Welcome international healthcare professionals

This site is no longer supported and will not be updated with new content. You are welcome to browse and download all content already included in the site. Please note you will have to register your email address to access the site.

You are here

Gene expression profiling in lymphoma diagnosis and management

Best Practice & Research Clinical Haematology, 2, 22, pages 191 - 210

The classification of lymphoid malignancies has evolved from a purely morphological scheme to the current WHO (World Health Organization) classification, which takes into consideration histological, immunophenotypic, genetic and clinical information. DNA microarray technology enables the simultaneous determination of the expression levels for thousands of genes (gene expression profile; GEP) and provides a powerful approach for investigating lymphoma biology and improving disease classification. Distinct molecular signatures for many lymphomas, as well as novel lymphoma subtypes have been identified. Molecular prognosticators have also been constructed. Many of the molecular subgroups of lymphoma also show distinct patterns of genetic abnormalities. We also briefly review the application of other genome-wide techniques to the study of lymphomas, such as high resolution array comparative genomic hybridization (aCGH) and next-generation sequencing, and how these technologies will complement each other in improving our understanding of the pathobiology of lymphoma. Specific therapeutic targets will likely emerge from the increased insight into the molecular pathogenesis of the different lymphomas, thus illustrating the utility of these global studies in advancing disease management strategies.

Keywords: molecular classification, molecular prognosis, diffuse large B-cell lymphomas, peripheral T-cell lymphoma, gene expression profiling, targeted therapy.

The diagnosis and classification of non-Hodgkin lymphoma is challenging, and there have been multiple classification systems proposed in the last 50 years. In 1966, Rappaport [1] established a widely used, morphologically based classification system that was followed later by the Lukes and Collins [2] and the Kiel classification [3] . The latter two classification systems were based on the conceptual advance in immunology and incorporated the lineage of the lymphoma cell in the classification. The multiplicity of classifications led to an international effort that resulted in the Working Formulation [4] that was designed to provide a translation between the different widely used classification systems. An international group of haematopathologists later proposed the Revised European American Lymphoma (REAL) Classification that subsequently evolved into the World Health Organization (WHO) [5] classification that had broad consensus among the haematopathology community as well as the participation of the oncology community. Both the REAL and WHO classification systems emphasised the delineation of disease entities utilising all the available information, including morphology, immunophenotyping, molecular and cytogenetics data and clinical findings. Even with this much more refined classification system, the response to treatment and survival of patients within each specific lymphoma category shows significant variability, indicating that there is further biological heterogeneity not captured by the classification system.

Development of a lymphoma is initiated by a genetic alteration in a cell that predisposes the cell to undergo further genetic alterations [6] . With time, the additional acquired abnormalities promote the development of a clone that has growth and/or survival advantage over other cells [7] . This eventually will develop into a clinical lymphoma. The complement of genetic abnormalities in a tumour is a major determinant of the biology of the tumour and its clinical behaviour. The cumulative genetic abnormalities in a tumour will lead to unique gene expression profiles or signatures that can be determined experimentally by microarray analysis. Based on this assumption, it is logical to postulate that the characteristics of a tumour and its clinical behaviour can be predicted by the unique gene expression profile of the tumour. Therefore, gene expression profiling (GEP) studies may improve our diagnosis and classification of lymphomas. If GEP can identify important oncogenic pathways in the tumour, it would also help to direct therapy against these pathways that may result in more effective management of the tumour while at the same time with reduced toxicity.

We describe below some of the major findings from GEP studies on lymphoma over the past decade. We also discuss some of the ancillary technologies that may complement GEP studies and how these new investigations will potentially affect our diagnosis, prognostication and management of lymphoma patients.

Microarray analysis: technical considerations

A microarray consists of numerous regularly spaced DNA probes immobilised on a solid surface. The transcript in the sample is labelled with a fluorescent dye and hybridised to the microarray. The fluorescent signal bound to the probe is a function of the concentration of the corresponding transcript ( Fig. 1 ). The probe on the array may be prepared off-line as cDNA or oligonucleotides and spotted on the solid surface [8] . It may also be synthesised in situ using photolithographic technique or inkjet synthesis. The oligonucleotide probes may consist of individual long ones in the range of 50–70 base pairs or sets of short probes for each transcript such as the 25-base-pair-long probe sets in arrays manufactured by Affymetrix (Sanata Clara, CA). Most of the laboratories are now using oligonucleotide arrays prepared from a purchased library that is spotted by a robot on the microarray or simply pre-fabricated microarray from various companies. The currently available commercial arrays are close to whole transcriptome coverage, and arrays that cover all the expressed exons are also available so that one can determine the expression of alternatively spliced forms of the gene in addition to the expression level of the total transcripts from that gene estimated often from 3′ biased probes.


Fig. 1 Outline of methodology involved in spotted cDNA and oligonucleotide microarray gene expression analysis. The three major steps involved in a microarray experiment are: (i) the immobilization of the DNA probes on chemically activated solid surfaces within a specific and defined region. (ii) Sample preparation, labeling and hybridization. The fluorescent labeled test and standard samples are hybridized simultaneously to the array. (iii) Data imaging and analysis.

Each microarray experiment, therefore, generates a vast amount of data and requires sophisticated data management and analytical tools, many of which are now publically available [9] and [10]. There is no single best tool for all purposes, and the appropriate tools for an experiment depend on the experimental design, the type of analysis that needs to be performed and the questions being addressed. It is beyond the scope of this article to have a detailed discussion of the analytical methods but the importance of validation needs to be stressed. Validation may be computational as in leave-one-out cross-validation (LOOCV) [11] . Important findings can be validated independently by specific measurements such as quantitative reverse transcriptase polymerase chain reaction (RT-PCR) and specific conclusions can be validated by studying an independent series of patients. Assessing certain biological (e.g., genetic alterations) or clinical variables on the cases studied may provide further independent evaluation of the conclusions.

With tens of thousands of parameters measured in each sample, the number of tumour samples studied is comparatively small and generally insufficient for confident statistical conclusions. This discrepancy in dimensionality is an important reason why validation of the results and conclusions is so important in microarray experiments. While it may not be practical or possible to study thousands of cases, the number of cases included should be as large as possible since it is not feasible to have meaningful conclusions from a very limited number of cases even with sophisticated analysis.

Questions have been raised regarding the reliability and reproducibility of microarray experiments and the feasibility of cross-platform comparison of results. With improvement of microarray technology, especially with commercial microarrays, carefully performed studies with strictly controlled experimental procedures are highly reproducible using the same array platform [12] . Cross-platform comparison is more challenging; however, a recent large multicentre study [12] , examining a number of microarray platforms, has found that the results are reasonably comparable even across different platforms. It is now generally recognised that while there may be substantial variations in the measurement of individual transcripts, the comparison of different signatures represented by large groups of transcripts is much more robust. Therefore, many of the early concerns regarding the reliability and reproducibility of microarray analysis have now been addressed. Properly and carefully performed microarray experiments represent a useful approach for GEP.

GEP by sequencing of the transcriptome was not a practical approach a few years ago. However, with the development of high-throughput sequencing [13] , and the marked reduction in the price and time required, it is now possible to perform expression profiling using this platform. There are certain significant advantages using this approach including the unbiased survey of all transcripts comprising different spliced isoforms, the ability to quantitate low-level transcripts and detection of mutations or polymorphisms that may have significant biological implications.

Major findings in gene expression profiling of lymphoma

In an early study of diffused large B-cell lymphoma (DLBCL), Alizadeh and co-workers [14] found that DLBCL can be divided into at least two subtypes: one of these expressed a set of genes that are characteristically expressed by the germinal centre (GC) B cells, the GC B-cell signature, while the other subset of cases expressed a set of genes also up-regulated by peripheral blood B cells activated in vitro by mitogens. When the clinical outcome of the cases was examined, cases that expressed the GCB-cell signature (GCB-like DLBCL) had a significantly better prognosis than cases that expressed the activated B-cell gene expression signature, (ABC-like DLBCL). This initial finding was subsequently confirmed by the study of a much larger series of DLBCL by Rosenwald and co-workers [15] . In the latter study, a group of cases could not be confidently classified into the GCB or ABC subgroup, and these cases were considered unclassifiable by GEP ( Fig. 2 ).


Fig. 2 Distinct DLBCL subgroups confirmed by gene expression profiling on a separate group of cases. (A) The presence of two distinct subgroups of DLBCL was confirmed on an expanded series of cases. Apart from the GCB and ABC subgroups, a third group called Type III (unclassifiable) is identified. These cases have low expression of genes characteristic of the GCB and ABC subgroups. Relative gene expression for each lymphoma biopsy sample is presented according to the color scale shown. (B) Significantly better overall survival of GCB cases compared to ABC and Type III cases is demonstrated [Reproduced with permission from Copyright 2003 Massachusetts Medical Society Fig. 1a and c].

Since even within the GCB or ABC group, cases were still heterogeneous regarding treatment response and survival, additional predictors of outcome were analysed in this series of patients and three signatures aside from the cell of origin classification were found to be predictive of survival. These include a large group of genes associated with cell proliferation, the proliferation signature; a group of genes associated with the tumour microenvironment, the lymph node stromal signature; and a group of genes coding for the major histocompatability complex class I and II molecules, the MHC class I and II signatures. High expression of the proliferation signature and low expression of the MHC signature are associated with poor survival, whereas high expression of the lymph node stromal signature is associated with better outcome. Since this study was performed on patients prior to the widespread use of rituximab in the treatment of DLBCL, there has been quite a bit of concern regarding the applicability of these signatures in predicting survival in patients treated with regimens containing rituximab.

A follow-up study was performed by Lenz and co-workers [16] on 233 patients and found that the GCB subgroup still has significantly better survival than the ABC subgroup even with rituximab treatment. This study also found that the lymph node stromal signature can be subdivided into two signatures: one of them (stromal I signature) reflects extracellular matrix deposition and tissue cell infiltration, and high expression is associated with better outcome. The other stromal signature (stromal signature II) reflects angiogenesis, and high expression of this signature is associated with poorer survival. While high-proliferation signature is still predictive of poorer survival in a univariate model, a multivariate model indicates that the DLBCL prognosticator can be represented by the cell of origin and the stromal I and II signatures ( Fig. 3 ) [16] .


Fig. 3 Gene-expression predictors of survival among patients with DLBCL treated with R-CHOP. Kaplan–Meier estimates of progression-free and overall survival are shown. Panel A shows that patients with germinal-center B-cell–like diffuse large-B-cell lymphoma had a higher probability of progression-free survival (left) and overall survival (right) than patients with activated B-cell-like diffuse large-B-cell lymphoma. Panel B shows a gene-expression–based predictor of survival among patients with diffuse large-B-cell lymphoma treated with R-CHOP. Kaplan–Meier estimates of progression-free survival (left) and overall survival (right) are based on a multivariate model derived from the germinal-center B-cell, stromal-1, and stromal-2 gene-expression signatures. Survival-predictor scores derived from this model were used to rank the cases of lymphoma, which were then divided into quartile groups as indicated. R-CHOP denotes rituximab plus combination chemotherapy with cyclophosphamide, doxorubicin, vincristine, and prednisone. Reproduced with permission from Copyright 2008 Massachusetts Medical Society Fig. 1.

Monti et al. [17] studied 175 DLBCL cases and identified three groups of cases by ‘consensus clustering’ that they termed ‘oxidative-phosphorylation (Ox-Phos)’, ‘B-cell receptor/proliferation’ and ‘host response’. The first two groups are characterised by the differential expression of functional sets of genes with the B-cell receptor/proliferation cluster showing increased expression of genes related to BCR signalling, cell proliferation and replication, DNA repair and transcription factors. The Ox-Phos cluster shows up-regulation of genes related to oxidative phosphorylation, mitochondrial function and the electron transport chain. The host response cluster was enriched in transcripts from stromal cells such as T cells, dendritic cells and macrophages and included many of the cases of T-cell-rich B-cell lymphoma. While the GCB and ABC DLBCL subgroups described by Rosenwald et al. [15] can be identified in this series of cases, there is no good correlation with the consensus clusters, indicated that these two classifications were based on different aspects of tumour biology.

There is also an uncommon type of DLBCL that tends to occur in the anterior mediastinum in young female patients. This tumour often has very characteristic morphologic features composed of large cells with pale cytoplasm and fine compartmentalised fibrosis. Two GEP studies [18] and [19] on primary mediastinal DLBCL (PMBCL) have finally resolved the issue whether this tumour represents a unique type of DLBCL. Both studies showed that this lymphoma is associated with a distinct gene expression profile different from the GCB and ABC subtypes of DLBCL ( Fig. 4 ). Interestingly, there is a significant overlap in the gene expression profile of PMBCL with Hodgkin lymphoma cell lines, suggesting shared biological processes between these two tumour types. A re-examination of the series of DLBCL cases by Rosenwald and co-workers [18] has identified a number of cases with PMBCL signature but clinically did not have a primary mediastinal presentation. These cases tend to occur in thoracic structures above the diaphragm and the signature may allow us to expand the spectrum of PMBCL to include these extra mediastinal cases.


Fig. 4 Molecular diagnosis of PMBL. (A) A PMBL predictor based on the expression of 46 genes can differentiate PMBL from other types of DLBCL. The probability that a sample is PMBL or other DLBCL based on gene expression is shown at the top. Note that there are DLBCL cases in the mediastinum without the PMBL signature indicating that they are non-PMBL that happened to present as a mediastinal mass. (B) Kaplan–Meier plot of survival showed PMBL to have a favorable prognosis compared to the ABC subtype of DLBCL [Reproduced with permission from Journal of Experimental Medicine (27): Figs. 2a and 3a].

There are other studies on subsets of DLCBL based on site of presentation, such as the central nervous system [20] or the skin [21] . Cutaneous DLBCLs, which involve the lower leg, have been found to have worse prognosis compared with those involving other areas of the skin and have been proposed to be classified separately as a distinct entity. It is interesting that the tumours involving the leg are invariably of the ABC subtype, while those involving other areas of the skin are of the GCB subtype [21] . The observation correlated well with the known worse outcome of the ABC subtype of DLBCL. Why the ABC subtype of cutaneous DLBCL would preferentially involve the lower leg is not clear.

About 10% of DLBCL aberrantly express the T-cell-associated antigen CD5. Kobayashi et al. compared the GEP of CD5+ DLBCL with the CD5– cases [22] and suggested that the CD5+ tumours constitute a distinct subset of DLBCL that may have worse prognosis. Primary effusion lymphoma is an uncommon entity usually associated with HIV-infected patients and is characteristically EBV positive. Several studies have been performed on small number of cases [23] and [24]. It would be interesting to investigate other uncommon DLBCL subtypes, but the major impediment is the difficulty of obtaining a sufficient number of high-quality samples with adequate ancillary information.

Mantle cell lymphoma is characterised by t(11;14) where the CCND1 locus on chromosome 11 is brought into the proximity of the immunoglobulin heavy chain gene (IgH) enhancers resulting in the over-expression of cyclin D1. GEP shows that this type of lymphoma has a well-defined signature [25] that can readily distinguish it from other types of lymphoma. In a study of over 90 cases of MCL, Rosenwald and colleagues have found a number of cases with the Mantle Cell lymphoma signature but without cyclin D1 over-expression. In a subsequent more detailed study, a small subset of cyclin-D1-negative MCL has been defined ( Fig. 5 ) [26] . Some of these cases have a high expression of cyclin D2 or D3. However, expression of cyclin D2 and D3 is not specific and is not as useful as cyclin D1 over-expression in identifying MCL. Therefore, a firm diagnosis of these types of cases still has to rely on the mantle cell lymphoma signature [25] . Interestingly, a few cases with translocation involving the cyclin D2 or D3 loci and over-expression of cyclin E associated with a cryptic N-MYC translocation have been reported [27] .


Fig. 5 Expression profiles of mantle cell lymphoma (MCL) signature genes in 6 cases of cyclin D1-negative MCL using Affymetrix U133 A/B arrays. These expression profiles are compared with 22 cases of cyclin D1-positive MCL, 78 cases of activated B-cell-like (ABC), 85 cases of germinal center B-cell-like (GCB), and 33 cases of primary mediastinal (PMBL) variants of diffuse large B-cell lymphoma, 193 cases of follicular lymphoma (FL), 14 cases of extranodal marginal zone lymphoma, MALT type (MALT), 6 cases of splenic marginal zone lymphoma (SMZL), and 14 cases of small lymphocytic lymphoma (SLL) (median expression levels of the MCL signature genes in these entities are shown). In the 6 cases of cyclin D1-negative MCL, each column represents a single lymphoma specimen and each row represents the level of expression of a single gene in the MCL signature. In the bottom panel, the gene expression levels of the D-type cyclins in the various entities and the 6 cases of cyclin D1-negative MCL are shown. [Reproduced with permission from Blood (American society of hematology)].

An attempt was made to identify a prognosticator for MCL, and a group of genes that is associated with cell proliferation has been identified to be the major determinant of prognosis ( Fig. 6 ). Patients with tumours that were in the top quartile of the expression of the proliferation signature had much worse prognosis than patients whose tumours were in the lowest quartile. There is a positive correlation between the proliferation signature average, the level of cyclin D1, deletion of the INK4A/ARF locus, suggesting multiple factors that affect prognosis may converge to alter cell cycle and proliferation.


Fig. 6 Gene expression profiling of MCL. (A) Expression of 20 genes involved in proliferation was used to define the proliferation signature. (B) Survival of patients in each quartile divided according to the proliferation signature average, showing the relationship between proliferation and survival [Reproduced with permission from Elsevier: Figs. 2a and b].

