Whole Genome Sequencing Reveals Recurrent Structural Driver Events in Peripheral T-Cell Lymphomas Not Otherwise Specified

Historically, the differential diagnosis between different nodal peripheral T-cell lymphoma (PTCL) subtypes based on morphological and phenotypic grounds has posed great challenges. In the last few years, our knowledge of the molecular bases of different PTCLs has significantly expanded. However, peripheral T-cell lymphomas not otherwise specified (PTCL-NOSs) are still regarded to as a heterogeneous category encompassing PTCL cases not fitting other, more homogeneous, subtypes. In fact, PTCL-NOS is one of the few lymphoma subtypes where no recurrent driver mutations have been reported so far. In order to better characterized the PTCL-NOS genomic landscape, we decided to investigate 11 PTCL-NOS patients by a whole genome sequencing (WGS) approach (median coverage 27X). Ten out of eleven samples were collected from FFPE blocks and 2 were removed from analysis: one due to low cancer cell fraction (CCF) and the other based on cluster generation issues during sequencing likely caused by a hyper-fragmented DNA. Among the remaining 9 cases, we extracted 59,617 somatic base substitutions (range 2,471-10,756, median 6,358 per patient) and 20,531 small insertion-deletions (indels) (range 84-6,397, median 1,580). We were able to characterize the spectrum of FFPE-induced artefacts, mostly composed of point mutations and indels within LINE-1 (L1) elements, predominantly of the L1PA family. This is a crucial quality control step that could be applied to similar future studies from archive samples. Four samples were heavily involved by FFPE-related artefacts and were excluded for this reason. Using a non-negative matrix factorization (NNMF) algorithm we investigated for the first time the PTCL-NOS mutational signature landscape. We did not find novel processes in this entity, but rather known processes operative in other lymphoid malignancies. Among those: signatures 1 and 5, deriving from the age-related process of spontaneous deamination of methylated cytosines; signatures 2 and 13 deriving from aberrant activity of the APOBEC family of DNA deaminases; signatures 17 and 8, pertaining to two yet poorly characterized processes. The contribution of different processes to the mutational spectrum of each case was profoundly heterogeneous. Combining our data set with 64 previously published whole exome sequencing cases (23 ALCL, 15 AITLs, 9 PTCL-NOSs and 16 EATL-II), we confirmed the lack of recurrent driver mutations among PTCL-NOS. Taking advantage of WGS data, we therefore focused on structural variants (SVs: inversions, translocations, internal tandem duplications and deletions) and copy number alterations (CNAs). We found 372 SVs, with a stunning median of 73 per sample (range 56-86). Even more interesting, at least one complex event was observed in all but one patients, including one whole genome duplication (WGD) and five chromothripsis events in three patients, suggesting a critical role of SVs in shaping the PTCL-NOS genome. We found that known onco-drivers were recurrently disrupted by such events: the most frequent target was CDKN2A, deleted in 4 out of 5 patients, 2 of which carried homozygous deletions. Interestingly, PTEN loss was observed in 2 out of 4 CDKN2A-deleted patients. Given the high prevalence of these deletions, we extended our observation to an independent validation set of ALCLs (n=56), AITL (n=22) and PTCL-NOS (n=59) investigated by FISH (n=36), next generation sequencing (n=25) or SNP6 array series (n=76). Overall, CDKN2A was deleted in 22/59 (37%) PTCL-NOSs cases, and in 17/22 (77%) both alleles were lost. PTEN was deleted in 12/59 (20%) PTCL-NOS cases, all of which also carried a CDKN2A loss. Strikingly, the co-occurrence of CDKN2A and PTEN was found only among PTCL-NOS, and in none of the other entities. With the limitations of the small sample size, the presence of CDKN2A bi-allelic deletions was associated with inferior survival (25% [95% CI: 9-66%] 5-y OS for deleted cases vs 52% [95% CI: 28-96%] for wt/hemizygous cases, p=0.042) among patients treated with an autologous bone marrow transplant front line program for advance stage and high-risk disease (n=19). Our observations point at SVs as a main driver of PTCL-NOS, often involving known cancer genes and their downstream pathways. Furthermore, our data highlighted recurrent gene deletions that may be relevant for differential diagnosis within this category of lymphomas.