Comparing reference standards for NGS cancer genomics

Cancer genomics is rapidly moving away from pure discovery to a much more translational perspective. An important area to consider, as the tools being used and developed in research labs are deployed to a more clinical setting, is an understanding of the reproducibility, sensitivity and specificity of these tools. A big help in this is the availability of high quality reference materials, which are likely to form an important part of any clinical lab QA program. Several companies sell reference standards but the most interesting in the context of this post are those from Horizon Discovery, AcroMetrix and Seracare; each of which is briefly reviewed below.

Horizon Diagnostics approach is one of several

What does an NGS reference standard need to offer: First and foremost is reproducibility of manufacture and use, these materials must not change over time, from lot-to-lot, or between runs. Ideally they would be clinically relevant as well. Multi-step controls would be nice, one that can be spiked into the sample before extraction, one after, and perhaps a prepared library to control for sequencing and informatics. The three discussed below are really designed to be included after DNA extraction, and before library prep and are then processed through the same workflow as the patient samples. 

It will be up to each lab to determine the performance criteria,  but the reference materials need to work across different genome technologies and NGS instrument platforms, assays and bioinformatics pipelines. Sanger sequencing, qPCR, Nanostring, etc - the more platforms a control is compatible with the better. I'm not sure anyone has generated a phased control yet but that might become more important over time.

For now our own use is as a process control in our Exome pipeline, and we're evaluating how well we can determine the sensitivity of an exome capture based on observed/expected allele ratios.

Horizon Diagnostics: Horizon Diagnostics works with highly characterised parental cell lines. These are genetically modified using their proprietary genome editing platform to include clinically relevant cancer gene mutations, exactly as they occur in patient tumours. A single-cell mutant clone is bulked up for use. In the multiplex reference standard the same parental line has been engineered with multiple mutations, and these clones have been mixed to get defined ratios of normal:mutant alleles, which are verified using digital PCR. The cell lines and standards are available in multiple formats including gDNA , FFPE sections, etc.

The Quantitative Multiplex HDx Reference Standard reference standard contains precise allelic frequencies for 24 oncology targets. We've been using this for a while in my lab and are currently writing up our approach to using this to QC exome sequencing results. Watch this space!

The newer cfDNA Reference Standard is provided as fragmented human genomic DNA at the same average size of 160 bp as cfDNA derived from human plasma. It is not a perfect control as it also contains lots of fragments below the size cutoff as shown in their application note (see below). But it is likely to prove popular as I'm not sure there is anything similar on the market yet and ctDNA is hot!

Seracare: SeraCare works with synthetic plasmid DNA created for each targeted gene which contains the engineered mutation, as well as an “Internal Quality Marker”. This additional QC marker is a 6bp 5’-ACATCG-3’ sequence present around 25bp from the  mutation, and allows both features to be analysed from a single sequence read. These should be at the same alleleic frequency allowing assay performance and analysis methods to be evaluated, before downstream clinical interpretation.

The Seraseq Solid Tumor Mutation Mix-I (AF20)  includes 26 variants including SNVs, SNVs as part of homopolymer tracts, deletions and insertions at 20% allele-frequency

AcroMetrix: AcroMetrix employ their MegaMix Technology, a proprietary high-multiplex blending protocol - although finding any information on this has proven to be beyond me. If the information for Acrometrix on the ThermoFisher website is so hard to find and verify, I'd worry about the actual controls in a complex NGS experiment!

From the information I could find online the AcroMetrix Oncology Hotspot Control has "500+ single nucleotide variants (SNVs), multiple nucleotide variants (MNVs), insertions, and deletions on a genomic background of cell line GM24385. The variants are 1 to 40 nucleotides in length and have been confirmed by Sanger sequencing." The control is a mix of both synthetic and genomic DNA. Cell line GM24385 is an NIST-selected cell line and has been characterised by the Genome in a Bottle public consortium.

SlideShare has decks from both Horizon Diagnostics and Acrometrix.

Choosing a reference standard: We're working with Horizon controls already and I've not tested the others so can only talk about choosing a control after declaring my possible bias. The most important criteria are likely to be whether you will design your assay to a control or vice versa. You may have flexibility as to what gets targeted, but to get the most out of these reference materials we're all going ot need the companies making them to talk to the comapnies developing assays. The closer the coverage the better we'll be able to use them.

I like the genome engineering method applied by Horizon as I think it most closely recapitulates a real tumour, and the DNA is prepared using the methods you're probably applying in your own lab. However the use of plasmids and other synthetic DNA do give exquisite control over the kinds of variants engineered into the control so expect strong competition from both SeraCare and AcroMetrix.

I also realy like SeraCare's 6bp 5’-ACATCG-3’ “Internal Quality Marker” and can see this internal QC of the QC as being powerful.

It is still early days for the use of such controls outside of clincal labs so definitley "watch this space".

PS: don't forget to complete the SciBlogReaders survey by Paige Jarreau.