Improvements in the predictive accuracy of the hydrophobic subtraction model of reversed-phase selectivity

A generalizable methodology for predicting retention time of small molecule pharmaceutical compounds across reversed-phase HPLC columns

Quantitative structure retention relation (QSRR) is an active field of research, primarily focused on predicting chromatography retention time (Rt) based on molecular structures of an input analyte on a single or limited number of reversed-phase HPLC (RP-HPLC) columns. However, in the pharmaceutical chemistry manufacturing and controls (CMC) settings, single-column QSRR models are often insufficient. It is important to translate retention time across different HPLC methods, specifically different stationary phases (SP) and mobile phases (MP), to guide the HPLC method development, and to bridge organic impurity profiles across different development phases and laboratories. In response to this need, we present a novel approach for retention time transfer across SPs and MPs, without requiring pre-existing Rt data on the target column. To achieve this, we developed an RP-HPLC based Genentech Multi-column Retention Time (GMCRT) database containing 51 small molecule pharmaceutical compounds analyzed on twenty SPs and multiple pH levels. The database incorporated the SP selectivity parameters from Hydrophobic Subtraction Model (HSM) - hydrophobicity (H), steric hindrance (S), hydrogen-bond acidity (A), hydrogen-bond basicity (B), ionic interaction (C) under two different pHs (2.8 and 7) and ethylbenzene (EB) retention factor. Two machine learning approaches, partial least squares (PLS) and artificial neural networks (ANN) were found to improve accuracy of Rt prediction on new SPs compared to the direct mapping approach that have been previously published, especially when the RP-HPLC columns have significant selectivity difference. As a comparison, our approach does not require pre-existing retention data on the target SPs and it is generalizable to any RP-HPLC columns with a set of known column selectivity parameters (https://www.hplccolumns.org/). The generalizability is achievable not only via the available retention data correlation among the twenty commonly-used RP-HPLC columns in GMCRT, but also via the retrainable mechanism of our ML models by adding Rt of the compounds of interest on the source columns into GMCRT, followed by predicting Rt on the target column. Thus, we propose a new QSRR framework that incorporates the physiochemical properties of SPs and MPs and makes the retention time prediction transferable across SPs and MPs. Such a framework is expected to open up possibilities for developing more comprehensive and generalizable models, and streamline RP-HPLC method development and lifecycle management across various pharmaceutical CMC development phases.

Introducing macrocyclic glycopeptide columns as unique achiral stationary phases: Insights from hydrophobic subtraction model and in-silico modeling

BACKGROUND

The need for stationary phases with unique selectivity in reversed-phase liquid chromatography has been of utmost importance to chromatographers for advancing the analysis of complex samples. Macrocyclic glycopeptide based stationary phases have been widely used for chiral separations with different chromatographic modes such as normal phase, reversed phase, and supercritical fluid chromatography. Given the multimodal retention mechanisms namely π-π complex interaction, hydrogen bonding, dipole-dipole interaction, and strong Coulombic interactions by which analytes are separated using the macrocyclic glycopeptides, these stationary phases are expected to provide novel selectivity when used under the reversed phase for achiral separations.

RESULTS

Herein, for the first time we have conducted a systematic study using the improved hydrophobic subtraction model (HSM) which incoporates dipole-dipole interactions to demonstrate the novel selectivity offered by four different macrocyclic glycopeptide based stationary phases, namely NicoShell, TeicoShell, TagShell, and VancoShell. A comparison of the HSM parameters for these columns has been made with 551 commercially available reversed phase stationary phases and the differences in the values point to the importance of adding these columns to the already existing arsenal. These stationary phases offer separations over a wide range of pH and show variability in selectivity depending on the pH of the mobile phase which make them versatile for method development in the reversed phase mode. Additionally, we have provided an actual example of a separation from an Amgen discovery project using the VancoShell column aided by computer-assisted modelling.

SIGNIFICANCE

This is the first report characterizing macrocyclic glycopeptides for achiral RPLC applications. The selectivity of these stationary phases were found to be unique when compared to other commercially available stationary phases thereby acting as their own class of columns. The unusual selectivity of the columns enabled separation of complex pharmaceutical samples.

