Targeted Sequencing Offers a Leap Forward in Cancer Research
UCSD researcher uses RainDance targeted sequencing of cancer hotspots to multiplex samples, boost sensitivity, specificity and utility, and dramatically lower per-sample costs.
When translational genomics expert Olivier Harismendy, Ph.D., joined the cancer center at the University of California, San Diego, in 2009, he quickly realized two things: first, that clinicians were eager for new technologies to better treat their patients; and second, that tremendous gains would be possible if he could figure out how to bring the benefits of next-generation sequencing to the clinic. “I knew from previous experience that if personalized medicine was going to appear, it was going to appear in the cancer field first,” recalls Harismendy, an assistant professor at UCSD.
Olivier Harismendy, Ph.D., Assistant Professor at University of California, San Diego Department of Pediatrics
That left him assessing his technology options. While Harismendy’s work would be a research project, he was intent on developing a pipeline that could one day have clinical utility. With that in mind, full genome sequencing was crossed off the list because it generated too much data to fit well in a clinical setting. Exome sequencing was still too expensive for routine use. In bringing next-gen sequencing to the clinic, Harismendy had to figure out “the most effective way to do it so people would use it and trust the results.”
Based on the state of molecular characterization of cancer samples at the time, Harismendy says that being able to study “even 20, 30, 40 genes would be a gigantic step forward.” So he headed to RainDance Technologies, a company he had already collaborated with on targeted sequencing methods.
Among the specifications Harismendy had in mind for the new approach was a significant improvement in sensitivity over the current standard of Sanger sequencing. Sanger was used regularly in the clinic, but even “a robust assay could only detect mutations present in about 20 percent of the DNA,” he says. Having taken a hard look at “the reality of the samples we can get in the clinic,” Harismendy knew that certain tumors were so diffuse that some samples might consist of 95 percent normal cells. In light of that, he was determined to find an assay far more sensitive than Sanger, and set a goal of detecting mutations in as little as 1 to 5 percent of a small biopsy or a heterogeneous tumor.
Along with collaborators at RainDance and Prognosys Biosciences, Harismendy set out on a project that would eventually be published in Genome Biology in December 2011 (“Detection of low prevalence somatic mutations in solid tumors with ultra-deep targeted sequencing”). The team used the COSMIC database to select cancer mutational hotspots, or “certain genes that are always mutated in the same domains,” Harismendy says. They chose more than 500 primer pairs that covered 71.1kb of sequence around 42 cancer genes, and that information was used to develop an assay on the RainDance microdroplet platform.
In addition to the “very robust, microdroplet-based PCR” that Harismendy had come to expect from RainDance, the team was able to eliminate much of the library prep — an important attribute for a workflow that could one day be translated into clinical use. “The way we designed this experiment with RainDance, there’s no library prep anymore. As soon as you’re done with your PCR, your PCR product already has the adaptor and bar code, so it’s very straightforward,” Harismendy says. “RainDance designed their instrument so it’s really error-proof. Everything is done automatically in a very streamlined protocol.”
Based on the study, Harismendy and his team determined that they could achieve their goal of detecting low-prevalence mutations in the 1 to 5 percent range while pooling up to eight samples in a single Illumina GAIIx lane. (Similar studies are now underway for MiSeq, and preliminary data on that platform — the first MiSeq data to be published — was included in the Genome Biology paper.) With that kind of multiplexing, costs per sample come down significantly. “This is the main reason to do targeting,” Harismendy says. “It really saves costs. For the same price you can obtain deeper coverage and sequence more.”
The team demonstrated better than 94 percent sensitivity and 99 percent specificity using the RainDance assay, an accomplishment that Harismendy attributes to the technology platform. “Sensitivity is directly linked to the performance of your targeting method,” he says. “What RainDance PCR enables us to do is to assay the different primer pairs in different droplets. That turns out to be a much more robust approach, with very little chance for the PCR primer pairs to interact, either between themselves or with each other’s product.” The assay’s sensitivity also stems from how equally represented the amplicons are: disparities could make it harder to detect low-prevalence mutations. “We had great uniformity of coverage,” Harismendy adds.
Given the results of this work, Harismendy has high hopes that targeted sequencing could one day help patients. “We definitely think this could have clinical utility in the future,” he says. “We hope to shift the paradigm by looking at the molecular profile of these tumors to increase the chances of these patients to receive the appropriate care.”
— Meredith Salisbury
Related Content: Ultra-Deep Targeted Sequencing and Potential Implementation in Cancer Webinar
On June 19th, Dr. Olivier Harismendy from UC San Diego discussed the development and implementation of an assay that targets ~100kb of mutational hotspots in cancer. He also showed its technical validity for the identification of low-prevalence mutations in heterogeneous cancer samples. Participants learned about the potential of this assay for a clinical environment and the challenges and solutions associated with its development and future implementation. Watch the webinar today.
