Tamoxifen, the breast cancer drug, may very well mark the beginning of a new era of "chemoprevention" in public health. But it is not clear how popular tamoxifen is really going to be. And some skeptics see a downside to depending too much on drugs to keep us healthy.

The idea of health in a bottle, prevention in a pill, has slowly but surely seeped into our notion of what medicine is all about these days. We take aspirin to prevent heart disease, calcium to prevent osteoporosis, and even a B-vitamin to prevent birth defects. Nevertheless, when the National Cancer Institute announced the early results of the first clinical trial of tamoxifen as a drug to prevent breast cancer, it seemed like a genuine breakthrough: here was a drug that stopped cancer, and one of the most dreaded cancers at that. But as is so often the case in science after the tantalizing results have made the big splash, troubling questions quickly swoop down on settled answers. The story of tamoxifen has become at once a shining symbol of the promise of chemoprevention and a revelation of just how complex and intricate a process drug discovery can be--scientifically, economically, and ethically.

On October 29, 1998, the FDA approved the use of the drug tamoxifen to reduce the development of breast cancer in healthy women who are at high risk of getting the disease, making it the first anticancer drug sanctioned for use in disease-free individuals. The decision was based predominantly on the results of the National Cancer Institute—sponsored Breast Cancer Prevention Trial (BCPT), which was stopped in April 1998--14 months early--having shown a dramatic difference in the incidence of breast cancer between tamoxifen-treated women and those on a placebo. This large-scale test of tamoxifen involved 13,388 high-risk women aged 35 and up and concluded that the incidence of invasive breast cancer was reduced by about half in those taking the drug: 44 percent for women aged 35—49, 51 percent for women aged 50—59, and 55 percent for women 60 and older. On the downside, taking tamoxifen also increased the risk of potentially fatal blood clots and endometrial cancer (cancer of the lining of the uterus), particularly in older women, who developed these conditions at three to four times the rate, respectively, of those in the placebo group.

Needless to say, the tamoxifen story was big news, and the headlines proclaimed optimistically: "Designer Hormone Holds Off Cancer," "Your Breast Cancer Risk Just Went Down," "Breast Cancer Breakthough." And for good reason: breast cancer is the most common form of internal cancer in this country. According to the American Cancer Society, an estimated 178,700 new cases of invasive breast cancer occurred in the U.S. last year. A women’s lifetime risk of developing the disease is about one in eight, or about 12.5 percent, and more women between the ages of 40 and 55 die of breast cancer than any other cause. "Before the age of 50, if you’re going to get something serious, chances are you’re going to get breast cancer," says Donna Spiegelman, associate professor of epidemiology and biostatistics at the School. "While the chances of getting it are very small, if something bad is going to happen, it’s probably going to be that." Now here was a drug scientifically proven to reduce breast cancer incidence among women at high risk--those with familial indications or other notable risk factors. For them this pill could provide an alternative to prophylactic mastectomy or frequent mammography screening. Was it too good to be true?

Tamoxifen is not a new drug. Prescribed for almost two decades as a breast cancer therapy to prevent recurrence and reduce mortality, it was originally synthesized as a contraceptive in the make-love-not-war era of the ’60s. "It was a great contraceptive in rats," muses Dr. V. Craig Jordan, director of the breast cancer research program at the Lurie Comprehensive Cancer Center at Northwestern University Medical School. "But it did exactly the opposite in women: it induced ovulation." And so it was put on the shelf--until Jordan, embarking on his research career fresh out of graduate school, contacted one of the Zeneca Pharmaceuticals scientists who created the drug: "So I called Dr. Arthur Walpole and said I’d like to turn ICI 46 47 4 [tamoxifen] into a breast cancer drug. Walpole said, ‘That’s what I want to do, too.’" Still, Jordan says many of his colleagues thought he was crazy. Confident that high-dose chemotherapy would cure cancer, they couldn’t comprehend why anyone would dedicate his career to a failed birth control pill--and for prevention, no less.

Jordan had his reasons. "Though people can say it’s the wisdom of hindsight, it actually wasn’t, because there were clues that hormones and breast cancer were connected," he says. For one thing, women with breast cancer seemed to fare better after removal of the ovaries, which are a significant source of the hormone estrogen. Although the exact mecha- nisms by which it works are still largely unknown, tamoxifen appears to act as an "anti-estrogen" in the breast. Some breast cancer cells are estrogen-sensitive, meaning that the hormone binds to certain receptors in these cells and stimulates them to grow and divide in a cancerous way. Tamoxifen appears to interfere with this binding process, muscling its way to the receptor sites with no corresponding malignant effects. Additionally, tamoxifen has a property Jordan dubs an "estrogenic tickle": while the compound acts against estrogen in the breast, it actually mimics the hormone in other parts of the body, likewise improving bone density and lowering cholesterol.

