A blood test in development can detect cancer and tissue of origin with few false positives but many false negatives in people suspected of having cancer before a histologic diagnosis is made, according to a new study. The test uses next-generation sequencing technology to detect DNA methylation patterns associated with cancer in cell-free DNA. It is being developed by GRAIL, Inc., which funded the study. Dr. David Thiel, chair of the department of urology at Mayo Clinic in Florida, in Jacksonville, described the latest results with the test April 28 during the American Association for Cancer Research (AACR) virtual annual meeting. He noted that many cancers are detected too late, with regional or distant metastases present at diagnosis. "The detection of cancer prior to the development of metastatic disease can significantly improve 5-year survival," Dr. Thiel said during a media briefing. For all cancers, the 5-year cancer-specific survival is 89% when diagnosed early compared with 21% when diagnosed late, although this can be a result of lead-time bias. For lung cancer, 5-year cancer-specific survival is 56% with early diagnosis and 5% with late diagnosis, he noted. "Our thought is that a simple, noninvasive, early cancer-detection test could potentially decrease cancer-related mortality," said Dr. Thiel. The GRAIL early-cancer-detection assay is currently being tested in several clinical trials. At AACR, Dr. Thiel presented data from a pre-specified subgroup from the company's Circulating Cell-free Genome Atlas (CCGA), a multicenter, case-control, observational study with 5-year follow-up of more than 15,000 adults with and without a diagnosis of cancer who provided a blood sample prior to clinical diagnosis. In this analysis, people with high clinical suspicion (HCS) of cancer were classified as clinically confirmed cancer (with 164 in the training set and 75 in the validation set) or clinically confirmed non-cancer (49 training and 15 validation). The group with clinically confirmed cancer in this HCS subgroup included people with multiple cancers across stages I to IV. In the group of confirmed non-cancer, all training and validation samples were correctly predicted as non-cancer (100% specificity), Dr. Thiel reported. In the confirmed-cancer group, cancer detection across all stages was 40.2% in training and 46.7% in validation, with 100% specificity. Excluding stage I renal cancers—which were overrepresented and subject to poor detection at early stages due to low tumor cfDNA fraction—detection across stages was 59.3% in validation. In stages II and above, detection was 78.9%, all at 100% specificity. For detected cancers, the tissue of origin (TOO) was predicted in 93.9% of samples in training, and 100% in validation. Of those with a TOO result, accuracy was 85.5% in training and 97.1% in validation. Dr. Thiel said the test has the potential to detect cancer and predict tissue of origin in people with high clinical suspicion of cancer ahead of confirmed pathologic diagnosis. It could potentially be used to accelerate the diagnostic workup of these patients. The test, he emphasized, is intended to be used "alongside guideline-recommended screening and not replace it." Population-scale studies to validate the performance of the test are in progress. Commenting on the results, AACR President Dr. Antonio Ribas said, "This is one of the areas of cancer research that is most rapidly becoming reality in the clinic. This abstract presents data on more than 15,000 patients that have suspicion of cancer and with a blood test they could not only confirm or not confirm that there was a cancer but also know where it originated and before many of these cancers would be diagnosable otherwise." "We are seeing the future in real time. I think this session (at AACR) will be remembered for when we saw the big data (showing) that these assays should be broadly applicable provided they have the appropriate regulatory-agency approval. This will not take that long and it will change our practice," said Dr. Ribas of the University of California, Los Angeles. The study was funded by GRAIL, Inc. Several authors have financial relationships with the company. —Megan Brooks Source
