Identification of Novel Germline and Tumor-Specific Nucleotide Variants and Copy Number Variation in Clival Chordomas by Exome Sequencing

Introduction: Chordoma is a rare, locally aggressive tumor with high recurrence rates. Available treatments are restricted to surgery and radiation. Our limited understanding of the tumor's underlying molecular pathophysiology compromises early prognosis and precludes the use of targeted molecular therapies.

Methods: We performed high resolution, paired-end, exome sequencing (Life Technologies SOLiD 5,500; > 100 times base depth) of DNA from five primary clival chordomas and matched blood of patients undergoing endoscopic endonasal surgery. BAM (Binary Alignment tab delimited) files were mapped to HG19 (Human Genome version 19), aligned (Burrows-Wheeler & Smith-Waterman algorithms) and analyzed using a battery of algorithms. Single nucleotide variants (polymorphisms:SNPs; substitutions:SUBs) and insertions/deletions (INDELs: < 5 bases) were tested for damaging effects using SIFT, SIFT-INDEL, and PolyPhen-2 algorithms. Base calls depths within exon borders were compiled across tumor normal pairs to statistically identify copy number variants (CNV) after correction for edge effects and false discovery rates (paired SAM test:FDR = 99%, 1.5-fold cutoff).

Results: We identified “damaging,” SUBs in the coding region of 36 genes in all germline and tumor samples. Among these were two mutations in the BRCA gene previously identified in breast and pancreatic cancers. All samples contained the benign SNP in brachyury (T), previously identified by Genome Wide Association Studies. Seven INDELs were delineated in all samples, including a deleterious, truncating insertion in cytoplasmic FMR1 interacting protein 2 (CYFIP2), a p53 inducible protein involved in apoptosis. Neutral frameshift insertions were present in the transcription factor piccolo (PCLO) involved in synaptic zone organization as well as integrin-linked kinase (ILK) and anaplastic lymphoma receptor tyrosine kinase (ALK) implicated in tumorigenesis. Five intronic SUBs were differentially detected in tumors vs blood comprising variants in signal transducing adapter molecule 2 (STAM2), Integrin αV (ITGAV), thymosin Beta 15B (TMSB15B), and 2 SUBs in general transcription factor IIi (GTF2I). Manual curation of the sequencing data identified these variants at significantly lower levels in blood samples.

Exon copy number analysis revealed significant losses in 150 genes including 15 transcriptional regulators. These included multiple homeobox and basic helix-loop-helix factors which orchestrate embryonic neural patterning and cell fate determination including iroquois homeobox 3 (IRX3), LIM homeobox 3 (LHX3), visual system homeobox 2 (VSX2), transcription factor 3 (TFE3), and left–right determination factor 1 (Lefty). CNV pathway analysis (IPA, Qiagen Corp) indicated a profound effect on the NOTCH pathway associated with neurogenesis and cancer including specific deletions in the NOTCH ligand, JAGGED2, and its receptor NOTCH1. Significant losses occurred in the tumor suppressors runt-related transcription factor 3 (RUNX3) and SIN3 Transcription Regulator (SIN3B) which have been associated with a variety of cancers (esophageal squamous cell carcinoma and olfactory neuroblastoma, respectively).

Conclusion: Our analysis revealed unique single base, INDEL and copy number variants in patients with clival chordoma. Preliminary analysis suggests underlying mechanisms associated with this tumor involve disruption of patterned neurogenesis at the transcriptional level and loss of critical tumor suppressor function. These findings provide several potential diagnostic and prognostic indicators for consideration as well as point to prospective targets for molecular therapies. Further study, including Sanger sequencing, will be needed to confirm these findings.