I guess the biggest thing to remember when judging whether a stat is relevant over a certain sample size is just to think about how often that stat is affected.
For example, VP$IP and PFR are affected every hand. Every pre-flop action you make has an impact on those stats. While something like W$@SD is rarely changed. You only see the flop in less than 1/4 of all the hands you play, and only go to showdown in about 1/4 of THOSE hands. So basically you need like 20 times more hands for W$@SD to become statistically relevant than VPIP and PFR.
As to a certain number, I start using basic reads from stats at around 50 hands, and don't make too many big decisions from stats until I have about 200 hands.