased usage of IGHV4-34 [50]. As IGHV4-34 is known to encode autoantibodies, such BCL2-uR FL might use responsiveness to auto-antigens as an alternative mechanism of BCR stimulation to sustain cell proliferation. The putative relationship between BCL2 translocation and BCR signalling deserves further exploration.

BCL6 rearrangement in FL: an enigma in pathogenesis

Translocations affecting 3q27, resulting in BCL6 rearrangement (BCL6-R), are found in many different types of indolent and aggressive B-cell lymphomas, including ~20% of BCL2-R-positive cFL [31, 35, 51], 22–35% of BCL2-uR FL [31, 34, 36, 51], 40% of FLBCL, and ~30% DLBCL, not otherwise specified (NOS) [52] (Figure 2). The FL subtype with unusual cytological features was reported to harbour BCL6-R more frequently than cFL [53]. It is also worth noting that FLBCL is more closely related to DLBCL, NOS in both biological and clinical terms [52].

Among BCL2-uR FL, BCL6-R-positive cases present at more advanced clinical stages, mostly with a classic FL cytology and growth pattern [36], but less frequent CREBBP and STAT6 mutations than those without BCL6-R [32] (Figure 2). Data are emerging that especially localised stage BCL2-uR FL represent a biologically separate group characterised by STAT6 and CREBBP mutations as primary drivers. In contrast, the pathogenesis of advanced-stage BCL2-uR FL is different and more determined by BCL6 dysregulation [32].

BCL6 is a master transcription regulator, orchestrating the GC reaction and coordinating GC B-cell differentiation. BCL6 drives the GC process and prevents GC B cells from plasma cell differentiation by transcriptional repression of BLIMP1 (Figure 2). Through these concerted actions, BCL6 contributes to FL development. Apart from IGH, BCL6 has >40 promiscuous translocation partners, including BCL2 and MYC, largely due to erroneous SHM and class switch activities [54]. It remains unclear whether all these BCL6 translocations with different partners are driver events, causing constitutive BCL6 expression, and how such genetic events contribute to different FL types at a molecular level.

FLs lacking BCL2 and BCL6 rearrangements: clues for a variant cell-of-origin?

FLs lacking both BCL2-R and BCL6-R show evidence of SHM, a characteristic feature of the GC process. However, in contrast to BCL2-R-positive FLs, a small subset of BCL2-uR FLs shows gene expression signatures reminiscent of late/post-GC cells, displaying increased IRF4/MUM1 and enrichment for post-GC pathways, including NF-κB signalling [31, 55]. These data suggest that the postulated normal counterpart of these BCL2-uR cases might differ from that of BCL2-R-positive FLs. This hypothesis is supported by miR profiling, showing similar downregulation of miR subsets in BCL2-R-negative FL and normal GC cells undergoing maturation to become plasma cells [55]. It is also supported by some reports of BCL2-R-negative FL transforming into activated B-cell-like DLBCL (ABC-DLBCL) [51].

Putative normal cell counterparts of FL subtypes and clinical implications

A subclassification based on normal cell counterparts has long been considered as irrelevant to FLs. Overall, FLs were indeed generally thought to resemble normal GC light zone (LZ) B cells, given the observed similarities in their gene expression profiles [56]. Nonetheless, some clues pointing to a more heterogeneous nature of a putative COO classification of FLs had been shown by epigenetic data [57]. More recently, this hypothesis was further supported by the characterisation of discrete steps of plasmablastic differentiation as well as the identification of a distinct subpopulation composed of memory B-cell precursors within the GC (Figure 2) [58, 59].

Transcriptomic studies of FL have identified gene expression signatures related to a putative normal cell counterpart, such as the ICA13 signature, which is a feature of the GC dark zone (DZ) [60] and associated with poor prognosis. Similarly, FLs with high FOXP1 are enriched in DZ- and ABC-related gene sets, correlated with unfavourable outcomes [61]. Furthermore, a meta-analysis study reported that molecular subtyping of FLs into GC- versus ABC-like subgroups has prognostic significance [62]. Accordingly, single-cell mass cytometry analysis of FL samples revealed two recurrent patterns, closely resembling GC B cells and post-GC memory B cells [63].

In a recent RNAseq study of FL, unsupervised clustering identified a signature enriched for genes expressed in ABC-DLBCLs and normal memory B cells and a signature enriched for genes expressed in GCB-DLBCLs [64]. These gene expression signatures subclassify FLs into ABC/memory B-cell-like and GCB-like subtypes, with the former displaying shorter survival following immunochemotherapy [64]. The hypothesis of distinct FL subgroups is also supported by genome sequencing showing DLBCL-like FL and contrained FL with distinguished mutation patterns, biological and clinical characteristics [65].