Dave and co-workers have determined the gene expression profiles of 191 cases of follicular lymphoma (FL) [28] . Interestingly, in this lymphoma, the major determinant of prognosis appears to be related to two gene expression signatures derived from the microenvironment. These two signatures (immune response-1 and -2) can be combined to form a molecular predictor of survival for FL. The immune response-1 signature contains a number of genes that are known T-cell transcripts, but the signature does not just reflect the presence of tumour-infiltrating T cells, and a high expression of these genes is associated with better outcome. Immune response-2 signature contains many transcripts expressed by myeloid and monocytic cells, and a high expression of this signature is associated with poor prognosis. Using this molecular prognosticator, a group of patients with poor prognosis and requiring special attention in management can be identified ( Fig. 7 ).


Fig. 7 Survival and genes associated with prognosis in follicular lymphoma. (A) The hierarchical clustering of survival-associated genes according to their expression in the training set of 95 follicular lymphoma biopsy specimens. The dendrogram shows the degree to which the expression pattern of each gene is correlated with that of the other genes; the colored bars represent sets of coordinately regulated genes, defined as gene-expression signatures. To the right of the dendrogram, the genes making up the immune response 1 and immune response 2 signatures that formed the survivor-predictor model are listed. (B) Kaplan–Meier survival curve for all the patients for whom these data were available. (C) Overall survival among the patients with biopsy specimens in the test set, according to the quartile of the survival-predictor score (SPS) [Reproduced with permission from Copyright 2003 Massachusetts Medical Society: Fig. 1a and c].

These studies highlight the importance of the tumour microenvironment in influencing the prognosis of patients with DLBCL and FL. This observation also provides additional impetus in designing therapy that may alter the microenvironment, in addition to targeting tumour cells per se. However, it should be pointed out that the failure to detect the tumour-associated predictors of survival in FL does not indicate that the tumour component has little or no influence on survival. This may be due to the large components of reactive elements in many follicular lymphomas that might have diluted the signal from the tumour. Furthermore, there are frequently multiple related subclones of follicular lymphoma in the same patient [29] and GEP of a tumour from a certain site may not be predictive of survival if another clone at a site that is not biopsied is the determinant of survival in that patient. Tumour/host interaction, on the other hand, may be a more consistent characteristic and is therefore delineated as an important prognosticator in these studies.

In 25–60% of patients, the FL transforms to an aggressive lymphoma associated with a marked worsening of survival [30] . A variety of genetic changes have been reported with this transformation such as c-MYC re-arrangement [30] , p53 mutation [31] , mutations in the 5′ untranslated region of the BCL6 gene [32] , mutations of the translocated BCL2 gene [33] or p15 or p16 deletions, mutations or hypermethylation [34] . It is not clear if these changes are directly associated with the transformation event or may reflect interval changes that occurred between the initial biopsy and the transformation. Other genetic aberrations that are important in transformation may not have been described yet. Lossos and co-workers found that [35] only a subset of transformed FL shows increased expression of cMYC and its target genes. This study also revealed that transformed FL has different gene expression profiles from de novo DLBCL. In an extended study using the same cases, array comparative genomic hybridisation (aCGH) was performed [36] . This study found a number of imbalances not previously described in FL transformation including gains of 9p23–p24, which is common in PMBL and classical HL, gains of 6p12–p21, a site including oncogenes PIM1 and CCDN3 and frequently targeted by translocation and copy number increase in B-NHL and multiple myeloma and gains of 17q21.33. None of the transformed cases had elevated copy numbers of the MYC gene locus on 8q24 [35] . Thus, over-expression of MYC cannot be accounted for by gene amplification. The authors suggested that one of the possible mechanisms could be related to changes in promoter methylation. The combination of aCGH and GEP studies on the same patients may help to elucidate the functional consequences of the genetic alterations as well as identification of target genes in the abnormal loci.

In another study of FL transformation, 12 matched pairs of FL and DLBCL were examined and all the DLBCL exhibited a GCB-like profile [37] . Several members of the RAS family, growth factors and cytokine receptors with known growth-promoting activity (C-MET, FGFR3, LTβR and PDGFRβ) and p38MAPK were found to have high expression levels in the DLBCL. A three-gene predictor (PLA2, PDGFRβ and Rab-6) for transformation was proposed. The authors also showed that inhibition of p38MAPK blocked the growth of t(14;18)+ cell lines and inhibited growth of transplanted tumours in non-obese diabetic/severe combined immunodeficient (NOD-SCID) mice, thereby suggesting that p38MAPK may be a promising target for transformed FL.

Several uncommon classes of B-cell lymphomas have been examined by GEP. In a study by Thieblemont and colleagues [38] , small lymphocytic lymphoma and splenic marginal zone B-cell lymphoma (MZL) were reported to be distinguishable by GEP. There are still many questions concerning the relationship between nodal, extranodal and splenic MZL and the heterogeneity within these entities. Development of accurate and biologically relevant class predictors will improve the diagnosis as well as our understanding of this group of lymphomas. However, MZL is challenging to study because of its relatively low incidence, frequently small biopsy size, admixture of normal reactive immune cells and involvement of many different extranodal sites.

Peripheral T-cell lymphoma

Peripheral T-cell lymphoma (PTCL) is uncommon and constitutes about 10–15% of all NHLs [39] and [40]. The diagnosis of PTCL is frequently challenging, and up to 50% of the cases are categorised as PTCL, unspecified (PTCL-U) as attempts to further characterise this group of disorders have been unsuccessful. The other PTCLs are classified into angioimmunoblastic T-cell lymphoma (AITL), anaplastic large cell lymphoma (ALCL), adult T-cell leukaemia and lymphoma (ATLL) and other rare entities that mostly present extranodally [5] .

Due to the low incidence of PTCLs, they are less extensively studied than their B-cell counterpart with far-less understanding of their pathogenesis and pathobiology. Unfortunately, with the exception of ALCLs that express anaplastic lymphoma kinase (ALK), this group of tumours also shows a poor response to standard chemotherapy with corresponding poor patient survival [41], [42], and [43] ( Fig. 8 ). Given the poor results with conventional chemotherapy like CHOP, it is imperative to obtain better understanding of the biology of this group of disease to discover novel treatment that may improve the outcome. New diagnostic tools, molecular genetic markers and prognostic models to better define these entities will assist clinical investigators in designing improved therapies and clinical trials in the future.


Fig. 8 PTCL and NKCL overall survival (OS) by diagnosis group. The OS is based on the clinical data from the international PTCL consortium (JCO, 2008 Vose et al).

Several recent GEP studies with small series of cases have begun to show that PTCLs can be segregated into distinct clusters [44], [45], [46], and [47]. One important finding is that many of the pathologically defined AITL tend to cluster together, supporting the concept that it is a distinct type of PTCL. In addition, comparing the expression profile of AITL with that of various types of T lymphocytes demonstrates enrichment in the expression of genes that are characteristic of follicular helper T (TFH) cells [44] ( Fig. 9 ). Interestingly, a T-cell-intrinsic abnormality with the over-expression of ICOS, a T-cell co-stimulatory molecule, in mice with mutated ROQUIN shows expansion of TFH cells and the development of a lymphoproliferative disorder reminiscent of AITL [48] .


Fig. 9 GSEA analysis. Enrichment curves and clustering results for the 2 significant gene signatures related to follicular T helper cells significantly overexpressed in AITL compared to PTCL-u. Reproduced with permission from Blood (American Society of Hematology. de Leval, L. et al. Blood 2007;109:4952–4963).

Among PTCLs, AITL, in particular, has increased numbers of Epstein–Barr virus (EBV)-infected cells, and many AITL patients show evidence of autoimmunity as well as a propensity to develop opportunistic infections, reflecting profoundly disturbed immunoregulation. EBV is present mainly or exclusively in B cells rather than the neoplastic T cells and, in some cases, a clonal B-cell proliferation emerges from the EBV-transformed population [49] . This failure to control transformed B cell may also be associated with similar defects in immune surveillance and elimination of neoplastic T cells in the tumour.

An immunosuppressive environment may be induced by a variety of mechanisms, including regulatory T cells and tolerogenic dendritic cells [50] . Immunosuppression may be mediated by the secretion of immunosuppressive molecules such as INDO [51] , interleukin 10 (IL10) [52] and transforming growth factor beta (TGFβ) [53] or through membrane molecules that transduce inhibitory signals, for example, CTLA4 and PD1. The gene expression profile of AITL indeed shows evidence of the presence T regulatory cells and the up-regulation of IL10 and TGFβ. There is also expression of PD1 on the tumour cells. AITL is characterised by a high expression of a large number of chemokines and cytokines and many of their receptors. These include CXCL13, CXCR4 and IL21 [54] , and CXCL13 has been frequently used as a marker for the diagnosis of this type of PTCL. AITL is also characterised by a B-cell expression signature probably from the EBV-transformed B cell in the microenvironment. Interestingly, AITL highly up-regulates the NF-kB pathway, most likely in both the tumour and the microenvironment suggesting that this pathway is very important in the survival and tumour biology of AITL [55] ( Fig. 10 ).


