[sandworm101]

>> so it is probably rare for cancer to go away on its own.

It is actually very common. It has happened to you many times without you knowing it. Cancerous cells can appear and be taken care of immediately by various natural mechanisms. There are mechanisms to correct errors in DNA before a cell divides, or a damaged cells can self-destruct. This happens all the time as skin cells deal with DNA damaged by UV light. And there are diseases that cause cancer not directly but because of breakdowns in these repair mechanisms (see DNA repair-deficiency disorder). If you looked at every cell in your body no doubt a great many are technically cancerous at any one time. A clinical case starts when the natural mechanisms fail and damaged cells divide enough to cause the disease we call cancer.

The distinction is medically important because if you go looking for every tiny little bit of cancer you will find a great many that would never give rise to a clinical case, that would never impact the patient.

[nonbel]

Have you looked into what combination of mutation rates and number of required mutations would be necessary for your model to work? Usually people say it is something like 10^-8 per bp per division and you need to collect 3-6 mutations to get cancer.

[sandworm101]

Isn't cancer predicated on mutations that result in uncontrolled cell growth? I imagine there are a great many mutations that result in cell death or an inability to properly divide. Those mutations wouldn't contribute to cancer risk. So maybe you need 3-6 but only amongst those bits of DNA that actually matter.

[nonbel]

Yes, that is correct. It would only be some "special" combinations of mutations, not just any set.