Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Cytogenetics and molecular genetics

Distinct genetic evolution patterns of relapsing diffuse large B-cell lymphoma revealed by genome-wide copy number aberration and targeted sequencing analysis

Abstract

Recurrences of diffuse large B-cell lymphomas (DLBCL) result in significant morbidity and mortality, but their underlying genetic and biological mechanisms are unclear. Clonal relationship in DLBCL relapses so far is mostly addressed by the investigation of immunoglobulin (IG) rearrangements, therefore, lacking deeper insights into genome-wide lymphoma evolution. We studied mutations and copy number aberrations in 20 paired relapsing and 20 non-relapsing DLBCL cases aiming to test the clonal relationship between primaries and relapses to track tumors’ genetic evolution and to investigate the genetic background of DLBCL recurrence. Three clonally unrelated DLBCL relapses were identified (15%). Also, two distinct patterns of genetic evolution in clonally related relapses were detected as follows: (1) early-divergent/branching evolution from a common progenitor in 6 patients (30%), and (2) late-divergent/linear progression of relapses in 11 patients (65%). Analysis of recurrent genetic events identified potential early drivers of lymphomagenesis (KMT2D, MYD88, CD79B and PIM1). The most frequent relapse-specific events were additional mutations in KMT2D and alterations of MEF2B. SOCS1 mutations were exclusive to non-relapsing DLBCL, whereas primaries of relapsing DLBCL more commonly displayed gains of 10p15.3–p12.1 containing the potential oncogenes PRKCQ, GATA3, MLLT10 and ABI1. Altogether, our study expands the knowledge on clonal relationship, genetic evolution and mutational basis of DLBCL relapses.

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoid neoplasm worldwide. Although majority of patients are cured by existing therapeutic regimens, still about 30–40% recur or are refractory to standard treatment. A DLBCL relapse is defined as appearance of new lesions after achieving complete remission for at least 3 months, and is linked to significantly increased morbidity and mortality.1 Standard treatment of relapsed patients consists of high-dose chemotherapy with autologous stem cell transplantation,2 and the 3-year overall survival is only about 50%.3

Relapses of lymphoid neoplasms, even after long-lasting clinical remission, are generally regarded as direct outgrowths of the primary tumor,4, 5, 6 although rare clonally unrelated relapses have been documented.7, 8, 9, 10 Such unrelated recurrences point to an increased general risk for patients to develop a second de novo neoplasm of the same diagnostic category, and raises the question of potential predisposition factors in these types of relapses. Clinically, the choice of the most effective treatment is challenging in such instances, that is, high-dose chemotherapy with autologous stem cell transplantation or another standard primary regimen.4, 7

The gold standard for determination of a clonal relationship is analysis of immunoglobulin (IG) V(D)J gene rearrangements, which are unique for mature B cells and are stably transferred to daughter cells during the process of clonal expansion.11 Clonal relationship can also be established at the genomic level by genome-wide approaches such as array-comparative genomic hybridization (aCGH) and next-generation sequencing (NGS).12, 13, 14

Despite the fact that relapses are a major cause of morbidity and mortality in DLBCL, their genetic mechanisms remain poorly characterized due to the paucity of paired primary/relapse samples. This is illustrated by the fact that only 20 matched DLBCL pairs in total were investigated by 3 recent studies addressing the genetic basis of DLBCL relapses.12, 15, 16 The remaining cohorts in these studies partly or entirely consisted of unpaired relapsed or treatment-refractory DLBCL samples. Despite a modest overlap between the results of these studies, more than one study showed recurrent alterations of B2M, CD58, SOCS1, STAT6, NFKBIE, FOXO1, PIM1 and BCL2 genes suggesting their functional importance. Underlying genetic differences between primaries of relapsing and non-relapsing DLBCL have not been previously addressed at the genome-wide scale.

In this study, we performed a comprehensive characterization of 20 paired (that is, primary and relapse samples) relapsing and 20 non-relapsing DLBCL cases and demonstrated clonally unrelated relapses at the genomic level as well as two distinct genetic evolution patterns in clonally related DLBCL recurrences. We observed recurrent group-specific genetic alterations between relapsing and non-relapsing DLBCL primaries as well as frequently shared mutations between primaries and relapses suggesting that the latter are probably early drivers of lymphomagenesis.

Materials and Methods

Samples

Formalin-fixed, paraffin-embedded (FFPE) and available fresh frozen (FF) lymphoma samples (Table 1) were retrieved from the archive of the Institute of Pathology at the University Hospital Basel, Switzerland. In the relapsing cohort (n=20 pairs) tissues from primaries and respective relapses following complete remission (median period of remission 22 months, mean 42, range 8–141) were included. The non-relapsing cohort (n=20) consisted of well-preserved diagnostic biopsies of patients who had a single occurrence of DLBCL and were relapse-free for at least 4 years (median 78 months, mean 80, range 49–119). A previously reported historic DLBCL cohort was used to evaluate GATA3 overexpression.17 The study was approved by the Ethics Committee of North-Western and Central Switzerland (EKNZ 2014-252).

Table 1 Clinical characteristics of the study cohorts

Immunohistochemistry

Immunohistochemistry was performed on serial tissue sections using an automated immunostainer Benchmark XT (Ventana/Roche, Tucson, AZ, USA) according to manufacturer’s protocols. Details on antibodies and cut-offs used for immunohistochemistry evaluation are summarized in Supplementary Table S2.

