The growing case for doing less: How harmless cancers are being overdiagnosed in America

Fortune, By Carolyn Barber on October 2, 2023


In 2009, Laura Esserman, a breast cancer surgeon and oncology specialist in San Francisco, co-published an article in the Journal of the American Medical Association (JAMA) suggesting that it was time to rethink routine screening for breast and prostate cancer. The current approach, she wrote, wasn’t reducing aggressive or late-stage disease as much as had been hoped. Instead, it was leading to overdiagnosis.

Almost 15 years later, “The amount of hate mail has gone down,” says Esserman, now the director of the University of California San Francisco Breast Care Center. “But what has happened is that women have voted with their feet. People have flown in from all over the world and all over the country for me to do nothing when they have DCIS.”

Ductal carcinoma in situ, a condition sometimes called non-invasive or stage-zero breast cancer, is a very early finding of disease in the cells that line the milk ducts of the breast. For decades, the diagnosis of DCIS has routinely led to surgery–a mastectomy or a lumpectomy (a partial breast resection) that’s often combined with radiation treatment and possibly, a five-year course of hormone medication.

The issue? In a study of 100,000 women who were followed for two decades, patients who’d been diagnosed with and treated for DCIS ultimately had about the same chance of dying from breast cancer as those in the general population. And while Esserman was making this case years ago, she now finds herself part of a growing number of researchers and experts who are pushing back on the medical industry’s tendency to overtest, overdiagnose, and overtreat patients for conditions that otherwise might never affect their lives.

“Getting a biopsy is not pleasant, and 75% of all the biopsies we do turn out to be nothing. You stick a needle in the breast and sometimes you’ll see these little calcifications that are benign–but incidentally, there’s focus on DCIS, and the next thing someone’s getting a bilateral mastectomy. You think those things don’t happen. They happen all the time,” Esserman says.

The overdiagnosis industry

An analysis published last year in the Annals of Internal Medicine estimated that among women aged 50 to 74 years, about 15% of screen-detected breast cancer is overdiagnosed, meaning it never would have caused any symptoms or problems. The Annals study estimated that significant numbers of women over 70 are potentially overdiagnosed for breast cancer, including nearly half of those aged 75 to 84. And that was down from a 2012 NEJM paper, which suggested that overdiagnosis occurred in nearly a third of all newly diagnosed breast cancers.

The pattern extends to multiple cancers, including those affecting the breast, prostate, melanoma, and thyroid. One common thread is that with the growth of diagnostic technologies, it’s now possible to test for these cancers in more refined ways and at earlier stages, including, several experts contend, stages at which the cancers often pose little or no risk to the patients involved.

In August, a meta-analysis of 18 randomized clinical trials involving 2.1 million people, published in JAMA Internal Medicine, concluded that “current evidence does not substantiate the claim” that common cancer screens (mammography, colonoscopy, prostate-specific antigen (PSA) testing, etc) save lives, by extending lifetimes with the possible exception of sigmoidoscopy, for colon cancer.

“In our exuberance to find these cancers, we have basically turned a lot of healthy people who are not destined to die from the cancers into patients,” says Ade Adamson, a cancer screening expert and dermatologist at Dell Medical School in Austin. Adamson notes that screening 1,000 women for breast cancer will statistically avoid one cancer-related death. However, “a significant portion” of the other 999 screened, he says, will end up having “false alarms” and may have to undergo further tests, including painful biopsies, anxiety, and sleepless nights over the prospect of having cancer. Ultimately, some will undergo unnecessary surgeries and treatments–and could even be harmed from all of the treatments prescribed.

“The general picture with all of these screening tests is that they are quite good at finding additional disease,” says Luc Morris, a surgical oncology specialist and researcher at New York’s Memorial Sloan Kettering Cancer Center. “But to show that that actually has a benefit to the population, you have to show that you’re decreasing the risk of dying–what we call all-cause mortality–and there’s only one cancer screening test (sigmoidoscopy) that decreases your risk of dying.”

Yet cancer diagnoses via screening in the U.S. have grown exponentially. A 2021 article in the New England Journal of Medicine found that the documented incidence of melanoma, “once a rare tumor,” was now six times higher than 40 years ago. However, the rate of mortality due to this skin cancer has remained generally stable. No randomized controlled trials or population-level studies have shown benefits for melanoma screening in reducing deaths from melanoma.

“I think, in general, we have overinvested in cancer screening,” says Gilbert Welch, an internist and lead author of the NEJM article. “Its benefits have been systematically overstated, its harms largely ignored…Screening has become a big business both for health systems and diagnostic companies.”

The issue can be vexing for health care professionals. The reality is that we’re better at detecting things than we are at knowing much about them, says Rita Redberg, a cardiologist at UC San Francisco Health and former editor of JAMA Internal Medicine. In the case of breast cancer, it is difficult to predict with certainty whether a tiny mass will become malignant, grow substantially, or even disappear entirely. Still, these findings prompt near-immediate action.

