Introduction

Since the 2016 WHO (World Health Organization) update, progress has been made in understanding the biology of Burkitt lymphoma (BL) and the concept of high-grade B-cell lymphoma (HGBCL) that allows some degree of refinement. Advances in integrative analysis of mutations, structural abnormalities, copy number, and gene expression signatures allow a more nuanced view of the heterogeneity of DLBCL, NOS as well as definitions of high-grade B-cell lymphomas. Relationships to specific types of normal counterparts in the germinal center are also being refined. In BL and HGBCL, the expression of TdT has taken on less prominence in driving classification based on recent molecular findings. The segregation of MYC and BCL2 from MYC and BCL6 double-hit lymphomas reflects studies that suggest important differences and the need for further study. Strategies for identifying MYC, BCL2, and BCL6 rearrangements and the importance of partner genes for MYC rearrangements are areas of active investigation. Finally, the criteria for identifying HGBCL, NOS are still problematic due to their heavy reliance on subjective assessment of the cytologic appearance of the neoplastic cells. The following reflects the conclusions the Clinical Advisory Committee (CAC) members reached following the discussions held in the drafting of the 2022 International Consensus Classification (ICC) revision to the 2016 WHO classification (Fig. 1).

Fig. 1
figure 1

International Consensus Classification approach to large B-cell/high-grade B-cell lymphomas. Lymphomas with a Burkitt-like, blastoid, or large-cell morphology that have MYC with BCL2 and/or BCL6 rearrangements are categorized as HGBCL, with either MYC and BCL2 (with or without BCL6R) or MYC and BCL6 R. HGBCL, NOS lacks these double hits and may have either Burkitt-like or blastoid morphology. TdT expression, but not CD34, may be seen in HGBCL. Burkitt lymphomas typically have a MYC R without BCL2 or BCL6 and a mature GC B-cell phenotype. B-lymphoblastic lymphomas are blastoid-appearing, usually TdT + and frequently but not always CD34 + . Cases with a large B-cell morphology that lack a DH, generally, are categorized as one of the DLBCL, NOS or as one of the more specific types of large B-cell lymphoma. The possibility of a large B-cell lymphomas with 11q aberrations should be considered particularly with a Burkitt-like morphologic appearance but with coarse apoptotic material in the tingible body macrophages and lack of a MYC R. Some cases, however, will more closely resemble a DLBCL. DLBCL, diffuse large B-cell lymphoma; R, rearrangement; HGBCL: high-grade B-cell lymphoma; DH, double hit

Full size image

Burkitt lymphoma

BL is a highly aggressive B-cell lymphoma related to dark zone germinal center B-cells with three recognized clinical variants, namely sporadic (sBL), endemic (eBL, occurring in equatorial areas in sub-Saharan-Africa and South America), and immunodeficiency-associated (iBL) typically seen in the setting of acquired immunodeficiency syndrome (AIDS). Given the near universal association of endemic BL with EBV infection and association with endemic malaria, host and environmental factors likely impact pathogenesis of BL and are areas of ongoing investigation [1, 2]. Clinical presentations differ somewhat based on clinical scenario. In eBL, the peak incidence occurs in children (age 6–8) with a male predominance. Tumors present as rapidly growing masses with a predilection for extranodal sites such as the jaw in younger children and abdomen in older children. A decline in jaw tumors has been observed [3,4,5]. Sporadic tumors demonstrate age-dependent incidence rates with peaks in young children (around age 10), younger adults (age 30–40), and older adults (age 70) in more recently diagnosed cases [6]. Again, there is a male predominance with abdominal and other extranodal sites being involved. In HIV-associated iBL, cases occur early in infection and appear to still occur in the era of highly active anti-retroviral therapy [7, 8]. Patients present in advanced stage with frequent bone marrow and CNS involvement [9].

Regardless of the clinical variant, pathologic features are similar with a relatively uniform population of intermediate-sized cells with round nuclei and multiple small nucleoli. There is scant to moderate amount of cytoplasm that, on Wright-Giemsa-stained smear preparations, is basophilic with small lipid vacuoles. A background “starry-sky” pattern is seen and numerous mitotic figures reflect the high proliferative index. Reports focusing on HIV-associated BL have noted an increased frequency of plasmacytoid differentiation, usually in the setting of EBV positivity and which may be associated with IRF4/MUM1 and cytoplasmic Ig expression (Fig. 2) [10, 11]. Rare cases of BL will have a florid background proliferation of epithelioid histiocytes and granulomas (Fig. 3) [12, 13]. In some cases, these may obscure the tumor, especially on small biopsies, and could lead to misdiagnosis as an infectious process or sarcoidosis. These cases appear to be nearly 100% EBV positive (Fig. 3) [13]. Interestingly, this feature has been found to portend a favorable prognosis, and even, in rare cases, spontaneous remission [12,13,14].

