I'd thought that individual cells becoming 'cancerous' was medium-common and our immune systems, when working properly, detected that and destroyed them - so one way of thinking about cancer is an insufficently active immune system. I know that's a vast oversimplification*, but if the thrust of it is correct the numbers here suggest that the inverse (immune system not on high alert) is protective against dementia.
*probably to the point of being laughably wrong (something I'm pretty good at). Michael Levin's lab has done some interesting work on cell/cell communication and how 'cancerous' cells can be forced into normal operation via manipulating the ion channels that cells use to communicate with each other, which leads me to wonder if there's something immune cells (macrophages?) can detect when a cell stops communicating with its neighbors appropriately. If it's not obvious, I'm out of my depth here. :-)