Novel blood tests could improve cancer detection
Click Here to Manage Email Alerts
Two novel blood assays demonstrated efficacy in detecting various cancer types at different disease stages, according to results of two studies presented at ASCO Breakthrough in Bangkok.
“There is something compelling about a single blood test that could potentially complement existing screening tests to find cancers early and allow earlier intervention with a growing array of effective cancer treatments,” Geoffrey R. Oxnard, MD, thoracic oncologist at Dana-Farber Cancer Institute and associate professor of medicine at Harvard Medical School, told HemOnc Today. “The Circulating Cell-free Genome Atlas study has showed us that this targeted methylation test for plasma DNA has the ability to detect the majority of cancers, especially high-risk cancers, with a false-positive rate below 1%, while also accurately predicting the tissue of origin for cancers detected.”
Methylation signatures
In the first study, Oxnard and colleagues evaluated use of targeted bisulfite sequencing of plasma cell-free DNA for simultaneous detection of multiple cancer types and tissue of origin localization. The preplanned substudy of the prospective, multicenter, observational, case-control Circulating Cell-free Genome Atlas (CCGA) study — funded by the manufacturer of the blood-based assay, Grail Inc. — included 927 participants, 654 of whom had cancer (> 20 tumor types, all stages) and 273 of whom were cancer free. The study also included 337 noncancer samples from the multicenter, prospective STRIVE study cohort of women undergoing screening mammography.
Researchers used high-efficiency methylation chemistry to enrich for methylation targets in plasma cell-free DNA, and they determined cancer status and tissue of origin with a machine leaning classifier. They combined methylation fragments characteristic of cancer and tissue of origin across targeted areas and designated a relative probability of cancer, as well as of a specific tissue of origin.
Results showed the blood-based assay had more than 99% specificity, which the researchers noted is required for population-level screening and represents a combined false-positive rate across all cancer types of less than 1%. Sensitivity of the test, at 99.3% specificity, was 55% (95% CI, 51-59) for all cancer types.
Overall sensitivity of the test was 18% (95% CI, 13-25) for stage I disease (n = 185), 43% (95% CI, 35-51) for stage II (n = 166), 81% (95% CI, 73-87) for stage III (n = 134) and 93% (95% CI, 87-96) for stage IV (n = 148).
The blood-based assay detected tissue of origin in more than 90% of tumors for which the tissue was provided. Specifically, the assay detected 39% (95% CI, 30-50) of breast cancers (n = 104), 66% (95% CI, 56-75) of lung cancers (n = 111) and 77% (95% CI, 64-88) of colorectal cancers (n = 53).
In addition, among the 96% of cases where the cancer was localized to an anatomic site, the tissue of origin call was correct in 93%.
Researchers also analyzed a prespecified group of 12 deadly cancer types — anal, bladder, colorectal, esophageal, head and neck, liver/bile duct, lung, ovary, pancreatic and stomach cancers, as well as lymphoma and plasma cell neoplasm — that account for two-thirds of cancer deaths.
At 99.3% specificity, the test detected 76% (95% CI, 72-81) of these cancers, including 39% (95% CI, 27-52) at stage I, 69% (95% CI, 56-80) at stage II, 83% (95% CI, 75-90) at stage III and 92% (95% CI, 86-96) at stage IV.
“Not all cancers are detected with a test like this — with lower sensitivity observed for stage I cancers in particular,” Oxnard told HemOnc Today. “We hope this means that blood-based cancer screening will end up detecting the more deadly cancers, which need urgent treatment, with a lower rate for overdiagnosis of low-risk cancers that do not need treatment. Ongoing studies are evaluating the performance of the test in a cancer screening population. We also hope to study the actual return of test results to patients in upcoming clinical trials.”
Risk score model
In the second study, Mao Mao, PhD, MD, founder and CEO of SeekIn Inc. in Shenzhen, China, and colleagues validated a multivariate cancer risk score model that incorporates shallow whole-genome sequencing data and plasma protein markers to identify cancer from a sample of blood.
According to study results, the model demonstrated 60% sensitivity and 98.5% specificity among a set of 76 patients with stage I to stage IV cancer and 152 otherwise healthy controls.
