Main

Lymphomas are a group of hematological malignancies that are derived from lymphocytes and occur predominantly in lymph nodes or other lymphoid structures. More than 50 different types of lymphoma were described in the 2008 World Health Organization Classification of Tumors of the Hematopoietic and Lymphoid Tissues.1 Lymphomas are heterogeneous at the clinical, morphological, and molecular level, and have overlapping features. Mechanistic studies have shown that lymphomas are driven or affected by abnormal genetic alterations, disordered epigenetic regulation, aberrant pathway activation, and complex tumor–microenvironment interactions.2, 3, 4 Hence, the diagnosis and classification of different lymphomas and related entities can be challenging. In addition, the molecular heterogeneity underlying lymphoma aggressiveness and progression leads to patients who are treated similarly having variable outcomes.5, 6, 7 Although biomarkers, especially protein markers detected mainly by immunohistochemistry and flow cytometry, have been used widely and have contributed greatly to diagnosis, classification, and prognostication of lymphomas, novel clinically applicable, reliable, and reproducible biomarkers for lymphoma diagnosis and prognosis are still needed for better supervision of clinical trials.

In this review, we summarize biomarkers that are related to alterations in lymphomas at the genetic, epigenetic, and protein level as well as the tumor microenvironment. We mainly concentrate on the diagnostic and prognostic value of these biomarkers in the most common types of lymphoma.

Biomarkers of genetic alterations

Genomic aberrations and relevant dysregulated oncogenic regulatory pathways account for many malignant phenotypes in lymphomagenesis.4, 8 With the wide application of advanced technologies, the identification of genetic alterations and related biomarkers has become available.4 Microarray-based technologies like gene expression profiling and massively parallel sequencing technologies like next-generation sequencing have enabled the discovery of novel biomarkers and the exploration of underlying molecular mechanisms of lymphomagenesis;2, 9 these findings also support better diagnosis and stratification of patients who may benefit from potential therapeutic strategies based on targeting specific alterations. Figures 1 and 2 summarize recurrent genomic and molecular biomarkers involved in B-cell, T-cell, and natural killer (NK) cell lymphomas. The prognostic effects of genetic abnormalities of MYC, BCL2, BCL6, and TP53 in diffuse large B-cell lymphoma are also presented using a large cohort data set in Figure 3.10, 11, 12, 13

Figure 1
figure 1

Schematic representation of B-cell development and summary of molecular and immunophenotypic biomarkers in B-cell non-Hodgkin lymphomas and Hodgkin lymphomas. The GC is an important structure during B-cell differentiation (black arrows direct the B-cell development). Most types of B-cell lymphoma are proposed to be derived from GC or post-GC B cells (purple arrows indicate the proposed cellular origin of B-cell lymphomas). FL, BL, and GCB-DLBCL are of GC origin, whereas ABC-DLBCL, PMBL, and cHL are inferred to be post-GC origin. MCL is thought to be derived from the mantle zone. Certain molecular features are relatively specific for given type of lymphoma and have diagnostic or prognostic potential. Recurrent gain of function (red) and loss of function (blue) molecular biomarkers of common types of B-cell lymphoma and HLs are summarized. Immunohistochemical biomarkers (green) that are of diagnostic value in B-cell lymphomas are also shown. ABC-DLBCL, activated B-cell-DLBCL; BL, Burkitt lymphoma; cHL, classical Hodgkin lymphoma; FDC, follicular dendritic cell; FL, follicular lymphoma; GCB-DLBCL, GC B-cell-like diffuse large B-cell lymphoma; MCL, mantle cell lymphoma; PMBCL, primary mediastinal B-cell lymphoma; TFH, T follicular helper cell.

Figure 2
figure 2

Schematic representation of T- and NK-cell development and summary of the molecular and immunophenotypic biomarkers in the most common types of peripheral T- and NK-cell lymphoma. T-cells originate from HSC and CLP in the bone marrow and then migrate to the thymus to undergo positive or negative selection. Eventually, single-positive T-cells enter the periphery (black solid arrows direct T-cell development; black broken arrows direct some inclusive NK/NKT-cell development). Most types of T-cell lymphoma are CD4-positive (purple arrows indicate the proposed origin of T-cell lymphomas). Certain molecular features are relatively specific for a given type or subtype of T-cell lymphoma and have diagnostic or prognostic potential. Recurrent gain of function (red) and loss of function (blue) molecular biomarkers of the most common types of peripheral T- and NK-cell lymphoma are summarized. Immunohistochemical biomarkers (green) that have diagnostic value in the most common types of peripheral T- and NK-cell lymphoma are also shown. ALCL, anaplastic large cell lymphoma; AITL, angioimmunoblastic T-cell lymphoma; ALK, anaplastic lymphoma kinase; ATLL, adult T-cell leukemia/lymphoma; CLP, common lymphoid progenitors; DN, double negative; DP, double positive; ENKTCL, extranodal NK/T-cell lymphoma; HSC, hematopoietic stem cell; NK, natural killer; NKT, natural killer T-cell; PTCL-NOS, peripheral T-cell lymphoma, not otherwise specified.

Figure 3
figure 3

Prognostic effects of genetic abnormalities and overexpression of MYC, BCL2, BCL6, and TP53 in diffuse large B-cell lymphoma. (a–b) MYC or BCL2 translocations (MYC-R and BCL2-R) correlated with significant poorer OS in DLBCL. (c) BCL6 translocations (BCL6-R) did not significantly correlate with poorer OS in DLBCL. (d) Patients with TP53 mutations (MUT-TP53) had significantly poorer OS in DLBCL than patients wild-type TP53 (WT-TP53). (e) MYC translocations correlated with significant poorer PFS in DLBCL. (f–g) BCL2 or BCL6 translocations did not significantly correlate with poorer PFS in DLBCL. (h) Patients with TP53 mutations had significantly poorer PFS in DLBCL than patients wild-type TP53. (i–j) Myc or Bcl-2 overexpression (≥70%) correlated with significant poorer OS in DLBCL. (k) High level of Bcl-6 protein (>50%) did not correlate with poorer OS in DLBCL. (l) p53 overexpression (≥ 20%) correlated with significant poorer OS in DLBCL. (m–n) Myc or Bcl-2 overexpression correlated with significant poorer PFS in DLBCL. (o) High level of Bcl-6 protein correlated with significantly better PFS in DLBCL. (l) p53 overexpression correlated with significant poorer PFS in DLBCL. The clinical and pathological data in this figure have been organized from International DLBCL Consortium Program. Results have been previously reported and now are organized for presentation in different formats.10, 11, 12, 13 DLBCL, diffuse large B-cell lymphoma; OS, overall survival; PFS, progression-free survival.

B-Cell Lymphoma 6 (BCL6)

BCL6 encodes a transcriptional factor that has a key role in germinal center B-cell differentiation and in the pathogenesis of germinal center-derived lymphomas. Translocation and mutation are the most common means by which BCL6 activity is dysregulated.14, 15 Both rearrangement and somatic mutation of BCL6 can be present simultaneously in diffuse large B-cell lymphoma, but somatic mutations occur independently of BCL6 rearrangement. Studies in vitro and in animal models have shown that aberrant Bcl-6 expression results mainly from BCL6 translocation.16, 17 High expression of BCL6 mRNA and protein have been shown to be associated with better prognosis for patients with diffuse large B-cell lymphoma, concordant with immunohistochemical stain.18, 19 Some reported results are inconsistent in the literature, possibly attributable to heterogeneity within BCL6 mRNA expression and/or post-translational regulation.16

Tumor Protein 53 (TP53)

Dysfunction of the TP53 tumor suppressor gene is common in many hematological malignancies. Despite a low frequency of mutation in hematological malignancies, TP53 mutation status is an independent prognostic biomarker in patients with diffuse large B-cell lymphoma, follicular lymphoma, and mantle cell lymphoma.20, 21 In diffuse large B-cell lymphoma patients, rather than TP53 deletion and loss of heterozygosity, TP53 mutation has been shown to be associated with a poorer prognosis in patients with either germinal center B-cell-like diffuse large B-cell lymphoma or activated B-cell-like diffuse large B-cell lymphoma, especially if the mutations occur in the Loop-sheet-helix, L3, and L2 motifs of the DNA-binding domain.20 These results are true in diffuse large B-cell lymphoma patients treated with either the cyclophosphamide, doxorubicin, vincristine, prednisolone (CHOP), or rituximab plus CHOP (R-CHOP) chemotherapy regimens.10 Thus, it is necessary to account for TP53 mutation status in the design of future therapeutic strategies for diffuse large B-cell lymphoma patients.

