Diagnostic and prognostic molecular pathology of lymphoid malignancies

Although the diagnosis of CLL/SLL is straightforward in most cases, genetic alterations are of great importance for prognostic stratification and influence therapy decisions. The presence or absence of somatic hypermutation in the rearranged IG heavy chain gene, taking a <98% identity with the germline sequence as cutoff, is of prognostic and therapeutic relevance [1]. In addition, some subsets of CLL with stereotyped B-cell receptors and CLL expressing IG lambda V3-21 with the R110 point mutation show distinct prognostic features [1, 66]. Although the cytogenetic and mutational landscape of CLL is highly diverse and shows differences between IGH-mutated and unmutated cases, alterations of TP53, usually bi-allelic (i.e., mutation and del17p) are the single most important prognostic and predictive parameter and need to be tested before initiation of therapy [64]. Of importance, even cases with small subclonal mutations (variant allele frequency (VAF) <10%) show a poorer outcome and response to therapy [85]. Mutational analysis of BTK, PLCG2, and BCL2 is recommended for CLL under certain targeted therapies (ibrutinib and venetoclax) [64].

LPL, in its classic form presenting as Waldenström’s macroglobulinemia, shows the canonical MYD88^L265P in 95 to 97% of cases, and CXCR4 is mutated in 30 to 40% [91, 104]. IgM monoclonal gammopathy of undetermined significance (MGUS) and non-IgM LPL show these mutations at lower frequency. Especially for IgM MGUS and LPL with low tumor burden, sensitive techniques need to be employed for detection [22]. The presence or absence of these two mutations has a significant impact on clinical features and therapy response and should be determined in all LPL cases. Whereas the IgM-expressing multiple myeloma is universally negative, rare cases of splenic [74], nodal, and extranodal marginal cell lymphomas in the ocular adnexal region show MYD88 mutations [76], as well as 2–3% of CLL, the latter cases associated with favorable biology [64].

Due to the lack of a specific immunophenotype or a defining or characteristic genetic alteration, diagnosis of marginal zone lymphomas (MZL) relies on a combination of clinical, morphological, and phenotypic features, but mutational profiling and translocation detection may help to separate them from other small B cell lymphomas. A subset of MALT lymphomas shows recurrent translocations frequently involving the MALT1 locus on 18q21; their frequency, as well as those of some recurrent mutations is strongly dependent on the primary organ manifestation [110]. In small B-cell lymphomas with nodal or bone marrow presentation, KLF2, NOTCH1/2, PTPRD, BIRC, TRAF3, CARD11, and IRF8 mutations favor a diagnosis of MZL. Of note, nodal MZL rarely may show MYD88 or BRAF mutations [49, 76, 96].

Conventional follicular lymphoma (FL) is characterized by the presence of a t(14;18)(q32;q21) translocation in about 85–90% of cases. The mutational profile is characterized by frequent alterations of epigenetic regulators, transcription factors, and others, whereas transformation is associated with a different set of genes, including TP53 mutations and MYC translocations[44, 49]. Although risk stratification models incorporating mutations, such as the m7-FLIPI [75], have been proposed, they are strongly dependent on the type of therapy. Therefore, genetic profiling currently is not of clinical or diagnostic relevance in typical FL, except for the detection of EZH2 mutations before treatment with the inhibitor tazemetostat [62]. However, it is useful for discerning FL variants, including the newly described BCL2 rearrangement negative, CD23+ follicle center lymphoma [12, 67], which shows a high frequency of STAT6 mutations, cutaneous follicle center lymphoma and pediatric type FL, which shows frequent alterations of TNSFRSF14, MAP2K1, and IRF8, distinct from adult FL [92].

Mantle cell lymphoma (MCL) is characterized by the presence of the t(11;14) translocation, usually easily demonstrable by the surrogate marker cyclin D1 overexpression. MCL can be divided into the conventional subtype, derived from naïve B cells with unmutated IGHV genes, usually SOX11+, whereas the leukemic non-nodal type shows somatic hypermutated IGHV genes, lacks SOX11, and shows less genomic complexity [65]. Clinically relevant for both subtypes are TP53 mutations, associated with a more aggressive clinical course, higher proliferation rate, and frequent blastoid morphology. The rare cyclin D1− MCL show rearrangements of CCND2 or CCND3, which should be demonstrated by FISH or other techniques, since immunohistochemistry is unreliable in this setting [57].

Hairy cell leukemia is virtually universally characterized by the BRAF^V600E hotspot mutation, with the exception of cases using the IGHV4-34 family, which may show MAP2K1 mutations and are similar to hairy cell leukemia variant [102].

Multiple myeloma (MM) and its precursor lesion, non-IgM MGUS, show great genetic heterogeneity, with 45–50% showing trisomies of odd-numbered chromosomes, and 40–45% showing recurrent IGH translocations with a variety of partners, which are currently detected by interphase FISH [20]. Due to the prognostic impact of genetic alterations, the ICC has formally divided MM into cytogenetic subgroups [12]. Mutational profiling currently is not of diagnostic relevance outside of clinical studies, but an absence of MYD88 mutations helps to exclude LPL [27].

Diffuse large B-cell lymphoma (DLBCL) is both the most common and most diverse lymphoma occurring in adults, and a number of parameters, including but not restricted to clinical site and pathologic features, are used for subclassification. One of the earliest, and at the time most paradigm-shifting, approaches to applying molecular tools to the characterization of DLBCLs, particularly those that belong to the largest category of DLBCLs, not otherwise specified (NOS), occurred almost a quarter-century ago with the identification of two major forms based upon their putative cell-of-origin (COO) [4]. This original study employed gene expression profiling (GEP) using DNA microarrays to identify and divide most of these DLBCL, NOS cases into those of either germinal center B-cell (GCB) or activated B-cell (ABC) origin. This bifurcation also had prognostic connotations, in that the latter predicted for worse outcomes [113]. At a mechanistic level, GCB-DLBCLs tend to harbor mutations in chromatin-modifying genes while ABC-DLBCLs are more likely to display mutations that result in activation of the NFKB and BCR pathways [40, 115]. Notwithstanding the availability of a commercial RNA-based assay (Lymph2Cx) [116] that has yet to be widely adopted, translation of the COO strategy into routine clinical use has been difficult. This has led to the development of a number of surrogate immunohistochemical (IHC) algorithms that, unfortunately, do not perfectly reflect the GEP-based classification [35, 59]. While the binary use of COO, using IHC algorithms, has been retained in contemporary classification schemes and is used in diagnostic practice, it has become clear that this seemingly simple separation may not be sufficiently robust to capture the complexity of the molecular underpinnings of DLBCL to be of independent clinical and therapeutic value [19, 72].

This complexity has more recently been unraveled using a compendium of genomic approaches by different investigators that have independently largely converged into somewhat overlapping classifications identifying up to seven major subtypes of DLBCL [13, 45, 93] (Fig. 3). The complexity and comprehensive genetic testing required to unravel these different subtypes is currently not readily adoptable by routine diagnostic laboratories, but publicly available online tools such as LymphGen may be used to facilitate classification [87, 114]. Of note, the LymphGen tool does not subtype 30-40% of cases of DLBCL into any of the major groups, in contrast to other proposed molecular subclassifications of DLBCL which assign all evaluable cases to a specific category. Many of the key or defining features in the various clusters are already available for testing in some or even most routine settings, since they can be ascertained by a combination of FISH (for BCL2 and BCL6 translocations, and TP53 loss) and targeted mutational analysis (evaluating MYD88, CD79B, EZH2, CREBBP, KMT2D, NOTCH1, NOTCH2, TP53, SGK1, TET2, and SOCS1, among others), although it is unlikely that such focused approaches fully capture the complexity of these subtypes. Nevertheless, a simplified 20-gene classifier has been described and has the potential to approximate the LymphGen approach to classify four main subgroups [60]. Conspicuously, and perhaps surprisingly, (almost entirely) absent as a major variable in the stratification of the different genomic categories of DLBCL is testing for a MYC translocation that has long been a key component of evaluation of this group of lymphomas. However, its presence does help refine one of the subgroups, and it remains an important initial genetic abnormality to evaluate in all aggressive B-cell lymphomas, particularly as a necessary first step in identifying the genetically-defined subtype of high-grade B-cell lymphoma (HGBL) of double-hit lymphoma (DHL), as discussed below. Provocatively, many of these major genomic subtypes of DLBCL share features evident in other, and often indolent, B-cell lymphomas such as marginal zone lymphoma, follicular lymphoma, and small lymphocytic lymphoma/chronic lymphocytic leukemia, as well as other defined DLBCLs, including those arising in extranodal immune-privileged sites or the mediastinum.

