Actually I am using SEMCAD X, but as I understand, making the same numerical computation with two different software any not give the same results. I am afraid you are not defining your boundary conditions the same in the two software. Here, I mean the Absorbing Boundary Conditions (ABC). The ABC are set as a uniaxial perfectly matched layer (UPML) mode with different strength thickness.
You didn't give me the setup frequency you used in HFSS. It will be the source of problem because you have in both figures closer pics with S11 values different.
If you are interested only for the 77-81 Ghz band make the simulation only in this band . but to compare between two simulation, it's recommended to do it in the same configuration conditions.
I insist to do a small step like 10 or 20 Mhz and if you have strong machine do the sweep type = Discrete and not fast
I think the issue will be related to the mishing and the mish size you use in both programs as you know that both programs are using different numerical methods, i.e. FEM and FIT.
I advise you this time to change the frequency setup and make it equal to 77.5 GHz or exactly the desired value corresponding to the peak you have in CST simulation corresponding to -45 dB and make the frequency in exact value expressed in MHz (it's more accurate than the Ghz)
In my experience with HFSS you want to set your solution frequency to the highest frequency of evaluation. So in your case 90 GHz (or thereabouts) appears to be your highest evaluated frequency, rerun your HFSS model but with 90 GHz as your solution frequency. Let me know if you are able to resolve your problem. Also, more details about each simulation would be helpful!