DNA extraction

Genomic DNA was extracted only from FFPE and FF tissues with >70% tumor content. Detailed information on that procedure is available in the Supplementary Material.

Assessment of IGH rearrangements and BCL2 translocations

A PCR-based assay for IG V(D)J rearrangements was performed with consensus FR2 and FR3 as well as J primers, as previously published.18 Fragments suggesting clonally unrelated recurrences were cloned into pGEM-T (Promega, Madinson, WI, USA) plasmids and Sanger-sequenced to unequivocally confirm different clonal origin. Sequences were analyzed with the IgBLAST tool.19 A multiplex PCR assay with primers targeting the major and minor breakpoint regions of BCL2 was used for detection of t(14;18).18 Additional information is available in the Supplementary Materials.

Array-comparative genomic hybridization

At least 100 ng of genomic DNA from each sample and 500 ng of commercial 46XX reference gDNA (Promega, Southampton, UK) were used as reported.20 Exact description of this procedure is available in the Supplementary Material. aCGH data were assessed with a series of QC metrics and subsequently analyzed using Agilent Genomic Workbench v.7.0 software with the aberration detection algorithm ADM2.21 Using derivative log2 spread ratio as the primary quality assessment criterion and manual inspection, probe-based aberration frequency data were exported for each sample set (primaries of relapsing DLBCL, relapses and primaries of non-relapsing DLBCL) and further processed by a custom workflow programmed in R software (available upon request).

Targeted next-generation sequencing, variant calling and filtering

A target-enrichment panel was designed to include genes and their hotspots that are most frequently mutated in B-cell lymphoid neoplasms according to the COSMIC database (release v70) and manual overview of the literature. 68 genes were included (Supplementary Table S3). Further information on sequencing procedure is available in the Supplementary Material. Mutation identification was performed by the Variant caller plug-in v4.4 from the Torrent Suite (Thermo Fisher Scientific, Carlsbad, CA, USA) using low stringency parameters, listed in Supplementary Table S4. Depth of coverage, coverage uniformity and number of variants called are summarized in Supplementary Table S1. At the initial quality filtering, variants with Phred-based quality score <50, strand bias >0.75 and number of variant-supporting reads <10 were discarded from the analysis. Depending on the analysis application, different filtering workflows were used at the final variant filtering stage as depicted in Figure 1. Only exonic non-synonymous variants were considered for the identification of recurrent mutations, while synonymous and non-exonic variants were additionally included for tumor evolution analysis. Pairs containing samples (five primaries and two relapses (Supplementary Table S1)) with suboptimal sequencing quality due to poor coverage uniformity (<70%) or exceptionally high numbers of somatic variants detected (more than two s.d.) were excluded from the evolutionary analysis. Only variants that were shared between such samples and their pair were used in the recurrent mutation analysis.

Figure 1
figure1

Variant exclusion criteria for different analysis applications. Synonymous mutations and variants located in non-exonic regions were used in evolutionary analysis, but they were excluded from recurrent mutation analysis. Other filtering steps were designed to exclude germline variants and possible false-positives in samples with suboptimal sequencing quality (see also Supplementary Table S1). VAF, variant allele frequency.

Evolutionary analysis

Lack of relationship was established when paired samples contained distinct IG chain rearrangements and/or no common chromosomal aberrations with identical breakpoints in all 22 autosomes and the X chromosome. Clonal relationship was established when paired samples contained the same IG rearrangement and at least one chromosomal aberration with an identical breakpoint. Presence of shared mutations was not considered evidence of clonal relationship, but mutations were used for evolutionary distance estimation. Tumor evolution via a common progenitor was assumed in clonally related cases when at least one homozygous or heterozygous deletion with a unique breakpoint was detected in the primary tumor, but was not present in the paired relapse sample. Detailed information on the calculation of evolutionary distances between tumor occurrences is available in the Supplementary Materials.

Statistical analysis

Information on the applied statistical methods is available in the Supplementary Materials.

Results

Relapsing DLBCL show more advanced tumor stages and stronger BCL2 expression

Available clinical and pathological characteristics of the patient cohorts are summarized in Table 1, Figure 2, Supplementary Table S5 and Supplementary Figure S1. Therapy information was retrievable for 36 of 40 patients. First line therapy with curative intent was administered in all but 1 patient and consisted of CHOP (n=2) or R-CHOP (n=23) with (n=4) or without irradiation (n=32). Depending on risk factors, intrathecal chemotherapy was administered in 10 patients (5 of them relapsed and 5 did not). Median age at initial diagnosis was 67 years for the relapsing DLBCL cohort and 68 years for the non-relapsing cohort; patients in the relapsing DLBCL cohort presented with more advanced disease stages at initial diagnosis (P=0.029).

Figure 2
figure2

Chronological overview of the relapsing DLBCL cohort. Cell of origin is represented by symbol shape and clonal relationship by color. Numbers centering the symbols represent respective C-MYC and BCL2 protein co-expression double-hit scores (immunohistochemistry). Only one case (case 8) had t(14;18) affecting BCL2 as determined by multiplex PCR assay. Stacked symbols (cases 1 and 5) represent two parts of the same tumor profiled by aCGH where intratumoral heterogeneity was detected (Supplementary Figure S2). At the right side, sizes of the amplified IGH rearrangement products are indicated for clonally unrelated cases. NA, not assessed; NP, not processed.

