Common cytogenetic alterations

Copy number alterations (CNAs) and karyotypic complexity are key features with clinical utility in other B-cell neoplasms such as CLL [37, 38]. In SMZL, a G-banding study of 330 cases found that 50% of cases harboured complex karyotypes (as defined by having ≥ 3 cytogenetic aberrations per metaphase cell), while 30% of cases harboured single aberrations, and 18% of cases harboured two [39]. The most common abnormalities include 7q22-q32 deletions and chromosome 3/3q gains (approximately 25%, Figure 2A, Table 2). Complex karyotypes involving chromosomes 1, 3, 7, 8, and 14 are also frequent [39–41]. TP53 deletions, occurring within the del(17p) minimally deleted region (MDR), are found in 8–20% of SMZL cases and are associated with adverse outcomes [39, 42].

Display full size

Figure 2. 

(A) Ideogram of recurrent copy number alterations in SMZL. Yellow indicates a trisomy, while red and green indicate common regions of deletions and gains, respectively; (B) schematic diagram of the recurrent 7q deletion in SMZL. The chromosome 7 ideogram indicates the common breakpoints of deletion found among SMZL cases. The panel below shows an example of a copy number plot of a case with clear lower segmentation means indicating the 7q deletion. A total of 38 coding genes and 6 microRNAs (miRNAs) are found in the 7q32 minimally deleted region (MDR)—those that were found to be transcriptionally under-expressed are indicated by the yellow highlight [43]. SMZL: splenic marginal zone lymphoma

In SMZL, the most prominent cytogenetic feature is del(7q), detected in 14–44% of patients [39, 44, 45, 48]. The breakpoints have been shown to vary from 7q21 to q36, but a 3.04 Mb MDR at 7q32.1-32.2 (127.03–130.07 Mb) has been identified, with another less common deletion region at 7q22 [43, 45, 46] (Figure 2B). A study comparing cases with or without del(7q) identified 38 coding genes within the 7q MDR [43, 46], and of those genes, 18 were significantly under-expressed in cases with the deletion and some of these had further evidence for hypermethylation [46]. Additionally, a cluster of six miRNAs in the del(7q) MDR exhibited reduced expression, including miR-593, miR-129, miR-182, miR-96, miR-183, miR-335, miR-29a, and miR-29b1 [46]. Integrated CNA and whole-exome sequencing (WES) analysis identified three mutated putative candidate genes of this deletion, CUL1, EZH2, and FLNC, with FLNC located within the mapped MDR [53].

The second most common aberration is the dup(3q), which is present in up to 34% of patients and is sometimes the result of unbalanced translocations [44, 48]. In contrast to other MZLs, where whole trisomy 3 is predominant, SMZL frequently exhibits a gain in 3q23, often in conjunction with a complex karyotype [39]. The breakpoints and partners in dup(3q) rearrangements vary, indicating secondary events [84]. Compared with the more aggressive diffuse large B-cell lymphoma (DLBCL), SMZL genomes infrequently harbour balanced chromosomal translocations, though IgH locus translocations can occur (9% of cases) in t(14;19)(q32;q13), t(6;14)(p21;q32), t(9;14)(p13;q32), and t(1;14)(q21;q32), targeting BCL3, CCND3, PAX5, and BCL9, respectively [39, 85]. The (14;19)(q32;q13) translocation was first discovered in CLL, with the identification of BCL3 as the translocation partner of the IgH locus [49]. A cluster of breakpoints in the 3’ region of BCL3 was shown to be associated with cases of atypical SMZLs with mutated IGHV status and complex karyotypes-this breakpoint did not, however, result in changes in expression of the BCL3 [86]. Hence, in such cases, it is still not clear what the relevant target genes of these translocations may be [50].

SMZL mutations cluster within common pathways

Genome-wide studies, employing unpaired whole-genome sequencing (WGS), WES, and targeted sequencing technologies, have revealed recurrently mutated genes and pathways in SMZL development. However, these approaches have limitations in exploring non-coding genome regions and in investigating mutational signatures contributing to disease. KLF2 (12–42%), NOTCH2 (10–25%), and TP53 (15%) are among the most frequently mutated genes [28, 32, 53–55] (Table 2). Recurrent truncating mutations are observed in TNFAIP3, KMT2D, and MYD88, where the latter results in the promotion of B-cell proliferation and survival by activation of NF-κB [72]. Additionally, other genes within pathways that relate to crucial B-cell functions are also mutated, including the NOTCH pathway, NF-κB signaling, chromatin remodeling, cytoskeleton organization, and toll-like receptor (TLR) signaling pathways.

