by Greg Gibson
Nature does not eliminate all of them, because it has much more severe problems to deal with, ones that act earlier in life and negatively impact more lives. Until recent times most breast cancer susceptibility mutations had a negligible impact on public health. Now that women live longer and have greater lifetime estrogen exposure, the mutations are contributing to the increased frequency of breast cancer. As small groups of people settled new lands, if a few of the founding men or women carried a susceptibility allele, it would easily have reached the kinds of frequencies that we see in, for example, Ashkenazi Jews for BRCA1. This is just a normal consequence of population genetics, not at all a unique property of cancer genes or Jewish habits.
Growth Factors and the Risk to Populations
Turning now to the vast majority of cases where breast cancer does not run in families, we can still ask whether genes might be involved. Groups all over the world have been tackling this question from different angles for a dozen years. For the most part they have been taking educated guesses and following up suggestive leads. In 2005, a consortium of 20 of these groups decided to pool their resources and look to see whether there were any consistent results in their combined set of more than 30,000 cases of breast cancer, admittedly mostly in Caucasian women.
The outcome was somewhat humbling. Just two genes out of the nine best guesses showed a consistent association with risk of cancer. The most convincing of these is a variant of the gene CASP8, which provides a modest protection against breast cancer for about a quarter of all women. The gene is involved in getting cells to commit suicide when damage to the DNA is sensed. It is the only case we know of to date where the protective variant is the less common one, where most women are at greater risk of cancer because they have the more common flavor of genetic variation.
The other gene encodes an important growth factor known as TGFβ, short for Transforming Growth Factor Beta. This gene is required for breast development in the first place. It is thought on the one hand to counteract the early development of tumors, but on the other to promote their aggressive growth after they are established. In this case, the “bad” version of the gene increases the risk of breast cancer by around seven percent in carriers and 17 percent in women with two bad copies. It is the less common allele in the population, but not by much, and about half of all women carry the risk factor.
About the same time, a group based in Cambridge, England, took a different approach. They decided to look at a lot more genes in their cohort of 4,400 patients and 4,400 age-matched cancer-free women. No single gene emerged as a significant risk factor. However, they did find a trend implicating two sets of genes as if they were gangs of teenagers looking for trouble in the neighborhood. It is tough to pin the crime on any one, but collectively they carry a smoking gun.
One group is involved in regulation of when and how often cells divide. It is not hard to imagine variation in such genes contributing somehow to cancer risk. The other group is involved in steroid hormone metabolism. This is interesting, because the leader of the gang is ESR1, the estrogen receptor. Recall that the combination of earlier menarche (the first menstrual period) and later menopause, both leading to altered estrogen levels in modern women, is now thought to be the leading candidate for the increased incidence of disease over the past several decades. If there are genetic variants that reduce estrogen exposure, it stands to reason that these would tend to be protective against breast cancer.
Subsequently, by 2007, the large consortium had a handle on another half dozen candidates after scanning the whole genome. Several of these don’t immediately make a lot of sense (one is a heck of a long way from any known genes), but three of them also implicate variation in the transfer of hormonal signals inside cells.
Much the strongest case is for FGFR2, a receptor for Fibroblast Growth Factor, which as the name implies is a protein that normally encourages skin cells to grow. In some breast cancers, FGFR2 is amplified in the genome, and there is evidence that different shapes of the protein produced by splicing together different bits of the gene, vary in their effectiveness. It now looks as though two-fifths of all women have some changes in a part of the gene that affects how it is expressed—either how much is made or how it is stitched together—and that these women are more likely to get breast cancer than the majority of women. Of all the women who will have breast cancer at some point in their life, almost one half will carry the “bad” allele of FGFR2, and for one quarter both alleles will be of the bad variety.
A large American study has confirmed that this is the case also for postmenopausal women who show no family history of the disease. This one gene, then, accounts for as much as one-seventh of the total genetic risk for breast cancer in the general population. That sounds like a lot, but keep in mind that most of the disease is sporadic, and not caused by genetic variation. In other words, there is no reason for anyone to rush out and get tested. FGFR2 is simply not predictive on its own, and the great majority of carriers will never get breast cancer. In fact, putting together all the evidence from the half dozen new genes explains only another five percent, in addition to the ten percent for the BRCA genes, of the increased risk even in familial breast cancer.
It is highly unlikely that any more genes remain to be discovered that have effects of the same magnitude. It is much more likely that hundreds of susceptibility factors are sitting around in the genome, each with extremely modest effects or only to be found in relatively few people. Every woman probably has some of them, but it is the total constellation that elevates or depresses her risk of cancer—well, that plus lifestyle factors and raw chance.
