Genetic mechanisms underlying tumor microenvironment composition and function in diffuse large B-cell lymphoma

Diffuse large B-cell lymphoma (DLBCL) is a biologically and clinically heterogenous disease. Gene expression analysis based on cell-of-origin classification identified 2 major molecular subtypes of DLBCL: activated B cell (ABC) and germinal center B cell (GCB). These subtypes have distinct clinical behavior and molecular features, reflecting differential pathogenesis.¹ More recently, studies of genomic alterations in DLBCL cells, including mutations, somatic copy number alterations, and structural variants, identified genetic classes within cell-of-origin subtypes.^2,3 Lymphomagenesis, progression, and treatment resistance were once thought to be driven almost entirely by these cancer cell–intrinsic alterations. Only in the past decade have researchers and clinicians come to recognize the important roles of nearby nonmalignant cells, in what is now called the tumor microenvironment (TME).

TME cells respond to the changes imposed by the presence of cancer cells and other cancer-influenced TME cells with adaptive changes in their gene expression, epigenetic landscape, and metabolome. These phenotypic adaptations typically include dysfunction for cells that normally would have antitumor roles (eg, an exhausted phenotype in cytotoxic CD8 T cells), whereas other cells become tumor-supporting, “cancer-associated” cells, often to such an extent that the tumor cannot survive or proliferate without them. This process is accompanied by signaling cross talk that drives the coevolution of malignant and nonmalignant cells.

The TME comprises numerous cell types, each of which consists of heterogeneous subsets with various phenotypes and diverse functions. Recent studies have found that TME cells show significantly increased heterogeneity of cell states and marked phenotypic expansions compared with cells in normal tissues.⁴ In lymphoma, malignant B cells also exhibit diverse cell states,⁴ which adds to the complexity of tumor-TME cross talk and to the challenge of deciphering it. Other layers of complexity are due to the establishment of long-range interactions (eg, effect of interleukins and/or metabolites on distal cells that eventually populate the TME),⁵ the biomechanical properties conferred by the extracellular matrix (eg, stiffness and porosity affecting the rate and degree of TME cell infiltration) and the physical features of the TME (eg, oxygen and metabolic gradients affecting TME cell functionality).

The proportions of cell subtypes and, to great extent, their functional state, which can include cytokine production and extracellular matrix remodeling activity, can be captured by their transcriptional signatures. Artificial intelligence–based computational techniques allow for the comprehensive analysis of big transcriptome data sets and provide biological frameworks for experimental and clinical data interpretation.^4,6 Using these techniques, researchers have described 4 major categories of “lymphoma microenvironments” (LME) in DLBCL, each carrying distinct clinical and biological characteristics: the germinal center–like LME (GC-LME), with cell types commonly found in germinal centers; mesenchymal LME (MS-LME), so-called for its abundance of stromal cells and extracellular matrix signatures; inflamed and immunosuppressive LME (IN-LME) dominated by inflammatory cells and exhausted and suppressed cytotoxic cells; and depleted LME (DP-LME), which overall has less prominent TME-derived signatures (Figure 1).⁶ Observations from this analysis indicate that the presence or absence of specific cell subtypes is insufficient to fully understand the relevance of the TME. For example, the 2 LME categories (MS- and DP-LME) with lower T-cell infiltration carry good and poor prognoses, respectively.

There is evidence that in the earliest stages of lymphoma, most of the cells in and around the site of tumor origin work to avert malignant transformation and/or limit malignant-cell proliferation, essentially acting as an “external” oncogenic checkpoint,⁷ similar to the intrinsic molecular checkpoints in healthy B cells.^8,9 This apparent external checkpoint function is not exclusively provided by immune cells; for example, cancer-associated fibroblasts (CAFs) in the TME secrete soluble factors such as transforming growth factor ligands¹⁰ to help curtail lymphoma progression. Under the influence of the incipient tumor, however, these cells begin to change and, in later developmental stages, make up a fully lymphoma-supportive TME. Lymphoma cells may drive this microenvironment evolution by deploying checkpoint-evading mechanisms, for example, by epigenetic reprogramming, as well as mechanisms that directly reprogram TME cells to become cancer-neutral or cancer-supporting entities (Figure 2). In support of this view of TME evolution, investigators have found that the GC-LME category of DLBCL, whose cellular constitution more closely resembles healthy lymphoid tissue, is more likely to respond to treatments and have better outcomes, whereas a category such as DP-LME, with a fully reprogrammed TME, is less likely to respond to treatments.^6,11 

An increasing body of clinical observations and murine lymphoma models indicates that specific somatic mutations leading to oncogene activation and loss of tumor suppressor genes are correlated with changes in TME composition and/or the functionality of TME cells.^12-14 Genetic variants associated with ethnicity and multifactorial systemic processes, such as aging, which drive phenotypic modifications in noncancer cells, may also affect the TME.

There are at least 3 types of mechanism by which mutations can affect TME composition and function (Figure 3): (1) mutations of membrane ligands or receptors; (2) mutations affecting signaling pathway transducers or scaffolding molecules that induce pathway rewiring and thus alter how cancer cells respond to TME cues; and (3) expression and/or secretion of short- and long-range acting molecules (these mechanisms will be discussed separately below). Note that most TME-relevant genetic changes have a combination of direct and indirect as well as short- and long-range effects on the TME; moreover, due to the genetic heterogeneity of DLBCL, TME changes typically are driven by multiple genetic alterations. In fact, the frequent co-occurrence of multiple genetic alterations have been used to defined “genetic classes” of DLBCL; for example: MCD (MYD88 and CD79B mutations), EZB (EZH2 mutations and BCL2 translocations, with and without MYC alterations), BN2 (BCL6 fusions and NOTCH2 mutations), ST2 (SGK1 and TET2 mutations), and A53 (aneuploid with TP53 mutations).³ Some of these genetic classes are more represented in specific types of TMEs (Figure 1); however, for most cases, the mutation responsible for a particular TME feature cannot be individualized.

Disruptions of antitumor immune mechanisms are among the most prominent drivers of TME transformation. These disruptions can occur from the earliest to the latest stages of cancerous progression, including in response to selective pressures exerted by immunotherapy. The vast majority can target the expression of the antigen processing and presentation pathway (APPP), as well as costimulatory and coinhibitory molecules. Mutations affecting the APPP generally contribute to a TME with low immune infiltration even for DLBCLs with high immunogenicity (eg, lymphomas with high activity of activation-induced cytidine deaminase).² In DLBCL, the most prevalent APPP alteration is the lack of expression of the class I major histocompatibility complex (MHC-I); it is estimated to affect ∼40% to 75% of patients, depending on the methodology used for assessment.^2,3,15,16 MHC-I molecules are heterodimers of one of the HLA-I heavy chains and β2-microglobulin (B2M) that present viral and tumor peptides to engage CD8⁺ cytotoxic T cells (CTC) through a T-cell receptor complex.¹⁷ Mutations in HLA-I and B2M account for 80% of MHC-I–negative DLBCL, more commonly seen in GCB-DLBCL, and/or harboring high mutational burden¹⁵ and/or with an MS-LME category.⁶ In addition, as many as 70% of DLBCLs that express MHC-I carry monoallelic HLA-I genetic alterations that limit the repertoire of neoantigen presentation to CTCs and, similar to MHC-I–negative DLBCLs, harbor high mutational and inferred neoantigen load.¹⁵ Although lack of HLA-I expression in GCB has not been shown to lead to lymphoma in animal models,¹⁵ disruption of MHC-I has been reported in the progression from follicular lymphoma to DLBCL.¹⁸ 

