Yeah, so basically conceptually I just look at it like a spectrum of pancreas cases, whereas if it was tail of pancreas you would include splenic hilum, head you include porta hepatis/duod/peripancreatic LNs, and if it is GB or cholangio, you include porta and medially the adjacent liver for the tumor bed. Essentially the rest of it stays the same, and you can include less of the peripancreatic stuff, but still use the contralateral aspect of vertebral body as a guide for that, and then of course celiac/PA nodes down to L1 or L2 or so. I don't think there are enough cases to have a well defined atlas, but I think this is a reasonable way to approach.
I think the traditional pancreas fields are too big, b/c it wasn't that we were having marginal failures, we just don't have the dose to control in-field disease. Let me see if I can put up a screenshot.