New Insights in Diffuse Large B Cell Lymphoma Pathobiology

: Diffuse large B cell lymphoma (DLBCL) is the most common non-Hodgkin lymphoma (NHL), accounting for about 40% of all cases NHL. Analysis of the tumour microenvironment is an important aspect of the assessment of the progression of DLBCL. In this review article, we have analyzed the role of different cellular components of the tumour microenvironment, including mast cells, macrophages, lymphocytes, in tumour progression of DLBCL. We examined several approaches to confront the available pieces of evidence; three key points emerged. DLBCL is a disease of malignant B-cells spreading and accumulating both at nodal and in extranodal sites. Both in patients with nodal and extranodal lesions, the subsequent induction of a cancer-friendly environment appears pivotal. DLBCL cell interaction with mature stromal cells and vessels confers tumour protection and inhibition of immune response while delivering nutrients and oxygen supply. Single cells may also reside and survive in protected niches in the nodal and extranodal sites as a source for residual disease and relapse. This review aims to molecularly and functionally recapitulate the DLBCL-milieu crosstalk, to relate niche and pathological angiogenic constitution and interaction factors to DLBCL progression. Author Contributions: Conceptualization, D.R., A.G.S., G.S. and A.V.; methodology, T.A., R.T., G.I., E.M; software, R.T., A.G.S.; validation, T.A., G.I. and A.V.; formal analysis, A.G.S., R.T. and G.I.; investigation, A.G.S.; resources, A.G.S., D.R., A.V.; data curation, D.R.; writing—original draft preparation, A.G.S., T.A., R.T., writing—review and editing, D.R., G.S. A.V.; visualization, A.G.S., R.T., G.I., E.M; supervision, D.R., G.S. and A.V.; project administration, D.R., A.G.S.; funding acquisition, D.R., A.G.S., G.S. and A.V. agreed published version the


Introduction
Diffuse large B-cell lymphoma (DLBCL) classified by the 2008 WHO classification as one of the B-cell lymphomas types is the most common non-Hodgkin B-cell lymphoma (NHL), accounting for about 40% of all cases of NHL [1]. DLBCL characteristically onsets with advanced stage, both in nodal and extranodal symptomatic disease, with a median age of 60, representing an important disease holding a practical objective of treatment represented by curative approach, while minimizing the toxicity profile [2]. Most of DLBCLs arise from germinal B cells at different stages of differentiation where recurrent genetic alterations contribute to the molecular pathogenesis of the disease [3,4]. Gene expression profiling technique allowed to identify at least two molecular subtypes of DLBCL with different prognoses [5]. The first is the lymphoma derived from normal germinal center B cells (GCB) and the second one is the lymphoma derived from the activated B cell (ABC) that arises from a post-germinal center B cell that is blocked during plasmocytic differentiation. The two subtypes have different oncogenic mechanisms [6].
Specific markers, including CD10, LM02, and BCL6 are expressed in GCB patients who have a better response to conventional chemotherapy, whereas ABC patients express lower levels of BCL6 and are refractory to chemotherapy [5,7]. ABC type showed constitutive activation of NF-kB which may be related to the presence of mutations of multiple genes regulating this pathway [8,9]. Constitutively activated STAT3 is correlated with a more advanced clinical stage and overall poor survival in DLBCL [10,11]. In ABC-DLBCL, the activation of Janus Kinases (JAKs)/STAT3 pathway correlates with autocrine production of intereukin-6 (IL-6) and IL-10, which promotes cancer progression [12,13]. STAT3 gene is a transcriptional target of BCL6 and is highly expressed and activated in ABC-DLBCL and BCL6 negative normal germinal center B-cells [12]. Moreover, STAT3 is strongly linked to tumor angiogenesis and metastasis and is related to poor prognosis in different tumors [14,15]. Activation of STAT3 contributes to hypoxia inducible factor 1 alpha (HIF-1α) and vascular endothelial growth factor (VEGF) expression in tumor cells, while VEGF in turn activates STAT3 in endothelial cells. Finally, STAT3 inhibits the expression of the anti-angiogenesis transcription factor p53 [16].