Of the classifiable samples (n=18) in the relapsing DLBCL cohort, 12 primary tumors were of non-germinal center B cell-like- (non-GCB) and 6 of GCB origin (Figure 2, Supplementary Table S5). In the non-relapsing DLBCL cohort, 18 of 20 cases could be classified—9 were of GCB origin and 9 of non-GCB (Supplementary Table S5; also see supplementary results for information on changes of cell of origin, and Supplementary Table S2 for the cutoffs used to evaluate immunohistochemical staining).

As evaluated by immunohistochemistry, primaries of relapsing DLBCL expressed higher levels of BCL2 than non-relapsing ones (mean positivity 74±29%, median 80% vs 46±31%, median 40%, P=0.006); when positivity was evaluated applying the cutoff value of 70%,22 this difference remained significant (14/19 positive cases in the relapsing cohort vs 6/19 in the non-relapsing, P=0.011). No significant differences in C-MYC, BCL6 or MIB1 protein expression of were observed between the cohorts (Supplementary Table S5).

Analysis of primary and relapsed DLBCL pairs

Analysis of genetic data within the individual pairs of primary and relapsed DLBCL identified clonally unrelated recurrences and two different genetic evolution patterns among the clonally related DLBCL relapses (Figure 2). Interestingly, intratumoral heterogeneity was evident in multiple instances at both primary and relapse stages (Supplementary Figure S2).

Clonally unrelated DLBCL relapses

Clonally unrelated relapses were detected in three patients (cases 1, 15 and 17). Primary and relapse tumors in these pairs showed no shared chromosomal copy number aberrations (CNA). Even when certain chromosomal loci were aberrant in both samples, breakpoint regions never matched (Figure 3a). Targeted NGS revealed no shared single-nucleotide mutations in cases 1 and 17, but two shared single-nucleotide variants in EP300 and CD79A in case 15. As expected, from the evolutionary point of view, clonally unrelated tumors showed maximal dissimilarity (Figure 4a). Lack of clonal relationship in cases 1 and 15 was additionally confirmed by fragment size and sequence analysis of the rearranged IG genes (Figure 3b and Supplementary Figure S3). However, attempts to amplify IG rearrangement regions in sample 17 were unsuccessful, likely due to heterozygous deletions of the IGH locus in both primary and relapse tumors and the deletion of the IGK locus in the primary tumor. Compared with other relapses (see later), clonally unrelated relapses tend to occur later after initial lymphoma presentation. Times to relapse were 103, 32 and 81 months in cases 1, 15 and 17, respectively.

Figure 3
figure3

Different modes of DLBCL relapse. (a) A genome overview plot of a clonally unrelated pair (case 1) is shown. Genome position is represented on the abscissa and the log2 ratio on the ordinate. Moving average of the primary (red) and the relapse (blue) is plotted. Regions above and below the log2 ratio value of 0 represent gain or loss of DNA, respectively. None of the aberrations match between primary and relapse tumors. For example, chromosome 3 is highly aberrant (magnified, color tint represent the start and end of the aberrant region) in both cases, but breakpoints of gains and losses do not match. (b) Amplified FR3 and FR2 regions of IGH V(D)J rearrangements show distinct sizes between both DLBCL occurrences and FR2 sequences align to different families of IG variable genes. (c) An example of a clonally related DLBCL pair (case 8), in which primary and relapse tumors emerged through early-divergent/branching evolution from a common progenitor. Both tumors share some chromosomal aberrations (chromosomes 6, 13, 21) confirming their shared origin, but aberrations unique to the primary tumor (green arrows) suggest early separation and independent evolution of the tumor-initiating clones. (d) An example of a clonally related DLBCL pair (case 4), in which the relapse emerged through late-divergent/linear evolution from a common progenitor but gained some additional chromosomal aberrations.

Figure 4
figure4

Genetic evolution patterns in relapsing DLBCL. (a) Schematic representation of clonally unrelated DNA relapses showing few or lack of unifying genetic alterations and a maximal genetic distance one from another. (b) Primary tumors belonging to the early-divergent/branching pattern of DLBCL recurrences have high numbers of private genetic alterations and, therefore, large genetic distances from the putative common progenitor. In these cases, relapses are genetically less distant to the common progenitor than primaries. (c) In the late-divergent evolutionary pattern, primary tumors are genetically very close to the putative common progenitor, while relapses show higher divergence. Numbers inside circles represent a combined count of mutations and CNA in the respective population. Circle sizes are scaled according to the number of genetic alterations. Dashed purple area symbolizes the time from putative divergence of populations to occurrence of primary lymphoma, which is unknown. Red, genetic alterations unique to the primary tumor. Blue, genetic alterations unique to the relapse. Gray circle, putative common progenitor. The genetic distance from the common progenitor is plotted on the y axis; at the top of the x axis for the primary tumor, at the bottom for the relapse.

Two patterns of clonally related DLBCL relapse

The first genetic evolutionary pattern within the group of clonally related relapses, early-divergent/branching evolution from a common progenitor, was detected in six cases. In addition to shared CNA with identical breakpoints between primary and relapse, primary tumors had homozygous and heterozygous deletions, which were not present in the subsequent relapse samples (Figure 3c) indicating evolutionary divergence of primary and relapse tumor clones. Taking into account also mutations detected by targeted NGS, most (5/6) of the primary tumors characterized by this pattern were more genetically distant from the putative common progenitor than relapses (Figure 4b and Supplementary Figure S4A), suggesting that the divergence occurred early. Clonal relationships in all six cases were confirmed by detection of identical IGH rearrangements and presence of shared mutations (median number of shared mutations, 5; range 1–23).