V-Myc Avian Myelocytomatosis Viral Oncogene Homolog (MYC)

MYC is one of the most important oncogenes in cancer. As a pleiotropic transcription factor, MYC is involved in almost every process of cell biology and oncology by interacting with thousands of target genes.22, 23 In the prototypical example of a MYC translocation in cancer, Burkitt lymphoma, MYC normally located on chromosome 8q24 is juxtaposed with the IGH on the derivative chromosome 14 or with the 'IG light chains' on the derivative chromosome 8. The t(8;14)(q24;q32), t(2;8)(p12;q24), or t(8;22)(q24;q11) lead to Myc overexpression.22, 24 MYC translocations with either IG or non-IG partners also occur in diffuse large B-cell lymphoma and other aggressive B-cell lymphomas,24, 25 and usually are associated with a poorer prognosis, particularly MYC-IG translocations.26, 27

B-Cell Lymphoma 2 (BCL2)

t(14;18)(q32;q21), involving IGH and BCL2 at chromosome 18q21, has been proven to be a molecular hallmark of follicular lymphoma28 and also can be detected in germinal center B-cell-like diffuse large B-cell lymphoma.29 In contrast, BCL2 is rarely translocated but frequently amplified in activated B-cell-like diffuse large B-cell lymphoma.30, 31 Fluorescence in situ hybridization analysis has confirmed a strong correlation between BCL2 rearrangement and germinal center B-cell-like subtype.32 The prognostic impact of BCL2 alterations in diffuse large B-cell lymphoma remains controversial and may depend on subtype, Bcl-2 expression, BCL2 gene polymorphism, and the treatment regimen.31, 33, 34 A minority of diffuse large B-cell lymphoma cases associated with both MYC translocation and t(14;18)/IGH/BCL2 and/or BCL6 translocation are commonly designated as double-hit or triple-hit lymphoma and affected patients usually have an aggressive clinical course and poor prognosis.26, 27, 35

Myeloid Differentiation Primary Response 88 (MYD88)

MYD88, an adaptor protein, has a role in facilitating Toll-like and interleukin 1 receptor signaling.36 Somatic mutation of MYD88L265P is the most frequent mutation found in lymphoplasmacytic lymphoma/Waldenstrom macroglobulinemia and activated B-cell-like diffuse large B-cell lymphoma.36, 37, 38 MYD88 mutations also occur in small (<5%) subsets of marginal zone lymphoma, mantle lymphoma, and chronic lymphocytic leukemia/small lymphocytic lymphoma. MYD88 mutation has not been observed to have prognostic impact in patients with diffuse large B-cell lymphoma, but MYD88 protein expression, regardless of the MYD88 genetic status, has been shown to be significantly associated with recurrence and shorter disease-free survival.37, 39

Spi-B Transcription Factor (SPIB)

SPIB, which encodes an ETS family transcription factor on chromosome 19, is expressed at substantially elevated levels and can be detected in 26% of activated B-cell-like diffuse large B-cell lymphoma, but only in 3% of germinal center B-cell-like diffuse large B-cell lymphoma and primary mediastinal B-cell lymphoma. Knockdown of SPIB in cell lines also indicates that SPIB is an oncogene dysregulated by chromosomal aberrations in activated B-cell-like diffuse large B-cell lymphoma.30

Tumor Necrosis Factor α-Induced Protein 3 (TNFAIP3)

TNFAIP3, which encodes the NF-κB inhibitor A20, is commonly mutated, deleted, or silenced epigenetically in activated B-cell-like diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma, mantle cell lymphoma, and classical Hodgkin lymphoma.40, 41, 42 Inactivation of TNFAIP3 has a role in dysregulation of the NF-κB pathway.

Major Histocompatibility Complex Class II Transactivator (CIITA)

CIITA, located at chromosome 16p13.13, is a non-DNA-binding coactivator of the major histocompatibility complex (MHC) class II promoter. CIITA is altered by chromosomal translocation in 38% of primary mediastinal B-cell lymphoma cases, which is associated with reduced MHC class II expression. Translocations involving CIITA are uncommon in de novo diffuse large B-cell lymphoma, and seen in <5% of cases.43, 44 Deletion or mutation of CIITA is also uncommon in diffuse large B-cell lymphoma.45, 46 Therefore, CIITA has the potential to act as a biomarker to distinguish primary mediastinal B-cell lymphoma from diffuse large B-cell lymphoma. However, CIITA is also translocated in 15–20% of cases of classical Hodgkin lymphoma.43

Other Genetic Biomarkers or Simplified Gene Expression Models in B-Cell Lymphomas

Several investigators have attempted to simplify analysis of gene expression profiling using smaller gene expression models. In particular, attempts to quantitate gene expression for activated B-cell-like/germinal center B-cell-like molecular sub-classification using formalin-fixed and paraffin-embedded tissue have become clinically practicable.47, 48, 49

Lossos et al.50 identified a six-gene model (LMO2, BCL6, FN1, CCND2, SCYA3, and BCL2) that could predict survival in CHOP-treated diffuse large B-cell lymphoma patients. Another study combined LIM domain only two (LMO2) expressed by tumor cells and tumor necrosis factor receptor superfamily member 9 (TNFRS9) expressed by microenvironmental cells to predict overall survival in patients with diffuse large B-cell lymphoma.51

Love et al.52 identified several recurrently mutated genes in Burkitt lymphoma including ID3, GNA13, RET, PIK3R1, ARID1A, and SMARCA4 as well as MYC. They found that ID3 is mutated in Burkitt lymphoma, but not in most cases of diffuse large B-cell lymphoma, implicating ID3 as a new tumor suppressor gene that has lost function via genetic alteration(s) and could be a new potential therapeutic target in Burkitt lymphoma. In addition to ID3 and MYC, TPST2 and RET mutations were also found predominantly in Burkitt lymphoma, whereas PIM1, CECR1 and MYD88 mutations occurred predominantly in diffuse large B-cell lymphoma. MLL3, TP53, and LAMA3 showed overlapping patterns of mutation in Burkitt lymphoma and diffuse large B-cell lymphoma.

The chromosomal translocation t(11:14)(q13;q32), involving CCND1 at 11q13, leads to overexpression of cyclin D1 and is the molecular hallmark of mantle cell lymphoma.53 Hartmann et al.54 established a gene expression-based predictor of survival in mantle cell lymphoma patients using a quantitative PCR with reverse transcription-based assay on formalin-fixed, paraffin-embedded tissue specimens. This model identified five genes including RAN, MYC, TNFRSF10B, POLE2, and SLC29A2. Elevated expression of TNFRSF10B was associated with a favorable outcome; in contrast, increased expression of RAN, MYC, POLE2, and SLC29A2 correlated with a poorer prognosis.

