Skip to main content
Premium Trial:

Request an Annual Quote

NGS Error Correction Method Described at AACR as Excitement Grows For Early Cancer Detection


WASHINGTON, DC (GenomeWeb) – Researchers from Johns Hopkins University this week described the methodology they have developed for subduing sequencing errors in order to allow accurate detection and characterization of mutations at very low frequencies in circulating cell-free DNA.

The approach, called targeted error correction sequencing, or TEC-Seq, involves in-solution capture of a targeted panel of tumor-related genes and exceedingly deep sequencing — as much as 30,000x — of cell-free DNA fragments.

In a presentation at the annual meeting of the American Association for Cancer Research held here, JHU researcher Jillian Phallen said that she and her colleagues developed a panel of 58 cancer-related genes that they have adopted for testing the approach in early cancer detection, demonstrating a limit of detection down to 0.05 percent and a specificity of nearly 100 percent across these regions of interest.

Phallen also described some early experiments with the method, which appear to bode well for its use in detecting mutations either in very early-stage cancers in order to monitor patients, predict outcomes, or screen healthy people for furtive, unknown tumors.

In collaboration with a number of other medical centers, as well as with liquid biopsy firm Personal Genome Diagnostics, Phallen and colleagues used TEC-Seq first in 44 healthy individuals and then in a cohort of 179 cancer patients, most of whom had earlier-stage disease.

In the healthy group, the method uncovered no tumor-related somatic mutations. The assay did, however, find some alterations in genes related to myelodysplasia in a subset of cases.

In a group of lung, breast, colorectal, and ovarian cancer samples the team detected measurable ctDNA at varying levels. About 56 percent of stage I and II breast cancer patients had measurable ctDNA, while 71 percent of early-stage colorectal cancer patients had positive signals. Similarly, 57 percent of early lung cancer patients, and 56 percent of ovarian cancer patients were positive for somatic mutations covered by the panel.

'We think theses results are very promising for bringing liquid biopsy toward early detection, but the question remains still what the comparison is between plasma and tumor," Phallen said.

In some patients in the cohort, she and her colleagues had matched tumor samples, and encouragingly, they saw a high concordance between the mutations they detected in plasma, and what had been seen in individuals' tumor tissue. Over 75 percent of the ctDNA alterations were present in the corresponding tumor overall, and almost 80 percent of patients had at least one alteration in circulation that matched in their tumor, Phallen said.

In one patient, the team was also able to examine multiple tissue samples, revealing that the discordance between the ctDNA and tissue results was due to tumor heterogeneity.

Importantly, TEC-Seq didn't pick up all, or even close to all of the early-stage cancers in these experiments. Asked by the audience about the 50 percent or more individuals that would be missed, Victor Velculescu, Phallen's collaborator at Hopkins and chair of the session, argued that the promise of blood-based cancer screening tools doesn't require perfection.

Even if a technology can move cancer diagnoses into the early stages for half of patients, or three-quarters of patients, it would be expected to have a large impact, given what clinicians know about the improvement in outcomes for those who are caught and treated early.

PGDx, which performed sequencing for these analyses, is a spinout of Hopkins, co-founded by Velculescu and other Hopkins researchers. Phallen confirmed that TEC-Seq is part of the technology that PGDx has licensed from Hopkins and is developing in its own commercial activities.

Monica Nesselbush, a scientist at PGDx, presented validation data for the company's current PlasmaSelect test in a separate mini-symposium on Wednesday. She also reiterated that TEC-Seq is an element of PlasmaSelect.

Though PGDx hasn't publicized plans for moving that test into the early detection space, Nesselbush did cite some of Phallen's data in early-stage cancers during her presentation.

Hopkins is not alone in developing sequencing error-correction methodologies for liquid biopsy. At other recent meetings, investigators from a number of institutions have described methods using barcoding or systematic tweaks to glean true variants from amplification error and other biases.

For example, researchers from Stanford University reported last year in Nature Biotechnology on their own error correction method for circulating tumor DNA sequencing — integrated digital error suppression (iDES) — which builds on a technique the Stanford team previously developed called CAPP-Seq, and which Roche has since acquired. According to the group, iDES enabled them to detect mutant alleles at a frequency of .004 percent.

Companies like Guardant Health also have proprietary methods for barcoding samples to boost sensitivity and reduce sequencing errors. Guardant is advancing an early detection test of its own through a multi-arm, multi-site trial called Project LUNAR, but it hasn't shared specific details on how assays it advances through LUNAR might differ from its current Guardant360 test.

In the meantime, there remain unanswered questions that methods like TEC-Seq and others will have to answer, clinicians at the meeting said across several other sessions.

Even if investigators can show that error correction strategies do allow sensitive enough detection of fragments of circulating tumor DNA to pick up occult cancers, the clinical field is still grappling with how it could or should use such information.