Nevertheless, the hypothesis of only two COO-based FL subtypes is hardly consistent with the heterogeneity of the GC compartment, which comprises multiple distinct subpopulations with different levels of relatedness to DZ and LZ cells, as well as to memory B-cell precursors [58, 66]. Moreover, FL cells represent a continuum expanding far beyond the dichotomy of classic LZ and DZ phenotypes [67, 68]. Therefore, it is inconceivable how numerous normal GC subpopulations could give rise to only two subtypes of FL, as reported thus far. The in-depth characterisation of other putative subtypes represents a stimulating field for future investigations.

DLBCL/HGBL with MYC and BCL2 rearrangements

WHO-HAEM5 has redefined the WHO-HAEM4R entity of high-grade B-cell lymphoma (HGBL) with MYC and BCL2 and/or BCL6 rearrangements (frequently also called double-hit and triple-hit lymphomas, though this name is obviously ambiguous) [1]. WHO-HAEM5 excludes cases with concomitant MYC and BCL6 rearrangements (without BCL2-R) from this entity, as these MYC-R and BCL6-R DH cases are genetically heterogeneous and as a group are essentially different from those with MYC-R and BCL2-R. Consequently, WHO-HAEM5 renames the entity diffuse large B-cell lymphoma/high-grade B-cell lymphoma with MYC and BCL2 rearrangements (DLBCL/HGBL-MYC/BCL2) with the morphological connotation to recognise their variable morphology (Figure 3). The definition of this entity is therefore based on the recognition of structural chromosome aberrations with breakpoints at MYC and BCL2 loci. DLBCL/HGBL-MYC/BCL2 may harbour an additional BCL6 translocation as these cases are like those with MYC/BCL2-DH and overwhelmingly of the GCB subtype, showing a gene expression profile similar to the characteristics of DZ B cells. Rare cases of DLBCL/HGBL-MYC/BCL2 may show intermediate or blastoid morphology and variable TdT expression [69]. These cases bear a mature B-cell phenotype (CD34 negative) and show a breakpoint distribution pattern and mutation profile like those of transformed FL [54, 70]. This would distinguish them from the precursor B-cell neoplasms designated as B-lymphoblastic leukaemia/lymphoma with MYC rearrangement grouped within the entity of the B-lymphoblastic leukaemia/lymphoma with other defined genetic alterations [71], which is considered a very rare entity virtually restricted to the young age group. Irrespective of morphological and immunophenotypic variations, most if not all DLBCL/HGBL-MYC/BCL2 biologically result from high grade transformation of BCL2-translocation positive clones, which may present as a FL or occult in situ follicular B-cell neoplasms [69].

Details are in the caption following the image
Algorithm for classification of aggressive mature B-cell lymphomas in WHO-HAEM5 in light of MYC, BCL2, and BCL6 rearrangement and complex 11q gain/loss patterns. HGBL, high grade B-cell lymphoma; R, rearranged; uR, un-rearranged.

Despite being a seemingly homogeneous entity, there is mounting evidence pointing to variable clinical outcomes and disparate clinical and biological impacts inferred by MYC translocations with different partners. The clinical outcome of DLBCL/HGBL-MYC/BCL2 varies considerably. Some studies convincingly demonstrate that DLBCL-DH/TH with IG::MYC show inferior survival compared to those with non-IG::MYC, with the latter group showing no significant difference in their overall survival from other DLBCLs, including those negative for MYC translocations [72, 73].

MYC translocation: the enigma of non-IG partners and their biological and clinical impact

Based on their COO signatures, DLBCLs can be classified into ABC and GCB subtypes, with the latter in general showing a more favourable survival. A subset of GCB-DLBCL, however, is clinically more aggressive and characterised by gene expression signatures associated with high MYC and DZ-like features, known as molecular high grade (MHG) or double-hit signature (DHITsig) [74, 75]. On the other hand, ~25% of DLBCL with MYC translocations, including those with MYC/BCL2-DH or MYC-SH (single hit), are conventional GCB but not MHG subtype by gene expression profiling [74]. These DLBCLs appear to show survival comparable to conventional GCB-DLBCLs [74]. In line with these observations, only ~45% of DLBCLs with MYC translocations show strong MYC protein staining in >70% lymphoma cells by immunohistochemistry, and only those with high MYC expression are associated with more adverse clinical outcome [76, 77]. The molecular mechanisms underlying such highly variable clinical outcome and heterogeneous MYC mRNA and protein expression among DLBCLs with MYC translocations are unclear.

