The Screening Dilemma
In the mid–19th century, a german pathologist named Rudolf Virchow discovered that leukemia was caused by the rapid multiplication of abnormal white blood cells. Just like that, with some autopsy samples and a light microscope, Virchow defined cancer — a process in which healthy cells mutate and then reproduce. Before this revelation and for many decades after it, cancerous tumors were found and cut out only when they became visible or palpable. But Virchow's notion that cancer cells start out normal and then go rogue laid a foundation for modern medicine's approach to the disease: early detection.
These days, we no longer have to wait for tumors to make themselves evident. We don't even have to wait for symptoms. Now doctors look for abnormal cells in healthy people, hoping to catch and remove them before they cause sickness, a strategy that has had remarkable results. Along with treatment advances, mammography has reduced the U.S. breast-cancer mortality rate by some 30% since 1989. Pap smears have helped lower the cervical-cancer mortality rate by 60% since 1975. The rate of death from colorectal cancer is also steadily dropping, thanks largely to screening.
It seems that we should be better off finding all cancer early. But this logic may be flawed. Virchow never imagined that modern medicine would have the tools to find tiny cancers at such early stages. The field now includes highly sophisticated blood tests, ultrasound, computed tomography (CT), X-ray, magnetic resonance imaging (MRI) and fine-needle biopsy. Paradoxically, we've become so adept at finding abnormal cells early that there are more cancer patients than ever before. About 4% of the U.S. population are "cancer survivors." "If we had a 100% sensitive test that could pick up everything a pathologist would call cancer, it's conceivable that most of us, if not all of us, would be found to have cancer," says Dr. Barnett Kramer, a medical oncologist and former associate director for disease prevention for the National Institutes of Health (NIH).
"One of the problems is what our definition of cancer is," says Dr. Otis Brawley, chief medical officer of the American Cancer Society (ACS). "Through all the iterations in science, all the advancements in imaging, in understanding diagnostics, in understanding how to do biopsies, we still use Virchow's definition of cancer."
Happily, we don't also use his definition of how the disease progresses. Virchow believed all cancers would eventually spread and lead to death. Yet as scientists have learned how to detect and treat cancer earlier, they have also learned that some cancers never cause any sickness at all. In rare cases, certain cancers can even disappear without treatment. "Early diagnosis has changed the face of what it means to have cancer," says Dr. H. Gilbert Welch, a clinical epidemiologist, cancer-screening researcher and internist at the Veterans Administration Medical Center in White River Junction, Vt. The problem, he says, is that "there are really bad cancers and there are really innocuous ones that never go anywhere, and we're not good at sorting them out."
So we continue to look for more cancers early and treat nearly everything we find as though it would be fatal not to. Patients don't complain. Why would they? Even though the U.S. has so many effective treatment options available — the best in the world, in fact — cancer kills some 600,000 Americans every year. Countermeasures like screening that can be administered in a controlled manner seem like antidotes not only to cancer but also to the disease's inherent unpredictability.
But looking for signs of illness in seemingly healthy people is complicated. Cancer screening is truly effective only if the growths found would eventually cause sickness and if finding those growths earlier increases the efficacy of treatment. Absent these two conditions, finding cancer via screening is what's known as "overdiagnosis," which is guaranteed to happen when screening is performed population-wide. Overdiagnosis causes harm ranging from unnecessary worry to death in rare instances. Says Welch, a professor at Dartmouth Medical School and the lead author of a new book titled Overdiagnosed: "Theoretically, we could spend every day looking for early signs of disease. And we're getting closer and closer to that."
Do No Harm, Unless ...
even in cases in which cancer screening has clearly saved lives, its precise use is controversial. With breast cancer, a debate rages over when women with no risk factors should begin mammography. In 2009 the U.S. Preventive Services Task Force (USPSTF), an independent government body, advised that women get routine mammograms every other year beginning at age 50. Previously, the group had said that mammograms should be annual and begin at 40, which the ACS and other advocacy groups still say is best. The USPSTF changed its advice after determining that the collateral damage of annual screening beginning at 40 wasn't worth the payoff.