Approximately 20% of DLBCLs carry genetic lesions disrupting the CD58; these lesions are seen more frequently in ABC-DLBCL and usually affect CD58 protein expression.¹⁹ CD58 interacts with CD2, a costimulatory surface receptor that is expressed by T and natural killer (NK) cells and is required for CD8⁺ CTC and NK cell–mediated cytolysis of DLBCL cells. Genetic lesions, specifically gains and translocations involving 9p24.1, lead to overexpression of PDL1 and/or PDL2 in up to 25% of DLBCLs.² DLBCLs harboring alterations in PDL1 are frequently of the ABC-DLBCL subtype, associated with decreased HLA-I and/or -II expression and higher infiltration of clonally restricted CD8⁺ T cells.²⁰ Although not so well characterized mechanistically, mutations in TNFRSF14 cause immune synapse alterations leading to recruitment of T follicular helper (TFH) cells and enrichment of stromal cells in the TME²¹ that may underlie their association with GCB-DLBCLs harboring the GC-LME category of TME.⁶ GC-LME also presents the highest proportion of dendritic cells (DC), and the most common lymphoma subtype with this category of TME is the EZB MYC-negative GCB-DLBCL.⁶ Remarkably, some EZB-class GCB-DLBCLs show a tendency to acquire specific N-glycosylation sites in their B-cell receptors (BCRs) during somatic hypermutation; this enables the BCRs, upon engaging the putative dendritic cell-specific ICAM-3 grabbing nonintegrin ligand in DCs, to provide supportive proliferation signaling to lymphoma cells.²² Approximately 5% of DLBCLs carry TMEM30A flippase mutations causing a phospholipid imbalance in the outer plasma membrane leading to increased BCR mobility and signaling.^23,TMEM30A loss-of-function (LOF) mutations and gene losses are associated with a DP-LME.⁶ CD70 costimulation of CD27⁺ T cells favors the immune surveillance of malignant B cells. CD70 losses and LOF mutations have been described in DLBCL, frequently in association with BCL6 alterations.² In Bcl6^tg/+ mouse models engineered to bear Cd70^−/− and Cd70^wt lymphomas, the loss of Cd70 decreased the expansion and persistence of CD8⁺ and CD4⁺ CTCs in their TME.²⁴ Several other ligands and receptors have been described as mutation targets in DLBCL, including CD83, CD44, and TNFSF9 (4-1BBL), but their contributions to specific aspects of the TME have not been fully characterized.

Mutations affecting transcriptional and/or epigenetic regulators can affect the expression of ligands and receptors.²⁵ Because B cells are professional antigen-presenting cells, they usually express MHC-II in a process driven by the transcriptional regulator CIITA. Mutations in CIITA, which cause loss of MHC-II expression, are enriched in EZB-class GCB-DLBCL and Epstein-Barr virus (EBV)⁺ DLBCL.^3,26,CIITA mutations are associated with reduced T-cell infiltration and an MS-LME category and in EBV⁺ DLBCL with dysfunction of CD4⁺ CTCs.^3,6,26,27 

Mutations in epigenetic regulators^28,29 (eg, EZH2, CREBBP, KMT2D, and EP300) commonly target the APPP pathway, causing, for example, partial or complete loss of MHC-I and/or MHC-II expression. These alterations are frequently associated with low T-cell infiltration TMEs, such as the MS- and DP-LME categories, as well as low cytotoxic activity from CD8⁺ T cells and NK cells (in cases with low MHC-I expression) and CD4⁺ T cells (in cases with low MHC-II expression).^12,30,31 In addition to epigenetic mutations affecting on chromatin, aberrant DNA methylation of APPP components has been described in DLBCL, and although it is not mechanistically linked to specific mutations, it tends to be more prevalent in ABC-DLBCLs and/or in tumors with high genomic instability and/or harboring a DP-LME.^6,12 Unlike genetic alterations targeting APPP components, epigenetic downregulation of these molecules can be restored using EZH2 inhibitors, histone deacetylase inhibitors, or hypomethylating agents.^6,32,33 Although in animal models APPP deficiency can phenocopy most of the immune TME effects of epigenetic mutations, in patients with DLBCL the improvement in the TME immune infiltration seems to be independent of the restoration of APPP expression. This observation suggests the possibility of additional mechanisms at play, probably related to tumor antigenicity (ie, number of tumor antigens and/or neo-antigens present in cancer cells). Other genomic alterations associated with a low TME cytolytic score (a measure of the transcriptional signature of cytotoxic T cells and NK cells) include 7q amplification, BCL2 mutations, BCL2 translocations in GCB-DLBCL, and TP53 loss of heterozygosity (LOH) and LOF mutations.^6,12 Contrarily, mutations in ETV6, ETS1, and DTX1 (particularly in GCB-DLBCL), are associated with increased TME cytolytic scores¹² (Figure 4). Although there are several mechanisms by which TP53 interacts with the TME (some discussed below), TP53 LOH and LOF mutations may decrease the expression of target genes such as those for some Toll-like receptors^34-36; in addition, gain-of-function TP53 mutations that induce ENTPD5³⁷ may modify the folding and expression of N-glycosylated membrane proteins.