Fig. 10 Enrichment of NF-κB target genes in AITL. GSEA revealed enrichment of NF-κB target genes in AITL compared to PTCLU (Iqbal et al, Unplublished data).

ALK + ALCL is a distinct entity characterised by the over-expression of ALK kinase due to its translocation to a number of partners that cause an inappropriate expression and activation of this kinase in lymphoid cells [56] . Approximately 80–85% of the ALK-positive ALCLs are associated with the t(2;5) (p23;q35), [57] which juxtaposes the nucleophosmin (NPM) gene at 5q35 involved in shuttling ribonucleoproteins from the cytoplasm to the nucleus, to the anaplastic lymphoma kinase (ALK) gene at 2p23, a tyrosine kinase receptor belonging to the insulin receptor superfamily [58] . The NPM-ALK fusion protein has been shown by immunohistochemistry to localise in the cytoplasm and the nucleus of the neoplastic cells, thereby providing a distinctive marker for t(2;5)-positive ALCLs [58] . Fifteen to twenty percent of ALK + ALCLs harbour variant fusion partners [5] . One major functional consequence of inappropriate expression of ALK kinase is the up-regulation of the STAT3 [59] and downstream pathways, including PI3K-AKT [60] with resulting changes in expression of its target genes. A recent report demonstrated that NPM1-ALK induces epigenetic silencing of STAT5A gene via STAT3 and that STAT5A protein can act as a key tumour suppressor by reciprocally inhibiting expression of NPM1-ALK [61] . Piva et al. [62] have shown that C/EBPbeta and the anti-apoptotic protein BCL2A1 are absolutely necessary to induce cell transformation and/or to sustain the growth and survival of ALK-positive ALCL cells. Few genome-wide and proteomic studies performed on limited number of patient samples or cell lines have identified a number of genes involved in proliferation and cell-cycle regulations (G1/S check points) and anti-apoptosis highly up-regulated in ALK + ALCL [63] and [64] compared to normal T cells. We have identified a gene expression signature that can identify the vast majority of cases of this entity and separate it from other types of PTCLs.

There is another uncommon type of PTCL that resembles ALK + ALCL morphologically and also expressed many of the immunohistochemical markers of ALCL, including CD30, clusterin and cytotoxic molecules, classified as ALK (-) ALCL [65] . In addition, it also shows the abnormal failure of expression of many T-cell-associated markers, as in ALK + ALCL [66] . The relationship between ALK– and ALK + ALCL is unclear but morphologically and immunohistochemically, ALK– ALCL seems to be quite distinct from other PTCLs. The GEP studies demonstrated that it does not form tight clusters with ALK + ALCL and other PTCL subgroups and the molecular classifier of ALK+ cases does not identify these cases. Salaverria et al. identified gains of 17p and 17q24-qter and losses of 4q13–q21, and 11q14 associated with ALK-positive cases (P = 0.05), whereas gains of 1q and 6p21 were more frequent in ALK-negative tumours (P = 0.03) [67] . Current evidence suggests that this is a disorder that is distinct from ALK + ALCL and other PTCLs, but there is no robust classifier that can identify these cases as a distinct molecularly defined entity. Further studies are necessary for its better characterisation.

ATLL has limited geographical distribution and is endemic in southwestern Japan, the Caribbean region, parts of the Middle-East, South America and Papua New Guinea [68] . It is characterised by infection and transformation of T-helper cells by the retrovirus HTLV1, which is believed to be the primary pathogenetic event in this type of lymphoma [69] . ATLL occurs in 2–4% of the HTLV-1 carriers with a long latent period, suggesting that additional alterations are required for its development. The viral gene TAX plays a critical role in transforming T cells [70] by regulating the genes involved in T-cell activation and proliferation, including IL-2, IL-2Rα, IL-5, IL15R, GMCSF and TNF-α [71] . Gene expression studies have recently showed markedly increased expression of tumoursuppressor in lung cancer 1 (TSLC1), caveolin 1 and prostaglandin D2 synthase [72] . The introduction of TSLC1 into a human ATL cell line-ED enhanced its aggregation and adhesion capability to vascular endothelial cells [72] . Another expression profiling study of uncultured lymphocytes from ATLL patients showed increased expression of genes linked to the cell cycle (CDC2, cyclin B), hypercalcaemia (RANKL, PTHLH), tyrosine kinase signalling (SYK, LYN) pathways and anti-apoptosis (BIRC5) [73] . The authors identified that TCF4 expression is linked to BIRC5 and introduction of BIRC5 shRNA into ATLL cells specifically down-regulated BIRC5 expression and decreased ATLL cell viability, suggesting BIRC5 as a rational clinical target in the treatment of ATLL. ATLL has a distinct gene expression signature with many of the transcripts induced by TAX. This signature can readily differentiate it from other types of PTCL [Iqbal, unpublished].

A gene expression profiling study on PTCLU has demonstrated that it can be clustered into three distinct clusters [74] . However, the clusters may reflect the differences in the tumour microenvironment rather than the unique characteristics of the tumour cells. This category of tumour is likely to be heterogeneous and requires a significantly larger number of cases for it to be characterised. A genetic study on pathologically diagnosed PTCLU and AITL indicates a substantial difference in the genetic profiles between these two categories of tumours, with PTCLU having more complex karyotypes and AITL more often characterised by simple gains [75] . However, while there are frequent abnormalities and distinct differences identified in these lymphomas, there are no specific disease-defining abnormalities.

There have been a number of attempts to define prognosticators for PTCLs. High expression of the proliferation signature has been reported to be associated with poor survival, while high expression of NF-kB signature has been reported to be associated with better outcome [55] and [76]. These studies were performed on a small number of cases and frequently they are not performed on a single molecularly defined entity. Therefore, it is uncertain how robust are these findings and how well they can be applied to current clinical management. Further investigations into a much larger series of patients with molecularly defined cases will be necessary to obtain reliable predictors of survival.

Many of the PTCL entities are rare. A large international effort will have to be mounted to perform meaningful studies on these entities. However, even with a large collaborative effort, there may not be sufficient cryopreserved tissues for investigation. Recently, there have been significant advances in the technology on GEP using formalin-fixed, paraffin-embedded (FFPE) tissues [77] . With the potential of extending GEP investigations to archival tissues, it may be possible to access and study a sufficient number of cases to advance our knowledge on the rare types of PTCL.

The integration of GEP with other global studies

While GEP has proven to be a powerful approach in furthering our understanding of the pathogenesis and biology of malignant lymphoma, it is quite clear that for a more comprehensive understanding of this disease, the availability of additional global information such as genetic alterations, mutational and methylation status will be very helpful or even essential.

Comparative genomic hybridisation (CGH) is a very useful technique in determining copy number abnormalities in the genome without having to culture living cells [78] . Traditionally, metaphase spread has been used as an indicator system and the resolution of this technique is in the megabase range [79] . Recently, a number of array-CGH platforms have been developed using DNA clones of large genomic fragments such as BAC clones [80] . Dr. Wan Lam's group at the British Columbia Cancer Agency has performed extensive development of this platform, and by using overlapping BAC clones, this group has been able to substantially increase the sensitivity of this system [81] . More recently, oligonucleotide arrays have been developed with extremely high resolution and are now available from companies such as Agilent (Palo Alto, CA, USA) or NimbleGen (Madison, WI, USA). Some of these arrays are developed not only to study single nucleotide polymorphisms (SNPs), but they can also be adapted to study copy number variation [82] . The SNP arrays, in addition to providing information on copy number changes, can also provide information on copy neutral loss of heterozygosity (LOH). Some large studies have been performed using high–resolution-array CGH and have provided highly useful information in acute lymphoblastic leukaemia [83] , diffuse large B-cell lymphoma [84] and NK cell malignancies [85] . The integration of the array CGH information with GEP data on the same cases is a very powerful approach in helping to identify the functional abnormalities caused by the copy number alterations, as well as in identifying the target genes in the abnormal regions. This is illustrated by a recent study [84] where using the combined data set, a number of highly probable target genes have been identified in many loci with common copy number abnormalities ( Fig. 11 ). In another study on NK cell malignancies, using a similar approach, PRDM1 has been identified as a likely target gene in 6q21 deletion, a very common abnormality in this type of malignancy [85] . In addition to abnormalities induced by single target genes, some of these copy number abnormalities may exert a global influence on tumour cell proliferation, growth and survival by the alterations of large groups of genes serving these functions.


Fig. 11 Identification of minimal common regions of CNA. Combining aCGH and GEP in the identification of abnormal pathways and genes: The recurrent genomic gains and losses with high resolution array CGH can identify the putative TSGs and oncogenes in these regions with the help of corresponding GEP data. Reproduced with permission from PNAS (Lenz et al 2008;105(36):13520--5.).