Pastore et al.55 integrated the mutation status of seven-genes (EZH2, ARID1A, MEF2B, EP300, FOXO1, CREBBP, and CARD11) with the follicular lymphoma international prognostic index (FLIPI) and established a clinicogenetic-risk stratification model for follicular lymphoma patients (known as the m7-FLIPI). This model exhibits a reasonable ability to recognize a high-risk group (28%) of patients with follicular lymphoma with 5-year failure-free survival, but needs to be further validated in an independent study.

The diagnosis of primary mediastinal B-cell lymphoma can be challenging because there is morphological and immunophenotypic overlap with classical Hodgkin lymphoma.56, 57 Specific characteristics of primary mediastinal B-cell lymphoma include STAT6 constitutive activation and JAK2 amplification.58, 59 In contrast, NF-κB1A mutation is rather specific for classical Hodgkin lymphoma, suggesting that NF-κB activation in classical Hodgkin lymphoma and primary mediastinal B-cell lymphoma may result from different gene mutations and other mechanisms.59, 60

Other Genetic Biomarkers or Gene Expression Models for Differential Diagnosis in Peripheral T-Cell Lymphoma

Approximately 30–50% cases of peripheral T-cell lymphoma are categorized as peripheral T-cell lymphoma, not otherwise specified, because of a lack of specific morphological features and key biomarkers. However, studies using gene expression profiling have shown that a subset of cases of peripheral T-cell lymphoma, not otherwise specified has profiles more consistent with angioimmunoblastic T-cell lymphoma, adult T-cell leukemia/lymphoma, anaplastic large T-cell lymphoma or extra-nodal NK/T-cell lymphoma.6, 61 Moreover, cases of peripheral T-cell lymphoma, not otherwise specified without a distinctive profile can be classified into two subsets based on distinct oncogenic pathways and associated with different outcomes. One subgroup, characterized by high expression of GATA3 and its known target genes (CCR4, IL18RA, CXCR7, and IK) is associated with poorer overall survival. Another subgroup, characterized by high expression of cytotoxic gene signature such as TBX21, EOMES and their target genes (CXCR3, IL2RB, CCL3, and IFNγ) is associated with a more favorable clinical outcome.6

Chromosomal translocations also occur in some types of T-cell lymphoma. In anaplastic lymphoma kinase (ALK)-positive anaplastic large T-cell lymphoma, a characteristic chromosomal translocation, t(2;5)(p23;q35) involving NPM/ALK, or other ALK translocations occur. As a consequence, ALK is aberrantly expressed by the lymphoma cells.62 In ALK anaplastic large T-cell lymphoma, ~20% of cases carry t(6;7)(p25.3;q32.3) involving DUSSP22 at 6p25.3 which was discovered by massively parallel genomic (next-generation) sequencing.63 DUSP22 is located within 40 kb of MUM1/IRF4 and is a dual-specificity phosphatase that physiologically has a role in inhibiting T-cell receptor signaling. Translocations involving DUSP22 result in reduced DUSP22 expression; these data suggest that translocation disrupts the tumor suppressor gene functions of DUSP22. A second subset of ~10% of cases of ALK- anaplastic large T-cell lymphoma carries rearrangements of TP63, a homolog of TP53, located at chromosome on 3q28. Fluorescence in situ hybridization may be utilized to identify these chromosomal translocations. The remaining cases of ALK anaplastic large T-cell lymphoma have neither DUSP22 nor TP63 translocations.64

Despite the overlap of morphology and immunophenotype, ALK+ and ALK anaplastic large cell lymphoma are associated with different patient outcomes. In general, ALK+ anaplastic large cell lymphoma patients have a better outcome than patients with ALK anaplastic large cell lymphoma as a single entity, but patients with DUSSP22/IRF4 locus abnormalities also have a relatively good outcome.64 Therefore, these translocations might be used as biomarkers not only in diagnosis of ALK+ anaplastic large cell lymphoma and ALK anaplastic large cell lymphoma, but also as molecular targets for therapeutic inhibitors. ALK-targeting drugs used to treat non-small cell lung cancer with EML4/ALK fusion may also benefit patients with ALK+ anaplastic large cell lymphoma.65, 66

Assessment of T-cell receptor rearrangement is necessary for establishing the diagnosis of many cases of T-cell lymphoma, most often accomplished by using PCR assays to demonstrate T-cell monoclonality. NK cell lymphoma, however, is an exception because NK cells do not rearrange their T-cell receptor genes. Killer cell immunoglobulin-like receptor (KIR), as one of NK cell receptors, can be used as a surrogate biomarker of clonal cellular expansion to distinguish reactive NK cell proliferations from neoplastic disease.67, 68 Restricted KIR expression can be immunohistochemically detected in extranodal NK/T-cell lymphoma.69

The t(5,9)(q33;q22) involving ITK and SYK has been found to be common in the follicular variant of peripheral T-cell lymphoma, not otherwise specified which has characteristic expression of T follicular helper cell markers similar to angioimmunoblastic T-cell lymphoma.70, 71 This translocation may serve as a diagnostic biomarker for the follicular variant of peripheral T-cell lymphoma, not otherwise specified.

Overall, genetic alterations that are relatively specific for a given type of lymphoma may be a defining feature or have prognostic implications, and are of great value in the evaluation of lymphomas.

Biomarkers related to epigenetic alterations

Epigenetic alterations are usually defined as heritable changes independent of alterations in DNA sequence.72 In numerous cellular processes of lymphomas, epigenetic aberrations have been shown to alter genes involved in signal transduction, DNA repair, cell cycle regulation, differentiation, invasion, and apoptosis.73, 74 With components of the epigenome being the focus of extensive studies, and epigenetic alterations involved in lymphomagenesis being understood in various types of lymphomas, altered patterns of DNA methylation, histone modification, and noncoding RNA expression are promising potential biomarkers for lymphoma diagnosis, assessment of prognosis, and prediction of response to therapy (Table 1).

Table 1 Biomarkers related to epigenetic alterations in lymphomas

P16/INK4

One of the most frequent molecular alterations of P16/INK4A is homozygous deletion.75, 76 However, methylation also has a major role in P16/INK4A inactivation in lymphomas and has been associated with a poorer prognosis in some studies.77, 78, 79 To date, the prognostic importance of P16/INK4A promoter methylation in lymphomas remains inconsistent and needs to be validated.78, 79

DNA Repair Enzyme O(6)-Methylguanine DNA Methyltransferase (MGMT)

MGMT, although rarely deleted or mutated, frequently loses its function as a result of epigenetic alteration. MGMT promoter hypermethylation has been found to correlate with a favorable prognosis in diffuse large B-cell lymphoma patients treated with multidrug regimens including cyclophosphamide.80

Kruppel-Like Factor 4 (KLF4)

KLF4 acts as a tumor suppressor gene in T- and B-cell lymphomas and is aberrantly hypermethylated in many types of lymphoma including follicular lymphoma, diffuse large B-cell lymphoma, Burkitt lymphoma and Hodgkin lymphoma.79, 81 These findings suggest a subtype-independent mechanism of lymphomagenesis. Epigenetic silencing of KLF4 may be of potential benefit to patients with B-cell lymphomas and particularly classical Hodgkin lymphoma.81

Myeloid/Lymphoid or Mixed-Lineage Leukemia 2 (MLL2)

MLL2 (also known as KMT2D) encodes human H3K4-specific histone methyltransferase and is inactivated by mutations in about 90% of follicular lymphoma and 32% of diffuse large B-cell lymphoma; there is no difference between germinal center B-cell-like diffuse large B-cell lymphoma and activated B-cell-like diffuse large B-cell lymphoma.79, 82

Myocyte Enhancer Factor 2 (MEF2)