The second and the most frequent relapse pattern (detected in 11 cases), late-divergent/linear evolution with or without progression, most closely reflects the current concept of DLBCL recurrence. In cases characterized by this pattern, primary tumors had no private CNA compared to relapses (Figure 3d), yet a limited degree of divergence was evidenced by small numbers of private mutations in the primary tumors. Nevertheless, these private mutations yielded only a minimal genetic distance between the primary tumor and a putative common progenitor population, which was in all cases smaller than the genetic distance between the common progenitor and relapse (Figure 4c and Supplementary Figure S4B). At relapse, the majority of cases (9/11) showed increased numbers of genetic alterations. Time to relapse was shorter in the group of late-divergent cases compared to cases of early-divergent/branching evolutionary pattern (median 17 months vs 54 months, respectively, P=0.018, Mann-Whitney test).

Recurrent genetic alterations in primaries of relapsing and non-relapsing DLBCL

The most frequent CNA in primaries of both relapsing and non-relapsing DLBCL were heterozygous deletions of chromosome 6q (affecting PRDM1) in 18 of 40 (45%) cases as well as gains of large parts of chromosome 18 in 16 of 40 (40%) cases. The most frequently mutated gene overall was KMT2D, which harbored mutations in 10/16 (63%) primaries of clonally related relapsing DLBCL, 1/3 (33%) primaries of clonally unrelated relapsing DLBCL and 8 of 20 (40%) non-relapsing DLBCL cases; mutations were scattered across many exons of this gene without hotspot. Other frequently mutated genes were CD79B, BTG1 and PIM1 (Figure 5 and Supplementary Table S6).

Figure 5
figure5

Non-synonymous variants in the studied DLBCL cases. A heatmap plot of all non-synonymous variants detected by targeted next-generation sequencing in DLBCL cases. Genes are ordered left-to-right according to the decreasing number of mutational frequency. Cases are ordered in groups according to the assigned type of relapse evolution and also according to cell-of-origin classification. Samples in the relapsing DLBCL cohort, which were evaluated only for shared mutations due to suboptimal sequencing quality are marked with asterisk (*). Variable number of mutations per gene are represented by changing color intensity (see color legend).

Losses of 6q22.31-q22.33 and gains of chromosome 7 were the most frequent CNA occurring early in the progenitor cell population before divergence in six clonally related DLBCL cases with early-divergent/branching evolutionary pattern. Mutations of KMT2D (7/16), CD79B (5/16), MYD88 (4/16) and PIM1 (4/16) were frequently shared in clonally related tumor pairs, suggesting that they occurred early and thus were potential drivers of lymphomagenesis. Of note, mutations in the latter three genes were restricted to non-GCB-DLBCL and were concurrent in 3 tumors.

Most of the tumors acquired several chromosomal gains and losses at relapse, but no recurrent CNA could be identified. Additional mutations in KMT2D were the most frequent relapse-specific sequence alteration, occurring in 7 of 16 (44%) evaluated relapses. The second most frequently mutated gene at relapse was MEF2B, affecting 3 of 16 cases (19%).

At the nucleotide level SOCS1 was found to be mutated in 5 of 20 (25%) non-relapsing DLBCL cases, but was not affected in any of the relapsing DLBCL (P=0.047). In contrast, BCL2 was found exceptionally mutated in primaries of relapsing DLBCL (4/19, 21%) and not in non-relapsing DLBCL (P=0.047). Three of these mutations were scored neutral by varint effect prediction algorithms and the respective cases had moderate-to-high BCL2 protein levels, while in one case (case 7) the detetcted BCL2 gene mutation was predicted to be deleterious by all applied algorithms and this case lacked BCL2 expression (Supplementary Table S7).

Comparison of CNA between primaries of relapsing DLBCL and non-relapsing DLBCL identified several differentially aberrant regions (Figure 6 and Table 2). Most significantly, relapsing tumors more frequently displayed gains of 10p15.3–p12.1. The affected region contains 130 protein-coding genes, among them 4 oncogenes with lymphomagenic potential: GATA binding protein 3 (GATA3), protein kinase C-theta (PRKCQ), mixed-lineage leukemia translocated to 10 (MLLT10 encoding transcription factor AF-10) and Abl-interactor 1 (ABI1). Due to immediate availability of anti-GATA3 antibody we tested the prognostic significance of GATA3 overexpression in 23 DLBCL cases with sufficient material from this study and in a well-characterized historic collective consisting of 250 cases.17 Although a fraction of GATA3-positive DLBCL samples was identified, its overexpression did not correlate neither with 10p15.3-12.1 gains nor with worse survival of the patients (see Supplementary Results and Supplementary Figure S5 and S6 for more detailed information). Reliable anti-PRKCQ, anti-MLLT10 and anti-ABI1 antibodies for immunohistochemistry were not available for us and the significance of these genes remains to be validated.

Figure 6
figure6

Differential comparison of chromosomal copy number aberrations between primary relapsing (n=20) and non-relapsing (n=20) DLBCL. The upper part of each graph plots the frequency of aberrations (%, amplification in positive, deletion in negative scale) against the chromosomal position (base pairs). In the bottom panel, P-values are plotted. Dark colors represent aberrations more frequently detected in the non-relapsing set of DLBCL, while light colors represent aberrations characteristic of the relapsing set of DLBCL. Genes affected in regions with P-values <0.05 are listed in Supplementary Table S8.

