A 14 Parameter Study of UHPLC’s for Method Development Transfer and Troubleshooting

BART: A transferable liquid chromatography retention time library for bile acids

Identification of unknown bile acids, especially the distinguishment between isomers, requires retention times of a large number of reference standards, which are often not commercially available. Meanwhile, published retention information cannot be directly transferred across labs due to the differences between liquid chromatography (LC) systems, such as different extra column volume and dwell volume. To improve this situation, a transferrable retention time library for bile acids named BART was developed. BART was composed of isocratic retention models of 272 bile acids and a software tool to predict their gradient retention times on various LC systems. The isocratic retention times of bile acids were acquired on a Waters BEH C18 column with mobile phases of acidic ammonium acetate buffer and acetonitrile, and fit to the quadratic solvent strength model (QSSM). Segmented linear gradient retention times were calculated with holdup time (t0), dwell time (tD) and actual gradient profile corrected using 21 bile acid calibration standards. In addition to the reference system where the isocratic retention times were acquired, this approach has been validated on four other LC-MS systems in four labs with two gradient methods. Average root mean square errors (RMSE) between predicted and experimental retention times were 0.052 and 0.054 min for the two gradients tested, which were 9-fold more accurate than referring to a static retention time library. The library is freely available at https://bafinder.github.io/.

In Silico Method Development of Achiral and Chiral Tandem Column Reversed-phase Liquid Chromatography for Multicomponent Pharmaceutical Mixtures

Tandem column liquid chromatography (LC) is a convenient, cost-effective approach to resolve multicomponent mixtures by serially coupling columns on readily available one-dimensional separation systems without specialized user training. Yet, adoption of this technique remains limited, mainly due to the difficulty in identifying optimal selectivity out of many possible tandem column combinations. At this point, method development and optimization require laborious "hit-or-miss" experimentation and "blind" screening when investigating different column selectivity without standard analytes. As a result, many chromatography practitioners end up combining two columns of similar selectivity, limiting the scope and potential of tandem column LC as a mainstay for industrial applications. To circumvent this challenge, we herein introduce a straightforward in silico multifactorial approach as a framework to expediently map the separation landscape across multiple tandem columns (achiral and chiral) and eluent combinations (isocratic and gradient elution) under reversed-phase LC conditions. Retention models were built using commercially available LC simulator software showcasing less than 2% difference between experimental and simulated retention times for analytes of interest in multicomponent pharmaceutical mixtures (e.g., metabolites and cyclic peptides).

Charged aerosol detection in early and late-stage pharmaceutical development: selection of regressionmodels at optimum power function value

In recent years, the use of quantitative liquid chromatography (LC) coupled charged aerosol detection (CAD) for poor UV absorbing analytes in multicomponent mixtures has grown exponentially across academic and industrial sectors. The ballpark of previous LC-CAD reports is focused on practical applications, as well as optimization of critical parameters such as: response dependencies on temperature, nebulization process, analyte volatility, and mobile-phase composition. However, straightforward approaches to deal with the characteristic nonlinear response of CAD still scarce. A highly overlooked parameter is the power function value (PFV), whose optimization enables a detection signal that is more linear with higher signal-to-noise ratio (S/N) and lower relative standard deviation (RSD) of area counts. Herein, a systematic investigation of different regression models (log-log, first-and second-degree polynomial) by both interpolation and extrapolation process in conjunction with PFV optimization throughout the development of LC-CAD assays is reported. The accuracy of the results via interpolation is always good (< 5%) when operating in the vicinity of the optimum PFV regardless the regression model choice. On the contrary, extrapolation process only worked when applying log-log regression at the optimum PFV (accuracy <5%). This outcome indicates that a first-order regression via interpolation can be a safe and simple choice for quantitative LC-CAD in highly regulated laboratories (GLP, GMP, etc.). Whereas a straightforward extrapolation combined with log-log regression can enable the deployment of high-throughput LC-CAD assays, especially but not limited to laboratories where the synthetic process route is undergoing rapid change and optimization (medicinal chemistry, discovery, biocatalysis, process chemistry, etc.). This approach is crucial in developing quantitative LC-CAD assays for poor UV absorbing pharmaceuticals that are sensitive, precise, accurate and robust across early and late-stage pharmaceutical development.

Reducing the influence of geometry-induced gradient deformation in liquid chromatographic retention modelling

Rapid optimization of gradient liquid chromatographic (LC) separations often utilizes analyte retention modelling to predict retention times as function of eluent composition. However, due to the dwell volume and technical imperfections, the actual gradient may deviate from the set gradient in a fashion unique to the employed instrument. This makes accurate retention modelling for gradient LC challenging, in particular when very fast separations are pursued. Although gradient deformation has been addressed in method-transfer situations, it is rarely taken into account when reporting analyte retention parameters obtained from gradient LC data, hampering the comparison of data from various sources. In this study, a response-function-based algorithm was developed to determine analyte retention parameters corrected for geometry-induced deformations by specific LC instruments. Out of a number of mathematical distributions investigated as response-functions, the so-called "stable function" was found to describe the formed gradient most accurately. The four parameters describing the model resemble the statistical moments of the distribution and are related to chromatographic parameters, such as dwell volume and flow rate. The instrument-specific response function can then be used to predict the actual shape of any other gradient programmed on that instrument. To incorporate the predicted gradient in the retention modelling of the analytes, the model was extended to facilitate an unlimited number of linear gradient steps to solve the equations numerically. The significance and impact of distinct gradient deformation for fast gradients was demonstrated using three different LC instruments. As a proof of principle, the algorithm and retention parameters obtained on a specific instrument were used to predict the retention times on different instruments. The relative error in the predicted retention times went down from an average of 9.8% and 12.2% on the two other instruments when using only a dwell-volume correction to 2.1% and 6.5%, respectively, when using the proposed algorithm. The corrected retention parameters are less dependent on geometry-induced instrument effects.

Revealing the inner workings of the power function algorithm in Charged Aerosol Detection: A simple and effective approach to optimizing power function value for quantitative analysis

In recent years, charged aerosol detection (CAD) has become a valuable tool for fast and efficient quantitative chromatographic analysis of drug substances with weak UV absorption. In analytical method development using CAD, the power function settings available in the instrument software are key for linearization of the signal response with respect to analyte concentration. However, the relatively poor understanding of the power function algorithm has limited a more widespread use of CAD for quantitative assays, especially in the late stage of method validation and GMP laboratories. Herein, we present an approach to understand the inner workings of the power function value (PFV), the PFV optimization algorithm, as well as a method to determine the optimum PFV based on the signals acquired at PFV = 1 (default CAD settings). The exponent and the constant in the PFV equation used for modeling follow a trend as a function of PFV. The CAD signal at any PFV was modeled based on the signal acquired at PFV = 1, the modelling was successful for two analytes at different concentration levels on two different CAD detectors of the same model. This method reveals the functionality of the PFV which substantially simplifies the workflow needed to optimize the detector signal. The accuracy between the experimental and theoretical results showed high correlation and always resulted in the same optimum PFV determined by both ways. The approach described in this investigation simplifies the selection of the optimum PFV at which the signal is more linear, the signal-to-noise is higher, and the area reproducibility is better. The power function algorithm elucidated herein enables determination of optimum PFV from minimal experimental output and excellent overall accuracy. This paper provides an approach that includes no data transformation outside the vendor software, a very important requirement to easily validate and report results in a GMP environment.