Most chemically induced cancers seem to have a long latency between exposure to the carcinogen and expression of the disease. Mesothelioma typically 30-40 years, bladder cancer about 20 years, for example.
Most carcinogens also seem to have a dose-response relationship, with both tumour formation andlatency.
The only way I've ever heard latency explained, however, is the number of cell activities in the presence of the mutagen which are required to give reasonable accumulation of errors, but this seems like the same level of variation as dose. The higher the dose the more likely an error so the fewer iterations required before such an error might reasonably occur.
Yet this doesn't seem to be how latency manifests in epidemiology at all. For a start it seem an absolute value - no one no matter how unlucky, gets cancer five minutes after exposure to the carcinogen, it's seems to take some absolute length of time no matter what. Also, whilst latency is dose dependent it seems much less so. Risk can vary between 1 in a few hundred at high exposures, to 1 in a few million at low exposures, yet latency varies hardly at all despite massive variation in dose.
So basically, what causes latency in carcinogenisis, what is happening in the body for those decades, and how does this differ from the randomness of simple dose-response carcinogenicity?