Table 2 Differentially aberrant chromosomal regions between relapsing and non-relapsing DLBCL

Discussion

First presentations and subsequent recurrences of lymphomas are generally regarded as direct outgrowths of the primary tumor clone.4, 6 However, cases of clonally unrelated relapses have been documented in indolent and aggressive lymphomas,5, 23 as well as in Hodgkin and composite lymphomas.9, 24, 25 Reports on unrelated relapses of DLBCL are scarce. One study found two equivocally unrelated recurrences within a group of 13 late-relapsing DLBCL,4 and another reported five late central nervous system relapses of DLBCL unrelated to the respective primaries.26 These studies relied on multiplex PCR-based determination of clonal relationship, a technique that, while quite robust in addressing clonal relationship, is based on the analysis of a single or a few genetic loci, providing no insight into the overall genetic composition of the respective diseases. aCGH microarrays have been previously used to establish relationship between multiple lesions or primaries and metastases in solid tumors.13, 27, 28 Here, we demonstrate that this approach can also be successfully applied to analysis of paired lymphoma samples. Our PCR-based clonality analysis combined with genome-wide copy number arrays and targeted massive parallel sequencing unequivocally proved the existence of clonally unrelated DLBCL recurrences and exposed the profound differences between genomes of primary and relapse tumors. Moreover, this approach advantageously clarified clonal relationship in instances where V(D)J rearrangements could not be amplified or interpreted, as in our case 17.

In agreement with other reports, the clonally unrelated relapses in our study occurred after relatively long intervals from the initial DLBCL diagnosis, thus being considered late (32 months) or very late (81, 103 months). The reasons for a second 'de novo' appearance of DLBCL in the same patient are unclear, but could be related to continuous exogenous hazards, genetic predisposition, treatment-induced DNA damage, acquired mutations in the (hemato-)lymphopoietic stem cell pool or prolonged immunosuppression as suggested by the special patient background of our reported cases (immunosuppression in case 15, immuno-privileged site lymphomas in case 17 and the very old age (80 years) at relapse of case 1). The presence of clonal somatic mutations at low frequency in non-malignant hematopoietic cells is a recently described phenomenon.29 One of the clonally unrelated cases in our cohort (case 15) bore two mutations, an inactivating nonsense mutation in EP300 and a missense mutation in CD79A, in both primary and relapse tumors, which were otherwise genetically distinct (different IGH rearrangements and sequences, different CNA profiles, 10 different private mutations, specific for either primary or relapse), had different cell-of-origin and a discrepant EBV status (EBV+ primary, EBV− relapse). Thus, it is possible that these two mutations were inherent to a putative B-lymphopoietic progenitor cell pool and increased the chance of this patient developing a second lymphoid malignancy. Additionally, the patient’s clinical history of immunosuppression due to solid organ transplantation suggests a second risk factor (that is, immunosuppression). Lack of clonal relationship (that is, second de novo DLBCL and not a progress of the initial DLBCL) may have had a role in the remarkably successful treatment of the relapse tumor with lenalidomide monotherapy, after which the patient achieved lasting clinical remission.30 Also notable in this context is the clinical history of patient 17 who was diagnosed with primary testicular DLBCL followed by an unrelated DLBCL of the central nervous system 81 months later. Interestingly, both lymphomas occurred at immuno-privileged sites, supporting possible underlying immune deregulatory mechanisms or immunogenetic defects as unifying risk factors. Importantly, clonally unrelated DLBCL recurrences in the same patient and demonstration of their different genetic constitutions raise a question of immediate clinical impact, whether salvage therapy should be adjusted to the distinct pathogenesis to avoid overtreatment. The assessability of genomic data by means of current technologies like those used in our study warrants respectively designed prospective trials to address this issue.

The existence of two different patterns of clonally related DLBCL relapses in our study, namely early-divergent/branching- and late-divergent/linear evolution, resonates well with other findings from genomic studies on paired transformed follicular lymphoma (FL) and relapsing DLBCL cohorts. Pasqualucci et al. reported that FL transformation to DLBCL more frequently occurs via divergent evolution from a putative common progenitor (n=10) than by linear progression (n=2).31 Okosun et al.32 found that transformed FL evolved from either rich (n=8) or sparse (n=2) ancestral common progenitor clones. Likewise, by sequencing rearranged V(D)J junctions of paired primary and relapse DLBCL samples, Jiang et al.12 identified early- (n=6) and late-divergent (n=7) patterns of clonal evolution. Thus, it is evident that clonally related DLBCL relapses follow at least two distinct paths of tumor evolution. So far there is no clear explanation what are the underlying reasons for this. One possibility could be that DLBCL relapses fall into one or another genetic evolution category depending on when (and how) resistance to treatment has emerged. It is likely that tumors, which follow early-divergent/branching evolution path, develop intrinsically resistant tumor clones at early stages of lymphomagenesis and that these clones are already present at diagnosis. In support of this hypothesis, Morin et al.16 have recently reported that some mutations that are present at low frequency in primary tumors become more frequent at relapse due to clonal selection and expansion. On the other hand, tumors with late-divergent/linear evolution could have acquired therapy resistance shortly before or during (and possibly because of) treatment. Intrinsic and acquired resistance occur via principally different molecular mechanisms33 and, therefore, if this hypothesis is correct, determining genetic evolution pattern of DLBCL relapses could have important implications to treatment, especially in the coming era of targeted therapy.

