“What if we could not only detect a cancer’s presence but divine its intent?” A science startup has developed an early-detection test for multiple cancers, using DNA found in blood samples. It’s a remarkable new technology—so what’s the catch? And, then, Robin Wright on Donald Trump’s statements about Iran. Plus:

Hannah Jocelyn
Newsletter editor

For a piece in this week’s issue, Siddhartha Mukherjee, a physician who writes about medicine and science for The New Yorker, investigated groundbreaking advancements in early cancer detection. I spoke with Mukherjee about how these discoveries might change the way we test for and treat the disease. Our conversation has been edited and condensed.

Current medical guidelines have us screening for only five cancers, and one at a time. A new test, from a startup called Grail, has identified fifty cancers, with a remarkably low rate of false positives. How do tests like this work?

Cancer cells, like all cells, shed their DNA into the bloodstream. And now, with new sophisticated technologies, we can detect even a minuscule amount of DNA in the blood. A process known colloquially as a “liquid biopsy” is a mechanism by which you can look at the DNA that’s been shed out of cancer cells somewhere in the body—whether that cancer is in the breast, ovary, or lung. There is a chemical signal of cancer in the blood, and, with liquid biopsies, science has become sophisticated enough to identify cancer from simply drawing blood rather than through imaging or physical exam. It’s just an amazing process.

How soon might this kind of test be widely available?

The hope is that this technology will become very widespread in the near future. It’s a wonderful technology, extraordinarily exciting. But that excitement has to be tempered by the idea that, even if you are able to detect cancers, we may not change cancer-mortality rates. My piece in this week’s issue really tries to separate the wonder of these incredible technologies—it’s amazing that we can do this essentially with just a pint of blood!—from the realities of how difficult it is to do early screening and to save lives.

What does an Enlightenment-era Presbyterian minister and mathematician have to do with this story?

You’re referring to Thomas Bayes, who was born in 1702. What he realized, using mathematics, is that the question we should be asking is not whether a test is accurate; it’s whether a test is accurate for a particular population of people. Because that number will vary, and that’s an incredibly important realization—the idea of prior probability. The more you refine probability, the more you refine the value of a test. If you apply screening in general to a large, random population where the prior probability is low, the chances are that the test will not perform well. But if you change the population—to, say, just women with a prior or family history of breast cancer—the likelihood is that the screening will bring up more positives.

It sounds counterintuitive, because we think of tests as absolute. But, really, it depends on the population being tested. And he figured this out by demonstrating, in a very philosophical way, that all knowledge is conditional knowledge, and that conditional knowledge is the important thing—especially in the real world. It’s an incredibly important insight because it should and can and will influence how clinical trials are run in the future.

Do you think there is such a thing as knowing too much about our health, in this era of technological advancement?

The truth about your health depends on what you know about your priors—your family history, your exposures. I don’t think we should know too much about our health, but what we do know should be informed by our understanding of conditional probability.

Read the story