Dear Florin,
Absolutely not! I mean it's not related to SU(2,2). And I don't see how it could be. Therefore, if you discover a connection I will be both amazed and pleased. Pleased because it would open up unexpected relationships between objects I think are unrelated.
SU(2,2) is the twistor branch on the SO(2,4) trunk. I informally call it the geometric branch. Since the Lie algebras so(2,4) and su(2,2) are isomorphic, twistors are related to the conformal group, which does not preserve the Minkowski metric.
Quantions are the algebraic branch. They are derived from so(2,4) by way of internal complexification. Their metric norm is Minkowskian.
You see now why I don't believe quantions and twistors will ever meet again, even though they are siblings -- their mother being so(2,4). Have a peek at Figure 26.1 on page 550 of my 2005 book.
As for the proof, it is not based on some "clever" new idea that could be easily conveyed. The idea of the proof is exactly the same as in the book. The difference is that I longer work with matrices. The formalism is different. Many things just 'fall into place' once that formalism is developed -- which I am doing these days, making sure that there is no mistake. As an analogy, consider 19th century differential geometry, where we (humans) had to do all the work, and compare it with tensor analysis, where the formalism does 99% of the work for us.
The secret to being good at mathematics is not smarts (there is plenty of that around) but laziness. The lazier you are, the harder you'll work at developing the tools that will do the work for you. The funny thing is that those tools end up BEING the next mathematics.
Incidentally, as part of the finalization effort, I think I also found a much simpler way of re-doing Part I of the four-part paper in Fizika, where composability is put to work. After I've gone through the details, making sure that I am not deluding myself, I'll be happy to send you the first draft.
Regards, Emile
PS: Why not drop the stuffy "Dr. Grgin"? On this forum we are all on first name basis, which is much nicer.