To validate the clinical utility of the model, researchers pooled data from a previous study that included 466 pregnant women who tested positive for multiple chromosomal aneuploidies and who underwent the protein markers test.
At a median follow-up of 399 days, 39 malignant maternal cancers were identified. Of these, the risk score model correctly identified 71.8% (n = 28) for a positive predictive value of 73.7% and specificity of 97.7%.
“Cancer is the second leading cause of death — we clearly need improved methods of cancer screening with the ultimate goal of early detection,” Minetta C. Liu, MD, chair of oncology research at Mayo Clinic in Rochester, Minnesota, as well as an author of the first study, told HemOnc Today. “The promising results presented here support the possibility of bringing multicancer detection assays into clinical practice. However, successful implementation of blood-based screening for multiple malignancies will require a paradigm shift for health care providers and patients. Pilot studies are needed to determine feasibility, cost-effectiveness, clinical workflows and general acceptance.” – by Jennifer Southall
References:
Mao M, et al. Abstract 45. Presented at: ASCO Breakthrough; Oct. 11-13, 2019; Bangkok.
Oxnard GR, et al. Abstract 44. Presented at: ASCO Breakthrough; Oct. 11-13, 2019; Bangkok.
For more information:
Minetta C. Liu, MD, can be reached at Mayo Clinic, 201 W. Center St., Rochester, MN 55902; email: liu.minetta@mayo.edu.
Geoffrey R. Oxnard, MD, can be reached at the Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA 02215; email: geoffrey_oxnard@dfci.harvard.edu.
Disclosures: Grail Inc. funded the study of the test that uses targeted bisulfite sequencing of plasma cell-free DNA. Liu reports research funding to her institution from Eisai, Grail, Janssen Diagnostics, Merck, Novartis, Roche/Genentech, Seattle Genetics and Tesaro; and travel, accommodations and expenses from Agena Bioscience, Celgene, Cynvenio Biosystems, Genomic Health, Grail, Menarini Silicon Biosystems, Merck and Pfizer. Mao reports employment, patents, royalties and other intellectual property with and research funding from SeekIn Inc.; a consultant/advisory role with Yuce Biotechnology; and stock and other ownership interests in SeekIn Inc. and WuXi Biologics. Oxnard reports consultant/advisory roles with AstraZeneca, DropWorks, Grail, Ignyta, Inviatia, Janssen, Loxo, Sysmex and Takeda; patents, royalties or other intellectual with Dana-Farber Cancer Institute; and honoraria from Foundation Medicine, Guardant Health and Sysmex. Please see the abstracts for all other authors’ relevant financial disclosures.
Perspective
Back to Top
Edward S. Kim, MD, MBA, FACP, FASCO
Cancer diagnostics have traditionally relied on imaging for screening followed by tissue confirmation. Screening tests, including mammography, colonoscopy and most recently CT scans for lung cancer, have improved survival. These tests are utilized to target high-risk populations for which the probability of benefit is maximized vs. the inherent risk of procedure. Development, testing and approval of diagnostic screening tests has been the aspiration of noninvasive screening tests, such as blood-based testing.
CA-125 has been a model for blood-based testing to monitor tumor activity. PSA, on the other hand, demonstrates the controversies associated with the risk for unnecessary treatments vs. benefits of screening tests. For patients diagnosed with cancer, plasma circulating DNA is utilized with confidence to assess tumor biomarkers. This approach is now being utilized to identify cancer in people without known cancer.
Studies report high rates of specificity and sensitivity among patients with and without cancer using plasma cell-free DNA with a targeted methylation sequencing assay identifying methylation fragments, as well as for multivariate cancer risk scores using blood-based whole-genome sequencing and protein marker data. The overall performance of these types of assays continues to improve. Not surprisingly, sensitivities improve with increasing overall stage of cancer.
Both studies report promising data for detecting cancer in people without known cancer.
The challenge of implementing a population-based screening test for cancer is cautiously optimistic. Prospective validation studies with specific high-risk populations will be needed. If cancer can be detected early or higher-risk individuals can be identified, it may open a new generation of chemoprevention strategies to reverse or prevent cancer development. This would, indeed, be a tremendous advance in cancer screening.
Click Here to Manage Email Alerts