It has also been noticed that microRNAs (miRNAs), a group of small regulatory RNAs of ∼22 nucleotides, regulate the expression of large groups of genes post-transcriptionally [86] . One of the main functions of miRNA is in the regulation of developmental processes, and many of these miRNAs are tissue- or developmental-stage-specific and have been shown to be useful in identifying the cell of origin in a wide variety of tumours, including lymphomas [87], [88], [89], and [90]. It has also been noticed that miRNA are quite often located in fragile sites in the genome and in regions often involved in copy number abnormalities and chromosomal breakpoints [91] . It is interesting that a number of these miRNAs have been identified to be possible target genes in these regions such as miRNA 15a and 16-1 in chronic lymphocytic leukaemia and the miRNA cluster 17–92 in the GCB type of DLBCL and MCL [84] . MiRNA 155 is coded by a non-protein-coding gene BIC that is a frequent avian leukosis virus insertional site [92] . This miRNA is often highly up-regulated in the ABC-DLBCL and in HL and is likely important in their pathogenesis [93] , but down-regulated in the Burkitt's lymphoma [94] . Several recent publications have suggested that expression of certain miRNA may be associated with poorer survival or more aggressive tumour [95] . Therefore, miRNA profiling may also be able to identify prognostically important signatures in lymphoma. It is quite clear that additional important information on the role of miRNA in cancer will be discovered in the future, and miRNA profiling will complement the data obtained in both GEP and array CGH studies.

Normal B lymphocytes undergo somatic hypermutation (SHM) in the GC stage of differentiation. In the normal physiological state, this affects the immunoglobulin genes and the BCL6 gene at a lower frequency [96] . For genes that are translocated to the immunoglobulin heavy-chain gene locus, such as cMYC and BCL2, SHM may affect the translocated partner genes as if they are part of the IgH gene. However, it has been found that in the neoplastic state, mutation not only affects the immunoglobulin gene locus but also genes that are not normally involved by SHM such as Pax5, PIM1 and RhoH. As in other tumours, TP53 mutations are found in lymphoma and mutations of the DNA-binding domain appear to be particularly relevant to the treatment outcome [97] . More recently, PRDM1 has been found to be frequently mutated in the ABC type of DLBCL [98] and [99] giving rise to a truncated non-functional protein. This gene is a master regulator of plasma-cell differentiation and the loss-of-function mutation observed may impair B-cell differentiation and contribute to lymphomagenesis. Mutants that may contribute to the activation of the NF-κB pathway have been described [100] , and additional activating mutations have recently been identified affecting CARD11, TNFAIP3 and BCL10[100], [101], and [102]. It is quite clear that many more mutations affecting critical genes that may alter important pathways will be detected in the future, especially using high-throughput sequencing [103] . This will obviously be an important aspect in the investigation of lymphomas and will likely provide explanations to the observed abnormalities in certain oncogenic pathways.

Abnormalities in global methylation are common in malignancies [104] . These include hypomethylation in many regions that may lead to the inappropriate activation of oncogenes and also genomic instability [105] . There is also inappropriate methylation in the promoter regions in many genes, including tumour suppressor genes that may facilitate the pathogenesis and progression of a tumour [106] . It is important to point out that data from limited methylation analysis of a genetic locus may not correlate well with gene expression [107] . A more comprehensive methylation analysis is necessary to obtain a complete picture of the methylation status of a gene and to more accurately interpret methylation data. Concurrent gene expression data would be extremely helpful in the interpretation of functional alterations associated with changes in methylation status in different regions of the gene locus.

Detection of oncogenic pathways by GEP and implications on therapy

Several studies have demonstrated the potential of GEP to identify oncogenic pathways in different types of cancers [108] and [109] and coordinated deregulation of multiple pathways may identify subgroups of patients with unfavourable outcome [109] . Furthermore, cell line experiments have demonstrated the correlation between pathway prediction and sensitivity to therapeutic agents targeting the pathway [109] . Thus, these studies have shown the potential of GEP to define oncogenic pathways, which may be novel therapeutic targets, and to identify the subset of cases that may benefit from a specific therapy. An example of this comes from the profiling studies of DLBCL, in which the ABC and PMBL subgroup of DLBCL showed an increased expression of subsets of target genes of the transcription factor NF-κB [18], [19], and [108]. Davis et al. [108] demonstrated constitutive activation of NF-κB in ABC-derived cell lines that underwent apoptosis when the NF-κB pathway was inhibited. Several NF-κB inhibitors are currently available and could be tested in clinical trials. Similarly, the GCB subgroup of DLBCL, which shows high expression of the BCL6 oncogene, may benefit from the recently designed peptide fragment that inhibits the function of BCL6 [110] . Recently, we have shown that survival in DLBCL after treatment with R-CHOP is influenced by the pattern of host/tumour interaction, including angiogenesis in the tumour microenvironment [16] . The subset of patients with high angiogenesis signature may be targeted by VEGF inhibitors such as bevacizumab. Similarly, high expression of several members of the RAS family and p38MAPK pathway in transformed FL patients revealed by GEP may benefit by pharmacological targeting of the p38MAPK pathway [37] , as the inhibition of p38MAPK has been shown to block the growth of t(14;18)-positive cell lines and inhibit tumour growth in the NOD-SCID mice. Profiling studies in MCL have demonstrated that patient survival is determined to a large extent by the expression level of the proliferation signature [25] . Therapeutically targeting cyclin D1/CDK4 by a p16INK4a-mimetic that could disrupt the proliferative machinery of MCL cells should prolong the survival of MCL patients.

NF-κB has been identified as particularly activated in AITL in our GEP study (Iqbal J, submitted). Indeed, bortezomib (a proteosome inhibitor with known NF-kB inhibitory function) has demonstrated some early promise in the treatment of PTCL [111] , and perhaps this class of drugs would be particularly effective in AITL. The frequent high expression of PDFGRα in PTCL-U [112] may also be a possible therapeutic target and imatinib (Gleevec) is currently under clinical trial for these cases.

More and more agents that specifically target biologic pathways are now becoming available for clinical studies. These developments converge to provide an opportunity for novel therapeutic intervention where oncogenic pathway signatures can be used to select specific pathway targeting agents. It is possible to rationally design clinical trials in which treatment cohorts are stratified according to molecular-profiling-defined criteria. This approach will test the feasibility of personalised treatment based on agents specifically targeting oncogenic pathways found in the tumour.


Microarray studies are promising, and it is anticipated that GEP and other genomic-wide studies will provide information that will have enormous impacts on molecular diagnostics and patient management. Molecular diagnostics should provide information on key factors that determine the biological and clinical behaviour of a tumour, so that the patient can receive treatment directed against these molecular abnormalities. It is not clear what platform will this information be used in the clinical setting. Useful signatures from GEP can be condensed and represented by a much smaller number of transcripts so that a diagnostic array with several hundred probes may be adequate for haematologic malignancies. Frozen tissues may be difficult to obtain for diagnosis in routine clinical practice, which may hinder the use of microarrays as a diagnostic platform, unless robust technique for GEP on paraffin tissue is developed. Alternative platforms using quantitative RT-PCR [113] , RNA protection assay [114] or immunohistochemistry [115] may also be developed to apply the knowledge gained from GEP studies to paraffin-embedded materials.

It is also not clear if a single diagnostic platform can be designed that will incorporate important information from multiple sources (e.g., GEP, aCGH, methylation and mutation analysis) or several assays will have to be used to capture all relevant information. However, it is likely that some form of diagnostic assay for haematological malignancies based on these global analyses will be adopted in clinical practice to provide additional relevant molecular information for patient management in the near future. It may be argued that the whole transcriptome array may not be necessary for molecular diagnostics, but if the goal is to have pathway-directed therapy, fairly extensive information has to be obtained from the specimen and an array with limited content may not be ideal. Furthermore, with a whole transcriptome platform, it is not necessary to make individual diagnostic arrays as different diagnostic algorithm can be applied on the same platform to obtain the desired information.

Practice points


  • GEP studies on lymphoma has provided promising results in the identification of new disease entities, in the development of molecularly-defined prognosticators, in the discovery of novel genes and pathways, and in the understanding of pathogenetic mechanisms.
  • DLBCL can be divided into at least two molecular subgroups, GCB and ABC subgroups, and should be regarded as distinctive clinicopathological entities.
  • PMBL can be molecularly defined as distinct subgroup of DLBCL, which at molecular level shows some similarity with Hodgkin’s disease.
  • MCL is a homogeneous disease at molecular level, but can be prognostically divided, based on the expression of proliferation gene signature.
  • FL and DLBCL prognostic models are influenced by the microenvironmental gene signatures.
  • PTCL represent a heterogeneous group of T-cell neoplasms, but now few entities can be molecularly associated with cellular background.
  • GEP will help diagnostic-medicine towards a mechanism-based diagnosis for individualized therapy.
Research agenda


  • Refine and confirm the utility and reliability of the previously-defined gene expression signatures.
  • Identify and validated unique gene expression signatures for additional, uncommon categories of lymphomas.
  • Incorporate of the molecular findings from other novel technologies to a unified diagnostic model.
  • Identify the ideal approaches to translate the findings this model into clinical practice.
  • Identify the molecular targets and develop novel therapeutic approaches for lymphomas.