Another genetically defined LBCL entity is associated with an 11q aberration which in most cases harbors the combination of a proximal gain at 11q23.2-11q23.3 and distal loss at 11q24-qter, although some only show one of these features [88]. This lymphoma was previously classified as a Burkitt-like lymphoma but that designation has subsequently been revised since these cases, though morphologically and immunophenotypically sharing features with Burkitt lymphoma (BL), actually lack mutations that are typically evident in BL, as discussed below [31]. Commercially available 11q FISH probes are imperfect in identifying this defining aberration, and copy number array or other genomic analyses may be necessary to make this diagnosis. Notably, while classified as a LBCL by the ICC, WHO-HAEM5 has designated this as a form of HGBL [2, 12]. The rationale of the ICC for classifying the cases with 11q aberration as LBCL, rather than using the morphologically inspired term HGBL is based on their clinically less aggressive course and favorable outcome after therapy [31]. LBCLs with IRF4 rearrangements are another subtype that requires genetic testing for its diagnosis, typically by FISH although they also harbor mutations of IRF4 that may further support the diagnosis [80]. This entity is handled differently by ICC, which categorizes such cases within follicular lymphoma, given its overlapping features and typically indolent behavior, as compared with WHO-HAEM5, which groups this together with other LBCLs.

One of the two originally designated entities of high-grade B-cell lymphoma (HGBL) was one with MYC and BCL2 and/or BCL6 translocations, which was also termed double-hit (DH) or triple-hit (TH) lymphoma. However, this has subsequently been split into two distinct categories (1) HGBL with MYC and BCL2 rearrangements, which account for 80–90% of such cases; and (2) HGBL with MYC and BCL6 rearrangements, which account for 10–20% of such cases, with substantial evidence allowing the former to be retained as a bona fide entity, while the latter seems to be more heterogeneous and has been relegated, by the ICC but not WHO, to a provisional entity [41]. Virtually all of the MYC+BCL2 HGBLs are of GCB origin, whereas only around 50% of the MYC+BCL6 HGBLs are. The mutational landscape of the two DHLs is also quite distinct. Since there are no pathologic features (including morphologic, COO, or proliferative index) that can reliably identify DHLs, all DLBCLs should be screened for these defining translocations. MYC break-apart probes (BAP) alone are inadequate for detecting all MYC translocations, and different commercially available probes vary in sensitivity [63]. The addition of dual-color dual fusion MYC::IGH FISH probes may increase the yield by detecting translocations missed by the MYC BAP [42]. While the MYC partner in Burkitt lymphoma (BL, see below) is essentially always an IG gene, this is not the case here, and non-IG genes are involved in many (at least 40%) cases, including BCL6, IRF4, and PAX5 [16]. The suggestion that DHLs with MYC translocations involving IG have a worse prognosis appears to be unresolved [58, 84]. While DH MYC+BCL2 HGBLs are defined by these translocations that are imperfectly identified by cytogenetics or FISH, they can also be identified, perhaps even more reliably, by transcriptional profiling and have variably been termed molecular high grade (MHG) or DHITSig wherein cryptic translocations, missed by cytogenetics or FISH, are evident, occurring in 20% of cases [24, 37, 94]. It remains to be determined when and whether such testing can be applied in the routine diagnostic setting. Such lymphomas that almost always are of GCB origin may also, together with Burkitt lymphoma (see below), fall under the umbrella term of dark-zone lymphomas or DZSig lymphomas, with these cases refining the DLBCL-COO classification by identifying patients within this category with an inferior prognosis [3]. HGBL that is not a DHL or THL, termed HGBL, NOS, while enriched in MYC translocations (seen in ~45%) do not have a consistent unifying genetic profile, and remains a problematic diagnosis that is heavily reliant on often subjectively determined morphologic features.

