by Greg Gibson
So, why should our genes break so often? The short answer here is that they really don’t: It is the sheer scale of the thing that matters. There are millions of cells in each of our bodies, and every time one cell divides, just a handful of mistakes are made in the copying of the DNA. Humans should be so lucky to design a machine that makes one mistake in every billion operations. It only takes a couple of mistakes, namely mutations that change one letter of the DNA code into another, to push a cell along the winding and inevitable road to cancer.
It is somewhat pointless then to talk about eradicating, or even preventing, cancer. All we can do is contain it, and hopefully fix it before it does too much damage in each individual case. Containing it includes minimizing the number of mistakes made, for example, by reducing exposure to carcinogens such as tobacco and ultraviolet radiation. But it also means ensuring that adequate health care and timely screening are provided so that active interventions can be pursued.
In fact, probably the greatest risk factor for cancer, and the one we can choose as a society to do most about, has nothing to do with the underlying biology, but rather concerns socioeconomic status. Cancer survival is directly related to quality health care as well as the decision to take advantage of that health care, for example, by pursuing early breast and prostate cancer screening options. Without adequate health insurance this does not happen. Without a certain level of trust and confidence in the medical profession, this does not happen, and in this respect race is also a factor. There might well be racial differences in predisposition to certain cancers, but if there are, the impact of the genetics is likely to pale in comparison to the differences in utilization of the health care.
Epidemiology and Relative Risk
One of the great heroes in the history of cancer research and public health is the little-known English physician Janet Lane-Claypon. In the first quarter of the 1900s, having become one of the first women to receive both M.D. and Ph.D. degrees, she carried out a remarkable series of studies that went a long way toward founding the field of epidemiology. One of these demonstrated the nutritional superiority of human breast milk over cow’s milk and led to her being commissioned by the British Ministry of Health to conduct the first large case-control study on the causation of cancer.
After surveying 500 cancer patients and 500 matched cancer-free women she was able to conclude, presciently, that such factors as age at menopause, age at first pregnancy, number of children, and breast-feeding have a significant impact on the incidence of breast cancer. Furthermore, she was also the first to demonstrate that early surgical intervention significantly improves long-term survival. All this in half a career, since marriage in her late 40s to a colleague in the Ministry meant, by bureaucratic rule, that she must abandon her position. Such was the fate of all too many women scientists in the middle of the last century.
The epidemiological risk factors identified by Janet Lane-Claypon are now thought to reflect lifetime exposure to estrogen and other hormones involved in female reproduction. With better nutrition, girls are reaching puberty earlier, and breast maturation commences at a younger age. Delayed first pregnancy and failure to breast feed both increase estrogen levels in the breast, as does prolonged time to menopause. Similarly, prolonged post-menopausal hormone replacement therapy is now widely thought to contribute as much as 20 percent of the risk of breast cancer in older women.
Risk is a difficult concept to get our heads around. Typically it refers to the extra fraction of people who have the disease because of some risk factor. It is formulated as the ratio of the prevalence in the at-risk group to the prevalence in the population at large. A ratio of 2 means that the risk is increased 100 percent and a ratio of 1.01 means that it has increased one percent. It is far from trivial to establish that a relative risk of one percent is significant. Suppose that there really aren’t any factors that make anyone in your town or subdivision or city block more or less likely to get cancer. Over a 20-year period, perhaps 500 of the 10,000 people in the neighborhood are stricken with the disease. If you arbitrarily split the community into two groups of 5,000 people each—say those who live in even-numbered houses, “the evens,” and those in the odd-numbered houses, “the odds”—then you would expect that 250 each of “evens” and “odds” would have cancer. However, it would not be particularly surprising to find 230 “even” and 270 “odd” patients, just because of random sampling. In this hypothetical case, there is more than a five percent difference in the frequency of cancer in the “evens” and “odds.” So, on the face of it, there is a relative risk that large of living in an odd-numbered dwelling and getting cancer.
Then, if you look at enough behavioral patterns, maybe you will discover that orange juice consumption or attendance at the opera is slightly less common in the cancer group than you would expect given the frequency of these activities in the whole community. After the statistical epidemiologists crunch the numbers, there might well be a significant correlation between not drinking orange juice or not enjoying high culture, and contracting cancer. In fact, it is all but inevitable that correlations like this arise by chance, usually at pretty marginal significance levels, but they translate into fairly large apparent risk factors. The only way around this is to replicate the study in different populations, and then to try to work out the mechanism that causes the correlation.
Of course, there is no guarantee that a strong correlation really implies causation in any case. If you look at a map of incidence of all types of cancer in America, painting states with low incidence red and relatively high incidence blue, it is virtually indistinguishable from the political map of George W. Bush’s America. Voting Republican causes a lot of things, but I’m pretty sure that offering protection from cancer is not one of them.
Brakes, Accelerators, and Mechanics
To see why cancer is a disease of the genes but not generally a hereditary disease, we need to understand how defective genes cause cancer and what makes them defective. Cancer is basically a collection of diseases, since the things that go wrong in brain tumors and breast tumors, and in lymphoma and prostate cancer, are quite different. What these diseases have in common is that the division and migration of cells gets out of control.