Pharmacogenetics and Breast Cancer
Many physicians would argue that we ought to be more interested in finding the genes that mediate how a person responds to cancer therapy than in finding the genes that cause cancer in the first place. Traditional cancer treatments were like taking a sledgehammer to dividing cells. Radiation breaks the DNA into fragments that the cells have a hard time putting back together as they are dividing, while the older chemotherapeutics generally inhibit the replication of DNA. These treatments cause hair loss, nausea, and other side effects because certain types of cells in the body are always dividing as a normal part of life, and these are affected by the nonspecific cancer treatments.
The new drugs by contrast are designed to target just the cancer cells. They do things such as interfering with the receptor proteins on the cell surface, messing up the communication pathways specifically inside cancer cells, enticing them to commit suicide, or disrupting the blood flow to growing tumors. In many cases, it is not exactly clear how the drugs work, and some really effective drugs fail in clinical application because of “off-target” effects elsewhere in the body.
Just how far genomic medicine has and has not come in the few years since the completion of the first draft human genome sequence is shown by title of an Act introduced to the U.S. Senate by Barack Obama. The Genomics and Personalized Medicine Act of 2006 (S-3822) proposed several initiatives, and a large amount of money, to accelerate the translation of genome science into clinical practices. It is foreseen that these will eventually be simultaneously tailored to the unique genetics of each patient and sensitive to the racial and environmental circumstances of the individual.
Upwards of 100,000 hospital patients lose their lives each year as a consequence of adverse drug responses. For example, 85 percent of the cases of childhood acute lymphoblastic leukemia can be treated with the drug 6-MP. Unfortunately, one in ten children carries a variant form of the TMPT gene, and as a consequence they are unable to metabolize the drug. If physicians know this, they can reduce the dosage and eliminate the adverse response, helping thousands of children a year. Who wouldn’t want to see the widespread application of such personalized medicine?
Contemporary cancer therapy is already tailored to molecular attributes of biopsy samples. FGFR2 is just one of the many receptors that provide hints about what the cancer cells are respon
ding to inside a person’s body. Two of the other most important biomarkers are HER2 and ER, components of the receptors for Epidermal Growth Factor and estrogen, respectively.
A quarter of all advanced breast cancers make too much HER2, and this is associated with increased recurrence rates and hence mortality. In one of the early success stories for medical biotechnology, Genentech set out to specifically inhibit HER2 activity by making a molecule that binds to and inactivates the receptor. The drug, trastuzumab, trademarked as Herceptin, is actually a modified antibody, just like the antibodies that your body normally uses to fight infections.
A course of treatment costs in excess of $70,000, so health care providers have been reluctant to approve its use for treatment of early stage cancers. Increasing evidence suggests though that early intervention can be highly effective, potentially saving not just lives, but hundreds of millions of dollars a year in care for terminally ill patients. For this reason, the pharmaceutical industry is extremely active in developing new inhibitors of HER2 and other receptors like it that are implicated in numerous cancers. GSK’s Tykerb, AstraZeneca’s Zactima, Novartis’s Gleevec, and Genentech’s Tarceva are just a few examples of such so-called tyrosine kinase inhibitors to watch for over the coming decade.
The chemotherapeutic drug of choice for combating estrogen responsive cancers has long been tamoxifen. This compound was actually first developed in the 1960s as a potential contraceptive pill. Effective for that purpose in rats, it turned out to stimulate ovulation in human women, not exactly a desirable property of a contraceptive. Further studies revealed it to be an effective antagonist of estrogen in breast tissue, and consequently an excellent drug for inhibiting the ability of the hormone to stimulate growth of Estrogen Receptor positive cancer cells.
A new generation of Selective Estrogen Receptor Modifiers (SERMs) is being introduced that get around some of the problems with tamoxifen, which is now also known to increase the likelihood of development of uterine and endometrial cancer. Eli Lilly has developed a similar drug known as raloxifene and marketed as Evista, which appears to be as effective in reducing recurrence of breast cancer, without the side effects. This drug is also unaffected by a common enzyme type, CYP2D6, that digests tamoxifen and reduces its effectiveness for some patients.
Down the road, genomics experts see a day when profiling cancers with a new technology known as microarray analysis will allow physicians to tailor particular drug regimens according to the entire molecular signature of the cancer. The idea is that the profile of hundreds of genes is likely to be more predictive than just the two or three that are currently examined. Hundreds of millions of dollars are being invested in this possibility, but it remains to be seen whether the technology will deliver on its promise. Currently the approach has limited approval for use with low-grade cancers where treatments are ever improving anyway.
This new approach also holds promise for guiding supplementary treatments when cancers evolve resistance to drugs such as Herceptin. Such is the competition among cells that when humans try to conquer their uncontrolled growth with drugs, the cells escalate the arms race by accumulating mutations that thumb their noses at the treatment. There are as many ways this can happen, as there are signaling pathways inside cells. Unlike normal cells, cancerous ones don’t care to behave as they should; they just want to survive and divide. By looking at all the genes at once, clinicians hope to be able to target just those processes that have gone particularly sour.
Why Do Genes Give us Cancer?
Why hasn’t natural selection ensured that the protective versions of all the genes associated with cancer development or progression are the predominant type in the human population? The answer to this question is quite possibly a good example of the disequilibrium between our modern genetic makeup and what might have been the ideal human genetic condition throughout our history as a species.
Female reproduction is one of the traits that evolved most rapidly in humans relative to other primates. Hormonally regulated processes such as the timing and cycling of menstruation and preparation for breast feeding changed greatly a few hundred thousand years ago. This almost certainly involved selection on genes involved in hormone production over a period of thousands of generations.
Advocates of fundamentalist Darwinian medicine would probably make the argument that breast cancer is better regarded as an example of genomic conflict. They would argue that since all organisms are attempting to maximize the number of offspring they have, mutations that cause estrogen to be produced earlier in girls will tend to advance menarche and hence lead to earlier pregnancy, increasing the number of children they have. However, since as adults they are more likely to have breast cancer, there will be an opposing force of negative selection, setting up a trade-off that leads to the balance that ensures that puberty comes in the midteens.
What is wrong with this type of argument? Let’s start by recognizing that earlier childbirth does not necessarily translate into having more children over a lifetime, and even if it does, it does not necessarily mean that the children will be more “fit.” Birth weight is a vital indicator of child mortality and health, and is a function of maternal health. Menarche generally occurs only after a girl reaches a total body fat level of 17 percent, and the regularity of menses is also a function of growth and nutrition. Nobody knows what the relationship between early motherhood and long-term fitness may have been during human evolution.
Next we must recognize that breast cancer is predominantly a postmenopausal disease, meaning that it affects women after childbearing age and hence is not selected against with respect to having children. True, it is not advantageous for a child to have her mother die young. We also know that given the fullness of time, nature can sift through genetic differences that have an impact of just a fraction of a percent on childbearing. So I am not saying that there is no selection against alleles that promote cancer late in life. But I am saying that it takes a lot more careful empirical observation and mathematical reasoning to establish the argument than just to make it casually.
Was the incidence of breast cancer high enough to trade-off against any possible benefits of early pregnancy? It is almost impossible to know and seems unlikely. Any trade-off argument is a gross oversimplification. Estrogen regulates hundreds of coordinated processes; timing of menarche and menopause are changing against a background of dramatic remodeling of reproductive strategies and of primate lifespan, not to mention mortality and health risks. A few hundred thousand years is a very short time to expect the genome to come to some sort of equilibrium.
What we can be confident about is that there is genetic variation affecting estrogen production, that this variation has been under selection throughout human history, and that any connection to breast cancer liability is likely to be incidental. Now throw in all the changes that modern society has wrought—excellent childhood nutrition, cultural taboos against teenage pregnancy, the obesity epidemic, stresses on nuclear families and social relationships, longevity—and whatever effect that variation had in the past is turned upside down. Don’t expect a new equilibrium any time soon, and don’t expect the genetic risk factors to go away either.
A couple of naïve but legitimate questions remain that deserve to be asked about why cancer is so prevalent, accounting for as much as a fifth of human mortality, if this natural selection thing is supposed to be so efficient. Why would there be cancer genes at all, and why hasn’t the evolutionary process done something about them? I do not have a quick sound-bite answer, other than to reiterate that cell division is so complex that there is ample scope for it to go wrong, and that the high incidence of cancer is really a modern phenomenon. Let’s recap the salient facts from this chapter that might help us to understand this better.
First, the terminology “cancer gene” is at best misleading and definitely inappropriate. It implies that there are for some reason genes whose job it is to cause cancer, just like there are viruses and bacteria that se
em to exist solely to cause misery. The reality is that hundreds of genes that are perfectly good and vital citizens of the genome unfortunately mutate into forms that contribute to cancer. The most insidious mutations affect genes whose function it is to protect the genome by repairing the DNA: When these stop working, the genome starts to fall apart, and cells lose control of their place in the organism.
Every single human carries mutations in several of these genes, but since a lot of redundancy is built into the control of cell division, it does not much matter. At least, not until new mutations build up in our cells during the course of life, knocking out these failsafe mechanisms. This is why chance plays the major role in determining who will get cancer: It is just a stochastic matter of who is unlucky enough to find themselves with a bad combination of mutations that the body cannot eliminate.
Second, we shouldn’t blame the genes for cancer: By far the greatest threat comes from environmental factors. Smoking or hanging out in smoky bars, not eating your greens, tanning in the midday sun, and the combination of early puberty with delayed pregnancy, are all to blame. These are things we can do something about, though reality usually steps in the way, and most of us make a Faustian pact to trade a little extra cancer risk for the pleasures of social engagement, healthy looks, or an independent career. The one thing we cannot do anything about, though, is the biggest environmental factor of all, and that is growing old.