The genetic hallmark of BL is a balanced translocation involving MYC and one of the three immunoglobulin genes (IGH, IGK, or IGL, in approximately 80%, 15%, and 5% of cases, respectively). The translocations occur in the germinal center in mature B-cells, either at the time of somatic hypermutation (SHM) or class switch recombination (CSR). In sporadic BL (see below), this event happens most often at the time of CSR (~75%), interestingly with a proclivity to involve IGHA, rarely seen in other MYC-translocated lymphomas [54]. While traditionally epidemiologically stratified into three types — sporadic, endemic, and immunodeficiency-associated BL — a perhaps more mechanistic (though imperfect) and somewhat overlapping approach might be to separate cases based upon the involvement (or not) of EBV. Thus, EBV-positive cases may have fewer, or at least different, additional driver mutations [33, 51]. BL lacking EBV tends to harbor canonical mutations affecting ID3, TCF3, and CCND3 that are also more common in pediatric than in adult cases [82], while EBV-positive BL tends to be enriched in mutations affecting DDX3X, GNA13, and FOX01 [101]. The mutations evident in EBV-negative cases are quite unique to BL, while different mutations in other GC-derived aggressive B-cell lymphomas such as those affecting genes encoding chromatin modifiers are not seen, which together might be used to bolster the diagnosis of BL in routine diagnosis [33]. BL that dichotomizes based upon EBV status and mutational profile also differ in their timing of the MYC translocation, occurring during SHM in EBV-positive BL and more typically during CSR in EBV-negative BL. Testing for TP53 mutations is also of value in BL, providing important prognostic information for risk stratification [68]. Cases previously reported as TdT+ BL are no longer classified as BL and rather should be designated as B-lymphoblastic lymphoma/leukemia with MYC rearrangements. They differ from bona fide BL at three levels: (1) the MYC::IGH translocation occurs during VDJ recombination in the bone marrow; (2) they harbor RAS mutations, and lack those prevalent in BL; and (3) they do not express a functional B-cell receptor [111]. An approach to the initial FISH-based genetic workup of aggressive B-cell lymphomas, including DLBCL-NOS, HGBL, and BL is portrayed in Fig. 4, and current genetic tests are summarized in Table 2.

A predominant genetic alteration has been identified in only a minority of nodal T-cell lymphomas, such as ALK or DUSP22 rearrangement (R) in anaplastic large T-cell lymphoma (ALCL). Although most nodal T-cell lymphomas have diverse molecular alterations, there are commonly affected cellular processes and signaling pathways [26]. In general, T-cell lymphomas are characterized by alterations in the TR and JAK/STAT signaling pathways, as well as mutations in epigenetic regulators [20]. These genetic alterations might have diagnostic, prognostic, or predictive impact in different entities (Table 3).

Systemic ALCL is divided in two groups based on the presence or absence of ALK-R on chromosome 2p23. The most frequent partner gene is NPM1 on chromosome 5q35.1, followed by a variety of other fusion partners that encode oncogenic ALK fusion proteins. The demonstration of ALK-rearrangement or protein by immunohistochemistry is diagnostic of the disease and has allowed the recognition of several histologic variants, including the common variant (60% of the cases), the lymphohistiocytic variant (10%), the small-cell variant (5-10%), the Hodgkin-like variant (3%) and cases with more than one pattern [25]. Constitutive ALK activation promotes tumorigenesis through the activation of multiple signaling pathways such as PLCγ, PI3K-AKT1, JAK-STAT, mTOR, and MEK-ERK [15, 78]. Most ALCL have clonally rearranged TR genes despite the failure to express TR molecules on the cell surface [9]. The mutational landscape includes recurrent mutations of TP53, epigenetic modifiers, and genes in the TR pathway [53]. More recently, it has been shown that ALK-positive ALCL has recurrent missense mutations in the extracellular domain of NOTCH1 (12%; distinct from the truncating mutations of the PEST domain in B-cell lymphomas), a potential therapeutic target [48].

