**3. Results**

## *3.1. Patient Information*

Patient characteristics can be seen in Table 1. The patient was a male with a history significant for a clinical diagnosis of neurofibromatosis type 1—patient had a plexiform neurofibroma, spinal neurofibromas, café au lait macules, and multiple first-degree relatives with neurofibromatosis type 1—and was 40 years old at the time of diagnosis of MPNST. He presented with a large tumor located in the left neck. Resection showed a high-grade malignant peripheral nerve sheath tumor, 10.2 cm in the largest dimension, with negative margins. The patient did not receive any adjuvant therapy for his MPNST following initial resection due to poor performance status. He recurred 21 months after the initial diagnosis and ultimately died secondary to complications from metastatic disease (33 months after initial diagnosis). Samples were taken in three di fferent locations within the primary tumor immediately following the inititial resection for the purpose of this study.

#### **Table 1.** Patient Characteristics.


 By French Federation of Cancer Centers Sarcoma Group Grading System (FNCLCC) [46]; OS = Overall Survival-time from diagnosis of MPNST to death.

#### *3.2. Histology of Biopsy Sites*

We first reviewed the H&E images of the tumor to correlate histology to the gross images of the tumor. H&E stained sections in Figure 2 show representative images of the three sampled areas. Area #1 demonstrates tissue of a spindle cell neoplasm of neural di fferentiation arranged in fascicles with elongated hyperchromatic nuclei and a mild to moderate amount of cytoplasm. The tumor purity of this sample was >95%. Area #2 shows spindled cells in a background of hemorrhage, a finding commonly seen in these high-grade tumors with a tumor purity of >95%. Area #3 represents an area of necrosis, another characteristic finding for MPNST. This sample showed >95% tumor purity.

**Figure 2.** H&E stained sections of the biopsy sites. H&E stained sections (20X) show areas (#1) of relatively uniform, spindled cells with fascicular growth pattern, characteristic for MPNST. Sampled area #2 shows evidence of hemorrhage within the tumor, a feature commonly seen in MPNST. Area #3 shows abundant tumor necrosis.

#### *3.3. Whole Exome Sequencing (WES), RNA Sequencing (RNA-Seq), and Copy Number Analysis*

We first interrogated the sequencing data to identify the germline NF1 variant within this tumor. Figure 3 shows a lollipop plot identifying the patient's likely NF1 germline variant based on exclusion of any variants with minor allele frequency >0.05 in the 1000 genomes database. Next, to

investigate intra-tumoral heterogeneity within the sample, RNA sequencing of the three sample sites was performed and is shown in Figure 4.

**Figure 3.** Location of NF1 germline variant. One intronic germline variant, NC\_000017.11:g.31296270C>T (rs11080149). was identified and is depicted in this figure.

**Figure 4.** RNA-Seq Heatmap. Normalized read counts by counts per million (CPM) in differentially expressed genes are depicted here. Distinct gene expression profiles can be appreciated in each biopsied area. Each column is depicted as list of genes.

Distinct gene expression profiles were observed in each of the areas sampled. The top 16 differentially expressed genes are listed in Table 2 and include a number of genes involved in transcription and translation. We next performed a copy number analysis of the three biopsy sites to determine whether or not different copy number alterations were observed in each area (Figure 5). Distinct copy number signatures can be appreciated in each of the three samples further illustrating intra-tumoral heterogeneity. Additionally, we evaluated the single nucleotide variants found in each of the samples. This broad overview of all somatic variants is depicted in the waterfall plot in Figure 6. Again, distinct somatic variants can be appreciated across different areas. We next explored the potential significance of these variants through further bioinformatics analysis. While the biological significance of each of these variants is uncertain, there is evidence that some of these variants may play a role in the pathogenesis. For each variant in a coding region, CBioPortal [47] was queried for each gene to determine if the somatic variant was in a functional domain. Additionally, the RNAseq data was queried to determine if the variant in a specific area of the tumor influenced the gene expression of that gene in a specific area. Finally, SIFT and Polyphen were used to predict pathogenicity. Table 3a,b list the somatic variants in the coding region that may play a role in the pathogenesis of this tumor based on the above criteria. For those mutaions in non-coding regions, the Ensembl Variant E ffect Predictor [33] was used to determine whether or not the variant would be predicted to a ffect gene expression. All of the identified variants were classified as modifiers, indicating that pathogenicity prediction is di fficult, thus the e ffects of these variants are unclear. (Table 3c). Further details of the somatic variants can be found in Supplemental Table S1. Next, a gene ontology analysis was performed. To do this, a list of genes in copy number aberrant (CNA) regions was extracted. For each area, the list of genes located in the CNA regions intersected with the di fferentially expressed gene list reported in the RNA di fferential expression analysis, and PantherDB [45] was utilized to identify pathways that may be a ffected by these genes. Table 4 displays the unique genes in each area with copy number aberrations and alterations in gene expression. Genes depicted in Area 1 have been reported in the literature to serve a myriad of functions in tumorigenesis, including base excision repair, nucleotide excision repair, and alternative splicing [48–55]. Those in Area 2 are involved in several di fferent pathways, including transcriptional regulation in addition to ribosomal and proteasomal function [56–60]. Finally, the genes in Area 3 consist of several ribosomal subunits and small nucleolar RNAs, suggesting that both translation and transcription are uniquely a ffected compared to other areas [61–63]. This analysis suggests that there may be di fferent functional programs at play across the three areas. Next, we manually reviewed the data to look for changes in other known drivers of MPNST including TP53, ATRX, EED, SUZ12, and CDKN2A. There were no copy number changes or somatic mutions in any of these genes. Finally, we performed a careful manual review of all of the shared and unique somatic variants and copy number alterations in each area in order to develop a predicted clonal evolution. Figure 7 depicts the predicted phylogenetic tree of the subclones from each area, representing the likely clonal evolution of the tumor.


**Table 2.** Top Differentially Expressed Genes. The gene-specific analysis was used to test for differential expression of genes or transcript between sample regions in Partek ® Flow ®. Statistical cutoff are made by these following parameters: *p*-value <= 0.05; FDR step up <= 0.05; Fold Change <−2 or >2.

**Figure 5.** Copy Number Variation Plot. Copy number variation plots for each biopsied site demonstrate distinct copy number signatures.

**Figure 6.** Somatic Variant Waterfall Plot. All somatic variants displayed on a waterfall plot. Each row represents a gene. Distinct somatic variant signatures are appreciated.

*CSK* 2 Chr15:74798671 frameshift p.(Glu25fs) Y *TSPAN9* 2 Chr12:3283047 frameshift p.(Leu218fs) Y







**Figure 7.** Phylogenetic Tree. A predicted phylogenetic tree of the tumor subclones.
