Second order behaviour of the tail of a subordinated probability distribution

Let G = [Sigma][infinity]n=0pnF*n denote the probability measure subordinate to F with subordinator {Pn}. We investigate the asymptotic behaviour of (1 - G(x))-([Sigma] npn)(1 - F(x)) as x --> [infinity] if 1 - F is regularly varying with index [varrho], 0