The NF-κB pathway is deregulated in SMZL

In B-cells, NF-κB activation influences signaling pathways and downstream genes fundamental to development, differentiation, and survival [87, 88]. NF-κB is typically sequestered in an inactive state in the cytoplasm by inhibitory IκB proteins. There are two pathways of NF-κB signaling: canonical and non-canonical. These pathways involve distinct NF-κB family members and are triggered by specific receptor-proximal signals, such as those emanating from the BCR, tumour necrosis factor-alpha (TNF-α), TLRs, and interleukin-1 (IL-1). It involves phosphorylation and degradation of IκB proteins by IκB kinase α (IKKα) and IKKβ, releasing and activating NF-κB heterodimers (e.g., p65/RelA) that translocate to the nucleus to regulate gene expression. Conversely, the non-canonical pathway is activated by stimuli like B-cell activating factor (BAFF) and a proliferation-inducing ligand (APRIL), which bind to receptors such as BAFF-receptor (BAFF-R) and CD40, leading to activation of IKKα. This results in the processing of p100 to p52-REL-B heterodimers, which regulate gene expression. The TNFAIP3 (A20) protein and other negative feedback mechanisms modulate its activation [89].

Constitutive NF-κB activation has been implicated in the growth of various B-cell leukaemia and lymphomas, with mutations in multiple genes playing a part. Constant NF-κB activation in MZ B-cells models results in abnormal B-cell expansion [90], and enhanced signaling is observed in other lymphoid malignancies including Hodgkin lymphomas, DLBCL, and multiple myeloma [91]. Genetic abnormalities in genes involved in this pathway including CARD11, TNFAIP3, TRAF3, and BIRC3 [53, 54, 92] have also previously been reported (Figure 3A).

Display full size

Figure 3. 

Recurrently mutated genes and pathways in splenic marginal zone lymphoma (SMZL). (A) Illustration of the NF-κB signaling pathway. Response to diverse stimuli, including ligands of B-cell receptors (BCRs) and toll-like receptors (TLRs) leads to the primary mechanism of nuclear factor kappa B (NF-κΒ) activation through the inducible degradation of the IκB, which is triggered by the site-specific phosphorylation of the multi-subunit IκB kinase (IKK) complex. Recurrent coding mutations in KLF2, TRAF3, TNFAIP3, BIRC3, CARD11, and BCL10 are recurrently found across SMZL; (B) illustration of the NOTCH signaling pathway. When NOTCH2 is activated by the binding of ligands such as Jagged ligands and delta-like ligands, its intercellular domain [Notch intracellular domain (NICD)] is cleaved and released into the cytosol, where it then traffics into the nucleus to act as a transcriptional regulator. Mutations in SPEN, NOTCH1, and NOTCH2 are recurrently found across SMZL; (C) types of unique mutations found across recurrently aberrant genes in SMZL; (D) distribution of mutations found across KLF2 (top) and NOTCH2 (bottom) in SMZL. TNFR: tumour necrosis factor receptor; BAFFR: B-cell activating factor receptor; NICD: Notch intracellular domain; NECD: Notch receptor extracellular domain; MAM: mastermind; TRAF3: tumour necrosis factor-receptor associated factor; FS: frameshift; RPBJ: recombination signal binding protein for immunoglobulin kappa J region. Figure 3A and 3B were created with Biorender.com

Note. Figure 3A was adapted from “Notch Signaling Pathway”, by BioRender.com (2020). Retrieved from: https://app.biorender.com/biorender-templates/figures/all/t-5f6d097dbedf1200aa63bc9c-notch-signaling-pathway; “NF-KB Signaling Pathway”, by Iwasaki A (2020). Retrieved from: https://app.biorender.com/biorender-templates/figures/all/t-5f5b7c3514d40300aa942984-nf-kb-signaling-pathway