Conflict of interest statement

None declared.


This work was supported in part by an NCI grant (5U01/CA114778), Department of Health and Human Services.


  • [1] H. Rappaport. Tumors of the hematopoietic system. Atlas of tumour pathlogy. (AFIP, 1966) Section III
  • [2] R.J. Lukes, R.D. Collins. Immunologic characterization of human malignant lymphomas. Cancer. 1974;34(4 Suppl) suppl:1488–1503
  • [3] A.G. Stansfeld, J. Diebold, H. Noel, et al. Updated Kiel classification for lymphomas. Lancet. 1988;1(8580):292-293 Crossref.
  • [4] National Cancer Institute sponsored study of classifications of non-Hodgkin's lymphomas: summary and description of a working formulation for clinical usage. Cancer. 1982;49:2112-2135
  • [5] Steven H. Swerdlow, Elias Campo, N.L. Harris, et al. WHO classification: pathology and genetics of tumors of haematopoietic and lymphoid tissues. (IARC Press, Lyon, France, 2008)
  • [6] P.L. Pearson, R.B. Van der Luijt. The genetic analysis of cancer. J Intern Med. 1998;243(6):413-417 Crossref.
  • [7] A.G. Knudson. Hereditary predisposition to cancer. Ann N Y Acad Sci. 1997;833:58-67 Crossref.
  • [8] H.P. Saluz, J. Iqbal, V.L. Gino, et al. Fundamentals of DNA-chip/array technology for comparative gene-expression analysis. Curr Sci. 2002;83:829-833
  • [9] G. Sherlock. Analysis of large-scale gene expression data. Curr Opin Immunol. 2000;12(2):201-205 Crossref.
  • [10] BRB-ArrayTools user guide, version 3.6.0. http://linus.nci.nih.gov/BRB-ArrayTools.html Available from:
  • [11] R. Simon. Microarray-based expression profiling and informatics. Curr Opin Biotechnol. 2008;19(1):26-29 Crossref.
  • [12] L. Shi, L.H. Reid, W.D. Jones, et al. The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements. Nat Biotechnol. 2006;24(9):1151-1161 Crossref.
  • [13] J. Shendure, H. Ji. Next-generation DNA sequencing. Nat Biotechnol. 2008;26(10):1135-1145 Crossref.
  • [14] A.A. Alizadeh, M.B. Eisen, R.E. Davis, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000;403(6769):503-511 Crossref.
  • *[15] A. Rosenwald, G. Wright, W.C. Chan, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med. 2002;346(25):1937-1947 Crossref.
  • *[16] G. Lenz, G. Wright, S.S. Dave, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008;359(22):2313-2323 Crossref.
  • [17] S. Monti, K.J. Savage, J.L. Kutok, et al. Molecular profiling of diffuse large B-cell lymphoma identifies robust subtypes including one characterized by host inflammatory response. Blood. 2005;105(5):1851-1861 Crossref.
  • *[18] A. Rosenwald, G. Wright, K. Leroy, et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J Exp Med. 2003;198(6):851-862 Crossref.
  • [19] K.J. Savage, S. Monti, J.L. Kutok, et al. The molecular signature of mediastinal large B-cell lymphoma differs from that of other diffuse large B-cell lymphomas and shares features with classical Hodgkin lymphoma. Blood. 2003;102(12):3871-3879 Crossref.
  • [20] A. Fujii, K. Ohshima, M. Hamasaki, et al. Differential expression of chemokines, chemokine receptors, cytokines and cytokine receptors in diffuse large B cell malignant lymphoma. Int J Oncol. 2004;24(3):529-538
  • [21] J.J. Hoefnagel, R. Dijkman, K. Basso, et al. Distinct types of primary cutaneous large B-cell lymphoma identified by gene expression profiling. Blood. 2005;105(9):3671-3678 Crossref.
  • [22] T. Kobayashi, M. Yamaguchi, S. Kim, et al. Microarray reveals differences in both tumors and vascular specific gene expression in de novo CD5+ and CD5- diffuse large B-cell lymphomas. Cancer Res. 2003;63(1):60-66
  • [23] U. Klein, A. Gloghini, G. Gaidano, et al. Gene expression profile analysis of AIDS-related primary effusion lymphoma (PEL) suggests a plasmablastic derivation and identifies PEL-specific transcripts. Blood. 2003;101(10):4115-4121 Crossref.
  • [24] W. Fan, D. Bubman, A. Chadburn, et al. Distinct subsets of primary effusion lymphoma can be identified based on their cellular gene expression profile and viral association. J Virol. 2005;79(2):1244-1251 Crossref.
  • *[25] A. Rosenwald, G. Wright, A. Wiestner, et al. The proliferation gene expression signature is a quantitative integrator of oncogenic events that predicts survival in mantle cell lymphoma. Cancer Cell. 2003;3(2):185-197 Crossref.
  • [26] K. Fu, D.D. Weisenburger, T.C. Greiner, et al. Cyclin D1-negative mantle cell lymphoma: a clinicopathologic study based on gene expression profiling. Blood. 2005;106(13):4315-4321 Crossref.
  • [27] I. Wlodarska, D. Dierickx, V. Vanhentenrijk, et al. Translocations targeting CCND2, CCND3, and MYCN do occur in t(11;14)-negative mantle cell lymphomas. Blood. 2008;111(12):5683-5690 Crossref.
  • *[28] S.S. Dave, G. Wright, B. Tan, et al. Prediction of survival in follicular lymphoma based on molecular features of tumor-infiltrating immune cells. N Engl J Med. 2004;351(21):2159-2169 Crossref.
  • [29] F. d'Amore, E. Chan, J. Iqbal, et al. Clonal evolution in t(14;18)-positive follicular lymphoma, evidence for multiple common pathways, and frequent parallel clonal evolution. Clin Cancer Res. 2008;14(22):7180-7187 Crossref.
  • [30] T. Yano, E.S. Jaffe, D.L. Longo, et al. MYC rearrangements in histologically progressed follicular lymphomas. Blood. 1992;80(3):758-767
  • [31] C.A. Sander, T. Yano, H.M. Clark, et al. p53 mutation is associated with progression in follicular lymphomas. Blood. 1993;82(7):1994-2004
  • [32] A. Matolcsy, P. Casali, R.A. Warnke, et al. Morphologic transformation of follicular lymphoma is associated with somatic mutation of the translocated Bcl-2 gene. Blood. 1996;88(10):3937-3944
  • [33] K.S. Elenitoba-Johnson, R.D. Gascoyne, M.S. Lim, et al. Homozygous deletions at chromosome 9p21 involving p16 and p15 are associated with histologic progression in follicle center lymphoma. Blood. 1998;91(12):4677-4685
  • [34] M. Pinyol, F. Cobo, S. Bea, et al. p16 (INK4a) gene inactivation by deletions, mutations, and hypermethylation is associated with transformed and aggressive variants of non-Hodgkin's lymphomas. Blood. 1998;91(8):2977-2984
  • [35] I.S. Lossos, A.A. Alizadeh, M. Diehn, et al. Transformation of follicular lymphoma to diffuse large-cell lymphoma: alternative patterns with increased or decreased expression of c-myc and its regulated genes. Proc Natl Acad Sci U S A. 2002;99(13):8886-8891 Crossref.
  • [36] J.A. Martinez-Climent, A.A. Alizadeh, R. Segraves, et al. Transformation of follicular lymphoma to diffuse large cell lymphoma is associated with a heterogeneous set of DNA copy number and gene expression alterations. Blood. 2003;101(8):3109-3117 Crossref.
  • [37] K.S. Elenitoba-Johnson, S.D. Jenson, R.T. Abbott, et al. Involvement of multiple signaling pathways in follicular lymphoma transformation: p38-mitogen-activated protein kinase as a target for therapy. Proc Natl Acad Sci U S A. 2003;100(12):7259-7264 Crossref.
  • [38] C. Thieblemont, V. Nasser, P. Felman, et al. Small lymphocytic lymphoma, marginal zone B-cell lymphoma, and mantle cell lymphoma exhibit distinct gene-expression profiles allowing molecular diagnosis. Blood. 2004;103(7):2727-2737 Crossref.
  • [39] A clinical evaluation of the International Lymphoma Study Group classification of non-Hodgkin's lymphoma. The Non-Hodgkin's Lymphoma Classification Project. Blood. 1997;89(11):3909-3918
  • [40] T. Rudiger, D.D. Weisenburger, J.R. Anderson, et al. Peripheral T-cell lymphoma (excluding anaplastic large-cell lymphoma): results from the Non-Hodgkin's Lymphoma Classification Project. Ann Oncol. 2002;13(1):140-149 Crossref.
