[In this post, I may ascribe agency to processes, inanimate objects or microorganisms: this is rhetorical, I know they don't "intend" anything. I will also use "evolution" in a fairly loose sense of "learning" (again, rhetorical agency) to overcome obstacles to survival].
Cancers are caused by mutations of cells, and "successful" cancers are caused by mutations that result in a, sometimes literally, killer combination of
- Suppression of their own apoptosis
- Invisibility to the immune system or resistance to immune attack
- Ability to co-opt biological resources such as bloody supply so they don't starve
The various "types" of cancer that we know and give names to are the end-points of available "mutation pathways" that cells can take to get to such a "successful cancer" state. Because lots of mutations result in cancers that don't "work" for some reason (e.g. immediately shut down by an immune response), they just don't result in a noticeable tumor. Therefore, we end up with a discrete set of "familiar" cancers, each representing a different path, and their frequency related to how likely it is that path is hit (i.e. how many mutations needed, etc).
Some cancers, the malignant ones, happen also to spread very effectively, though this is not specifically required by the above success criteria, so we also see benign tumors which don't spread but still can be said to "succeed" in that the cells survive.
Once mutated, cancers are presumably fairly genetically static within a single case, otherwise we'd see a storm of cancer variants developing within patients over a relatively short period as an evolutionary arms-race took place, much like the race between antibiotics and bacteria.
- Some cancers have quite "advanced" mechanisms for spreading1
- Some cancers are extremely resistant to human treatment
How did these "dangerous" cancers becomes so dangerous? A dangerous cancer seems to be "fully capable" of effective spread in the first instance, rather than undergoing a complex evolutionary battle with the host's own anti-cancer abilities or some exogenous treatment.
Furthermore, even cancers that "succeed" can't effectively propagate themselves like bacteria, because they're always in the same body2. If anything, cancers that do not kill the host should be selected for, since people with genes "near" dangerous mutation opportunities would be less likely to reproduce effectively3. So (naively, since it's clearly not the case), if people do evolve such that they can get cancer4, cancers would tend towards being both benign and generally non-life threatening.
Even if they do somehow "evolve" to spread and threaten the host (as they indeed evidently have), why are they then also often so good at evading non-host treatment?
It seems like two warring armies (host vs. cancer) that have evolved together using hand weapons, evenly matched for centuries. The cancer invents the crossbow (mutates) and starts to win the war. Then, aliens (doctors) suddenly turn up with UAV surveillance and cruise missiles (imaging and treatment) and side with the host, but the cancer still wins much of the time, even though they're facing an Outside Context Problem that by rights they have no innate ability to counter.
So what results in these cancers being so adept at spreading themselves without some adversarial evolutionary process (à la bacteria vs. antibiotics)? Furthermore, how are they so often also resistant to human-directed treatment? Is it just because the human treatment is crude (i.e. not so much aliens but a large but unruly peasant milita that often doesn't really help that much)? Is there more "evolution" going on within a single cancer patient than I imagine?
Expansion in response to comments: if my assumption that the cancer cells are not usually "evolving" once formed is indeed incorrect: is that evolution then usually the cause of treatment escape? And does that mean that treatment-escaping cancers are freely evolving, in that they could take a number of case-dependent paths to escape (as opposed to the original recognizable cancer "type" which took one of a limited number of available mutation pathways through the fairly static environment of human biochemistry to become that cancer in the first place)?
- For example, the use of matrix metalloproteases to escape into the bloodstream, the use of cytoskeletal locomotion and various other highly specialized mechanisms that I would normally associate with a long and involved adversarial evolutionary arms race on a far longer timescale than one patient's cancer progression
- Certain transmissible cancers like in Tasmanian devils notwithstanding
- The obvious mechanism for this is they die or become critically disabled before having children, but it could also be disadvantageous for elders to die too early, especially for humans, when they assist in rearing their own descendants
- Perhaps as a by-product of some otherwise-advantageous genetics, like sickle-cell anaemia and malaria resistance, where the occasional lethal downside of in a certain percentage of people is outweighed by the statistical benefits.