NEW YORK (GenomeWeb) – The Arkansas Research Alliance has received a $764,000 contract from the US Food and Drug Administration for a feasibility study aimed at developing improved methods and tools for screening for lung cancer.
The alliance is a consortium of five research universities in Arkansas, namely the University of Arkansas for Medical Sciences, University of Arkansas at Little Rock, Arkansas State University, University of Arkansas at Pine Bluff, and the University of Arkansas as well as the FDA's National Center for Toxicological Research.
The funds will support the first year of a planned three-year project that aims to analyze genomic data from smokers, non-smokers, and individuals with compromised immune systems as well as animal models to understand the genetic mechanisms that underlie certain lung cancer subtypes.
Donald Johann, principal investigator and associate professor of medicine and biomedical informatics in the UAMS hematology and oncology department, told GenomeWeb that for the first year of the project the researchers plan to work with four to eight current smokers with non-small cell lung cancer of the adenocarcinoma subtype. In years two and three, they hope to enroll an additional six to 12 individuals per year for the study. Additional funding for the remaining two years is contingent on the project's performance in the first year.
Ultimately, the goal is develop more effective screening methods that can detect tumors earlier so that patients can get treatment faster, Johann said. Solid tumors can grow for 20-30 years before they can be detected clinically, he said, and by then the disease is usually very advanced and much harder to treat.
"The field of precision medicine is really bringing a new way of looking at disease and much more effective therapies [that will enable us] to look at a patient's cancer in a very rational type way and assign therapies rather than just looking at it in a categorical way and assigning therapies categorically," he said.
Furthermore, for nearly two decades, Arkansas has reported a higher than average incidence of lung cancer, so there is a public health impetus for the project, he added.
Hoping to shorten the time to diagnoses, the investigators plan to use liquid biopsies to try to detect tumor-shed products in patients' blood samples that indicate the presence of cancer. Liquid biopsy testing will be repeated multiple times in patients over the course of the study, and data will be collected and analyzed to track tumor evolution, changes in mutation profile, and treatment response. The benefits of this type of testing is that it's much lower risk than a standard cancer biopsy, which can compromise patients' well-being, and they are also quick and convenient, requiring a simple blood draw that can be repeated as often as necessary, Johann said. It also avoids sampling bias associated with standard tumor biopsy because of the spatial heterogeneity of tumors, which the research hope will provide a more balanced picture of the tumor.
The researchers also plan to use a co-clinical trial approach that involves combining and analyzing data from patients and a number of associated mouse models that each patient has been paired with. Specifically, Johann said that each patient will have at least a small group of patient-derived xenograph mouse models — mice with cancers that are derived from the patient's tumor; and then if the patient's tumor has classic mutations, such as KRAS mutations, then they'll also be paired with genetically engineered mouse models for the mutation. The exact number of mice selected per patient will be based on guidelines laid down by the National Institutes of Health. Both patients and their mouse models will receive the same treatment and undergo the same types of genetic testing, and the data collected will be integrated and analyzed to explore the effectiveness of the therapies.
For this project, the researchers will collect routine clinical information from patients. They'll also perform extensive molecular profiling of solid tissue biopsy samples collected from the patients and their associated mouse models including performing whole-exome sequencing, RNA sequencing, and methylation sequencing, Johann told GenomeWeb. They'll also do liquid biopsies on the patients and mice and perform DNA-based sequencing on these samples, he said. For methylation, the researchers plan to use the Illumina 450K BeadChip array platform, and for liquid biopsies, they'll do a combination of both DNA sequencing and digital droplet PCR, Johann said. Whole-exome, whole-genome, and RNA-seq will all be performed on an Illumina HiSeq 3000.
The clinical trial itself will be run at UAMS, which has the requisite infrastructure in place to run the program. However, all consortium members will collaborate on the development of the bioinformatics pipelines and high-performance computing infrastructure needed to analyze, interpret, and store the terabytes of data that the researchers expect to generate. All of the computational pipelines that will be constructed for the project will follow best practices put together by researchers at the Broad Institute and the McDonnell Genome Institute at Washington University in St. Louis, Johann said. The list will include pipelines for whole-exome sequencing, RNA-seq, and methylation data analysis.
The data analysis will be done using a combination of open-source and custom bioinformatics tools to handle alignment and variant calling tasks. For variant calling, the researchers are taking a consensus approach that will involve evaluating the results of multiple variant callers. Specifically, they'll use Strelka, an Illumina-developed method for somatic SNV and small indel detection from matched tumor-normal samples and MuTect, which is used for identifying somatic point mutations, Johann said. They'll also use a consensus approach to identify insertions and deletions in the data using Strelka and another tool called Indelocator, which is designed for calling short indels in NGS data.
For RNA sequencing data, the researchers plan to use the Tuxedo software suite, which includes tools like TopHat, Cufflink, and Cuffdiff, as well as the STAR aligner. For analyzing methylation data, they'll likely use the Minfi software or one of its derivatives, Johann said. The researchers are also making use of data from existing public repositories including the Ensembl and Entrez databases, Johann said.
Furthermore, the researchers are currently developing a series of custom tools that will allow integrative analysis of the different omics and clinical datasets collected as part of the project as well as a custom database for hosting data after its been processed, curated, and cleaned — though the FDA will decide whether or not the data from this project will be made publicly available, Johann said.
In terms of hardware, UAMS has a local high-performance compute cluster that it will use for the project, but it also has a business associate's agreement with Google, which is providing some HIPAA-compliant cloud space for use by the researchers, Johann said.