Development and validation of a directed, rapid and highly multiplexed assay to detect copy number variations in clinical samples

AGBT Meeting, February 25-28, 2015

Authors

Josh D. Haimes1, Namitha Manoj1, Abel Licon1, Joshua A. Stahl1, Brian A. Kudlow1, Maria Cueller2, Milhan Telatar2

1ArcherDX, Inc., Boulder, CO; 2Molecular Diagnostics Lab, City of Hope Medical Center, Los Angeles, CA


Abstract

Copy number variation (CNV) results from oncogene amplification or tumor suppressor gene deletion and is a common mode of gene deregulation in cancer (1). Several techniques have been employed to determine copy number, including array comparative genomic hybridization (aCGH) and quantitative PCR (qPCR); however, none of these methods are amenable to high-throughput, directed CNV detection. We developed a directed next-generation sequencing (NGS)-based method to rapidly and quantitatively measure the copy number of tens and potentially hundreds of genes simultaneously. This complete workflow, found in the Archer™ Universal DNA Kit, is powered by Anchored Multiplex PCR (AMP™) chemistry and processes dozens of samples in about 6 hours. By ligating a molecular barcode to randomly fragmented input DNA and then using AMP to simultaneously enrich for several regions of each target gene in test samples by counting unique molecular barcodes associated with each target region.

We validated our methodology with a 25-gene panel on a subset of NCI-60 cell lines by comparing our copy number measurements to those determined by both aCGH and qPCR. Results from both orthogonal methods strongly correlated with data from our NGS-based method. We multiplexed hundreds of samples on a single MiSeq® run and detected CNVs, both amplifications and deletions, of 2X magnitudes (and often lower) at extremely high confidence, indicating that this panel is amenable to highly multiplexed screens of potentially hundreds of samples. Furthermore, we demonstrate that our NGS-based CNV detection workflow and analysis is compatible with DNA extracted from formalin-fixed, paraffin embedded (FFPE) samples, suggesting that this system could be adapted for use in clinical applications.

Published in conjunction with City of Hope™

Figure 1 - Introduction

A. Genomic regions interrogated in the 25-gene oncogene/tumor suppressor panel. Target genes are listed in red, and control genes are listed in blue.

Figure 1A Genomic Regions

B. Overview of CNV detection strategy. Target gene regions are uniquely tagged with molecular barcodes (MBCs) and enriched for NGS analysis by Anchored Multiplex PCR (AMP). The number of unique MBCs associated with each target region is used to deduce copy number for each target region relative to a control sample.

Figure 1B Overview of CNV Detection Strategy

C. Archer Universal DNA Workflow. Library preparation is carried out using Archer's fully lyophilized reagents. Input DNA, as little as 10 ng, is enzymatically sheared, tagged and amplified into sequencer-ready indexed libraries, allowing hundreds of samples to be multiplexed. Total library preparation time is about 6 hours for dozens of samples.

Figure 1C Universal DNA Workflow

D. Overview of Archer Analysis for CNV detection. CNV analysis is fully automated, permitting sensitive detection of CNVs without a need for bioinformatics training.

Figure 1D Overview of Archer Analysis for CNV detection

Figure 2 - Assay Validation on MCF-7 and HT-29 Cell Lines

A. Archer NGS-based CNV assay correlates closely with published aCGH data. Copy number of target genes was measured in MCF-7 and HT-29 cell lines with the Archer CNV assay, and copy number calls were compared to published aCGH data (2). The correlation between copy number calls is shown in the graphs.

Figure 2A Graph 1
Figure 2A Graph 2

B. Archer CNV assay correlates closely with qPCR copy number measurements. Copy number of target genes was measured in MCF-7 and HT-29 cell lines with the Archer CNV assay. A single qPCR probe for each target was then used to measure copy number in the same genomic DNA samples. The correlation between copy number calls is shown in the graphs.

Figure 2B Graph 1
Figure 2B Graph 2

C. High concordance between Archer NGS assay and published aCGH data over complete panel. Table shows copy number calls across complete panel in MCF-7 cells.

Gene Copy Number (aCGH) Copy Number (Archer)
APBA1 0.66 0.84
APP 0.69 0.82
ATRN 0.75 0.82
CCND1 1.36 1.22
CDKN2A 0.19 0.10
CLNK 0.88 0.82
EGFR 0.72 0.83
ERBB2 0.76 0.57
FGFR1 0.69 0.60
FGFR2 1.00 1.01
FGFR4 1.57 1.46
GLTP 1.30 1.03
LONP1 0.82 0.55
MET 1.13 1.28
MYB 0.68 1.06
MYC 2.38 3.05
MYCN 1.00 0.95
NEDD4 1.29 1.42
NETO1 0.34 0.54
PTEN 1.00 1.21
RREB1 1.28 1.18
RTCB 1.00 0.98
SGCG 1.14 1.17
UNC79 1.38 1.55
XYLB 1.00 0.82
Pearson 0.926

Figure 3 - Archer CNV Assay is Highly Multiplexable

A. Multiplexing options available with Archer Universal DNA Kit. Archer NGS assays facilitate highly multiplexed CNV measurements with 144 P5 adapter indexes and 8 P7 adapter indexes for a total of 1,152 dual-index combinations.

