That study wasn't longitudinal, following the same people over years and measuring their metabolites. If you measure a bunch of things you can be almost certain to find some statistical correlations that can be mistaken for causality. To think that something happens in everyone exactly at 44 and 60 is ludicrous.