Targeted detection of copy number variants and fusion transcripts greatly expands the ability to detect oncogenic drivers in NSCLC

Association of Molecular Pathology Annual Meeting 2015

Authors

Josh Haimes, Laura Johnson, James Covino, Namitha Manoj, Marc Bessette, Elina Baravik, Abel Licon, Ryan D. Walters, Brady P. Culver, Joshua A. Stahl and Brian Kudlow

ArcherDX, Inc., Boulder, CO


Introduction

The genetic mutations of non-small cell lung cancer (NSCLC) include single nucleotide variations (SNVs), insertions and deletions (indels), copy number variations (CNVs) and rearrangements. A tumor is typically driven by a single class of mutations, and the proto-oncogenes in each class are diverse. Thus, to comprehensively and economically understand a given NSCLC sample, an assay needs to assess disparate mutation types multiplexed across many gene targets. We present results from a NSCLC screen using a pair of targeted NGS panels that, when combined, permit simultaneous detection of SNVs, indels, CNVs and rearrangements across oncogenes and tumor suppressors relevant to solid tumors. We found driver mutations harbored in many genes across each class of mutation, demonstrating the benefit of a targeted sequencing approach with comprehensive detection ability. In addition, we show the complementarity of two panels when used in combination to interrogate individual samples.


Download the Poster

Comprehensive tumor profiling with a pair of Archer NGS panels.

From total nucleic acid purified from FFPE, rearrangements are detected from RNA; in a parallel workflow, SNVs, indels and CNVs are interrogated in DNA. Anchored multiplex PCR allows for detection of rearrangements of oncogenes with novel and unknown partners, and molecular barcodes are used to collapse PCR amplicons into unique detection events, which gives an accurate representation of the input and aids in highly sensitive CNV detection down to two-fold changes.

Tumor profiling with a pair of Archer NGS panels

Archer Analysis provides push-button bioinformatics.

Features include high-level sample information such as QC result (1), ability to visualize variants (2), visual copy number reports (3) and structural consensus images for rearrangements (4).

1.

Archer Analysis QC Result (1)

2.

Archer Analysis visualizes variants (2)

3.

Archer Analysis visualizes copy number reports (3)

4.

Archer Analysis structural consensus (4)

PreSeq™ is a qPCR assay for FFPE input quality.

Nucleic acid from FFPE can be degraded and crosslinked; input quality drives NGS assay sensitivity. PreSeq assigns a DNA QC Score to report the amount of amplifiable molecules available for library generation, which is used to determine how much nucleic acid should be used in an NGS assay and is highly predictive of Archer assay sensitivity.

PreSeq is a qPCR assay for FFPE input quality

Methods

SNVs, indels and CNVs were detected in total nucleic acid extracted from archived FFPE NSCLCs across 67 oncogenes and tumor suppressors using the VariantPlexâ„¢ Solid Tumor targeted NGS assay. Gene rearrangements leading to functional mRNA fusion transcripts were detected in a parallel workflow with the RNA-based FusionPlexâ„¢ Solid Tumor assay. Variant, fusion and CNV detection was performed using the Archer Analysis bioinformatics software.


Results

SNVs, indels, CNVs and rearrangements identified as genetic drivers across 62 NSCLC FFPE samples.

Archived FFPE samples were screened by NGS for variants in DNA and RNA by Archer VariantPlex and FusionPlex assays, respectively. Driver mutations were identified in 40 (64.5%) of 62 samples. All samples in which a driver mutation was not identified also failed to meet Archer Analysis library complexity QC metrics; the false negative rate is increased in these libraries. Variants are reported by gene for SNVs (amino acid change and allele frequency) and CNVs (copy number fold change; 1 = 2 copies = normal). The allele frequencies for the EGFR indel are reported as well as rearrangement partners. Mutations are colored by class (blue = SNV; purple = CNV; green = indel; red = rearrangement) and by PreSeq DNA QC Score. * = stop codon.

Mutations identified as genetic drivers in 62 NSCLC FFPE samples
Color coded mutation type
DNA QC Score color scale

Distribution of mutation by class, gene and amino acid variant in 50 NSCLC FFPE samples.

