The Race to Diagnose Cancer With a Simple Blood Test
Five years ago, a team of researchers pored over the results of a prenatal genetic test given to more than 125,000 healthy pregnant women and made a stunning discovery. The blood test, marketed by gene-sequencing giant Illumina, was designed to detect chromosome anomalies associated with conditions such as Down syndrome by analyzing fragments of fetal DNA circulating in the mother’s blood.
In 3,757 of the tests, the scientists found at least one abnormality. But in 10 of those cases, further analysis revealed that the fetuses were in fact normal.
“In every one of those 10 cases, it turned out there was an undiagnosed cancer” in the mother, says Alex Aravanis, who at the time of the study was the senior director of research and development at Illumina.
To Aravanis and the other scientists, the unexpected result suggested a whole new opportunity: a single blood test for detecting multiple types of cancer before a person has any symptoms. “This was really important background science that [suggested] this might be possible,” says Aravanis, a founder and chief scientific officer at Grail, a Menlo Park, California–based startup launched three years ago by Illumina.
A so-called pan-cancer blood test would address a major barrier in substantially reducing the toll of the world’s second leading killer: Most cancers are diagnosed at advanced stages, when the prognosis is poor. Catching cancer early, when it has the best chance of being cured with surgery, could, theoretically, prevent cancer deaths and reduce the high costs of treating the disease.
“We do very well in improving lives of people whose cancers are screened and detected before symptoms occur,” says David Ahlquist, MD, a gastroenterologist at Mayo Clinic. “Five-year survival rates increase dramatically if cancer is detected at an early stage.”
But most cancers — including some of the most lethal — aren’t prevalent enough to justify regularly screening for them in healthy people. At least not with current methods. But in an October article published in the journal Nature Precision Oncology, Ahlquist argues that a universal screening test for early detection could be cost effective and “fill an enormous existing gap in cancer control.”
He’s hardly the only one to suggest it. For more than a decade, scientists in labs around the world have been working on versions of so-called liquid biopsies — tests that search for evidence of cancer in a person’s bloodstream. Some of these tests are used today to help guide the treatment of people already diagnosed with cancer — applications that by themselves appear poised to transform cancer care. But a liquid biopsy that detects the disease at its earliest stage in apparently healthy people is the ultimate pipe dream—and one that is attracting investors and skeptics alike.
In the past decade, the ability to sequence DNA from tumor tissue has fueled major advances in how cancer is treated, ushering in an era of medicine known as precision oncology, in which doctors use genetic data to match cancer patients to drugs that target the specific cell mutations underlying their disease.
But while the biopsies required to obtain this tissue are considered the gold-standard practice, they are invasive procedures that typically require surgery or needles. Often there’s insufficient tissue to analyze or a tumor is precariously located, making a biopsy unsafe. Tissue biopsies can detect the presence of cancer, but they’re inadequate for monitoring people for relapse or identifying the need for alternate treatments.
In the midst of these challenges, there’s an ongoing push to develop sophisticated liquid biopsies that can detect and analyze cancer DNA in a vial of blood. The bet is that such tests will not only overcome the shortcomings of conventional biopsies but also expand the role of precision oncology well beyond drug-to-tumor matching. By some estimates, more than 15 million people already diagnosed with cancer could be candidates for liquid biopsies. Medical and industry experts also predict that at least for people at a high risk of cancer — about 35 million Americans are in this category due in part to aging or family history — a blood test to check for cancer will soon become part of a regular physical exam.
“This could be done in a way that could save an enormous number of lives and save costs for the health care system. Or it could be done in a way that is disastrous.”
“This is one of the hottest areas in all of medicine,” says Eric Topol, MD, director of the Scripps Research Translational Institute in La Jolla, California. He estimates that, all told, some 50 companies are working on liquid biopsies. While their ambitions differ, all are competing for a portion of what industry estimates suggest is a total market of $35 billion in the United States alone.
More research is needed to support use of liquid biopsies. Some observers are skeptical that early detection tests will be useful as a tool in the general population. By some estimates, around 1 percent of the U.S. population has cancer lurking in their bodies. If a test has just a 1 percent false positive rate — the rate at which a test finds cancer when there is none — the number of cancers detected could be offset by the number of people erroneously told they have the disease.
Another worry is that such tests would pick up DNA from indolent tumors that pose no threat to people who might then undergo the cost and risk of treatment. And if a test detects evidence of cancer before symptoms arise, doctors will have to find it; a potentially invasive process that could expose patients, for instance, to the cost and risk of whole-body radiation.
But proponents are confident in the potential payoff. “You get a healthy individual and you say, ‘You have cancer,’” says Nickolas Papadopoulos, a liquid biopsy researcher and professor of oncology and pathology at Johns Hopkins Medicine. “That’s never good news, but it’s much better than detecting it a few years from now, when you can’t do anything about it.”