Mechanistic studies to understand peak tailing due to sulfinic acid- and carboxylic acid-silanophilic interactions in reversed-phase liquid chromatography

Silanophilic interactions are a primary contributor to peak tailing of acidic pharmaceutical compounds, thus a thorough understanding is especially important for reversed-phase liquid chromatography (RPLC) method development. Herein, a sulfinic acid compound that exhibited severe peak tailing in RPLC with acidic mobile phases was carefully studied. Results indicated that the neutral protonated form of the sulfinic acid is involved in the strong interaction that leads to peak tailing, but that tailing can be mitigated with a blocking effect achieved through use of acetic acid modifier in the mobile phase. Peak tailing was also observed with other structurally-similar sulfinic acids and carboxylic acids but was, in general, less severe with the latter. The Hydrophobic Subtraction Model (HSM) was applied to six commercial C18 columns that exhibited different tailing behaviors for the sulfinic acid compound in attempts to identify key sites of interaction within the stationary phase. A combination of heated acid column wash experiments and density functional theory (DFT) calculations indicate that the differential interactions of the acids with vicinal silanol pairs in the stationary phase play a major role in peak tailing.

An attempt to apply a subtraction model for characterization of non-polar stationary phase in supercritical fluid chromatography

This study verified the feasibility of using a subtraction model to characterize the non-polar stationary phases (including C4, C8, and phenyl-type) in supercritical fluid chromatography (SFC). The model with 6 terms was expressed as log α = η'H + θ'P + β'A + α'B + κ'C + σ'S, where a term θ'P indicating dipole or induced dipole interaction was intentionally supplemented. Ethylbenzene and SunFire C8 were respectively defined as the reference solute and column. A 7-step modeling procedure was proposed: in the first 6 steps, except σ'S, by the use of a bidirectional fitting method, other parameters were calculated based on the equation: log α = log (k_i/k_ref) ≈ η'H + θ'P + β'A + α'B + κ'C; and in the 7th step, residual analysis was employed to describe the σ'S term according to the equation: σ'S = log α_exp. - log α_pre. Furthermore, six columns that were not involved in modeling process and 12 compounds with unknown retention were used for methodology validation. It showed good predictions of log k, as demonstrated by adjusted determination coefficient (R²_adj) from 0.9927 to 0.9998 (column) and from 0.9940 to 0.9999 (compound), respectively. The subtraction model emphasized the contribution of dipole or induced dipole interaction to the retention in SFC, and it obtained the σ'S term through residual analysis. Moreover, it made reasonable physical-chemical sense as the linear solvation energy relationship (LSER) model did, with the distinct advantages of better fitting and more accurate prediction. This study provided some new insights into the characterization of non-polar stationary phases in SFC.

A fast stop-flow two-dimensional liquid chromatography tandem mass spectrometry and its application in food-derived protein hydrolysates

Food-derived bioactive peptides have many outstanding features like high safety, easy absorption, etc. However, explorations of the peptides are suffering from the limited knowledge of sample composition and low efficiency of separation techniques. In this work, a fast stop-flow two-dimensional liquid chromatography tandem mass spectrometry (2DLC-MS) was designed and constructed in-house. For chromatographic system optimization, the effects of column pairs and fraction transfer volumes on separation performance were studied. The pair of Protein BEH SEC and HSS T3 columns was found of high orthogonality. The peak capacity detected by the optimized 2DLC reached 1165 (for corn protein hydrolysates), indicating high resolving power. Moreover, the number of peptides identified from corn, soybean and casein protein hydrolysates reached as high as 8330, 8925 and 7215, respectively, demonstrating the high potential of the system. This would help reveal the peptide composition and facilitate the research on exploring bioactive peptides from food-derived protein hydrolysates.

Development of tandem-column liquid chromatographic methods for pharmaceutical compounds using simulations based on hydrophobic subtraction model parameters

The liquid chromatography (LC) analysis of small molecule pharmaceutical compounds and related impurities is crucial in the development of new drug substances, but developing these separations is usually challenging due to analyte structural similarities. Tandem-column LC (TC-LC) has emerged as a powerful approach to achieve alternative separation selectivity compared to conventional single column separations. However, one of the bottlenecks associated with use of tandem column approaches is time-consuming column pair screening and selection. Herein, we compared critical resolution (R_c) in single column vs. TC-LC separations for a given set of small molecule pharmaceutical compounds and developed a column selection workflow that uses separation simulations based on parameters from the hydrophobic subtraction model (HSM) of reversed-phase selectivity. In this study, HSM solute parameters were experimentally determined for a small molecule pharmaceutical (Linrodostat) and ten of its related impurities using multiple linear regression of their retentions on 16 selected RPLC columns against in-house determined HSM column parameters. R_c values were calculated based on HSM database column parameters for a pool of about 200 available stationary phases in both single-phase column (2.1 mm i.d. × 100 mm) or tandem column paired (two 2.1 mm i.d. × 50 mm) formats. Four column configurations (two single and two tandem) were predicted to achieve successful separations under isocratic HSM separation conditions, with a fifth tandem pair predicted to have a single co-elution. Of these five potential candidates, one tandem pair yielded compete baseline resolution of the 11-component mixture in an experimental separation. In this specific case, the tandem column pairs outperformed single-phase columns, with better predicted and experimental R_c values for the Linrodostat mixture under the HSM separation conditions. The results reported in this study demonstrated the enormous selectivity potential of TC-LC in pharmaceutical compound separations and are consistent with our previous study that examined the potential of tandem column approaches using purely computational means, though there is room for substantial improvement in the prediction accuracy. The proposed workflow can be used to prioritize a small number of column combinations by computational means before any experiments are conducted. This is highly attractive from the point of view of time and resource savings considering over 200,000 different tandem column pairings are possible using columns for which there are data in the HSM database.

