For a same drainage process, why ECO2N with hysteresis modeling and ECO2N without hysteresis modeling have very different results?

I studied the reservoir response to different injection processes. To just test the simulators' capability, I used TOUGH2-ECO2N with the hysteretic capillary pressure and relative permeability functions and without these functions to simulate the same continuous injection process, respectively. All the parameters were set to the same except the parameters for wetting branches. But to my surprise, the supposed-to-be identical results were very different. The pressure difference can be as high as 8 MPa, and the distances of CO2 migration are different from each other by 100 meters. I can't understand why there are such big differences.  What should I do the find out the problem? Thanks.