NEW YORK – Buoyed by data from a newly published study, Stanford University spinout Foresight Diagnostics is pushing forward a novel liquid biopsy sequencing technology that promises to significantly increase the sensitivity of minimal residual disease detection in both hematologic cancers and solid tumors.
Called phased variant enrichment and detection sequencing, or PhasED-seq, the technique was developed by Stanford researchers Max Diehn and Ash Alizadeh, based on the concept that the co-location of two or more mutations on circulating tumor DNA fragments could be harnessed to reduce sequencing errors and increase sensitivity.
Liquid biopsy technologies have proven themselves successful in advanced cancer, but applications to detect residual disease in patients treated with the intent of a cure have posed a greater challenge due to patients' relatively low levels of circulating tumor DNA.
In recent years, several approaches for minimal residual disease detection have emerged to meet this challenge, reporting minute detection limits. But there is clinical evidence that these tools are still not sensitive enough. Across cancer types, studies have found that significant numbers of patients without detectable ctDNA still eventually experience disease progression, implying that the technologies used have yielded a false-negative result in these cases.
PhasED-seq pushes ctDNA detection limits even further, enhancing sensitivity by up to 100-fold, depending on the comparator, by targeting what are dubbed "phased variants" — defined as two or more SNVs located on the same DNA fragment.
"Seeing all these molecules with multiple mutations on them … led us to realize that this is a really powerful strategy for improving the detection limits for minimal residual disease detection, because the existing methods are all limited by the fact that they have a certain background error rate that's inherent to the sequencing methods. And you can't look below that error rate," Diehn explained. "These phased variants have orders of magnitude lower error rates than the conventional ways that groups, including ours, have previously done this."
For example, for a single mutation, the error rate might be 1 in 50,000, but for two mutations within 10 bases of each other on the same molecule to occur by error, the chance is 1 in 50,000 squared.
According to the study, the approach is similar to duplex sequencing in that it reduces sequencing errors by requiring the detection of two concordant events. But unlike duplex sequencing, which looks at two complementary strands of DNA, PhasED-seq interrogates target mutations on the same strand, regardless of whether the complementary strand is present.
"A key challenge we found for cell-free DNA applications using duplex sequencing is that the efficiency is not super high for recovering both the [forward and reverse] molecules. You only recover them about a third of the time at best, so this is the same idea, but instead of going across two DNA strands, we're going within a DNA strand, so the efficiency is much higher for recovering that structure," said Alizadeh.
In their publication in Nature Biotechnology last week, Diehn, Alizadeh, and colleagues reported that they could find phased variants across diverse cancer types. But B-cell lymphoma has become the team's first commercial target because these cancers feature pairs or larger groups of variants that occur consistently in "stereotyped" genomic regions.
"The phased variants cluster in the vast majority of patients in these highly stereotyped regions that allow the average patient to be captured by a relatively small panel," Alizadeh explained. This means a universal, or "off-the-shelf" assay can work across patients with no need to customize individual assays. MRD detection is also a crucial paradigm for these patients, up to 40 percent of whom experience recurrence on standard-of-care treatment.
Aurora, Colorado-based Foresight, which closed a $12.5 million Series A financing in June, was spun out by Diehn, Alizadeh, and two of their former postdocs, David Kurtz and Jake Chabon, and had been operating in stealth mode prior to the publication of the new study.
Their research showed that PhasED-seq nearly doubled the identification of B-cell lymphoma patients with residual disease compared to CAPP-seq, an older liquid biopsy sequencing approach also developed by the Stanford researchers and acquired by Roche in 2015.
The study analyzed 678 specimens from 213 patients with B-cell lymphomas tracked longitudinally. Investigators reported that PhasED-seq detected and quantified ctDNA at levels below one part per million, a limit of detection 40- to 100-fold more sensitive than competing ctDNA detection approaches.
Comparing PhasED-seq to CAPP-seq in 88 DLBCL patients after two cycles of immuno-chemotherapy, the older technology identified less than half of those who went on to experience a recurrence, while PhasED-seq detected ctDNA in almost 90 percent of these recurrent cases.
In an additional 19 DLBCL patients tested following the completion of their therapy, standard sequencing identified only two out of five who later experienced recurrence whereas PhasED-seq caught all five.
Just as importantly, the remaining 14 patients where PhasED-seq did not detect ctDNA have remained disease free after more than two years of follow-up, the authors wrote.
In a statement, Foresight cofounder and study co-author Kurtz said that the sensitivity and accuracy demonstrated in the study are such that the PhasED-seq platform could become the new standard of care for clinicians managing patients with B-cell lymphoma.
"While many patients … are cured using standard therapies, there remains a great need to accurately and rapidly identify those who are not cured so that alternative treatment regimens can be initiated in a timely way to increase the likelihood of treatment success," he added.
"This also has important implications in drug development where PhasED-seq could be employed as a companion diagnostic to guide treatment decisions during or after curative intent therapies," added Chabon, the firm's cofounder and CEO.
Although a lack of predictable, stereotyped locations for phased variants in non-DLBCL cancers means that patient-specific assays will probably be necessary, Foresight said it does intend to pursue MRD applications in other tumor types.
Commercial activity in this area, which already includes patient-specific, personalized assay approaches from companies like Natera and others, has been mostly focused on colorectal cancer thus far. However, research is also ongoing in breast cancer, lung cancer, and other tumor types where curative-intent treatment is used in early-stage cases.
In their study, the Foresight investigators also tested the potential of PhasED-seq to detect MRD in a pilot cohort of 36 samples from six solid tumor patients (five with lung cancer and one with breast cancer). The authors wrote that they could detect ctDNA at levels of sensitivity similar to what they saw in the B-cell lymphoma samples, down to less than one part per million, in one pre-treatment plasma sample from a stage IB NSCLC patient.
"The key question now is what things we can achieve that overcome the limitations of existing MRD methods for solid tumors," Alizadeh said.
"Technically, it’s a little harder because we have to do whole-genome sequencing first and then build a custom capture panel, whereas with lymphoma, we have one off-the-shelf capture panel," said Diehn. "But that personalized approach is no different than what other groups in the field are doing, like Natera and Archer, and we have the advantage that we can get much better error rates and a much higher sensitivity than the competing methods."
Although the existence of phased variants isn't proprietary, Diehn and Alizadeh said that Foresight is operating under a patent filed for the application of phased variant detection to MRD.
"I think part of the reason people have missed these [as a target] in the past is that there are some challenges with both the bioinformatics as well as the molecular biology recovery of these molecules. Just from a technical standpoint, oftentimes mutation callers penalize these molecules because if you have many mutations in a single molecule, that can be an artifact, that's probably a bad molecule," Diehn added.
While mutation-based techniques have led the pack, liquid biopsy researchers and commercial firms have also turned to epigenetic signals for MRD detection, with Guardant Health notably launching the first commercial methylation-based test in the space earlier this year.
In theory, PhasED-seq could potentially be combined with epigenetic methods. "That's something that would be interesting to explore," Diehn said.
But those epigenetic-based methods don’t have the low error rates of even non-phased variant mutation approaches, he added. "They have some other advantages — in particular that you potentially don't need to sequence the tumor first. But the problem they have is that everything that's been published so far suggests that the error rates are not as good as even just a single nucleotide variant."
"We firmly believe that from the perspective of overcoming the key challenge that MRD methods currently have, which is that there's too many false negatives — right after surgery, the majority of patients who ultimately will recur don't have detectable ctDNA — that's what PhasED-seq offers," Diehn said.