Abstract
In the course of process development for phentermine base, samples were observed to fail the UV specification. Analysis using the HPLC purity profile method did not yield unusual results. However, re-analysis using UV-photodiode array detection indicated the presence of an impurity that almost exactly co-elutes with the phentermine peak. Comparison of these samples with previously manufactured lots known to pass the UV test, along with area comparison of the impurity, permitted its determination by GC. Previous efforts had tentatively identified this peak 2,3,3 trimethylindolenine. Further evaluation using GC(EI+)MS confirmed the identity of the peak as 3,4-dihydro-3,3-dimethylisoquinoline, formed via a Bischler-Napieralski type reaction (cyclodehydration of acyl derivatives). The impurity was eliminated from the process by minimizing those factors that contribute to its formation. Assay of the