It's not clear from the question what the biological source of the maltose binding protein (MBP) is. As you may know E. coli MBP is used as a fusion partner in an expression-purification system marketed by New England Biolabs (this becomes relevant below).
My initial thought upon reading your question was to suggest dialysis. However I found some hints online that this might be problematical and eventually found this on a FAQ at the New England Biolabs site. Looks like this may be your answer.
Question: In order to rebind MBP to the column, the maltose must be removed. Can this be done by dialysis?
Dialysis does not work very well to remove maltose from maltose-binding protein. This is a general phenomenon of binding protein/ligand interactions; after the free ligand is gone, ligand that is released from the binding site usually finds another binding site before it encounters the dialysis membrane (28). We have determined empirically that binding the fusion to a chromatography resin and then washing away the maltose is much more effective. We prefer standard chromatography (e.g. DEAE) as the separation step, since it can separate the Factor Xa and MBP from the protein of interest. In case MBP co-elutes with the protein of interest, we include a large volume washing step to remove the maltose before starting the salt gradient. This way, the mixture can be run over an amylose column afterward if necessary.
So, in summary they are suggesting that you bind the protein to an ion exchange column; wash with lots of buffer; then elute the cleaned-up protein (using salt or a change of pH - you will have to know something about your protein's behaviour on the column to do this).