Ignoring any correlations between general health and weight, I would naively think that #cancer risk would be proportional to weight (or rather, weight of the tissue in question): the chance of creating a mutation that is effective at creating a malignancy should be roughly constant per cell per unit of time, so the total rate of that happening should scale with number of cells.

Is there an obvious reason why this scaling is wrong, or is it not observable due to the health-weight correlation? (I've spent a few minutes trying to look it up, but found a mountain of experimental results correlating BMI with cancer risk only.)