The TNF-receptor associated factor (TRAF) proteins act as docking molecules for the interaction between the cytosolic regions of various tumour necrosis factor receptors (TNFRs) and TLRs. Studies have shown the action of the A20 protein, encoded by the TNFAIP3 gene, inhibits NF-κΒ activation in the TNFR1 signaling pathway with the help of several binding partners including TRAF1 and TRAF2 [93, 94]. TRAF3 has recurrently mutated in a variety of B-cell malignancies, including SMZL (21%) [95]. Inactivation of TRAF3 has been found to increase the survival of mature B-cells, leading to the development of B-cell lymphomas in murine models [70]. Mutational clusters were found within the C-terminal domain of the protein, which is required for BIRC3 protein recruitment. Alternatively, truncating mutations within the RING domain of BIRC3 occur in 10% of SMZLs and result in the loss of ubiquitin ligase activity, further stabilizing NF-κB-inducing kinase (NIK) and resulting in the constitutive upregulation of NF-κB targets [93, 96]. Mutations that result in the activation of positive regulators of the NF-κB pathway, such as in CARD11 and IKBKB, have also been identified in a proportion of SMZL cases [93, 97].

Genomic subtypes

Bonfiglio et al. [59] identified subgroups of SMZL based on the enrichment of particular gene mutations clustering into specific biological pathways. The authors studied 303 splenic-derived tumour specimens from an international multicentre study (IELSG46) using a multi-omic approach. This study used a pathology expert panel, confirming the WHO classification for all cases. Using genetic and phenotypic disease features, they were able to identify a number of genetic subtypes, with differing survival outcomes. Most importantly, of the four total clusters, two prominent subgroups were proposed, termed NNK (mutations in NF-κB, NOTCH, and KLF2, 58.2% of all cases), and DMT [with DNA-damage response, mitogen-activated protein kinases (MAPK), and TLR modules, 32% of cases]. These sub-types exhibited different patterns of chromosomal lesions, immunogenetic features, transcriptional signatures, immune microenvironments, and clinical outcomes, with inferior survival in the NNK group.

The co-occurrence of mutations within the NNK cluster, which accounts for the majority of cases, alludes to a possible role of KLF2 as a master regulator of both NOTCH and NF-κB signaling, suggesting the possible cooperation between these lesions in dysregulating MZ B-cell differentiation. As KLF2, an important transcription factor, is mutated in the immune-suppressive NNK SMZL subgroup, future studies might focus on how deregulation of KLF2 or downstream targets results in immune cell reprogramming, recruitment, or evasion, as has been shown for CREBBP and EZH2 in other B-cell lymphomas [82].

Diagnosis

No single genomic or immunogenetic marker is pathognomonic for SMZL. However, in the context of the differential diagnosis of a primary splenic lymphoma, the use of IGHV1-2*04 and/or del(7q) are very strong pointers for the diagnosis of SMZL, while a combination of these markers with or without mutations of genes within the NF-κΒ pathway is not seen in other splenic lymphomas. As discussed earlier in this review, transcriptomic studies may also have diagnostic potential. DNA methylation studies of normal B-cell maturation and more common B-cell tumours have identified both disease-specific signatures and differing patterns within diseases that have clinical implications. A recent study, published as an abstract, investigated DNA methylation profiles in splenic lymphomas with the pathological diagnoses of HCL, HCL-V, SDRPL, or SMZL, and identified 5 distinct methylation subgroups [128]. While the majority of the 170 SMZL cases formed a distinct subgroup, 29 were included in subgroups encompassing the new WHO classification of SBLPN. These cases were depleted in NOTCH2, KLF2 mutations, and IGHV1-2*04 usage, suggesting either that they could be reassigned to the new WHO subgroup or there were differences in the criteria used to diagnose SMZL.

Prognosis

Early studies indicated that both cytogenetic abnormalities and genomic mutations may have prognostic significance although the results were not always concordant among studies [54–56]. A cytogenetic study identified karyotypic complexity, 14q abnormalities and del(17p) as markers of a shorter overall survival in univariate, but not in multivariate analysis which included standard clinical and laboratory parameters and IGHV mutational status [39]. A multivariate analysis of 175 patients showed that NOTCH2 mutations and 100% IGHV identity were associated with a shorter time to first treatment, while TP53 mutations were associated with shorter overall survival [29]. It was, however, unclear whether TP53 abnormalities impacted the survival of most patients with SMZL who received treatments that act through p53-independent mechanisms.

