By the time surgical oncologist Phyllis Napoles operates on people with pancreatic cancer, the disease is usually advanced, the prognosis poor. But in October 2020, Napoles entered the operating room at Sutter Health in Sacramento, California, to work on a different kind of patient.
Identified with stage two pancreatic cancer as part of a trial for a new kind of blood test that screens for cancer, Jim Ford was surprised by the diagnosis. Unlike most people who find out that they have the disease, the 76-year- old retired car salesman had no symptoms and had been out golfing just a week before.
Pancreatic cancer is only rarely discovered so early. Once somebody has symptoms, Napoles says, survival rates can be as low as 3 percent and the disease is often untreatable. But she was able to completely remove Ford’s tumor, which was about the size of his thumb. Now a year after surgery, radiation, and chemotherapy, Ford is cancer-free—illustrating the potential for a new generation of blood tests to someday swell the ranks of survivors from a wider range of cancers.
“This is something I never thought would evolve in my career, and I'm still young in my career so I’m still hopeful,” says Napoles, who graduated from medical school in 2006, finished her fellowship training in 2013 and specializes in pancreatic cancer. “This is so dramatic. This is really going to change all the statistics we have on pancreas cancer survival and detection.”
Using advances in genetic sequencing and artificial intelligence, a dozen or more companies are working on blood tests, sometimes called liquid biopsies, that can pick up cancer signals circulating in minuscule concentrations in the bloodstream. The demand for these tests, if they end up being useful, is enormous. According to estimates by the nonprofit think tank Information Technology and Innovation Foundation, the potential market for this kind of cancer-detection technology is worth more than $6 billion and is expected to almost triple in value by 2025.
The only blood test for cancer-screening currently available outside of trials is the one that identified Ford’s cancer. Called Galleri, the company says the test can detect 50 types of cancer in a sample of blood. Created by the California-based healthcare company Grail, the test is now available by prescription in the U.S. for people with an elevated cancer risk. It will also available as part of a large study in England, which is currently recruiting participants.
The goal of these new blood tests is to save lives by catching cancers earlier, especially those that don’t currently have reliable screening tests. In the United States, there are now ways to screen people at high risk for five types of cancer: breast, colon, prostate, cervical, and lung—by blood or other types of tests, like mammograms. But of the roughly 600,000 cancer deaths that occur in the U.S. each year, more than two-thirds are caused by cancers that have no good screening options, studies show. They are usually not discovered until they have metastasized.
Already, doctors are using liquid biopsies to scan the blood of their cancer patients for information to help determine which treatments to use and whether cancers have returned after treatment. This new wave of blood tests attempts to detect cancer in people who have never been diagnosed before.
Yet, the tests also raise concerns—about false positives that lead to risk-laden and unnecessary follow-up tests, about finding early-stage cancers that might never advance far enough to require treatment, and about discovering cancers with no effective treatments. What happens if people use these tests without medical supervision? What happens when the only way to confirm a positive blood test is with highly invasive surgery?
Establishing that the tests are both accurate and useful will be crucial, says Shivan Sivakumar, a medical oncologist at the University of Oxford in the United Kingdom. “It's important to have hype. It’s important to get excited about these things,” he says. “But you really do need to know if what we're doing is actually working or not. I probably still sit on the fence.”
Cancer signals in blood
Liquid biopsies initially were developed for testing the blood of people who already have cancer to understand the biology of their disease. As cancers grow in the body, some cancer cells die and shed DNA, and liquid biopsies use various strategies to detect this circulating tumor DNA.
These kinds of liquid biopsies can also be used to see if a cancer has specific mutations that can be targeted by certain drugs, says Geoff Oxnard, a thoracic oncologist and head of clinical development at Foundation Medicine, a biotechnology company focused on genetic analyses. Following treatment, liquid biopsies can help track specific mutations within a tumor to understand how a cancer is responding to that treatment. They can detect resistance, guide subsequent therapy, and reveal when a patient relapses.