Here we show the latest findings on disease evolution of DLBCL, by providing a specific focus on the role of the new players within the cancer immune microenvironment in order to envision novel theragnostic windows.
provided by the use of any ancillary diagnostic techniques. In particular, modern histopathological diagnostics of lymphomas requires knowledge and combination of morphological, phenotypic molecular, cytogenetics and clinical profiling. This methodological approach constitutes the founding principle of the World Health Organization (WHO) and has been translated into the "blue book" "WHO Classification of Tumors of the Hematopoietic and Lymphoid Tissues" [1]. Recent progresses in understanding the immunogenetic mechanisms and genetic molecular alterations of hematopoietic and in particular lymphoid neoplasms have allowed a pathogenetic approach to the DLBCL taxonomy. Many lymphomas are considered distinct entities, characterized by immunophenotypic profiles and known genetic alterations, identifiable with laboratory techniques now widely used, with good reproducibility. DLBCL parallels the complex NHLs biological architecture, being differentiated in the GCB type and the ABC/non-GC type, by means of an immunohistochemical algorithm; a distinction that can influence the therapeutic choice [1,25,26]. Furthermore, the coexpression of MYC and BCL2 identifies a new prognostic "subset" ("double-expressor" lymphomas) [27]. Although the understanding of the mutation scenario has also been widened and deepened, the translational relevance in the clinical subset still represents an unmet medical need.
Recently, NGS studies uncovered different profile of genomic alterations to be relevant in both the GCB and non-GCB/ABC subtypes [28]. Alteration in Histone-Lysine N-Methyltransferase (EZH2) as well as the translocation of BCL2 and GNA13 mutation are fundamental molecular fingerprint described in GCB. Conversely, MYD88, CD79a, CARD11, TNFAIPA3 mutations play a pivotal role in non-GCB/ACB by activating the BCR and NFKB pathways [25]. The importance of a subdivision of diffuse large cell B lymphomas, NOS in the two groups (GCB and non-GCB/ABC), is confirmed. This distinction, with possible therapeutic consequences, can be obtained in routine diagnostics by applying an immunohistochemical algorithm based on a relatively simple and reliable antibody panel (CD10, BCL6 and IRF4/MUM1) [28]. Moreover, among the DLBCL NOS, the immunohistochemical coexpression of MYC, BCL2 deemed biologically and clinically relevant identifies the category double expressor disease, harboring an unfavorable prognostic impact [27].

Molecular pathogenesis: novel insights
Double/triple hit high grade B-cell lymphoma (HGBL-DH/TH) constitute approximately 8% of DLBCL, harboring MYC, BCL2 and/or BCL6 translocations. Most of them belong to GCB molecular subgroup and clinically, despite the generally superior prognosis of GCB-DLBCLs, patients with HGBL-DH/TH have a poor outcome [29]. Double hit lymphomas show a distinct gene expression profile when dissected by RNAseq. In 157 de novo GCB DLBCLs, including 25 HGBL-DH/TH-BCL2, were analyzed to define gene expression differences between HGBL-DH/TH-BCL2 and other GCB-DLBCLs [30]. When RNAseq was applied to RNA extracted from fresh frozen biopsy samples and 104 genes that were most significantly differentially expressed between HGBL-DH/TH-BCL2, other GCB were identified [30]. Double-hit gene signature-positive (DHITsig-pos) DLBCLs are characterized by a peculiar cell of origin and a distinct mutational landscape, after genetic features association with DHITsig status. DHITsig-pos tumors were universally positive for CD10 staining, and the majority were MUM1 (IRF4) negative. CD10+/MUM1-cases were significantly more frequent in DHITsig-pos tumors. Genes associated with the GC intermediate zone had higher expression within the DHITsis-pos tumors. These findings demonstrate that DHITsig-pos tumors are B cells transitioning from the GC dark zone to the GC light zone. Along with the expected enrichment of mutation in MYC and BCL2, mutations of genes involved in chromatin modification (e.g. CREBBP, EZH2, DDX3X, TP53 and KMT2D) were more frequently harbored by DHITsig-pos tumors [30][31][32] In the frame of this thinking, regulation of chromatin status plays a pivotal role in the correct development and differentiation of mature B-cells and has been extensively investigated with therapeutical purposes. In B-cell tumors, a plethora of mutations affect genes involved in chromatin regulation and in normal B-cell development [31][32][33]. Specifically, EP300 and CREBP are main acetylation regulators and therefore