Not knowing what technology (and probably not using
it anyhow) I'd suggest first looking into the STI mask
fab Booleans, which should be described somewhere in
your PDK docs. Ask the foundry if you don't find this.
It will tell you everything that goes into the STI mask
and one of these may be a "block" layer.
A question is, what ends up between the edge of
active and the new location of STI? And what does
it do? Since this is abnormal to the foundry, you may
get zero help in predicting and no prior data.
If you're talking about "unusual environments" as
your VT shift concern implies, I'd be concerned about
proving that your edge leakage is any better with
STI pushed out away from active - you would still
have some thick oxide over the device edge, to
cause similar trouble.
If you're just talking strain effects, and these are
making devices unstable or unrepeatable, maybe you
want a better foundry that has these in hand.