It has been perceived that MYC translocation dysregulates its transcriptional control due to super-enhancer effect or promoter substitution by its partner genes. It is unclear why MYC translocation in certain cases fails to cause enhanced MYC mRNA and protein expression. Some of the MYC translocation-positive cases with high MYC mRNA but without strong MYC protein staining bear mutations that impair the epitope for MYC antibody binding, giving a false weak or negative MYC protein expression by immunohistochemistry [78]. A proportion of MYC translocation-positive cases truly lack high MYC mRNA and protein expression [76, 77], and in such cases the translocation may not place the MYC gene in a transcriptionally highly active region [79].

Among DLBCL/HGBL-MYC/BCL2, IG::MYC accounts for 55–58% of cases, commonly involving the IGH and infrequently the IGL or the IGK loci as the translocation partner. In such cases, MYC is constitutively activated by the strong transcriptional effect of the IG gene super-enhancer (Figure 4). Such super-enhancer-mediated transcriptional activation, unlike promoter substitution, is independent of the genomic orientation of the MYC and IG genes and to a certain extent also of the ‘linear’ distance between the two genes [80]. In addition, MYC protein half-life is frequently prolonged by somatic mutations that affect its N-terminal phosphorylation sites, consequently impairing proteasome-mediated MYC degradation [81].

Details are in the caption following the image
Examples of DLBCL with MYC translocation in association with different partners. (A) Super enhancer hijack: MYC is placed in vicinity of active IGH super enhancer, which drives MYC overexpression. (B) Promotor substitution: MYC expression is under the transcriptional control of HNRNPA1, a ubiquitously expressed RNA binding protein. (C) MYC is fused with BCL6 in an opposite orientation. There are no structural alterations in the 5’ MYC transcriptional regulatory region, although uncertain on the transcriptional effects by super enhancer downstream of MYC and in the translocated BCL6 region. Nonetheless, this translocation does not cause high MYC expression (unpublished data from Du lab, May 2023).

It remains unclear whether MYC is similarly dysregulated in non-IG::MYC translocation cases. The full spectrum of non-IG::MYC translocation loci remains to be investigated, although several non-IG partners have been identified, including BCL6, ZCCHC7, and RFTN1 [54, 82, 83]. The impact of these non-IG::MYC translocations on MYC transcription control is largely unclear (Table 1). Breakpoint analysis of MYC translocations revealed that not every non-IG::MYC translocation places MYC in the vicinity of an active super-enhancer or yields a genomic configuration that results in promoter substitution causing constitutive MYC expression (Figure 4, unpublished data from Du lab, May 2023). This may potentially explain the previously mentioned intriguing findings, including lack of high MYC level in certain MYC translocation-positive DLBCLs, and why the clinical outcomes of DLBCL-MYC/BCL2 are heterogeneous and depend on the MYC translocation partner IG::MYC versus non-IG::MYC [72, 73, 76, 77].

Given the preceding discussion, interphase FISH is an imperfect tool to investigate MYC translocation as it does not provide information on MYC transcriptional activation, particularly in non-IG::MYC translocations that account for 40–45% of MYC translocation in DLBCL/HGBL [72, 73, 81]. This represents a critical knowledge gap, and a comprehensive characterisation of non-IG::MYC translocations and their effect on MYC expression and impact on clinical outcome would improve the precision of using MYC translocation as a prognostic marker and a selection criterion for personalised treatment.

For current routine histological diagnosis, interphase FISH remains the most widely used test to investigate MYC translocation. It is a practical dilemma whether to investigate all cases of DLBCL/HGBL or only those with GCB-COO or high MYC protein and, if the latter, to define the cut-off value of MYC expression for case selection [76]. Nonetheless, it is relevant to routinely investigate whether MYC translocation is associated with IG (both heavy and light chain) loci given the significant associations with adverse clinical outcome. In the absence of knowledge of genomic configuration, the prognostic potential of non-IG::MYC translocation is better interpreted in conjunction with MYC protein expression level.

