Those who died would still have been at risk into their 80s; many of them didn't make it though.
As a pure thought experiment using made up numbers, let's suppose that people are divided into two equal-sized groups: group A has a 75% chance of developing cancer after age 60, whereas group A has zero risk. We have no idea who is in what group. So it looks like people have a 37% chance of getting cancer after 60. Now suppose many of those who get cancer after 60 due to this risk end up passing away. They are of course from group A, and so more of the group B is represented among the survivors going past 80. Since in the 80+ group, the mix of A:B is no longer 50%, but has fewer A people, the risk of cancer is lower.