modulate gene expression, as well as histone methylators such as KMT2D, SUZ12 and EZH2 [35][36][37]. These genes are mutated in 25-30% of DLBCL cases. Notably, CREBBP and EP300 positively modulate multiple biological programs in the germinal center, through acetylation of histone and nonhistone proteins. Moreover, CREBBP and EP300 mutations contribute to lymphomagenesis by perturbing the expression of genes that are relevant to normal biology (i.e. BCL6 and p53). Inactivation of CREBBP and EP300 rarely coexist in human DLBCL, suggesting that cells require a certain amount of acetyltransferase activity [38]. Remarkably, GC B cells essentially require a minimum amount of acetyltransferase activity [39] and CREBBP mutated B cells are addicted to the residual activity of EP300, envisioning potential therapeutic windows driven by CREBBP-mutated GC B cells on EP300 [39]. Thus, double KO of CREBBP and EP300 is required to abrogate GC formation detected by BCL6 immune staining. Furthermore, CREBBP deficient cells are preferentially sensitive to inhibitors targeting HAT/BRD domains of CREBBP/EP300 [39]. In DLBCL with CREBBP genetic inactivation by mutation, pharmacologic inactivation of EP300 may lead to lymphoma cell death. Additionally, EP300 polymorphism has been uncovered to decrease the balance between acetylation and deacetylation in the tumor niche, impacting disease progression [40]. Epigenetic dysregulation can therefore represent one of driver lesions in high risk DLBCL and the restoration of physiological chromatin remodeling is an attractive target for novel therapy.

Molecular prognostic models
Efficient clinical prognostic tools have been uncovered to be relevant in driving patient management. The international literature has highlighted some molecular characteristics of the DLBCL that condition their prognosis and, in perspective, the therapy [1]. Adequate histological diagnosis must include in the report an evaluation of the parameters useful to guide the therapeutic choice in order to confirm the cell of origin, its immunophenotype, the presence of double expressors as well as the proliferation index and sometimes specific FISH characteristics addressed by BCL2, BCL6, MYC and IG-heavy/kappa/lambda (IGH/IGK/IGL) DNA probes [1,56]. The prognostic impact of the biological characteristics holds relevant translational consequences. To this end, a proper stratification included specific characteristics of investigation on the cancer cells that have been uncovered to be CD20 and/or CD79a expressing B lymphocyte [26,57]; additionally, anti-CD5 deemed important when expressed, so allowing the identification of a clinically more aggressive CD5+ DLBCL subset [57]. Moreover, while characterizing the cell of origin phenotype CD10, BCL6, MUM1 play a pivotal role, by driving the GC-type identification, differentiated by CD10 and/or bcl6 expression in >30% of DLBCL cells, while their low expression along with > 30% expression of MUM1 documentation indicate a non-GC-type [58].
MYC/BCL2 evaluation in DLBCL using immunohistochemical staining have been employed to exactly define double expression and identifying subgroup with dismal prognosis, often belonging to the non-GC-type subgroup [59]. A percentage of cells with intense MYC positivity> 70% is often associated with translocation [60].
The percentage of Ki67 positive tumor cells (clone MIB1) should also be considered. In the event of uneven distribution in the tissue, it is advisable to report a percentage value representative of the average, while signaling the uneven distribution of the positivity signal [61].
Several alternative prognostic models already exist for DLBCL. A new one has been uncovered to be significant, showing in 199 cases the relevance of the immunohistochemistry according to Hans algorithm and MYC/BCL2 evaluation. The cell of origin evaluated by Nanostring, FISH analysis evaluating BCL2, BCL6 and c-MYC and the targeted sequencing from custom platform based on univariate analysis identifying gene mutations significantly correlated to poor or favorable prognosis [62]. According to that stratification system the authors elaborated an m3D-IPI uncovering sex, age, extranodal sites, LDH, advanced stage, double hit, and mutation in KMT2D, PIM1 and MEF2B which significantly related to high risk disease in R-CHOP treated patients. Despite statistically powered validation studies are required, this novel approach performed better than traditional IPI in this patients' cohort (C-index 0.87 vs. 0.77, respectively). The increasing number of biological acquisitions, combined with clinical characteristics of patients, will allow a better treatment tailoring.
