Depending on how heavy a retoucher you are, replacing the background layer (or updating a Smart Object) is impractical. For instance, if you have pixel layer edits (merged layers, sharpening, composites, cloning, etc) above you're dead.
If I want to make LR edits on top of a TIF, I make a Virtual Copy of the TIF, edit that, then render it as a TIF (Edit Copy with LR Adj) and drag that layer on TOP of my original layered TIF. Name the Layer "VC Adj" or some such thing.
This way all you have to do is click the top layer on/off to return to the previous state, and you have all your masks at hand to reuse if you need to make additional PS adjustments on top of that.
That's the PS end of things. What I haven't figured out yet is whether to save the VC and how to make sense of all the VC plus TIF iterations in LR. I end up with multiples of huge PS files, redundant VCs, etc.
One might ask why bother even rendering the VC as a TIF, or bringing it into your master document. The reason is to have a LR exit strategy. Don't make yourself beholden to any single brand.