Have you calculated the concentration of your current injections to determine if you are now and were then exceeding the capacity for a column of specific dimensions? Rood estimates a .32 mm diameter column with .5 um film at 125-250 nanograms per compound (peak). If you are overloading your column and have compounds with similar retention times you will have problems with coelution. This can be sneaky, since amino acids require derivatization that may increase mass by a great deal without a concomitant increase in the element you are analyzing.
In particular, a number of papers that look at compound-specific dD measurements advocate a peak height of 1-4V to achieve linearity. However, if those authors computed the amount of sample they injected, they would find they were exceeding the column capacity by a factor of 10. That is fine if you are looking at one or two compounds that aren't near one another but in the case of the amino acids that is a highly non-trivial issue. It's a dynamic trade-off between the parameters that work best for the IRMS and the GC.
The two most obvious solutions are to a) buy a very high capacity column such that an injection at a concentration below capacity yields large enough peaks or b) carefully calibrate peak height against isotope ratio of a standard of known value.