How I Found Clinically Actionable Variants in cfDNA from Cancer Patients
Wednesday, September 27, 2017 - 16:46
By: Eric Sieser, PhD - Field Application Scientist, Partek Incorporated

Precision medicine offers the promise of selecting treatments based on the genetic characteristics of an individual. Cancer, which is estimated to be responsible for over 600,000 deaths in 2017 within the US alone, is a disease where precision medicine could be instrumental in improving disease outcomes.  Genomic profiling has identified numerous somatic mutations with clinical relevance, such as EGFR mutations that guide gefitinib therapy in lung cancer, but only a small fraction of cancer drugs approved by the FDA have molecular predictors of efficacy. To realize the promise of precision medicine in cancer, continued research is critical. 
 
One major challenge for identify and utilizing somatic mutation markers of drug efficacy is obtaining tumor material from cancer patients. Conventional biopsies are invasive, expensive, and impractical for tracking dynamic changes to the mutational landscape of a tumor over time and in response to treatment. In many cases, the use of cell-free DNA (cfDNA) can circumvent these issues by providing access to tumor DNA through a simple blood draw. In cancer patients, a fraction of this fragmented free-floating DNA is tumor-derived (ctDNA) and can provide insight into the mutational characteristics of both primary and metastatic sties. 
 
Next-generation sequencing (NGS) is the optimal platform to survey mutations. Unlike microarrays, NGS does not rely on surveying only a fixed set of loci known to be relevant to cancer. But the application of NGS to mutational analysis of cfDNA has its challenges. First and foremost, only a fraction of cfDNA is tumor-derived. Second, the amount of ctDNA in the blood can vary substantially both from individual to individual and throughout disease progression in a single person. Addressing these limitations requires the application of appropriate sequencing technologies and data analysis tools. Exome or targeted sequencing is likely more appropriate for cfDNA analysis than whole genome sequencing because it remains cost-prohibitive to obtain sufficient sequencing depth for accurate variant calling with whole genome sequencing. 
 
The use of a robust variant calling pipeline is equally critical to ensuring that called mutations reflect the tumor and are not technical artifacts.  Partek® Flow® provides an optimal analytical platform for mutation detection in cfDNA as it provides tools to process your data from raw fastq files to prioritized variants.  Alignment with BWA or Bowtie2 allows for the accurate mapping of reads to ensure high-quality data is available for variant calling. Coverage reports offer a way to determine if there is sufficient depth in the sequencing to capture variants in the ctDNA fraction of cfDNA. In the case of cfDNA samples, variant callers such as Freebayes and LoFreq provide the ability to detect low frequency alleles, and a consensus set of variants from both callers can be utilized to gain additional confidence in putative mutations.  Prioritizing variants can be accomplished through annotation with transcript models, vcf-based variant databases such a ClinVar or COSMIC, and functional effect prediction tools like SnpEff. All this and more can be accomplished by building simple visual pipelines, while maintaining the full power of the underlying command-line tools.
 
A recent publication by Butler et al. offers a good case study of the utility of NGS for identifying clinically actionable variants in cfDNA from cancer patients and how Partek® Flow® can be employed to detect these critical somatic changes. Using exome sequencing, the authors examined two cancer patients, one with metastatic sarcoma and another with breast cancer, to assess how well cfDNA sequencing replicates mutations found in tumor tissue. The patient with metastatic sarcoma illustrated that mutations identified in cfDNA correlate well with mutations found in the primary tumor; clinically actionable variants in KRAS and PIK3CA were found in both cfDNA and primary tumor samples. The breast cancer patient with metastatic disease provided a more complex scenario. A primary tumor biopsy was obtained prior to estrogen deprivation therapy and both metastatic tumor tissue and cfDNA were obtained after the patient became resistant to the treatment. Both the metastatic tissue and the cfDNA presented a clinically actionable mutation in ESR1 that confers resistance to estrogen deprivation therapy whereas the primary tumor did not. For this patient, serial blood draws and cfDNA analysis could have been used to dynamically track the evolution of the tumor and guide adjustments in treatment. Using Partek® Flow® and both cfDNA samples from the Butler et al. study, we performed an independent analysis. We were able to identify the above-mentioned clinically actionable variants and other mutations in the cfDNA samples that were reflective of the tumor DNA from both cancer patients.  To learn how we analyzed the cfDNA using a state of the art bioinformatics pipeline in Partek® Flow®, please watch our webinar.