Genome-wide analysis uncovers novel recurrent alterations in primary central nervous system lymphomas

Authors

Esteban Braggio

Scott Van Wier

Juhi Ojha

Ellen McPhail

Yan W. Asmann

Jan Egan

Jackline Ayres da Silva

David Schiff

M. Beatriz Lopes

Paul A. Decker

Riccardo Valdez

Raoul Tibes

Bruce Eckloff

Thomas E. Witzig

A. Keith Stewart

Rafael Fonseca

Brian Patrick O’Neill

Doi

10.1158/1078-0432.CCR-14-2116

PMID: 25991819 · DOI: 10.1158/1078-0432.CCR-14-2116 · Journal: Clinical Cancer Research (2015)

TL;DR

Braggio et al. performed a comprehensive genomic study of 19 immunocompetent primary central nervous system lymphoma (PCNSL) patients using array-CGH, whole exome sequencing, mate-pair WGS, targeted sequencing, and FISH. They uncovered novel PCNSL-specific recurrent biallelic inactivation of TOX and PRKCD, a high prevalence of MYD88 mutations (79%) and CDKN2A biallelic loss (60%), plus a set of alterations shared with systemic DLBCL (e.g. TNFAIP3, PRDM1, GNA13, TMEM30A, TBL1XR1, B2M, CD58, CD79B, CARD11, IGH-BCL6 translocations). BCR/TLR/NF-κB pathway components were altered in >90% of PCNSL cases, highlighting potential therapeutic targets.

Cohort & data

  • N = 19 immunocompetent PCNSL patients (HIV− and EBV−), newly diagnosed and confined to the CNS, retrieved from the Mayo Clinic Tumor Registry (IRB 08-001933) and the University of Virginia (IRB 14225). Cancer type: PCNSL; compared throughout to systemic DLBCLNOS.
  • Dataset: pcnsl_mayo_2015. aCGH microarray data deposited in GEO under accession GSE28952.
  • DNA from frozen tissue (7 cases) or FFPE (12 cases); 3 frozen samples required whole-genome amplification due to insufficient yield. 9 of 19 biopsies provided tumor-free tissue for matched-normal validation sequencing.
  • Assays/methods applied:
    • array-cgh-agilent-1m — 18 cases, Human Genome 244A + Sureprint G3 (Agilent); CNA calling via Nexus RANK segmentation; CNVs filtered using TCAG plus 10 in-house HapMap controls.
    • whole-exome-seq — 10 cases, SureSelect 50 Mb capture, 100 bp paired-end on HiSeq2000, mean 65.6M mapped reads in capture, median 80% of exome at ≥30× coverage. SNVs called with SomaticSniper; indels with gatk-somatic-indel-detector; annotated by snpEFF/PolyPhen-2 via the TREAT workflow.
    • mate-pair-seq, targeted-dna-seq (Ion Torrent PGM, MYD88 coding regions at ~2,005× average depth), sanger-sequencing for validation of CARD11/CD79B/PRKCD/TNFAIP3 variants, and interphase fish for IGH-BCL6, PRKCD, and TOX rearrangements.

