October 15, 2016 (Vol. 36, No. 18)
Comprehensive Sequencing and Analysis of Circulating, Cell-Free DNA in Longitudinal Studies
With the advent of liquid biopsy assays, it is now possible to monitor the treatment response of oncology patients in research studies, where mutations found in primary tumors can be tracked and novel mutations that appear during disease progression can be identified.
Liquid biopsy assays utilize methods that detect low frequency mutations present in cell-free DNA that is typically limited to low input quantities of only 10–20 ng. Most incorporate targeted next-generation sequencing (NGS) to enable cost-effective, deep coverage of the target loci for sensitive detection of low frequency variants.
Targeted NGS can be performed using either a multiplexed amplicon workflow or a hybridization capture workflow, the former being ideal for targets in the Kb size range, where the latter is suited for targets in the Mb size range. In order to achieve sensitivity across the target region, the assay must produce uniform, comprehensive coverage of the target region from low DNA input quantities. A key aspect of achieving this sensitivity is through proper sample collection and storage prior to analysis.
In addition to sensitive mutation detection, proper tracking of samples for liquid biopsy studies is critical. At the onset, a minimum of three samples from the same individual are analyzed for their genetic profile; tumor tissue, normal tissue, and circulating, cell-free DNA (cfDNA), followed by monitoring of cfDNA and other biopsies during longitudinal studies. Proper tracking is achieved by determining a SNP genetic fingerprint to confirm that samples are properly matched. This can be readily obtained when sequencing a large targeted panel, but not typically when sequencing a small targeted panel, thus requiring a separate SNP-based assay and ensuring proper labeling of datasets to accurately maintain sample source.
As more studies are designed to track response rates in oncology patients, it is critical to select sensitive assays with sample tracking tools for these longitudinal studies.
Sample Identification
Searching for an integral, fail-proof method of sample identification combined with an efficient, targeted enrichment technology that allows low frequency mutation detection from cfDNA samples, Swift Biosciences developed the Accel-Amplicon™ Sample_ID panel. A genetic fingerprint provided by the 104 exonic and gender specific amplicons is ideally utilized as a low percentage spike-in to Swift Accel-Amplicon Panels such as the 56G Oncology Panel v2. This results in sample identification from low depth sequencing of germline targets while still enabling high depth of coverage for somatic mutation detection.
This provides an efficient, single-tube assay for analyzing somatic mutations in oncology specimens while generating the genetic fingerprint within the same sequence file. The 56G Oncology Panel v2 includes 263 amplicons that cover both hotspot loci and contiguous regions over 56 oncology-related genes, which represent >16,000 single nucleotide variants referenced in the COSMIC database. This panel can achieve a 1% limit of detection when using 10 ng of cfDNA, and the workflow consists of two incubations in a single-tube format to reduce the likelihood of error in sample tracking.
Library Preparation
When a liquid biopsy study demands a larger target region to include loci that enable discovery of novel variants or target genes implicated in disease, Swift Biosciences offers the Accel-NGS® 2S Hyb library preparation kit that can be coupled with various commercially available hybridization capture panels for target enrichment such as IDT, Agilent, or NimbleGen.
Cell-free DNA is typically extracted from blood plasma or other body fluids and predominantly produces fragmented DNA at approximately 170 bp. The Accel-NGS 2S kit is ideally suited for cfDNA because it offers superior efficiency of NGS adapter ligation at low inputs, typical of cfDNA, with up to a 90% library conversion rate to maximize the likelihood that low frequency variants will be present in the library. The kit has been validated to detect a 1% mutation frequency from 10 ng cfDNA when paired with the IDT xGen® Pan-Cancer panel. The high efficiency also leads to fewer PCR duplicate reads when sequencing low input liquid biopsy samples.
In addition, Swift now offers a unique adapter indexing module that includes a sample index for multiplexed sequencing and a molecular index (MID) to uniquely tag each input molecule captured into library during ligation. Use of the MID indexing module is particularly advantageous for cfDNA because the molecular identifier can be used for accurate marking and removal of PCR duplicates from the data while preserving the naturally occurring fragmentation duplicates, which should be included in data analysis.
The molecular identifier can also enable higher specificity for low frequency variant detection by identifying random PCR and sequencing errors for their removal prior to variant calling. This capability requires a depth of sequencing where each unique fragment has multiple PCR duplicates that can be used to generate a consensus sequence, thereby raising the level of confidence in variant calling.
Figure 1. Even at low sequencing depth, the minimal sequence-dependent bias of the Accel-NGS® 2S adapter attachment results in even coverage across the genome. Data was normalized to mappable chromosome content.
Methylation Analysis
An alternative approach to detection of somatic mutations in liquid biopsy is to determine methylation density from cfDNA. Researchers have identified two unique methods of methylation analysis in oncology, one is to measure global methylation status as tumor-derived cfDNA is hypomethylated relative to cfDNA derived from healthy individuals, and second where distinct methylation signatures can shed light onto the tissue of origin of the cfDNA fragments. The advantage of using global hypomethylation patterns to determine overall cancer burden is the process requires only 5 ng of cfDNA and a low depth of sequencing. With these applications, whole genome bisulfite sequencing (WGBS) or targeted bisulfite sequencing by NGS is performed.
Bisulfite treatment of DNA introduces a sequence change to distinguish methylated vs. nonmethylated cytosine. This process both denatures and fragments the DNA, so it is critical to choose a library prep method that can capture the limiting complexity available in liquid biopsy samples. Most workflows treat a prepared library with bisulfite, which fragments the majority of library molecules leading to significant loss.
Swift Biosciences has a post-bisulfite library prep that generates the NGS library on single-stranded bisulfite converted fragments to maximize recovery of limiting samples by greater than 10-fold compared to traditional methods. The Accel-NGS Methyl Seq kit has been validated to track hypomethylation status of oncology sample derived cfDNA from 5 ng of cfDNA and 10M paired end reads.
Figure 2. Circos plot represents hypomethylation status on chromosomes 1–22 between five healthy controls and a metastatic colorectal adenocarcinoma sample.
Sample QC
Regardless of the application used for performing NGS analysis on liquid biopsy samples, proper handling of samples and QC to determine the quantity and integrity of each DNA sample is important. We have validated methods using Streck Cell-Free DNA BCT collection tubes for isolating blood plasma, paired with a commercially available circulating nucleic acid DNA extraction kit.
All Swift library preparation kits include an ALU repeat qPCR assay for determination of liquid biopsy sample quantity and integrity. The assay uses two amplicons, 115 bp and 247 bp, where DNA concentration is determined using the 115 bp assay, and where the ratio of 247 bp/115 bp determines whether significant high molecular weight cellular DNA has contaminated the cell-free DNA fraction, which would reduce sensitivity of detecting mutations or methylation status of cell-free DNA. Since the sensitivity of detection of rare events is dependent on the number of input genome copies (1% allele frequency=30/3,000 copies from 10 ng input), and sample purity, sample QC is highly recommended.
Laurie Kurihara, Ph.D. ([email protected]), is director of R&D at Swift Biosciences.