Here are the odds. To save the life of one woman in her 40s, 1,904 would have to undergo annual screening. Beyond inconvenience and overexposure to radiation, this excess mammography would lead to false positives; psychological stress, including depression; and unnecessary surgery. In addition, much of the abnormal cell growth detected in women in their 40s could have been detected in their 50s with no adverse effects from the delay.
Not surprisingly, many women railed against the new USPSTF guidelines. Understanding a statistic is one thing. Accepting even the slim chance that you could die of a treatable disease to spare others unnecessary harm is less clear-cut. No woman cares about collateral damage when her life could be at stake.
Colonoscopy, the examination of the colon from within to look for cancer and precancerous polyps, is an accepted standard of care. Doctors often advise average-risk patients to get a colonoscopy at age 50 and, if nothing suspicious is found, every 10 years after that. (See Dr. Oz, page 50.) Yet this advice is given despite the fact that no long-term randomized trial — the industry standard for amassing bulletproof data — has shown colonoscopy, a highly invasive procedure, to be more effective at saving lives than a simple test looking for blood in fecal material or a sigmoidoscopy, a procedure in which just a portion of the colon is examined. One advantage of colonoscopy is that doctors can screen for cancer and intervene if they suspect it, removing suspicious lesions while the patient is still on the examination table. (The USPSTF says colonoscopy, sigmoidoscopy and the fecal-blood test can all be effective.)
Other kinds of screening, like for prostate cancer, are even more contentious. Multiple scientific trials have proved that the ubiquitous prostate-specific antigen (PSA) test saves very few lives, if any. PSA tests are notoriously unreliable, detecting potential cancers where there are none and returning normal results in some men who have malignancies. The doctor who discovered the existence of PSA, a protein that, when elevated, is sometimes an indication of cancer, has disavowed the test, calling it a "profit-driven public-health disaster."
Of men who undergo routine PSA testing, a staggering 17% are eventually diagnosed with cancer, and most of these are treated with radiation or surgery. At least half experience complications such as erectile dysfunction or incontinence. Much of this is needless suffering, since the vast majority of men diagnosed with prostate cancer will not die of it, even if it is left untreated. In fact, for every 1,000 men ages 55 to 70 who undergo annual PSA tests for 10 years, only one life might be saved, according to Welch, who has conducted extensive study on the topic. Meanwhile, an estimated 150 to 200 will have an unneeded biopsy, and 30 to 100 will undergo radiation treatment or have their prostate removed unnecessarily. Despite its ineffectiveness, some 30 million American men have a PSA test every year, partly because it's the best we have for now and prostate cancer remains a deadly disease, killing about 32,000 men annually in the U.S.
Impotence and urinary dysfunction are awful, but the history of poorly designed cancer screening includes tests with much more dire consequences. In the 1960s, many doctors advised American smokers to get chest X-rays to check for lung cancer, the No. 1 cancer killer in the U.S. In the 1970s, the ACS followed suit, and millions heeded the advice. Yet a decade later, studies showed that Americans who were screened for lung cancer via chest X-ray actually had a slightly higher mortality rate than those who were not screened. The excess deaths were due in part to the risky surgery that patients underwent when something was found.
As recently as the 1980s and '90s, doctors in Japan and Quebec screened babies for neuroblastoma, a cancer of the nerve tissue that's the most common type of cancer in infants. The screening revealed alarmingly high cancer rates. "People thought there was a neuroblastoma epidemic," says the ACS's Brawley. "Kids started getting biopsied and then getting surgery, which means basically being filleted open for an 18-month-old." The result was the same as with lung-cancer screening. Mortality rates were slightly higher among the screened population because of deaths caused by surgery. The screening was discontinued. Doctors, it turned out, couldn't distinguish between fatal neuroblastoma and the far more common kind that simply vanishes over time.
Cascades and Incidentalomas
Dr. Jim Mold did not want a PSA test. A family doctor trained in geriatrics, Mold had published journal articles about the hazards of screening and of medical intervention for men with prostate cancer. So it made sense that even though he was 58, solidly in the risk group for prostate cancer and in his doctor's office for a series of unrelated blood tests, he had chosen to forgo a PSA test. He got one anyway. It had inadvertently been added to his order. He noticed at the last minute but didn't protest.
