To screen or not to screen.

Abstract

The American Cancer Society (ACS) and other nonprofit organizations have long asserted that screening increases the odds of the early detection of certain cancers, thereby increasing the likelihood that those cancers can be cured. However, controversy recently has flared over the true benefits and limitations of screening, particularly for breast and prostate cancers. In October 2015, the debate boiled over when researchers at Oregon Health and Science University in Portland published a provocative analysis.1 Vinay Prasad, MD, MPH, the article's lead author and an assistant professor of medicine in the university's division of hematology/oncology, argued that overall mortality rather than disease-specific mortality should be the true benchmark for success. If the former measure is used to gauge a screen's benefits, he says, “then actually nothing meets that bar, unfortunately. Nothing.” Dr. Prasad and other critics also have cited the potential downsides of screening, such as false-positive findings, overdiagnosis, patient anxiety, unnecessary costs, and harm from the tests or interventions. Compounding the problem, he says, is that overall survival rates associated with screening have been routinely disguised or hidden from patients. “We let them believe that we know for sure it improves survival,” he says. “I think we have totally misled the average person who comes to our office.” Otis Brawley, MD, chief medical officer of the ACS, says he believes screening is still beneficial for certain cancers. As an example, he points to the 20% decline in lung cancer mortality and 7% decline in overall mortality seen in the 2011 National Cancer Institute-sponsored randomized study of lowdose computed tomography screening among patients with an extensive smoking history.2 But Dr. Brawley agrees that physicians must do a better job of distinguishing between disease-specific and overall mortality and of spelling out both benefits and harms in their discussions with patients. “Even in those diseases where I believe screening is beneficial,” he says, “I am very concerned that the benefits of screening have been horribly exaggerated and harms associated with screening have been tremendously underemphasized.” The task of talking to patients about the pros and cons of screening tests is “still a work in progress,” says Robert Smith, PhD, vice president of cancer screening at ACS. However, he argues that cancer organizations have made strides in tempering their messages and in avoiding overpromising. “At ACS, we're pretty conscientious about describing the benefits and the limitations and harms of screening,” he says. “But in general, every cancer we screen for has been judged to have benefits that outweigh harms.” He also strongly disagrees with using all-cause mortality as an inherent measure of an intervention's worth. “A woman going for breast cancer screening is not imagining that this is going to prevent her from dying of anything. She's concerned about dying from breast cancer, and that's why she gets screened for breast cancer,” Dr. Smith says. “A preventive health intervention is designed to reduce the risk of an adverse outcome for a single cause. If it actually contributes to multiple causes, further reducing your risk of dying from anything, then that's what we call icing on the cake.” In fact, he says, the medical literature suggests that overly pessimistic decision aids that deemphasize benefits and overemphasize harms can add to patient confusion. However, Dr. Prasad insists that patients need to hear both the good and the bad in order to make up their minds. A trial that gauges cancer-specific mortality only, he contends, may miss unintended downstream consequences of the intervention. A recent meta-analysis from The BMJ led by researchers at the University of Toronto, for example, examined the risk of second malignancies in patients who had initially received radiotherapy for prostate cancer. The examination of 21 trials found a slightly elevated risk of second malignancies of the bladder, colon, and rectum compared with unexposed patients, primarily among those treated with external beam radiation.3 “You may slightly increase death from a different cancer or, as other data have shown, you may slightly increase death from cardiovascular disease or even suicide, and those deaths don't get counted against you,” Dr. Prasad says. For that reason, he says, overall mortality should be the gold standard when examining all of the downstream effects. Dr. Smith rejects that argument, asserting that all-cause mortality on its own reveals nothing about whether there may be unintended effects that are otherwise unmeasured. Nevertheless, he agrees that mortality data could help identify clusters of additional causes of death that may be attributable to the intervention. “That's actually a smart thing to do,” he says. “And in fact, many years ago we did discover excess breast cancer, excess heart disease, and lung disease deaths in women treated for breast cancer that were the result of collateral effects from radiation therapy.” The results, he says, led to a “wholesale change” in how physicians targeted the radiation beams. “We let [patients] believe that we know for sure [screening] improves survival. I think we have totally misled the average person who comes to our office.”–Vinay Prasad, MD, MPH Dr. Brawley says that the quality of a screening process also plays a widely underappreciated role in determining whether the screening benefits outweigh the risks. If women are screened at a mammogram van or clinic that does not have access to their prior images, “every mammogram is like a first mammogram with a very high false-positive rate,” he says. “If you just have last year's mammogram, the false-positive rate in a good mammographer goes down dramatically.” Calculating the potential benefits and harms has been further complicated by an inability of researchers to agree on which parameters or models to use for studies, or on which studies to include in meta-analyses. A 2013 review by the Cochrane Breast Cancer Group, for example, suggests that 2000 women would need to be screened every year for 10 years to prevent 1 breast cancer death.4 Conversely, the European Screening Network (EUROSCREEN) working group suggests that 7 to 9 lives could be saved by a biennial screening of approximately 1000 women, a nearly 20- fold difference.5 The controversies, and charges of underinflation or overinflation, are even more dramatic for estimates of breast cancer overdiagnosis. The numbers “are all over the place,” Dr. Smith says, varying from zero to greater than one-half of all cancers diagnosed via screening, depending on the methodology used. Newer screening trials might not help to resolve the controversy either. Although some of them have done a better job of ascertaining cause of death, Dr. Prasad says, none are big enough or designed to rigorously examine overall mortality. There is a good reason for that, Dr. Brawley says. “When I have designed screening studies, I have found that in order to have an overall mortality endpoint, the study has to last far too long to be practical or be far too big to be practical.” For example, designing a prospective mammography trial examining overall mortality instead of cancer-specific mortality of women in their 50s would likely require a doubling of the study length— from 15 to 30 years—and a tripling of the enrollment. If such endpoints were required, Dr. Brawley says, “We wouldn't have any studies because we couldn't afford to do them.” By focusing on groups at higher risk who may be most likely to benefit, Dr. Prasad says, a screening trial's power will grow more quickly. For larger studies targeting the general population, he says, registry-based randomized trials could help to lower the costs. In addition, he says, even a $2 billion mammography trial would correspond to a small fraction of the annual sums spent on screening; if such a trial debunked the putative benefits of a costlier screening regime, that money might be well spent. However, there is an even bigger issue: “If we started that trial today, would the answer be useful to us 30 years from now or will we have some new technology that has usurped mammography?” Dr. Brawley asks. Dr. Smith echoes the sentiment, noting, “As a practical matter, our studies are designed not only to be affordable but also to deliver the intervention to the population as soon as is feasible and practical yet is efficacious.” Nevertheless, he agrees that trial cohorts should be followed long-term to look for late effects that might not appear until 15 or 20 years after the intervention. Modeling studies eventually may offer some resolution, such as those by Ruth Etzioni, PhD, at the Fred Hutchinson Cancer Research Center in Seattle, Washington, that are estimating the extent of overdiagnosis associated with prostate cancer screening and quantifying the role of screening versus treatment in explaining declines in the cancer's mortality. In the meantime, Dr. Brawley says the way forward should include an effort to transparently spell out what is known and unknown and to understand that disease-specific mortality is an imperfect surrogate for potential survival benefits. Physicians, Dr. Smith says, also should receive incentives to talk to patients about the relative merits and potential costs of a particular screen. Despite his sharp criticism, Dr. Prasad emphasizes that he is not saying cancer screening is always doomed to fail. In high-risk populations, he says, it may yet prove its mettle. But patients should be told the truth regarding risks and benefits. “I think people just deserve the right to know,” he says.