Basically three types of events can happen in the earliest stages of turning normal cells into cancerous ones: The brakes can fail, the accelerator pedal can get stuck to the floor, and the mechanics can go out of business. When a combination of these things happens, all sorts of other problems arise, and a cell mass that may have been relatively benign, just growing into a lump where it first appeared, starts to metastasize. This means that some of the cancer cells begin to migrate to other parts of the body where they set up secondary tumors, greatly exacerbating the problems.
Framing this discussion of cancer biology in terms of brakes, accelerators, and repairs should help make it clear that the term “cancer gene” is something of a misnomer. It implies that there are genes for cancer, when the reality is that these genes are actually essential for growth and development but can cause cancer when something goes desperately wrong. At least one percent of our genes, and by some estimates more than ten percent, are primarily involved in controlling the timing and mechanics of cell division. This is literally hundreds if not thousands of genes.
Why so complex? Think for a second about what happens after conception. You start out life as a single cell and in the space of a few weeks become billions of cells in carefully orchestrated organs with distinct numbers of retinal, muscle, neuronal, skin, and blood cells. It should be obvious that there is plenty of opportunity for things to go wrong. The more so since the development of an animal is totally self-regulating: There is no template or design that a builder can refer to. Rather, the organism unfolds with only a billion years of experience as a guide to what works.
Coordination of the parts is achieved by thousands of different signals telling each cell what its neighbors are doing and what is going on in the rest of the
body. Cells can divide only after the DNA has been replicated and scanned for errors, and when it is clear that various checkpoints have been cleared. A core set of twenty or so genes orchestrate cell division in all animals (and most of these are the same in plants as well), but more than ten times this number of genes provide the checks and balances that keep the process under control.
The brakes in this process are technically known as tumor suppressors. These are genes whose normal job is to stop one cell from dividing into two cells, either because the body is not ready for more cells, or because all is not right inside the mother cell that is getting ready to divide. When one of these genes is mutated so that a functional protein is no longer made inside the cell, it is as if the brake lining is gone, so cell division just keeps rolling along.
The most famous of the tumor suppressors are p53 and Rb. Most if not all tumor suppressor mutations are recessive, meaning that both copies of the gene must be nonfunctional for cancer to develop. Simply put, one copy is enough, so you can think of the second copy as a backup. In fact, most of our genes are like this. Rb provides an apparent exception, because in half of all cases of the eye cancer retinoblastoma, which affects about one in four million children and is responsible for three percent of all cancers up to the age of 15, the cancer seems to be hereditary. Worse than that, familial retinoblastoma is often transmitted in a dominant manner, and when it is observed, most often both eyes are affected by independent tumors. The major breast cancer susceptibility genes, BRCA1 and BRCA2, are like this as well.
The famous oncologist Alfred Knudson came up with an explanation for this phenomenon in 1971. Knudson’s hypothesis has since become the centerpiece of a general theory for the increased incidence of cancer with age. It supposes that it takes at least two “hits” for a cell to start to become cancerous. One mutation is not enough; both copies must be knocked out, and this is exceedingly unlikely to happen early in life. However, if an individual is born with one mutant copy in all of their cells, then it only takes a single second mutation for a cancer to arise. It is apparently almost inevitable that a second mutation occurs in children who inherit one bad copy of the Rb gene. When this happens, a tumor starts to develop in the eye, and if untreated by laser or some other type of microsurgery, leads to blindness or can spread to the brain.
By contrast, people who have unilateral, nonfamilial retinoblastoma are unlucky enough to get two mutations affecting Rb in their own lifetime. Tens of other tumor suppressor genes can suffer the same fate. It is even possible that two mutations in different tumor suppressors combine to initiate a cancer. As we get older, the chance of two hits occurring in one of our hundreds of millions of cells increases, and so does the likelihood of getting cancer. There is precious little we can do about this.
On the other side of the axle are the accelerators. These are genes whose normal role is to push cells through their natural cycle of growth and division. Technically known as proto-oncogenes, they typically pick up information from outside of the cell in the form of growth factors and send the signal into the part of the cell where other gene products orchestrate the replication of DNA and rearrangement of the furniture that is necessary for cells to divide.
Like the tumor suppressors, these genes are essential for life. The term proto-oncogene means that these genes are primed to become cancer-causing genes, or oncogenes. Oncogenes were first identified in chicken and mouse viruses, giving rise to an early theory that cancer is often caused by viruses. This is true in some circumstances, most notably cervical cancer, but more generally it turns out that the viruses were laboratory artifacts that had picked up activated forms of normal genes with names such as Ras, Src, and Jnk.
Oncogenes alone are capable of promoting cancer, but it would not be accurate to say they act alone. They are so crucial that a lot of redundancy is built into the system, so losing both copies of one of the genes can actually be tolerated quite well. We can engineer mice to have activated oncogenes in all of their cells, and such mice develop tumors predictably, but only a relatively small number of cells become cancerous. So other “hits” must be required here as well. Nevertheless, there is a key difference between oncogenes and tumor suppressors: The mutations that cause a proto-oncogene to become an oncogene are ones that, instead of destroying the gene, make it active all the time.