Recently, since limited data was available on comprehensive genetic signatures, Chapuy et al. proposed a novel molecular gene signature deconvoluting the DLBCL heterogeneity. While dissecting the complex genomic architecture these authors uncovered an integrated approach combining recurrent mutations, somatic copy number alterations (SCNAs) and structural variants (SVs) analyses to efficiently reveal DLBCL taxonomy and highlighted 5 genetically distinctive clusters (C1-C5) [63]. Specifically, these genetically distinct DLBCL subsets predict different outcomes, provide novel insights into lymphomagenesis and suggest certain combinations of targeted therapies [63,64]. In more details, among ABC-DLBCL, C1 subtype DLBCL deemed to be associated with favourable prognosis and is characterized by MYDnon-L265P, NOTCH2, SPEN mutations and BCL6 SVs and this phenotype might origin from marginal-zone lymphoma and from an ancestor of extrafollicular origin [63]. Conversely, C5 subtype DLBCL correlated with unfavourable clinical outcome, harbouring BCL2 gain , MYD88 L265 , CD79B mut , TBL1XR1 mut and is associated with extranodal tropism and genes overexpressed in the BCL2 overexpressing group [65,66]. Contrariwise, within the GCB-DLBCLs, C4 subtype was associated with more favourable PFS, is characterized by a mutation in NF-kB, JAK/STAT and RAS pathway components and histone genes. C3 subgroup paralleled the C5 dismal prognosis, being associated with BCL2 SV and mutations, PTEN CN loss and mutation as well as chromatin-modifying enzymes alterations. Lastly, Chapuy et al. also identified remarkable feature from a C2 subtype with a distinct clinical trajectory, being composed by bi-allelic TP53 inactivation, 9p21.23/CDNKN2A copy loss and increased genomic instability reflected by recurrent SCNAs and frequent genome doublings [63]. Next, to validate the genetic substrate in an independent dataset and develop a robust molecular classifier allowing prediction in new samples, Chapuy et al. also genetically confirmed identity associated marker genes and biology of the c1-c5 DLBCL clusters in a combined larger cohort [32,67]. This independent analysis sanctioned a parsimonious probabilistic classifier able to prospectively identify the C1-C5 DLBCL subtypes in newly diagnosed patients [67].

Tumor microenvironment and angiogenesis
Based on several compelling evidences highlighting the impact of DLBCL niche in nursing cancer cells, by promoting a favorable stromal environment, several prospective clinical studies are needed to validate the clinical utility of the stromal gene expression profile in DLBCL and dissect subtypes which would profit the most from anti-angiogenic and milieu-targeting strategies [68,69]. Nonetheless, it is well known that the presence of immune and inflammatory cells contributes to modulate tumor growth and invasion in haematological malignancies and DLBCL [70][71][72]. Analysis of the tumor microenvironment has become an important aspect in the assessment of progression of DLBCL. Different components of the microenvironment have been considered in DLBCL including mast cells and TAMs to establish several correlations between prognostic significance, stage-related tumor progression and differences in treatment outcome [73,74].
Lymphomas include more than 40 lymphoproliferative disorders, and angiogenesis plays a critical role in their progression and prognosis [75,76].
The state-of-the-art knowledge on the crucial mechanisms promoting angiogenesis and mediating immunosuppression during DLBCL development, progression [77,78], and sensitivity to drugs [26,79], needs further in-depth analysis. Solid and haematological neoplasms propagate and progress through several vicious cycles feeding into surrounding tumoral milieu [80][81][82][83], and emergent knowledge pinpoint to angiogenesis and immunosuppression as simultaneous process in response to this reciprocal loop [84,85] and to a plethora of paracrine and exogenous stimuli [86][87][88].
