But perhaps they should have said earlier because, from their own data, the PRS risk gradient appears to change continuously across the age range. Supplementary Table 4 shows that the odds ratio per standard deviation went from 1.75 to 1.60 to 1.52 to 1.44 as age increased from the 40s to 50s to 60s to 70s (P ¼ 3.44 Â 10 À10 ). On the log(odds ratio) scale, these risk gradients are 0.56, 0.47, 0.42, and 0.37, which is an almost linear decline of approximately 0.06, or 14%, per decade. This is an important observation, not the least because it is contrary to what has been found for breast cancer where the risk association for currently the best PRSs for overall, and for estrogen receptor negative cancer is not discernibly dependent on age (the risk gradient decreases by approximately 0.01 per decade; P ¼ 0.39). 2 This null age-dependency implies that, for breast cancer, the PRS might not be predicting death from breast cancer. On the other hand, for CRC, the negative age-dependency implies that the PRS is likely predicting death from this disease, and across a wide age range. Note that, if a PRS is predicting disease mortality, the average PRS would be expected to decrease with age at diagnosis for cases, and age for controls, but at a slower rate for the latter. 3 It is even more important to take into account this negative age-dependence because, as the authors state, "Analyses of PRS, along with environmental and lifestyle risk factors, might identify younger individuals who would benefit from preventive measures." Their data show that this statement is not only relevant to the prevention of CRC before age 50 years, but to prevention of CRC being diagnosed at least across the next 3 decades: there appears to be no threshold. As indicated by the strength of association with age and its P value shown previously, there is a vast amount of risk information on CRC diagnosed after age 50 years (89% of the study sample) that has not been used or commented on in the published paper.