Jordan would go on to show that the drug could stop the estrogenic effects on human breast tumors grown in the laboratory as well as prevent mammary cancer in rats. By the mid-’70s, researchers were doing clinical trials of tamoxifen in women with breast cancer, and in 1978 the once-obscure drug hit the market for the treatment of advanced cases of the disease. More than ten years later it received FDA approval as an adjuvant therapy in early-stage cases. Recent clinical trials continue to confirm its therapeutic value for women diagnosed with breast cancer.

The recent FDA approval of the drug for risk reduction in healthy women caps what Jordan calls "a terrific 30-year story." Three decades is certainly a long haul, but it seems that when it comes to tamoxifen, longer is better. Animal studies indicate the advantages of long-term therapy, and an overview of all the clinical trials of tamoxifen last year confirmed this point in human populations. One year of tamoxifen does virtually nothing to prevent contralateral breast cancer--cancer in the opposite breast after an initial breast cancer diagnosis. "Two years is good, five years is great," says Jordan. "It produces a 47 percent decrease in contralateral breast cancer. That’s almost exactly the same number as in the prevention study, so all of the biology absolutely fits. And if everyone had done the trials for one year or two years, it wouldn’t have been as good." In fact, the NCI’s Breast Cancer Prevention Trial was originally designed to follow women on a daily dose of tamoxifen or placebo equivalent for five years. Yet because the trial was cut short, the average length of time women were taking the drug was 47 months, according to NCI.

But if one of tamoxifen’s biggest proponents notes that "longer is better," why did NCI stop its prevention trial 14 months early, when the participants had an average of only four and a half years of follow-up? The answer touches upon one of the issues that makes a human clinical trial so complex. "That’s really one of the dilemmas in prevention science now--that it is socially and politically unacceptable to continue a trial where there is substantial evidence of benefit," says David Hunter, professor of epidemiology at the School and director of the Harvard Center for Cancer Prevention. "So you have to stop the trial, and that means that the longer term questions about long-term consequences may be forever unknown."

As do most clinical trials today, BCPT had certain mechanisms built into its design to periodically monitor its results and end the trial if a certain threshold of benefit or harm was attained. "You don’t want to keep a study going unnecessarily long when people could benefit from the results of the study," notes Spiegelman. "Nor do you want to keep a study going unnecessarily long if it’s hurting people. So we need those sorts of systems in place because we are experimenting on human beings, which is a very tricky thing to be doing." [See Comment] In April 1998 NCI determined that it had an ethical responsibility to "unblind" the trial early in light of tamoxifen’s dramatic results, so that high-risk women in the placebo group could have the option of taking the drug.

Consequently, the wealth of unknowns that remain have some advocates worried. "We don’t know what would happen to these women over a longer stretch of time, including the length of time, past five years, when earlier studies of tamoxifen showed that more serious side effects became more frequent and the benefits seem to decline," says Amy Pett, executive director of the Massachusetts Breast Cancer Coalition, an advocacy group. Indeed, Pett says it is a major mistake to suggest that tamoxifen prevents breast cancer when all that the trial showed (despite its name) was a reduction in incidence: "We just don’t know whether tamoxifen prevents breast cancer. All we know is that it delayed its appearance for some women for four and a half years. But we don’t know much more than that." Hunter acknowledges that it could be that tamoxifen only delays the inevitable: "There is the theoretical possibility that what happens in a trial like this is that you freeze the growth of the tumors that are about to come to clinical attention, and at the moment you start taking the tamoxifen, if you just take it for a short period of time, you just delay the onset of tumors--you don’t actually prevent them." According to Hunter, much of what we know about breast cancer says that it is a 10-, 15-, 20-year process, and even the use of tamoxifen in a therapeutic context has been predominantly over a five-year period; there’s not a lot of experience with women taking it for longer. There are some indications that tamoxifen’s beneficial effects may persist beyond a five-year treatment period; for example, in an overview of randomized trials published in the May 16, 1998 issue of the Lancet, Richard Peto of Oxford University found that therapeutic tamoxifen given to post-surgical breast cancer patients for five years substantially improved their ten-year survival rates.

A point on which everyone appears to agree is that tamoxifen may not be for everyone. BCPT was a trial in healthy but high-risk women, and both NCI and Zeneca Pharmaceuticals, a Delaware subsidiary of the London-based Zeneca Group P.L.C. that makes the drug, stipulate that tamoxifen should be used only by women in this category. However, researchers generally agree that women overestimate their risk of getting breast cancer, as well as the odds that the disease will be fatal. According to a recent survey by Cancer Care, Inc., in New York, nearly 70 percent of Americans mistakenly believe that breast cancer is the leading cancer killer in women (it’s actually lung cancer, which kills more women annually than breast and ovarian cancers combined). "We have a national preoccupation with breast cancer," comments Dr. Sigrid Tishler, assistant professor at Harvard Medical School and a Harvard Vanguard Medical Associates oncologist, and although she is quick to note that a certain amount of concern is realistic, she worries that it might be "way out of proportion."

So how do you rationally determine if you fit the bill as a woman at "high risk"? The designers of the prevention trial used a computer program based on a statistical model called the Gail Model, which is considered to be the best available means for estimating an individual woman’s risk of developing breast cancer. Developed by Mitchell Gail and his colleagues at NCI in 1989, the original model bases its results on the responses to five risk-relevant facts: a woman’s age; her age at first menstruation; her age when she first gave birth; the number of breast cancers in her mother, sisters, and daughters; and the number of breast biopsies she’s had and their results. The software developed by NCI to determine eligibility for BCPT--the Breast Cancer Risk Assessment Tool, or "risk disk"--uses a slightly tweaked, seven-question version of the model to compute the risk of developing breast cancer, first over the next five years and then over the course of a lifetime (see cancer calcuations). The results are compared to the numbers of a woman of the same age having no risk factors for the disease. If a woman had a risk factor of 1.7 or above--which corresponds to the degree of risk for a typical 60-year-old woman--she met the criteria for the NCI study; on average, participants’ risks were twice that level. Zeneca has now made this software available at no charge to clinicians to aid risk assessment of their patients (to order, call 1-800-898-9702), and although women can order a free diskette for their own individual use from the NCI Web site (http://cancertrials.nci.nih.gov), they are encouraged to use it as a way to talk to their doctors about risk assessment.

However, it may be that all that glitters isn’t gold when it comes to the "gold standard" of breast cancer risk modeling. Back in 1994 Spiegelman was the lead author and Hunter a coauthor of a paper that evaluated the effectiveness of the Gail Model based on the number of observed breast cancer cases in the School-affiliated Nurses’ Health Study ("Validation of the Gail et al. Model for Predicting Individual Breast Cancer Risk." Journal of the National Cancer Institute, April 20, 1994; 86:600—7). What they discovered is that the model had significant shortcomings. "It’s good for choosing, out of 100,000 women, 20,000 who are at moderately high risk," comments Hunter, "but most of these women will still never develop breast cancer." In other words, if you took 100,000 women and plugged their information into the model, it could effectively sort them into high, medium, and low risk, but according to Hunter there are very few risk profiles for any disease that are so compelling as to more accurately predict the risk for a certain individual. "The idea that you can generalize the Gail Model and only give tamoxifen to women who meet some threshold is a very arbitrary idea," he notes. "It’s not like some magic mathematical model that says these women are at ultra-high risk versus women who are at low risk."

This "arbitrary threshold" is giving women’s health advocates pause. Remember that the Gail Model, even NCI’s modified computer version, has nothing to do with tamoxifen; the model determines risk alone, not who should take the drug. Amy Pett argues that the limit set to allow women into the prevention trial--the risk of a 60-year-old woman--was determined by NCI and figures predominantly in Zeneca’s marketing campaign. "So it’s not just a problem with the model that you use for the risk assessment, but how you define high risk," she says, stressing that the cut-off mark should be made by the woman herself after careful assessment of personal costs and benefits. "Does an individual woman feel that the risk of a 60-year-old woman is great enough to warrant taking a ‘risk reduction’ drug that may cause endometrial cancer and potentially fatal blood clots? Or is she unwilling to chance tamoxifen’s side effects until her risk reaches the level of, say, a 70-year-old woman," Pett muses. Spiegelman agrees and wonders if a model specifically designed to address tamoxifen might be better. "What we really need is a Gail Model—like program that takes into account--if women really want to make informed decisions based on their individual risk--all these adverse events and also the beneficial events in addition to the breast cancer one," she says. "To get someone to do this might be a very valuable service to women, a very valuable public health service."

According to Mary Lynn Carver, a spokesperson for Zeneca, to add information about tamoxifen’s side effects to the disk would be difficult because the kind of data one would need on the side-effects side of the coin doesn’t exist. "You can’t say definitively who’s going to have problems and who’s not," she remarks. Carver acknowledges that NCI and other research groups are grappling with these issues and a risk-benefit model is something now under development.