Blood tests that look for specific cancer mutations have been approved by the Food and Drug Administration and commercially available in the U.S. for about five years. In 2020, the FDA approved two liquid biopsy tests that are the first to screen for multiple mutations at once, including one developed by Oxnard’s company. They use next-generation genetic sequencing (a rapid approach that sequences many parts of the genome simultaneously) to detect mutations driving the cancer’s growth. The Foundation Medicine test was approved as a companion diagnostic for therapies to treat lung, breast, ovarian, and prostate cancers. Compared to traditional biopsies, which extract tissue for analysis, Oxnard says, these tests are less invasive, and they often deliver results more quickly.
The development of technologies that can find cancer signals in the blood of people with diagnosed cancer opened up another enticing prospect: the possibility of finding the first signs of early-stage cancer in people without a diagnosis. Applying the technologies developed for liquid biopsies to blood-based cancer screening could address an enduring problem, researchers say: Most of the time, cancers are diagnosed when the disease is already advanced and causing symptoms—when the prognosis is often dismal.
People with pancreatic cancer that has spread, or metastasized to other organs, for example, can expect to live for between three and 12 months, Sivakumar says. If the disease is discovered early—which happens in fewer than one in five cases—life expectancy can extend to three or four years. People who get lung cancer can also benefit from effective treatments if the disease is found early, but it rarely is. Blood tests for early cancer detection? “I think it would be incredibly transformative,” Sivakumar says.
Needle in a haystack
For years, developing these kinds of tests was hindered by a major challenge: the amount of cancer DNA that circulates in the blood is incredibly tiny. “You’re looking through a potpourri of DNA floating in the blood and most of that DNA is the patient’s DNA, their haystack of DNA,” Oxnard says. “You’re hunting through the haystack of DNA for the needle.”
Once scientists figured out how to detect, amplify, and sequence DNA quickly and affordably— steps aided by leaps in the field of next-generation sequencing—they also needed to work out what they were looking for. There are various strategies to do this.
One approach is to search for certain known cancer mutations in circulating tumor DNA that has been isolated from a blood sample. The strategy is used by a multi-cancer blood test called CancerSeek, which searches for cancer-related genes as well as proteins that can be elevated in response to the tumor. Developed by researchers at Johns Hopkins University, an early version of the test detected 26 cancers of various types during a yearlong trial of about 10,000 women, ages 65 to 75, with no history of cancer, according to research published in 2020. The test, not yet available outside of trials, did not lead to excessive follow-up testing, which is a common concern of screening programs. Just 0.22 percent of participants had false positives that triggered unnecessary and invasive diagnostic procedures, the researchers reported.
Grail has taken a different approach with its Galleri test, looking not for mutations but for compounds called methyl groups that attach to segments of DNA, turning genes on or off, says Joshua Ofman, Grail’s chief medical officer and head of external affairs. Each person has their own methylation pattern, but specific genes tend to get more or less methylated in predictable ways when someone has cancer. Mutations can also appear as a normal part of aging, even without cancer. And methyl markers are far more common than mutations in those genes, Ofman says. In a region of DNA that has just two or three mutations, there may be tens of thousands of methyl markers turning on tumor-promoting genes and turning off suppressor genes in similar ways across many kinds of cancer. That makes it easier for machine-learning algorithms to recognize them.
Scanning for methylation groups has multiple advantages, Grail’s research suggests, including specificity. Certain patterns of methylation correlate with different types of cells, so algorithms can provide a fingerprint of where in the body the cancer is. In one validation study that included more than 15,000 people, the false positive rate was just 0.5 percent.
Examining data from an ongoing trial that includes more than 6,600 people over age 50, Grail reported in June that the Galleri test had found 29 cancers of 13 types, including breast, colon, liver, and lung cancers, as well as leukemia. Forty percent of those were still localized—in stage one or two, according to an interim analysis from the trial, called Pathfinder. It's the same trial that identified Ford with pancreatic cancer.
