Statistics of Human erythroleukemia: is the two-hit model of mouse leukemogenesis valid in human disease?