Fig. 2
figure 2

Burkitt lymphoma with plasmacytic differentiation in a HIV patient. (A) H&E shows more pleomorphism than is typical for Burkitt lymphoma, including cells with plasmacytoid features. (B) MUM1 staining highlights plasmacytic differentiation. Both images, 1000 × oil

Full size image
Fig. 3
figure 3

Burkitt lymphoma, EBV + , associated with granulomas in parotid biopsy (CD20 + , CD10 + , BCL2-, FISH studies showed IGH::MYC but no BCL2 or BCL6 rearrangement). (A) There is a diffuse lymphoid infiltrate associated with scattered granulomas. (B) At higher magnification note the relatively uniform intermediate-sized cells associated with mitotic figures and tingible body macrophages. The granulomatous areas also included smaller T-cells. (C) The tumor is EBER positive (A H&E, 100 × , B H&E, 1000 × oil, C EBER in situ hybridization with hematoxylin counterstain, 400 ×)

Full size image

The characteristic immunophenotype is CD19 + , CD20 + , CD5-, CD10 + , BCL6 + , BCL2 negative (or very weak), surface immunoglobulin (sIg) positive, and TdT negative. The molecular hallmark of BL is the translocation of MYC with an IG [2] gene, mostly commonly IGH (approximately 80%), followed by IGK (approximately 15%) and IGL (approximately 5%) [1]. As noted above, EBV is found in essentially 100% of eBL and in approximately 30% of sBL and iBL.

While these basic features remain unchanged, the 2022 ICC update notes that cases previously reported as TdT + BL should be distinguished from BL based on several genetic findings including (1) recent molecular analysis demonstrating IG::MYC rearrangements due to aberrant VDJ recombination rather than class switch recombination (CSA) or somatic hypermutation (SHM), (2) lack of mutations commonly seen in BL but rather recurrent NRAS/KRAS mutations, and (3) lack of functional B-cell receptor [15, 16]. Such TdT + cases should be designated as B-lymphoblastic lymphoma/leukemia with MYC rearrangement to recognize their biology and allow clinicians to consider appropriate treatment options.

Studies characterizing the molecular genetic landscape of BL at the genomic and transcriptomic levels have provided insight into the physiological counterpart and important cellular pathways targeted by recurrent genetic alterations. With regard to the IG::MYC rearrangement, sequence analysis in sBL showed class switch recombination, particular into the IGHA switch region (an uncommon event in other MYC-rearranged GC B-cell derived lymphomas) to be most common (CSR, 73%) followed by SHM, 27%, with the SHM being seen most frequently in translocations with IGL partner gene, compared to IGH [17]. Thus, in line with other studies, BL derives from a mature GC-derived B-cell. These MYC rearrangements also result in loss of repressive BCL6 binding sites in the MYC promoter and, along with mutations in MYC that alter protein phosphorylation sites involved in proteosomal degradation, account for aberrant BCL6 and MYC protein co-expression. Recurrent mutations in coding genes affect a limited set of genes in BL including MYC, ID3, CCND3, TP53, SMARCA4, FBXO11, ARID1A, and DDX3X being seen in more than 20% of tumors [17]. Important pathways that are genetically or epigenetically altered in addition to MYC include constitutive activation of the BCR/PI3K/AKT (TCF3, ID3, FOXO1), cell cycle regulation (CCND3, CDKN2B, TP53), RAS, and Gα13 signaling (GNA13, ARHGEF1, S1PR2, RHOA, and P2RY8) [17]. Genomic studies have also suggested the greater role of EBV compared to geographic origin in distinguishing different subsets of BL with potential therapeutic implications [18].

Treatment of pediatric BL is similar to pediatric DLBCL with favorable outcomes using risk-adapted intensive chemotherapy regimens, with addition of rituximab in some settings [19]. Treatment of adult BL consists also of intensive immunochemotherapy regimens. Outcomes are generally good (75–85% overall survival) with these strategies but challenges remain for patients with CNS involvement and older patients (> 60 years), and in the relapse setting [20].

