A cell-free DNA test based on the presence of DNA methylation has proven highly specific as a multicancer detection test and appears especially good at detecting high-risk malignancies. In most cases, it can also accurately pinpoint the tumor’s tissue of origin, researchers reported at the 2019 ASCO Annual Meeting.

“Our hope is that someday, with this assay, individuals at risk for cancer can simply get a single blood test (as a complement to other screening methods), and it would find cancer with high specificity,” Geoffrey R. Oxnard, MD, Associate Professor of Medicine at Harvard Medical School and a thoracic oncologist at Dana-Farber Cancer Institute, told The ASCO Post.

Geoffrey R. Oxnard, MD

Minetta C. Liu, MD

“We see 90% accuracy overall for identifying the tissue of origin, which is critical for guiding the efficient downstream workup for a positive signal,” added fellow investigator Minetta C. Liu, MD, Research Chair and Professor in the Department of Oncology at the Mayo Clinic, Rochester.

Screening tests for early cancer detection are often criticized due to the risk of overdiagnosis. What is needed is a highly specific test with a low rate of false-positive results—one that will minimize the detection of indolent cancers and spare patients unnecessary testing, according to Dr. ­Oxnard. “Specificity should be the rule if our goal is to detect cancer in asymptomatic, otherwise healthy persons,” he said.

Substudies of CCGA Project

At the ASCO meeting, Drs. Liu and Oxnard reported the results of substudies of the Circulating Cell-Free Genome Atlas (CCGA) project, a prospective, multicenter study of more than 15,000 individuals (56% with cancer). The aim is to develop a noninvasive cancer detection assay.

Dr. Liu reported the results for 2,301 subjects with more than 20 cancer types (12 prespecified and considered high-risk cancers) across all stages, of whom 1,422 had cancer, and 879 did not.¹ Dr. Oxnard described a longitudinal follow-up of 1,320 distinct participants that evaluated the prognostic significance of detection by the assay.²

DNA Methylation Is Key

In the initial CCGA discovery substudy, researchers evaluated several approaches to cancer detection, ultimately identifying whole-genome bisulfite sequencing for DNA methylation as the most effective approach and the one that is being developed further. DNA methylation helps regulate gene expression, with some regions of hypermethylation and some regions of ­hypomethylation.

"We see 90% accuracy overall for identifying the tissue of origin, which is critical for guiding the efficient downstream workup for a positive signal."

— Minetta C. Liu, MD

Tweet this quote

“Mutations and copy number changes can be found in white blood cells, and those findings can show up in individuals without cancer as well, leading to false-positives. Methylation patterns, however, are unique to the tumor DNA, avoiding problems with specificity; they not only detect but localize the tumor,” Dr. Oxnard explained. “Methylation has not panned out for use in precision cancer care, but it’s a great signal for the presence of cancer,” he added.

Identifying the Tissue of Origin

Dr. Liu’s study of the targeted cell-free DNA methylation assay found that it detected a strong signal for the 12 prespecified aggressive cancer types at early stages, with a specificity of at least 99%. In addition, the test identified the tumor’s tissue of origin with high accuracy across all cancers. The 12 prespecified cancer types included anorectal, colorectal, esophageal, gastric, head and neck, hormone receptor–negative breast, liver, lung, ovarian, and pancreatic, as well as multiple myeloma and lymphoid neoplasms. Together, these malignancies account for approximately 63% of all cancer deaths in the United States.³

Dr. Liu emphasized that the targeted methylation assay detects both advanced disease and early-stage cancers, “which is essential because our goal is to facilitate early detection and improve clinical outcomes.” At 99% specificity, the sensitivity for these 12 high-risk cancers ranged from 59% to 86% at early stages (stages I–III).

For all 20 cancer types, the overall detection rate in the substudy, across all stages, was 55%, and 94% of participants received a tissue of origin location (with 90% accuracy). “Accurately determining the tissue of origin was consistent regardless of stage,” Dr. Liu said. “Knowing the tissue of origin will streamline the subsequent diagnostic workup of patients who receive a positive result from the screening blood test.”

For example, the targeted methylation assay’s sensitivity for the following prespecified tumors, at stage IV and stages I to III, follows:

• Breast cancer: 100% and 64%
• Lung cancer: 90% and 50%
• Pancreatic cancer: 95% and 78%
• Ovarian cancer: 100% and 67%
• Head and neck cancer: 92% and 86.

The investigators are following the controls who do not yet have cancer. They are seeing that some who demonstrated “early signals” of cancer are indeed being diagnosed with cancer over time,⁴ said Dr. Liu.

