We had a thread about this back shortly after I had first joined this forum. Roughly 15 to 30% of random mutations in proteins are stability increasing, with the remaining 70-85% being destabilizing. See:
Needless to say, one in seven to one in three aren’t bad odds at all. For hyperthermophile organisms, the percentage of mutations that are stabilizing on protein structure is on the lower end (~10%), as you would expect since they have already evolved for a long time under selection for high thermostability.
Now it has to be said, and this is extremely important to understand, that protein stability is not directly proportional to protein function. A protein can become both too stable and too unstable to function properly. A degradation/reduction in protein stability thus cannot automatically be said to constitute a degradation in protein functional capacity. So even if polar bear APOB mutations can be identified which is predicted by some algorithm to result in a reduction in protein stability, that does not imply the protein is suffering a degradation in it’s functional capacity.