Any answer is going to be semantic in nature - your definitions of upstream, tributary, network, etc. are going to dictate the answer.
From a network (continuously connected features) point of view, I'm going to go with there is no difference and you cannot have a water body upstream of another without it being tributary. If you do have one that isn't connected, it's called a sink and it breaks things into separate networks. That situation certainly does happen in nature (all standing water to a degree), so from a physiographic point of view yes you can have a water body that is upstream (topographically higher in the same catchement) but not connected and therefore not tributary.
I typically only view 'tributary' as applied to flowing water and not still bodies. By definition, upstream implies within the same catchement, such that water eventually all flows to that point. Tributary essentially means flows into. I can think of no case where water would be upstream but not flow into and still be part of the network.
A stream could split via natural or manmade means, but everything above the split would then be upstream and tributary to everything downstream along both paths even if water would not naturally take that path. That also assumes that your definition of tributary allows for non-natural flows. Nearby we have diversion structures that bring water from one side of the continental divide to the other, however we do not consider those rivers on the source side to be tributary to those the water flows into on the other side.
In the case of a saddle, as Hornbydd mentioned above, that source node would be upstream to both channels, not just one. But the moment you leave that node you are no longer 'instream' with one of the channels, so you could not say one flowing in the opposite direction is upstream of the other. *Up*stream inherently implies flow and direction, so while that geometric situation can exist, you can't apply the terminology without considering direction of flow.