Run to run yes. But if you’re looking at 99th percentile I think you can safely assume what ever the best situation for that spec was, that’s what happened.
And if you look at every spec the same way you can safely assume that’s what happened on those runs as well.
If you look at the Dungeon data in this post you don’t see the same type of variation you see in raids where some specs are vastly better for one type of fight and not for another. For the most part the spec that does the most damage does the most damage in every dungeon. As you go down the list for any particular dungeon there would be only minor swaps in the specs position from their average ranking.