Archived FFPE was screened by NGS for variants in DNA and RNA using Archer VariantPlex and FusionPlex assays, respectively. Fifty total driver mutations were found, some in very low-quality FFPEs. Frequency of variants is reported by class of mutation and broken down by gene within each class. KRAS and EGFR SNVs were the most frequent variants; these groups are further stratified by amino acid change. Number of samples represented by each chart is indicated (n).

Oncogenic drivers by mutation class
SNVs by gene
Rearrengement by gene
Indel by gene
Amplification by gene

Screen results summary

  • SNVs represent 62% of all driver mutations found in these NSCLC samples and were distributed across 7 genes. Together, VariantPlex and FusionPlex assays permitted detection of 60% additional drivers by also assaying for CNVs, indels and rearrangements.
  • Driver mutations were found in 64.5% of all NSCLC FFPEs interrogated with DNA QC Score greater than ~10
  • KRAS G12 SNVs comprised the single largest variant type; half of these variants are G12C
  • EGFR was the most commonly amplified gene (38™ of all amplifications); FGFR1 copy number gain was assessed but not represented in this sample group; CNV is a gene-diverse mutation class (7 idfferent targets)
  • EGFR del E746-A750 was the only indel driver detected, representing 8™ of all drivers
  • Rearrangements are the most rare mutation class in this sample group
  • Tumor suppressor variants were common, and highly sensitive CNV analysis resulted in high-confidence detection of single-copy losses; a myriad of other non-driver changes were also observed

EML4-ALK fusion harboring Crizotinib resistance mutation.

Archer Analysis reports fusion partners found in the Archer Quiver™ database of known fusions. Expression imbalance adds confidence to fusion call (1). Crizotinib resistance variant reported (2). Library passes QC metrics indicating sufficient library complexity was observed to meet sensitivity requirements (3). Additional support for fusion: visualization of transcript, translated protein sequence, link to Quiver Fusion Database, link to BLAST consensus (4). Fusion consensus sequence (5). FISH data supports fusion (6).

Archer Analysis reports fusion partners

4.

Additional support for fusion

5.

Fusion consensus sequence

6.

FISH data supports fusion

CNV analysis explains high allele fraction KRAS G12C variant.

A KRAS G12C variant was detected at 75% allele fraction in a 90% tumor cellularity sample. A 2x focal amplification found by CNV analysis explains the high allele fraction.

CNV analysis explains high allele fraction KRAS G12C variant

Copy number variants confirmed by overexpression.

CNVs were detected in NGS libraries prepared from FFPE DNA. Overexpression was confirmed by performing expression analysis (relative to normal lung) using NGS data generated from RNA in the same samples. NGS CNV data is shown for the MET-amplified sample as well as supporting FISH data.

Graph of Copy Number Variants

MET NGS
MET FISH

Orthogonal confirmation of EGFR L858R variant by detection in teo biomolecules: DNA and RNA.

. An EGFR L858R variant was detected independently by two primers in an NGS library prepared from FFPE DNA. The variant was confirmed in a parallel library preparation using RNA. Detection of a variant originating from two distinct biomolecules provides orthogonal validation.

Orthogonal confirmation of EGFR L858R variant

Conclusions

ArcherDX carefully developed the VariantPlex and FusionPlex NGS workflows to detect genetic variations in DNA and RNA, respectively, along with a complete bioinformatics software suite. When used together, the paired workflow delivers an information-rich dataset to simultaneously detect SNVs, indels, CNVs and rearrangements, which greatly increases the number of samples in which driver mutations can be detected in genetically complex tumor types such as NSCLC. The complementary nature of VariantPlex and FusionPlex assays permits orthogonal validation of results through observation in two biomolecules and increases confidence via expression imbalance of fusion transcripts and overexpression of amplifications.

How to contact us

Address

2477 55th Street, Suite 202

Boulder, CO 80301

Phone

Phone: (877) 771 1093

Phone: (303) 357 9001

All content © 2017 ArcherDX, Inc.

For Research Use Only. Not for use in diagnostic procedures. For Research Use Only. Not for use in diagnostic procedures.