'
P5 1 P5 2 P5 3 P5 4 ... P5 144
P7 1 1-1 1-2 1-3 1-4 ... 1-144
P7 2 2-1 2-2 2-3 2-4 ... 2-144
P7 3 3-1 3-2 3-3 3-4 ... 3-144
P7 4 4-1 4-2 4-3 4-4 ... 4-144
P7 5 5-1 5-2 5-3 5-4 ... 5-144
P7 6 6-1 6-2 6-3 6-4 ... 6-144
P7 7 7-1 7-2 7-3 7-4 ... 7-144
P7 8 8-1 8-2 8-3 8-4 ... 8-144

B. Highly sensitive CNV calls are achievable even at minimal sequencing depth. MCF-7 CNV libraries were sub-sampled to sequencing depths equivalent to indicated multiplexing (assuming 15M productive reads per MiSeq run). P-values for significant CNV calls are plotted against the read depth. Even at sequencing depths equivalent to running 1500 samples on a single MiSeq run, modest CNVs (e.g. FGFR1) are called with p < 0.01.

Target Copy Number
CDKN2A 0.09
ERBB2 0.53
FGFR1 0.55
FGFR4 1.34
MYC 2.81

Figure 4 - A Simple QC Assay Predicts Ability to Call CNVs in FFPE Samples

A. Overview of qPCR input QC assay. Probe-based measurement of quantity of amplifiable DNA present in an FFPE input. Amplicon size is 66 bp.

Figure 4A Overview of qPCR input QC Assay

B. qPCR QC assay predicts ability to call CNVs. Receiver operating characteristic (ROC) plots demonstrating predictive power of input QC assay in identifying samples in which a CNV of magnitude 3X(left) or 4X (right) could be called by the Archer CNV assay.The minimum magnitude of a CNV is a function of baseline noise, which increases as input quality decreases. Data is representative of 43 archived FFPE samples.

Figure 4B Graph 1
Figure 4B Graph 2

C. qPCR assay predicts magnitude of detectable CNV. The sensitivity of the Archer CNV assay, expressed as the minimum detectable CNV, for a given sample is linearly correlated with the Cq value of input QC qPCR. Poor quality samples result in higher input Cq values, and this is correlated with reduced CNV sensitivity. Total input quantity was equivalent for all samples.

Figure 4C qPCR predicts magnitude of detectable CNV

Figure 5 - Detection of CNVs in Patient-Derived FFPE Samples

A. Copy number report for four clinical FFPE samples. Copy number variants identified in each of four clinical samples are shown. Each bar represents a single probe, where the magnitude of the copy number (log 2) corresponds to the height of the bar and the significance is shown by color (z-score, scale below).

Figure 5A Z-score scale
Figure 5A CNV Graph 1
Figure 5A CNV Graph 2
Figure 5A CNV Graph 3
Figure 5A CNV Graph 4

B. Description of the samples in A. Input quantity and expected CNVs for each sample in A.

Sample pPCR QC Input (ng) Input Type Tissue Tumor Cellularity CNVs
JH3741 (Pass) 100 FFPE N/A N/A CCND1+, MYC+
JH3745 (Pass) 100 FFPE Breast 60% ERBB2+
JH3803 (Pass) 200 FFPE N/A N/A PTEN-
JH3742 (Pass) 40 FFPE N/A N/A None

C. qPCR confirmation of detected CNVs. CNVs detected in the three samples were validated by qPCR with a single primer set for each of the indicated genes.

Sample Target qPCR AMP
JH3741 CCND1 10.82 5.69
CDKN2A 1.58 1.13
MET 1.07 1.43
MYC 16.63 6.19
JH3745 CDKN2A 0.46 0.98
ERBB2 4.05 7.88
MET 0.54 1.23
JH3803 ERBB2 0.93 0.90
PTEN 0.56 0.57
FGFR2 1.01 0.85

Figure 6 - Screen of 47 Patient-Derived FFPEs Identifies Several CNVs

A. Archer assay detects copy number variants in patient samples. Forty-seven archived clinical FFPE samples were screened with the qPCR QC assay. Libraries were generated and sequenced for the 21 samples that passed the qPCR QC assay as well as 10 of the 16 samples that failed QC. Samples are ordered according to input quality determined by qPCR QC. QC pass threshold was determined by ROC analysis. Copy number is reported for all samples; those with greater than 2-fold copy number for control genes in samples that passed QC was 1.03 +/- 0.33 and for samples that failed QC 1.70 +/- 3.42, indicating that samples that failed QC have insufficient signal-to-noise ratio to support CNV calling.

Figure 6A Copy Number Variants in Patient Samples

B. qPCR confirmation of selected CNVs. Selected CNVs identified by Archer CNV assay were confirmed by qPCR. Copy number measurements tightly correlate with qPCR measurements for high quality samples (top) but not for samples that fail input QC (bottom).

Figure 6B Graph 1
Figure 6B Graph 2

Conclusions

  1. Archer AMP technology permits sensitive and quantitative CNV detection in cell lines and clinical FFPE samples.
  2. The 25-gene panel used in this study is highly multiplexable and successfully detects known copy number variants.
  3. A qPCR-based input QC allows pre-screening of archived FFPEs for samples of adequate quality.

References

  1. Chromosomal abnormalities in cancer. Frohling S, Dohner H, N Engl J Med. 2008 Aug 14;359(7):722-34.
  2. High resolution copy number variation data in the NCI-60 cancer cell lines from whole genome microarrays accessible through CellMiner. Varma S, Pommier Y, Sunshine M, Weinstein JN, Reinhold WC. PLoS One. 2014 Mar 26;9(3):e92047.

For Research Use Only. MiSeq® is a registered trademark of Illumina, Inc. Archer™ and FusionPlex™ are trademarks of ArcherDX, Inc.

Subscribe to stay in the loop







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.