Lymphoproliferative disorders [89,90] and DLBCL [91] is no exception. Accordingly, strategies combining anti-angiogenic therapy and immunotherapy seem to have the potential to tip the balance of the tumor microenvironment and improve treatment response of lymphoid malignancies [21,22,92,93]. These pieces of evidence prompted an intense translational investigation aimed to targeting angiogenesis and the immune system in a coordinated fashion, based on the preclinical insights available [94,95].
Transformation from indolent B cell lymphoma to aggressive DLBCL and poor prognostic subgroups within DLBCL is associated with increased VEGF expression [99]. In aggressive subtypes of DLBCL, VEGF-A-producing CD68+ VEGFR1+ myelo-monocytic cells are closely associated to new-formed blood vessels [68]. In DLBCL, the average MVD correlates with the intensity of VEGF and VEGFR-1 and VEGFR-2 expression in tumor cells [100]. Other studies in DLBCL found no correlation between MVD and VEGF expression [101]. The transcript level of the soluble isoforms of VEGF, VEGF121 has a major impact on the prognosis of ABC-like DLBCL, and low VEGF121 expression was associated with a significantly better survival than in high level ones [91]. Moreover, 57 genes involved in immune response and T cell activation were decreased in patients with high VEGF121 expression in both ABC-like and GBC-like subtypes of DLBCL [91].
In a meta-analysis of 8 studies conducted on 670 patients, positive VEGF expression in blood circulating lymphocytes and lymph nodes correlated with shorter survival in newly diagnosed DLBCL [102]. In another study performed on 149 newly diagnosed DLBCL, high serum VEGF level was associated with poorer prognosis [103]. VEGF-A and VEGFR-1 negative patients had an improved overall survival compared to VEGF-A and VEGFR-1 positive ones [104]. Polymorphism in the VEGFR-2 gene may be associated with better survival in DLBCL patients [105].
Borges et al. [106] demonstrated an association between increased expression of pro-angiomiRs miR-126 and miR130a along with anti-angiomiR-328 and the subtype-non-GCB. Moreover, they found higher levels of the anti-angiomiR-16, R-221, and R-328 in patients with low MVD and stromal-1 signature.
More recently, Lupino et al. [107] demonstrated that the overexpression of SPHK1, one of the two isozymes responsible for the production of sphingosine-1 phosphate (SP1), a bioactive sphingolipid metabolite acting as a potent inducer of angiogenesis [108], correlates with an angiogenic transcriptional program in DLBCL.

Correlations between angiogenesis, VEGF expression and response to therapy
Immunodeficient mice engrafted with human DLBCL treated with antibodies against human or murine VEGFR-1 or VEGFR-2, showed a significant 50% reduction of tumor mass after treatment with human anti-VEGFR-1. By contrast, inhibition of murine VEGFR-1 resulted in a similar tumor reduction, but inhibition of human VEGFR-2 had no antitumor effect [109].
In patients affected by DLBCL treated with anthracycline-based chemotherapy no correlation between increased MVD and VEGF expression in tumor cells has been demonstrated. Moreover, high VEGF and VEGFR-1 expression identified a subgroup of patients affected by DLBCL with improved overall survival and progression-free survival [100]. In patients with DLBCL treated with R-CHOP, high serum level of VEGF was associated with adverse outcome, having lower values in survivors than in non survivors [110]. Additionally, high MVD determines a poor outcome in DLBCL in patients treated with R-CHOP [97]. Bevacizumab inhibits tumor growth, either alone or in combination with chemotherapy in untreated DLBCL [111].

Targeting Angiogenesis and the Immune System in DLBCL: a single center experience
Recently, we demonstrated that there is a significant increase in tryptase-positive mast cells and CD68-positive TAMs as well as a significant increase in MVD and a positive correlation in chemo-resistant non-responder when compared with chemo-sensitive responder DLBCL patients ( Figure 1) [112].  [112].
Moreover, we uncovered CD3-positive T cells to be decreased while comparing bulky (patients with bulky disease are defined by the presence of a large nodal tumor mass > 10 cm or mediastinal disease) and non-bulky groups ( Figure 2) [113], suggesting that a reduction in T cells in bulky disease patients contributes to loosen the immune control over the tumor resulting in increased cell proliferation and large tumor masses [114].
