If you look at the test results they are measuring the combination of lens and body. Hard to do otherwise. Makes comparison between manufactures difficult and less than meaningful.
Yes, and sorry, but I think you are missing the point. For every measurement, all generated with a body attached, the Canon lens comes out on top, in some cases by a significant margin. Yet, the Score is a tie. So...the score is fabricated, pulled from their nether orifices, etc.
If you go over to photozone.de's page, they'll quite clearly state that the measurements (or scores) from one camera system (make + sensor size) cannot be compared with another. Given this I can't see why the same wouldn't be true for DxO. What does that mean? That you cannot compare a score of 25 for the Nikon lens with a score of 25 for the Canon or 22 for the Sony lens.
To be able to compare each lens properly would require each lens being mounted on the same camera.
How do they calculate their lens score? Would love to know but I'm pretty sure that it is corporate secret. For the Nikon one to be so high must mean that the readings are somehow weight on the sensor (e.g pixel size.)