flash12 Don't care (X) refers to the set of input for which the output of the function does not matter or a set of inputs that is known to never occur. Now, that being said, technically it leaves a function non-deterministic as we do not know what output the function will give on these set of inputs. However, when a circuit is realized it can only be deterministic; meaning you're in a way forced to define the output for the function even at those positions for which the input is known not to occur.
Now, we do something clever, in order to minimize our circuitry we assume the values of these don't cares in such a way that the overall expression for the function has minimum literals and minterms and thus a smaller circuit.
Coming back to the question, if you closely look, you'll see that the octet formed for deriving term A in the answers just takes all those don't cares to be 1. This does not mean you have changed the function. The function is known to never take those values and we can thus play with them for our benefit. Just draw K-Map and you'll see.
Please note, this a very intuitive yet useful way of thinking about don't cares. I suggest that you read more about them from standard text