Typically in medicinal chemistry research, we are looking for tool compounds (drugs) that will affect only one enzyme. Then you know that whatever effect you had on the biological system by administering the tool compound was caused by the interaction of that tool compound with the one enzyme it could interact with. If the tool compound causes the desired effect, you know that your hypothesis about the activity of the enzyme was correct, and you can modulate the activity with the tool compound. If the tool compound causes an undesired effect, you will know that the undesired effect was modulated by the enzyme the tool compound targeted, and will be a side effect of any administration of the tool compound.
It gets much messier when your tool compounds hit interact with more than one enzyme. It might cause a desired effect by interacting with one enzyme, and an undesired effect by interacting with another, or it may show no activity at all because the desired effect at one enzyme was counteracted by an undesired effect at another.
In short, if you do not believe the GPCRs are associated with your disease state, you do not want your drug to interact with them at all.