How Effective Are Routine Mammograms at Delaying Death?

BY Jason Kane  December 25, 2012 at 2:45 PM EST


Photo of doctors looking at mammogram results by BSIP/UIG via Getty Images.

In the last 30 years, mammograms have led 1.3 million women to seek treatment for cancer that would never have harmed them, according to a recent report in the New England Journal of Medicine. “At best,” the report says, mammograms have had “a small effect on the rate of death from breast cancer.”

It’s an assertion the American College of Radiology and other prominent groups call “flawed and misleading”, adding that it could confuse women who should be receiving the test and could cost lives. On Tuesday evening’s PBS NewsHour broadcast, health correspondent Betty Ann Bowser explores this latest firestorm in the oncology community and what ordinary women should take away from all of the conflicting information.

In the meantime, we’ve asked three of the nation’s most influential doctors whether women over the age of 40 should be screened yearly. Dr. Sandra Swain of the American Society of Clinical Oncology, says yes. Dr. Archie Bleyer, co-author of the New England Journal of Medicine report, says no. And below, Dr. Barnett Kramer, director of the National Cancer Institute’s Division of Cancer Prevention, explains the scientific basis for both arguments.



Dr. Barnett Kramer is the director of the National Cancer Institute’s Division of Cancer Prevention. He previously served as director of the NIH Office of Disease Prevention from 2001 to 2010. He heads NCI’s Physician Data Query Screening and Prevention Editorial Board since the early 1990s. He is also an investigator with the National Lung Screening Trial and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial.

Dr. Barnett Kramer: One of the most important goals in cancer research is to find a way to prevent cancer before it ever occurs. As director of the National Cancer Institute’s Division of Cancer Prevention, I’d be thrilled if we had effective means of preventing the over 100 different types of cancer, each of which has unique characteristics and behaviors.

While we’ve made strides in our understanding of how to prevent a number of cancers, a lot of research and hard work remains to fully achieve our goal. We also have to pursue additional strategies to reduce cancer mortality. One of those strategies is screening to detect cancer at sufficiently early stages, and to treat or remove screen-detected cancers before they metastasize and spread to other organs.

For a few cancers, screening methods have proven to be extremely effective. For cervical cancer, early detection with Pap smears or HPV testing clearly reduces mortality from the disease. For colorectal cancer, screening with endoscopy and removal of pre-cancerous polyps also reduces mortality from the disease.

For breast cancer, screening, most often with a mammogram, can be effective but the magnitude of mortality benefit is not as large or as clear cut as it is for cervical and colorectal cancer. And the balance of benefits and harms for women at various ages is sometimes debated. For example, there are specific downsides of using mammography (particularly at young ages), such as the detection of non-life-threatening tumors, which can in turn trigger unnecessary treatment that may be harmful. The detection of non-life- threatening tumors is referred to as “overdiagnosis.”

Overdiagnosis occurs in general when there is a substantial reservoir of asymptomatic, indolent disease (disease that never progresses to a lethal form) detectable by a screening test that dips into that reservoir. Trends in breast cancer rates since the 1980s (when mammography became commonly practiced) suggests that overdiagnosis of breast cancer has been occurring at a national level.

There has been a substantial increase in early stage disease, but only a small decrease in late stage disease, and no apparent decrease in metastatic breast cancer at the time of diagnosis. This is a pattern that would be expected with overdiagnosis. Other examples of overdiagnosis include prostate cancer (where the PSA test led to a huge jump in new cases when it was introduced) and a tumor known as neuroblastoma that occurs in infants.

Because overdiagnosis involves the detection of non-life-threatening cancers, it leads to higher cure rates, since the treatment of such tumors ‘cures’ cancers that did not need treatment in the first place. It also leads to an artifactual increase in survival rates, thus exaggerating the apparent benefits of screening.

Another phenomenon that inflates the apparent benefit of mammography is known as lead-time bias. Since survival is measured from the date of cancer diagnosis, and mammography advances the date of diagnosis, survival increases even if the date and cause of death are not changed. For example, if a type of cancer always caused death on the fourth anniversary of the diagnosis, the five-year survival rate would be zero percent. However, if a screening test advanced the date of diagnosis by 3 years, but did not change the date of death by a single day, the five-year survival would be 100 percent.

It is important to emphasize that the efficacy of breast cancer screening has been established in multiple randomized controlled trials, the gold standard of study designs. The debate about the benefits of mammography generally centers not on whether there is a breast cancer mortality benefit, but the balance of risks and benefits by age and the frequency of screening intervals to maximize the benefits and minimize the harms. It is in this area that NCI research is key, especially in conducting and supporting clinical trials, where we can provide the research evidence upon which others can formulate recommendations.

At NCI, we do not issue screening guidelines, but rather have cancer screening summaries in professional and patient versions of evidence in what is commonly referred to as the PDQ. The PDQ, which stands for Physicians Data Query, can be accessed here.

Today, an important area of research funded by the NCI uses the new and emerging tools of molecular biology and genomics to try to identify molecular fingerprints of indolent versus rapidly progressive tumors identified by screening. If successful, such patterns could be used to maintain the benefits of screening while decreasing the harms associated with overdiagnosis.

Related Content