High-grade B-cell lymphoma

The 2016 WHO classification introduced the category HGBCL with MYC and BCL2 and/or BCL6 rearrangements, frequently termed double-hit (DH) and triple-hit (TH) lymphomas (DHL/THL) depending on whether or not both BCL2 and BCL6 rearrangements are present [2, 21]. This category included many of the cases previously classified under the WHO 2008 as “B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and Burkitt lymphoma” as well as a subset of DLBCL. The 2016 DHL/THL category thus unified cases with varying morphology based on their genetic findings and generally poor prognosis [21,22,23,24,25,26,27,28]. These DHL/THL may occur de novo, or as transformation of a prior low-grade lymphoma, usually follicular lymphoma (FL) [29].

Morphologically, DH/TH cases may resemble blastoid cells of lymphoblastic lymphoma, centroblasts of DLBCL, intermediate-sized, uniform cells of BL, and those that overlap between the three (Fig. 4). The 2016 WHO indicates that whether a given case has high-grade cytology (Burkitt-like and lymphoblastoid) or large-cell (DLBCL-like) cytology should be noted in the pathology report, because data from several studies have suggested that, among DHL/THL, those with high-grade cytology have a worse prognosis than those with large-cell cytology [26, 27, 30]. However, there is widespread acknowledgement in the field that there is poor reproducibility in identifying what represents “high-grade” cytology [31].

Fig. 4
figure 4

High-grade cytology compared with large-cell cytology in B-cell lymphomas. Panels (A) and (B) demonstrate two cases with high-grade cytology showing intermediate-sized cells with open chromatin and inconspicuous nucleoli. These cases may or may not have tingible body macrophages imparting a starry-sky appearance similar to what would be expected in Burkitt lymphoma. In contrast, panels (C) and (D) show two cases of large B-cell lymphoma cytology which should not be considered “high-grade” in the absence of “double-hit” (DH) cytogenetics. Panel (C) demonstrates frequent tingible body macrophages and tumor cells with open chromatin and amphophilic cytoplasm which may be mistaken for Burkitt lymphoma; however, the cells are unequivocally large. In the absence of DH, this would be best considered diffuse large B-cell lymphoma. Panel (D) also shows large-cell cytology but with coarse chromatin. Although there are frequent mitoses and apoptotic bodies, this case should again be diagnosed as large B-cell lymphoma in the absence of a DH. All H&E, 600 × oil

Full size image

In the 2022 ICC, the DHL/THL group is split into two distinct categories: (1) HGBCL with MYC and BCL2 rearrangements (HGBCL-DH-BCL2) and (2) HGBCL with MYC and BCL6 rearrangements (HGBCL-DH-BCL6) based on biologic differences [15]. The prior recommendation to note whether they have high-grade or large-cell cytology remains given its potential prognostic impact. There is substantial evidence to maintain the HGBCL-DH-BCL2 cases as distinct from other DLBCLs even if they have some overlapping features with a subset of DLBCLs. As explained below, there is debate as to whether the HGBCL-DH-BCL6 cases are a uniform and distinct disease group. As such, HGBCL-DH-BCL6 is considered a provisional entity, maintained to encourage further study of these tumors. Cases with MYC and both BCL2 and BCL6 rearrangements (TH) should be categorized as HGBCL-DH-BCL2, as the role of BCL6 rearrangement in these cases remains uncertain [32].

High-grade B-cell lymphoma with MYC and BCL2 rearrangements

HGBCL with MYC and BCL2 rearrangements (HGBCL-DH-BCL2) accounts for 80–90% of DH/TH lymphoma cases but only 2% of non-Hodgkin lymphomas overall. The median patient age is approximately 60 years, but cases in teenagers and young adults have been reported. Cases in younger children are exceptionally rare. There is a slight male predominance (60%). Most patients present with advanced-stage (III/IV) disease. These are aggressive lymphomas, with poor prognosis using conventional R-CHOP chemotherapy [23,24,25,26,27,28, 33, 34]. There are small subsets of patients with DHL, defined based on selected clinical and laboratory findings, however, who do well [30, 35, 36].

As described above, the morphology is variable, and the genetic findings alone are sufficient to classify cases as HGBCL-DH-BCL2 with two exceptions. First, cases of otherwise typical grade 1–2 or grade 3A follicular lymphoma with MYC and BCL2 rearrangements have been reported [37, 38], but should still be categorized as FLs and not as a HGBCL. Although some studies have suggested a more aggressive course than typical for indolent FL, data is limited and variable and, as a group, they are clearly not as aggressive as classic HGBCL-DH-BCL2 [37,38,39]. Whether grade 3B FL with these rearrangements are more like a HGBCL-DH-BCL2 is uncertain. Second, there are very rare cases of what are believed to be true B-lymphoblastic leukemia/lymphoma (B-ALL/LBL) with DH cytogenetics. However, very strict clinical and pathologic criteria must be applied to identify these cases, as HGBCL-DH-BCL2 lymphomas can have phenotypic overlap with ALL and can even express TdT (discussed below) [40,41,42,43].

