The short answer is that blood withdrawals can't be masked. But if they're done during the season, they generally are followed immediately by a transfusion, so there is no change in parameters. The withdrawal/transfusion cycle is used in effect to store blood beyond the several week limit that it is stable in the refrigerator. At any one time, the rider always has some freshly withdrawn blood stored and ready for a transfusion.
The only time a withdrawal would not be followed by transfusion is to obtain the original stored blood. This problem can be minimized by starting slowly, just withdrawing a 100 ml or so, then upping it to 200 ml and transfusing back the 100, and so on. If done incrementally, the change in parameters is too small to detect. It also has less effect on training, which is an issue with large withdrawals. Moreover, the process is usually begun in the off-season, when testing is less frequent.
A second problem is that when the rider uses the stored blood for actual performance enhancement, the transfusion is not preceded by withdrawal. Thus there is no longer any stored blood; the cycle is broken. This often limits the rider to PE transfusion for only a single major event, or very few such events. This is why freezing cells is such a big advantage. One can accumulate a large amount of stored blood, enough for several transfusions during the season. This also means a lot of withdrawal during the off-season to get all this blood. But red blood cells can be stored frozen for years, so it's possible to accumulate this blood over the course of more than one off-season.