DLBCL with MYC and BCL6 rearrangements

As mentioned earlier, WHO-HAEM5 excludes cases with MYC/BCL6 from the DLBCL/HGBL-MYC/BCL2 entity, the same approach as in the International Consensus Classification [1, 84]. These cases are included as a genetic subtype of DLBCL, NOS or HGBL, NOS depending on their morphological features. DLBCL with MYC/BCL6 translocations are heterogeneous in their molecular subtype, with 31, 31, 23, and 15% being GCB, ABC, unclassified, and MHG, respectively [81]. These cases also show a mutation profile clearly different from that of DLBCL/HGBL-MYC/BCL2 [81]. In addition, the clinical outcome of DLBCL with MYC/BCL6 translocations varies and is controversial among different studies [73, 85-87], reflecting at least in part the heterogeneous biology and the small number of cases investigated in individual studies.

BCL6 translocation has ‘promiscuous’ partners, with more than 40 genomic loci identified so far. Among these, MYC is one of the BCL6 translocation partners, which may exhibit as t(3;8)(q27;q24) or more complex t(3;8;14)(q27;q24;q32) by classic karyotyping [86, 88]. By interphase FISH with MYC/BCL6 fusion probes, ~30% of DLBCLs with MYC/BCL6 translocations are due to a direct genomic fusion between the MYC and BCL6 loci (Figure 4, unpublished data from Du lab, May 2023). As the genomic configuration of the translocation is not deciphered, intervening IG or other partner sequences cannot be excluded, and the potential effect on MYC and BCL6 are unclear (Table 1) [89]. In addition, the partners of MYC and BCL6 translocations in the remaining cases without MYC/BCL6 fusion are largely unknown and, hence, again uncertain on their effects on MYC and BCL6 expression. Similar to MYC translocations in DLBCL, not every case with BCL6 translocation shows high BCL6 expression, although this has not been comprehensively investigated.

MYC and BCL6 translocations are most likely acquired due to the off-target effects of SHM and class switch recombination when B cells undergo GC reactions and are likely a secondary event in DLBCLs [77]. Many of the point mutations introduced by the SHM process, including those in well-known lymphoma-associated genes, are not necessarily pathogenic or driver changes. This notion may also apply to structural alterations. It is likely that not every MYC and BCL6 translocation, particularly those not involving IG as the translocation partner, is an activation change resulting in their enhanced expression. Addressing this critical question will provide a molecular basis for distinguishing translocations with disparate pathogenic potential and their precision application in lymphoma classification and prognosis.

Apart from translocation, BCL6 function may also be dysregulated by somatic mutations that occur in transcriptional regulatory regions and thus impair their binding by transcriptional repressors (BLIMP1, BCL6), thereby preventing BCL6 downregulation [90, 91]. It remains to be investigated to what extent these transcriptional regulatory regions’ mutations enhance BCL6 expression in primary lymphoma cells and impact their molecular profile.

High-grade B-cell lymphoma, NOS

HGBL, NOS represents a heterogeneous entity of aggressive mature B-cell lymphomas with blastoid morphology or features intermediate between Burkitt lymphoma (BL) and DLBCL. Its definition has not changed substantially from WHO-HAEM4R to WHO-HAEM5. The cellular morphology is variable, exhibiting more pleomorphisms in nuclear size and nucleolar content than is generally acceptable for BL. The cytoplasm is usually not as basophilic as in BL, and some cases lack this feature, which can be best appreciated in Giemsa stains that highlight blastoid features optimally. Cytoplasmic vacuoles are few or absent in cytological imprints [92, 93]. A starry sky pattern due to macrophages with apoptotic bodies may be present, as well as many mitotic figures. Cohesive growth is usually absent. By definition, HGBL, NOS does not harbour a MYC rearrangement in combination with BCL2-R.

HGBL, NOS is rare, representing <2% of lymphomas, and interobserver variability in assessing the diagnosis is widely recognised. Following a consensus pathology review of the Lymphoma/Leukaemia Molecular Profiling Project (LLMPP), nearly half of HGBL, NOS were reclassified as DLBCL (44%) or BL (5%) [94]. Therefore, the diagnosis should be made with caution and only when it is impossible to confidently classify a case as DLBCL or BL. In cases suspicious of HGBL, NOS, the differential diagnosis may include a range of lymphoid tumours composed of medium-sized cells (Figure 5). Lymphoblastic leukaemia/lymphoma and the blastoid variant of MCL need to be included in the differential diagnosis and should be excluded [95]. Lack of TdT and CD34 excludes a lymphoblastic leukaemia/lymphoma, and lack of cyclin D1 (largely) excludes MCL. The differential diagnosis with BL may be difficult. Variation in the size and shape of the nuclei can be accepted in BL, and cases with such morphological characteristics can still be diagnosed as BL if they show a combination of an isolated IG::MYC fusion and a typical immunophenotype (CD10+, BCL6+, BCL2−, Ki67 >95%) [96]. Such cases often show a molecular BL or intermediate gene expression profile between DLBCL and BL and are associated with an excellent prognosis [97, 98]. This is particularly true for some paediatric lymphomas that may also show morphological features intermediate between DLBCL and BL and are better classified as BL. On the other hand, cases with strong expression of BCL2 and that lack CD10 expression generally have a high level of cytogenetic abnormalities or other characteristics unusual for BL and are better classified as HGBL, NOS [99].

