SPIE does require the multi-page abstract as far as I know - in fact, this is one of the reasons some of the groups I worked with don't publish there, because it's a lot of work for a paper that isn't in a formal journal.
The technical answer to your question is that your submission could be withdrawn if you don't submit the abstract, as per their guidelines:
Conditions of Acceptance
By submitting an abstract, I agree to the following conditions:
An author or coauthor (including keynote, invited, oral, and poster presenters) will:
• Register for the conefrence & pay the fee.
• Attend the meeting.
• Make the presentation as scheduled in the program.
• Submit a manuscript (6 pages minimum for LASE & OPTO; 4 pages minimum for BiOS; 20 pages maximum) for publication in the Proceedings of SPIE in the SPIE Digital Library.
However, your main question as to what really happens if you don't submit the abstract? Here is my guess:
I have tried to access various presented papers on the SPIE online proceedings after the conference, only to find them missing the full 6-page abstract (only having the 1-paragraph abstract). This implies that they do not retract/adversely alter your submission's status if you fail to turn in this full six page abstract. I don't have any better evidence than this though. (These were not invited papers either, but usually papers from companies who may not have wanted to divulge too many details in a full 6-page abstract.) It's possible that they submitted the full paper after the conference and I just didn't check again later.
Also, if you've already published the work in another high-impact journal, then I'd have thought it would be pretty easy for you to put the 6-page abstract together, and it won't impact your future publication plans too much. You can include new material in your presentation but omit this from the 6-page paper to prevent jeopardizing future publication plans.
Lastly, in case you do want to include new material in the paper, sometimes journals will accept papers that have previously been submitted to conferences, with the understanding that much of the work has been published before but at lower quality and usually lower standards than the Journal. One often does more rigorous analysis, includes additional details and work (measurements or simulations) for the subsequent journal pub. I often find conference papers with accompanying higher-quality (often longer) journal papers by the same groups, on the same subject, but with additional info/rigor in the Journal paper.
I spoke to an SPIE rep at the conference, and he said that you can indeed submit "anything you want" (my words) in the submission, and that even if the paper was too short they wouldn't penalize you in any way.
He did say that if you did that all the time they may become concerned, but in general the page numbers were a "guideline", not something to have your presentation rejected over.