Health June 1996

Good News and Bad News About Breast Cancer

Women are more alarmed than they need to be about the chances that they will develop breast cancer. But they are also more confident than they should be that the advances medicine has made in treating the disease and prolonging life mean that it can be cured

In 1935, 26.2 out of every 100,000 women died of breast cancer. That was a long time ago, of course, when life expectancy for women was sixty-four years. Americans now live longer, which means that diseases of the elderly are more common. Although breast cancer affects many young women, it is still principally a disease of middle and old age—the median age at diagnosis is sixty-four. Statisticians must adjust raw data about incidence and mortality to compensate for underlying demographic shifts. In 1992, the latest year for which figures are available, the adjusted rate of mortality was 26.2 women per 100,000—the same as in 1935. (The death rate rose a bit from 1986 to 1989 and declined by about the same amount from 1989 to 1992.) Since 1935 medicine has seen improvements in surgical technique and anesthesia, and the introduction of mammography, radiation therapy, and chemotherapy, along with an enormous jump in general medical knowledge. But all this progress has had no effect on the chances that an individual woman will die of breast cancer. To my way of thinking, the constancy of the death rate in the face of rising incidence and aggressive treatment is a strong hint that we need to approach the disease in another way.

When I discussed the evidence relating to mammography, radiation, and chemotherapy, I was in the realm of fact—although some colleagues may disagree with my interpretation. In what follows I am moving into unsteadier terrain. Of course, I believe that my view of breast cancer is correct, and much of it is shared by other specialists. But I will be satisfied if I can persuade readers that the mere existence of a coherent alternative explanation raises questions about the mainstream view.

As I have said, almost all cancer researchers think that the disease is triggered by an accidental change in the DNA of at least one cell. That cell divides, producing two cells, then four, then eight, and so on, with the volume of the tumor doubling in each successive generation. By the time the tumor has doubled twenty-three times, the original cancer cell has multiplied to more than eight million. At that point the tumor is about an eighth of an inch in diameter, just big enough to be detected (sometimes) by a mammogram.

Such a tumor is not very dangerous by itself; the danger lies in the metastases. The question is how soon the tumor metastasizes.

For almost thirty years John S. Spratt, a cancer surgeon now at the University of Louisville Department of Surgery, has been measuring the growth rates of breast tumors. In one of his most recent studies, performed in collaboration with researchers from Heidelberg, Germany, Spratt examined the progressive mammograms of 448 women who had tumors that turned out to have been visible in mammograms made before the tumors were diagnosed. (The women's doctors were not necessarily at fault for missing the tumors; in many cases mammographic imagery is ambiguous.) Comparing first, second, and even third mammograms provided evidence of how fast the tumors grew in the intervals. The median doubling time was 260 days, but the range was considerable: the fastest tumor doubled in ten days, the slowest in 7,051 days—that is, almost twenty years. These figures have striking implications.

Consider a woman who is unlucky enough to develop a single cancerous cell on her forty-third birthday. If the woman is especially unlucky, the cell has a fast doubling time of, say, thirty days. Twenty-three cell generations later the tumor might be visible on a mammogram; in another six or seven doublings it will be just big enough to feel. By then the woman will be forty-five. She will probably die before her fiftieth birthday. If, though, the woman develops a cancer with a slow doubling time of, say, 360 days, twenty-three doublings will take about twenty-three years, at which point the tumor might be seen with mammography. The tumor will be palpable in another six or seven years, meaning that without mammography it probably would not be detected until the woman was in her mid-seventies. By that age some people have already died of other causes.

I have simplified these calculations considerably. Spratt and his German colleagues found that breast cancers do not grow at a constant rate but instead slow down as time passes. Yet the principle holds that tumors that begin with fast doubling rates grow faster than tumors that begin with slow doubling rates at comparable stages of formation.