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 9 July 2020 doi:10.20944/preprints202007.0168.v1 Furthermore, by microscopic imaging, we uncovered tumor vessels in ABC samples but not GBC samples to be coated by FVIII and STAT3 positive endothelial cells [115]. Evidences from our group revealed a positive correlation not only between STAT3 expression and CD3, CD8, CD68, but also between D163 positive cells in ABC and the GBC group ( Figure 4) [116].
Additionally, in ABC group, we found also a positive correlation between CD8 and CD34 and between Ki67 and CD68/CD163 positive cells ( Figure 5).

Discussion
Overall, data generated by our group corroborated previous findings, pointing towards a higher STAT3 expression is associated with a higher CD163 and CD8 positive cells infiltration, which induces a strong angiogenic response in ABC-DLBCL as compared with GCB-DLBCL [116]. Preliminary results generated in our and other labs uncovered enhanced angiogenesis to be a strong regulator of lymphoproliferative disorders prognosis due to direct and indirect activation of cell Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 9 July 2020 doi:10.20944/preprints202007.0168.v1 survival [115][116][117]. The cell-adhesion-dependent DLBCL milieu interaction nurses DLBCL proliferation, by supporting immune-surveillance evasion [118]. Independent data provided compelling evidences that in the intimate interaction between stromal cells, the malignant clone creates a permissive immune microenvironment within the lymphoma niche which starts a vicious cycle hijacking anti-tumor activity [21, 119,120]. Mechanistically, endothelial cells, by expressing TIM-3, HB-EGF [120][121][122] and a plethora of surface and soluble factors prompt a defective immunosurveillance and in turn allow for the persistence and proliferation of lymphoid neoplastic cells [123][124][125], envisioning novel therapeutic windows [126,127]. Moreover, the initial observation that the expression level of the adhesion molecules by the malignant lymphoma cells can predict disease outcome in extranodal DLBCL [128] prompt further investigation, especially in peculiar clinical disease phenotype, namely central nervous system (CNS) involving DLBCL [128].
Remarkably, CNS spreading represents a paradigmatic extranodal localization with peculiar pathobiology involving adhesion-molecule deregulated expression [129], hyperactivation of angiogenesis fueling pathway [130] along with truncal genomic signature [131] that can contribute to drug sensitivity and resistance [132][133][134], as in other malignancies [135][136][137]. Therefore, given that the aberrant expression of adhesion molecules also on bone marrow endothelial cells of patients with lymphoid and myeloid neoplasia has been discovered to predict poor clinical outcome [138][139][140][141], it is tempting to speculate a vicious cycle in DLBCL by paralleling the neoplastic cells behavior [128] and that the described molecular signature [36,142] have more interactions among themselves than what would be expected for a random set of gene encoding proteins drawn from the genome [143].
Based on these findings and on several compelling evidences to investigating how deregulated adhesion-mediated system would contribute to more aggressive disease several attempts uncovered junctional adhesion molecule role in mediating disease aggressiveness [141, 144,145]. In line with previous results [128,146,147], preliminary data from our lab demonstrate that direct contact of environmental cells with DLBCL cells would enhance adhesion molecules levels, thus preventing both the direct and indirect cell invasiveness and epithelial-mesenchimal transition and extra-nodal dissemination (unpublished data). Even more interesting, the cell adhesion molecule junctional adhesion molecule-A (JAM-A) presents remarkable features [148]: can interact with itself if expressed on two opposing cell types. Furthermore, if JAM-A is shed by a cell, the soluble form of the JAM-A molecule can bind to cell-bound JAM-A, which in turn notably enhances its binding capacity [149][150][151]. Remarkably, consistently with Peng-Peng Xu et al. [128] JAM-A appears related to extra-nodal involvement in DLBCL being selectively expressed in those cases. The therapeutic effects of blocking angiogenesis, endothelial adhesion system, JAM-A and its cognate shedding regulator ADAM17 were mainly observed in preclinical models but not in patients and, therefore, they must be interpreted with caution [151][152][153][154][155]. In a clinical setting, adhesion system and neoangiogenesis, along with competent CD8 T cells and dendritic cells had an increased OS and time to progression [99]. Thus, it is likely that invasiveness potential along with new blood vessel formation (i.e., angiogenesis) within DLBCL environment, a recognized hallmark of disease progression, mirroring the cancer evasion from T cell immune surveillance [156].
