I mean roughly in reference to the underlying mechanism it directly addresses, not all the downstream effects. And even that was, admittedly, sloppy, because there's some complex feedback loops involved. I guess it would be more accurate to say it is a maintenance medicine and not a complete cure, and so stopping taking it unmasks the continuing condition that is treating.
It's not like it's impossible unmedicated. Plenty of people have and will do it. Obviously most people are unable to. I was always surprised just how few prediabetics did though.
Source? Everyone I know who stopped taking it rebounded a bit, but not to where they were. And no literature shows 100% rebound to my knowledge.