Close scrutiny of tumor doubling times could explain why the earlier diagnoses provided by mammography seem to provide so little prolonging of individual lives, despite the statistical appearance of benefit caused by earlier diagnosis. Although mammography is able to spot tumors as small as an eighth of an inch, which contain eight million cells, the average size at diagnosis with mammography is about 600 million cells. Such a tumor is only a bit more than a quarter of an inch across, but it has already doubled almost twenty-seven times and may have been in the body for decades. The average size of tumors detected by palpation is about 45 billion cells and about an inch and a quarter in diameter; these tumors have doubled an additional eight or nine times. To argue that earlier diagnosis provides an important benefit, one must believe that the tumor is considerably likelier to spread in those eight or nine later doublings than it was in the preceding twenty-seven.

There is no evidence that this is the case; indeed, the small amount of available data suggests that this view is wrong. With mammography we can see breast tumors earlier than we could before. But it is illogical to assume that our newfound ability to observe breast tumors between the twenty-seventh and thirty-fifth doublings means that they are especially likely to spread during this time or afterward and not before. If tumors are more likely to metastasize after rather than before mammography can detect them, the burden is on mammography advocates to demonstrate it. Meanwhile, I believe that the reasonable course is to assume that breast cancer can spread at any time in its development, and that metastasis has probably already occurred by the time we are able to detect the primary tumor. If this view is correct—and I should stress that studies to prove it have not yet been conducted—then it would explain why research has had such difficulty proving that mammography can save women's lives.

Similarly, examination of tumor doubling times could explain why chemotherapy boosts five- and ten-year survival rates but has little impact on the annual percentage of women who die of breast cancer—that is, why it helps women with the disease to live longer but leaves them just as likely to die of metastatic breast cancer in the end. My best guess is that adjuvant chemotherapy wipes out 95 to 99.9 percent of the residual cancer cells in a patient's body. (It doesn't get them all because the remaining tumor cells are innately resistant to chemotherapy.) Expressed as a percentage, the figures are impressive, but the actual impact is surprisingly slight. Suppose that a woman's tumor has metastasized and that the new tumor has grown to a million cells—a lot of cells, but not enough to be seen by the naked eye, or palpated, or spotted by any current imaging method (CAT scan, ultrasound, magnetic resonance imaging, and so on). If chemotherapy kills 99 percent of the cancer cells in a woman's body, this prophylactic treatment will reduce the metastasis from a million cells to 10,000. The remaining cells, which are resistant to chemotherapy, will keep on proliferating, more than likely at the same rate. In six and two thirds cell generations the tumor will have grown back to a million cells and the woman will be right back where she was before treatment began.

Cruelly, chemotherapy helps least those who need it most. If a woman's cancer has the short doubling time of thirty days, the six and two thirds doublings the tumor needs to recover from chemotherapy translate into about 200 days. Because chemotherapy is often administered monthly for six months, the gain is roughly equivalent to the length of treatment. Producing so much suffering, chemotherapy would in this case be a dubious exercise. If the woman's cancer has a doubling time of 360 days, however, she would gain 2,400 days, which is six and a half years. That sounds like a good payoff, but does she need it?If cancer were diagnosed in that woman at sixty-four (the median age of diagnosis), her slow-growing metastases would probably not become life-threatening for twenty to twenty-five years, when she would be in her late eighties. Because most people do not live that long, there would be little point in subjecting her to a round of chemotherapeutic treatment that would give her another seven years when she probably would die of something else in the meantime. If chemotherapy has little impact on a woman's chance of surviving either aggressive or indolent tumors, is it any wonder that it makes few inroads on mortality?

At the same time, chemotherapy should not be dismissed. The calculations above, for doubling times of thirty and 360 days, represent extremes; I used them to illustrate my point. A more representative example would apply Spratt's median doubling time of 260 days to my hypothetical forty-three-year-old woman. If, as before, the first cancer cell develops on her birthday, the resultant tumor could take eighteen to twenty years to show up on a mammogram. (Such calculations are necessarily inexact, because individual tumors do not always grow at the same rate.) The woman would then be in her early sixties—near the median age of diagnosis. With no treatment other than lumpectomy, she would be likely to die before the age of seventy-five. But if chemotherapy gave the woman the time it would take the tumor to double another six or seven times, the onset of life-threatening metastatic disease would be postponed until the woman was at least eighty; antihormonal therapy might buy an equivalent number of years. In real terms the achievement would probably be smaller, because people in their eighties are likely to die of heart disease or some other condition. Nonetheless, the woman would have gained five to ten years of life. This is a precious gift, one that she and her doctor can justly celebrate.

