Yeah, it looks like the only way your going to figure this out is to examine the response, byte by byte, as seen by the client (web browser). I'd be saving the Byte stream, converted to text (toString), then use WinDIff to examine the differences between
the responses of both pages.
From memory, Acrobat implements web browser integration as a
[Browser Helper Object] (BHO). But I'm not sure that sending the PDF to the browser via Content-Disposition and ContentType will cause the browser to use the Acrobat BHO. I suspect Internet Explorer will be using ShellExecute instead, like from Windows
(file) Explorer. That'll also mean that there will likely be security issues, as well (Trusted sites list, etc).
As for differences between Acrobat 5 and 7, I'm not sure. But I do know that when you change the "Browser integration" option in v7, it kicks off an instance of Installer to add or remove the Acrobat BHO. In v5, that was not the case, and the only way I used
to remove the Acrobat BHO was to use SpyBot's Advanced tools, which has a tool to do this - it's been a while, though.