NEW YORK – A group of researchers led by investigators at the Broad Institute has developed a sequencing method, coined "minor allele enriched sequencing through recognition oligonucleotides" (MAESTRO), that can help track thousands of personalized cancer mutations in a patient's blood sample.
The team aims to integrate MAESTRO, which combines mutation enrichment with error-correcting duplex sequencing, into a liquid biopsy workflow for routine patient minimal residual disease (MRD) testing following surgery.
Viktor Adalsteinsson, associate director of the Broad's Gerstner Center for Cancer Diagnostics, explained that detecting trace levels of residual disease often requires an enormous amount of sequencing to detect low-frequency mutations. The group therefore sought to develop a more efficient sequencing approach that could help detect those cancer mutations.
After sequencing a primary tumor to identify a patient's specific mutations, Adalsteinsson said his team creates short MAESTRO probes for up to 10,000 mutations. These probes are then used to enrich sequencing library molecules that carry unique barcodes on the top and bottom strand for those with mutations, followed by sequencing. He noted that the overall process takes roughly three to four weeks to generate a tailored assay that clinicians can then use to monitor for MRD.
"We've seen that tracking more mutations can increase the sensitivity for tracking MRD and tumor DNA from a blood draw, but that can become costly, given the number of normal DNA [molecules] in blood," Adalsteinsson explained. "MAESTRO, by enriching the mutations of interest, enables us to test for far more mutations than we could envision doing beforehand."
In a study posted on BioRxiv in January, he and his team established MAESTRO's performance by comparing it to duplex sequencing, which uses long probes to capture both mutant and wild-type alleles with similar efficiency.
"MAESTRO enrichment for rare mutations … doesn't waste sequencing reads on the excess normal DNA that could be present in the sample," Adalsteinsson said. "At the same time, [it affords] the high accuracy of duplex sequencing that requires the mutation to be present on both strands of the DNA duplex."
His team showed MAESTRO's ability to detect low-abundance mutations and track thousands of patient-specific mutations across the genome.
After performing whole-exome sequencing (WES) on tumor biopsies matched with normal DNA from 16 patients, they initially built a 40-mutation panel for both MAESTRO and conventional duplex sequencing and found that both methods identified mutations similarly well.
The group then compared MAESTRO's ability to trace ultra-rare mutations to conventional duplex sequencing using 438 mutations. They found that MAESTRO identified about 80 percent of the mutants detected with conventional duplex sequencing — but with much less sequencing — and also detected additional mutations.
Adalsteinsson's team then aimed to see if tracking all genome-wide tumor mutations could improve MRD detection from cfDNA, using a group of 16 breast cancer patients enrolled in a clinical trial. Overall, they saw that tracking all tumor mutations with MAESTRO uncovered more mutations per patient in cfDNA compared to conventional duplex sequencing.
"In our analysis of pre-operative breast cancer [patients], we showed how many tumor mutations in blood we can detect when tracking all genome-wide mutations," Adalsteinsson said. "This provides a large signal enhancement, which helps exclude [clonal hematopoiesis of indeterminate potential] mutations and provides unique advantages for MRD detection."
Manish Kohli, a professor of oncology at the Huntsman Cancer Institute who was not involved with the study, praised MAESTRO's ability to search for thousands of mutations at the same time in a single blood sample.
Kohli also pointed out the Adalsteinsson's team only searched for single-nucleotide mutations and did not share MAESTRO's potential performance for other types of alterations, such as copy number variants, indels, and amplifications. Certain tumors have amplifications, deletions, or fusions, which can be "clinically useful to detect and monitor for MRD," he noted.
While Adalsteinsson acknowledged that his team did not explore other genomic alterations, he said that it "conceptually should be feasible" to use MAESTRO to search for alterations like structural variations.
"When tracking MRD in patients undergoing therapy, we have the opportunity to achieve [high] sensitivity in individual tests for each patient," Adalsteinsson said. "So far, we've found among the different kinds of alterations we could track, mutations were highly specific for MRD, provided we removed those associated with clonal hematopoiesis."
Kohli also noted that the researchers only analyzed samples from a very small cohort of breast cancer patients, rather than testing a large number of patients and a variety of different cancer types. He emphasized that the tumor type may matter when using MAESTRO, as other types of cancer — for example, medulloblastoma and CNS tumors — do not shed as much ctDNA as breast cancer.
Adalsteinsson agreed that a wide discrepancy exists between cancer type and stage regarding how much ctDNA is shed into a patient's bloodstream. However, he believes that using MRD testing to guide improved care for each patient "stands to be a universal paradigm" that could allow delivery to patients who stand to benefit the most from therapy while sparing those who "don't need to suffer from its negative side effects."
At the same time, he noted that his team will need to determine whether the method's high analytical sensitivity will translate into better clinical sensitivity and detect MRD in more patients. The researchers will therefore pursue large retrospective and prospective studies to see if they can correlate MRD detection with future recurrence across different cancer types.
"We hope that with the ultrasensitive test, we could enroll patients early when they have MRD at a timepoint where they have fewer resistance mechanisms in the body, and where therapy could be more effective and intercept MRD before it leads to cancer recurrence," Adalsteinsson said.
Kohli said he is fascinated by MAESTRO's ability to detect mutations with high sensitivity and specificity, but he argued that many of these mutations are likely to be passenger instead of driver mutations. He believes that independent investigators will therefore need to adapt and look at MAESTRO in different settings to make the method inexpensive and usable in the clinical space.
"There's no doubt that [MAESTRO] is an improvement on conventional sequencing and detection of low-abundance mutations in the blood," Kohli said. "But to make it clinically available and call it 'low-cost,' it needs more work in different tumor types, stages, and states, and [be] widely available."
Integrated MRD workflow
While the study focused on MAESTRO's feasibility, Adalsteinsson highlighted that his team is also developing additional sample prep methods to address accuracy issues within the sequencing workflow.
To improve sequencing, they have created a method called "Concatenated Duplex Sequencing," which links both strands of each DNA duplex together so they can be sequenced together at lower costs. Rather than adding adaptors and separately amplifying each strand, Adalsteinsson said the technique instead links the strands for sequencing.
"Traditionally, one would need to sequence large numbers of DNA molecules to find those which originated from the same DNA, "Adalsteinsson said. "With our technique, we think that we can make that substantially more efficient and believe this may have applicability not only for liquid biopsy-based testing, but for other types of sequencing-based tests, as well."
In addition, his team is building a technique called "Duplex Repair," which he said ensures the original base pairing of each DNA duplex stays as intact as possible.
"We wanted to come up with a new approach that would limit the likelihood that false mutations from one strand could be copied to both strands of the duplex before sequencing," Adalsteinsson explained. "Duplex Repair helps to ensure that the original nucleobase composition of each DNA duplex is … not artificially compromised by sample prep methods."
He said that his team is performing ongoing studies to evaluate the feasibility of Duplex Repair and Concatenated Duplex Sequencing in the cancer diagnostic space.
Adalsteinsson noted that both techniques could potentially be used independently of MAESTRO for situations like deeper cancer panel sequencing, but he ultimately envisions a sequencing workflow integrating all three methods to help improve MRD detection using a patient's blood sample.
He and his colleagues have filed patents for MAESTRO to be used for MRD-based detection, but he declined to comment on potential commercial avenues. However, he said that his group is eager to partner with academic and commercial collaborators to further develop the integrated platform for additional cancer types and applications.
While the envisioned mutation tracking workflow will require an initial tumor genome from a patient and generation of unique MAESTRO probes, Adalsteinsson said that "substantial strides" in DNA synthesis technology are helping to drive down the initial cost of the platform.
"There will be upfront costs, [but] we anticipate them being amortized over multiple serial samples per patient," Adalsteinsson said. "By providing a high accuracy at minimal sequencing costs, that will provide unique advantages by being able to do this [workflow] routinely and repeatedly."