ALK-negative ALCL, by definition, lacks ALK rearrangement, and although it is considered a single disease entity, molecular data indicates the existence of several distinct subgroups [73]. Similar to ALK+ALCL, activation of the JAK-STAT signaling pathway seems to play an important role and is in many cases secondary to STAT3, JAK1, and JAK3 mutations, but also through rearrangements involving ROS1, TYK2 of FRK. Two recurrent chromosomal translocations with characteristic clinicopathological features are recognized. ALK-negative ALCL with DUSP22-R (20–30%), has monomorphic cytology, absence of cytotoxic molecules and PD-L1 expression, lacks JAK-STAT pathway activation, has a characteristic gene expression signature and recurrent MSC mutations [55, 56]. The 2022 ICC considers ALK-negative ALCL with DUSP22-R a genetic subtype of ALK-negative ALCL with usually favorable outcome [12, 26]. FISH analysis for DUSP22 is recommended to identify this genetic subgroup (Fig. 5). The second recurrent translocation involves TP63 (5–8%) which encodes a p63 fusion protein and is associated with aggressive clinical behavior and poor outcome [73]. Truncated ERBB4 overexpression is observed in 24% of ALK-negative ALCL and represents a potential therapeutic target [89].

TFH lymphoma has distinctive clinical, morphological, immunophenotypic, and genetic features. Angioimmunoblastic T-cell lymphoma (AITL) is the prototype and best-characterized subtype [26]. The mutational landscape of TFH lymphoma includes genes involved in epigenetic pathways such as TET2 (up to 90%), IDH2 (20–45%), and DNMT3A (20–30%), mutations in the small GTPase RHOA (50–70%) and mutations in the TR signaling pathway including PLCG1, CD28, FYN, and VAV1 [105]. Recurrent translocations have also been identified including ICOS::CD28, ITK::SYK, and VAV1 fusions [23]. These genetic alterations are seen across the morphological spectrum of TFH lymphoma supporting the unifying concept of one disease entity. However, IDH2 mutations are exclusively identified in AITL with characteristic morphology (medium to large cells with clear cytoplasm) and immunophenotype (strong CD10 and CXCL13 expression) (Fig. 6) [97], whereas ITK::SYK fusions are mostly identified in the follicular type [98]. TET2 and DNMT3A are the most frequent mutations observed in clonal hematopoiesis (CH) [29], suggesting that CH plays an important role in the pathogenesis of TFH [52]. CH also explains the association of myeloid neoplasms with TFH lymphoma, particularly after cytotoxic therapy [52]. TFH lymphoma is probably the only T-cell lymphoma where mutational analysis is currently recommended to support the diagnosis [12]. In difficult cases, the presence of RHOA and/or IDH2 mutations supports the diagnosis of TFH lymphoma. In the interpretation of molecular findings, it is important to recognize the low allelic frequency often observed in ROHA and IDH2 mutations (VAF <5%) and to avoid misinterpretation of TET2 and DNMT3A as tumor-specific mutations [97].

PTCL, NOS represents a heterogenous group of lymphomas with variable morphology, immunophenotype, cytogenetics and molecular features. The diagnosis of PTCL, NOS often requires the confirmation through TR clonality analysis, supported by mutational analysis [100]. Recently, 2 molecular subgroups were identified through GEP analysis, namely PTCL-TBX21 and PTCL-GATA3 [38]. PTCL-TBX21 is characterized by low genomic complexity, higher frequency of epigenetic modifier genes (TET1, TET3, DNMT3A), and more favorable outcome. In contrast, PTCL-GATA3 is associated with high genomic complexity, gains in STAT3 and MYC, frequent mutations in TP53 and PRDM1, and poor outcome [36]. A novel immunohistochemical algorithm has been proposed to reproduce these two molecular groups using 4 antibodies; TBX21, CXCR3, GATA3, and CCR4 [5]. This algorithm needs validation but likely will be integrated in routine diagnosis in the future. Cytotoxic PTCLs are included within PTCL, NOS and molecularly cluster mainly in the PTCL-TBX21 group.

Primary nodal EBV-positive T- or NK-cell lymphoma is now separated from PTCL, NOS and is considered a provisional entity in the ICC [12, 79]. By definition, it is a systemic disease without nasopharyngeal involvement and affecting mainly elderly or immunocompromised patients. The most common mutated genes are TET2, PIK3CD, DDX3X, and STAT3. A characteristic finding is recurrent losses of 14q11.2 where the TRA/D genes are located, supporting the T cell derivation of this lymphoma [69].

Some neoplasms derived from mature T and NK cells, mostly from innate immune cells, affect predominantly extranodal sites [21]. These include extranodal NK/T-cell lymphoma (ENKTCL), nasal type, the group of gastrointestinal lymphomas, hepatosplenic T-cell lymphomas (HSTCL), and the rare group of γδ T-cell lymphomas and subcutaneous panniculitis-like T-cell lymphomas (SPTCL). Other neoplasms involve predominantly the peripheral blood and bone marrow such as T-cell large granular lymphocytic leukemia (T-LGLL), T-cell prolymphocytic leukemia (T-PLL), adult T-cell lymphoma/leukemia (ATLL) and aggressive NK-cell leukemia (ANKL). The diagnosis of these diseases is based on their characteristic clinicopathologic features. Besides TR clonality assessment (with the exception of ENKTCL if derived from NK-cells and ANKL, which lack clonal TR rearrangements), molecular analysis is not required for the diagnosis. Nevertheless, extranodal T- and NK-cell lymphomas share the activation of the JAK/STAT pathway, a potential therapeutic target [20]. In difficult differential diagnosis, the mutational pattern, although not specific, might favor one disease over another (Table 4).

Enteropathy-associated T-cell lymphoma (EATL) is characterized by frequent mutations in the JAK/STAT pathway, most commonly involving JAK1 and STAT3 and rarely JAK3, STAT5B, TYK2, or SOCS1. KMT2D and TET2 are also frequently mutated in EATL [17]. In contrast, monomorphic epitheliotropic intestinal T-cell lymphoma (MEITL) is characterized by mutations or deletions of SETD2 (>90%), resulting in reduced or absent of expression of lysine 36 in histone H3 (H3K36me3). Other characteristic co-occurrent mutations are STAT5B (<60%) and JAK3 (35–50%). The absence of H3K36me3 expression by immunohistochemistry is an excellent surrogate marker for SETD2 inactivation [81].

ENKTCL and ANKL share a similar mutational profile, frequently affecting the JAK/STAT pathway (STAT3, STAT5B, JAK3), epigenetic regulators (BCOR, KMT3D, ARID1A, EP300), tumor suppressor genes (TP53, MGA) and the RNA helicase DDX3X [21, 61, 79].

The differential diagnosis of HSTCL, T-LGLL, and γδ T-cell lymphoproliferations is not always straightforward. Characteristic genetic alterations of HSTCL are the presence of isochromosome 7q (80%) and trisomy 8 (50%) regardless of the αβ/γδ derivation [77]. Inactivating mutations in SETD2 (70%), and in other chromatin-modifying genes (INO80, TET3, SMARCA2) are frequently identified (62%). Other mutations frequently observed are STAT5B (31%), STAT3 and PIK3CD and rare cases also carry TP53, URB5, and IDH2 (5–10%) mutations [20]. In contrast, T-LGLL carry STAT3 mutations in up to 50% of the cases and co-occur frequently with KMT2D and SETD1B mutations [14]. STAT5B mutations are mainly associated with the indolent CD4+ T-LGLL and a subgroup of CD8+ aggressive T-LGLL [8].

T-PLL is characterized by 14q chromosomal inversions with breakpoints in the long arm at q11 and q32 seen in 80% of the cases. These alterations involve the TCL1 family genes and less often MTCP1 gene, which is homologous to TCL1A/B at Xq28. Recurrent mutations in the JAK/STAT pathway are observed up to 75% of the cases. Monoallelic deletions and/or mutations of ATM are common [20, 99]. Demonstration of TCL1 chromosomal alteration by FISH or TCL1 expression in T-cells by immunohistochemistry is diagnostic of T-PLL.

Germline mutations of the HAVCR2 gene have been demonstrated in patients with SPTCL [28]. Two hot spot mutations have been identified, one predominates in East Asia (HAVCR2^Y82C) and the other in Europeans (HAVCR2^I97M). HAVCR2 encodes for TIM3, an inhibitory receptor expressed on interferon-γ producing T-cells involved in regulating inflammation. Patients with HAVCR2 mutation are younger (<30 years), often associated with hemophagocytic lymphohistiocytosis (HLH) and worse prognosis [95]. The investigation of HAVCR2 mutations may allow risk stratification and provide a target for therapy.