Researchers in the UK have developed a targeted sequencing approach to detect mutations in circulating tumor DNA that they believe could eventually be developed as a clinical assay for early cancer diagnosis or to monitor disease progression and treatment response.
Their strategy, tagged-amplicon deep sequencing, or TAm-Seq, is a two-step amplification process that combines the Fluidigm Access Array with Illumina sequencing.
In a study published last week in Science Translational Medicine, the team described how it technically validated the technique on 46 formalin-fixed paraffin-embedded samples of patients with advanced ovarian cancer, demonstrating they could reliably detect mutations down to about a 2 percent allele frequency.
Additionally, they applied the assay to several clinical plasma samples, highlighting its potential clinical use.
In a metastatic cancer sample from a patient who had multiple primary tumors, they demonstrated that the assay could identify the origin of the metastasis. In a separate case, they showed that the technique could detect a druggable mutation in circulating tumor DNA that had not been seen in the primary tumor; and in a third demonstration, they used the assay to monitor 10 mutations in the blood of a metastatic breast cancer patient undergoing treatment over 16 months.
The researchers are now in the process of evaluating the technology in a clinical lab setting, testing the assay on many more patient samples. James Brenton, a senior author of the study and head of the Functional Genomics of Ovarian Cancer Laboratory at Cancer Research UK, said that the assay would have two major clinical applications. It will "allow us to look for the emergence of mutations that may be therapeutically relevant," and it could potentially be used to "speed up diagnosis for women who might have ovarian cancer," he told Clinical Sequencing News.
To design the TAm-Seq assay, the researchers first created primer sets to generate amplicons from regions of interest — in this case 5,995 base pairs of sequence that included known cancer genes such as TP53, PTEN, EGFR, BRAF, KRAS, and PIK3CA. Then they performed a limited-cycle pre-amplification to capture all the starting molecules, followed by individual amplification to purify and select for the targets.
Next, the researchers used Fluidigm's Access Array to amplify the targeted regions from 47 samples. Following amplification, samples were pooled and sequenced.
In the study, the team used Illumina's Genome Analyzer for the sequencing step, but they plan to test the assay on both the HiSeq 2000 and the MiSeq, depending on the number of samples being run at one time. They are also considering the Ion Torrent PGM.
Tim Forshew, a co-lead author of the study and post-doctoral researcher at the Cancer Research UK's Cambridge Research Institute, told CSN that the team settled on the two-step amplification process because using the Fluidigm system alone "dilutes the DNA," but the pre-amplification PCR step had "all the traditional issues with multiplex PCR," including some regions amplifying better than others and capturing non-specific targets.
Adding the Fluidigm Access Array after the PCR step helped to normalize regions of coverage, he said. The system also "allows you to set up lots of separate PCR reactions" from different patient samples.
Olivier Harismendy, an assistant professor at the University of California, San Diego, told CSN that the method is technically similar to an assay he developed for targeting rare mutations in cancer genes, dubbed UDT-Seq (CSN 12/21/2011).
What is unique about TAm-Seq, he said, is the application to circulating tumor DNA. "That was the really novel part," he said. So far, researchers who have looked at circulating tumor DNA have relied on assays designed for one mutation, while the next-gen sequencing approach can be more comprehensive and allows multiple samples to be analyzed at the same time, he said.
Because the assay sequenced tumor DNA from blood samples, the so-called "liquid biopsy" could be used in cases where a tumor biopsy sample is unobtainable or not recommended because it may pose a risk to the patient, Harismendy added.
However, before it can be used clinically, the researchers will need to demonstrate its clinical utility, he said.
To establish sensitivity and specificity of the assay, the team tested it on 47 FFPE samples from patients with ovarian cancer. The team screened for mutations in the coding regions of TP53, PTEN, EGFR, BRAF, KRAS, and PIK3CA. Additionally, they conducted Sanger sequencing of the TP53 gene in the same cases, and found that 100 percent of the mutations were concordant.
