Keynote Address: Using Technology to Intervene Earlier in Cancer and Improve Survival Rates

Most cancers are diagnosed in the later stages, but if this disease could be diagnosed earlier and with greater accuracy, patient outcomes could be much better, said Joshua Ofman, MD, MSHS, chief of corporate strategy and external affairs, Grail, Inc. “I know that nothing can keep you up at night like the concerns people have about cancer,” Ofman said, as he began his keynote address during Patient-Centered Oncology Care, the annual multistakeholder gathering presented by The American Journal of Managed Care^® in Philadelphia, Pennsylvania.

Although there are eff ective screening methods for some cancers, most are not screened for at all, and the process is inefficient: screening for one type of cancer at a time. Furthermore, screening rates are “suboptimal,” Ofman said. As a result, most cancers are diagnosed in later stages, when outcomes are poorer. Finding cancer in later stages means the cost of care will be dramatically higher, too.

On the fl ip side, however, there are concerns that when cancers are diagnosed early, there can be overdiagnosis or cancers diagnosed that might not cause harm. Ofman pointed to controversies over recommendations for mammography and prostate-specifi c antigen testing.

According to the data Ofman presented, approximately 2000 Americans die each day due to cancer. By 2020, cancer care is projected to cost more than $150 billion dollars.

“So, we have got to transition our current healthcare system away from a ‘break it and fix it’ healthcare system and toward a healthcare system that focuses on prevention and early detection,” he said. “And if we don’t do that, these numbers are not going to change.”

New technologies are the key to focusing on early detection. Sequencing the human genome and reading the code of DNA is allowing the healthcare system to find mutations and changes, but we are not yet at a place where we can distinguish cancer from noncancer just based on mutations, Ofman explained.

The next step in technology has been the convergence of machine learning and artifi cial intelligence. This technology is allowing Grail to look into blood to identify fragments of DNA that degrading cancer cells release. However, healthy cells also degrade and release fragments. Any test of these cell fragments would have to be able to discriminate a cancer cell from a noncancer cell with high specifi city in order to create few false positives. And that test would have to be able to tell clinicians what tissue it was from to localize the cancer.

In a study, Grail examined the methylation patterns in DNA. Methylation patterns change the activity of a DNA segment and tell cells what cells to become and what tissue to go to. Grail also looked at RNA, mutations, and DNA code in order to determine the best way to detect cancer.

The Circulating Cell-Free Genome Atlas Study included more than 15,000 participants with and without cancer and compared how efficient the technologies were at detecting cancer, how many false-positive test results there were, and how well they could identify where the cancer was localized.

The study revealed that DNA methylation was the best method, and adding on any of the other methods, such as mutations, did not improve the test. Then, Grail trained a machine learning algorithm to detect the cancer and determine its location.

The test can diagnose or find more than 24 different types of cancer, but Ofman presented on a subset of the 12 deadliest cancers that account for about 70% of cancer mortality. The test detected cancer across these 12 cancers with 76% sensitivity and the false-positive rate was below 1%. In comparison, the false-positive rate for mammography is 10%. The test was also able to accurately localize the cancer in more than 90% of cases.

The low false-positive rate is important, as the test identifies early-stage cancers. Ofman presented data that showed patients whose cancers were diagnosed by the Grail test had a worse mortality rate, which appears to mean that the Grail test is diagnosing harmful and lethal cancers.

“And that’s really important for an early-detection test, because as you know, one of the biggest concerns with cancer detection right now is that we’re fi nding too many false positives,” Ofman said. “And we’re fi nding early cancers that probably don’t need to be treated as aggressively as they’re being treated. So, we can be fairly confi dent based on these early data that the cancers the Grail test is finding required treatment, which is exactly what you want to know.”

What does this test mean for overall public health? If everyone in the United States from age 50 to 79 were eligible for screening and they were all given the Grail test, or a test like it, almost half a million cancers would be identifi ed, and many of these cancers would not have been identifi ed without this test because there is no capability to detect the cancer that early.

If the same population went through current screenings, such as mammography for breast cancer, Cologuard (Exact Sciences Corporation, Madison, WI) for colon cancer, or low-dose computed tomography for lung cancer, the tests would diagnose 150,000 cancers but produce about 9 million false positives. Adding the Grail test to the current screening assessments would find 3 times as many cancers but only add 1 million false positives.

Grail also looked at how its test could intercept cancers at an earlier stage and what impact this finding would have. Looking at the participant data of those likely to die in the next 5 years from cancer, Grail found it could intercept at an earlier stage for 68% of cancers. This interception would fl ip the current distribution of diagnosis, which is mostly happening at stage III or stage IV. This test has the potential to reduce the cancer mortality rate by 37%.

“This is all a projection, and it’s early modeling, and there’s much more to come, but this paints the magnitude of the opportunity that we’re staring into, to focus on early detection,” Ofman said.

Download PDF: Keynote Address: Using Technology to Intervene Earlier in Cancer and Improve Survival Rates