Ahlquist predicts it will take a decade for such a test to clear the necessary scientific and regulatory hurdles, but he believes “with the different tools that are now available,” an early detection test “is in our grasp.”
Finding cancer in blood is difficult. Tumors naturally shed their DNA into a person’s bloodstream in the form or tumor cells or fragments called cell-free DNA or circulating tumor DNA. But the telltale tumor DNA is present at minute levels, and its signal can be obscured in a cacophony of confounding noise from other molecules circulating in blood. Levels of tumor DNA also vary widely among people with the same tumor types, with as many as 15 percent of advanced cancer patients lacking sufficient levels to obtain a result.
This poses a problem for interpreting any liquid biopsy, but it becomes especially worrisome for companies that want to create a test for early detection. Even a small fraction of false positive results could lead to thousands of people bearing the anxiety, cost, and risk of an unnecessary and futile workup to find the disease.
“This could be done in a way that could save an enormous number of lives and save costs for the health care system,” says Alexis Borisy, a partner at Third Rock Ventures, a Boston-based venture capital firm. “Or it could be done in a way that is disastrous.”
Guardant Health is among the frontrunners in the race for a diagnostic blood biopsy. The biotechnology company introduced its Guardant360 liquid biopsy in 2014. Oncologists have since ordered the test more than 70,000 times to match people with metastatic cancer to targeted therapies. The company is also developing a broader line of tests to monitor people for recurrence of their disease as well as for early detection.
Helmy Eltoukhy, CEO of Guardant, likens the difference between tissue and liquid biopsies to the difference between wired and wireless technology. “Everything that wireless has allowed us to do — it’s pretty mind-boggling,” he says. “We think the same thing will happen when people get access to tumor information” from blood.
Grail is another key player in the liquid biopsy face-off. In contrast to Guardant’s broad market approach, the deliberately named Grail is betting exclusively on a test for early detection. It’s a big bet: The company has raised more than $1.5 billion in venture capital and already has studies underway that involve more than 165,000 people to prove whether the test is effective. Grail has presented what it says are promising initial results from a subset of patients in a 15,000-person study. The data, presented last June at a major cancer meeting but not yet published in a scientific journal, indicate that early prototypes of the company’s test can detect roughly 40 to 80 percent of some 10 different cancers that haven’t spread and that a false positive rate under 1 percent is “feasible.” Results from a separate data set presented later in the year were comparable, the company says.
“That gives us a lot of confidence that we can develop the test” for early detection, says Aravanis, the chief scientific officer. Grail initially hoped to have a test on the market by 2019, but it currently isn’t disclosing a timetable, which some expect is likely several years away.
At Johns Hopkins University, researchers recently published promising results for an early detection liquid biopsy called CancerSeek. In a January 2018 study in the journal Science, the researchers reported that the test picked up 70 percent of eight common cancers among 1,005 people already diagnosed with disease. The eight cancers — breast, lung, colon, esophageal, pancreatic, ovarian, liver, and stomach — account for 60 percent of all cancer deaths. The sensitivity — the test’s ability to detect cancer — ranged from 33 percent for breast cancer to 98 percent for ovarian cancer.
Specificity, or the ability to rule out disease that isn’t there, was 99.14 percent: Among another 812 people without cancer who were tested, only seven were reported to have cancer.
“Most people don’t have cancer, so you can’t afford to have many false positives,” says Papadopoulos, of Johns Hopkins Medicine. “We wanted it to be 99 percent or higher.”
The CancerSeek test screens for segments of 16 cancer-related genes and eight proteins associated with specific tumor types — a panel kept deliberately small because it improved accuracy and would keep the cost below $500, Papadopoulos says. The protein analysis is a novel aspect of CancerSeek and, among other things, helped narrow the location of the cancer to a small number of sites in 83 percent of patients, Papadopoulos and his fellow researchers reported in the Science study.
“No test is 100 percent perfect,” Papadopoulos says. “This test is going to be used in the whole context of the individual being tested.” Several companies have licensed technologies from Hopkins related to cancer genetics developed by Papadopoulos and other Hopkins scientists. They have received equity or royalties from some of those companies.
Other contenders in the broader liquid biopsy space include Foundation Medicine and Genomic Health, two companies that pioneered the DNA analysis of tissue biopsies to guide cancer treatment. Mayo Clinic is working on a test with corporate partner Exact Sciences; labs in Canada, Europe, and Asia are also in the hunt.
Not all of the news is positive. Studies looking at whether rival tests yield similar results have been mixed, raising concerns about accuracy. Companies’ tests have different gene panels, for example. If they were given to the same patient, would they provide the same result? If not, which would be correct? “I don’t know,” says Anirban Maitra, a pathologist at MD Anderson Cancer Center in Houston, Texas. “As more patients are getting profiled,” he says, “I hope there will be a more systematic comparison that will help us decide what is real and what is not.”
