top of page

Quick thought about LC method development

Updated: Jun 29

The draft version of ICH Q2 sees roboustness as not a validation element, but a step of development. Reading that brought back some memories about developing an LC method that became a disaster when implemented. What did we wrong? By the book, nothing. But what we didn't consider, is the real roboustness check.

The companies pay good money for liquid chromatographs to compensate the potential lack of roboustness coming from the method itself: the precision of the pump, the injector etc. are more than enough to keep the method within its intended parameters.

So what can be the bottleneck? What is the only thing the pharma companies have no influence on at all? That's right, it's the column.

The lab gets familiarized with a certain brand, and everything goes well with the tests, until...the newly purchased column produces slightly different retention. The key is the batch of the silicagel they use for the column manufacturing. And sometimes these slightly different retentions can cause huge System Suitability Test problems, getting the lab incapable of generating QC results on the product.

The best solution to circumvene this scenario is simple (but not too short): stress-test your method with multiple column brands. And what kind of test to use? the ultimate stress-test: inject your sample solution at least 50-100 times, and evaluate not only the chromatographic profile, but also the pressure. If you observe only a slight trend in either the resolution, the retention time or the pressure on the column, your method is not roboust - with the chosen column. So choose wisely, because that method can stay with you for years.

4 views0 comments

Recent Posts

See All

PQR, QMR, CPV, VMP/VMR, Computerized systems' periodic reviews, who knows what else...we manufacture these summaries, but exactly what for? Sometimes we spend so much time compiling these documents th

I just received a question that looked easy at first, but the laws and guidelines are surprisingly vague on this topic. The question is: how to choose between retest period and shelf-life for an API (

bottom of page