For over two centuries, opponents of evolution have searched for examples of natural complexity that could have only been created by design. Reverend William Paley was fond of the eye, with its lens, retina, and other components all beautifully fine-tuned to work with one another. These days, the Intelligent Design camp tries to invoke blood clotting cascades or the flagella that bacteria use to move around in the same way. (See here for some refutations of these arguments.) Ironically, one of the most successful, intricate examples of complexity in nature is something creationists never mention: a tumor.

Cancer cells grow at astonishing speeds, defying the many safeguards that are supposed to keep cells obedient to the needs of the body. And in order to grow so fast, they have to get lots of fuel, which they do by diverting blood vessels towards themselves and nurturing new vessels to sprout from old ones. They fight off a hostile immune system with all manner of camouflage and manipulation, and many cancer cells have strategies for fending off toxic chemotherapy drugs. When tumors get mature, they can send off colonizers to invade new tissues. These pioneers can release enzymes that dissolve collagen blocking their path; when they reach a new organ, they can secrete other proteins that let them anchor themselves to neighboring cells. While oncologists are a long way from fully understanding how cancer cells manage all this, it's now clear that the answer can be found in their genes. Their genes differ from those of normal cells in many big and little ways, working together to produce a unique network of proteins exquisitely suited for the tumor's success.

All in all, it sounds like a splendid example of complexity produced by design. The chances that random natural processes could have altered all the genes required for a cell function as a cancer cell must be tiny--too tiny, some might argue, to be believed. And surely the only way that a cell could become cancerous naturally would be for all the genes to change at once. After all, what good is it for a cell to be able to increase blood flow towards itself if it can't grow quickly? Getting so many genes to change at once makes an impossibility an absurdity. By this sort of reasoning, you'd conclude that cancer is the work of a supernatural designer.

And yet, despite all its appeals, creationists don't like to bring up cancer. Perhaps that's because they prefer to use the warm and fuzzy examples of complexity in nature instead of the pain-causing, life-ending ones. I'm no theologian, so I'll leave the religious implications of all this to others. But as a science writer, I do want to talk about what this means about creationism and evolutionary biology as sciences. Creationists say that they want to be taken seriously as scientists. But one mark of important scientific ideas is the important new scientific research it generates. Cancer is a case in point. Creationism in any of its flavors has never led to an important hypothesis about cancer. Evolutionary biology, on the other hand, is generating a wealth of new ideas about potential ways to fight cancer.

Martin Nowak of Harvard University and his coauthors offer a nice roundup of these ideas in a paper appearing in this month's Nature Reviews Cancer. (Nowak has posted a pdf of the cancer paper here, on his publications page. His other papers are worth checking out, too. He's done brilliant work on the evolution of everything from HIV to human language.)

Nowak and his co-authors argue that you can't understand cancer unless you recognize it as an evolutionary process. As cells divide, they mutate on rare occassion (roughly one out every 10 billion cell divisions). Most of these mutations will kill a cell, so that the genomes in most of the new cells in your body are identical to the old ones. But a few of these mutations can allow a cell to divide more quickly than its neighbors. They begin to outcompete the ordinary cells for resources, becoming even more common. These cancer cells continue to mutate, so that there's lots of genetic variation in a growing tumour. In a few cases, these mutations make cells better adapted to a cancerous existence, and the offspring of these cells come to dominate the tumor. As the tumor matures, new kinds mutations may be favored--ones that let it metastatize, for example, or withstand the abuse of chemotherapy.

The same basic dynamics of evolution by natural selection that can alter a species are at work in the cells of a tumor. Obviously, however, the two cases of evolution are not identical. The mutations that alter a species are the ones carried down in sperm and eggs from one generation to the next; the mutations to cells in the rest of the body (the soma) are irrelevant. Cancer, on the other hand, is all about somatic evolution. And while ordinary evolution can last for billions of years, each case of somatic evolution ends with the death of the body in which it takes place.

That said, though, Nowak and his colleagues show how evolutionary dynamics can tell us a lot about how cancers get started and spread. One crucial fact about cancer is that the evolutionary arena where it gets its start is a microscopic one. Our organs are generally composed of millions of little compartments, each containing a few thousand cells. Colon cancer, for example, begins in so-called "crypts" that line the intestines. Normally the crypt is in a delicate balance. A single stem cell at the base of the crypt divides every day, producing a fresh colon cell. The older cells move up towards the surface of the intestines to make room, dividing themselves as well. The oldest cells near the top of the crypt die off in an intricate self-destruct sequence of biochemistry.

The evolution of cancer cells has a different trajectory depending on the size of their compartment. In a big compartment with lots of cells mixing together, natural selection will favor cancerous mutants, which will quickly spread--and possibly spread to neighboring compartments. In a small compartment like a crypt in the colon, supplied by just a few stem cells, cancer may grow more slowly because the cells are more likely to self-destruct before they can cause much trouble. (In fact, the architecture of our tissues overall may be adapted to keeping cancer in check this way.)

Another factor in the spread of cancer are the genes themselves. For example, one common sort of mutation found in cancer cells causes the cells to do a bad job of repairing their DNA. At first, this seems like a very dangerous mutation for a cancer cell to have, since it means that the cell risks mutations to the many genes that it needs to stay alive. Nowak and his colleagues find, however, that bad repairs have a benefit that makes them worth the cost. To understand why, bear in mind that each of our cells has two copies of each gene, inherited from mother and father. In order for cancer to progress, both copies of certain genes have to get knocked out in a cell. This is a remote possibility for most cells, but, according to Nowak's calculations, not for ones that have become genetically unstable. (Genetic instability, Nowak's work also shows, is responsible for cancer's extraordinary capacity to evolve protection against drugs.)

Nowak's work is elegant and fascinating, but as he admits, it's just the beginning of an understanding of how cancer evolves. (He's not the only one pursuing it--this article in the March 15 issue of The Scientist describes how other scientists are pursuing similar lines of research.) It's worth pursuing further, because it may make it possible to predict precisely particular cases of cancer will progress, and help reveal which line of attack will work best.

It will be interesting to see how the members of certain state boards of education react to this kind of medicine. Will they hold off on chemotherapy until they find out what insights their creationist friends have gotten about cancer? If they do, they'll be waiting a dangerously long time.

Originally posted at http://www.corante.com/loom/archives/002449.html. Licensed under a Creative Commons License.