What is it that's soldered onto the board and could that be causing your leakage? Is the PCB trace large in area, directly opposite ground plane, does it get very close to other traces etc? Is there a solder resist?
If I had a circuit with resistances in the ~10M range, I would anticipate solder flux to be the cause of occasional problems, not so much below that, and above that I would go to special lengths to try and control the problem. But ofcourse, its all very geometry dependent, especially if there's anything fine-pitch involved, or its high layer count.
I would expect a slightly more consistent measurement (assuming you are talking about measuring the same node multiple time on the same board etc) If you're getting inconsistent results measuring ~100nA, make sure all your cables are kept separate from one another and absolutely was far as possible from anything at an appreciable voltage. Could the SMU be set in a range not quite suited to such a low current? Make sure there's nothing light sensitive on the circuit also. Its well worth repeating measurements and checking and different voltages to help work out where your inconsistency is coming from.