HGBCL-DH-BCL2 are mature lymphomas, which virtually always express pan B-cell markers PAX5 and CD19, and are usually CD20 positive. By flow cytometry, surface light chain expression is found in up to 80% of cases [44]. Bright CD38 is a distinctive feature of some cases [44]. CD34 is absent and if present should prompt strong consideration for B-ALL/LBL.

TdT expression has been reported in 11–14% of DHL/THL [40, 41]. Expression of TdT, however, does not provide sufficient support for reclassification to B-ALL/LBL, in contrast to the WHO 2016 recommendation. Although commonly considered a marker only of immature cells, the majority of TdT-positive HGBCL-DH represent either transformations from prior follicular lymphoma, relapses of aggressive B-cell lymphomas that did not previously express TdT, or co-occurrence with a similar mature aggressive lymphoma without TdT expression [30, 40, 41, 43, 45]. Importantly, these cases never express CD34, and many have other features of maturity such as surface light chain and CD20 expression [30, 40, 41, 43, 45]. Each of these scenarios supports these TdT-positive cases to have originated from a mature B-cell, with acquisition of TdT expression as a secondary event in the course of clonal evolution. Limited mutational studies have suggested a frequently branching clonal evolution [40, 42]. Additionally, targeted mutation analysis performed on 6 cases of TdT-positive LBCL shows mutation profiles similar to GCB-type DLBCL and distinct from B-ALL/LBL [40]. Furthermore, the presence of somatic hypermutation in the small number of cases tested also supports these to be mature lymphomas [42]. Rare reports of true, otherwise clinically and pathologically classic, B-ALL/LBL with DH cytogenetics are challenging to reevaluate retrospectively. However, TdT alone should not be used as a de facto evidence of immaturity, and the overall clinicopathologic features should be considered before rendering a diagnosis of B-ALL/LBL with DH cytogenetics.

HGBCL-DH-BCL2 cases are almost 100% of germinal center B-cell (GCB) origin as classified by either the Hans algorithm immunohistochemistry (IHC) or gene expression profiling (GEP) methods [46]. CD10, BCL6, and BCL2 are positive in over 90% of cases, while IRF4/MUM1 is usually negative. MYC is usually expressed using an IHC cutoff of ≥ 40%. However, importantly, neither expression of MYC nor BCL2 is sensitive or specific enough to identify these rearrangements. Several mechanisms accounting for MYC negativity in tumors with MYC rearrangement have been described including low MYC mRNA levels and a MYC variant (NS11S) which disrupts the epitope of the antibody used for IHC, thereby making MYC expression appear low albeit with maintained mRNA and protein levels [47] (see further discussion of the significance of protein expression of MYC and BCL2 in DLBCL in Dirnhofer et al. from this article series) [46, 48]. The relatively uniform phenotype of the HGBCL-DH-BCL2 cases hints at their underlying uniform biology which, as discussed below, supports their derivation from a follicular lymphoma-like clone.

With recognition of HGBCL-DH-BCL2 as a distinct, aggressive entity, an active area of study has been the unique biology of these tumors, and their relationship to both BL and DLBCL, NOS. HGBCL-DH-BCL2 typically have a complex karyotype consistent with an underlying biologic complexity beyond the constitutive expression of MYC and BCL2 [49]. Mutations in BCL2 (~ 80%), CREBBP (~ 50%), EZH2 (~ 50%), and TNFRSF14 (~ 45%) are frequently seen in HGBCL-DH-BCL2 and overlap with mutations typical of FL and GCB DLBCL [29, 32, 50]. In addition, HGBCL-DH-BCL2 also harbor mutations in ID3 (20–40%), CCND3 (10–20%), which are common also in BL, and FOXO1 (~ 25%) which is known to regulate the germinal center dark zone along with CCND3 [29, 32, 51, 52]. MYC mutations are frequent (~ 40%) [29]. In addition to these well-established recurrent mutations, whole exome sequencing studies (WES) performed by Künstner et al. identified mutational drivers which were not previously identified in panel-based studies. TP53 mutations (25–30%) are commonly seen in HGBCL-DH-BCL2, in striking contrast to the HGBCL-DH-BCL6 (discussed below). Künstner et al. also compared their results to the molecular DLBCL clusters from Chapuy et al. and Schmitz et al. and report significant overlap between the EZB/C3 groups and the HGBCL-DH-BCL2 [32, 53,54,55]. These results continue to support a FL-like clone as the origin of the HGBCL-DH-BCL2 tumors. Finally, triple-hit lymphoma cases in general show mutational similarity to HGBCL-DH-BCL2, but with overlap to HGBCL-DH-BCL6 cases as well, possibly suggesting alternative routes of clonal evolution [29, 32].