MEF2, a calcium-regulating gene, encodes transcription factors and recruits histone-modifying enzymes.79, 82 MEF2B mutations have been linked to lymphoma. These mutations are somatic and are detected mostly in germinal center B-cell-like diffuse large B-cell lymphoma and follicular lymphoma, suggesting that MEF2B mutations have a role in enhancing the malignant transformation of germinal center B cells to lymphoma.82 In addition, one study showed somatic mutations of MEF2B in a subset of diffuse large B-cell lymphoma with consequential dysregulation of BCL6 expression, suggesting that inhibition of MEF2 could be an alternative to inhibit BCL6 activity for patients with diffuse large B-cell lymphoma and MEF2B mutation.79, 83

Enhancer of Zeste 2 Polycomb Repressive Complex 2 (EZH2)

EZH2 is one of the most commonly mutated epigenetic modifiers and is mutated in 27% of follicular lymphoma84 and 6-14% of diffuse large B-cell lymphoma.85, 86 EZH2Y641 mutations can enhance H3K27 trimethylation activity of the polycomb repressive complex 2 in diffuse large B-cell lymphoma and follicular lymphoma, mainly occurring in germinal center B-cell-like diffuse large B-cell lymphomas. EZH2 mutations have not been observed to have prognostic impact, but protein expression may be a better prognostic indicator of overall survival. The combination of an EZH2 inhibitor and demethylating agents might be a useful therapeutic option for patients with an EZH2 hypermethylation phenotype.87, 88

Other Epigenetic Gene Signatures

Other aberrant epigenetic gene signatures have been identified in several studies. For example, five hypermethylated genes (SOX9, HOXA9, AHR, NR2F2, and ROBO1) were found to correlate with higher proliferation, increased chromosomal abnormalities, and poorer survival in patients with mantle cell lymphoma.89

Using quantitative real-time methylation assays, Bethge et al90 analyzed the methylation status of a colorectal cancer biomarker panel (CNRIP1, FBN1, INA, MAL, SNCA, and SPG20) and showed that methylation of SNCA and SPG20 separate lymphoma from healthy control samples with high sensitivity (98%) and specificity (100%), suggesting that SNCA and SPG20 methylation status could be suitable for early detection and monitoring of patients with non-Hodgkin lymphomas. Meanwhile, promoter methylation of CNRIP1 was found to be associated with poorer overall survival in patients with diffuse large B-cell lymphoma and potentially could be used as a prognostic factor. Another concise methylation signature set that includes TNF alpha pathway biomarkers and downstream biomarkers was shown to be helpful in distinguishing germinal center B-cell-like diffuse large B-cell lymphoma from activated B-cell-like diffuse large B-cell lymphoma.91 Hypermethylation of SMAD family member 1 (SMAD1) has been reported to predict chemotherapy resistance in patients with diffuse large B-cell lymphoma,92, 93 and functional assays have confirmed that chemotherapy responsiveness can be induced through SMAD1 expression in chemotherapy-resistant diffuse large B-cell lymphoma cells.92

Using a combination of whole-exome sequencing, RNA sequencing analysis and targeted deep sequencing, Palomero et al94 identified frequently mutated genes involved in epigenetic alterations in peripheral T-cell lymphoma, including ten-eleven translocation 2 (TET2), DNA methyltransferase 3A (DNMT3A), isocitrate dehydrogenase 2 (IDH2), and RHOA. RHOA is mutated in up to 67% of angioimmunoblastic T-cell lymphoma and 18% of peripheral T-cell lymphoma, not otherwise specified. TET2 mutations occur in both angioimmunoblastic T-cell lymphoma and peripheral T-cell lymphoma, not otherwise specified with similar frequency and are associated with advanced-stage disease, thrombocytopenia, high International Prognostic Index scores, and a shorter progression-free survival. IDHR172 mutation occurs mainly in angioimmunoblastic T-cell lymphoma and has potential as a specific biomarker of this disease.94, 95, 96 IDH2 mutations in angioimmunoblastic T-cell lymphoma have not been associated with overall survival.97

Biomarkers related to epigenetic regulation have great potential for use in the diagnosis of lymphomas and for prognostication. However, the relationship between mutations and epigenetic status of these genes, and the value of these alterations for diagnosis and prognosis in lymphomas needs to be further explored.

MicroRNA and lncRNA as Biomarkers for Diagnosis and Prognosis in Lymphomas

MicroRNAs, a class of small noncoding RNA, regulate numerous biological processes at the post-transcriptional level by targeting a broad set of messenger RNAs and affecting a multitude of cellular transcripts and pathways.98, 99 Long noncoding RNAs (lncRNAs) have specific tissue expression patterns and nuclear location, and exhibit an important role in regulating gene expression at different levels and in malignant transformation, mainly involving regulation of epigenetic processes.100, 101, 102 Growing evidence supports a role for dysregulation of expression of microRNAs and lncRNAs in the development and progression of lymphomas and the rationale that microRNAs and lncRNAs may serve as promising biomarkers for diagnosis and prognosis in lymphoma patients (Table 2).5, 98, 101, 103

Table 2 MicroRNAs and LncRNAs involved in lymphomas

miR-155

miR-155 is one of the best recognized miRNA in lymphomas, particularly in diffuse large B-cell lymphoma. miR-155 acts as an onco-miR in the pathogenesis and aggressiveness of diffuse large B-cell lymphoma.104 Levels of miR-155 in activated B-cell-like diffuse large B-cell lymphoma were found to be significantly higher than that in germinal center B-cell-like diffuse large B-cell lymphoma, suggesting that miR-155 is diagnostically useful to distinguish activated B-cell-like diffuse large B-cell lymphomas from germinal center B-cell-derived tumors and may explain the poor prognosis of activated B-cell-like diffuse large B-cell lymphoma patients.105 In transgenic mice models, the critical role of constitutive expression of miR-155 in activated B-cell-like diffuse large B-cell lymphoma tumorigenesis has been shown.106 Furthermore, a correlation between miR-155 expression and resistance to R-CHOP therapy in diffuse large B-cell lymphoma patients has been observed.5

miR-17-92 Cluster

The miR-17-92 cluster, including six microRNAs (miR-17, miR-18a, miR-19a, miR-20a, miR-19b1, and miR-92-1), is encoded by locus 13q31.3 and is overexpressed as a consequence of 13q31.3 locus amplification107 in several types of lymphoma, including germinal center B-cell-like diffuse large B-cell lymphoma, follicular lymphoma, mantle cell lymphoma,4, 30, 108 anaplastic large cell lymphoma,109 and Burkitt lymphoma,110 but never in activated B-cell-like diffuse large B-cell lymphoma.4 Additionally, the miR-17-92 cluster is expressed more highly in ALK+ anaplastic large cell lymphoma compared with ALK- anaplastic large cell lymphoma,109 supporting a role for miR-17-92 in sustaining the oncogenic characteristics of STAT3 through the ALK-STAT3-microRNA-17-92 pathway in ALK+ anaplastic large cell lymphoma.111 Overexpression of the miR-17-92 cluster correlates with a poorer overall survival in patients with mantle cell lymphoma.108

miR-15a/16-1

Deregulation of miR-15a/16-1, resulting from deletion of chromosome 13q14, is implicated in the pathogenesis of chronic lymphocytic leukemia/small lymphocytic lymphoma, a disease that more typically presents with leukeumic disease rather than lymphomatous disease.112 Deletion of 13q14 and consequent low expression of miR-15a/16-1 have been associated with favorable prognosis for patients with chronic lymphocytic leukemia/small lymphocytic lymphoma, whereas other chromosomal abnormalities, such as 17p deletion, 12q trisomy, and 11q deletion, correlate with a poorer prognosis.113

Other MicroRNAs and MicroRNA-Based Classifiers

High expression of miR-18b and downregulation of the miR-29 family are independent predictors of poor outcome in patients with mantle cell lymphoma.114, 115 Notably, a combination of miR-127-3p with Ki67 can provide a novel prognostic model for mantle cell lymphoma. Similarly, the combination of miR-615-3p and the Mantle Cell Lymphoma International Prognostic Index has prognostic value.116

