NEW YORK (GenomeWeb) – Two independent groups have published proof-of-concept studies demonstrating the feasibility of using targeted next-generation sequencing of circulating tumor DNA as an assay to detect mutations in patients' tumors.
A study led by researchers from Memorial Sloan-Kettering Cancer Center and published in the Annals of Oncology this month showed that an NGS-based assay analyzing ctDNA can capture mutations present in both primary and metastatic tumor sites and can measure changes in mutational load in response to therapy.
Separately, a study led by a group from the International Agency for Research on Cancer, based in France, and published this month in Clinical Cancer Research, demonstrated that targeted sequencing of ctDNA can identify actionable mutations in lung cancer patients.
The two studies add to a growing body of evidence that sequencing circulating tumor DNA may be a good way to noninvasively detect cancer mutations and could help identify targetable mutations in tumors that may be hard to biopsy, monitor a patients' tumor burden in response to therapy, and diagnose disease progression earlier.
Already, at least one company, startup Guardant Health, is commercializing targeted sequencing-based ctDNA assays. Its first product, Guardant 360, analyzes 54 oncogenes and tumor suppressor genes.
And academic groups are honing their own methods for sequencing ctDNA, such as a group from Johns Hopkins developing a low-coverage whole-genome sequencing approach to detect structural variants and a group from Yale University that is using targeted sequencing as a tool for early detection and diagnosis.
"The applications of the technology are incredible," Jorge Reis-Filho, a surgical pathologist at Memorial Sloan-Kettering and senior author of the Annals in Oncology study, told Clinical Sequencing News. "There are possibilities to use this technique for disease monitoring, predicting whether patients will relapse or not after receiving a given therapy, to define the mechanisms of resistance, and to help us define what will be the next therapy for a patient," he said.
In the Annals of Oncology study, the MSKCC team used an NGS assay targeting 300 cancer genes to evaluate a breast cancer patient's primary tumor and metastatic sites, as well as ctDNA at different time points throughout therapy.
The 66-year-old patient had ER-positive, HER2-negative, grade 2 mixed invasive ductal-lobular carcinoma with bone and liver metastases at diagnosis. Following three lines of chemotherapy, array-based molecular profiling in 2011 revealed an AKT mutation that made her eligible for Genentech's pan-AKT inhibitor Ipatasertib. A blood sample was taken at baseline before she was treated with Ipatasertib, at two months, six months and eight months when progression occurred.
The MSKCC team used its targeted sequencing assay IMPACT, which uses the Illumina HiSeq 2000 to assess 300 cancer genes. They analyzed DNA from archival samples of the primary and metastatic tumors that were taken at diagnosis as well as the four blood samples that were collected throughout therapy.
They identified 15 somatic non-synonymous mutations in the primary tumor, all of which were also detected in the liver metastasis. Two mutations, to FLT4 and MAP2K2, were detected in the liver metastasis but in the primary tumor were present in only 12 of 47 reads and 40 out of 113 reads, respectively, so could not be reliably called.
However, ctDNA sequencing of the baseline sample identified all mutations found in both the primary and metastatic sites at allele fractions above 20 percent, with the exception of an MLL3 mutation.
In addition, sequencing ctDNA at various time points showed how the tumor was responding to therapy. For instance, at two months after starting Ipataserib treatment, the fraction of mutant alleles decreased compared to the baseline sample. But, at six months, the mutant alleles increased to similar levels seen at the baseline sample. Importantly, the AKT mutation, the drug target, had an allelic fraction of 57 percent after six months of treatment, even higher than the baseline sample of 39 percent. The increase in allele frequency of the drug target was seen before disease progression was diagnosed, "providing evidence to suggest that increases in disease burden can be detected earlier by ctDNA analysis than by classical biochemical and radiologic assessments," the authors wrote.
Reis-Filho said that he has since analyzed ctDNA from additional breast cancer patients in a prospective study that is currently being reviewed for publication. In that study, he said that the team designed a targeted assay that analyzed genes specific to breast cancer to monitor patients' response to neoadjuvant chemotherapy.
Reis-Filho said that while sequencing ctDNA holds a lot of promise, there are still a number of limitations. One limitation his group has found is that so far it has been unable to detect mutations from brain metastases in the blood, possibly because of the blood-brain barrier, he said. In addition, while the technique seems to work well on patients with a high disease burden, there has been some evidence that patients with early stage disease may not have detectable amounts of tumor DNA in their blood, he said.
An easier biopsy
Being able to identify tumor mutations from a blood sample is also important for the purely practical reason that some tumors are difficult to biopsy.
In thoracic oncology, it is difficult to get a good tumor sample, Sebastien Couraud, a senior physician in pulmonology and oncology at Lyon Sud University Hospital, told CSN. "Most patients' tumors are very deep in the thorax, so you can only get small biopsies, and often the tumor sample is too poor quality to accurately test for somatic mutations," he said.
Couraud and colleagues described proof of concept for a targeted ctDNA assay for lung cancer in the Clinical Cancer Research study. Their test used Thermo Fisher's Ion Torrent PGM to evaluate hotspot regions from five genes known to be altered in lung cancer — EGFR, BRAF, ERBB2, KRAS, and PIK3CA. The team tested it on 107 plasma samples collected from a cohort of lung cancer patients that had never smoked. They also ran the assay on matched tumor DNA from 68 of the patients.
The researchers identified 50 mutations in the tumor sample, 26 of which were also detected in the blood sample, for a sensitivity of 58 percent and specificity of 87 percent.
Couraud said that the main problem with the test's sensitivity had to do with the bioinformatics. At the time the team developed the test, he said that they did not have software for calling somatic mutations, so had to use software that was developed for germline mutations. A second issue, he said, was with the test's ability to call deletions, in particular EGFR deletions, which are common in lung cancer.
"We're working on improvements and hope to have a better test in a few months," he said.
Nonetheless, even the first iteration of the test was able to detect mutations that traditional biomarker assays were unable to detect because they were present at less than 25 percent frequency. The majority of the mutations were found in EGFR and no sample had more than one mutation in the tested genes. More than half of the mutations, 29 out of 50, were insertions or deletions.
In addition, two of the mutations that were considered to be false positives in the ctDNA were upon further analysis found in the primary tumor but at allele frequencies under 5 percent.
Couraud said that his team will continue to work on improving the test and is also considering testing other sequencing technology to see if that results in improved sensitivity, particularly for detecting deletions. "We hope to eventually translate the test into routine practice," he said. "It would be very exciting for us to have a noninvasive diagnostic of lung cancer somatic mutations."