Integrative Genomic Analyses of Malignant Peripheral Nerve Sheath Tumors

Abstract

[eng] Malignant peripheral nerve sheath tumors (MPNSTs) are rare soft-tissues sarcomas of the peripheral nervous system (PNS) with poor prognosis and lack of therapeutic options. Complete resection with wide margins of the tumor is essential for a good prognosis of MPNSTs like for many other soft-tissue sarcomas. MPNSTs develop either sporadically or in the context of neurofibromatosis type 1 (NF1), a genetic condition affecting 1:3000 people worldwide. The diagnosis of MPNSTs can be challenging, especially outside the NF1 context since these are rare tumors and there are other tumor entities with overlapping histology. NF1 patients develop neurofibromas: benign tumors of the PNS caused by the complete inactivation of the NF1 gene in a cell of the Schwann cell (SC) lineage and they can be cutaneous neurofibromas (cNF) and plexiform neurofibromas (pNF), the latter being able to progress towards an MPNST. In the context of NF1, a clear implication of CDKN2A in the progression from a pNF toward an atypical neurofibroma (ANNUBP) was established, which is considered a premalignant lesion. However, whereas all ANNUBPs bear the complete inactivation of CDKN2A, different genomic analyses of MPNST identified the loss of the CDKN2A/B locus in, at most, 70-80% of MPNSTs analyzed, suggesting an alternative path for MPNST tumorigenesis. In addition, whereas pNF and ANNUBP genomes were essentially 2n, the genomic structure of MPNSTs was characterized for having highly rearranged and hyperploid genomes, highlighting the importance of analyzing their genomic structure to understand MPNST biology. The main objective of this thesis was to gain insight into the genomic characteristics of MPNSTs, using different genomic analyses and bioinformatic approaches. In particular, we were interested in better understanding the process of MPNST tumorigenesis focusing on: the possible molecular paths of MPNST formation, and obtaining a better genomic definition of what is an MPNST. We intended that this genomic information would help, in the future, the development of new precision medicine strategies for MPNSTs, in their differential diagnostics and patient management. To understand if there were more than one molecular path in MPNST tumorigenesis we analyze Single nucleotide polymorphism (SNP)-array, Whole exome sequencing (WES), and Whole genome sequencing (WGS) of different MPNST cell lines and primary tumors. We detected the complete inactivation of CDKN2A in all MPNST studied mainly by

structural variants (SV), and specifically by inter-chromosomal translocations. We were able to detect this kind of inactivation due to the use of WGS technology. The inactivation of CDKN2A in a set of different MPNSTs and an ANNUBP reinforced the idea that the complete inactivation of CDKN2A (encoding both p16INK4 and p14ARF proteins) is a bottleneck in the MPNST tumorigenesis. It also uncovered a striking recurrence of inactivation by inter-chromosomal translocations, especially concentrating in a hotspot close to exon 2. To gain insight into the genomic characteristics of MPNSTs, and to generate tools for developing precision medicine strategies, we genomically characterized 9 of the most widely used MPNST cell lines. We discovered the misidentification of one of the cell lines. In addition, a deep genomic analysis along with a methylome-based classification and the study of cell-identity expression markers challenged the identity of common MPNST cell lines, some of them re-classified as melanomas or other soft-tissue sarcomas, which had different treatment responses compared to genuine MPNST cell lines. These results alerted us to the potential heterogeneity of tumor entities under an MPNST label. In addition, this analysis opened an opportunity to incorporate these molecular techniques along with a thorough histological characterization for revising MPNST differential diagnosis. To translate these results to primary MPNSTs, we collected and genomically analyzed 19 tumors clinically diagnosed as MPNSTs to gain insight into the process of MPNST tumorigenesis and provide a genomic definition of a classic MPNST. To assess any potential heterogeneity regarding tumor identities, we first classified these samples according to the expression of transcription factors as they are good markers of cell identity, and then analyzed the genomic characteristics of the different groups of tumors. The first cluster (C1) was formed by 14 tumors clustering together with 5 genuine MPNST cell lines, sharing specific genomic characteristics. The other one (C2) was heterogeneous and composed of 5 tumors clustering together with non-MPNST cell lines having completely different genomes. Due to the high recurrence of genomic features in C1 tumors, we could establish a genomic definition of classic MPNST. Classic MPNSTs have highly rearranged and hyperploid genomes represented by even copy numbers and recurrent copy-neutral LOH regions (possibly resulting from the loss of chromosomal

regions before a tetraploidization event). They also have the complete inactivation of CDKN2A by inter-chromosomal translocations, as well as the complete inactivation of NF1 and Polycomb repressive complex 2 (PRC2) tumor suppressor genes. Their genomes present low mutational frequency with the absence of associated mutational signatures associated and also, the absence of activating mutations either by point mutations or fusion genes. C2 constituted a heterogeneous group of tumors, differing in several C1 genomic characteristics, possibly indicating distinct tumor entities. In conclusion, in this thesis we achieved a better understanding of the genomic features of MPNST, and provided a genomic definition of a classic MPNSTs. We gained insight into MPNST tumorigenesis and also provided a genomic resource of a set of MPNST cell lines to perform translational research and precision medicine strategies. We think that our findings will help in the differential diagnosis of MPNTS and improve patient management.