As lymphoma-specific deaths account for less than one half of total deaths in SMZL, Bonfiglio et al. [59] used a classifier and survival relative to a matched general population to assess excess mortality in patient subgroups based on genomics. After adjusting for the effects of age, sex, and year of diagnosis, the 10-year life expectancies of patients with NNK-, DMT-, and “immune-suppressive”-SMZLs were reduced by 21.0%, 14.5%, and 20.4% respectively. Cases with both the NNK genotype and an “immune-suppressive” microenvironment, had a relative survival of 70.8% at 10 years. Furthermore, 99.4% of mutations detected in the spleen were also identified in blood or marrow samples, offering less invasive avenues for biopsy procedures.

Additionally, epigenetic alterations have emerged as valuable clinical indicators, serving either as a metric for cellular proliferation [117], or as a global measure of hypermethylation (High-M) [115]. Both these studies emphasise how global changes in the epigenome, such as those correlated with cell proliferation, may have greater prognostic significance than any individual genomic abnormality. A recent study by Hopper and colleagues [129] employed bulk RNA-sequencing on a cohort of indolent lymphomas, including 48 SMZL tumours. They identified five putative patient clusters, three of which were strongly enriched for an SMZL diagnosis. One cluster, which the authors termed LGBCL5, was enriched for genes involved in cell cycle control, T-follicular helper cells, mutations in TNFAIP3 and BCL2, and M1 macrophages in the tumour microenvironment (TME) inferred from the CIBERSORTx software. This cluster was also associated with a higher incidence of transformation to a high-grade lymphoma [129].

Transformation

Histological transformation to a DLBCL is associated with a poor outcome in SMZL. In murine models, concurrent inactivating mutations of BLIMP1 and members of the NF-κB pathway can promote lymphomagenesis, leading to DLBCL with similar characteristics to human disease [130]. Genes involved in a proliferation signature were also found to be some of the most overly expressed upon the transformation of FL into DLBCL in human samples [131]. A study of 36 cases identified karyotypic complexity as the only risk factor for transformation in a multivariate analysis [21]. A more recent study [51] employed copy number arrays and a next-generation sequencing (NGS) panel to investigate genomic abnormalities in 41 SMZL patients with high-grade transformation, of whom 18 were studied at both diagnosis and at transformation (SMZL-T). The median time from diagnosis to transformation was 2.4 years (range 0–17) and the median survival from transformation was 6.55 years. SMZL-T clones arose by divergent evolution from a common altered precursor cell. Compared to SMZL, SMZL-T had higher genomic complexity. The most frequent abnormalities in SMZL-T were TNFAIP3 (59.4%), KMT2D (46.9%), TP53 (34.4%), ARID1A (31.3%), and KLF2 (31.3%) mutations, losses of 9p21.3 (CDKN2A/B, 40.6%) and 7q31-q32 (34.4%), and gains of 1q (40.7%). Four cases with a TP53 mutation at diagnosis, acquired a del(17p) at transformation. The main genomic variables predicting shorter survival from transformation were KLF2 mutations, a complex karyotype, and a trend for MYC gains. These results identify a subgroup of patients in whom more regular follow-up could be indicated, although it is unclear whether earlier treatment would improve outcomes.

What is the non-coding mutation landscape of the disease?

The research community is yet to generate the genome-wide data on coding and non-coding mutations in SMZL, from matched WGS of tumour and germline duos. The previous WGS study by Kiel et al. [55] did not include matched germ-line material from the SMZL patients, which is critical to generating a catalogue of the mutational landscape of any tumour, particularly when it comes to the non-coding space. This is because comparison to germline sequencing is required for adequate accuracy and sensitivity for the detection of high-confidence somatic variants. In contrast, in other B-cell tumours, such as CLL, larger cohorts with matched germline material have identified an expansive panel of genomic regions and genes targeted by non-coding mutations, such as PAX5 [132] and BACH2 [133]. Further insights into the key mutational signatures from paired WGS would be crucial in understanding the mechanisms involved in disease development and could potentially yield insights into malignant transformation. From a patient perspective, mutational signatures could provide insight into exposure to potential genotoxins helping to manage cancer risk, and might identify suitable treatment options for patient subgroups, such as poly- adenosine diphosphate (ADP)-ribose polymerase (PARP) or check-point inhibitors in patients with mutational signatures that suggest a deficiency in homologous recombination and mismatch repair, respectively.

What drives clinical outcomes in poor-risk SMZL?

