The obvious sensible precaution for testing your eventual plan is to shoot and develop another roll, or half a roll etc, in the same way and then to do the test on the non-important roll - before applying the results to the original. The only thing I tried was Selenium, for a roll with accidental two stops under exposure and 'tired' developer. There was a slight increase in density which made it a little easier to print, but if there is almost no silver on the neg (in the shadows in my case) then not much can help. Good luck!