I'm using use a videogame as an example for this post, but there are a lot of other examples available.
The engine has been upgraded to support 2x polygons than the original, the mechanics are more complex (but still similar), but more important, the game has a lot of content. You can find dozens of terminals with notes, logs and messages that just add lore to the game. Hundreds of conversations (with multiple paths), hundreds of npcs, world-shifting decisions (for example, you can blow a bomb and take out a whole town for the rest of the game, and the story keeps being logical after that)...
If they decided to rebuild from scratch a new engine, we would probably haven't seen the game until a year or two later. By using an existing, improved one, they have clearly focused on making the content as rich as possible.
Sometimes the best solution is to create something new. But sometimes the best approach is to take something that really works, and use it (using does not mean not being able to make it better).
Note: The game engine is not free of flaws. For example, on pc there are some problems with graphic drivers that I myself have suffered when downloading and installing the latest NVidia drivers (at least it was all fixed fast). But the overall is a better game that it would have been if developed with the same resources and time constraints with a fully new engine.