Here’s a recent paper illustrating an impurity problem that I’ll bet none of us out there have experienced! It describes process work on an antifungal molecule, efinaconazole, which is now available as a generic drug. The authors were checking the purity of various batches of the final product and kept noticing a small peak shouldering the desired one under LC conditions. They estimated it at about 0.06% of the samples (it showed up every time!) and eluted just a bit faster than the real drug.
This of course needed to be tracked down. The standard is generally 0.15% for a known impurity and 0.10% for unknown ones. And this peak was getting too close to that cutoff. Chemical means of lowering its presence did not change the amounts at all. There was no change in the amounts seen in crystals versus the “mother liquor” in crystallization experiments, which is rather odd.
Analytically, it appeared that the impurity was one Dalton heavier than the parent, which is a real puzzler. More mass spec work narrowed down the increased weight to the piperidine region, but there was no evidence that the piperidine starting material had any M+1 impurity. This all points to the problem being outside of starting material impurities and outside of impurities introduced during the synthesis process.
High-resolution (Orbitrap) mass spec data revealed the mass matched exactly one deuterium swapped in for a single H down in the piperidine ring. A large-scale HPLC purification on 100g of efinaconazole produced 6mg of “pure impurity”, and NMR spectra confirmed the deuterium assignment. The natural abundance of deuterium is 0.015%, and there are four possible substitutions next to the nitrogen, giving your 0.06% impurity, revealed by its very slightly different physical behavior. Yikes!
Source: Hacker News · Summarized by HeadlinesBriefing