As for the titular question, I think that the theoretical answer must be "no," but the practical one might be "maybe" instead.
Theoretically, IQ testing doesn't depend on the person's LEVEL of attainment/achievement, but is adaptive to reflect the developmental level at which the subject presents. So it wouldn't be appropriate to IQ test a high school student using the same tool as a preschooler.
Pragmatically, however, there are not an unlimited series of endlessly adaptive tools to choose from, and therefore, if a tester selects on the basis of an assumption about attainment/achievement (e.g. that a subject can read independently) which turns out to be false, then yes, the scores might be invalid.
In that situation, using a different tool might well lead to higher scores.
That's not an "improvement" in IQ, though-- it's correction of a measurement error.
