In theory, the answer to your question is yes, they can be used.  Combinations of such databases is par for the course in the statistical analysis of such trials.

But in practice, it is not so great.  Those earlier trials added up to no more than 300 patient years (if I remember correctly).  Take only the patients with low HDL, diabetes and pre existing CVD issues, and the "useful" patient years are far less than 100 patient years, I am guessing.  Also, there is a "standard of care" issue here.  I don't know how you correct for that.  We had concluded that given the low MACE rates for the entire trial, one of the explanations could be that the standard of care had changed sufficiently enough, that results from earlier trials may no longer be useful.

In short, every bit helps, but I am skeptical that it will move any needle.