High-grade B-cell lymphoma with MYC and BCL6 rearrangements

HGBCL-DH-BCL6 accounts for only 10–20% of DH/TH and in the 2022 ICC is considered a provisional entity [15]. In general, the molecular features of HGBCL-DH-BCL6 are less well-established and clinicopathologic features are less uniform than those of the HGBCL-DH-BCL2 group. The decision to separate these cases from HGBCL-DH-BCL2 stems primarily from their distinct biology [26, 46, 56]. The demographic data of patients with HGBCL-DH-BCL6 are similar to the HGBCL-DH-BCL2 cases, although there may be a particular predilection for extranodal disease in this group [57, 58]. Because of their relative rarity, and because they have typically been combined with the more numerous HGBCL-DH-BCL2 cases for clinical trials, data on their prognosis is challenging to establish. While several recent studies have suggested a very poor prognosis, similar to HGBCL-DH-BCL2, others suggest such lymphomas have similar prognosis to DLBCL, NOS [22, 26, 27, 56, 57].

Morphologically, the majority of HGBCL-DH-BCL6 have DLBCL morphology with a smaller subset showing high-grade morphology [57]. As in HGBCL-DH-BCL2, the cells are phenotypically mature B-cells (express CD19, CD20, CD79a, PAX5), which do not express CD34, and only rarely express TdT, a scenario which again should not independently prompt reclassification as B-ALL/LBL. In significant distinction from HGBCL-DH-BCL2, by GEP, COO is GCB in only about half of HGBCL-DH-BCL6 cases, and ABC/non-GC in the remaining half although by Hans algorithm higher percentages of GCB cases are seen [46, 57, 58]. CD10 is less common than in HGBCL-DH-BCL2 cases, and IRF4/MUM1 expression is seen in 40–90%. MYC expression is seen in up to 90%, and BCL2 expression in 20–80% [46, 57, 58].

The MYC gene partner in HGBCL-DH-BCL6 includes both IG and non-IG partners as with HGBCL-DH-BCL2 (discussed below). However, unlike in HGBCL-DH-BCL2 cases, HGBCL-DH-BCL6 includes a subset of cases in which the BCL6 gene is actually the MYC partner gene resulting from a t(3;8)(q27;q24) which has been termed a “pseudo-double hit” [59, 60]. This MYC::BCL6 translocation has been found to account for up to 30% of cases of DH/TH with MYC and BCL6 rearrangements [61]. Thus, of the HGBCL-DH-BCL6 group, a substantial number harbor this “pseudo-DH” rearrangement, further supporting the separation of these cases from HGBCL-DH-BCL2 in the 2022 ICC [15]. Although there is some literature to support the aggressive behavior of these pseudo-DH tumors, data remains sparse and further study is needed before a recommendation can be made to separate these cases from HGBCL-DH-BCL6 [59]. Additionally, many of the cases (as many as ~ 70%) harboring the MYC::BCL6 also have BCL2 rearrangements (THL), making it challenging to draw conclusions about the significance of the MYC::BCL6 alone [59]. As such, FISH testing to determine whether a HGBCL-DH-BCL6 has MYC::BCL6 translocation is not required.

In contrast to HGBCL-DH-BCL2, the mutational landscape of HGBCL-DH-BCL6 is more heterogeneous and knowledge comes from limited case numbers. Although recurrent mutations in HGBCL-DH-BCL2 are seen in some cases of HGBCL-DH-BCL6, they are found at much lower frequencies: CREBBP (~ 14%), KMT2D (~ 18%), EZH2 (~ 5%) [29, 32, 52]. Overlap with BL mutations such as ID3 (~ 30%) and CCND3 (~ 30%) is similar to what is seen in HGBCL-DH-BCL2, and emphasizes both an overlap with the HGBCL-DH-BCL2 and the distinction of these cases from other DLBCL, NOS [29, 32, 52]. Compared to HGBCL-DH-BCL2, HGBCL-DH-BCL6 shows infrequent BCL2 mutations [32, 62]. While some HGBCL-DH-BCL6 tumors cluster with EZB lymphomas, others cluster with the BN2 subtype reflective of marginal zone lymphoma-like biology, suggesting that there may be two distinct pathways of transformation (one from low-grade FL and one from marginal zone lymphoma) leading to the same HGBCL-DH-BCL6 endpoint. Finally, nearly half fall into other groups or remain unclassified by these algorithms [32] highlighting the heterogeneity within this group and distinction from HGBCL-DH-BCL2 and from DLBCL, NOS.