This review outlines several studies highlighting the molecular features of poor-risk SMZL. Taking these studies together, a key aggressive subset can be defined, harbouring distinct, partially overlapping clinico-biological characteristics. Immunogenetically, these cases are defined by a highly biased usage of the IGHV1-2*04 allele, with intermediate levels of SHM that appear to persist throughout tumour development. At the genomic level, these cases harbour del(7q), genome complexity, short telomeres and mutations within key regulators of marginal-zone development and NF-κΒ signaling (NOTCH2 and KLF2). This group of cases also harbours high levels of global hypermethylation, including deregulation of the PCR2 complex, and evidence of elevated historical tumour proliferation with the capacity for future cellular growth. However, not all cases carry all these features, so it will be important to understand which has the greatest impact on disease biology and clinical outcome; these studies will require high-quality biological material for omics and functional analysis, from well-characterised patients with extensive follow-up, drawn from a large international consortium.

Can we deconvolute tumour and normal cells in the SMZL TME?

In humans, the MZ is located at the interface between the red and white pulp, where macrophages and resident B-cells can rapidly respond to antigens in the circulating blood supply. How the development of SMZL in the spleen might deregulate normal interactions between different cell populations remains poorly studied. In other tumour types, single-cell sequencing (SCS) technology has revealed insight into cellular components, functions, and interactions by circumventing noise caused by genetic and phenotypic variations across individual samples. In SMZL, it could enable direct measurement of molecular characteristics in thousands or even millions of individual cells, facilitating the characterization of cell types and revealing insights into sub-clonal dynamics, gene regulatory networks, transcription dynamics, and proteomic profiles. The precise identification of sub-clones could be crucial for understanding shifts in gene expression and for gaining insights into the SMZL TME and inflammatory cell milieu, where preliminary studies implicate a high-risk proinflammatory TME [134]. The importance of the TME and its contribution to different disease sub-types is shown in the study by Bonfiglio et al. [59], showing an “immune-suppressive” class encompassing macrophage, cytokine, and T-cell signatures, and an “immune-silent” class dominated by B-cell signatures. The presence of an ‘immune-suppressive’ subgroup opens the possibility of checkpoint inhibitor treatment in these patients.

The analysis of IGHV1-2*04 tumours in the spleen could further confirm the presence of ongoing SHM by using VDJ sequencing, and documenting interactions with T-cells by using TCR single-cell tracking. This type of immune repertoire analysis should maximize the development and deployment of future immunotherapies. SCS information could be powerful in predicting tumour progression, and transformation and for predicting drug sensitivity or resistance. As tumour cells evolve under treatment pressure, spontaneous mutations can confer drug resistance to a subset of cells, where single-cell proteomic analysis could enable a more accurate correlation between gene mutations and protein expression, elucidating diverse clinically relevant mechanisms. The application of spatial biology technologies could provide further insight into the organization and cell-to-cell interactions between tumour, immune and stromal cells, as well as their distributions within the SMZL TME. Cellular confirmation will ultimately be required, and whilst these studies are also rare, due to the paucity of variable material for study, they have already shown impaired T-cell and myeloid cell function in a proportion of patients [135–137].

When will these new molecular insights be translated for the benefit of patients?

The findings outlined in this review have significant clinical relevance and promise to enhance diagnosis and patient stratification in the future, provided that standardized testing protocols can be established. Achieving this goal presents challenges, necessitating the use of assays that capture the most clinically informative immunogenetic and (epi)genomic data. A single platform could be developed to perform: IGHV analysis, identification of gene mutations and CNAs linked to survival (del(7q), KLF2, NOTCH2, TP53, TNFAIP3), as well as those aiding in the differential diagnosis (e.g., MYD88 for WM, BRAF for HCL, NOTCH1 and SF3B1 for CLL, PTPRD for NMZL, and MAP2K1 for HCL-V), and assessment of proliferative history using DNA methylation data. As genomic knowledge expands, these platforms could be augmented with additional prognostically relevant and therapeutic target lesions, potentially facilitating treatment monitoring. While comprehensive multi-omics analysis, including epigenetic profiling, may offer the most precise patient identification for precision medicine, its adoption must be economically justified by selecting optimal treatment approaches. Regardless of the technical approach, each molecular marker must undergo rigorous validation across multiple cohorts, retrospective studies, and prospective trials to establish robust associations with disease outcomes. Analytical and clinical validation, followed by international harmonization, regulatory approval, and ongoing accreditation, are essential steps. Furthermore, guidelines for interpretation and reporting, along with appropriate training for clinical referral and patient communication, are imperative for effective implementation in clinical practice.