Aiken et al.'s work on focus.
Gulwani's PLDI09 control-flow transformation. Rybalchenko's Path Invariants paper?
Reps undecidability paper. Shows undecidability of flow-sensitive, context-sensitive, field-sensitive points-to analysis with no dynamic memory (I think) and with at least 2 fields (car and cdr in this case). I'm fairly confident that extending this undecidability result to the flow-insensitive case should be straightforward, as the proof doesn't rely on reasoning about strong updates of heap locations. However, going to a single field from 2 fields would require a significantly different proof, since the reduction from PCP relies on having an alphabet with at least two letters.
Chakavarthy undecidability paper. Shows undecidability of flow-sensitive single-field points-to analysis with dynamic memory within a single method. Leaves open the question of decidability of the flow-insensitive version of the problem. At some point, I convinced myself that this flow-insensitive problem (single field, dynamic memory) was somehow equivalent to context-sensitive single field with no dynamic memory, but I can't fully remember the argument. As far as adapting the proof style used here to flow-insensitive analysis (i.e., reduction of Hilbert's tenth problem), the issue is that the reduction for the flow-sensitive case uses destructive updates to heap locations to check must aliasing and perform "counting" (see p. 123 of linked PDF). Reasoning about such destructive updates requires flow sensitivity, since the analysis must reason about the state of the heap location before and after the updates.
One other question: is single-method flow-sensitive single-field points-to analysis with dynamic memory decidable if the program is functional, i.e., each heap location is assigned only once? What about the flow-insensitive version? Here I'm not sure what havoc loops can still wreak.
Note that even with no dynamic memory, we still end up with an infinite lattice in the case of context-sensitive analysis of recursive methods.
Also see this paper: On the complexity of partially-flow-sensitive alias analysis