We had some bending ply core baseboard delaminate a couple years ago on the concave side of of a curved wall. The convex side was fine.
We had a double layer of 1/8" thick white oak veneer applied to the face of a double layer of 3/8" bending ply. As the oak shrunk with moisture change it remained stable on the convex side because shrinkage just pulled it tighter to the curve. On the concave side though shrinkage allowed the oak to move away from the wall surface thus putting pressure on the bending ply layer structure. The grain direction was going length wise down the wall so that the only shrinkage involved was over the 4" width of the material. In effect the oak was able to curl away from the wall.
One could consider that the same was taking place here. If the wood had been glued direct to the MDF substrate there would have been no way for it to move. My guess is the paper backer was the weak link much as our bending ply. It acted as an weak barrier between veneer and MDF allowing for the veneer to surface to move and fail.
Think of the paper barrier's strength in terms of paper being used in a joint where the surfaces are going to have to be separated later..... like when mounting blanks for turning on a lathe. Paper is put between the two because it will split on center when the two wood parts are split apart with a chisel.
Even in a climate controlled space there will be some moisture variation over a 5 year period. Perhaps a different veneer wood species would have been more forgiving in this situation. I suspect the paper allowed for enough wood movement to stress the grain structure, thus resulting in surface failure.
BH Davis