Details are in the caption following the image
Separation of HGBL, NOS from similar entities. While there is a possible morphological overlap between HGBL, NOS, HGBL-11q, BL, DLBCL/HGBL-MYC/BCL2, and some forms of DLBCL, NOS, these entities are clearly defined and separated by their genetic constitutions.

Another entity to be differentiated from HGBL, NOS is high-grade B-cell lymphoma with 11q aberrations (HGBL-11q), characterised by gains in 11q23.2-q23.3 in concert with telomeric deletions of 11q24.1-qter. Finally, cases of otherwise typical DLBCL, NOS harbouring an isolated MYC translocation should be classified as DLBCL, NOS.

Few data on the molecular landscape of HGBL, NOS are available, since these cases are combined with DLBCL/HGBL-MYC/BCL2 in most reports. Isolated MYC rearrangements have been reported in 8–58% of cases (Figure 3) [77, 96, 100, 101] and MYC amplification in 32% [100]. MYC amplification concurrent with IGH::BCL2 or amplification of both oncogenes has been described [102]. Translocations of BCL2 and BCL6 are reported in 10–18% and 12–18% of HGBL, NOS, respectively [100, 102]. Approximately 50% of HGBL, NOS cases bear a DHITsig or similar gene expression profile [94, 103] and, in a more recent study, a subset of HGBL, NOS clustered with DLBCL/HGBL-MYC/BCL2 by principal component analysis [103]. These HGBL, NOS comprise both MYC translocation-positive and -negative cases [103]. Similarly, over 40% of HGBL, NOS cases express both MYC and BCL2 proteins, representing double expressor lymphomas [96, 103, 104]. As routine interphase FISH is unable to detect cryptic MYC or BCL2 rearrangements, a subset of these HGBL, NOS with DHITsig might be DLBCL/HGBL-MYC/BCL2 [81, 83].

Taken together, the phenotypic features, mutational spectrum, and gene expression profiles suggest that HGBL, NOS contains several biologic subgroups, underscoring the heterogeneity of this diagnostic label (Table 1) [105]. Considering its rarity and the challenges in differentiating other related entities by conventional molecular analysis, collaborative networks with innovative and comprehensive research approaches are required to ‘distill’ the ‘true’ HGBL, NOS cases, if such an entity does indeed exist, and to interrogate their pathobiology.

High-grade B-cell lymphoma with 11q aberrations (HGBL-11q)

HGBL-11q is an aggressive mature B-cell lymphoma consisting of medium-sized cells in majority of cases and harbouring a characteristic chromosome 11q gain/loss pattern in the absence of MYC-R (Figure 6). Currently, only cases with telomeric loss and/or solely telomeric loss of heterozygosity (LOH) are recognised as bona fide HGBL-11q, and this concept remains to be further corroborated in future studies. The presence of an 11q23.3 gain alone – without concomitant 11q24 loss – is considered non-specific and, therefore, not defining for HGBL-11q. In such cases, high-resolution array-based analyses are necessary to decipher the genomic alterations, providing more robust evidence of, for example, telomeric LOH (Figure 6) [106].

Details are in the caption following the image
Examples of HGBL with 11q aberrations. Detection of gains of 11q23.3 and loss of 11q24.1 by interphase FISH (upper right panel) and OncoScan assay (lower panel).

Currently, it is unclear how to deal with rare cases of HGBL that have both 11q aberrations and a MYC rearrangement (Table 1). Cases of bona fide IG::MYC-positive BL have been described that have acquired characteristic 11q aberrations as a secondary event. This phenomenon has served as an argument for the present classification to exclude cases with MYC rearrangement from the definition of HGBL-11q. However, such cases have been reported more recently, suggesting that MYC rearrangement can occur as secondary alteration in HGBL-11q [107, 108]. Therefore, this issue should be treated as an open question to be resolved in further research. The number of such cases may be underestimated, as screening for 11q aberrations is recommended for MYC-R-negative B-cell lymphomas that are morphologically and immunophenotypically similar to BL, but not for MYC-R-positive cases [108].