Using a Bayesian algorithm, Iqbal et al5 developed a 27-miRNA classifier that can distinguish Burkitt lymphoma from diffuse large B-cell lymphoma and an 8-miRNA classifier that separates diffuse large B-cell lymphoma subgroups. The activated B-cell-like diffuse large B-cell lymphoma signature includes miR-155 and miR-542-3p. miR-155 has been known for its association with activated B-cell-like diffuse large B-cell lymphoma as mentioned above; this is the first report of miR-542 in activated B-cell-like diffuse large B-cell lymphoma. Six miRNAs (miR-28-3p, miR-28-5p, miR-129-3p, miR-589, miR-331-5p, and miR-597) are upregulated in germinal center B-cell-like diffuse large B-cell lymphoma. This miRNA classification of diffuse large B-cell lymphoma subgroups has been shown to be 90% concordant with gene expression profiling-based classification and is reproducible in formalin-fixed and paraffin-embedded tissues.5

A set of miRNAs (miR-222, miR-181a, miR-129-5p, and miR-18a) has been shown prognostic value for diffuse large B-cell lymphoma patients,117 whereas another miRNA set was peculiar to hepatitis C virus-associated diffuse large B-cell lymphoma.118 In this hepatitis C virus-associated subgroup, miR-138-5p was associated with longer overall survival and miR-147a, miR-147b and miR-511-5p were associated with shorter overall survival. miR-222 expression has been shown to be consistently associated with an inferior overall survival in lymphoma patients.5

DNA methylation of the miRNA gene promoter has been implicated in lymphomagenesis. Asmar et al119 reported that a large number of miR-34s are downregulated by promoter hypermethylation in diffuse large B-cell lymphoma. Intriguingly, they identified a new TP53/MIR34A 'double hit' diffuse large B-cell lymphoma subgroup, which has concomitant MIR34A methylation and TP53 mutation; this subgroup had significantly poorer overall survival, irrespective of treatment after rituximab. Thus, MIR34A methylation combined with TP53 mutation can be used to identify an aggressive subgroup of diffuse large B-cell lymphoma which may benefit from epigenetic therapy.

Besides tumor biopsy specimens, microRNA can be assessed in serum, plasma and bone marrow smears.120, 121 A five-miRNA serum signature has been shown to be predictive of prognosis for diffuse large B-cell lymphoma patients treated with R-CHOP.122 Tissue miRNAs seem to be more reliable in the detection and staging of cancer, whereas circulating miRNAs show higher stability and can be measured non-invasively.123

In peripheral T-cell lymphomas, some unique diagnostic miRNA classifiers have been constructed. One 7-miRNA signature (miR-512-3p, miR-886-5p, miR-886-3p, miR-708, miR-135b, miR-146a, and miR-155) is mainly implicated in ALK+ anaplastic large cell lymphoma. miR-155 is present at higher levels in ALK versus ALK+ anaplastic large cell lymphoma.124 Another 11-miRNA signature (miR-210, miR-197, miR-191, miR-512-3p, miR-451, miR-146a, miR-22, miR-455-3p, miR-455-5p, miR-143, and miR-494) can distinguish ALK anaplastic large cell lymphoma from other peripheral T-cell lymphomas.125

In summary, some miRNA-based diagnostic and prognostic classifiers have been developed in various types of lymphomas and are expected to improve clinical decision making. Studies on lncRNAs in lymphomas remain rare. With further understanding of their biology and function, lncRNA profiling may also contribute to the diagnosis and prognostication of lymphomas.

Immunophenotypic biomarkers

The basic diagnosis, differential diagnosis and classification of lymphomas are commonly based on histopathological features and the results of immunohistochemistry or flow cytometry immunophenotypic analysis. A number of immunophenotypic biomarkers are routinely used, including T-cell markers (CD2, CD3, CD4, CD5, CD7, CD8, and T-cell receptors), B-cell markers (CD19, CD20, CD22, CD79, and Pax-5), cytotoxic markers (such as TIA-1, granzyme B, and perforin), germinal center markers (CD10, Bcl-6, and LMO2), and plasma cell markers (such as CD38, CD138). In addition, certain markers are highly associated with specific diseases such as CD15, CD30 in classical Hodgkin lymphoma; ALK-1, CD30 in anaplastic large cell lymphoma; cyclin D1 and SOX-11 in mantle cell lymphoma; and follicular T helper cell markers including CXCL13 and PD-1 in angioimmunoblastic T-cell lymphoma. Immunophenotypic markers commonly used in lymphoma diagnosis and classification are shown in Figures 1 and 2. Many protein biomarkers also have been shown to predict lymphoma patient outcomes or are useful in guiding personalized therapy. Figure 3 shows the prognostic value of overexpression of Myc, Bcl-2, Bcl-6, and p53 in diffuse large B-cell lymphoma based on a well-organized large cohort of patients.10, 11, 12, 13

CD30

CD30, a member of the tumor necrosis factor receptor superfamily, is a transmembrane cell-surface marker expressed by activated B or T cells in normal tissues and is highly expressed by tumor cells in classical Hodgkin lymphoma and anaplastic large cell lymphoma, and a subset of diffuse large B-cell lymphoma and Epstein-Barr virus-driven lymphoproliferative disorders.126 Immunohistochemical assessment for CD30 is used commonly as a valuable biomarker for classical Hodgkin lymphoma and anaplastic large cell lymphoma diagnosis; soluble CD30 in serum and/or body fluids can independently predict disease progression and poor outcomes of patients with CD30+ lymphoma.126 As a result of the physiological highly restricted distribution of CD30 expression and dysregulated signaling pathway in lymphoma subtypes, brentuximab vedotin, an antibody drug conjugate composed of anti-CD30 linked to monomethyl auristain E, has been shown to be an effective therapeutic drug for patients with CD30+ lymphomas, especially for patients with relapsed/ refractory Hodgkin lymphoma, systemic anaplastic large cell lymphoma, as well as a subset of patients with refractory/resistant CD30+ diffuse large B-cell lymphoma.126

Bcl-6, Bcl-2, and Myc

Bcl-6 has a critical regulatory role in the programming of germinal center B cells and is considered a unique marker for B cells at the germinal center stage of differentiation. Anti-bcl-6 antibody is often used to mark the germinal center B cells in lymphoma diagnosis, but is not specific for germinal center B-cell-like diffuse large B-cell lymphoma because it is also expressed by post-germinal center cells.17, 127 Bcl-6 is also expressed by the neoplastic T-cells of angioimmunoblastic T-cell lymphoma which are derived from normal follicular T helper cells and can be used as a marker of angioimmunoblastic T-cell lymphomas.16 In addition, Bcl-6 expression together with a simple karyotype can be used as a marker of better survival in patients with IG-MYC positive high-grade B-cell lymphomas.128

Myc and Bcl-2 overexpression are attributable to chromosomal translocations such as t(8;14)(q24;q32) and t(14;18)(q32;q21) and are of specific diagnostic value for Burkitt lymphoma and follicular lymphoma, respectively.22, 28 Overexpression of Myc and/or Bcl-2, however, is often identified in lymphomas with the absence of these translocations. Immunohistochemically, data are inconsistent regarding any prognostic value for Bcl-2, Myc, or Bcl-6 expression as a single marker in lymphomas. Other studies, however, have shown that coexpression of Myc/Bcl-2 in diffuse large B-cell lymphoma is associated with aggressive clinical features and inferior outcomes, independent of those cases with cytogenetically proven (traditional) MYC and BCL2 double-hit diffuse large B-cell lymphoma.129 Myc/Bcl-2 coexpression occurs more frequently in activated B-cell-like diffuse large B-cell lymphoma and may explain the poorer prognosis of activated B-cell-like diffuse large B-cell lymphoma,129 whereas Myc/Bcl-6 coexpression occurs in both germinal center B-cell-like diffuse large B-cell lymphoma and activated B-cell-like diffuse large B-cell lymphoma and does not correlate with outcome. Similarly, Myc expression in mantle cell lymphoma is often associated with aggressive histological variants and poorer prognosis.130

