Mary Chris Jaklevic is a reporter-editor at HealthNewsReview.org. She tweets as @mcjaklevic.
We often tell our readers to run for the hills whenever they see a news story touting a “simple blood test.”
You won’t find a better example of why you need to be wary than an exchange between a reporter and a researcher at a news conference at the American Society of Clinical Oncology’s annual meeting over the weekend.
A news release heralded a study to map blood biomarkers in people with cancer as bringing the world “one step closer to being able to detect early lung cancer from a simple blood test.”
The headline even referred to a “blood test” for lung cancer.
But when asked some pointed questions that any consumer might wonder about, researcher Geoff Oxnard, MD, of the Dana-Farber Cancer Institute was compelled to explain that this test isn’t a reality.
Alice Goodman, a reporter for the ASCO Post, asked Oxnard how blood analysis stacked up to a low-dose CT scan, the current screening method, in detecting lung cancers.
Oxnard acknowledged that’s unknown. He noted that unlike a true screening, where apparently healthy people are checked to find out whether or not they have cancer, analysis of the new approach has so far included people whose cancer status was already known, “so we can’t really make that comparison.”
Goodman then asked how Oxnard envisioned the “blood test” being used.
The researcher attempted to clarify: “This is not a blood test. This is a sequencing method that needs to be turned into a diagnostic.”
The journalist asked if the words “simple blood test” used in the news release were misleading.
Oxnard responded: “It has the potential to be a simple blood test.”
Goodman expressed confusion, prompting Oxnard to insist: “It will be a blood test.”
In this case the research is being sponsored by Grail Inc., a startup backed by Microsoft Corp. co-founder Bill Gates and a venture fund of Amazon.com Inc. founder Jeff Bezos.
The company’s narrative that a blood test will be capable of detecting early-stage cancers with ease neatly aligns with ASCO’s recent concerns about low uptake of lung-cancer screening. Highlighting that early and unpublished research at ASCO’s major conference gave Grail and its researchers a surge of news coverage.
Yet some experts say based on the company’s latest data, an actual screening test appears to be years off, if it arrives at all.
Richard Hoffman, MD, director of the internal medicine division at the University of Iowa Carver College of Medicine, said in an email the rates at which the markers under study detected known cancers “are not very impressive.”
Those rates ranged from 38% to 51% for stage I-IIIA cancers, but the news release didn’t break out data for stage I cancers, which is the early stage at which you’d expect to find cancers via low-dose CT screening — the current standard.
Other omissions make it difficult to tell how well a blood test might work in the real world, he said. For example, it’s unclear whether there was an appropriate control group of heavy smokers that would indicate how accurately the markers classify those without lung cancer, and whether patients in the study were asymptomatic, as would be those subject to a screening.
Hoffman has written previously about the hurdles of developing and implementing a cancer screening test based on blood.
Predictably, some news outlets reported on Grail’s research uncritically, downplaying the many caveats.
“The coverage is typical; overly optimistic early in the reports and then some provisos later in the story,” said HealthNewsReview.org contributor Harold DeMonaco, who’s also written about the massive challenges involved with Grail’s goal of creating a blood test for cancer.
DeMonaco, via email, took issue with the heavy use of positive quotes from researchers.
For example, the Guardian’s ‘Holy grail of cancer research’: doctors positive about early detection blood test, quoted a researcher on the project saying, “We hope this test could save many lives.”
That story misled readers with a grandiose headline, failed to mention significant limitations and harms, and neglected the potential cost, according to a HealthNewsReview.org review.
That Guardian’s “holy grail” reporting made its way onto some local TV news stations.
More balanced was CNN’s Blood test offers hope of finding cancers before symptoms develop, which offered several caveats. Yet the message distilled in a tweet seemed premature:
Coverage from some news outlets “did a good job in consistently indicating that these tests were a long way from being used in clinical practice,” Hoffman said.
Notable was STAT’s Grail’s cancer blood test shows ‘proof of principle,’ but challenges remain, which refers to the company’s “out-of-the-park goal.”
STAT quoted several independent experts including one who wondered if looking for changes in blood markers would catch cancers early enough for effective treatment and pointed out that the “true measure” of a test would be whether it led to higher cure rates.
STAT also raised the problem of overdiagnosis — detecting cancers that would never threaten health:
Ideally, a blood test would not flag those cancers, since a diagnosis would send patients for more testing and, in many cases, risky treatment for no purpose. Research to determine that lies well in the future.
Also commendable was Forbes’ Biotech Firm Grail Takes The First Steps In Its Quest For A Blood Test For Cancer, which Hoffman said “gets credit for explaining the population impact of false alarms” — results indicating a potential cancer that isn’t there.
In screening, it’s critical for a test to correctly identify people who don’t have cancer, known as specificity. Even a seemingly high rate of specificity can generate a lot false alarms.
Forbes usefully showed the math:
If an eventual Grail blood test for lung cancer found 90% of lung cancers and had a specificity of 99.5%, it would still warn ten times as many healthy people as those who actually have the disease if given to everyone. That’s because only 60 people per 100,000 patients develop lung cancer. So this test would identify 54 people with cancer, and tell 500 healthy people they have cancer when they don’t.
That got to the heart of what many see as the problem with screening.
Hoffman noted that Forbes was “the only story to address the next step—a chest CT scan” and “nicely pointed out the huge evidence gap between diagnosing cancers and saving lives.”
But Hoffman disagreed with its assertion that it’s “easy to look for lung cancer” with CT scans, noting that the National Lung Screening Trial of such scans “found that 95% of the abnormal scans were false positives—and lung cancer can be diagnosed only with invasive diagnostic procedures.”
HealthDay’s Could a Blood Test Spot Lung Cancer Early? and Reuters’ Grail blood test shows promise in early detection of lung cancer had relatively balanced tones but with less sourcing and detail than STAT and Forbes.
The impact of all this on consumers isn’t totally clear. DeMonaco suggested the hype might contribute to the public’s poor understanding of the value of cancer screening tests, including a lack of understanding of potential harms. “Overly optimistic reporting reduces that ability of the public to make informed decisions,” he said.
Hoffman was less concerned, although his comments suggest a mistrust of hopeful speculation.
“Given that these tests are many years from being used in clinical practice–if ever, the practical impact of uncritical coverage is pretty minimal,” he said. “I’m sure some people will be asking their doctors for these tests, but otherwise reading these stories is like going to a world’s fair and touring the model city of the future.”
[Editor’s note: This post was revised to clarify Richard Hoffman’s rationale for why a control group of heavy smokers would be useful.]
More HealthNewsReview.org coverage of ASCO’s 2018 annual meeting: