I would suggest that the tester figure out not just whether that V > NV split is statistically significant, but also how uncommon it is within the high-ability (GCA>120) population. It's in the norming tables. (I can do it, too, but this is the tester's job and part of what they were being paid for, and it only takes a few minutes.)
Standing on one foot (ie, not looking at the norming tables), I'm going to predict that it is statistically significant but common in the high-ability population and probably clinically irrelevant. We're not all good at everything -- it's normal to have strengths and weaknesses. That's why there *are* subtests.
Without other data (real-world and testing data) to suggest that it's an actual processing deficit, I think it's a huge leap to assume that this is a problem.