Key findings

  • Highly complex karyotype. Median 21 copy-number abnormalities per patient (range 10–47). Sixteen chromosomal abnormalities were present in >25% of cases (10 deletions, 6 gains).
  • CDKN2A/B (9p21.3) deletion was the most common CNA: 83% of cases. Homozygous deletion of CDKN2A (and CDKN2B) was found in 55% of cases; another 21% had monoallelic loss.
  • Chromosome 6 abnormalities were prominent: 6p21 (HLA genes) deletion in 9/18 (50%); 6q21 (PRDM1) and 6q23 (TNFAIP3) losses in 10/18 each (55%); 6q14.1 (TMEM30A) deletion in 8/18 (44%).
  • PCNSL-specific biallelic PRKCD inactivation. Two of 10 WES cases had splice-donor mutations (one with an additional missense p.W608S), each co-occurring with monoallelic deletion of PRKCD; another 3/18 aCGH cases carried focal monoallelic deletions (<1 Mb). No PRKCD recurrence has been reported in nodal DLBCL; COSMIC lists only 1/1,058 hematologic samples (p.A163T in ALL, 0.01%).
  • PCNSL-specific TOX biallelic loss. 11% of cases had homozygous deletion of TOX (8q12); 17% additional cases showed focal monoallelic deletion. No TOX homozygous deletions were detected in 754 cell lines (including 127 hematopoietic/lymphoid) in the Cancer Genome Project dataset.
  • MYD88 activating mutations were near-ubiquitous. Mutated in 7/10 WES cases (70%) and in 11/14 cases (79%) after adding 4 targeted-sequenced cases. Nearly all were the L265P hotspot; two were V217F and M232T. Prevalence is ~2× that reported in nodal ABC-DLBCL.
  • Other recurrently mutated NF-κB pathway genes: CD79B 40%, PIM1 40%, BTG2 30%, CARD11 30%, GNA13 20%, TNFAIP3 20%, BRAF 10%. 3 of 4 CD79B-mutant cases co-mutated MYD88.
  • Recurrent homozygous deletions in PCNSL included HLA genes (4/18, 22%), PRDM1, TMEM30A, TOX, and CD58 (each in 2/18, 11%); plus individual HDs of B2M, ETV6, TNFAIP3, and TNFRSF10A (Table 1).
  • High-level amplifications were rare (4 events total): 1p31.1–p22.2 (including BCL10), 7p22.3–p22.2 (including CARD11), 12q14.1–q21.33 (including BCL7A region), and 18q21.2 (TCF4) — each in one sample.
  • Gains of 3p14 (FOXP1) and 3q27 (BCL6) were noted, with breakpoints inside or 5′ of these genes suggesting their involvement in unbalanced translocations. Other recurrent gains: 7q21–q31 (39%), 11q (28%), 12q (50%), 19q13 (33%); loss of 10p15.3–p14 in 33%.
  • IGH-BCL6 translocations detected by FISH in 2/11 PCNSL cases analyzed.
  • BCR/TLR/NF-κB pathway alteration in 93% (≈>90%) of PCNSL when integrating mutation and CNA data. Pathway enrichment also implicated proliferation, immune response, apoptosis regulation, and lymphocyte differentiation.
  • Mutually exclusive copy-number patterns: gain of 12q was mutually exclusive with deletion of 10p15.3–p14 and with gains of 3p14 (FOXP1) and 11q (P = 0.0004); losses of 3p21.31 (PRKCD) and 6q23 (TNFAIP3) were rare in 12q-gained cases.
  • Survival association: deletion of 6q21 (PRDM1) was associated with shorter overall survival in univariate analysis (log-rank P = 0.001), but the authors caution this is in a small, heterogeneously treated cohort.

Genes & alterations

  • PRKCD — biallelic inactivation (splice-donor mutations + p.W608S missense; monoallelic deletions). Novel PCNSL-specific finding; not recurrently mutated in nodal DLBCL.
  • TOX — homozygous deletion (11%) and focal monoallelic deletion (17%) at 8q12. Novel PCNSL-specific finding; loss tentatively associated with shorter OS.
  • MYD88 — activating mutations in 79% (11/14), predominantly L265P, plus V217F and M232T. ~2× nodal ABC-DLBCL prevalence.
  • CDKN2A / CDKN2B — 9p21.3 deletion in 83%; homozygous in 55%.
  • PRDM1 — 6q21 deletion in 55%; homozygous in 2/18. Deletion associated with shorter OS.
  • TNFAIP3 — 6q23 deletion in 55%; somatic mutations in 20%.
  • TMEM30A — 6q14.1 deletion in 44%; homozygous in 2/18.
  • TBL1XR1 — 3q26.32 deletion (recurrent).
  • B2M — homozygous deletion in individual cases.
  • CD58 — homozygous deletion in 2/18 (11%).
  • CD79B — activating mutations in 40%; frequently co-occur with MYD88 mutations (3 of 4 CD79B-mutant cases).
  • CARD11 — mutations in 30%; one case with high-level amplification at 7p22.
  • BCL6 — IGH-BCL6 translocation in 2/11 FISH-screened cases; recurrent 3q27 gains with breakpoints suggesting translocation.
  • FOXP1 — recurrent 3p14 gain with breakpoints suggesting translocation.
  • GNA13 — mutations in 20%; recurrent homozygous deletion.
  • PIM1 — mutations in 40%.
  • BTG2 — mutations in 30%.
  • BRAF — mutations in 10%.
  • BCL10, BCL7A, TCF4, ETV6, TNFRSF10A — each implicated in one or two cases as part of focal amplification or homozygous deletion events.