  • [41] M. Reiser, A. Josting, M. Soltani, et al. T-cell non-Hodgkin's lymphoma in adults: clinicopathological characteristics, response to treatment and prognostic factors. Leuk Lymphoma. 2002;43(4):805-811 Crossref.
  • [42] J.M. Vose. International Peripheral T-cell Lymphoma (PTCL) Clinical and Pathology Review project: poor outcome by prognostic indices and lack of efficacy with anthracyclines. Blood. 2005;106:811a
  • *[43] International peripheral T-cell and natural killer/t-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol. 2008;26(25):4124-4130
  • *[44] L. de Leval, D.S. Rickman, C. Thielen, et al. The gene expression profile of nodal peripheral T-cell lymphoma demonstrates a molecular link between angioimmunoblastic T-cell lymphoma (AITL) and follicular helper T (TFH) cells. Blood. 2007;109(11):4952-4963 Crossref.
  • [45] B. Martinez-Delgado. Peripheral T-cell lymphoma gene expression profiles. Hematol Oncol. 2006;24(3):113-119 Crossref.
  • [46] P.P. Piccaluga, C. Agostinelli, A. Califano, et al. Gene expression analysis of angioimmunoblastic lymphoma indicates derivation from T follicular helper cells and vascular endothelial growth factor deregulation. Cancer Res. 2007;67(22):10703-10710 Crossref.
  • [47] P.P. Piccaluga, C. Agostinelli, A. Califano, et al. Gene expression analysis of peripheral T cell lymphoma, unspecified, reveals distinct profiles and new potential therapeutic targets. J Clin Invest. 2007;117(3):823-834 Crossref.
  • [48] D. Yu, A.H. Tan, X. Hu, et al. Roquin represses autoimmunity by limiting inducible T-cell co-stimulator messenger RNA. Nature. 2007;450(7167):299-303
  • [49] L.M. Weiss, E.S. Jaffe, X.F. Liu, et al. Detection and localization of Epstein-Barr viral genomes in angioimmunoblastic lymphadenopathy and angioimmunoblastic lymphadenopathy-like lymphoma. Blood. 1992;79(7):1789-1795
  • [50] S. Rutella, S. Danese, G. Leone. Tolerogenic dendritic cells: cytokine modulation comes of age. Blood. 2006;108(5):1435-1440 Crossref.
  • [51] P. Hwu, M.X. Du, R. Lapointe, et al. Indoleamine 2,3-dioxygenase production by human dendritic cells results in the inhibition of T cell proliferation. J Immunol. 2000;164(7):3596-3599
  • [52] R. Kuhn, J. Lohler, D. Rennick, et al. Interleukin-10-deficient mice develop chronic enterocolitis. Cell. 1993;75(2):263-274 Crossref.
  • [53] M.S. Sundrud, A. Rao. New twists of T cell fate: control of T cell activation and tolerance by TGF-beta and NFAT. Curr Opin Immunol. 2007;19(3):287-293 Crossref.
  • [54] J. Dupuis, K. Boye, N. Martin, et al. Expression of CXCL13 by neoplastic cells in angioimmunoblastic T-cell lymphoma (AITL): a new diagnostic marker providing evidence that AITL derives from follicular helper T cells. Am J Surg Pathol. 2006;30(4):490-494 Crossref.
  • [55] B. Martinez-Delgado, M. Cuadros, E. Honrado, et al. Differential expression of NF-kappaB pathway genes among peripheral T-cell lymphomas. Leukemia. 2005;19(12):2254-2263
  • [56] S.W. Morris, M.N. Kirstein, M.B. Valentine, et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 1994;263(5151):1281-1284 New York, NY
  • [57] R. Rimokh, J.P. Magaud, F. Berger, et al. A translocation involving a specific breakpoint (q35) on chromosome 5 is characteristic of anaplastic large cell lymphoma (‘Ki-1 lymphoma’). Br J Haematol. 1989;71(1):31-36 Crossref.
  • [58] S.W. Morris, M.N. Kirstein, M.B. Valentine, et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 1995;267(5196):316-317 New York, NY Crossref.
  • [59] S.W. Morris, L. Xue, Z. Ma, et al. Alk+ CD30+ lymphomas: a distinct molecular genetic subtype of non-Hodgkin's lymphoma. Br J Haematol. 2001;113(2):275-295 Crossref.
  • [60] A. Slupianek, M. Nieborowska-Skorska, G. Hoser, et al. Role of phosphatidylinositol 3-kinase-Akt pathway in nucleophosmin/anaplastic lymphoma kinase-mediated lymphomagenesis. Cancer Res. 2001;61(5):2194-2199
  • [61] Q. Zhang, H.Y. Wang, X. Liu, et al. STAT5A is epigenetically silenced by the tyrosine kinase NPM1-ALK and acts as a tumor suppressor by reciprocally inhibiting NPM1-ALK expression. Nat Med. 2007;13(11):1341-1348 Crossref.
  • [62] R. Piva, E. Pellegrino, M. Mattioli, et al. Functional validation of the anaplastic lymphoma kinase signature identifies CEBPB and BCL2A1 as critical target genes. J Clin Invest. 2006;116(12):3171-3182 Crossref.
  • [63] S.D. Bohling, S.D. Jenson, D.K. Crockett, et al. Analysis of gene expression profile of TPM3-ALK positive anaplastic large cell lymphoma reveals overlapping and unique patterns with that of NPM-ALK positive anaplastic large cell lymphoma. Leuk Res. 2008;32(3):383-393 Crossref.
  • [64] T. Gaiser, C. Thorns, H. Merz, et al. Gene profiling in anaplastic large-cell lymphoma-derived cell lines with cDNA expression arrays. J Hematother Stem Cell Res. 2002;11(2):423-428 Crossref.
  • [65] H.D. Foss, I. Anagnostopoulos, I. Araujo, et al. Anaplastic large-cell lymphomas of T-cell and null-cell phenotype express cytotoxic molecules. Blood. 1996;88(10):4005-4011
  • [66] I. Bonzheim, E. Geissinger, S. Roth, et al. Anaplastic large cell lymphomas lack the expression of T-cell receptor molecules or molecules of proximal T-cell receptor signaling. Blood. 2004;104(10):3358-3360 Crossref.
  • [67] I. Salaverria, S. Bea, A. Lopez-Guillermo, et al. Genomic profiling reveals different genetic aberrations in systemic ALK-positive and ALK-negative anaplastic large cell lymphomas. Br J Haematol. 2008;140(5):516-526 Crossref.
  • [68] F.A. Proietti, A.B. Carneiro-Proietti, B.C. Catalan-Soares, et al. Global epidemiology of HTLV-I infection and associated diseases. Oncogene. 2005;24(39):6058-6068 Crossref.
  • [69] G. Franchini, C. Nicot, J.M. Johnson. Seizing of T cells by human T-cell leukemia/lymphoma virus type 1. Adv Cancer Res. 2003;89:69-132 Crossref.
  • [70] I. Azran, Y. Schavinsky-Khrapunsky, M. Aboud. Role of Tax protein in human T-cell leukemia virus type-I leukemogenicity. Retrovirology. 2004;1:20 Crossref.
  • [71] M.L. Gatza, J.C. Watt, S.J. Marriott. Cellular transformation by the HTLV-I Tax protein, a jack-of-all-trades. Oncogene. 2003;22(33):5141-5149 Crossref.
  • [72] H. Sasaki, I. Nishikata, T. Shiraga, et al. Overexpression of a cell adhesion molecule, TSLC1, as a possible molecular marker for acute-type adult T-cell leukemia. Blood. 2005;105(3):1204-1213
  • [73] C.A. Pise-Masison, M. Radonovich, K. Dohoney, et al. Gene expression profiling of ATL patients: compilation of disease related genes and evidence for TCF-4 involvement in BIRC5 gene expression and cell viability. Blood. 2009;
  • [74] B. Ballester, O. Ramuz, C. Gisselbrecht, et al. Gene expression profiling identifies molecular subgroups among nodal peripheral T-cell lymphomas. Oncogene. 2006;25(10):1560-1570 Crossref.
  • [75] M. Nelson, D.E. Horsman, D.D. Weisenburger, et al. Cytogenetic abnormalities and clinical correlations in peripheral T-cell lymphoma. Br J Haematol. 2008;141(4):461-469 Crossref.
  • [76] M. Cuadros, S.S. Dave, E.S. Jaffe, et al. Identification of a proliferation signature related to survival in nodal peripheral T-cell lymphomas. J Clin Oncol. 2007;25(22):3321-3329 Crossref.
  • [77] T. Haque, D. Faury, S. Albrecht, et al. Gene expression profiling from formalin-fixed paraffin-embedded tumors of pediatric glioblastoma. Clin Cancer Res. 2007;13(21):6284-6292 Crossref.