Aberrant expression of the transcription factor MYC due to genetic or other alterations, leads to profound changes in the TME, mostly resulting in immune depletion. For example, the proportion of DP-LME in EZB class GCB-DLBCLs increases to ∼85% from ∼17% when these lymphomas carry MYC alterations.⁶ Most “double-hit” B-cell lymphomas carrying rearrangements of MYC and BCL2 (ie, HGBL-DH) and/or those expressing MYC-BCL2 activation signatures (eg, DHITsig-positive³⁸) show not only low infiltration of CD4⁺ T cells³⁸ and CD8⁺ T cells but also low cytolytic activity in the scant T cells found in these TMEs.¹⁴ Moreover, for both HGBL-DH and DHITsig-positive lymphomas, the presence of either a DP- or IN-LME categories (the latter containing exhausted/suppressed CD8⁺ T cells) is independently associated with a worse prognosis (than other LME categories).⁶ In this context, it is difficult to weight MYC’s specific contribution to TME features, because BCL2 rearrangements and EZH2 mutations, which a proportion of “double-hits” B-cell lymphomas also carry, can contribute to a DP-LME, as previously mentioned. Nonetheless, specific alterations appear to be directly regulated by the transcriptional activity of MYC, including secretion of immune-suppressive cytokines and metabolites^14,39 and higher expression of T cell and phagocytic immune evasion and checkpoint molecules such as PD-L1, CD47, and CD24.^40,41 CD24, which is expressed in ∼10% of DLBCL, functions as a “do not eat me” signal to avoid tumor-cell phagocytosis by tumor-associated macrophages (TAM) that recognize membrane-resident sialoglycans in tumor cells.³⁹ CD24 is a putative MYC target gene and, accordingly, CD24⁺ DLBCLs have a high proportion of MYC-rearrangements and/or MYC overexpression.⁴² In contrast, the TME of a murine B-cell lymphoma model phenocopying genetic alterations of the flippase TMEM30A showed increased TAM infiltrates, suggesting that alterations in phosphatidylserine exposure could increase signaling to phagocytic polarized TAMs.²³ 

Certain mutations act on the microenvironment by causing the release of short- and long-range acting molecules into the extracellular space. TP53 LOH and LOF mutations are associated with an immune-depleted and protumoral TAM-polarized LME,⁴³ a characteristic shared by several solid tumors and hematological malignancies carrying TP53 alterations.⁴⁴ In addition to modifications in cellular membrane proteins, one of the most important mechanisms by which decreased p53 function affects the TME is the modification of the secretome of cancer cells,⁴⁵ which ultimately weakens the infiltration and/or function of T cells and myeloid and stromal cells. This effect occurs by the amplification of the senescence associated secretory phenotype (SASP) in cancer cells.⁴⁶ SASP’s net effect, favoring a protumoral or antitumoral TME, is context dependent. Interleukin-6 (IL-6) and IL-8 are mediators of amplified SASP that can induce proliferation of lymphoma cells and reprogramming of CAFs and TAMs into protumoral entities.⁴⁵ Alternatively, only in cells with wild-type p53, SASP can induce the secretion of type-1 interferons and other cytokines that improve the immune infiltration of tumors.⁴⁵ Specific gain-of-function TP53 mutations have been reported to have protumoral effects, including the suppression of tumor infiltration by cytotoxic T and NK cells as well as the favoring of M2-like TAMs resulting in immune evasion.⁴⁷ The mechanism in this case involves the prevention, by mutant p53, of the formation of the STING/TBK1/IRF3 complex required for secretion of type 1 interferons. Reactivation of the cGAS-STING pathway by overexpressing TBK1 reverses the TME effects and improves cancer immunity.⁴⁷ 

