The following is a guest post by Dr. Richard Hoffman, who is Professor of Internal Medicine and Epidemiology, and Director of the Division of General Internal Medicine for the University of Iowa Carver College of Medicine/Iowa City VA. Building on yesterday’s post from Harold DeMonaco, Hoffman raises some new concerns about the concept of a universal cancer screen and the obstacles that such a test would face to prove its utility and effectiveness.
Modern-day knights of the round table, including a former director of the National Cancer Institute and several titans of industry, are questing for a Holy Grail of cancer control—blood tests for “pan-cancer” screening. While financial experts see this as a potential unicorn for investors, the health benefits for patients may prove more illusory.
The scientists behind the start-up company to be called “Grail” hope to use genetic sequencing to detect “just a few molecules of errant DNA in a vial of blood.” In a hype-laden news release, the company alludes to the importance of achieving a very high accuracy level with the new tests, but doesn’t specifically spell out the well-known challenges posed by false positive and false negative results.
Some of the news stories about the announcement — and I looked at several from major (and not-so-major) outlets including TIME, Bloomberg, CNBC, Fast Company, and Business Insider — addressed these concerns, but most seemed to accept the company’s boastful claims at face value and seemed awed by the billionaires and the bucks that are backing this venture. None that I read — other than a thoughtful and thorough New York Times report by Andrew Pollack and a piece in Forbes by Matthew Herper — went anywhere near what I would consider to be the major concerns and drawbacks that such a test presents.
The issues are complex.
First of all, despite claims that this technology is an alternative to tissue-based biopsies, a positive result will undoubtedly lead to invasive diagnostic testing because clinicians generally require tissue diagnoses before initiating what are often risky treatments. And the screening will be successful only if a potentially harmful cancer is found and cured. I emphasize harmful because we’ve learned from our experiences of overdiagnosing prostate and breast cancer that we can find cancers so indolent that treatment will lead to only harms and no benefits.
Another concern is that the few molecules of errant DNA could be heralding such an early stage of cancer that the body’s immune system might be able to successfully destroy these cells. Thus, these positive findings could lead to unnecessary invasive testing to find a non-existent cancer.
Finally, I would be concerned that “pan-cancer” screening tests might identify markers that could be associated with multiple cancers. This is a problem with the fecal DNA tests for colorectal cancer screening. These tests can detect a genetic abnormality in stool that could arise from a malignancy in the colon—or from upper gastrointestinal or airway malignancies. Such findings could prompt extensive diagnostic work ups with limited guidance as to where a cancer has most likely arisen.
An even bigger issue for me is knowing whether testing is actually beneficial before launching a new screening program. Establishing the accuracy of the test is just the first step in that pathway. We also need to demonstrate that early detection and treatment will, on average, increase the lifespan of the persons being screened. All too often we rely on survival time—how long someone lives after cancer diagnosis–as a metric of screening effectiveness. The problem is that any test finding a cancer before it causes symptoms will increase survival time merely by moving up the time of diagnosis. When increased survival time does not translate into increased longevity we are dealing with a lead-time bias, meaning that patients will be diagnosed with a cancer and subject to the harms and costs of treatment years before necessary.
I would argue that implementing an unproven screening program could violate the medical affirmation to “first, do no harm.” We need to conduct randomized trials comparing disease-specific mortality between screened and unscreened populations. Unfortunately for screening enthusiasts and investors, these studies require enrolling tens of thousands of patients and following them for at least a decade in order to know whether screening is efficacious—and to be able to appreciate the potential harms of screening. The Bloomberg report notes that a large-scale clinical trial of a single cancer condition will be initiated in 2017 with expectations for having the test on the market in 2019. To me, this implies that the clinical trial will just evaluate the accuracy of the test, but provide no meaningful information on whether using this test improves patient outcomes.
A test that could cost nearly $1,000 could be a boon to investors seeking a unicorn but might not be a Grail for patients seeking to make wise healthcare decisions.
Dr. Hoffman gratefully acknowledges editing assistance with this post from Kevin Lomangino.