But because no such model yet exists, the present version was a solid and economical tool that the FDA chose to use, argues Carver, in defining who should consider taking tamoxifen; it is the only risk-qualifying model that has been through the rigors of a clinical trial. Tishler, the Harvard Vanguard oncologist, agrees: "It may not be a perfect model, but it’s a way of starting to deal with risk assessment." For now, she says, it should be used with caution in individuals. When advising her own patients, using a program like the Gail Model would be only one small piece of the risk-assessment puzzle, which might also include looking at extended family history, genetics, lifestyle, environment, and personal needs and wants. Only then, according to Tishler, can a woman really make an informed decision on any course of action, be it prophylactic mastectomy, lifestyle change, increased screening, or tamoxifen. Hunter sees problems with that approach: "Well, the conventional answer is: ‘Tell them what we know, and they can make up their own mind.’ But decision scientists who specialize in this even have a hard time deciding what the risks and benefits are--not to mention someone who is not a specialist in the field." Yet for now, the consensus seems to be that women need to arm themselves with as much information as possible, talk to their health care professionals, and carefully consider their own wants and needs before making any decisions about reducing breast cancer risk.

One such consideration may be purely economic: the cost of tamoxifen is about $1,000 per year, with a recommended five-year course of therapy. That FDA approval for preventive use could financially benefit the pharmaceutical company is obvious. "Tamoxifen is approved for risk reduction in women at a high risk for breast cancer," notes Pett. "If high risk is defined as having the level of risk of any woman over 60, and there are 25 million women over 60, that’s a huge market for a drug." She worries that more women than appropriate will take the drug because its marketing plays on women’s fear of breast cancer.

However, the reality of the matter may be a bit more conservative. From fiscal years 1997 to 1998, U.S. sales of tamoxifen (which Zeneca is selling under the brand name NOLVADEX) rose about 19 percent from $310.6 million to $368.6 million. For a drug this is solid but not gangbuster growth, according to Robert Parente, a securities analyst at Leerink, Swann & Co., a Boston-based financial firm specializing in bioscience. In comparison, 1998 sales of Zeneca’s best-selling drug, the antihypertensive Zestril, were over $1 billion. Still, because the FDA just approved tamoxifen for risk reduction in October 1998, it’s probably too soon to say what the tamoxifen market will be.

Tishler notes that she has not been seeing as much interest in the drug for preventive use as she might have expected and cites increased press about the side effects, particularly the increase in uterine cancer, as a possible cause. The answer may actually lie in that ever-present public health predicament: prevention is no easy sell. Says Parente, "In the U.S. health care market, the pound of cure’s often worth a lot more than the ounce of prevention."

Chemoprevention: Wave of the Future or
Tough Pill to Swallow?

Every morning now for about four years, Linda, a 59-year-old Boston-area resident, has taken two little white pills in hopes of staving off breast cancer. Her mother and an aunt died of breast cancer; a cousin is a survivor after chemotherapy. "I have been waiting for the other shoe to drop from a very young age," she says. She was motivated to enroll in BCPT and, depending on randomization, take tamoxifen after hearing a talk by cancer experts at her temple: "All bells and whistles went off." Linda had a strong hunch she was on the active-drug, not the placebo, arm of BCPT because as soon as she started her blood-cholesterol level dropped and her hair turned white (something never cited as a side-effect, according to Carver). When the news broke last year that tamoxifen seemed to work, she says she did not jump for joy or anything like that, but adds that it was nonetheless "very nice to hear it had an effect and it was a positive effect." Linda’s muted response may say something about the limits of tamoxifen and today’s chemoprevention more generally; that for people haunted by high risk of a potentially life-threatening disease, these drugs may be welcomed as a reprieve but are not likely to be embraced as an absolution.

Certainly, Pett, at the Massachusetts Breast Cancer Coalition, thinks it is wrong to place our prevention hopes on tamoxifen and drugs like it. "It’s a real concern that something that is carcinogenic itself is considered prevention," she says, also commenting that a pharmaceutically focused research agenda reduces the chances of finding the root causes of breast cancer. But more than likely, chemoprevention is here to stay, faults and all: the field just holds too much promise. "The idea is that we can build on what we have," says Jordan. "I think what we have now is very compelling, but what I hope is that a future generation of scientists will say, ‘Look, this is a success story; now how are we going to get rid of the other 50 percent of breast cancer?’" Indeed, NCI has started recruiting participants for the five-year, 22,000-subject Study of Tamoxifen and Raloxifene (STAR) trial comparing the prophylactic benefits of tamoxifen with another potential preventive drug that may have a better side-effect profile: raloxifene (an osteoporosis drug sold as EVISTA by Eli Lilly & Co.).

Additionally, the science of chemoprevention may itself provide insights that might otherwise remain unexplored. Says Tishler, "Oh, I think it’s a very important field. The more we understand about the molecular underpinnings of cancer, the better we will be able to understand how to target and prevent the development of clinical cancers." Chemoprevention enthusiasts like Jordan see the field going beyond cancer and tackling coronary heart disease, osteoporosis, and host of other afflictions. "So it possibly does far more than focus on breast cancer," he says. "It focuses on women’s health." And it’s hard to argue against the benefits of that.