Endothelial-progenitor-cell trafficking has been uncovered to be implicated in DLBCL progression [157,158], especially in the early disease phases [100,159]. Several clinical trials in DLBCL tested the effects of angiogenesis-targeting agents, such as bevacizumab which have been used in combination with other agents, including B-cell targeting agents [111,160]. Several clinical trials in DLBCL tested the effects of angiogenesis-targeting agents, such as bevacizumab which have been used in combination with other agents, including B-cell targeting agents [161]. Nonetheless, the lack of clinical effect in the randomized study gained by the addition of antiangiogenic approach to chemo-immunotherapy. The tumour milieu might be predictive of response to anti-angiogenesis in DLBCL, being beneficial in DLBCL with high relative expression of a set of endothelial markers and angiogenic gatekeepers (the 'stromal-2' subtype) that correlate with enhanced vasculogenesis [17,69]. Furthermore, since compelling evidences pinpoint structural abnormalities in the endothelium to impair antitumor immunity by forming barriers to immune surveillance [162] the tumor-associated endothelium is nowadays also described as a caretaker that synchronizes the entrance and egress of the immune cells within the neoplastic niche [163,164]. Therefore, while defining DLBCL also based on quantity and quality of immune cell infiltrates might provide novel rational to overcome the lack of clinical success gained by angiogenesis targeting agents so far identifying the abnormalities in the DLBCL endothelium impairing the crosstalk with adaptive immunity. Targeting these abnormalities can improve the success of immune-based therapies for different cancers as well as DLBCL by improving immune-vascular crosstalk for DLBCL, enhancing anti-lymphoma immunotherapy using antiangiogenesis [165]. Thus, further studies of antiangiogenic approaches in B-NHL and DLBCL should not be denied [161]. Indeed, while preventing secondary immunodeficiencies [166,167], this evidence provides the translational rationale to overcome the scanty effect of the anti-angiogenic approach in DLBCL obtained so far by novel angiogenesis targeting via RAS pathway inhibition, while combining immune-modulatory agents (IMiDs, i.e. lenalidomide) when appropriate [168,169]. Assuming the different angiogenic impacts on a given disease stage, it would be worth tailoring the vasculogenic manipulation in the early DLBCL with the high-risk phenotype [78]. In this frame of thinking, one critical effect of corrupted angiogenesis is represented by disease dissemination, within and outside the original niche localization, driving intra-and extra-nodal adhesion dependent manifestation in DLBCL. Finally, the judicious use of antiangiogenics to normalize tumor vasculature might represent a reprograming tumour microenvironment strategy to improve next generation immunotherapy for DLBCL.

Conclusions
Lymphomas constitute a large group of more than 40 lymphoproliferative disorders, classified on the basis of morphologic, immunologic, genetic, and clinical criteria. The importance of tumour milieu and angiogenesis in lymphoproliferative disorders has been studied in relation to their impact on the prognosis of patients, suggesting high relevance in different types of lymphomas. Literature data concerning the angiogenesis of NHL is limited compared with HL, with most of the studies performed by retrospective immunohistochemical analysis, where evidence of correlation between cellular components of the microenvironment and increased vascularity has been established. Within the different types of B cell lymphomas, angiogenesis may be prominent in aggressive rather than indolent subtypes.
Current frontline DLBCL therapy although fairly successful (70-80% remission rates with the standard R-CHOP chemotherapy regimen) is frequently followed by relapse (40% of cases within 2-3 years), with an often refractory DLBCL. The anti-angiogenic and microenvironment-directed therapy represent important tools for the treatment of human lymphomas. However, a significant number of patients are resistant, whereas those who respond have minimal benefits. Nevertheless, these new findings may point towards a potential Achilles' heel of DLBCL which, in future, might be exploited therapeutically in the relapsed/refractory setting and in the extranodal dissemination.