ADA4255 Current Noise Characteristics at Higher Frequencies

The current noise specification is only listed to 10Hz in the ADA4255 datasheet. I have been testing a EVK and have some questions about the behavior. 

Setup 1: 

I have an AD7768 EVK connected to the output of the ADA4255 EVK. The AD7768 is sampling at 250ksps and is using the internal wideband brick wall anti-aliasing filter which limits the input bandwidth to roughly 110kHz. When I take a sample, the AD7768 software collects 2^17 samples for plenty of spectral resolution. I have both inputs of the ADA4255 EVK shorted to ground so that I can take a noise measurement. I can set the gain of the ADA4255 using the ACE software. I am using both EVKs as is without any modification. The results I see are exactly as expected from my understanding of the specifications. The spectrum looks nice and flat, and the RMS value reported by the AD7768 Software is 82uVrms at a gain of 1, and 728uVrms at a gain of 128, which is expected at this bandwidth based on the amplifier voltage noise specifications.

                           Unmodified ADA4255 EVK at a gain of 1                                                                                                       Unmodified ADA4255 EVK at a gain of 128

Setup 2:

The setup is similar to above, with the only difference being I replaced the 0 Ohm input resistors on the ADA4255 EVK with 100k Ohm resistors. I want to verify the impact of Johnson noise and current noise of the amplifier. Based on my calculations, the addition of these resistors should add 57.4nV/root(Hz) of Johnson noise, and 20nV/root(Hz) of input current noise from the amplifier specification of 100fA/root(Hz) at 10Hz. This should result in values around 86uVrms at a gain of 1, and 2.7mVrms at a gain of 128. However, the results of this test are not what I was expecting. Now the noise spectrum is no longer flat, but has some frequency dependent behavior as is shown in the plot with a gain of 128 below. The RMS value reported by the evaluation software at an amplifier gain of 128 was 20mVrms, almost 8 times higher than expected. The RMS value recorded at a gain of 1 didn't change significantly from Test Setup 1.

                                 Modified ADA4255 EVK at a gain of 1                                                                                    Modified ADA4255 EVK at a gain of 128

My question is what is causing this extra noise? Could it be input current noise of the amplifier? Since the datasheet only reports input current noise to 10Hz, I suppose the input current noise spec could get worse at around 1kHz which is where I see the noise floor rise in the above plot. What is the current noise performance out to 100kHz? Is low input impedance critical to the noise performance of the ADA4255? 

Thank you for the assistance!

Alex