**4. Discussion**

Despite advances in our understanding of the pathobiology of MPNST and the identification of seemingly promising therapeutic targets using a single model system in preclinical studies, no investigational agents have demonstrated efficacy following translation to human clinical trials. One element that has largely been ignored in the study of MPNST has been the possible existence of intra-tumoral heterogeneity. No single study in MPNST has focused on intra-tumoral heterogeneity. However, spatial intra-tumoral heterogeneity has become an area of interest in the study of other solid malignancies to begin to understand clonal evolution [91–95]. Within the NF1 field, researchers are beginning to appreciate the importance of understanding spatial and temporal heterogeneity. For example, Peacock et al. performed a genomic analysis of serial samples from one patient who developed an MPNST. Samples were taken at four timepoints (benign plexiform neurofibroma, MPNST pre-treatment, MPNST post-treatment, and MPNST at time of metastasis) [96]. They observed early hemizygous microdeletions in *NF1* and *TP53* with progressive amplifications of *MET*, *HGF*, and *EGFR,* highlighting the potential role of these pathways in progression. Additionally, Carrió et al. have started to examine intra-tumoral heterogeneity in PNF (plexiform neurofibromas), ANF (atypical neurofibroma) and ANNUBP (atypical neurofibromatous neoplasms with uncertain biological potential), the precursors to MPNST. They performed SNP-array analysis and exome sequencing on multiple biopsies of eight PNF, of which some had areas consistent with ANF or ANNUBP. Their data suggested that loss of a single copy of *CDKN2A*/*B* in *NF1* null cells is sufficient to start ANF development and that total inactivation of both copies is necessary to form ANNUBP [97]. Our study represents the first look at spatial intra-tumoral heterogeneity within an MPNST. We have demonstrated differing mutational profiles, copy number alteration signatures, and gene expression profiles within the three areas sampled. The differing mutation profile includes a variety of single nucleotide variants, including missense, frameshift, and synonymous variants. The role of synonymous variants in the tumorigenesis of MPNST is uncertain. However, there is increasing evidence that synonymous variants can alter gene expression and protein function and thus cannot be simply disregarded [98–101]. Additionally, several of the genes in Table 3a,b have previously been implicated in cancer [102–115]. For example, in Area 2, *CSK* was found to have a frameshift variant in its functional domain. *CSK* encodes a C-terminal Src kinase that has previously been found to act as a tumor suppressor in both breast cancer and prostate cancer [112–114]. Interestingly, in the context of breast cancer, Smith et al. showed that C-terminal Src kinase loss facilitated tumorigenesis by altering expression of the *PRC2* complex subunits, *EZH2* and *SUZ12* [113]. Based on these data, it is possible that alterations in *CSK* could be another way in which the *PRC2* complex is affected in MPNST. Another gene, *CCL16*, is involved in chemotaxis of human monocytes and lymphocytes. This chemokine was shown to delay mammary tumor growth and reduce rates of metastasis in mouse models [115], raising the possibility of decreased immune surveillance of our patient's MPNST secondary to a non-functional *CCL16*. In addition to the differences in single nucleotide variants, there were differences in copy number alterations across the three areas with Area

2 showing the most distinct signature in terms of copy number gains and losses. The degree to which each somatic variant, differentially expressed gene, and copy number aberration contributes to the biologic heterogeneity of the tumor remains uncertain. However, future work in our lab will be geared at elucidating this information. Finally, there was a distinct difference in gene expression among the three areas with gene ontology studies pointing toward differences in translation and protein targeting.

Taken together, these data point toward the existence of intra-tumoral heterogeneity and sugges<sup>t</sup> that further investigation into this phenomenon is warranted. Additionally, these data sugges<sup>t</sup> that there should be some caution taken in interpreting sequencing that comes from a single biopsy site. The advent of single cell sequencing has allowed for more rigorous evaluation of intra-tumoral heterogeneity in other cancers including acute leukemias [116,117], as well as in some solid malignancies [118,119]. Future work will be geared at using this data as the foundation to better understand clonal heterogeneity along with single cell sequencing to comprehensively evaluate intra-tumoral heterogeneity and clonal evolution of MPNST.