Cyclin D1 and SOX-11

Cyclin D1 is overexpressed as a consequence of the t(11;14)(q13;q32) and is a diagnostic hallmark of mantle cell lymphoma.131 Rare cases of mantle cell lymphoma are reported that are negative for cyclin D1, but have a gene expression prolife consistent with mantle cell lymphoma and these tumors can overexpress cyclin D2 or D3. A subset of these cases has a cyclin D2 translocation.132 In addition, SOX-11 is a marker of mantle cell lymphoma independent of cyclin D1.21, 133 Therefore, the combination of cyclin D1 and SOX-11 has utility for the diagnosis of mantle cell lymphoma, and SOX-11 is especially helpful for cyclin D1-negative cases.134 SOX-11 expression in mantle cell lymphoma patients treated with intensive chemotherapy is associated with improved survival. This observation may be explained by the fact that SOX11-target genes may affect transcriptional regulation of WNT and other biological pathways.135

Cyclin D1 is also overexpressed in about 3% of diffuse large B-cell lymphoma cases, which usually have a post-germinal center or activated B-cell phenotype, negative for CD10 and positive for Bcl-6 and multiple myeloma oncogene 1(MUM-1/IRF4); expression of cyclin D1 in diffuse large B-cell lymphoma is not a consequence of t(11;14) as the translocation is absent.131, 136

p53 and MDM2

Whether expression of p53, usually assessed by IHC, can predict survival in lymphoma patients remains inconsistent. In diffuse large B-cell lymphoma patients treated with R-CHOP, p53 overexpression predicted a worse survival and could be used to stratify patients into different prognostic groups.10 Interestingly, in diffuse large B-cell lymphoma patients, MDM2 expression was shown to not be an independent predictor of patient outcome, but MDM2 expression combined with mutated-type p53 expression could predict significantly poorer survival.137

p63 and Cancerous Inhibitor of Protein Phosphatase 2A (CIP2A)

TP73L, also known as p63, is expressed in all cases of primary mediastinal B-cell lymphoma, but not in classical Hodgkin lymphoma, suggesting that TP73L is useful for distinguishing primary mediastinal B-cell lymphoma from mediastinal classical Hodgkin lymphoma.138 CIP2A, a human oncoprotein, is expressed extensively in several cancers including diffuse large B-cell lymphoma. The extent of CIP2A staining is associated with the proliferation rate and the stage of diffuse large B-cell lymphoma, and may be the result of high tumor load during the disease development.139

Immunophenotypic Algorithms

Gene expression profiling can divide diffuse large B-cell lymphoma cases into at least two distinct molecular subtypes: germinal center B-cell-like subtype and activated B-cell-like subtype.140 As gene expression profiling is not routinely available in clinical practice, a number of investigators have developed immunophenotypic algorithms as surrogates for gene expression profiling (Figure 4). Bcl-2, Bcl-6, CD10, MUM-1/IRF4, forkhead box protein 1 (FOXP1), and GC expressed transcript 1 (GCET1) have been integrated into such immunohistochemical algorithms to subclassify diffuse large B-cell lymphoma basing on cell of origin.141, 142, 143, 144 Among these algorithms, the Hans algorithm is most often used and was established on a CHOP-treated cohort of patients.142 The concordance between gene expression profiling and immunohistochemical algorithms, however, is variable. The Hans algorithm matches gene expression profiling-defined subtypes in about 80% of cases. Other algorithms, such as Choi and Visco/Young algorithms, show a higher concordance (~90%) with gene expression profiling results.141, 142, 143, 144 Despite these limitations, these immunohistochemical algorithms are used because they are readily available, simple and of low cost.

Figure 4
figure 4

Molecular classification of DLBCL by immunophenotypic algorithms (a–f). Gene expression profiling has divided DLBCL into at least two subtypes: GCB and ABC. In an attempt to drive this concept in routine practice, Bcl-2, Bcl-6, CD10, MUM-1, FOXP1, and GCET1 have been integrated into immunohistochemical algorithms for DLBCL diagnostic workup, based on the expression of specific biomarkers at sequential stages of differentiation of mature B-cell through the GC (ae). ABC, activated B-cell; DLBCL, diffuse large B-cell lymphoma; FOXP1, forkhead box protein 1; GCB, germinal center B-cell; GCET1, germinal center expressed transcript1; GEP, Gene expression profiling; IHC, immunohistochemistry; MUM-1, multiple myeloma oncogene 1 (also known as interferon regulatory factor 4).

Besides serving a role in these classifiers, some of these biomarkers also have independent prognostic value. The expression of LMO2 and GCET1 has been shown to correlate with better survival in patients with diffuse large B-cell lymphoma. By contrast, expression of MUM-1/IFR4 and FOXP1, post-germinal center markers, are markers of poorer prognosis.142, 145 Interestingly, FOXP1 was shown to be a poor prognostic factor in the pre-rituximab era,146 but not in patients treated with rituximab.147

Immunophenotypic Biomarkers Involved in Oncogenic Signaling Pathways

Expression of some components of oncogenic signaling pathways also has been shown to have prognostic significance. Expression of cyclin D2, cyclin D3, protein kinase c-B, caspase-9, and survivin correlate with a worse prognosis in diffuse large B-cell lymphoma.148, 149, 150, 151 In particular, survivin was implicated in the inferior outcome of activated B-cell-like diffuse large B-cell lymphoma patients treated with R-CHOP.152 Constitutive activation of the NF-κB pathway (Figure 5) contributes to lymphomagenesis in several lymphoma types including diffuse large B-cell lymphoma, classical Hodgkin lymphoma, primary mediastinal B-cell lymphoma, and anaplastic large cell lymphoma.153, 154 Expression of components of the NF-κB pathway, such as p50 (NFκB1) and p65 (RelA), has been associated with activated B-cell-like diffuse large B-cell lymphoma and a worse prognosis,40, 155 whereas p52/RelB may identify a subgroup of germinal center B-cell-like diffuse large B-cell lymphoma patients with a good prognosis.156 Nuclear expression of c-Rel was not shown to correlate with REL amplification or a consistent prognostic impact in patients with diffuse large B-cell lymphoma treated with R-CHOP.157, 158 Aberrant activation of the B-cell receptor signaling pathway, leading to the NF-kB, MAPK, and NFAT pathways, has an important role in lymphomagenesis of many types of lymphoma, particularly in chronic lymphocytic leukemia/small lymphocytic lymphoma and diffuse large B-cell lymphoma. Bruton tyrosine kinase (Btk), one of critical components of the B-cell receptor pathway, has become a major therapeutic target. Its inhibitor, ibrutinib, shows promising efficacy in patients with relapsed/refractory chronic lymphocytic leukemia/small lymohocytic lymphoma, mantle lymphoma, follicular lymphoma, activated B-cell-like diffuse large B-cell lymphoma, lymphoplasmacytic lymphoma/Waldenstrom macroglobulinemia, primary diffuse large B-cell lymphoma of the central nervous system, and marginal zone lymphomas which also have aberrant Btk activity.159, 160

Figure 5
figure 5

Dysregulated signaling pathways involved in lymphomas. Several aberrantly activated regulatory pathways are involved in lymphomagenesis. Among them, the NF-κB and BCR pathways are most commonly aberrantly activated. Some key components of these pathways and relevant upstream and downstream molecules are emerging as biomarkers to yield prognostic and therapeutic information. BAFF-R, B-cell activating factor receptor; BCR, B-cell receptor; BTK, Bruton’s tyrosine kinase; CARD11, caspase recruitment domain family, member 11; IKK, IκB kinase; MALT1, mucosa-associated lymphoid tissue lymphoma translocation protein 1; NEMO, NF-κB essential modifier; NIK, NF-κB-inducing kinase; mTOR, mammalian target of rapamycin; PI3K, phosphatidylinositol 3 kinase; TCR, T-cell receptor; TLRs, Toll-like receptors; TNFR, tumor necrosis factor receptor.

