So Cancer Screening Doesn’t Actually Save Lives?

Cancer screening has been showing up in the news a lot lately.

First, there was Kim Kardashian’s Instagram advertisement for a company named Prenuvo that does full-body MRI scans on healthy people to look for cancers (and anything else they can find). Another company getting attention for cancer-hunting, but in their case using DNA rather than imaging, is Grail. Grail was originally formed as a spinoff of the DNA testing company Illumina, then became a unicorn, raising over $2B in investment before being re-acquired by Illumina in 2021. So clearly, a lot of moneyed people think that Grail is going to make a lot of money with their screening tests. And then today I saw this truly bizarre news story from England about putting cancer screening messages on underwear labels, proof that the normally-rational National Health Service has some blind spots.

As you can tell from the title and tone of this newsletter, I’m really skeptical about whether any of this testing is medically beneficial. The Wall Street Journal’s Alex Janin responded to the Kardashian Insta incident with this article. And JAMA Internal Medicine just published this opinion piece on multi-cancer screens such as the one sold by Grail. Both articles explain how the harms from these tests may exceed their benefits. The JAMA Internal Medicine piece argues that ethically, Grail should limit its test to randomised clinical trials until such time as there’s strong evidence for net benefit. (Note: Grail is apparently funding a large clinical trial of its test in England, so I give them credit for that much. I cynically suspect they’re doing it because they know there’s no way the UK National Health Service would ever pay for their test in the absence of randomised trial-level evidence. But in the mean time, here in the US, Grail is pushing their test hard via marketing rather than clinical trials. And my guess is that useful outcomes data from the UK will take at least a decade, because the main benefits of screening show up down the road.)

So what about more traditional cancer screening, esp. mammography and colorectal cancer screens? They work, right? With all the public service ads you encounter on television, radio, print, internet, billboards, sides of busses, and on and on, these must be good tests, right? Well, it’s not clear that that’s true. In that same JAMA Internal Medicine issue was this article, estimating the average amount of life gain from these more common cancer screening tests. And they found that these tests don’t give the average person much if any additional length of life.

Why don’t the tests work? If you want a deep dive into the explanation, I recommend this excellent book by Dr. Gilbert Welch: Should I be tested for cancer? Maybe not, and here’s why. If you don’t have time for a book (noting that it’s actually a pretty quick read), here’s an excellent interview with oncologist Dr. Vinay Prasad on Russ Roberts' EconTalk podcast.

And if you don’t have time for either the book or the podcast episode, here’s my own very brief explanation:

First of all, cancers are heterogeneous. Some are so slow growing that they’ll never kill you (at least, not before something else gets to you first). These cancers are therefore harmless, unless they’re caught by a cancer screening test, at which point you’re likely to be subjected to surgery and/or chemotherapy and/or radiation. And these can obviously cause harm to someone who doesn’t actually need them. Treatment for these slow-growing cancers happens a lot; it probably represents the majority of cancers caught by screening programs. Ironically, even though  these patients are being harmed by unnecessary treatment, they are almost always happy, because they attribute their survival to the treatment rather than the fact that the cancer was never going to kill them in the first place. Why do doctors keep treating these tumors? Partly because there’s just enough unpredictability with cancer, that you can’t guarantee that a low-grade lesion that you’re looking at under the microscope might not turn out to become aggressive down the road. Even when the best available statistics are telling you that this is very unlikely. And partly because cancer is so scary to most patients that they want to err on the side of "getting rid" of it. One exception to this rule: there are some low-grade cancers where there’s broad agreement that they won't turn aggressive, such as in the case of most thyroid carcinomas. For these, there’s a growing movement for pathologists to call them something other than "cancer" so that doctors and patients won’t feel obligated to treat them.

At the other end of the spectrum are the aggressive cancers that kill you regardless of whether you try to screen for them. These represent a smaller portion of cancers caught by screening, because they typically progress from undetectable to symptomatic in too short a time for an annual screening program to pick them up. Just like with the nonaggressive tumors, screening for these more aggressive ones is useless.

Are there any cancers for which screening programs could actually reduce mortality? Yes, but not many. Cervical cancer is probably the best example: most cervical cancers are medium-slow growing, in that they’ll kill you eventually, but it will take a decade or two before it happens. Plus the fact that they can start when you’re young, meaning that there’s time for them to progress before you might die of heart disease or stroke or Alzheimers or something else. Plus the fact that they can be surgically cured in the early stages, but aren’t curable in late stages. (Contrast this with testicular cancer, which is highly treatable even in late stages, and so for that reason is pointless to screen for.) Some breast and colon cancers probably fall into this intermediate category, but it’s hard for doctors to separate them cleanly from the slow and fast groups, which complicates the calculus for breast and colorectal cancer screening.

So if you’re a woman in an applicable age range, then absolutely get tested for cervical cancer (HPV test). You probably don’t need an annual Pap smear, especially if you’re HPV negative. The specific recommendations for cervical cancer screening depend on your age, and are summarised here. And of course, both boys and girls should be getting immunised against HPV.

But for other cancers, screening should be a personal choice. Mammography *might* reduce your risk of dying of breast cancer slightly, but even if it does, it reduces it so little that clinical trials haven’t shown a strong effect. Colorectal cancer screening might likewise reduce your mortality risk slightly, but it’s a small effect. And for heaven’s sake, don’t start with a screening colonoscopy unless you’re at elevated risk (such as having a family history of early colon cancer). Instead, get the FIT stool test like I did recently. It’s much more cost-effective, doesn’t require you to take a day off of work, and doesn’t expose you to the small but non-negligible risk of perforation from the procedure.

As for screening for prostate cancer, thyroid cancer, pancreatic cancer, etc.: Unless you’re at elevated risk (family history or genetics), I suggest you channel Nancy Reagan and Just Say No.

Final note: I hate that I have to say this, but doctors have been sued for recommending against cancer screening. So please be aware that reading this newsletter does not mean that you and I have a doctor-patient relationship. This newsletter does not constitute professional medical advice. It’s just my scientifically informed opinion, which I also share broadly with friends and family members. If, after reading this newsletter, you have questions about cancer screening, I formally recommend that you discuss them with your actual doctor, with whom you have an actual doctor-patient relationship.