Poor-Prognosis Cancers Most Likely to Be Detected

Dr. Oxnard reported an exploratory longitudinal analysis of survival from 1,320 participants with cancer from the discovery substudy. Researchers analyzed overall survival in patients with cancer detected vs not detected, using whole-genome bisulfite sequencing, set at 98% specificity, and for clinical stage (IV vs I–III), diagnostic method (symptom- vs screen-detected), sex, age, and histologic grade.

“We followed our subjects to determine what type of cancers we are detecting and what type we are missing,” he explained. “We hope we are finding the most important ones.”

Of 1,320 participants with all types of cancer and with 1 year of survival follow-up, the prototype whole-genome bisulfite sequencing assay detected 506 (38%), of whom 230 (17% overall) died. Among the 230 who died during follow-up, 188 (82%) had cancer detected by the assay.

Patients whose cancers were detected by the assay had a 2-year overall survival of less than 50%, whereas for those whose cancers were not detected, 2-year survival exceeded 90%. The poor prognosis associated with cancer detection with cell-free DNA was seen in both cancers that presented symptomatically and those found via screening. Across all stages of disease, cancers detected by cell-free DNA whole-genome bisulfite sequencing were associated with significantly worse survival than those not detected by the blood test.

The fact that life-threatening types of cancers were much more likely to be detected “suggests that cell-free DNA–based detection with this methylation assay may be an indicator of prognosis,” Dr. Oxnard suggested. The test also was good at detecting cancers that currently lack effective screening paradigms, such as ovarian and pancreatic cancers, he added.

The multivariate analysis confirmed the prognostic implications of detection with the assay: Detected cancers had double the risk of death (hazard ratio = 2.6; P < .001) when accounting for clinical stage, cancer type, histologic grade, age, sex, and method of diagnosis. Methylation-based cancer detection carried comparable prognostic significance to clinical stage (hazard ratio = 3.4; P < .001).

“We are finding the cancers associated with a high risk of dying and not those associated with little risk. With this assay, for example, we won’t be overdiagnosing such things as thyroid nodules. We may be avoiding some of the overdiagnoses that are seen with some existing cancer screening methods,” explained Dr. Oxnard. “Our hope is to find those early-stage cancers that need timely treatment, hopefully catching them at a time when we can impact survival.”

Next Steps

Trials are being planned or are underway to evaluate how the assay performs in clinical practice. Currently enrolling is SUMMIT, a prospective, observational, longitudinal cohort study being conducted in London with approximately 50,000 participants, aged 50 to 77 years, not diagnosed with cancer. About half the participants will be at high risk of cancer due to a significant smoking history, and the other half will have no history of smoking.

“There’s a sense of urgency to nail this down and get it to patients,” Dr. Oxnard said. Dr. Liu added, “Early cancer detection is a priority. Selfishly, as a clinician, I want this test appropriately incorporated into clinical care as soon as possible.” ■

DISCLOSURE: Dr. Oxnard has received honoraria from Foundation Medicine, Guardant Health, and Sysmex; has served in a consulting or advisory role for AstraZeneca, DropWorks, GRAIL, Ignyta, Inivata, Janssen, Loxo, Sysmex, and Takeda; and has a patent pending for “Noninvasive blood-based monitoring of genomic alterations in cancer,” for which Dr. Oxnard receives a portion of the licensing fees. Dr. Liu’s institution has received research funding from Eisai, GRAIL, Janssen Diagnostics, Merck, Novartis, Roche/Genentech, Seattle Genetics, and Tesaro; and has been reimbursed for travel, accommodations, and other expenses by Agena Bioscience, Celgene, Cynvenio Biosystems, Genomic Health, GRAIL, Menarini Silicon Biosystems, Merck, and Pfizer.

REFERENCES

1. Liu MC, Jamshidi A, Venn O, et al: Genome-wide cell-free DNA methylation signatures and effect on tissue of origin performance. 2019 ASCO Annual Meeting. Abstract 3049. Presented June 1, 2019.

2. Oxnard GR, Chen X, Fung ET, et al: Prognostic significance of blood-based cancer detection in plasma cell-free DNA: Evaluating risk of overdiagnosis. 2019 ASCO Annual Meeting. Abstract 1545. Presented June 3, 2019.

3. GRAIL announces positive new data with multi-cancer early detection blood test from CCGA study. Available at https://grail.com/press-releases/grail-announces-positive-new-data-with-multi-cancer-early-detection-blood-test-from-ccga-study/. Accessed August 7, 2019.

4. Cohn AL, Seiden M, Kurtzman KN, et al: The Circulating Cell-free Genome Atlas (CCGA) Study: Follow-up on non-cancer participants with cancer-like cell-free DNA signals. 2019 ASCO Annual Meeting. Abstract 5574. Presented June 1, 2019.