FISH probe strategy in high-grade and large B-cell lymphoma

A sensitive, efficient, and cost-effective FISH probe strategy to identify DH/TH lymphomas in the clinical setting has been the subject of much debate since the WHO 2016 publication highlighted the need to identify these aggressive lymphomas. Prior to the establishment of the DH/TH category, FISH for MYC was typically performed at the pathologist’s discretion only in cases that resembled BL or those with high Ki-67 proliferation index or mitotic rate. However, it has now been widely established that these “high-grade” morphologic features are inadequate for screening, as approximately 10% of otherwise typical DLBCL harbor MYC rearrangements [26, 46]. Screening by COO has been proposed as an efficient and cost-saving measure, as GCB lymphomas account for virtually 100% of the HGBCL-DH-BCL2 cases using GEP; however, this strategy will miss approximately 50% of the HGBCL-DH-BCL6 cases that are of non-GCB origin and a small number (up to 4%) of HGBCL-DH-BCL2 which may be misclassified by the Hans algorithm [22, 26, 27, 46]. Finally, neither expression of MYC, nor BCL2, nor BCL6 is sufficient to screen for their respective rearrangements. As such, the 2022 ICC recommends screening all large B-cell and high-grade B-cell lymphomas for the possibility of HGBCL-DH of either type [15].

The approach to screening typically begins with a sensitive MYC FISH breakapart probe (BAP). However, available commercial MYC BAP probes vary in sensitivity depending on the extent of the breakpoint region covered in the probe design [61, 63, 64]. Even the most sensitive MYC BAP, however, will miss a subset of rearrangements due to cryptic insertions or other complex genomic rearrangements that are too small or out of the coverage area of the probes. Performing MYC::IGH dual-color, dual-fusion (DF) FISH in addition to MYC BAP can detect the 4–12% of MYC rearrangements that are missed by BAP alone [30, 65]. The addition of MYC::IGL and MYC::IGK DF probes likely also increases sensitivity, although data are sparse due to the lack of widespread availability of these probes.

The use of MYC::IGH, MYC::IGL, and MYC::IGK probes to detect MYC rearrangements, however, belies the diversity of MYC gene partners and complexity of possible rearrangements, especially in the non-BL setting [63]. In BL, the MYC partner is virtually always an IG gene, whereas in non-BL DLBCL/HGBCL the partner genes are non-IG in at least 40% of cases and include a variety of genes such as BCL6, PAX5, IRF4, BCL11A, IKZF1, IMMP2L, and RFTN1 among others [22, 26, 61, 63, 66]. The true “false negative” rate of current MYC FISH strategies is almost certainly underestimated with one estimate being 20%, and increasingly, in the research setting, molecular methods are allowing for detection of rearrangements cryptic to FISH [65, 67].

The clinical significance of the MYC partner gene has also been the subject of much debate. Whether the canonical concept of juxtaposition of IG enhancers to the MYC gene leading to overexpression of MYC protein occurs in a similar fashion with enhancers of non-IG genes is uncertain. Several studies have suggested that, in cases with DLBCL morphology, the prognosis of MYC::IG translocations is inferior to those with MYC-non-IG translocations, and in fact that DH/TH lymphomas with MYC-non-IG have no difference in prognosis to DLBCL, NOS [22, 27]. However, others have shown discrepant results, and especially in a cohort enriched with high-grade cytology tumors, there was no difference in survival among DH/TH lymphomas with IG vs non-IG MYC rearrangement [26]. In summary, although many in the CAC group felt compelled by the data supporting worse prognosis in MYC::IG cases, this remains an unresolved question. Importantly, the lack of widespread availability of IGK and IGL probes precludes routine testing for all IG rearrangements in clinical practice, and, as such, the 2022 ICC does not require identification of the MYC partner [15].