Clinical implications

  • BCR/TLR/NF-κB pathway components were altered in >90% of PCNSL, supporting targeted therapy directed at this axis (e.g. BTK, IRAK4, PI3K inhibition) as a rational strategy in PCNSL — though the paper does not test any drug.
  • The PCNSL-specific MYD88 L265P prevalence (~79%) is higher than reported in nodal ABC-DLBCL and makes MYD88-dependent signaling a particularly attractive target in PCNSL.
  • Loss of PRKCD expression has previously been linked to radiation resistance in cervical cancer (via MiR-181a); the authors speculate this could extrapolate to PCNSL, suggesting PRKCD status as a candidate biomarker for radiation responsiveness.
  • Deletion of 6q21 (PRDM1) was associated with shorter overall survival (univariate log-rank P = 0.001), nominating it as a candidate prognostic marker pending validation.

Limitations & open questions

  • Small cohort (N = 19) with heterogeneous treatment exposure limits power for survival associations; the PRDM1/OS link is explicitly flagged by the authors as requiring confirmation in larger cohorts with robust age, performance status, comorbidity, and treatment annotation.
  • FFPE/amplified DNA in a subset of samples yielded lower WES coverage than fresh-frozen DNA; some cases had only 18-case aCGH or 10-case WES coverage, so prevalence estimates have wide confidence intervals.
  • No functional validation of TOX or PRKCD loss in PCNSL models — the pathogenic role is inferred from recurrence and from prior literature in T- and B-cell biology.
  • No matched germline DNA was available for 10 of 19 cases, which limited somatic-vs-germline calling for those samples.
  • Targetable-pathway claim is not therapeutically tested in this work — actual drug response in BCR/TLR/NF-κB-altered PCNSL remains an open question.
  • Comparison to systemic DLBCL is qualitative (literature-based) rather than against a paired internal DLBCL cohort.

Citations from this paper used in the wiki

  • “We performed a comprehensive genomic study of tumor samples from 19 immunocompetent PCNSL patients. Testing comprised array-comparative genomic hybridization and whole exome sequencing.” (Abstract)
  • “Biallelic inactivation of TOX and PRKCD were recurrently found in PCNSL but not in systemic DLBCL, suggesting a specific role in PCNSL pathogenesis.” (Abstract)
  • “we found a high prevalence of MYD88 mutations (79%) and CDKN2A biallelic loss (60%).” (Abstract)
  • “Overall, BCR/TLR/NF-κB pathways were altered in >90% of PNCSL, highlighting its value for targeted therapeutic approaches.” (Abstract)
  • “PCNSL cases showed a complex genome with a median of 21 CNA per patient (range 10–47).” (Results)
  • “The most common abnormality was the deletion of 9p21.3 (CDKN2A and CDKN2B), which was found in 83% of cases.” (Results)
  • “11% of PCNSL showed homozygous deletion of TOX and another 17% of cases showed focal monoallelic deletions in the gene.” (Results)
  • “mutations were found in 11 of 14 cases analyzed (79%). The mutations were found in the residue L265P, with the exception of two mutations, affecting the residue 217 (V217F) and the residue 232 (M232T).” (Results, MYD88)
  • “Deletion of 6q21 (PRDM1) was associated with shorter OS in univariate analysis (P = 0.001; Supplementary Figure S3).” (Results)
  • “Microarray data is deposited in GEO dataset under accession GSE28952.” (Methods)

This page was processed by crosslinker on 2026-05-14.