How did you model the feed line? Did you try to isolate the feedline in real measurement?
- - - Updated - - -
No. I'm talking about the cable effect, as discussed e.g. in this thread
https://www.edaboard.com/threads/129541/
An asymmetrical feed line becomes part of the antenna and changes S11 and radiation diagram completely. This is actually a two-part question, 1. which feed configuration did you simulate, 2. what is the real measurement configuration?
I presume that you performed correct VNA calibration to a reference plane at the antenna input, or apply at least port extension.
Another point would be to check the plausibility of your simulation.
- - - Updated - - -
I only used a waveguide port on CST to model the microstrip line feeding.
I'm not sure which configuration you are describing here. I don't see a microstrip line with your antenna.
In fact the simulation would be designed to fit best the real application. That's not typically the VNA measurement setup, at least it isn't for a portable device antenna.
The VNA measurement needs additional means to avoid wrong results by cable effect, like the ferrite tubes discussed in the linked thread, more have been discussed at Edaboard in other previous threads.