While the genomic landscape of CNS lymphomas at diagnosis has largely been defined, there remains a significant gap in our understanding of the molecular pathways that mediate therapeutic resistance and disease progression. CNS lymphomas generally exhibit multifocal dissemination as an end-stage manifestation. To elucidate key genetic and pathophysiologic pathways of CNS lymphoma that mediate resistance, we have conducted what may be the first study to apply next generation sequencing to compare the genetic features of PCNSL and SCNSL at the time of diagnosis with multifocal lesions isolated from relapsed, treatment-refractory disease from matched patient specimens at final lymphoma progression, isolated from whole brain autopsy specimens.

Whole-exome sequencing from diagnostic specimens (4 brain and 1 retinal biopsy) were compared with matched whole brain autopsy specimens in 5 patients: 4 with PCNSL and 1 with SCNSL. A median of 6 distinct anatomical CNS lymphoma-containing brain regions were analyzed/whole brain autopsy specimen. Median age was 57 (range 51-81 yrs). Two patients died after primary refractory disease to methotrexate-based induction and three died after median of four progressions; each of these later patients progressed after lenalidomide, and one also received pomalidomide followed by ibrutinib before final disease progression. Four cases were from UCSF and 1 from Stanford.

Tumor tissue and uninvolved normal brain was microdissected from formalin-fixed, paraffin-embedded blocks of initial biopsy prior to therapy and multiple distinct tumor regions at time of autopsy. Multiplex library preparation was performed from extracted genomic DNA and hybridization-capture of pooled libraries performed using NimbleGen SeqCap EZ Human Exome Probes Bait Library. Captured libraries were sequenced as paired-end 100 base pair reads and sequence reads mapped to the reference human genome build GRCh37 (hg19) using Sentieon DNAseq. Somatic variant detection including single nucleotide variants and insertions/deletions was performed. Single nucleotide variants and insertions/deletions were verified using Integrated Genome Viewer. Copy number analysis was performed using CNVkit and visualized using Nexus. High-confidence somatic nonsynonymous mutations were identified and pathway analysis applied using gene sets for canonical pathways and Gene Ontology (GO) terms downloaded from Molecular Signatures Database) at http://www.broad.mit.edu/gsea/msigdb. Lymphoid-specific signatures were curated by Staudt lab (http://lymphochip.nih.gov/signaturedb/) or previous publications. Fisher's Exact test was used to calculate enrichment p values for each of those gene sets and the BH method used for False Discovery Rate (FDR) control.

A mean of 214 lymphoma-specific mutations per specimen was identified (range 56-1,764). Notably, of the 283 distinct molecular pathways analyzed, only 1 pathway was significantly enriched for mutations in tumors from each of the 5 CNS lymphoma whole brain autopsy specimens compared to pre-treatment diagnostic biopsies: the Ikaros 1(IKZF1) in centroblast pathway (p<0.05). The 2 cases of CNS lymphoma that had been exposed to an IMiD most proximate to brain autopsy (< 6 weeks) exhibited the strongest enrichment for mutations in IKZF1 pathway (p<0.005). Genes recurrently mutated in the IKZF1 pathway in relapsed CNS lymphomas included CD79A/B, IGLL5, BTG2, and TMEM30A. Mutations in the BCL6 and BACH2 transcriptional repressor pathway were significantly enriched in CNS lymphomas at autopsy compared to diagnostic specimens in 3 out of 5 cases. We also detected marked regional heterogeneity in DNA copy number aberrations in lymphoma from whole brain autopsy specimens. High level focal amplification of 9p24.1 targeting PD-L1 was enriched in only one case: a highly refractory PCNSL that had been exposed to pomalidomide followed by ibrutinib within two months of autopsy; amplification of PD-L1 was not detected in the diagnostic specimen.

While multiple genetic aberrations are associated with disease progression in CNS lymphoma, our study suggests that mutational activation of the IKZF1 pathway may be a unique canonical mechanism in the evolution of resistance to multiple interventions, including first-line methotrexate. This observation may have significant therapeutic implications. Supported by NCI and Leukemia Lymphoma Society.

Disclosures

Rubenstein:Genentech: Research Funding; Celgene: Research Funding; Bristol Myers Squibb: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution