Cancer — 11

The test with the most extensive clinical experience to date is known as the GRAIL Galleri. This blood test interrogates DNA methylation via bisulphite sequencing for about one hundred thousand sites that are associated with cancer. It uses AI to determine whether there is a “cancer signal” and, if present, to help localize its likely organ source. In a prospective clinical trial of over sixty-six hundred participants aged fifty years and older, 92 percent of whom were white, a cancer signal was found in ninety-two individuals (1.4%). Of these, thirty-five individuals (0.5%) were found by additional workup to have cancer, which is a positive predictive value of only 38 percent. The AI prediction of cancer site of origin accuracy was 97 percent. Given the diverse sites of origin, including bone, head and neck, ovary, liver, uterus, blood, breast, lung, and gastrointestinal tract, that’s quite good. However, the small number of cancer cases detected were mostly at stage 3 or 4, already metastatic.

The fifty-seven individuals with a false-positive cancer signal indicate a low false-positive rate (specificity of 98%). Of interest, it is possible that the abnormal cell-free methylation signal in some individuals may reflect a transient appearance of microscopic cancer that was quashed by the person’s immune system. Anecdotes of individuals who had a false positive and on repeat testing were negative suggest longer-term follow-up to see how common this is. Strengthening our immune system might be a preventive strategy worth pursuing.

The problem with this test that is marketed for $949 for people aged fifty and older is that the yield is remarkably low. Only five per one thousand people test positive, and, of these, just two per one thousand were actually diagnosed early, in stage 1 or 2, before the cancer spread. The proportion of false negatives among every one thousand people tested is unknowable, but, and I’m sorry to say it, anyone who gets back a report of “no cancer signal detected” should not have confidence that they are not incubating a cancer.

The detection rate for early cancer (stage 1 and 2) is poor compared with later stages. Picking up cancer at the microscopic stage, before it can be seen on a scan or induce symptoms, or potentiate spread, is the key to improving outcomes, using interventions to change the natural history of cancer progression. Simply put, detection is not equivalent to favorably altering outcomes. That is why randomized trials are necessary, such as one by the National Health Service in the UK of 140,000 participants and the National Cancer Institute in the United States of 225,000 participants. These very large trials of healthy, asymptomatic people are again using age as the main inclusion criterion. It will be years before such trials will be complete for clinical outcomes, and there will be questions about the diversity of the participants, since we know the occurrence and behavior of many cancers are shaped by a person’s ancestry. But the problem of lacking compelling evidence from randomized trials is perhaps not the biggest problem. None of the tests are yet FDA approved or reimbursed by insurance carriers.