I haven’t tried that, so I might be off here.
Any VR rendering is based on the concept that the viewer is on a central position of the projection, if that is spherical or cube based. Camera=Viewer—>Rendering
This central position can’t be changed without a complete distortion of the viewing experience. Like sitting in the center of a globe and watching the earth map. If one moves now to the edge of that globe things are at least out of scale in both direction (larger and smaller). Globe or Cube is interchangeable here in terms of middle point—Viewer/Camera/Rendering
With being in the middle point of the projection enables the size of the globe or cube to be variable. (Similar to the problems motion track a 3D motion with a tripod based camera. The z information is irrelevant and not in existence for the camera.)
All objects that you “pinpoint” via External Compositing Tag, in X,Y,Z;H,P,B, plus scale (Matrix) will be very much defined in space, not variable at all. Those will not move with the projection, in fact they are absolute defined in space and eventually with their animation. If you even rotate the camera, the projection will rotate as an result, and will go out of sync with the information of the External Compositing tag.
How it works: the position and orientation of the camera must be connected to the Spherical projection or Cube projection, so all parts stay synchronized. Any change of that camera must change the resulting projection in the same way. In Ae it is easy to connect those movement. It is easily reconfigurable, with a rendered map, placed like a cube side, and with the external compositing tag. Then set the position of the cube to the camera position that all cube sides are as a cube in C4D around the camera. (easily done with a Cloner, a cube as Clone Matrix (Polygon Center), a clone/child and an External Compositing Tag set to Include Child. I have tested it an it works. If the cube (projection is not connected to the animated camera, it moves out of the cube and end of story. Since that is a pest to do in Ae with all the unit to pixel etc changes, 3rd party options do all of that work via script/plug in constructions.
These 3rd party options to use these spherical or cube maps in Ae (mostly convert those), you have to check which one allows for that, OR in case of a cube projection this needs to be set up with the 3D connection to the camera. Some have not an option, so you need to set up your own conversion in C4D, which is easily set up, just ask me in the Q&A forum and I answer it. Over time I have developed for my photography a collection of converters based on Cinema 4D. No external converter needed. C4D is my image machine, it is pretty much above most options anyway!
Set up the cube and the External Comp Tag will be visible, even if it is outside of the cube; Perhaps scale down the “frame” to see its precision.
The simplest version would be that the camera for the VR in C4D isn’t moving (nor rotating) at all and sticks to the 0,0,0 point with no exception. Then the External compositing information should work in any scale even. I have requested more information from one of the 3rd party plug in developers. I will update my post here if I get something workable. (Update, I tried the one, but I had only three clicks and the trial was exhausted. I wait for the reset of that 3 click trial)
All the best
Edit: Have a look here:
The measurements are quite different, and to place anything from one space into another, things need to be in sync.