Although the contributions of individual mutations to specific features of the TME are obscured by the complexity of the genetic constitution of most DLBCLs, some murine models offer an opportunity for clarification. For example, deletion of p53 (Trp53) or Blimp (Prdm1) in a MYD88^L265P-like mouse model generates extranodal ABC-DLBCL–like tumors harboring distinct TMEs.⁴⁸ Although Prdm1 loss leads to an abundance of T and other immune cells with activated/exhausted phenotypes, recapitulating the IN-LME category seen in patients, the Trp53 deletion leads, instead, to a low infiltration of T cells, consistent with the DP-LME category.⁴⁸ An analysis of patients with DLBCL has found that the majority of ABC-DLBCLs with MYD88^L265P showed either an IN-LME (43%) or DP-LME (44%).^6,48 

Similar to TP53 alterations, MYC overexpression in solid tumors can suppress STING-mediated interferon signaling and tumor immune infiltration.^49-51 However, the mechanisms here may include metabolic reprogramming. In MYC-overexpressing DLBCL, the increased biosynthesis of nucleotides and other metabolites required for biomass gain and cell proliferation is coupled with a higher secretion of metabolites such as adenosine, inosine, and other nucleotides, which may work as metabolic immune checkpoints.¹⁴ Adenosine is a product of the activity of endonucleotidases and/or ectonucleotidases (ie, CD39 and CD73) on adenosine triphosphate and can be released in the hypoxic tumor microenvironment.⁵² Inosine, which has a prolonged half-life, is produced via the deamination of adenosine by adenosine deaminase and is frequently found in circulation in patients with DLBCL.¹⁴ Through the direct activation of purinergic receptors on immune cells, which enhances the polarization of myeloid and T-cell subsets,⁵² this nucleoside exerts immunomodulatory effects⁵³ that can be broadly considered proangiogenic and immunosuppressive. Thus, interference with the MYC metabolic program might affect lymphoma cells not only directly but also indirectly by improving lymphoma immunity.

In patients with DLBCL, CREBBP and EP300 mutations are associated with a decreased peripheral blood leukocyte-to-monocyte ratio⁵⁴ and with a DP-LME that contains higher proportion of M2-like polarized TAMs.⁶ Murine xenograft models of CREBBP and/or EP300 mutant DLBCLs result in NOTCH pathway rewiring, higher CCL2 secretion, and TAM M2-like polarization and recruitment.⁵⁴ Similarly, MYD88^L265P causes NF-kB pathway rewiring that may lead to upregulation of membrane molecules such as CD44 and galectin-3, as well as the secretion of IL-6 and IL-10, which, in turn, may activate JAK/STAT3 in lymphoma cells as well as in the TME cells.^55,56 

Although cancer cells act on nontumor cells to reprogram them to become cancer-tolerogenic and eventually cancer cell supporting (Figures 3 and 4), the susceptibility of these nontumor cells to reprogramming and the ultimate phenotypic effect can be affected by genotypic variants and functional alterations in the genomes of these cells. For example, there are differences in TME function and composition associated with age and ethnicity.⁵⁷ Common mechanisms that may explain these TME differences include single nucleotide variants (SNVs) and functional alterations in the genome of aged cells. Because immunity and inflammation are shaped by the functions of genes that can contain SNVs, such SNVs can also affect the function and composition of the TME.^58-60 The best known SNVs associated with DLBCL are in the TNFA and IL10 genes.⁶¹ 

Although the relevant studies have varied significantly in their sample sizes, statistical powers, ethnicities and SNVs covered, and overall methodology (including their consideration of the potential role of epistasis⁶¹), recent investigations of solid tumors indicate that some differences in gene expression within the TME are associated with genetic variants that are unequally distributed among ethnicities. For example, a germ line dinucleotide polymorphism in the INFL4 gene affects the interferon response pathway in the TME of prostate cancer and associates with shorter survival in Black patients.⁶² A similar mechanism could be operating in DLBCL, potentially contributing to racial disparities in outcomes.⁶³ Of note, some differences in TME composition and function could also be attributed to the unequal distribution of specific mutations in cancer cells among ethnic cohorts. For example, in DLBCL, 6 driver genes (ie, ATM, MGA, SETD2, TET2, KMT2C, and DNMT3A) are more commonly mutated in patients with African ancestry than those with European ancestry.⁶⁴ As noted above, some of these mutations are associated with specific TME features (Figure 3).

