Of 287 people whose blood test came back positive, 173 turned out to have cancer. That ratio, 60.3%, is the headline number from PATHFINDER 2, a study published on September 22 in Nature Medicine. It is a positive predictive value: the chance that a positive result is right. It does not mean the test finds six in ten cancers, a mistake that is easy to make.
The test is a multicancer early detection (MCED) blood test. It looks for patterns in tiny pieces of DNA shed by cells into the blood, then guesses where in the body a signal came from, so doctors know where to look. The study enrolled 35,878 adults aged 50 and older, and 32,007 of them made it into the performance analysis. Signals turned up in 17 broad categories of cancer, including ovarian, lung, breast and colorectal, plus many that have no recommended screening test.
The numbers that matter
Specificity came in at 99.64%, so healthy people were rarely flagged by mistake. Where the test did flag a true cancer, it got the origin right, within its top two guesses, in 91.3% of cases.
Sensitivity tells a more modest story. Across all cancers diagnosed in the 12-month window, the test caught 39.3%. For a prespecified group of 12 cancers, the figure was 69.8%, and the authors stress that this applies only to that group. Performance varied by cancer type and stage.
Among 151 participants with a newly diagnosed primary cancer spotted by the test, 53% were at stage 1 or 2 while 70.9% were no further along than stage 3. In addition, 72.8% of the cancers it found belonged to types the US Preventive Services Task Force does not back with a grade A or B recommendation.
What the study cannot say
“One of the challenges in cancer screening is that most cancers do not have an established screening test,” said Dr. Karthik Giridhar, a Mayo Clinic medical oncologist and senior author. In his view the study helps show how a blood test might complement existing screening. He added a caution: for now the tests cannot stand in for established screening.
The design has limits. PATHFINDER 2 had a single arm, with descriptive endpoints, no comparison group and no formal hypothesis testing. Randomized trials and longer follow-up are needed to learn what the test is worth in routine care, and it does not show that using it lowers cancer deaths. The version tested is an earlier one that a commercial version has now replaced. GRAIL Inc. designed and funded the work and took part in analysing the data, and Giridhar reports consulting and advisory ties to the company.
The encouraging part is that a single blood draw found signals from cancers that screening usually misses. Patients should keep up their recommended screening, and talk to their doctor about which tests suit them. A negative result does not rule cancer out.
Sources
- Multicancer Blood Test Detects 17 Cancer Types in Large Prospective Study (Mayo Clinic)
- Multicancer Blood Test Detects Signals From 17 Cancer Types (Medindia)
- Performance and safety of a multi-cancer early detection test: the PATHFINDER 2 study (Nature Medicine)
