NEW YORK (GenomeWeb) – Researchers from Genentech have developed a new method for sensitive detection of circulating tumor DNA by combining a PCR amplification step with an approach they call microfluidics multiplex PCR sequencing (MMP-Seq).
Originally developed as a tool for sequencing challenging archived tissue specimens, MMP-Seq combines microfluidic multiplex PCR–based target enrichment using technology from Fluidigm with massively parallel sequencing.
In the team's new study, which appeared online earlier this month in the Journal of Molecular Diagnostics, investigators reported on their adaptation of MMP-Seq for the quantitation of ctDNA, via the addition of a PCR preamplification step and the use of a Bayesian model for error characterization and enhanced variant calling.
In the Genentech team's analysis, the MMP-Seq ctDNA method showed an overall sensitivity of 92 percent and specificity of 100 percent using mutation calls from matched tumor tissues as a benchmark. Investigators believe the approach could be especially promising for disease monitoring, an area of increasing interest for various players in the liquid biopsy space.
Yulei Wang, the study's lead author and a senior scientist in oncology biomarker development at Genentech, said in an email that ctDNA became a target of interest for her team soon after the initial development of MMP-Seq — both because of the promising clinical applications of liquid biospy and because of the challenge it poses to standard sequencing workflows.
PCR-based methods offer the highest sensitivity for detecting rare mutations in circulation, but development of individual assays limits detection to single mutations or at most a handful of targets.
But a wide variety of next-gen sequencing strategies have also emerged, along with error-correction methods to allow high-throughput, multiplex detection of ctDNA at very low frequencies. Leading approaches include targeted amplicon sequencing (TAm-seq), spun out from the University of Cambridge in 2014 and commercialized by Inivata, the barcoding-based Safe-Seqs (the Safe Sequencing System) developed at Johns Hopkins, and CAPP-Seq, a technique being commercialized by Roche in its Avenio circulating cell-free DNA assays.
Although Genentech is a Roche subsidiary, "we don't see CAPP-Seq and MMP-Seq as competing products," Wang said. [Various] ctDNA technologies have their pros and cons, and there may not be a 'one-size fits all' solution but rather an approach of 'fit for purpose' is more appropriate."
As opposed to CAPP-Seq, which Roche has moved forward as a set of specific kits targeted at the clinical research community, MMP-Seq might serve better as a flexible, customizable platform that users can tailor to address specific scientific questions in ongoing trials, she argued.
The original MMP-Seq method for preserved tissue samples requires splitting DNA samples into 48 PCR chambers for target enrichment. But for blood-based mutation detection this wouldn’t work, given the extremely low DNA input, Wang and her coauthors wrote. An additional preamplification step solves this problem.
Unlike some approaches that rely on hybridization techniques for pre-sequencing target enrichment, Wang and her team's method employs a PCR preamplification. The study authors argued that using PCR rather than a hybridization approach provides for a more flexible and more affordable platform. "While hybridization-based platforms generally have better specificity and sensitivity due to their ability to cross-reference multiple overlapping fragments and the low rates of hybridization errors, the resulting library preparation workflows are usually too costly and complicated to be widely adopted in the clinic," they wrote.
Wang estimated that a typical hybridization-based method would require 10 mL of plasma and cost $1,500-$2,000 per sample, whereas the MMP-Seq ctDNA workflow only requires 2 mL of plasma and costs less than $100 per sample.
"For clinical applications, such as monitoring disease progression and treatment response, multiple time points need to be analyzed, and cost-effectiveness is an important consideration for real-world applications," she wrote.
In their study, Wang and her colleagues used drug response monitoring as a test case for the clinical utility of the MMP-Seq plus PCR preamplification approach.
To calculate the sensitivity of the approach, the team compared calls using MMP-Seq to subjects' known tissue mutation status in a cohort of patients in the Phase I trial of the a Genentech investigational anti-MUC16 therapy DMUC5754A, assaying 81 plasma samples collected from 12 ovarian and 10 pancreatic cancer patients prior to and during treatment with the drug.
The team created a ctDNA MMP-Seq panel to interrogate the 29 SNVs previously identified in the matching tissues of these patients. They also developed a strategy for assessing the error rate of their method, capturing both position- and substitution-associated system errors, that could then be integrated back in to enhance the confidence and accuracy of variant calling.
Overall, eight out of the 10 patients with tissue mutations also had the same alterations present in their blood at above-background frequencies, including three patients who each had two driver mutations in both their tissue and the matching plasma samples. A patient who was known to be KRAS wild type based on tissue showed no KRAS mutations in circulation.
For one pancreatic cancer patient who did not have tissue mutation information available due to poor biopsy quality, blood testing revealed a high-frequency KRAS p.G12R mutation at baseline, which was present in all four subsequent plasma samples collected after treatment.
Using tissue mutation status as a benchmark, that translated to a sensitivity of 92 percent and a specificity of 100 percent for the MMP-Seq liquid biopsy method, albeit calculated from a very small group of test subjects.
The investigators also looked at the potential of the method for monitoring treatment response and disease progression, using one ovarian cancer patient and three pancreatic cancer patients who had both baseline and on-treatment longitudinal plasma samples available for ctDNA analysis.
The team compared what they saw in ctDNA with two protein markers that have also been proposed as disease biomarkers for ovarian and pancreatic cancers — HE4 and CA-19-9.
Changes in ctDNA levels, in general, matched those of the two proteins the authors wrote, except in one patient.
The results provide an initial proof of concept for using the MMP-Seq workflow to profile ctDNA as a surrogate for disease burden monitoring, the authors argued, especially in tumor types where no circulating protein biomarker is available.
"We believe our MMP-Seq ctDNA workflow constitutes a significant technical advance," Wang said. "The highly multiplexed nature of the method [yields] broad content, high sensitivity and accuracy, high sample throughput, and low per-sample cost," she added.
Moving forward, the Genentech team is planning to continue to test the method, but Wang said that it is too early to speculate on potential development into a commercial assay.