At the Journal Club for Condensed Matter Chandra Varma has a helpful commentary on
recent experimental papers reporting evidence that the mixed valence compound SmB6 is a topological insulator (see my earlier post and a recent Nature News article).
A few things I learnt from Varma's commentary.
The strongly correlated properties are not central to the topological properties. These are rather a property of the effective band structure which arises in a slave boson treatment or the Varma-Yafet variational wavefunction.
The Fu-Kane conditions for a topological insulator most likely hold in the mixed valence limit which is at the extreme of particle-hole asymmetry. In this limit the Kondo temperature is of the order of the hybridisation energy, in contrast to the Kondo limit when it is an order of magnitude smaller.
Some caution is in order because the existence of actual surface states [e.g. from ARPES] have not yet been definitively established.
Definitive signatures of the strongly correlated state might be seen in new low energy resonances that could arise from non-magnetic impurity substitution.
p.s. On his website Piers Coleman has a nice talk giving the background theory which preceded the experiments.