Imagine you could go to your doctor once a year for a blood test that would detect nearly every common type of cancer at an early stage. If you had a positive result, some further testing could pinpoint the cancer so you could have it treated before it spread. I would be ecstatic to have a simple tool for my patients that could detect cancers like ovarian and pancreatic that currently have no effective screening approach.
No surprise, then, that a host of companies are working on molecular diagnostics to detect early-stage cancers or pre-cancers using non-invasive methods. The most popular approach is screening blood for circulating tumor DNA (ctDNA), which several companies claim to reliably detect at very low levels in the blood. Grail Bio, a spin-off of DNA sequencing company Illumina, is the largest player in the field. Grail is working to develop a “pan-cancer” screen that it hopes will be used in the general population, potentially for millions of patients worldwide.
Despite the excitement I feel about the potential of ctDNA cancer screening, I can’t help but feel skepticism also. If these tests are to be applied for routine screening in the general population, they can’t just be good—they have to be nearly perfect. When screening for a fairly uncommon condition like pancreatic or ovarian cancer, even a one or two percent false positive rate would mean that many times as many people with a positive test result would be falsely positive as truly positive. These false positive results would lead to additional testing (CT and MRI scans, biopsies, and possibly even unnecessary surgeries), not to mention unquantifiable and unnecessary anxiety. What’s more, these tests may detect early-stage cancers that pose no real threat to patients. Overdiagnosis is both costly to the system and harmful to patients.
The PSA test represents a cautionary tale for these companies. A blood-based marker for prostate cancer, the PSA is similar in concept to ctDNA-based screening, although it is less specific for cancer than ctDNA. For years, the PSA was considered a key part of any middle-aged man’s annual physical. Yet in 2009, data from a large trial demonstrated that men who were screened had the same or higher risk of death as men who were not screened. Based on these data, the United States Preventive Services Task Force (USPSTF) recommended against screening for prostate cancer. PSA testing rates plummeted. Medicare has even considered penalizing doctors who order screening PSAs.
Cologuard, a colon cancer screening tool developed by Exact Sciences, provides an alternate example. The test can detect abnormal DNA in stool samples, and when not present, allows a patient to avoid the discomfort and inconvenience of a colonoscopy. Despite winning FDA approval in August 2014, uptake of Cologuard has lagged behind its scientific promise. The USPSTF, possibly in light of past experience with tests like the PSA, initially only recommended the test as an alternative to colonoscopy in select circumstances. Exact Sciences’ share price fell from a peak above $30 to below $8 on the USPSTF announcement. In 2015, only 104,000 Cologuard tests were performed (compared with 14 million colonoscopies). It was only last month that the USPSTF revised its recommendations and put Cologuard on an equal footing with colonoscopy.
Yet I worry that, because Cologuard has a 13 percent false positive rate, it will lead to many follow-up colonoscopies and add to patients’ anxiety and uncertainty. Because the Cologuard must be done every three years instead of every ten for colonoscopy, it is possible that the Cologuard won’t be a substitute for colonoscopies, but instead an expensive supplement.
To mitigate the problem of false positives, the simplest approach is to target people with a higher risk of cancer. This may mean people with cancer in remission, those with a known mutation like BRCA, or people with a strong family history of cancer. Because these groups are more likely than the general population to have cancer, there will be fewer false positives for each true positive cancer detected.
Guardant, another ctDNA detection company, has embraced this approach, as Xconomy’s Alex Lash has written. Similarly, PapGene hopes to use Pap smear samples to test women with BRCA mutations for ovarian cancer.
While it is less ambitious than Grail’s approach, this targeted strategy offers companies a faster path to market and reduces the risk of overdiagnosis that comes with general population screening. Better to study the test in a high-risk population more likely to benefit before exposing the general population to a test with poorly understood performance.
Once the tests have demonstrated excellent performance in high-risk patients and have won USPSTF support, I will happily offer them to my own patients as an effective screening tool.
Alex Harding is a resident physician in the primary care track of Internal Medicine at Massachusetts General Hospital. He has no financial interests to disclose. Follow @alexharding7