The fiendish protein products no longer respond to signals from outside the cell; they just do their thing anyway. It is as if the cruise control gets stuck in the on position, or the accelerator pedal is stuck to the floor. There are many more ways to break a gene than to activate it like this, so thankfully, activating mutations are rare enough that only a fraction of us get cancer. The other good thing about oncogenes is that they often result in an abnormally shaped protein that can be targeted by a very selective drug that stops the protein from being active.
The third wheel of cancer genetics is the team of mechanics that repair DNA. If the DNA is broken, then so too eventually will be the instructions for making the brakes and accelerators, and many other things can also go wrong. When the major gene for familial colon cancer was identified a few years ago, DCC1 (Deleted in Colon Cancer 1) turned out to encode a DNA repair enzyme.
Throughout your life, bad things happen to your DNA. Ultraviolet radiation in sunlight causes pairs of adjacent Ts to link together, and if these are not fixed, the code will change next time the cell divides. Similarly, all sorts of toxins that we eat, chemicals in tobacco smoke, and other carcinogens get into the nooks and crannies of the double helix and break it or otherwise chemically modify it. Having broken repair enzymes is not a good thing. Why the colon and breast should be particularly susceptible to this kind of problem is anyone’s guess, but in all likelihood DNA repair is eventually damaged in most advanced stage cancers. Unfortunately, no drugs can take the place of the broken enzymes, so cancer biologists have to try to alleviate the consequences of DCC and BRCA, rather than fix the root problem.
I do not want to leave the impression that mutation of any of these three major components is sufficient for cancer. Current thinking is that such mutations are necessary to start a long process of tumor progression. Slowly at first, but gradually building and eventually snowballing, a mature tumor bears little resemblance to the normal cells in our bodies, as it comes to operate on its own terms. By the time a tumor is actually observed, it is vastly different from the initial cell that managed to escape the shackles imposed on it by the rest of the body. Large chunks of the chromosomes may be lost or duplicated, and tens or hundreds of genes might have picked up mutations.
An intense classical Darwinian struggle goes on inside the tumor. Any new mutation that improves the rate at which the cancer cells propagate is likely to result in that cell outgrowing the others. If you think about it, the nucleus of a cancer cell is the ultimate in selfish DNA. That one cell that starts a tumor may in the course of ten years have more descendants than the first germ cells that founded modern humans 100,000 years ago. In the language of Richard Dawkins, tumor cells are the ultimate replicators. If the sole purpose of existence were to reproduce, you’d be well advised to become a cancer cell.
However, the strategy is ultimately self-defeating. Organisms have evolved complex mechanisms to suppress cancer cells because short-term reproduction is not in the best interests of long-term survival. The community of cells is a nonselfish one in which the various parts work together in harmony. This is a metaphor for life that I find far more enticing and accurate than the nature red in tooth and claw metaphor that so often characterizes popular portrayal of evolution.
Familial Breast Cancer
Only about ten percent of breast cancer runs in families, in the sense that sisters and daughters of an affected woman have elevated risk compared with the general population. Of this fraction, one-fifth can be attributed to two genes, BRCA1 and BRCA2. Women who inherit one bad copy of either of these genes have a lifetime risk of ovarian or breast cancer exceeding 85 percent: Each ge
ne thus accounts for about 1 in every 100 cases of the disease. That is worth worrying about if you are in an affected family, but the flip side of this is that 97 percent of breast cancer has no known genetic basis.
Here and there, studies have by now implicated mutations in at least ten different genes in promoting specific cases of familial breast cancer. Literally thousands of known mutations are in these genes, but a handful of relatively common ones account for most of the disease that runs in families. For example, a not particularly uncommon deletion of a single nucleotide in CHEK2 is pernicious but not as bad as the BRCAs, only increasing risk perhaps twofold. If you carry this mutation, the odds are still pretty good that you’ll be all right. The other mutations are so rare that they are found in only a handful of individuals. Many of the proteins that these genes encode work together as a mechanical complex that prevents something called genome instability. If the cell’s mechanics don’t work appropriately, then after a while the chromosomes basically either fall apart or divide inappropriately, leading either to abnormal growth or cancer.
Our best explanation for the relatively high incidence of these breast cancer susceptibility alleles in the human population, somewhere around two percent of all people, is genetic drift. Just like every other gene in the genome, upwards of 1 in 100,000 people are born with a new mutation in BRCA1 or BRCA2 or any of the other yet-to-be-identified tumor suppressors in the group. These individuals have the same predisposition to cancer as people who inherit a mutation from their parents instead of contracting their own. Given the late onset of disease, there is no reason why most of these people won’t have just as large families and normal lives as everyone else. Consequently, the mutations can hang around in the gene pool without doing too much harm, and some of them can drift to more common frequencies.
