Does Screening Mammography Work?

The American Cancer Society (ACS) urges women 40 and older to have annual mammograms. "Mammography is especially valuable as an early detection tool because it can identify breast abnormalities that may be cancer at an early stage," writes the ACS in Cancer Facts and Figures 2001. "Numerous studies have shown that early detection saves lives and increases treatment options. The declines in breast cancer mortality have been attributed, in large part, to the use of regular screening mammography." Supposedly, these recommendations are based on extensive randomized controlled trials (RCTs) showing that mammography extends lives.

Now, however, Ole Olsen and Peter C. Gøtzsche of the Nordic Cochrane Collaboration, Copenhagen, have taken an independent look at the data supporting the use of screening mammography and found it lacking {Lancet 2001;358:1340-1342 Their full report is available online at http://image.thelancet.com/lancet/extra/fullreport.pdf.

There is simply no reliable evidence, they write, that screening mammography actually reduces deaths among women who receive it. "Mammographic screening is of uncertain benefit, they write, "and leads to greater use of more aggressive treatment. By detecting cancers early, mammographic screening is widely believed to lead to reduced mortality from breast cancer and to less aggressive treatment." But their review "failed to find a decrease in overall mortality, and the best trials also failed to find a reduction in breast cancer mortality. In fact, screening leads to more aggressive treatment," not to "increased treatment options," as the ACS contends.

If you look in the cancer textbooks, not to mention innumerable websites, you will find vigorous recommendations for screening mammography. So how did these two authors reach such conclusions so at variance with accepted opinion? They carefully scrutinized the quality of each of the RCTs that allegedly proves the value of mammography. Were the patients in the two arms comparable at the start? Were they properly randomized? Was there an unbiased assessment of the outcome of the trials?

On the basis of such detailed assessments, they classify clinical trials into four groups: high quality, medium, poor and flawed. Of the seven existing randomized trials of screening mammography, none in their opinion was of high quality, two were of medium quality (those from Malmö and Canada), three were of poor quality (Two-County, Stockholm, and Göteborg) and two were flawed (New York and Edinburgh).

One of the problems they uncovered was that if a patient died in the control group her death was more likely to be ascribed to breast cancer than similar patients in the treated group. This made the outcome in untreated patients appear worse than it necessarily was.

One danger of mammography, they say, is that screened women are more likely to receive radiation therapy than unscreened women. Radiation, they write, "is expected to increase overall mortality because of cardiovascular adverse effects. These deaths were not counted as deaths related to screening in the trials we assessed." In other words, if a woman gets screened, she is more likely to be diagnosed with breast cancer. If she is diagnosed with breast cancer, she is likely to get adjuvant radiation therapy. But because of that she is at greater risk of damage to her cardiovascular system {Lancet 2000;355:1757-70}.

The main outcome measured in screening trials is the chance that a woman will die of breast cancer. But this ignores the likelihood of other causes of death. The authors consider more meaningful the data on overall mortality. In the two trials with the best methodology, there was no difference in the relative overall risk of death between those who received mammograms and those who didn't. In the Swedish trial, there was originally an imbalance in the ages of participants. When these were adjusted the benefit of mammography evaporated. Olsen and Gøtzsche conclude: "The reliable evidence does not indicate any survival benefit of mass screening for breast cancer."

How, then, can the utility of screening mammography be justified? While the best trials failed to find an effect, those trials with poor-quality data "found a marked difference." Thus, advocates have relied on faulty data to reach erroneous conclusions. But results from different quality groups "should not be combined," the authors caution.

They state that screening leads to more aggressive treatment, increasing the number of breast operations by up to 30 percent. The reason for this increase is that "screening identifies some slow-growing tumors that would never have developed into cancer in the women's remaining lifetimes, as well as cell changes that are histologically cancer but biologically benign." Furthermore, more accurate mammography could lead to "additional overtreatment," because of the detection of even more early and questionable lesions. "The problem cannot be avoided," they add.

Gøtsche and Olsen have thus re-opened the question of mammography's utility and safety. As you might expect, an extraordinary battle is shaping up around their findings. It has led to a serious division within the Cochrane Collaboration, an international organization devoted to rigorous and systematic reviews of clinical trials. This collaboration has "broken down badly," wrote Richard Horton, MD, editor of the Lancet. Certain Cochrane editors added statements to Gøtzsche and Olsen's analysis "which lent support to arguments in favor of screening and excluded data about the effects of screening on subsequent treatment" This outrageous behavior "erodes the academic freedom of these investigators," said Horton.

Where does this leave screening mammography? Some people will argue that today's mammography techniques are so far superior to the older techniques that historical data is simply not applicable. This sort of argument is often produced when clinical trials do not confirm one's theories! It simply makes the refutation of any evolving treatment impossible. Besides, as Richard Horton notes, "women should not be expected to forgive old-and disproven-screening practices" {Lancet 2001;358:1284-5}.

Does this mean that no woman should have a mammogram? Not at all. Each case must be considered on its own merits. But it does point to the weaknesses in the arguments currently being mustered on behalf of mammography screening of huge populations. New, less invasive (and, we might add, less painful) techniques are desperately needed. And further clinical trials need to be done to see if, using the most modern equipment and techniques, all-cause mortality can be reduced by screening mammography.