In the multistage model of lymphomagenesis, prior accumulated genetic and/or phenotypic diversity in aged hematopoietic tissues⁶⁵ should be incorporated to evaluate the somatic coselection of lymphoma and TME cells more accurately (Figure 2). Recent studies are beginning to mechanistically elucidate the complex interrelationship between an aged local and systemic microenvironment and cancer initiation and progression.⁶⁶ Somatic mutations accumulate during the human lifespan in a wide variety of healthy tissues. In hematopoietic cells, certain mutations, all of which are also found in hematological malignancies, can drive clonal expansion causing a condition called clonal hematopoiesis of indeterminate potential (CHIP).⁶⁷ CHIP confer a modest risk of developing a hematological neoplasm, and more strongly, it associates with low-grade chronic inflammatory conditions.^67,68 Murine models prone to atherosclerosis and engineered with CHIP mutations (eg, TET2 and JAK2) have increased production of proinflammatory cytokines, such as IL-1B and IL-6, and activation of macrophages.^69,70 In DLBCL, preliminary studies suggest that patients aged >75 years present more frequently with an IN-LME category.⁷¹ The mechanisms driving this polarization in older individuals are not fully elucidated; however, the low-grade chronic inflammation associated with age-related conditions (ie, the “inflammaging” phenotype^72,73) is a common feature of the TME of several tumors.⁷⁴ How these features in the aged TME affect DLBCL prognosis and the response to immunotherapies remain to be fully investigated.⁷⁵ 

As noted, substantial biological, biochemical, and biomechanical alterations occur in the TME during cancer progression (Figure 2). These can be due to genetic changes in cancer cells and/or to changes in external variables, such as aging and treatment. Similarly, to survive and proliferate, cancer cells respond and adapt to TME imposed conditions. This results in the establishment of a cross talk between cancer and LME cells, which evolve in a process of coselection; the TME selecting for certain cancer cell phenotypes and vice versa. In this process, acquisition of mutations and other alterations (eg, epigenetic reprogramming) in cancer cells can offer “shortcuts” that contribute to the evasion of microenvironmental constraints and accelerate TME reprogramming. For example, comparison of sequential biopsy samples obtained from patients with follicular lymphoma during transformation into DLBCL has revealed the acquisition of B2M aberrations with loss of MHC-I expression decreasing immune infiltration and facilitating immune evasion.¹⁸ To cite another example, epigenetic heterogeneity promotes the acquisition of aggressive traits in lymphoma cells.⁷⁶ Aggressive DLBCLs harboring DNA hypermethylation and downregulation of MHC-I, MHC-II, and the transforming growth factor β transducer Smad1^77,78 develop a TME with low immune infiltration and low number of Tfgb-producing CAFs.⁶ At the same time, changes in the TME composition and/or functionality influence the phenotype of lymphoma cells, for example, by inducing the expression of Myc, DNA damage-response programs, and immune checkpoints.⁷⁹ This influence on lymphoma cells, in turn, modifies tumor cell–to-TME signaling, creating topologically diverse TME-tumor interfaces that sustain lymphoma intratumoral phenotypic heterogeneity.

The process of mutual selection during lymphoma progression results in a less complex TME in terms of cellular subtypes; as a consequence, relapsed DLBCLs have a phenotypically more homogenous TME, often resembling a DP-LME type.⁶ This process often imposes profound epigenetic reprogramming at the chromatin and DNA methylation levels in TME cells. For example, most tumor-reactive CD8⁺ cytotoxic T cells eventually exhibit an exhausted DNA methylation signature.^80-83 In addition, antitumor immune memory may be compromised in the TME, because the generation of memory T cells and NK cells is under epigenetic regulation.^84,85 Myeloid cells including TAMs and myeloid-derived suppressor cells can be epigenetically reprogrammed within the TME.^86,87 TAMs can have protumoral and antitumoral properties that may depend on reversible epigenetic reprogramming. Although relatively little is known about epigenetic TAM reprogramming, several epigenetic regulators have been implicated in the phenotype and activation of macrophage subsets.⁸⁸ Similarly, the development of protumoral phenotypes in stromal cells such as CAFs and endothelial cells has been reported to be epigenetically driven.⁸⁹ 