HGBL-11q has been described as being particularly frequent in post-transplant patients, who develop B-cell lymphomas with Burkitt-like morphology. It has also been described in the setting of HIV infection and, more recently, in primary immunodeficiency syndromes (inborn errors of immunity), such as ataxia-telangiectasia [108]. Whether there is indeed a direct mechanistic association to immune deficiency/dysregulation needs to be further established. Epstein–Barr virus is consistently negative.

The morphological spectrum of HGBL-11q varies from blastoid to intermediate to rare cases of large B-cell lymphomas. Cases with intermediate morphology display cellular pleomorphism ranging from cases very similar to BL to others more intermediate between BL and DLBCL. Only rare paediatric cases show large cell morphology. In a recent study, coarse apoptotic bodies were reported as characteristic of HGBL-11q and to represent an important morphological feature to suspect the diagnosis and prompt FISH analysis [109]. This is a helpful feature, but its absence does not exclude the diagnosis [108].

HGBL-11q usually displays a GC phenotype (CD10+, BCL6+) and a high proliferation index (Ki67 >90%). MYC expression may vary from case to case, always displaying weak staining. A recent study demonstrated that 46% of HGBL-11q expressed LMO2, a GC marker, and this may represent a useful marker to support HGBL-11q over BL in difficult cases [108]. In contrast to BL, mutations in the ID3-TCF3 pathway, recognised as the biological hallmark of BL, are not detected in HGBL-11q. The mutational landscape of HGBL-11q is closer to that of GCB-DLBCL, with GNA13 mutations observed in about 50% of cases [110].

The outcome has been reported to be excellent in young patients treated according to protocols designed for BL [111, 112], while variable in adult patients undergoing therapy designed for DLBCL [111, 113, 114]. In the adult cases, TP53 mutations appear to be associated with dismal outcome [108]. Nonetheless, the mutation profiles and their clinical impact, as well as the genetic target of the 11q aberrations in HGBL-11q, remain to be fully characterised.

Conclusion and future prospects

As summarised in Table 1, the writing of the WHO-HAEM5 by its multidisciplinary team has enabled the identification of certain ‘open questions’ associated with the entities discussed above, with these questions forming the basis of further research and the targeted collation of evidence. These queries include the following: (1) dissecting distinct entities from each other within heterogeneous classes/families (e.g. splenic lymphomas); (2) understanding why genetic rearrangements that are present in most classic lymphomas (e.g. FL and BCL-2 rearrangements) are absent in a minority and discerning what alternative oncogenetic mechanisms result in these tumours' development; (3) examining dysregulation of master transcription regulators, such as BCL-6 in FL, and their influence on clinical course; (4) consideration of molecular subtypes of entities based on putative COO, enabled through newer technologies (e.g. possible differing COO of FLs, arising from the DZ and LZ); (5) comprehending the significant role of MYC translocation in DLBCLs/HGBLs with MYC and BCL2 rearrangements, as well as the underlying structural genomic alterations (and their incidence) in DLBCLs with MYC and BCL6 rearrangements; and (6) teasing out the distinctive features of rare entities, such as HGBL, NOS, to enable clear distinction from the mimics, e.g. HGBL-11q.

Each of these questions will involve huge amounts of effort and collaborative work between research groups, investigating well-annotated sample cohorts, which may be difficult to collate if the entities are rare. However, through the identification of these knowledge gaps or ‘known unknowns’ in our understanding of the pathogenesis of some lymphomas (and their subtypes), the scientific community can take on these queries in a targeted and precise way. Comprehensive approaches in genetic and molecular investigations will help both to mitigate controversial findings and to accelerate discoveries, establishing consensus on these challenging questions. To achieve these, a prerequisite is to standardise genetic and molecular methodologies and their data analysis, helping to reduce conflicting data and, hence, to focus on true pathological questions. The aggregated data from the research will lead to a modification of our understanding and, ultimately, the categorisation frameworks, i.e. WHO-HAEM6 and beyond.

Acknowledgements

We would like to thank Jasmine Makker for help on river flow figure and bibliography preparation. Research in the Du lab was supported by grants from Blood Cancer UK (19010) and Cancer Research UK (C8333/A29707).

    Author contributions

    All authors were actively involved in manuscript writing and preparation, and approve its submission for publication.