NEW YORK (GenomeWeb) – A group led by University of Pittsburgh Medical Center researchers has developed a next-generation sequencing assay that targets specific somatic mutations associated with late-stage breast cancer.
The assay, called MammaSeq NGS, comprehensively covers known driver hotspots targeting 79 genes and 1,369 mutations which the team believes could help cancer researchers identify clinically actionable mutations and potential therapeutic targets.
"We tried and whittled this [number] down to not only the ones we believe are drivers, but the ones that are therapeutically actionable," UPMC pharmacology and chemical biology professor Adrian Lee explained. "We designed it thinking about advanced disease because there's a kind of a gap in knowledge in advanced disease when we're doing this."
In a study published earlier this month in Breast Cancer Research, Lee, the study's senior author, and his colleagues integrated mutation calls from primary tumors in TCGA and studies focused on breast cancer in order to establish a list of somatic mutations.
Lee said that his team developed the NGS assay mainly for therapy selection and research purposes, including monitoring how breast cancer mutations alter over time and if certain enriched mutations exist in specific breast cancer subtypes.
The team initially collected solid tumor samples from 46 patients and 14 circulating tumor DNA (ctDNA) samples from seven patients in separate cohorts. The group used 1 to 4 milliliters of plasma to isolate the patients' ctDNA, which they then quantified using Thermo Fisher Scientific's qubit dsDNA HS assay kit. Extracting germline DNA (gNDA) to act as a control, the team used 20 nanograms of the ctDNA for library preparation.
The team then sequenced the 46 tumor samplesusing MammaSeq, calling up 4,970 variants across all the patient samples.
However, Lee noted that the researchers prioritized clinically important mutation genes using certain criteria: each needed to be among significantly mutated genes in primary and metastatic samples, was clinically actionable, was of functional importance in cancer, was identified in more than five primary tumors or two metastatic tumors, and had been found in both primary and metastatic lesions.
Specifically, Lee and his team removed multiple somatic errors including germline variants, non-coding and synonymous variants, and variants with an allele frequency higher than 90 percent. The team eventually identified a total of 592 protein-coding mutations.
Within the solid tumor cohort, the researchers noted that the total number of mutations was skewered toward a subset of samples, as 408 of the 592 mutations were found in four of the 46 tissue samples. They noted that mutations ranged from one to 128 per sample and identified PIK3CA and ESRI as the most commonly mutated genes in the cohort.
The team then applied the OncoKB Precision Oncology Database to determine how many of the mutations had "putative clinical utility." In total, Lee and his team found 28 clinically actionable variants annotated by OncoKB — 26 single nucleotide variants and two ERBB2 amplifications — distributed across 20 of 46 cases in the solid tumor cohort. However, the team annotated most of the actionable variants as stage III cancer biomarkers.
The team then sequenced the 14 ctDNA samples from patients with metastatic cancer using MammaSeq to a mean depth of 1810x, while they sequenced the gDNA controls to a mean depth of 425x.
Performing variant calling on the ctDNA and gDNA, the team removed patient-matched variants present in both sample types. Applying the same filtering pipeline included in the solid tumor cohort, the team identified a total of 43 somatic mutations, with up to 26 mutations per sample. Similar to the solid tumor cohort, they found PIK3CA and ESRI as some of the most commonly mutated genes.
The team then applied droplet digital PCR to validate specific ctDNA mutations in the samples. Taking two nanograms of each ctDNA sample, the team amplified the sample before performing ddPCR on ESR1-D536G, PIK3CA-H1047R, and FOXA1-Y175C mutations.
Lee acknowledged that his team dealt with multiple limitations in the study, including a small sample size and inability to associate mutations with patient outcome. In addition, the study authors noted that they were unable to completely capture all mutations "given rapid advances in the field," as well as the potential for false-positive results due to issues with detection of rare events.
While the team highlighted that it used extra deep sequencing to reduce the false-positive rate, Lee said that it will need to perform additional studies to improve MammaSeq's clinical sensitivity and specificity in ctDNA. He pointed out that targeted DNA sequencing panels like MammaSeq do not cover as many genes as other tools such whole exome or genome sequencing.
Lee also noted that his group had challenges working with different breast cancer subtypes in the solid and ct-DNA cohorts. Despite the issues, the team believes the pilot study will help researchers quickly and cost-effectively detect somatic mutations in patients' solid tumor and ctDNA.
"Lots of groups are trying to understand the concordance between solid tumor biopsies and liquid biopsies, which is quite challenging," Lee said. "We now have large studies [at UPMC] where we now collect plasma off of every patient with advanced breast cancer at every progression to develop a large cohort to advance these early preliminary studies."
While Lee emphasized that his team does not plan to commercialize MammaSeq, he argued that the platform differs from other research-use-only panels because of its focus on targeted breast cancer somatic mutations
"By only focusing on breast cancer mutations, [we] reduce the number of sequencing reads on genes not found to be altered in cancer, and thus offer increased sensitivity for breast cancer mutations given the same number of sequencing reads," Lee said.
Although Lee noted that his team has trademarked the MammaSeq name, he noted that the technology is currently an open-source tool that anyone interested in breast cancer research can use in their own labs.
"As researchers can perform [MammaSeq] in their own laboratories, they can negotiate the best price they can with core facilities or run it in their own laboratory if they have equipment," Lee explained.
Lee said that groups who want to know the distribution of clinically actionable variants in their breast cancer patient cohorts would likely find the assay useful. Instead of examining more genes in fewer tumors, he highlighted that identifying more tumors with fewer genes may help researchers find unique subtypes and begin applying clinically actionable mutations in the diagnostic space.
Lee and his group therefore believe that MammaSeq will be a useful and cost-effective platform that researchers could eventually apply to identify and track clinically actionable mutations in breast cancer.
"The ultimate goal is to use these types of assays to identify mutations that are associated with therapeutic response in patients with late-stage breast cancer," Lee said. "Use of this assay in a research cohort may increase knowledge toward this clinical goal."