When the test turned up a positive result, Ofman adds, cancer was confirmed 44 percent of the time. (Even with low false-positive rates in validation testing, screening tests like these can end up flagging plenty of people who don’t have cancer because the percentage of true positives in asymptomatic people is extremely low.)
In one analysis, published in March, Grail researchers estimated that a multi-cancer early detection test could avert 104 deaths per 100,000 people each year—amounting to 26 percent of all cancer-related deaths.
The test is now available by prescription in the U.S., as long as people are willing to pay up to $949 for it. In England, the publically funded National Health Service is collaborating with Grail to enroll 140,000 people in a pilot study that will assess its performance.
In August the biotechnology company Illumina acquired Grail. The company’s plan is to roll out the test around the world as quickly as possible, and make it widely available, especially in underserved communities, says Francis deSouza, CEO of Illumina. (DeSouza is a member of the board of directors of The Walt Disney Company, which is majority owner of National Geographic Partners.) He is moved by stories like the one Napoles tells. “That's what fuels my passion, he says. “To say: Look, that cancer was a death sentence, and now it doesn't have to be.”
Galleri and CancerSEEK are the most high-profile blood tests in development that aim to detect multiple types of cancer at once, Sivakumar says. Other groups are working on blood tests for single cancers. As the industry takes off, however, some researchers caution against getting too excited too soon.
Studies of the efficacy of early detection are notoriously hard to do, Sivakumar says. It can take a long time to get results, and they often end up being disappointing. For instance, in a study published earlier this year, researchers in the U.K. found that screening for ovarian cancer made no difference in reducing the number of women who died from it. It took 10 years to get that result.
Screening tests of all kinds can lead to false positives that have psychological and physical ramifications, studies show. Even when tests accurately detect cancers early, they might not lead to better outcomes if diseases, like ovarian cancer, are aggressive and hard to treat, Sivakumar says. “If you're going to screen for cancer, you better make sure that you have a very good intervention to offer, as well,” he says.
A number of screening programs haven’t ended up helping even though they detected problems, adds Margaret McCartney, a general practitioner in Glasgow, Scotland, and senior associate at the University of Oxford’s Centre for Evidence-Based Medicine. One notorious example is the implementation of mass screening programs for thyroid cancer in parts of Europe and Asia, including in the aftermath of nuclear disasters at Chernobyl and Fukashima, as well as a national screening program in Korea. The programs led to a large uptick in diagnosed cases of thyroid cancer—along with related surgeries and anxiety. But they had no impact on numbers of lives saved. “The problem with doing any kind of screening study is that it's very easy to find problems,” McCartney says. “But does that actually help you to make people's quality or quantity of life better?”
It is easy to get wrapped up in the sexiness of innovation at the expense of its usefulness, McCartney adds. People like Ford, who are identified with cancer early, fuel a sense of possibility, even if epidemiological data eventually fail to show net benefits for society.
The major challenge of screening is to provide information that will make a difference, Oxnard says. Pushing the field forward will require clear data to establish accuracy. “It's tantalizing that we could find a cancer signal, but can we do it in a way that never gets it wrong?” he says. “What's it going to take to get trust and adoption around those kinds of applications? That's what we're all trying to figure out.”
For Ford, the issues seem less complicated. He is back to working as many as 20 hours a week at a golf course, and he plays regularly. Agreeing to be in a trial was unusual for him: normally, he doesn’t want to be bothered. Now, he tells everybody he sees about his experience and recently encouraged his sister-in-law to join the trial, too.
“What if I had said no, like I usually did?” he says. “I would probably be stage four by now. And stage four is almost totally uncurable.”
Editor's Note: This story originally misspelled the name of the medical oncologist at the University of Oxford in the United Kingdom. It is Shivan Sivakumar.