An equally bad title would be, kids having sex are more likely to text frequently.

This is simple to clarify. The tests to detect these proteins are not 100% perfect but they get more sensitive every year. So just because they are not detectable in certain patients does not mean they are not there. So, this paper just came out recently, that means it was probably submitted for review 3 months (or more) and the actual experiments may have taken place at least a year ago. So in that time, the tests for these proteins have improved yet all the samples that were collected a year or more ago have been used up. Thats only part of the reason they claim "can be 100%"; the other part is that there is already strong evidence to indicate that these proteins are involved in Alzheimer's so they should be present even if they were not detected.