I'm going to use this example again:
When the 70-200 f/2.8L II IS lens came to the market and was tested, it got a lower score than the version I lens. Later, DxO mark used a different CAMERA to test them, then the v2 finally scored higher.
Well, their Lens Scores are an even larger, stinkier pile of steaming BS than their Sensor Scores. Even the name itself is intentionally misleading, since the primary determinants of the Lens Score are the T-stop of the lens and the low light performance of the camera body on which the lens is tested.
Last time I checked DxOMark, the very best lens that Canon makes was the EF 100mm f/2. Yes, the famous $499 100mm f/2 known by professional photographers all over the world as the very best lens that Canon makes ... not. That lens is from 1991.
Now that honor has gone to the EF 35mm f/2 IS. Yes, the $599 lens is better than ANY other lens that Canon makes ... according to DxOMark ... and no one else.
P.S. Nothing against either of those lenses (both excellent), but giving them the highest scores of all lenses in the entire Canon EF system is pretty much proof that the DxOMark scoring is faulty.