Pharmacological restoration of epigenetically repressed pathways in lymphoma and/or TME cells offers an opportunity to disengage their cross talk therapeutically.⁹⁰ This concept has been demonstrated in lymphoma models and patients by restoring the expression of epigenetically downregulated APPP components using EZH2 inhibitors, histone deacetylase inhibitors, or hypomethylating agents.^{6,17,32,33,91} Similarly, in poorly immune-infiltrated TMEs such as DP- and MS-LMEs, the activation of cGAS-STING and other nucleotide sensing pathways triggers the release of inflammatory cytokines and other mediators that improve TME immune infiltration.^6,32,33,91 This “epigenetic priming” therapeutic strategy⁹² often results in the reprogramming of most cell subtypes in the TME, curtailing their lymphoma cell-supporting pathways with a net potentiation of the antineoplastic effect. Other than epigenetic treatments, TME reprogramming can be achieved by targeting signaling or metabolic pathways that are hyperactivated in these cells such as NF-kB, JAK/STAT, mTOR, and PI3K.^6,93,94 For the MS- or IN-LME categories, therapeutic strategies targeting aberrant immunosuppressive and/or inflammatory cells could potentially shift the TME into an antilymphoma and proimmunity state.⁹⁵ This is exemplified in the results of the GUIDANCE-01 trial, in which newly diagnosed patients with DLBCL harboring an IN- or DP-LME benefited, on average, from the addition of a targeted agent (lenalidomide or ibrutinib) to R-CHOP (rituximab-cyclophosphamide, doxorubicin, vincristine, prednisone).¹¹ This study also showed that among oncogenic signaling pathways, NF-kB was upregulated in IN-LME and the BCR signature was activated in DP-LME, which was counteracted by ibrutinib treatment.¹¹ In general, any treatment, whether targeted to the TME or not, has the potential to modify the composition and function of TME. For example, anti-CD20 therapies typically kill normal CD20⁺ B cells in the TME, and a recent finding demonstrates that these cells are critical for antitumor immune responses in DLBCL.⁹⁶ Given the increased awareness of the TME’s importance in the treatment of DLBCL and its dynamic evolution, it is becoming imperative to characterize the TME when initiating a new therapeutic strategy. Moreover, this could be critical to provide an alternative option for targeted agents in patients with DLBCL-NOS (diffuse large B-cell lymphoma, not otherwise specified) lacking genetic features.¹¹ 

From the studies discussed in this review, it has become clear that oncogene activation and/or loss of tumor suppressors not only can drive intrinsic programs in DLBCL cells but also can have profound effects in shaping the TME. This TME-shaping effect can be due to the direct modification in the expression of cell-cell signaling molecules and, more frequently, to the epigenetic, metabolic, and signaling reprogramming they induce. However, the complexity of these interactions, which include not only TME cells interacting with lymphoma cells but also with other TME cells, helps explain why the genetic constitution of the tumor is often insufficient to predict the type of TME that surrounds it. In mechanistic terms, certain mutations specifically affect TME composition and/or function, but this seems more the exception than the rule. Nevertheless, identifying the mechanisms underlying the causal relationship between genetic drivers and their TME may uncover novel therapeutic targets. With increasingly sophisticated analytical and experimental research technologies, especially those that reveal spatiotemporal variations, it should be possible to uncover a multitude of genetic alterations that have specific effects on the TME. This knowledge will help to improve the precision of therapeutic approaches that target the TME as well as the lymphoma cell itself.

Funding was provided by National Institutes of Health, National Cancer Institute grant R01CA242069.

Contribution: L.C. wrote the manuscript.

Conflict-of-interest disclosure: The author declares no competing financial interests.

Correspondence: Leandro Cerchietti, Hematology and Oncology Division, Medicine Department, New York-Presbyterian Hospital, Meyer Cancer Center, Weill Cornell Medicine, Cornell University, 1300 York Ave, New York, NY 10065; email: lec2010@med.cornell.edu.