Having just gone through this process myself I will try and offer a little help that worked for us.
Go through your Vectorworks file and figure out where you want the pivot point to be. (in our case I created a simple box symbol symbol, and made it a lighting instrument) This allowed me to leverage a worksheet to make a nice list of all my pivot points that I could print out. (pivot point name for organization, and then X,Y and Z position - in inches based on document units)
Exported that list to Excel. Reason we did this was so we could swap Y and Z positions before entering them into Vision (as they have different coordinate systems) and that always gets confusing to remember.
Then in Vision we created a New layer in the Scene Graph for each rotation transform we wanted. For each new layer, we named them accordingly (ie. Truss 1 Transform Rotate) and adjust the X,Y and Z position of the layer to match with the pivot point we figured out from Vectorworks.(this is where having the pivot point figured out already sped things up, kinda)
For each transform we exported a .mvr that only included the truss, objects and lights needed for that transform. You merge each .mvr in to your file one at a time. After each import you can drag the .mvr layer "into" the associated layer you created to nest it inside the transform layer.
Adjust the DMX transform data as needed in your high level transform layer to get the rotations as you need them, and the nested layers follow suit.
We found it helpful to export our Truss in a vertical orientation from Vectorworks using the same pivot point we wanted. That way we could default the DMX transform channel in the console to 50%. This allowed us to rotate the truss both positive and negative directions without needing to nest yet another dummy layer for transforms.
I hope this helps.
I hope that there can be work done in this area to make this process easier. The future of integrated previz is already here, it would be nice to have a better workflow.