  • [78] J.P. Maciejewski, G.J. Mufti. Whole genome scanning as a cytogenetic tool in hematologic malignancies. Blood. 2008;112(4):965-974 Crossref.
  • [79] O.P. Kallioniemi, A. Kallioniemi, D. Sudar, et al. Comparative genomic hybridization: a rapid new method for detecting and mapping DNA amplification in tumors. Semin Cancer Biol. 1993;4(1):41-46
  • [80] D. Pinkel, D.G. Albertson. Array comparative genomic hybridization and its applications in cancer. Nat Genet. 2005;(37 Suppl):S11-S17 Crossref.
  • [81] A.S. Ishkanian, C.A. Malloff, S.K. Watson, et al. A tiling resolution DNA microarray with complete coverage of the human genome. Nat Genet. 2004;36(3):299-303 Crossref.
  • [82] Affymetrix I. High-resolution chromosome copy number analysis using GeneChip mapping arrays. Affymetrix Product Family GeneChip Chromosome Copy Number Analysis Tool 2005.
  • [83] J.C. Strefford, H. Worley, K. Barber, et al. Genome complexity in acute lymphoblastic leukemia is revealed by array-based comparative genomic hybridization. Oncogene. 2007;26(29):4306-4318 Crossref.
  • [84] G. Lenz, G.W. Wright, N.C. Emre, et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc Natl Acad Sci U S A. 2008;105(36):13520-13525 Crossref.
  • [85] Iqbal J, Can Kucuk, DeLeeuw RJ, et al. Genomic analyses reveal global functional alterations that promote tumor growth and novel tumor suppressor genes in natural killer-cell malignancies. Leukemia, in press [online publication 5 February 2009, doi:10.1038/leu.2009.3].
  • [86] A. Eulalio, E. Huntzinger, E. Izaurralde. Getting to the root of miRNA-mediated gene silencing. Cell. 2008;132(1):9-14 Crossref.
  • [87] R. Garzon, C.M. Croce. MicroRNAs in normal and malignant hematopoiesis. Curr Opin Hematol. 2008;15(4):352-358 Crossref.
  • [88] R. Malumbres, K.A. Sarosiek, E. Cubedo, et al. Differentiation-stage-specific expression of microRNAs in B-lymphocytes and diffuse large B-cell lymphomas. Blood. 2008;
  • [89] J. Zhang, D.D. Jima, C. Jacobs, et al. Patterns of microRNA expression characterize stages of human B cell differentiation. Blood. 2009;
  • [90] C.H. Lawrie, N.J. Saunders, S. Soneji, et al. MicroRNA expression in lymphocyte development and malignancy. Leukemia. 2008;22(7):1440-1446 Crossref.
  • [91] G.A. Calin, C. Sevignani, C.D. Dumitru, et al. Human microRNA genes are frequently located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci U S A. 2004;101(9):2999-3004 Crossref.
  • [92] W. Tam, D. Ben-Yehuda, W.S. Hayward. bic, a novel gene activated by proviral insertions in avian leukosis virus-induced lymphomas, is likely to function through its noncoding RNA. Mol Cell Biol. 1997;17(3):1490-1502
  • [93] P.S. Eis, W. Tam, L. Sun, et al. Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc Natl Acad Sci U S A. 2005;102(10):3627-3632 Crossref.
  • [94] J. Kluiver, E. Haralambieva, D. de Jong, et al. Lack of BIC and microRNA miR-155 expression in primary cases of Burkitt lymphoma. Genes Chromosomes Cancer. 2006;45(2):147-153 Crossref.
  • [95] G.A. Calin, M. Ferracin, A. Cimmino, et al. A MicroRNA signature associated with prognosis and progression in chronic lymphocytic leukemia. N Engl J Med. 2005;353(17):1793-1801 Crossref.
  • [96] L. Pasqualucci, A. Migliazza, N. Fracchiolla, et al. BCL-6 mutations in normal germinal center B cells: evidence of somatic hypermutation acting outside Ig loci. Proc Natl Acad Sci U S A. 1998;95(20):11816-11821 Crossref.
  • [97] K.H. Young, D.D. Weisenburger, B.J. Dave, et al. Mutations in the DNA-binding codons of TP53, which are associated with decreased expression of TRAILreceptor-2, predict for poor survival in diffuse large B-cell lymphoma. Blood. 2007;110(13):4396-4405 Crossref.
  • [98] W. Tam, M. Gomez, A. Chadburn, et al. Mutational analysis of PRDM1 indicates a tumor-suppressor role in diffuse large B-cell lymphomas. Blood. 2006;107(10):4090-4100 Crossref.
  • [99] L. Pasqualucci, M. Compagno, J. Houldsworth, et al. Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma. J Exp Med. 2006;203(2):311-317 Crossref.
  • *[100] G. Lenz, R.E. Davis, V.N. Ngo, et al. Oncogenic CARD11 mutations in human diffuse large B cell lymphoma. Science. 2008;319(5870):1676-1679 New York, NY Crossref.
  • [101] U. Novak, A. Rinaldi, I. Kwee, et al. The NF-{kappa}B negative regulator TNFAIP3 (A20) is inactivated by somatic mutations and genomic deletions in marginal zone B-cell lymphomas. Blood. 2009;
  • [102] Q. Zhang, R. Siebert, M. Yan, et al. Inactivating mutations and overexpression of BCL10, a caspase recruitment domain-containing gene, in MALT lymphoma with t(1;14) (p22;q32). Nat Genet. 1999;22(1):63-68
  • [103] E.R. Mardis. The impact of next-generation sequencing technology on genetics. Trends Genet. 2008;24(3):133-141 Crossref.
  • [104] J.I. Martin-Subero, M. Kreuz, M. Bibikova, et al. New insights into the biology and origin of mature aggressive B-cell lymphomas by combined epigenomic, genomic and transcriptional profiling. Blood. 2009;113(11):2488-2497 Crossref.
  • [105] M. Ehrlich. Cancer-linked DNA hypomethylation and its relationship to hypermethylation. Curr Top Microbiol Immunol. 2006;310:251-274 Crossref.
  • [106] J. Boultwood, J.S. Wainscoat. Gene silencing by DNA methylation in haematological malignancies. Br J Haematol. 2007;138(1):3-11 Crossref.
  • [107] B.L. Pike, T.C. Greiner, X. Wang, et al. DNA methylation profiles in diffuse large B-cell lymphoma and their relationship to gene expression status. Leukemia. 2008;22(5):1035-1043 Crossref.
  • [108] R.E. Davis, K.D. Brown, U. Siebenlist, et al. Constitutive nuclear factor kappaB activity is required for survival of activated B cell-like diffuse large B cell lymphoma cells. J Exp Med. 2001;194(12):1861-1874 Crossref.
  • *[109] A.H. Bild, G. Yao, J.T. Chang, et al. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 2006;439(7074):353-357 Crossref.
  • [110] L.C. Cerchietti, S.N. Yang, R. Shaknovich, et al. A peptomimetic inhibitor of BCL6 with potent anti-lymphoma effects in vitro and in vivo. Blood. 2008;
  • [111] P.L. Zinzani, M. Tani, G. Musuraca, et al. Phase II study of proteasome inhibitor Bortezomib (Velcade®) in patients with relapsed/refractory T-cell lymphoma: preliminary results. Blood. 2006;:108a
  • [112] P.P. Piccaluga, C. Agostinelli, P.L. Zinzani, et al. Expression of platelet-derived growth factor receptor alpha in peripheral T-cell lymphoma not otherwise specified. Lancet Oncol. 2005;6(6):440 Crossref.
  • [113] R. Malumbres, J. Chen, R. Tibshirani, et al. Paraffin-based 6-gene model predicts outcome in diffuse large B-cell lymphoma patients treated with R-CHOP. Blood. 2008;111(12):5509-5514 Crossref.
  • [114] R.A. Roberts, C.M. Sabalos, M.L. LeBlanc, et al. Quantitative nuclease protection assay in paraffin-embedded tissue replicates prognostic microarray gene expression in diffuse large-B-cell lymphoma. Lab Invest. 2007;87(10):979-997 Crossref.
  • *[115] C.P. Hans, D.D. Weisenburger, T.C. Greiner, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275-282 Crossref.


Departments of Pathology and Microbiology, Center for Lymphoma and Leukemia Research, University of Nebraska Medical Center, Omaha, NE 68198-3135, USA

Corresponding author. Amelia and Austin Vickery professor of Pathology, Co-director – Center for Lymphoma and Leukemia Research, Department of Pathology and Microbiology University of Nebraska Medical Center, Omaha, NE, USA. Tel.: +1 402 559 7684; Fax: +1 402 559 6018.