CHICAGO (GenomeWeb) – At this week's annual meeting of the American Society of Clinical Oncology, researchers from Memorial Sloan Kettering Cancer Center shared new data on a method that they highlighted as a first step toward a test that can detect cancer early in otherwise healthy people.
The study was funded in part by molecular diagnostics firm Grail, which announced its intentions in the early detection space in 2016 but has remained mostly tight-lipped about the scientific details of its research and development efforts.
The investigators reported that when they sequenced a larger portion of the genome — 508 genes — and at a higher depth — 60,000x coverage — they could find at least one circulating cancer mutation in about 90 percent of patients, and could detect nearly 75 percent of the genetic changes known to be present in these patients' tumors overall.
Divided by tumor type, the method picked up circulating mutations in 97 percent of the breast cancer cases, 85 percent of lung cancer cases, and 84 percent of those with metastatic prostate cancer. On average, that played out to finding a circulating cancer mutation in 89 percent of the full cohort.
The MSKCC researchers also reported that they saw a strong correlation between the fraction of cancer tissue cells harboring a particular mutation and the ability to detect that mutation in plasma. In other words, clonal mutations, which represent earlier nodes of tumor evolution, seemed to be more prevalent and easier to detect in circulation than rarer subclonal mutations.
Based on the differences between the breast, lung, and prostate cancer results, the MSKCC data seems to corroborate evidence from other studies that different tumor types may present varying challenges to early detection via circulating DNA, perhaps based on biologic differences that affect how readily theses different cancers shed DNA into circulation.
In terms of performance of the ultra deep and ultra broad assay, the MSKCC results were better, though not vastly superior to what some other companies have reported using narrower, shallower, and consequently cheaper sequencing approaches.
For example, researchers from Johns Hopkins presented data at the annual meeting of the American Association for Cancer Research earlier this year in which they compared tissue sequencing data and plasma sequencing using their targeted error correction sequencing, or TEC-Seq method. TEC-Seq also involves ultra-deep sequencing, but only about half as deep as that reported by the MSKCC team.
In a small group of patients, the JHU researchers found that over 75 percent of the ctDNA alterations they detected were also present in the corresponding patient's tumor. Also, almost 80 percent of patients had at least one alteration in circulation that matched their tumor.
Presenting the new method at ASCO this weekend, MSKCC investigator Pedram Razavi said that this study represents only a first step. The research team, with Grail, is also looking at whole-genome sequencing and assessment of DNA methylation, toward their ultimate goal of detecting of early, potentially curable cancers.
This requires a different approach than most currently available liquid biopsy assays, Razavi argued. "It's critical for the assay to be able to detect mutations without prior knowledge of a tumor's sequence … and it also should cover large portions of the genome to increase likelihood of finding mutations in light of known heterogeneity of cancers from one patient to another."
"It's also very critical to sequence the genome very deeply to increase sensitivity for finding mutations at very low levels," he added.
According to Razavi, the high-intensity cell-free DNA sequencing approach involves two separate sequencing steps, one performed on cell free DNA and another performed on white blood cells from the same sample tube. Comparing the two gave the team a de novo strategy for mutation calling.
In the study, to compare results to tissue mutations, researchers also performed sequencing on both tumor and normal tissue samples using MSKCC's existing MSK-IMPACT assay, a 410-gene panel.
Overall, the method covers 2.1 megabases, which Razavi said is at least 10 times more than most current cell-free DNA assays and yields about 100 times more data compared to commonly used methods.
Despite the potential of the assay in early detection, Razavi and his coauthors reported at ASCO on its performance in late-stage cancer patients — sharing no data that spoke directly to the question of detecting early occult tumors. Yet, commenters participating in a briefing on the results said that the design of the assay looks like it is moving in the right direction.
In the meantime, several clinicians at the ASCO meeting echoed in separate sessions that liquid biopsy methods, regardless of breadth and depth, are not yet appropriate for clinical use.
Stanford radiation oncologist Maximilian Diehn, for example, said that while cancer screening is one of the most exciting and high-profile applications of cell-free DNA sequencing, clinicians should take care that we aren't there yet.
"I'm very hopeful that eventually we will have some screening assays that utilize circulating nucleic acids. However, we are not there yet, and really screening should not be done on patients using ctDNA assays outside of clinical trials," he cautioned.
This has not swayed some companies from launching commercial blood-based cancer detection tests, however. Notably, Pathway Genomics has been selling a testing service it calls CancerIntercept Detect since 2015, and continues to market it despite receiving a letter from the US Food and Drug Administration raising concern over a lack of published validity data.
Another company, CellMax Life, has also been offering a circulating tumor cell-based test for colorectal cancer screening in Asia, which it said it planned to eventually launch in the US.
In an email, Grail spokesperson Charlotte Arnold confirmed that the high-intensity sequencing method presented by MSKCC is part of the company's assay development efforts. Specifically, the work the MSKCC team is doing is intended to generate data that will inform the company's future test development, she wrote.
MSKCC is also one of the institutions participating in the company's circulating cell-free genome atlas (CCGA) effort, which seeks to characterize the landscape of cell-free DNA profiles in at least 7,000 people with cancer and 3,000 people who do not have cancer.
In addition to the presentation on the yields of the high-intensity sequencing approach, Ravazi and his MSKCC and Grail coauthors also shared a poster at ASCO on something that many researchers have highlighted as a challenge for future liquid biopsy-based detection tests — the identification of somatic cfDNA variants derived not from occult tumors but from clonal hematopoiesis.
Clonal hematopoiesis is a term coined in recent years to describe somatic mutation events and clonal expansions that share features with known blood cancers, but which may or may not ever progress to malignant or symptomatic disease.
In the analysis presented at ASCO, the MSKCC and Grail researchers analyzed 151 metastatic cancer patients and 47 normal controls (blood bank donors). They applied the same 508-gene panel and 60,000x sequencing to both groups, and detected 1,072 variants through the two cohorts that were also present in sequenced white blood cells.
The allele frequency of these variants and the genes mutated both supported that they likely arose from clonal hematopoiesis. For both the cancer and non-cancer individuals, the average number of overlapping variants was five, and their frequency was positively associated with age for both groups.
According to the authors, the findings add to evidence that future cell-free DNA cancer screening tests will need to account for this phenomenon and accurately distinguish between tumor-derived and white blood cell-derived variants to avoid false positives.