We hypothesized that genetic factors are present in the primary DLBCL tumors that determine whether they will or will not recur. To test this hypothesis, we compared CNA and mutational data acquired from primaries of relapsing and non-relapsing DLBCL and identified genetic events that occurred more frequently in the one or the other group. Among differential CNA in multiple chromosomes, gains of the short arm of chromosome 10 in the primaries of relapsing DLBCL group appeared the most significant. Although precise driver identification among more than 130 protein-coding genes affected in the region is difficult, at least 4 potential candidates deserve mentioning: GATA3, PRKCQ, MLLT10 and ABI1. All of them except PRKCQ are included in the cancer census gene list34 highlighting their functional role in oncogenesis. PRKCQ has an established role in lymphomagenesis due to its interactions with various members of signaling pathways, which are important to lymphoma development.35, 36 At the nucleotide level SOCS1 mutations were exclusive to the non-relapsing DLBCL cases. This is in line with the observation that truncating mutations in tumor-suppressor SOCS1 are prognostic for a favorable DLBCL outcome.37 Indeed, two out of five affected cases had nonsense mutations truncating the SOCS1 protein, and the remaining cases had damaging missense mutations in the kinase inhibitory domain, which is required for binding Janus kinases and suppressing their activity.

Recently, three studies addressing the genetic background of DLBCL relapse have been published and identified some early drivers as well as numerous mutated genes potentially contributing to DLBCL recurrences.12, 15, 16 In one study mutations within histone modifiers were implicated as likely drivers of DLBCL.12 We partially confirm this finding by identifying KMT2D as the most commonly mutated gene in both primaries and relapses in our cohort. Moreover, our suggestion that PIM1, MYD88 and CD79B are early drivers of lymphomagenesis is in good agreement with findings presented previously.15, 16 Yet, there is generally only a modest overlap between relapsing DLBCL studies, which might be explained by variable study designs, relatively small tumor cohorts and high genetic heterogeneity of the disease. For example, none of the six relapsed GCB-DLBCL cases included in our cohort had any STAT6 D419 mutations that were recently reported by Morin et al.16 Furthermore, although we detected IRF4 mutations in 3 cases, our data suggests that IRF4 mutations are not relapse-specific as reported by Mareschal et al.,15 but shared between primary and relapse tumors and also present in non-relapsing DLBCL. Finally, SOCS1 mutations were implicated to relapses in at least two previous studies,15, 16 but we found SOCS1 to be mutated exceptionally in non-relapsing DLBCL as discussed earlier. This discrepancy might be explained by the inclusion primary mediastinal B-cell lymphoma, which frequently has SOCS1 mutations, within the relapsed/refractory sample cohorts of the mentioned studies. In our study no primary mediastinal B-cell lymphoma cases were included.

Numerous studies already addressed the negative prognostic value of BCL2 expression in DLBCL22 (and references therein). The significant differences in BCL2 protein expression and frequencies of BCL2 gene mutations between primaries of relapsing and non-relapsing DLBCL observed in our cohort corroborate these observations, reappraising the negative prognostic importance of BCL2 in DLBCL.

In conclusion, genome-wide CNA profiling and targeted sequence analysis of paired primary and relapse DLBCL samples reliably demonstrates, for the first time, the existence of clonally unrelated DLBCL recurrences at the genomic level. Further, our results show that genetic evolution of DLBCL at relapse occurs through early or late divergence from a common progenitor clone. Finally, we demonstrate differentially aberrant regions and mutated genes between relapsing and non-relapsing DLBCL.

References

  1. 1

    Friedberg JW . Relapsed/refractory diffuse large B-cell lymphoma. Hematology Am Soc Hematol Educ Program 2011; 2011: 498–505.

    Article  Google Scholar 

  2. 2

    Philip T, Guglielmi C, Hagenbeek A, Somers R, Van der Lelie H, Bron D et al. Autologous bone marrow transplantation as compared with salvage chemotherapy in relapses of chemotherapy-sensitive non-Hodgkin’s lymphoma. N Engl J Med 1995; 333: 1540–1545.

    CAS  Article  Google Scholar 

  3. 3

    Gisselbrecht C, Glass B, Mounier N, Singh Gill D, Linch DC, Trneny M et al. Salvage regimens with autologous transplantation for relapsed large B-cell lymphoma in the rituximab era. J Clin Oncol 2010; 28: 4184–4190.

    Article  Google Scholar 

  4. 4

    de Jong D, Glas AM, Boerrigter L, Hermus M-C, Dalesio O, Willemse E et al. Very late relapse in diffuse large B-cell lymphoma represents clonally related disease and is marked by germinal center cell features. Blood 2003; 102: 324–327.

    CAS  Article  Google Scholar 

  5. 5

    Geurts-Giele WRR, Wolvers-Tettero ILM, Dinjens WNM, Lam KH, Langerak AW . Successive B-cell lymphomas mostly reflect recurrences rather than unrelated primary lymphomas. Am J Clin Pathol 2013; 140: 114–126.

    Article  Google Scholar 

  6. 6

    Shioyama Y, Nakamura K, Kunitake N, Kimura M, Terashima H, Masuda K . Relapsed non-Hodgkin’s lymphoma: detection and treatment. Radiat Med 2000; 18: 369–375.

    CAS  PubMed  Google Scholar 

  7. 7

    Libra M, De Re V, Gloghini A, Navolanic PM, Carbone A, Boiocchi M . Second primary lymphoma or recurrence: a dilemma solved by VDJ rearrangement analysis. Leuk Lymphoma 2004; 45: 1539–1543.

    Article  Google Scholar 

  8. 8

    Nishiuchi R, Yoshino T, Teramoto N, Sakuma I, Hayashi K, Nakamura S et al. Clonal analysis by polymerase chain reaction of B-cell lymphoma with late relapse: a report of five cases. Cancer 1996; 77: 757–762.

    CAS  Article  Google Scholar 

  9. 9

    Obermann EC, Mueller N, Rufle A, Menter T, Mueller-Garamvoelgyi E, Cathomas G et al. Clonal relationship of classical hodgkin lymphoma and its recurrences. Clin Cancer Res 2011; 17: 5268–5274.

    CAS  Article  Google Scholar 

  10. 10

    Siddiqi IN, Ailawadhi S, Huang Q, Shibata RK, Kang H, Shibata D . Deep sequencing reveals lack of a clonal relationship between a metachronous classical hodgkin and diffuse large B-cell lymphoma. Clin Lymphoma Myeloma Leuk 2014; 14: e87–e93.

    Article  Google Scholar 

  11. 11

    Cobb RM, Oestreich KJ, Osipovich OA, Oltz EM . Accessibility control of V(D)J recombination. Adv Immunol 2006; 91: 45–109.

    CAS  Article  Google Scholar 

  12. 12

    Jiang Y, Redmond D, Nie K, Eng KW, Clozel T, Martin P et al. Deep-sequencing reveals clonal evolution patterns and mutation events associated with relapse in B-cell lymphomas. Genome Biol 2014; 15: 432.

    PubMed  PubMed Central  Google Scholar 

  13. 13

    Torres L, Ribeiro FR, Pandis N, Andersen JA, Heim S, Teixeira MR . Intratumor genomic heterogeneity in breast cancer with clonal divergence between primary carcinomas and lymph node metastases. Breast Cancer Res Treat 2007; 102: 143–155.

    Article  Google Scholar 

  14. 14

    Ostrovnaya I, Olshen AB, Seshan VE, Orlow I, Albertson DG, Begg CB . A metastasis or a second independent cancer? Evaluating the clonal origin of tumors using array copy number data. Stat Med 2010; 29: 1608–1621.

    PubMed  PubMed Central  Google Scholar 

  15. 15

    Mareschal S, Dubois S, Viailly P-J, Bertrand P, Bohers E, Maingonnat C et al. Whole exome sequencing of relapsed/refractory patients expands the repertoire of somatic mutations in diffuse large B-cell lymphoma. Genes Chromosomes Cancer 2015; 55: 251–267.

    Article  Google Scholar 

  16. 16

    Morin RD, Assouline SE, Alcaide M, Mohajeri A, Johnston RL, Chong L et al. Genetic landscapes of relapsed and refractory diffuse large B cell lymphomas. Clin Cancer Res 2015; 22: 2290–2300.

    Article  Google Scholar 

  17. 17

    Nagel S, Hirschmann P, Dirnhofer S, Günthert U, Tzankov A . Coexpression of CD44 variant isoforms and receptor for hyaluronic acid-mediated motility (RHAMM, CD168) is an International Prognostic Index and C-MYC gene status-independent predictor of poor outcome in diffuse large B-cell lymphomas. Exp Hematol 2010; 38: 38–45.

    CAS  Article  Google Scholar 

  18. 18

    Meier VS, Rufle A, Gudat F . Simultaneous evaluation of T- and B-cell clonality, t(11;14) and t(14;18), in a single reaction by a four-color multiplex polymerase chain reaction assay and automated high-resolution fragment analysis: a method for the rapid molecular diagnosis of lympho. Am J Pathol 2001; 159: 2031–2043.

    CAS  Article  Google Scholar 

  19. 19

    Ye J, Ma N, Madden TL, Ostell JM . IgBLAST: an immunoglobulin variable domain sequence analysis tool. Nucleic Acids Res 2013; 41: W34–W40.

    Article  Google Scholar 

  20. 20

    Juskevicius D, Ruiz C, Dirnhofer S, Tzankov A . Clinical, morphologic, phenotypic, and genetic evidence of cyclin D1-positive diffuse large B-cell lymphomas with CYCLIN D1 gene rearrangements. Am J Surg Pathol 2014; 38: 719–727.

    Article  Google Scholar 

  21. 21

    Lipson D, Aumann Y, Ben-Dor A, Linial N, Yakhini Z . Efficient calculation of interval scores for DNA copy number data analysis. J Comput Biol 2006; 13: 215–228.

    CAS  Article  Google Scholar 

  22. 22

    Visco C, Tzankov A, Xu-Monette ZY, Miranda RN, Tai YC, Li Y et al. Patients with diffuse large B-cell lymphoma of germinal center origin with BCL2 translocations have poor outcome, irrespective of MYC status: a report from an International DLBCL rituximab-CHOP Consortium Program Study. Haematologica 2013; 98: 255–263.

    CAS  Article  Google Scholar 

  23. 23

    Mao Z, Quintanilla-Martinez L, Raffeld M, Richter M, Krugmann J, Burek C et al. IgVH mutational status and clonality analysis of Richter’s transformation: diffuse large B-cell lymphoma and Hodgkin lymphoma in association with B-cell chronic lymphocytic leukemia (B-CLL) represent 2 different pathways of disease evolution. Am J Surg Pathol 2007; 31: 1605–1614.

    Article  Google Scholar 

  24. 24

    Ganzel C, Pogrebijsky G, Krichevsky S, Neuman T, Yehuda D Ben . Separate diagnoses of Hodgkin lymphoma and non-Hodgkin lymphoma in an individual patient might not signify a common clonal origin. Exp Hematol 2012; 40: 698–704.