Few liquid biopsy tests — including the ones already on the market — have been validated by the rigorous clinical trials necessary to establish their value in day-to-day practice. Yet a steady stream of research into what can be learned from circulating tumor DNA points to its potential to transform cancer care. Use of the early liquid biopsy tests — for treatment matching and monitoring — is growing, and insurers are beginning to pay for them. Guardant says Blue Cross Blue Shield, Cigna, and Medicare are among insurers that cover its $3,000 test for at least some cancers.
Oncologists say liquid biopsies could have an especially significant impact in monitoring people who have already been treated for cancer to get an early read on whether they are responding to their medication or if the disease is coming back.
Currently, imaging tests are used to assess how a person’s disease is reacting, but some studies show that circulating tumor DNA levels rise weeks to months before progression would be visible on a scan. Plus, with frequent blood tests, “you can see things changing over time versus a single measurement” from tissue or a scan, says MD Anderson’s Mitra. “We want to be armed as much as possible to know when the disease is going south, so we can change therapy or give patients other options.”
A recent study of 48 children with a devastating brain cancer called diffuse midline glioma makes a compelling case. In such patients, a biopsy to direct therapy or monitor a child’s response “is brain surgery,” says Sabine Mueller, a pediatric neuro-oncologist at UCSF Benioff Children’s Hospital in San Francisco and a co-senior investigator on the study. A blood test could offer a less invasive approach.
The children in the study were diagnosed with a tissue biopsy and treated with radiation. A liquid biopsy test was used to monitor their response. The test successfully detected mutations associated with the disease; in 12 children, the test revealed a significant drop in tumor DNA after treatment, an indication the cancer had receded. That result was confirmed by MRI in 10 of the 12 kids.
Mueller calls the results “exciting” but says more research is necessary before the test can be used in patient care. “It’s definitely not ready for prime time,” she says.
Some similar tests are on the market and could cut costs dramatically. Last February, Genomic Health introduced a liquid biopsy for people with advanced prostate cancer. The test, developed by Epic Sciences of San Diego, detects a genetic mutation called AR-V7 that predicts a poor response to the drugs Xtandi and Zytiga. The drugs have significantly extended the lives of men with advanced prostate cancer but carry list prices of more than $100,000 a year.
A 2016 study by researchers at Memorial Sloan Kettering Cancer Center in New York showed that men with the AR-V7 mutation are better off switching to a much cheaper chemotherapy regimen. “We want to outsmart the cancer by being proactive as opposed to being outsmarted by the cancer,” says Steven Shak, chief medical officer and chief scientific officer at Genomic Health. Medicare began covering the test in December.
People with early stage tumors that are surgically removed before the disease metastasizes are also ideal candidates for liquid biopsy monitoring. Surgery is intended to cure such patients, but if surgeons miss even a few tumor cells, they can become the seeds of an almost certain relapse.
For instance, people with Stage 2 colon cancer generally aren’t prescribed chemotherapy after surgery, but research shows that 15 percent still eventually relapse. The problem is there’s no way to determine which ones. A recent study of 178 Stage 2 colon cancer patients found that those who had tumor DNA in their blood after surgery were 18 times as likely to suffer a relapse as those with no circulating tumor DNA. A liquid biopsy soon after surgery could offer critical guidance on who should get additional treatment.
“We haven’t proved that it works clinically,” says Bruce Chabner, MD, emeritus clinical director of the Massachusetts General Hospital Cancer Center in Boston and co-author of a recent paper on the promises and challenges of liquid biopsies. “But the studies are logical, and they’re going to be done.”
Research is moving forward, but the prospect of a superior diagnostic test that finds cancer in people before they even have symptoms remains the holy grail. While he commends the efforts to create such a test, Maitra, the MD Anderson pathologist, says that of the various possible uses for liquid biopsies, early detection “is on the shakiest ground.”
Proponents acknowledge the challenges but say they are encouraged by early data. To improve the accuracy of their tests they’re using advanced computational methods and extending sequencing beyond tumor DNA to a variety of gene regulators that can affect cancer growth. While prior studies have used people who already knew their cancer status, both the Grail and Johns Hopkins research groups have launched studies involving tens of thousands of patients who haven’t been diagnosed with cancer. The expectation is that a small fraction of participants will develop the disease over the course of the trials, enabling researchers to see how well the tests perform in a real-world setting.
Even if these and other studies establish that such tests reliably detect early stage disease, it won’t be enough. To convince doctors and patients to use them and insurers to pay for them, researchers will still have to show that the tests are clinically useful — that they lead to treatments that cure disease, reduce costs, and improve or prolong patients’ lives.
“That’s what is required to move the standard of care,” says Eltoukhy, CEO of Guardant. “Everything else is a waste of time for patients and the health care system.”