The WHO 2017 Blue Book specifically states that copy number alterations (CNA) of MYC, BCL2, or BCL6 are not equivalent to rearrangement for classification of DH/TH lymphomas and the 2022 ICC retains this concept [2, 15]. FISH testing invariably leads to identification of CNA in tested genes. The majority are copy number gains of 3–10 copies per cell, many due to extra copies of whole chromosomes common in lymphoma in general [68]. Difficulty in elucidating the significance of these CNA is in part due to the lack of consistency as to how such gains are reported by different laboratories and the likely variable significance of whole chromosome versus single gene locus gains in altering gene expression [48]. In addition, studies looking at the significance of “amplification” have used various definitions. Interestingly, at least one study has shown that while BCL2 CNA were reported to be strongly associated with increased BCL2 expression, MYC CNA were not associated with such an increase [47]. This finding further casts doubt on the significance of such CNA. While some studies have found CNA of MYC to yield poor prognosis similar to rearrangement, others do not support this assertion [68,69,70,71,72,73,74,75]. There are some data to suggest that high amplification of MYC (as defined by uncountable FISH probe signals) is significant; however, these cases are rare and there are insufficient data to draw definitive conclusions [47, 71].

Gene expression signatures to identify high-grade B-cell lymphoma

Recently, two separate groups have identified gene expression signatures of “high-grade” lymphomas which encompass most, but not all, HGBCL-DH-BCL2 along with BL and some GCB-DLBCL. While termed “double hit signature” by Ennishi et al. [76] and “molecular high grade” by Sha et al. [77]), the patterns of gene expression in these signatures are reminiscent of dark zone (versus light zone) germinal center B-cells and could be consider dark zone signatures. The double hit signature (DHITsig) was developed on the gene expression features distinguishing HGBCL-DH-BCL2 from GCB-DLBCL [76]. Only half of the DHITsig cases identified by the classifier were HGBCL-DH-BCL2 by FISH [76]. However, in 30% of non-HGBCL-DH-BCL2, DHITsig tumors were subsequently found to have MYC or BCL2 rearrangements that were cryptic to FISH [67]. Interestingly, in spite of many not harboring the rearrangements, the DHITsig patients have significantly shorter time to progression, overall survival, and disease-specific survival than the DHITsig-negative GCB-DLBCL cases suggesting aggressive biology may be more closely tied to the signature than DH cytogenetic status [76]. The mutation profile of the DHITsig cases suggests enrichment of mutations in MYC and BCL2, as well as chromatin modifier genes including CREBBP, EZH2, and KMT2D along with DDX3X and TP53 and with low expression of major histocompatability (MHC) genes. A subsequent study applied the DHITsig to a series of HGBCL, NOS and found the DHITsig in 55% of these cases [31] further highlighting the heterogeneity of HGBCL, NOS.

In a parallel study, “molecular high grade” (MHG) signature was developed from the molecular Burkitt lymphoma gene expression signatures [78, 79]. Their MHG classifier also identifies a group of predominantly GCB tumors with GEP similar to centroblast (dark zone) germinal center B-cells which show poor prognosis that extends beyond the effect of the DH/TH cytogenetics [77]. Again, 51% of MHG cases lacked a MYC rearrangement and only 36% were DH. MHG lymphomas were enriched in similar mutations to those in the DHITsig group, and showed similar downregulation of MHC genes. Both the DHITsig and MHG classifiers identify groups which show considerable overlap with the molecular signature-based EZB/C3 clusters of DLBCL published by Chapuy et al. [53]. and Schmitz et al. [54] and discussed in more detail in regard to DLBCL by Dirnhofer et al. within this article series.

These novel findings identified by two independent groups underscore the common germinal center dark zone biology shared not only by the HGBCL-DH-BCL2 lymphomas but also by a subset GCB-DLBCLs with equally poor prognosis to the DH cases and BL. They raise the obvious possibility of adapting such an assay to the clinical realm, whether that be a GEP-based approach or a sequencing approach, reflecting the mutations characteristic of this subset. Regardless of the approach, it is clear that our current ability to identify both HGBCL-DH and HGBCL, NOS is insufficient to identify all of the tumors with aggressive biology that may benefit from more intensive therapy. The potential role and usage of MHG and DHITsig profiles remain to be determined as the dark zone biology is shared by a diverse group of tumors with different biology and clinical behavior, requiring interpretation in the context of other parameters. For example, BL, a biologically and clinically very different type of lymphoma, also displays this gene expression profile [80, 81]. In fact, when applied to a cohort of R-CHOP-treated GCB DLBCL, DHITsig only predicted poor survival if combined with TP53 abnormality, whereas a subset of DHITsig-positive tumors which lacked EZH2 and BCL2 abnormalities actually had an excellent prognosis [82].

