AGBT 16, Day 1: Into the Clinic
Thursday AM, February 11, 2016
AGBT Guest Blogger: Meredith Salisbury(Bioscribe)
Hundreds of not-yet-partied-out AGBT attendees descended on the Mediterranean Ballroom this morning to learn about the latest in clinical genomics from those in the trenches. What they heard confirmed that genomics continues to have a real and measurable impact in patient care, but that significant challenges remain — many of them associated with the non-trivial vagaries of biology.
From the British Columbia Cancer Research Centre, Sam Aparicio kicked off the session with a talk about clonal dynamics and evolution in cancer patients. Genome sequencing, he said, has been extremely important for revealing how cancers change over time and how heterogeneous tumors can be. But there’s still a lot to learn about the individual clones behind cancer operate, and for that Aparicio has turned to NGS-based and single-cell approaches to characterize clonal dynamics, population structure, evolution, drug resistance, and more. In a study of patients with triple negative breast cancer, his team discovered that patients who appeared clinically the same had a wide variety of clonal complexity that may shed light on why similar patients can have radically different disease outcomes. By grafting several tumor fragments from breast cancer patients into mice, he was able to show that clone location matters, and that clonal prevalence in mouse models often differed dramatically from that seen in the donor patients. He also found that clonal dynamics appear to be deterministic for genotype. Aparicio is now using a microfluidics approach to investigate individual cells, with hopes of scaling the method to work across many single cells to reconstruct their genomes at single-base resolution. In the future, Aparicio believes that clonality features may be an important part of cancer diagnosis and prognosis.
Continuing the cancer theme, Luis Diaz from Johns Hopkins presented data on somatic mutations in solid tumors and the utility of circulating tumor DNA to hone medical treatment for these patients. In a current study based on a successful digital PCR project from several years ago, Diaz and his colleagues are hoping to show that liquid biopsies can be used to identify the patients at greatest risk of recurrence (and who may benefit from aggressive chemo treatment) as well as patients at minimal risk. In the study, ctDNA is analyzed after patients have had their tumors resected. Looking at 14 tumor types across more than 680 patients, Diaz told attendees that his team was able to detect a significant proportion of cases likely to recur just by measuring ctDNA in plasma for a range of cancers, including breast, endometrial, ovarian, and lymphoma. He noted that major limitations include tumor location (higher rates of detection come from sera closest to the tumor), the likelihood of the tumor to shed DNA, and the lack of targeted therapeutics. In a separate project, he recounted the remarkable story of using a checkpoint inhibitor drug on patients with mismatch repair tumors marked by very high rates of mutation. More than half of these patients, for whom virtually all other treatments had failed, saw a noticeable response to the therapy.
A talk from Memorial Sloan Kettering’s Franck Rapaport focused on a rare and usually fatal type of adult leukemia, B-cell ALL. He used the FoundationOne Heme platform to study gene associated with this cancer and with gene fusion events across nearly 200 samples, identifying both known and novel genetic changes responsible for the cancer. His team found that samples with two common mutations, the BCR-ABL1 fusion and an MLL fusion, never had other mutations associated with these cancers. A third group of samples was characterized by a long tail of various mutations that explained the phenotype. Taken together, he reported that this information has relevance for prognosis and treatment of this type of cancer.
In the final cancer talk, John Martignetti from the Icahn School of Medicine at Mount Sinai spoke about a new approach to classifying patients with endometrial cancer. Current methods involve traditional pathology, but Martignetti believes that molecular markers could have a real impact in accurately sorting higher-risk patients from the others. Using a mix of technologies, from Ion Torrent sequencing to Swift Biosciences and more, he and his team used gene panels to identify biomarkers, which were later annotated for functional significance and used to train prediction algorithms. The final tools showed good concordance for prognostic prediction, and may be useful along with circulating tumor DNA for patient surveillance over time. Martignetti also showed brand-new data from a lavage approach to find tumor DNA in patients, noting that he expects to move to single-cell analysis in the future.
The session also included two non-cancer talks, with the first from Charles Chiu at the med school at the University of California, San Francisco. He presented data on the use of metagenomic sequencing for diagnosing infectious disease from a recent project designed to improve diagnosis of things like pneumonia and sepsis, for which the cause often goes undetected. Chiu’s wide-net approach is a dramatic shift from the ‘one test for one bug’ mentality that has long dominated infectious disease diagnosis, he said. With clinical genomics and a new bioinformatics pipeline called SURPI, his team was able to rapidly diagnose infectious diseases from plasma, cerebrospinal fluid, and bronchial lavage samples — often in patients who had undergone dozens of tests without getting conclusive answers. One limitation he pointed out is the lack of high-quality reference genomes in databases for matching pathogen DNA. Looking ahead, Chiu said that he plans to continue emerging work with nanopore sequencing for infectious diseases and to expand the clinical study to three hospitals and add a cost/benefit analysis.
In the final talk of the session, Katia Sol-Church from the Nemours A.I. DuPont Hospital for Children presented the recent discovery of new syndrome responsible for a rare recessive skeletal dysplasia. Next-gen sequencing failed to identify the genetic mutation behind the disorder due to a gap in the human reference genome, Sol-Church said, but her team did the legwork and implemented Sanger sequencing to find the series of deletions in XYLT1 that explained the patients’ phenotype. Bisulfite sequencing was an important second step, revealing a methylation pattern that explained why children who were heterozygous for the deletion had the phenotype. Moving forward, her team may try to de-methylate the region to prove their hypothesis.