Next, they sequenced libraries from six different diluted mixtures of six FFPE samples with a different known point mutation in the TP53 gene to a mean depth of 5600x. All 33 mutations were identified at allele frequencies greater than 1 percent, including six mutations present at an allele frequency below 2 percent, with one false positive called at an allele frequency of 1.9 percent. Thus, they established a positive predictive value of 100 percent for the assay for mutations with an allele frequency greater than 2 percent.
Clinical Plasma Samples
The team then tested the assay on specific clinical cases from patients' blood samples. First, they looked at samples from seven patients with high grade serous ovarian cancer who had high levels of circulating mutant TP53 DNA, as assessed by digital PCR. The TAm-Seq assay was able to detect the TP53 mutations in all seven patients.
Additionally, from one patient, an EGFR mutation was detected in the plasma sample at an allele frequency of 6 percent. The same mutation was not found in the original biopsy from the primary tumor that had been surgically removed 15 months prior to the blood sample. When the researchers re-examined additional tumor specimens collected at the time of initial surgery using the TAm-Seq assay, they identified the EGFR mutation in two samples at an allele frequency of 0.7 percent, but not in the other six samples. They were unable to examine the relapse tumor because a sample was unavailable.
The EGFR mutation was "not a typical hotspot mutation," said Forshew, so it "would have been missed" with other strategies.
The analysis was performed retrospectively, but had the assay been performed real-time in a clinical setting, the mutation could have potentially informed treatment with a tyrosine kinase inhibitor, said Muhammed Murtaza, co-lead author of the study and a doctoral candidate at the Cambridge Research Institute.
From a second set of plasma samples from 62 patients with high grade serous ovarian cancer, the team established that the assay has 97.5 percent sensitivity for detecting mutations at allele frequency greater than 2 percent. The assay detected 39 point mutations, out of 40 that were detected with digital PCR. The one missed mutation was due to sampling error.
Brenton said the assay could potentially serve as a way to diagnose cancer earlier. TP53 mutations in particular are extremely common in high grade serous ovarian cancer, present in around 97 percent of all cases. However, those mutations have been found throughout the gene and are not confined to hot spot locations, making them more difficult to analyze with other techniques.
Whether identifying TP53 mutations earlier would impact patient survival is another question that must be answered before the assay can be implemented clinically. Brenton said that his team is now looking to collaborate with hospitals to implement the assay in a diagnostic lab to test its clinical utility.
In another experiment, the team demonstrated that the assay could track mutation levels over time throughout a patient's treatment regimen.
From a woman with metastatic breast cancer, the team sequenced the tumor genome to identify 10 mutations to track throughout her treatment course. The mutations followed a distinct pattern of decline during treatment and increase after therapy was stopped and the disease progressed.
Finally, they used the assay to study plasma samples from a patient with two primary cancers — bowel and ovarian — which were resected at the same time. After a five-year remission, a pelvic mass of uncertain origin was identified, from which it was not possible to perform a biopsy. So the patient was put on an ovarian cancer chemotherapy regimen to which she responded.
The researchers then used TAm-Seq to retrospectively analyze FFPE samples from the primary tumors collected at initial surgery and plasma samples collected serially at the time of relapse. The analysis showed that the patient's plasma at relapse contained the TP53 mutation identified in the primary ovarian tumor but not the PIK3CA, KRAS, or TP53 mutations that were in the primary bowel cancer.
"Had these results been available, uncertainty and treatment delays may have been avoided," the authors wrote in the study. Additionally, had PIK3CA or KRAS mutations been identified instead of the TP53 mutations, it would likely have led to a different chemotherapy treatment, and may have also "opened the possibility of enrollment into a trial for targeted therapy" for mTOR, PI3K, or MEK inhibitors, they added.
Murtaza said that, going forward, the researchers would like to improve the assay by increasing its ability to detect rare mutations at an allele frequency threshold below 2 percent. He noted that one potential strategy would be to combine the assay with a technique such as COLD (co-amplification at lower denaturation temperature) PCR, which enriches for rare mutations. Additionally, newer algorithms have been developed that can detect variants present at a frequency of 1 in 1,000.
"We'd like to integrate the assay into clinical trials that measure how the tumor responds to treatment," Murtaza said.