Immunohistochemical Biomarkers Helpful in the Differential Diagnosis of Common Types of Peripheral T-Cell Lymphoma

Biomarkers that can be assessed by immunohistochemistry are extremely helpful for distinguishing between two entities that share morphological and histopathological features, and they can be applied simply as a part of routine clinical practice. For example, there can be overlap between classical Hodgkin lymphoma and angioimmunoblastic T-cell lymphoma and the latter can have Epstein-Barr virus (EBV) -positive large cells that resemble Hodgkin Reed-Sternberg cells both morphologically and phenotypically.161 In this situation, accurate diagnosis may be aided by immunophenotypic expression of CD10, PD-1 and CXCL13 in angioimmunoblastic T-cell lymphoma.162, 163 In addition, CXCL13 and PD-1 are particularly useful in the differential diagnosis between angioimmunoblastic T-cell lymphoma and EBV+ diffuse large B-cell lymphoma, angioimmunoblastic T-cell lymphoma versus EBV+ classical Hodgkin lymphoma, and angioimmunoblastic T-cell lymphoma versus the follicular variant of peripheral T-cell lymphoma, not otherwise specified. Cases of the follicular variant of peripheral T-cell lymphoma, not otherwise specified do not have expansion of follicular dendritic cell meshworks outside the follicles and no prominent inflammatory background, but Hodgkin-like cells can be present.161, 163 SAP, an additional immunohistochemical marker of germinal center T cells, also has diagnostic value for angioimmunoblastic T-cell lymphoma, especially in combination with PD-1 and CXCL13.164 Other follicular helper T-cells biomarkers characteristically expressed by angioimmunoblastic T-cell lymphoma cells include ICOS, CD200, and c-Maf.165, 166, 167 In practice, these immunohistochemical biomarkers are of great importance in the diagnosis of angioimmunoblastic cell lymphoma and the differential diagnosis with other lymphomas.

T-cell-restricted intracellular antigen-1 (TIA-1), granzyme B, and perforin are cytotoxic biomarkers that are commonly used to diagnose NK/T lymphomas. Anaplastic large cell lymphoma is another type of lymphoma in which the neoplastic cells are positive for cytotoxic biomarkers. For anaplastic large cell lymphoma, clusterin is another potentially useful diagnostic marker.168 Granzyme H more recently has been shown to be a biomarker of NK and T-cell lymphoma and may be useful in the differential diagnosis between NK/T lymphomas and other peripheral T-cell lymphomas with cytotoxic features.169

The Jun family (c-Jun, JunB, JunD) is a member of the activator protein-1 (AP1) pathway which is involved in cell differentiation, proliferation, survival and apoptosis.170 c-Jun, JunB, and JunD are overexpressed in several cancers, including lymphomas.171, 172 An early study indicated that c-Jun and JunB were expressed strongly in the tumor cells of classical Hodgkin lymphoma and anaplastic large cell lymphoma, but not in B-cell non-Hodgkin lymphomas, indicating their potential to serve as diagnostic tools and therapeutic targets.171 Increased coexpression of galectin-1 and c-Jun in tumor cells of classical Hodgkin lymphoma and anaplastic large cell lymphoma patients has been observed, suggesting that this combination of biomarkers can distinguish classical Hodgkin lymphoma and anaplastic large cell lymphoma from other lymphomas with similar morphological and molecular features. It seems reasonable to suggest that inhibition of galectin-1, an AP1 target, may represent a potential treatment strategy for patients with classical Hodgkin lymphoma and anaplastic large cell lymphoma.173 Moreover, serum levels of galectin-1 were associated with tumor burden and adverse clinical features in a large cohort of classical Hodgkin lymphoma patients.174

Taken together, immunophenotypic biomarkers, mostly assessed by immunohistochemistry, have important roles in the diagnosis and prognostication of lymphomas as well as serving as effective drug targets. However, due to limitations of immunohistochemistry methods, such as the variable quality of antibodies and the difficulty in standardization, objective scoring and interpretation of staining results, the reproducibility and effectiveness of many immunohistochemistry assays need to be confirmed.

Microenvironment-related biomarkers for prognostication and immunotherapy

The tumor microenvironment refers to the extracellular context in which the malignant cells reside, which is commonly a plethora of non-malignant cells and extracellular matrix surrounding the malignant cells. Overwhelming evidence has shown that tumor cells and benign reactive cells in the microenvironment interact with each other dynamically and reciprocally through a variety of cytokines and chemokines that are secreted or expressed by malignant or tumor-infiltrating non-malignant cells (Figure 6). Many studies also have suggested that the microenvironment influences clinical outcome.7, 175, 176 Subsequently, biomarkers reflecting the microenvironment of lymphomas have been discovered and used to predict the clinical outcome of patients. Improved quantitative immunohistochemistry (multiOmyx and Opal assays), in situ hybridization and gene expression profiling are powerful approaches to investigate microenvironmental biomarkers. Table 3 summarizes some of the microenvironmental biomarkers related to lymphoma patient outcomes.

Figure 6
figure 6

Schematic representation of cHL showing the interaction between HRS cells and their microenvironment through cytokines/chemokine signaling and a summary of microenvironment-related biomarkers for diagnosis and prognosis in lymphomas. Malignant cells and their extracellular context interact with each other dynamically and reciprocally through a variety of cytokines and chemokines secreted or expressed by malignant cells or tumor-infiltrating non-malignant cells in lymphomas, particularly in cHL. Many benign tumor-infiltrating cells that show diagnostic or prognostic significance in lymphomas can be located or numbered through some specific biomarkers detected by immunohistochemistry or gene expression profiling. APRIL, a proliferation-inducing ligand; BCL11A, B-cell lymphoma/leukemia 11A; BLC, B lymphocyte chemoattractant; CCL, Treg, regulatory T-cell; cHL, classical Hodgkin lymphoma; CSF1, colony stimulating factor 1; CTL, cytotoxic T-cell; FASL, Fas ligand; FDC, follicular dendritic cell; HRS, Hodgkin and Reed-Sternberg; IFNγ, interferon gamma; IL, interleukin; NK, natural killer; PD1, programmed death 1; PDL1, programmed death ligand 1; STAT1, signal transducer and activator of transcription 1; SPARC, secreted protein acidic cysteine-rich; Th1, T helper 1; TNFα, tumor necrosis factor-α.