The results showed Mold's PSA was elevated, which typically leads to a biopsy. But Mold was not typical. He understood the test's limitations. So he reread the medical literature on the subject and consulted his colleagues. He avoided sex and treated himself with antibiotics, both of which would have brought his PSA levels down in the absence of cancer. He had another PSA test. Elevated. So he opted for a biopsy. Cancer. Prostate cancer is often slow-growing, so some patients choose watchful waiting over surgery or radiation.
"I was trying to imagine myself living — hopefully 40 more years — with a cancer growing inside of me," remembers Mold. He knew that surgery or radiation could end his sex life and impair his ability to urinate. He knew studies showed that there was a good chance his cancer would never hurt or kill him. Still, he says, "I couldn't resist." He had his entire prostate removed. "I've done really well, and I'm really grateful that I had it done, but I don't know if it was needed," says Mold, now 62. "They say, 'Well, we can stop at any point.' No, you really can't."
Ironically, when he was a young doctor, Mold was one of the first clinicians to write about "the cascade effect," in which patients enter the health care system for one problem or even routine testing and end up getting shuttled through myriad related or unrelated interventions. This can happen in virtually any area of medicine, but cancer screening is particularly risky territory because healthy people are often caught in the net. "Once you've committed to testing, it makes sense to commit to everything else," Welch says. We all imagine we could be the 1 in 1,000 or 1 in 10,000 whose life could be on the line.
"The most important decision is whether or not to be screened," says Dr. Matt Handley, a family doctor and associate medical director for quality and informatics at Group Health, a Seattle-based health-and-insurance system with a zealous adherence to care based on hard evidence.
The chances of getting caught in a diagnostic cascade are increasing thanks to advanced imaging technology. For colorectal cancer, in addition to screening tests that analyze feces and survey the colon internally, there is now virtual colonoscopy. This screening method uses radiation, via CT, to view the organ from outside the body and reconstruct it digitally. CT colonography, as it is called, has advantages. No sedatives are required, and a patient can skip the part in which a doctor threads a scope and light through the rectum. The procedure is also cheaper than a colonoscopy. Whatever savings colonography might appear to present, however, are far outweighed by the cost of following up on all the abnormalities that can show up in the resulting images. These discoveries are known in the medical field as incidentalomas and are typically harmless. As many as 16% of patients undergoing their first virtual colonoscopy are found to have them. Cue the cascade.
Incidentalomas take up so much time, energy and money that some doctors are questioning whether imaging technology has advanced too far. Some doctors are even ignoring the images of the rest of the abdomen created by a CT colonography. They don't want to look, for fear of what they might find. At the same time, imaging-equipment manufacturers are creating higher-resolution scans, increasing even further the chances that something innocuous could be found.
Dr. G. Scott Gazelle, a radiologist at Massachusetts General Hospital who has a Ph.D. in health policy, points out that higher-resolution CT scans require more radiation. "We're starting to question how good the images need to be," he says. Efforts are under way at many hospitals to reduce radiation dosage, but this follows more than a decade of sharp increases. Radiation exposure is part of the reason the USPSTF says it cannot determine whether CT colonography causes more harm or benefit; it assigned the procedure an I rating for "insufficient evidence."
Incidentalomas have the potential to cause even more upheaval for people who undergo CT to screen for lung cancer. Gazelle says there's no doubt that these scans, while exposing patients to many times the radiation of a chest X-ray, can reduce lung-cancer mortality in smokers and former smokers. (A randomized trial to assess lung-cancer-screening CT is under way.) "The question is, At what cost? And by how much?" he says. In addition to the staggering expense of tracking down all the incidentalomas that are likely to be found — lungs are notoriously full of such strange-looking nodules — the physical risks are high. Three to five percent of people who undergo surgery to cut out pieces of their lungs die from the procedure. "Multiply that by the smokers and former smokers in the U.S. population and you could get tens of thousands of deaths," says Gazelle.