But consider the breast-cancer patients doctors most dread seeing—women in their thirties or forties. Such cases are relatively uncommon; breast cancer owes its status as the leading killer of women in this age group mostly to the even lower likelihood that they will be killed by anything else. Nonetheless, the individual tragedy of a disease that strikes down young, vibrant people makes it disproportionately urgent to treat them. Sadly, younger women in whom cancer is diagnosed are more likely than older women to have fast-growing tumors, because slow-growing tumors are usually still too small to detect. Given the probable doubling rates, these women will be lucky if we can give them an extra five years. Five years is, of course, much better than nothing—but much less than the thirty or forty years these women will lose. Anyone who treats younger breast-cancer patients knows that we will not have made major progress in the treatment of this disease until we can give women like these, with fast-growing cancers, thirty doubling times rather than six or seven.


In effect, mammography today provides our definition of breast cancer. Any tumor spotted on a mammogram is treated, almost reflexively, with surgery, radiation, and often chemotherapy and antihormonal drugs. Thinking about doubling times suggests the inadequacy of this approach. When doctors diagnosed breast cancer by palpation in annual exams, they found principally fast-growing tumors—ones whose average doubling time, according to the work of Spratt and other researchers, was about ninety days. Despite decades of work, medicine still is unable to treat this kind of cancer effectively. Today the spectrum of breast cancer is different. Perhaps because of the hormonal changes created by the changes in women's lives, physicians are increasingly likely to observe the "new" cancer described above, which is slower-growing and much less dangerous. These cancers, because they progress so much more slowly, have a radically different impact on women's lives. For that reason we should be more discriminating in how we treat them.

"New" may be a misnomer for this slow-growing breast cancer. Although its incidence has risen, I suspect that it has been with us for at least fifty years; we just weren't able to see it. In fact, I would not be surprised if someday we are all found to harbor somewhere in our bodies several small, slow-growing tumors that will never cause us any problems. (They are beaten to the punch by cardiovascular disease or faster-growing cancers.)

Among the most important varieties of the new breast cancer is the in situ tumor—the small, localized, almost nongrowing tumor that at the time of diagnosis has seemingly neither become invasive nor developed the capacity to metastasize. Prior to mammography, as noted earlier, in situ tumors accounted for only one to two percent of all breast-cancer diagnoses, whereas today in communities where people see doctors often and have lots of tests, in situ tumors account for at least 10 percent of all breast-cancer diagnoses. After lumpectomy and radiation, only one out of ten in situ malignancies recurs in the next five to eight years.

Most of my colleagues celebrate this as a triumph, because it appears that we are catching cancers earlier than ever and curing more of them. They may be right. But consider this—if one out of ten in situ cancers recurs after treatment, nine out of ten do not. If my view is correct, even without treatment many or most in situ cancers would never have grown big enough to be detected by palpation, let alone to pose a threat to life. They might even have become invasive and metastasized, but the metastases would also be too small to be detectable and would never be lethal—rendering the recurrence rate and thus the question of treatment ultimately unimportant to survival. As a result, mammography is only leading physicians to diagnose an ever-larger number of harmless tumors. Patients who otherwise would never have known they have cancer may needlessly suffer through the unique pain, anxiety, disfigurement, and expense associated with modern medicine and cancer. For all we know, the chief effect of mammography has been to disguise our inability to cure the old cancer, by burying it in cases of new cancer.


Because of the prevalence of in situ and other slow-growing breast cancers, women who receive a positive mammogram should not despair. Two thirds to four fifths of positive mammograms lead to biopsies that do not reveal cancer. Even if the biopsy indicates cancer, the patient should keep in mind that not all tumors are truly dangerous, and she should strive to learn what kind of tumor she has.