High-grade B-cell lymphoma, not otherwise specified

With the dissolution of the WHO 2008 category BCL, unclassifiable, with features intermediate between DLBCL and BL, there remained cases which did not fall into the HGBCL-DH category but also did not fit neatly into DLBCL, NOS, BL, or B-ALL/LBL. The WHO 2016 created the high-grade B-cell lymphoma, not otherwise specified (HGBCL, NOS) category for these cases [21], and the 2022 ICC retains this category to be used as a diagnosis of exclusion [15]. HGBCL, NOS cases typically have intermediate-sized cells, often with blastoid or Burkitt-like cytology (Fig. 4A, B) [2, 21]. DLBCL with starry-sky morphology and/or high proliferation index are not recategorized as HGBCL, NOS (Fig. 4C, D). Identification of HGBCL, NOS is subjective and requires well-fixed tissue biopsies as the distinction between intermediate and large cell size is imprecise, and cell size, as well as chromatin characteristics, can both be affected by differences in specimen type, tissue fixation, processing, section thickness, and stain quality among other pre-analytic factors. It is a diagnosis to be used very sparingly and there is limited data concerning these cases. In a multi-institutional study from the Lymphoma/Leukemia Molecular Profiling Project (LLMPP) with central pathology review (CPR) on 61 tumors submitted as HGBCL, NOS, the reviewers reclassified 48% as DLBCL and 5% as BL [31] highlighting the challenging nature of these distinctions.

Data on HGBCL, NOS remains sparse, but unsurprisingly points toward heterogeneity within these tumors. Patients appear to be older adults (median age 70), many of whom have advanced-stage disease [83, 84]. Rates of response to standard DLBCL-like therapy remain difficult to ascertain because of the recent adoption of this category and lack of complete alignment with historical categories; however, one early study suggests superior outcomes with more aggressive therapy regimens [83].

In a German study, also with CPR, HGBCL, NOS showed frequent expression of CD10 (65%) and BCL6 (90%), but interestingly only one HGBCL, NOS had the characteristic phenotype/MYC rearrangement typical of BL [85]. However, this may reflect an increased emphasis on phenotype/genotype vs morphology in this study. In the LLMPP HGBCL, NOS study, COO results showed 57% GCB, 25% ABC, and 18% unclassifiable tumors by GEP [31]. Only 54% of HGBCL, NOS cases showed a DHITsig profile [31]. In fact, the mutation profile, GEP, and phenotypic features were similar between cases classified as HGBCL, NOS and those reclassified as DLBCL emphasizing the heterogeneity of this category and potential flaw in using morphology to distinguish these tumors [31]. The German series and LLMPP series showed 13% and 46% of HGBCL, NOS harbored a single-hit MYC rearrangement, respectively, emphasizing likely differences in morphologic criteria used for inclusion [31, 85]. Of note, in spite of the higher percentage of cases with MYC rearrangement in the LLMPP series, this was the same percentage as was seen in their cases reclassified as DLBCL, emphasizing that MYC rearrangement alone does not explain the high-grade morphologic features and also highlighting the limits of classification based strictly on subjective morphologic features.

In summary, the 2022 ICC retains the HGBCL, NOS category with the acknowledgement that these cases likely do not represent a true, uniform biologic entity. However, until the field progresses to a molecular-based classification scheme, the consensus among CAC participants was that this category should remain but be used very sparingly.

Conclusion

Several updates to the 2016 WHO classification of BL and HGBCLs are recommended by the ICC. These include recognizing that TdT expression in DHL/THLs does not necessarily warrant a diagnosis of B-lymphoblastic leukemia/lymphoma. While HGBCL-DH-BCL2 remain relative well-defined, the best FISH strategies remain open to debate, realizing that cases will be missed if only MYC breakapart or MYC::IGH dual-fusion probes are used. Variation in performance of commercially available probe sets exists and laboratorians should be aware of their limitations. The so-called pseudo HGBCL-DH-BCL6 exist due to the t(3;8) translocation and the 2022 ICC segregates HGBCL-DH-BCL2 from HGBCL-DH-BCL6. HGBCL, NOS remains problematic due to an imprecise definition that relies heavily on morphologic features. While identifying molecular Burkitt/high-grade gene expression signatures may become a part of our future diagnostic armamentarium, their precise application in conjunction with our other morphologic, phenotypic, and genetic tools remains to be determined.