The "Fastest-Increasing" Cancers
In 2010, about 45,000 americans were diagnosed with thyroid cancer. That's about three times the diagnosis rate in 1975. But the mortality rate for the disease was the same. There was no more thyroid cancer than before; doctors were just looking and finding more of it. This means that when it comes to lives saved, thyroid-cancer screening may be doing little or no good. In fact, it's probably mostly causing harm.
A recently launched public-awareness campaign called Check Your Neck identifies the disease as "the fastest-increasing cancer in the U.S." Doctors screen for thyroid cancer by palpating the neck, but most necks are lumpy, and it can be hard to tell by touch whether a thyroid is enlarged. Once there's uncertainty, the cascade can take over. Remarkably, many of us, possibly even most of us, will develop thyroid cancer at some point in our lives, but very few of us will die of it. In a 1985 study, researchers examined the bodies of 101 people who had died of causes other than thyroid cancer and found that a third of them contained cancerous thyroid cells. Because of the sampling method, the researchers knew that they were certainly missing some cases, meaning the percentage was even higher, and yet none of those people were killed by the disease.
Still, the vast majority of people in whom thyroid cancer is diagnosed undergo radiation treatment or have their thyroid removed. The surgery leaves some patients hoarse and all forever dependent on medication. Something similar is happening with melanoma, a skin malignancy that kills about 9,000 Americans every year. Awareness campaigns like Melanoma Monday, sponsored by the American Academy of Dermatology, helped raise the melanoma-diagnosis rate 30% from 1975 to 2007. The mortality rate? Unchanged.
The Business of Screening
Among all the reasons overscreening is taking place, the least discussed — and most disturbing — is money. Back in the 1990s, when Brawley, now of the ACS, was an assistant director of the National Cancer Institute, he visited a large research hospital in Atlanta. There, a marketing expert explained that providing free PSA tests to 1,000 men at a local mall could lead to millions of dollars in subsequent revenue for the hospital. The income would come from biopsies, surgeries, radiation and even urinary-sphincter implants in men who experienced complications. This kind of strategy is common, according to Brawley.
Some health centers and urology practices use giveaways to entice men to get PSA tests. In recent years, men have scooped up tickets to Atlanta Hawks, Buffalo Sabres and Tampa Bay Rays games in exchange for getting tested. A nonprofit national organization called Zero, for "zero prostate cancer," tries to get the word out about the benefits of PSA testing, parking a mobile testing unit outside sporting events and churches. The organization doesn't charge patients for tests but accepts donations from urologists, Big Pharma and Beckman Coulter, a PSA-test manufacturer.
The downstream costs of cancer-screening campaigns like this are enormous. Says Welch: "It may lower costs for an individual patient" if minor surgery to remove a suspicious early growth makes major, long-term cancer treatment unnecessary, "but because there are so many more patients created, that effect is overwhelmed." Doctors sometimes encourage screening in part because they believe it could protect them from liability. In addition, the new Affordable Care Act requires insurers to cover "preventive services" at full cost, meaning most patients will pay nothing out of pocket for procedures like mammograms, PSA tests and colonoscopies. This could drive up screening rates even further.
Welch and Handley are urging change upstream. According to a meta-analysis published in 2009, patients are 20% less likely to undergo PSA testing once they understand the potential harms, benefits and uncertainties of it. The VA offers male patients over 50 a DVD and a booklet titled Is a PSA Test Right for You? The material contains the statement "If you find out you have prostate cancer later in life, you will most likely die with the cancer, but probably not because of it."
Advanced screening methods are putting more of us in a similar situation. Many more of us are finding out we have cancer. But even if we can survive cancer, can we live with it? Says Kramer, formerly of the NIH: "The term cancer is so fearsome, many people can't accept the concept that you don't do anything about it." Combine that fear with the American medical system's seemingly limitless capacity for testing and intervention, and excess is inevitable.
"In the U.S. in particular, we just feel like more is always better," says Diana Miglioretti, a biostatistician and investigator for Group Health who studies cancer screening. "There is an uncomfortableness with ambiguity, so we're always looking for that perfect test to save a life."