Although scientists are divided in their opinion of its accuracy, I believe that one of the most promising gauges of risk is the "S-phase fraction," which is a rough measure of a tumor's doubling time, derived from a technique known as flow cytometry. Technicians calculate this measure by chopping up a small amount of tumor tissue in a kind of specialized blender, staining the cell nuclei with a dye, and squirting the result one cell at a time through a very fine nozzle. The cells shoot through a thin laser beam, each nucleus casting a shadow as it crosses the light. Computers record the shadows with enough detail to discern the approximate percentage of cells in the sample that are dividing. From these data physicians can infer whether the cancer is aggressive (a doubling time kf sixty days or less), moderate (sixty-one to 150 days), indolent (151 to 300 days), or very indolent (more than 300 days). Because tumor growth rates may change over time, and metastases do not necessarily march in lockstep with the tumors that spawned them, the actual situation faced by patients is more complex than indicated by this summary. Nevertheless, Ibelieve that the broad principle holds:armed with information about a tumor's growth rate and the patient's age at diagnosis, doctors could often be more informative than they are now about what lies ahead for their patients.

If doctors could accurately gauge tumor growth rates, using any agreed-upon test, my strong hunch is that about a third of the tumors now detected would be found not to need treatment beyond removal of the tumor itself. Perhaps another quarter of women in whom breast cancer is diagnosed could gain considerable time—enough to take them safely into old age—with antihormonal therapy alone. The remainder could be helped by the combination of surgery, radiation, and chemotherapy, though not nearly as much as the providers of these treatments or their patients would hope. For this last group of women, I am very sorry to say, modern medicine has less to offer than newspaper headlines suggest. The outcome was dictated well before diagnosis—by the date the first cancer cells developed and by the rate at which they grew.

If, as I suspect, a woman's fate is set very early in the development of a tumor, it seems implausible that advances in detection will have an impact on the disease. One can always hope that science will develop a wonder drug that eradicates tumors completely, even when they can't be seen or felt. But for the present I think we should focus research on improving our ability to distinguish between women with breast cancer who can benefit from aggressive treatment and the larger number who will gain little from it no matter what we do.

The ultimate hope is preventing this awful disease, perhaps by modifying the contemporary hormonal environment that seems to promote it. Researchers at the University of Southern California have been examining ways to lower young women's exposure to reproductive hormones. Another approach is to use an anti-estrogen drug like tamoxifen preventively in postmenopausal women whose histories of breast problems indicate that they are at high risk of developing cancer; I am involved in a study that is examining this kind of treatment. The risks in changing the hormonal balance of millions of women are considerable, however, and it seems likely that any new preventive for breast cancer will have its own side effects.

Advances in this area will surely be slow, but they may be the only realistic hope for eventually lowering the death rate from breast cancer. Meanwhile, we should carefully consider whether we are misleading some women with messages of unwarranted hope at the same time that we are needlessly terrifying and hurting other women by diagnosing and treating a condition that will never pose a threat to their lives.

Presented by

Why Is Google Making Human Skin?

Hidden away on Google’s campus, doctors at a world-class life sciences lab are trying to change the way people think about their health.

Join the Discussion

After you comment, click Post. If you’re not already logged in you will be asked to log in or register with Disqus.

Please note that The Atlantic's account system is separate from our commenting system. To log in or register with The Atlantic, use the Sign In button at the top of every page.

blog comments powered by Disqus


Why Is Google Making Skin?

Hidden away on Google’s campus, doctors are changing the way people think about health.


How to Build a Tornado

A Canadian inventor believes his tornado machine could solve the world's energy crisis.


A New York City Minute, Frozen in Time

This short film takes you on a whirling tour of the Big Apple


What Happened to the Milky Way?

Light pollution has taken away our ability to see the stars. Can we save the night sky?


The Pentagon's $1.5 Trillion Mistake

The F-35 fighter jet was supposed to do everything. Instead, it can barely do anything.

More in Technology

More back issues, Sept 1995 to present.

Just In