    Article  Google Scholar 

  25. 25

    Libra M, De Re V, Gasparotto D, Gloghini A, Marzotto A, Milan I et al. Differentiation between non-Hodgkin’s lymphoma recurrence and second primary lymphoma by VDJ rearrangement analysis. Br J Haematol 2002; 118: 809–812.

    Article  Google Scholar 

  26. 26

    Lossos A, Ashhab Y, Sverdlin E, Amir G, Ben-Yehuda D, Siegal T . Late-delayed cerebral involvement in systemic non-hodgkin lymphoma: A second primary tumor or a tardy recurrence? Cancer 2004; 101: 1843–1849.

    Article  Google Scholar 

  27. 27

    Wa CV, DeVries S, Chen YY, Waldman FM, Hwang ES . Clinical application of array-based comparative genomic hybridization to define the relationship between multiple synchronous tumors. Mod Pathol 2005; 18: 591–597.

    CAS  Article  Google Scholar 

  28. 28

    Waldman FM, DeVries S, Chew KL, Moore DH, Kerlikowske K, Ljung BM . Chromosomal alterations in ductal carcinomas in situ and their in situ recurrences. J Natl Cancer Inst 2000; 92: 313–320.

    CAS  Article  Google Scholar 

  29. 29

    Steensma DP, Bejar R, Jaiswal S, Lindsley RC, Sekeres MA, Hasserjian RP et al. Clonal hematopoiesis of indeterminate potential and its distinction from myelodysplastic syndromes. Blood 2015; 126: 9–16.

    CAS  Article  Google Scholar 

  30. 30

    Läubli H, Tzankov A, Juskevicius D, Degen L, Rochlitz C, Stenner-Liewen F . Lenalidomide monotherapy leads to a complete remission in refractory B-cell post-transplant lymphoproliferative disorder. Leuk Lymphoma 2016; 57: 945–948.

    Article  Google Scholar 

  31. 31

    Pasqualucci L, Khiabanian H, Fangazio M, Vasishtha M, Messina M, Holmes AB et al. Genetics of follicular lymphoma transformation. Cell Rep 2014; 6: 130–140.

    CAS  Article  Google Scholar 

  32. 32

    Okosun J, Bödör C, Wang J, Araf S, Yang C-Y, Pan C et al. Integrated genomic analysis identifies recurrent mutations and evolution patterns driving the initiation and progression of follicular lymphoma. Nat Genet 2014; 46: 176–181.

    CAS  Article  Google Scholar 

  33. 33

    Asić K . Dominant mechanisms of primary resistance differ from dominant mechanisms of secondary resistance to targeted therapies. Crit Rev Oncol Hematol 2016; 97: 178–196.

    Article  Google Scholar 

  34. 34

    Futreal PA, Coin L, Marshall M, Down T, Hubbard T, Wooster R et al. A census of human cancer genes. Nat Rev Cancer 2004; 4: 177–183.

    CAS  Article  Google Scholar 

  35. 35

    Felli MP, Vacca A, Calce A, Bellavia D, Campese AF, Grillo R et al. PKC theta mediates pre-TCR signaling and contributes to Notch3-induced T-cell leukemia. Oncogene 2005; 24: 992–1000.

    CAS  Article  Google Scholar 

  36. 36

    Briones J, Moga E, Espinosa I, Vergara C, Alvarez E, Villa J et al. Bcl-10 protein highly correlates with the expression of phosphorylated p65 NF-kappaB in peripheral T-cell lymphomas and is associated with clinical outcome. Histopathology 2009; 54: 478–485.

    Article  Google Scholar 

  37. 37

    Schif B, Lennerz JK, Kohler CW, Kreuz M, Bentink S, Melzner I et al. SOCS1 mutation subtypes predict divergent outcomes in diffuse large B-cell lymphoma (DLBCL) patients. Oncotarget 2012; 4: 35–47.

    PubMed Central  Google Scholar 

Download references

Acknowledgements

This study was supported by Krebsliga beider Basel and Stiftung zur Krebsbekämpfung Zürich. We thank Sibylle Tschumi, Tanja Dietsche, David Jucker, Luca Quagliata and Bruno Grilli for their help in performing NGS and FISH investigations. The chromosomal copy number aberration data reported in this paper has been deposited at the National Center for Biotechnology Information Gene Expression Omnibus (accession no. GSE65720). The targeted sequencing data has been deposited to Sequence Read Archive (accession no. SRP067323).

Author contributions

DJ performed the research, analyzed and interpreted the data, and wrote the manuscript; JG, CR and TL analyzed the data and revised the manuscript; VP contributed to sequencing analysis; FS-L analyzed the clinical data and revised the manuscript; SD performed histological and immunohistochemical analysis, revised the manuscript; AT conceived and designed the study, performed histological, immunohistochemical and FISH analysis, analyzed and interpreted the data, and revised the manuscript.

Author information

Affiliations

Authors

Corresponding author

Correspondence to A Tzankov.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies this paper on the Leukemia website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Juskevicius, D., Lorber, T., Gsponer, J. et al. Distinct genetic evolution patterns of relapsing diffuse large B-cell lymphoma revealed by genome-wide copy number aberration and targeted sequencing analysis. Leukemia 30, 2385–2395 (2016). https://doi.org/10.1038/leu.2016.135

Download citation

Further reading

Search

Quick links