Table 3 Microenvironment-related biomarkers for lymphoma prognosis

CD21+ Follicular Dendritic Cells

Classical Hodgkin lymphoma is usually characterized by rare neoplastic cells (often <1%) within abundant reactive cells including follicular dendritic cells that comprise the microenvironment. Alavaikko et al177 reported that absence of CD21+ follicular dendritic cells in classical Hodgkin lymphoma predicted an unfavorable outcome. Another study confirmed this association and further showed that not only the number of CD21+ follicular dendritic cells, but also the extent of destruction of normal lymph node architecture, correlates with prognosis.178

Granzyme B+ Tumor-Infiltrating Cytotoxic T Cells

As a microenvironmental biomarker, granzyme B is expressed by tumor-infiltrating cytotoxic T cells that are present in several lymphoma types. Studies have shown a correlation between the number of tumor-infiltrating cytotoxic T cells and clinical outcome in patients with classical Hodgkin lymphoma or anaplastic large cell lymphoma.179, 180

CD117+ Mast Cells

CD117, expressed strongly by mast cells, can be used to evaluate the numbers of mast cells infiltrating in the microenvironment of lymphomas. One study showed that the number of mast cells was associated with unfavorable outcome of patients with classical Hodgkin lymphoma after primary therapy.181 In contrast, another study showed that mast cell infiltration in diffuse large B-cell lymphoma predicts a more favorable outcome.182

CD68+ or CD163+ Macrophages

CD68 and CD163 are expressed mainly by macrophages and can be detected easily by immunohistochemistry. Several studies have shown that an increased number of CD68+ or CD163+ macrophages is associated with poorer prognosis in patients with classical Hodgkin lymphoma or follicular lymphoma.183, 184 However, there are some inconsistent results,185 suggesting the need to validate study cohorts or methods. For diffuse large B-cell lymphoma, macrophages appear to have a dual, treatment-specific role. CD68+ macrophage numbers together with high pretreatment CD68 mRNA levels can predict a favorable outcome for diffuse large B-cell lymphoma patients treated with chemoimmunotherapy.186 Interestingly, CD163+ macrophages are useful as a prognostic indicator in pediatric classical Hodgkin lymphoma patients and this observation seems to be dependent on EBV status. It was demonstrated that high numbers of CD163+ and granzyme B+ cells are associated with inferior progression-free survival in patients with EBV negative rather than EBV positive cases of classical Hodgkin lymphoma.187

FOXP3+ Regulatory T Cells

An increased number of tumor-infiltrating FOXP3+ regulatory T cells has been shown to convey a better prognosis in many types of lymphoma, including follicular lymphoma, Hodgkin lymphoma and NK/T-cell lymphomas.188, 189, 190 It was speculated that tumor-infiltrating FOXP3+ regulatory T cells might suppress neoplastic cells in a TGF-β-mediated fashion and thereby form an inhibitory tumor microenvironment in lymphomas, at least in NK/T-cell lymphomas.190 A similar predictive trend can work for NK/T-cell lymphoma patients after IMEP chemotherapy alone and IMEP-based chemo- or chemo-radiotherapy.190 In patients with diffuse large B-cell lymphoma, the prognostic impact of tumor-infiltrating FOXP3+ regulatory T cells is debatable.191, 192

PD-1 and PD-L1

PD-1 (CD279) is expressed by tumor-infiltrating T cells in several types of lymphoma, whereas PD-L1 and PD-L2 (CD274 and CD273), the ligands of PD-1, are expressed by the malignant cells and tumor-infiltrating antigen presenting cells. High numbers of PD-1+ infiltrating T cells have been associated with inferior overall survival of classical Hodgkin lymphoma patients, respective of the disease stage.193 Furthermore, the existence of PD-1+ reactive T cells rosettes is recognized as a characteristic of nodular lymphocyte predominant Hodgkin lymphoma.194 In diffuse large B-cell lymphoma or follicular lymphoma, the prognostic value of PD-1+ tumor-infiltrating T-cell subsets is controversial,195, 196 but the expression of PD-L1 in tissue and PD-L1 levels in peripheral blood have been reported to be of independent prognostic or disease monitoring value.197, 198 Importantly, immunotherapeutic trials blocking PD-1 are successfully ongoing in lymphomas and have also seen great progresses in solid tumors. Agents that block PD-L1 are in extensive development.199, 200

CD58 and Tumor Necrosis Factor Receptor Superfamily Member 14 (TNFRSF14)

CD58 is the receptor of CD2 which is expressed mainly by T cells and NK cells.201 Decreased expression of CD58 can be observed in two-thirds of activated B-cell-like diffuse large B-cell lymphoma and one third of germinal center B-cell-like diffuse large B-cell lymphoma. Inactive CD58 may contribute to immune escape by disrupting the interaction between CD58 and CD2 and consequently facilitate other genetic defects.201, 202 Another genetic alteration involved in immune escape is TNFRSF14 which functions as a tumor suppressor and regulates immune responses. TNFRSF14 inactivation (via mutation or deletion) has been reported to be associated with poorer outcomes in patients with follicular lymphoma.203

EBV and Other Viruses

Although the pathological role of EBV infection in some types of lymphoma is well established, the contribution of EBV infection to clinical outcome is controversial. EBV positivity tends to be associated with unfavorable outcome in elderly patients with classical Hodgkin lymphoma, whereas the outcome is more favorable in young patients.204, 205 Additional studies have offered supporting evidence to confirm this trend in classical Hodgkin lymphoma and other lymphomas, like diffuse large B-cell lymphoma.206, 207, 208 EBV positivity is seen in ~5% of diffuse large B-cell lymphoma and a worse outcome was reported in Asian, East European and African American patients, but not in North American patients. EBV positivity also has been reported in occasional cases of peripheral T-cell lymphoma, many cases of anaplastic large cell lymphoma and plasmablastic lymphoma, but clinical outcome in these patients needs better validation.

In addition to EBV, other viruses also have been shown to be associated with lymphomagenesis.209, 210, 211 For example, patients with hepatitis B virus-associated diffuse large B-cell lymphoma have unique clinical features and poorer outcomes. Diffuse large B-cell lymphoma patients infected by hepatitis C virus also show good therapeutic responses to antiviral drugs. Human herpesvirus 8 (also known as Kaposi sarcoma-associated herpesvirus) infection is a characteristic of primary effusion lymphoma which occurs mainly in immunocompromised patients. HHV8 infection is also common in multicentric Castleman disease.

Microenvironment-Related Gene Expression Profiling Signatures

Several microenvironment-related gene expression profiling signatures have been established. STAT1 and ALDH1A1, expressed by macrophages, have been shown to correlate with adverse outcome.212 CD8B1, CD3D, CTSL, CD26, SH2D1A as a cytotoxic T-cell signature was shown to be associated with an inferior outcome in some studies.212, 213, 214 LYZ and STAT are expressed by tissue monocytes and activated macrophages. High levels of LYZ and STAT1 expression were correlated with prolonged failure-free survival and better outcome in patients with classical Hodgkin lymphoma.215

Gene expression profiling-based stromal gene signatures have been identified and shown to be in accord with gene expression profiling-based subgroups of diffuse large B-cell lymphoma. The activated B-cell-like subtype has an unfavorable stromal-2 signature with expression of angiogenesis markers (such as CXCL12, KDR, and CD31). In contrast, the germinal center B-cell-like subtype has a favorable stromal-1 signature with extracellular matrix deposition (biomarkers such as FN1, MMP9, CTGF, and SPARC).216

The interaction between tumor cells and their microenvironment is dynamic and complex, which may lead to uncertainty about the evaluation of microenvironment-related biomarkers. It is necessary to take cellular and matrix factors into consideration when such microenvironmental biomarkers are used.217, 218

Conclusions

The application of new generation detection platforms provides an opportunity for the assessment of genomic and transcriptomic biomarkers in lymphomas. Discovery of molecular biomarkers not only enhances our understanding of pathogenetic mechanisms in lymphomas, but also provides opportunities to refine classification, to improve accuracy of diagnosis, and to stratify the outcomes of lymphoma patients, especially in combination with classical immunophenotypic biomarkers. Nevertheless, the discovery and exploration of additional novel biomarkers lead to challenges as we need to select, validate, and translate these markers into clinical practice. Other factors, such as methods, reproducibility, quality of reagents, data interpretation, and cost efficiency, need to be taken into consideration. In particular, biomarkers used to guide treatment decisions must be reliable and reproducible. The integration of diagnosis, prognostication, and the assessment of therapeutic targets will become the norm in the assessment of lymphomas and these tasks will depend on effective application of diagnostically reliable and therapeutically meaningful biomarkers.