by S. Lochlann Jain
The 2007 San Antonio Breast Cancer Symposium (SABCS) raised all of the fascinating questions around the disease and the politics of numbers. Of the many terrifying aspects of cancer, one is the way it forces people to inhabit statistics, that is, to live in that place defined by chance and prognosis. I learned this with my first research into chemotherapy treatments for breast cancer: the results of randomized controlled trials (RCTs) uniformly showed survival increases of about five to seven months with chemotherapeutic regimens. The question I had was, why would anyone spend six months in treatment for only five months of increased survival?
It turns out—unclear in the statistics about survival “increases” and “rates”—that five months present an average. Each individual will either be one of the few people for whom the treatment works or not. Thus, the politics of chemotherapy: how does an individual put herself into a population-based statistic? Of the approximately 65,000 node-negative women a year who receive chemotherapy for breast cancer, only around 5,000 will benefit from it.1 Each person hopes to be in that number, but it’s a gamble. In the terms of risk/benefit that oncologists by definition speak in (the profession has been guided since its inception by statisticians), it is a gamble that many consider worthwhile, although there have been notable exceptions, such as breast cancer activist Rose Kushner, who refused chemotherapy because she believed it was too toxic. Besides, it’s really all there is. So, we are all busy and the oncologist just needs to know: do you want it, or not?
So how do patients, researchers, and physicians make sense of the kinds of information produced by RCTs? Despite the status of RCTs as the “gold standard” in evidence-based medicine, in most instances they are far from what many of us think of as unbiased. The lack of standardized definitions of such basic data points as “time to progression” or “relapse-free survival” means that trials are very difficult, if not impossible, to compare; people are not only analyzed in their groups as having received treatments, but also in terms of “intent” to treat; statistics are widely manipulated, and so on.2,3 The faults and fluidity of RCTs are well known among scientists, and the recent book False Hope, on the failure of RCTs in the introduction of high-dose chemotherapy with autologous bone marrow transplant (HDC/ABMT) as a treatment for breast cancer clearly outlines these issues for a lay and professional reader.4
RCTs in medicine derived from an agricultural practice. How were fertilizers to be compared when patches of land had different exposures to rain, sun, and other environmental factors? When divided into narrow strips, the researcher could randomly decide which strips to fertilize. Thus, by statistically randomizing the other environmental factors, they could—at least in theory—isolate the fertilizer’s effects. The more patches of land used, the more likely it was that factors other than the fertilizer were truly randomized.
Thus, RCTs turn up useful evidence only on large populations in which other factors really are randomized and in which similar groups are produced for comparison. The question remains very open in breast cancer research, as demonstrated at SABCS: how many factors can be included in a trial’s eligibility requirements and still turn up useful information? Can useful evidence be gathered when a study includes both pre- and postmenopausal women, all hormonal statuses, and stages I to III? Judging from current trials, researchers apparently think it can be. But there is much evidence to the contrary, and this was demonstrated in spades in San Antonio.
When I asked oncologists in the conference center about what they found most exciting about the conference, several physicians said they would immediately start changing their practices around the use of anthracyclines because of a study showing that the majority of breast cancer patients received no benefit from this aggressive treatment. (See “From the Executive Director,” in this issue.) One physician, for example, told me that he would stop giving anthracycline-based chemotherapy to his stage I and II cancer patients. When I asked about stage III, he said he would continue to use them because he would want to “shoot from both barrels.” (I regret not asking what he meant: surely either a treatment is shown to work or it is shown not to work?) There were many other presentations at the conference, and there are many excellent summaries of the conference (read, for example, BCA’s conference coverage). But because chemotherapy is at the heart of SABCS—indeed at the heart of oncology and the rise of this profession—let me give a bit of background to this “breakthrough” study on anthracyclines.
The Italian oncologist Gianni Bonadonna revolutionized breast cancer treatment in 1976 by demonstrating that the chemotherapy regimen known as CMF (as compared to nothing) reduced recurrence rates from 24 percent to 5 percent after 14 months of follow-up. Despite hot contestation of the statistical methods and the actual numbers (which some argued at the time were more like 16 percent, and that others argued, with longer follow-up, would show no significant difference), the study led to the nearly universal adoption of CMF in the United States. Debates in Europe included research into the use of hormonal therapies and questioned the universal use of CMF for women whose characteristics were different from those of the control groups. The ethical question of treating everyone when it would benefit perhaps 10-12 percent of those women remained hotly contested in both Europe and the United States. Because there were no other options and perhaps because of the financial incentive and/or fear of medical malpractice claims, CMF became the standard protocol.
Aside from the approval of tamoxifen in 1986, and the recent approvals of aromatase inhibitors (AIs) and Herceptin, the main improvement in early stage and locally advanced cancer since then has been the addition of anthracyclines to the chemotherapy “arsenal.” Thus anthracyclines such as epirubicin and doxorubicin have been widely used as substitutes for methotrexate (one of the chemicals in CMF) and are now offered as the state-of-the-art treatment protocols. These have been found to give a 4-5 percent survival advantage over nonanthracycline-based therapies.
Back to San Antonio. The big news this year was that due to a systematic review of RCTs of chemotherapy regimens that differentiated between HER2 subtypes (HER2-positive versus HER2-negative), “the published data demonstrate a remarkably consistent finding. Specifically, the incremental efficacy benefit attributed to anthracycline-based therapies is restricted to the HER2-positive subgroup.…The use of anthracyclines in the adjuvant treatment of all breast cancer is not supported by the existing data. Given the known long-term cardiac and leukemogenic/MDS toxicities of anthracyclines and the lack of an incremental benefit in non-HER2/topoIIa coamplified cancers (which is 92 percent of the overall breast cancer population), other approaches to the adjuvant treatment of breast cancer should now be adopted.”5 TopoIIa is another tumor marker used to characterize an individual tumor. The FDA recently approved a genetic test to assess the TopoIIa status of breast cancer tumors.
In my role as a patient advocate, I found it curious that this would not have turned up sooner, and in my role as a researcher, I was curious about how the organization of evidence-based medicine would allow this seemingly large, seemingly egregious, and seemingly straightforward error to have occurred.
So, if we are basically back to the chemotherapeutic treatments of the 1970s (not to dismiss the AIs, tamoxifen, and the benefits of anthracyclines to HER2-positive subgroups), then we are back to early detection. In its efforts to avoid the politics of cancer causation, the American Cancer Society has been informing Americans about cancer since 1913 through campaigns focusing on early detection. Many other organizations are now doing likewise, and to this day their instructions take the form of: when you find a lump or have bleeding, go to your doctor. What we don’t hear about in these campaigns is how your doctor should proceed and what you should do if your concerns are dismissed by your physician. At the various support groups I have attended, I have heard again and again that women with later-stage cancers have been told they were too young to have cancer, that lumps that are malignant don’t hurt, and that they should come back in a year. (I am curious about why these experiences don’t count as “evidence.”) For this reason, breast cancer misdiagnoses are among the most common medical malpractice claims in the United States.
But is this oversight inevitable? On the other end of the spectrum of presentations, that is, the small poster sessions, there were three of interest, one on cognitive dysfunction after chemotherapy, another on hemorrhages in the eye and changes in color recognition that seem to result from tamoxifen and AIs, and the third on “the missing exam.” The last of these posters noted that over 30,000 breast cancers each year are missed by mammograms and that only about half of American doctors routinely complete breast exams. According to the posters, doctors who do these exams claim that they cut them short because they worry about their sexual implications.
What if we started taking this side of the equation seriously? In a recent article in the New Yorker,Atul Gawande argued that hospitals that used simple checklists (that included things like hand washing, patient draping, etc.) reduced infection rates literally to zero.6 Surely, giving all women in the United States access to a high-quality breast exam each year would result in the early detection of the more than 200,000 diagnoses of all stages made each year. This alone might reduce the number of deaths by perhaps thousands each year at a fraction of the cost of sending people through years of chemotherapy and other treatments.
While everyone would like to think of the hypersexualization of breast cancer (through the pink ribbons, various ads, and so on) as separate from the “true science” of breast cancer, the “missing exam” makes it clear that they are two sides of the same coin.
And this is where San Antonio, for me, was utterly depressing. The hegemony of the RCT as a method in oncology meant that presentation after presentation reported studies of the same old agents with the most marginal differences in survival benefits. There were several studies that infinitely parsed the staging and grading of cancers (to what end, since there were no differences in treatments?). There was virtually no research on the side effects of the new and old drugs. When I asked the physician panel about this in the patient advocate meetings, I was told that more money is needed—that patient advocates need to raise more money. Other researchers said that we need more people signing up for trials, a claim I heard at least three times. Often I read that childhood leukemias are now virtually curable because such a high percentage of children were in trials during the 1960s and ’70s, but only about 3 percent of people with breast cancer enter trials. Many reasons have been suggested for this discrepancy, including lack of support in hospitals for administrative oversight (such as linking patients with trials) and patient and physician compensation (time, side effects, and so on). However, no amount of patient flesh and no amount of money can improve cancer treatments if the trials are simply testing the same old things for fractions of increments of survival increases. As several doctors have been calling for at these meetings for many years, a paradigm shift is in order.
S. Lochlann Jain is a professor of medical anthropology at Stanford University, currently writing a book on cancer research and politics. She has published Cancer Butch and Living in Prognosis, which are available at www.lochlannjain.typepad.com/lochlann. She was told twice in Texas that she looks like K.D. Lang. Lochlann welcomes correspondence of all kinds at: firstname.lastname@example.org (and there is nothing she would like more than to be wrong about certain parts of this article)
2 Time to progression is a measure of the time after a disease is diagnosed (or treated) until the disease starts to get worse. Relapse-free survival is a measure from the date of diagnosis to the date of any recurrence. Both measures are used as surrogates for overall survival, although this has not been validated.
3 Intent to treat analyses of clinical trial results include all data from participants in the groups to which they were randomized even if they never received the treatment.
4 Richard A. Rettig, et al., False Hope: Bone Marrow Transplantation for Breast Cancer (Oxford University Press, 2007). The book considers RCTs in relation to high-dose chemotherapy in part due to the inability to find human subjects, but it also gives a good critique of the method and its applications.
5 Dennis Slamon, et al., “Role of Anthracycline-based Therapy in the Adjuvant Treatment of Breast Cancer: Efficacy Analyses Determined by Molecular Subtypes of the Disease,” presented at SABCS on December 13, 2007, abstract #13.
6 Atul Gawande. “The Checklist,” New Yorker, December 10, 2007.