Reliability of simulated robustness testing in fast liquid chromatography, using state-of-the-art column technology, instrumentation and modelling software

• Published in: Journal of pharmaceutical and biomedical analysis Vol. 89; pp. 67 - 75
• Main Authors: Kormány, Róbert, Fekete, Jenő, Guillarme, Davy, Fekete, Szabolcs
• Format: Journal Article
• Language: English
• Published: England 01.02.2014
• Subjects: Amlodipine - chemistry; Chromatography, Liquid - instrumentation; Chromatography, Liquid - methods; Models, Chemical; Reproducibility of Results; Software; Column interchangeability; DryLab; Robustness; Modelling software; UHPLC; Method development
• ISSN: 0731-7085; 1873-264X
• DOI: 10.1016/j.jpba.2013.10.029

The goal of this study was to evaluate the accuracy of simulated robustness testing using commercial modelling software (DryLab) and state-of-the-art stationary phases. For this purpose, a mixture of amlodipine and its seven related impurities was analyzed on short narrow bore columns (50×2.1mm, packed with sub-2μm particles) providing short analysis times. The performance of commercial modelling software for robustness testing was systematically compared to experimental measurements and DoE based predictions. We have demonstrated that the reliability of predictions was good, since the predicted retention times and resolutions were in good agreement with the experimental ones at the edges of the design space. In average, the retention time relative errors were <1.0%, while the predicted critical resolution errors were comprised between 6.9 and 17.2%. Because the simulated robustness testing requires significantly less experimental work than the DoE based predictions, we think that robustness could now be investigated in the early stage of method development. Moreover, the column interchangeability, which is also an important part of robustness testing, was investigated considering five different C8 and C18 columns packed with sub-2μm particles. Again, thanks to modelling software, we proved that the separation was feasible on all columns within the same analysis time (less than 4min), by proper adjustments of variables.