ORLANDO – Researchers at Ohio State University have developed a rapid single-molecule nanopore sequencing assay that can analyze blood samples from acute myeloid leukemia patients for clinically actionable and prognostic markers in a few hours.
The team, led by OSU's Esko Kautto and James Blachly, designed the test on Oxford Nanopore Technologies' MinIon sequencer, aiming for same-day analysis of AML patients' DNA so the detected genomic variants may be used to guide treatment or inform their prognosis.
At the American Society of Hematology's annual meeting this week, Kautto presented findings from a proof-of-principle study in which he and his colleagues compared the assay's ability to gauge AML-associated alterations against short-read sequencing using Illumina's MiSeq sequencer and capillary electrophoresis. Although the test needs further validation and assessment in larger cohorts before it can be used in the clinical setting, the researchers are hoping that in the future, their assay may be a faster and more cost-effective alternative to the current gold standard approaches.
AML has a 24 percent five-year survival rate. For many years, chemotherapy was the only way to treat this genetically heterogeneous disease, but more recently, a number of molecularly targeted drugs have been approved for patients with certain genetic mutations.
However, AML patients often present with aggressive disease and therefore must receive treatment quickly, often the same day they are diagnosed. Current next-gen sequencing tests can be time consuming, which can hold up delivery of precision medicine approaches. "Targeted therapeutics need targeted diagnostics," Kautto said, noting that NGS approaches at major academic cancer centers can take more than a week and that private oncology offices without access to in-house testing may need to refer patients. "As you can imagine, delays in treatment or the initiation of broadly-acting treatments could possibly be detrimental to patients with aggressive disease such as AML."
Kautto believes that the nanopore assay, if his group is able to successfully develop it for clinical use, could be particularly useful for molecular diagnosis of AML in resource-limited settings, for clinical trial screening, as a companion diagnostic to guide treatment with specific drugs, and for monitoring minimal residual disease to assess whether patients can go off a drug or go on a different treatment.
Aiming to develop a test that is faster than current approaches, the researchers decided on a nanopore sequencing assay that relies on an amplicon-based library and a computational pipeline that immediately starts analyzing patients' data as it streams off the instrument. "That way, within a few hours we have a lot of sequencing data analyzed and the final report of any detected mutations in the sample," he said. "In contrast, by the time we already have results, a traditional NGS-based assay would still be in the middle of library preparation."
Although nanopore sequencing is said to be faster compared to traditional NGS, and more affordable in terms of instrument costs, it is also reputed to have a higher raw read error rate.
Kautto estimated that the out-of-the-box error rate for nanopore sequencing was around 10 percent, which is high when trying to detect low-frequency variants. "We spent a lot of time profiling the errors and optimizing the strategy to mitigate them because we felt that was one of the largest challenges in this," he said.
Kautto and colleagues established a cutoff for quality scores and filtered out lower quality data to reduce the error rate. "This is really possible because the algorithm that converts the raw signal that comes off the instrument to a nucleotide sequence has a built-in qualitative measure of how confident it is in the base call it makes," he explained. As a result, researchers could see the reads with lower-quality scores and exclude them from their analysis.
Additionally, most of the mutations the researchers looked at were missense mutations, which he said tend to have lower error rates than indels.
Kautto and colleagues compared the ability of the nanopore assay against MiSeq and capillary electrophoresis in detecting AML-implicated alterations in eight genes — DNMT3A, FLT3, IDH1, DH2, JAK2, NPM1, NRAS, KRAS — in 72 wild-type and mutated samples. He discussed the performance of the test with regard to three types of alterations: JAK2 hotspot mutations, NPM1 frameshift insertions, and FLT3 internal tandem duplications.
With regard to detecting JAK2 hotspot mutations, the nanopore assay and the Illumina assay showed a high correlation in eight mutant samples, with allele fractions ranging widely, from 4 percent to 94 percent.
The researchers also assessed the nanopore assay's ability to gauge an NPM1 frameshift insertion that is mutated in 30 percent of patients and is associated with an increased risk of relapse following remission. Although nanopore sequencing tends to have high error rates for indels, Kautto said, this is a distinct 4-nucleotide frameshift insertion at the terminal exon. "This mutational signature is distinct enough where we can pick it up in background noise, and it really doesn't randomly show up in our data," he said.
Although the detected variant allele fraction was lower than expected in this case, "this can be captured with a linear model and adjusted," Kautto said.
FLT3 internal tandem duplication mutations in AML confer poor prognosis and can be targeted by multiple drugs. The mutation results in a duplication of multiple parts of exon 13 to 15, but this insertion is long and can be challenging to detect with traditional NGS. Therefore, the gold standard method is capillary electrophoresis.
An evaluation of 48 samples for this mutation showed that 24 were positive by capillary electrophoresis. The nanopore assay detected all of them, but traditional NGS detected only six of the 24 as positive. Kautto noted that the nanopore assay detected variant allelic ratios similar to capillary electrophoresis and that both determined almost identical lengths for the internal tandem duplications.
Importantly, the results from the nanopore assay were available within four to six hours instead of days with traditional NGS, the researchers reported in an abstract.
There may also be cost advantages to the approach. For this research, Kautto's group used MinIon flow cells, barcoding the samples, so multiple samples could be pooled and analyzed in one run. However, Flongle flow cells enabling analysis of individual samples would cost below $250 per sample, the researchers estimated.
"For research purposes, we use a barcoded approach," Kautto said. "Ideally, there are smaller chips developed that would be lower cost and that we could run a single patient on. In the clinical testing setting, if we ever get into those, it would be safer to run these single, smaller chips to … avoid sample contamination."
In the future, he and his colleagues plan to validate the test using more samples from OSU's Leukemia Tissue Bank and from external cohorts in collaboration with other universities. They also hope to make improvements to in silico computational steps and library preparation, optimize per-target coverage, and refine the analysis of other genes, including the coding regions of TP53.
"We do hope to be able to eventually get our assay CLIA and/or CAP validated," Kautto said over email, though he didn't provide a timeline.