When a Mammogram Is Riskier Than Cancer

Overdiagnosis, or the identification of cancers that would not have become evident in the patient’s lifetime, represents one of the most significant harms of screening. Yet until recently overdiagnosis has gone almost entirely unmentioned in public messages about mammography.

Why the silence? Basically, the proponents of screening have been afraid of providing “mixed messages” that might confuse the public. Sadly, this fear has prevented the kind of full and open discussion that is essential to patient/consumer education.

It was in the mid-1980s that mammography went from being a diagnostic test for patients with symptoms to a screening test for all women. For a century, doctors had been telling women to get treatment for breast cancer as soon as they became aware of it, but now they were told to go further by searching for cancer in the shadows. Increasingly sophisticated technology would be used to find breast cancer in its most embryonic stages.

Overdiagnosis happens when a cancer is identified that either does not progress (possibly even regresses) or develops so slowly that the patient dies of something else before ever developing symptoms. These cancers are sometimes referred to as pseudodisease. The idea that some cancers don’t need to be caught is difficult for many people, even some medical professionals, to grasp. The phenomenon has been observed throughout the history of screening, however, for lung, prostate and breast cancer alike.

Stage Zero

In the breast, ductal carcinoma in situ (stage zero breast cancer) is a particularly problematic condition. In DCIS, the cancer cells are confined to the interior of the milk ducts (the cells’ “in situ” or “original” location). Typically there is no lump or other noticeable symptom, and as long as the tumor remains confined to the milk ducts, it cannot spread beyond the breast. DCIS has a 10-year survival rate of 98 percent.

Healthy women often harbor hidden DCIS lesions. Autopsy studies done on women who died of a variety of causes other than breast cancer suggest that up to 39 percent of women 40 to 70 may unknowingly have DCIS.

Mammography, it turns out, is very good at identifying it. The small lesions often contain crystals of calcium salts, which appear clearly on mammography as clusters of white specks known as microcalcifications -- even in dense breast tissue. In fact, whereas mammography may miss as much as 25 percent of invasive breast cancers, it misses only about 14 percent of DCIS.

From 1980 to 1998, when screening mammography came into widespread use, the incidence of DCIS increased sevenfold, from five to 37 cases per 100,000 women. DCIS now accounts for 20 percent to 25 percent of all new breast cancer cases, about 50,000 to 60,000 cases a year. More than half a million American women live with the diagnosis today.

For all that, very little is known about the behavior of DCIS. Seeking it out and treating it has not proportionally lowered the incidence of invasive breast cancer. Early stage incidence remains high, but rates of advanced breast cancer have dipped only slightly. Apparently, although most invasive breast cancers first pass through a DCIS phase, most DCIS cases do not progress to invasive breast cancer. Only about one-third to one-half do.

The trouble is, we have no reliable way to separate harmless DCIS cases from potentially dangerous ones. As a result, all women who get the diagnosis undergo cancer therapy. Although women with harmless DCIS cannot benefit from treatment, they are still subject to its potential risks.

Surgical Overreaction

The majority of patients have a lumpectomy, with or without radiation therapy. Just under one-third have a mastectomy, sometimes a double mastectomy. From 1998 to 2005, the share of DCIS patients who had both breasts removed more than doubled, from 2.1 percent to 5.2 percent. In other words, the practical effect of our DCIS-detecting efforts has been to send a huge new class of women along a pathway that involves disfiguring surgery, often a long course of high-dose radiation and, for some, treatment complications.

What is needed is a change to our knee-jerk response to DCIS, which encourages and even requires radiologists to find as much of it as possible and surgeons to treat all of it aggressively.

Perhaps it would help to change the name. Many influential voices in the scientific community suggest the scary word “carcinoma” should be dropped. One proposed alternative, “IDLE tumor” (for InDolent Lesions of Epithelial Origin) highlights the sluggishness the condition often displays.

In talking to a patient about DCIS, doctors should avoid suggesting that she is fortunate for having discovered her cancer in the earliest possible stage. Instead, they should explain the biological uncertainty of DCIS and the lack of clarity about how best to treat it. We know that patients with DCIS grossly overestimate the risk that their condition will spread or recur. A 2008 study found that 28 percent thought it was “moderately likely” their cancer would metastasize; in fact, the risk is close to zero. This explains why some patients take the “I may not be so lucky next time” approach and have both breasts removed.

We radiologists, too, have to shift our view of DCIS, which is informed by the unproven notion that finding DCIS today prevents invasive cancer tomorrow. We are constantly on the lookout for DCIS that is as small as the proverbial “period at the end of this sentence.” It has been observed that the rate of DCIS detection in the U.S. is higher than in other developed countries with active mammography screening programs -- three times higher, for example, than in the U.K.

Standard radiology teaching requires that a biopsy be recommended for all but the most obviously benign-looking microcalcifications. If the radiologist thinks that there is more than a 2 percent likelihood of DCIS, the usual practice is to recommend a biopsy, regardless of whether the abnormality is the size of a dot or the size of a dime.

Watchful Waiting

A more rational approach would be short-term follow-up mammography, in which the patient found to have microcalcifications is monitored at six-month intervals. If a particular cluster of microcalcifications does not change during a few years of follow-up, doctors can assume it is nothing dangerous. This would require a radical change in thinking among radiologists, and cooperation from oncologists, surgeons and other medical professionals who care for women with breast cancer.

Finally, concerted research is needed to develop ways to distinguish trivial DCIS cases from potentially more serious ones, and to find appropriate ways to treat the condition. Unburdened by the problem of overdiagnosis and overtreatment, mammography could become an even more effective tool.

From 1990 to 2007, the death rate from breast cancer in the U.S. decreased by 31 percent. How much of that decline has been due to screening is hotly debated. Some research suggests treatment has been all the difference, some suggests screening may account for almost half the improvement. The truth probably lies somewhere in the middle.

What we know is that screening helps women older than 40, with greater benefit for those over 50. Yet women of all ages should be made aware of mammography’s risks. Doctors who favor screening should not be afraid to discuss its limitations openly and honestly. Patients expect it and will thank us for it.

(Handel Reynolds is a breast radiologist at Piedmont Hospital in Atlanta. This is the last of three excerpts from his new book, “The Big Squeeze: A Social and Political History of the Controversial Mammogram,” which will be published on Aug. 7 by Cornell University Press. The opinions expressed are his own. Read Part 1 and Part 2.)

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

    To contact the author of this story:
    Handel Reynolds at

    To contact the editor responsible for this story:
    Mary Duenwald at

    Before it's here, it's on the Bloomberg Terminal.