“It triggers biopsies; it triggers treatments,” Redberg says. “We have a whole medical system that, once you buy the equipment (and) train on how to do something, it gets reimbursed–it’s kind of an unstoppable train.”

“I don’t think people feel better from all these screening tests we do, and they lead to a lot of complications,” Redberg says. “I think we would make a lot more of an impact if we worked on public health campaigns to stop smoking, stop vaping, increase physical activity, and improve our diets. That would really reduce cancer–and people would feel better.”

For Laura Ferris, professor and director of the clinical trials unit in dermatology at the University of Pittsburgh Medical Center, the overarching complication of screening is that, “While we can say that there is likely overdiagnosis of melanoma, we don’t have much that we can do about it on an individual patient level. I. cannot distinguish between a melanoma that is overdiagnosed and could be left on the patient and one that is likely to behave aggressively.”

For decades, the PSA blood test was recommended to screen men for prostate cancer despite scant evidence that it was actually helping to lower all-cause death rates. A meta-analysis of five randomized controlled trials concluded that PSA screening, at best, had a small effect on prostate-specific mortality–one death avoided for 1,000 men screened over 10 years–and no effect on overall survival.

Until about a decade ago, most low-risk prostate cancer initially detected by PSA-based screening (low-risk is the most common type identifiedwas treated with surgery or radiation, with both incontinence and erectile dysfunction as common side effects. In recent years, clinical practice has evolved so that more men in the U.S. (60% per one study) are choosing active surveillance (AS), which involves close monitoring of the cancer without treatment until or if it becomes necessary. But as the study’s authors note, the use of AS varies widely by urology practices and is barely existent in some, which the authors called “suboptimal.”

Behind all of these numbers lies a larger question: Do we need to find every cancer? At autopsy, about 40% of men over 80 have incidental cancer in their prostate, explains Morris, but they died from other causes. ”And 10 to 30% of us have (clinically insignificant) thyroid cancers when we die.” New diagnoses of thyroid cancer in the U.S. are triple what they were 50 years ago, and recent literature links this to both the increased use of diagnostic imaging and fine-needle aspiration biopsies.

A better way

Increasingly, physicians and researchers are looking for better, more patient-aware approaches not only to cancer identification but also to the results. Morris and his team at Memorial Sloan Kettering have published the largest U.S. data set of more than 500 patients with papillary thyroid cancer treated by active surveillance. The results? “Eighty percent of these tumors just sit there and don’t grow,” Morris says. “And for those that do, we can do the identical operation as we would have done on day one.”

In the case of breast precancers, a prospective randomized trial currently in progress, COMET, will compare the results of women diagnosed with DCIS who undergo standard treatment (surgery and/or radiation) with those who remain on active surveillance, with regular checkups to be sure the DCIS hasn’t turned into invasive cancer. Patients in both groups may also choose a hormone-blocking treatment that for some women prevents DCIS from spreading.

The idea of simply monitoring for changes “is a total reset of what we think the disease is,” says Shelley Hwang, director of the breast oncology program at Duke Health and the principal investigator of the trial. “Because of that, it’s really hard to change hearts and minds. But I think we need to change both.”

Hwang told me that while more than 50,000 women are diagnosed with DCIS each year (roughly 65 million American women undergo mammograms annually), a much smaller percentage of that total goes on to become any form of invasive cancer. Currently, though, almost any woman diagnosed with DCIS would be scheduled for a mastectomy, lumpectomy, and/or radiation therapy, and possibly hormonal treatment.

“I think the strategy right now is, let’s treat 100% of the patients, and hopefully we’ll have helped some,” Hwang says. “The alternative strategy is to let the biology dictate which ones need treatment–and don’t harm the others.”

At UCSF, Esserman has worked in collaboration with others for years to develop treatments for invasive breast cancers that might be offered before surgery–hopefully, instead of surgery. “Then I thought, ‘Why can’t we do the same thing in DCIS?’” she told me. A study called RECAST DCIS, planned to begin this fall, will pair active surveillance of those diagnosed with DCIS with endocrine therapy, and Esserman and many others believe that knowing more about patients’ personal histories, including genetic background, can and should factor into any decisions about care. “I think this is a very thoughtful way, an actionable way to make change,” says Esserman.

Increasingly, it is becoming apparent that mass screening hasn’t achieved what many hoped it would, and for too many Americans, it does more harm than good when applied to early-stage cancers.

A more balanced and nuanced approach that better stratifies individuals on the basis of risk and patient preference is surely welcome. As Hwang says, “Let’s screen the patients who we know will be at highest risk, and back off on the patients who have the lowest risks.” But none of it will come easily in a country that long ago established screening–and its resulting treatments–as a standard of care.

This article originally appeared on