From time to time I have done image recovery work for forensic purposes. In cases where high ISO is used or details have to be recovered from very dark areas I have run into this type of problem many times. There is nothing unusual or unexpected with these images.
It looks like LR has attempted to compensate for vignetting (I'm guessing the lens profile option is set to auto). With high ISO, corrections for vignetting will result in edges looking milky. This is because dynamic range is so low at these ISO levels. Normally the difference in DR between the center of an image and the boosted edges is unnoticeable, but with very high ISO the difference becomes significant. I have found that in such cases you have to correct for vignetting by boosting exposure at the edges (the normal fix) plus you have to raise the contrast at the edges. In any case, vignetting is much harder to fix at high ISOs, and the difference at the edges is very visible even when corrected. Essentially, with vignetting correction, the edges are boosted an additional 1-3 stops. That's the equivalent to the edges being exposed at over ISO 200K, give or take, in your examples. DR is less than 5 EV at ISO 50K, at 200K the range is probably around 2 EV - which isn't much DR to work with. Shot and amp noise will be much more visible at the edges because actual image data is so restricted. Also, distortions in color and reduction of DR will be twice as bad for reds and blues because there are half as many red and blue pixels as green ones in the sensor.