Analytical quality by design-compliant retention modeling for exploring column interchangeabilities in separating ezetimibe and its related substances

There are several potential advantages of using experimental design-based retention modeling for chromatographic method development. Most importantly, through the model-delivered systematic understanding (Design Spaces), users can benefit from increased method consistency, flexibility and robustness that can efficiently be achieved at lesser amount of development time. As a result, modeling tools have always been great supplementary assets and welcomed by both the pharmaceutical industry and the regulatory authorities. Most recently published chapters of ICH however - Q2(R2) and Q14 (both currently drafts) - evidence a further paradigm shift, specifying the elements of model-based development strategies in the so-called "enhanced approach". The main aim of this study was to investigate the impact of stationary phase chemistries on chromatographic method performance in the application example of ezetimibe and its related substances. A commercial modeling software package (DryLab®) was used to outline three-dimensional experimental design frameworks and acquire model Design Spaces (DSs) of 9 tested columns. This was done by performing 12 input calibration experiments per column, systematically changing critical method parameters (CMPs) as variables such as the gradient time (tG), temperature (T) and the ternary composition (tC) of the mobile phase. The constructed models allowed studying retention behaviors of selected analytes within each separation systems. In the first part of our work, we performed single optimizations for all nine stationary phases with substantially different surface modifications based on their highest achievable critical resolution values. For these optimum points in silico robustness testing was performed, clearly showing a change of CMPs, depending on the column, and specified optimum setpoint. In the second part of our work, we simultaneously compared the three-dimensional virtual separation models to identify all method parameter combinations that could provide at least baseline separation (R_{s, crit.}>1.50). These overlapping areas between the models described a common method operational design region (MODR) where columns were considered completely interchangeable - in terms of their baseline resolving capability - regardless of their exact physicochemical properties. A final optimized, column-independent working point within the common MODR was selected for verification. Indeed, experimental chromatograms showed excellent agreement with the model; all columns in the common condition were able to yield critical resolution values higher than 2.0, only their retentivity (elution window of peaks) was found different in some cases. Our results underline that a profound understanding of the separation process is of utmost importance andthat in some cases, adequate selectivity is achievable on various stationary phases.

Application of the "Method Operable Design Region" (MODR) approach for the development of a UHPLC method for the assay and purity determination of risperidone in risperidone drug substance and other formulations

To understand the role of analytics in drug development, regulatory bodies also started using the approach of Quality by Design (QbD) during analytical method developments. The present study deals with the development of the "Method Operable Design Region" for assay and purity determination of risperidone in risperidone drug substance and formulations usingy UHPLC. Five different column chemistries, five different pH buffers, oven temperatures from 25 to 45°C, and different organic modifier composition, column lengths and flow rates were studied and statistically evaluated using Fusion QbD software. The final method parameters were selected by performing multivariable changes in a single run and evaluated using the Monte Carlo simulation approach. The uniqueness of this method is that it is mass compatible, a total of 10 peaks are separated within a short run time of 12.0 min and it uses a "Platforming approach", which means the use of a single method for testing the drug substance, different strengths of a drug product and different formulations. The same method can be also used for the determination of the preservative (benzoic acid) in risperidone 1 mg/ml oral solution. The use of the QbD approach is aligned with the US Pharmacopeia <1220>, BP supplementary chapter 2022 and the International Conference on Harmonization Q14 guidelines for life cycle management of analytical methods.

LC Troubleshooting Odds and Ends from 2020

Sometimes our approach to troubleshooting specific problems has to change in response to changes in high performance liquid chromatography (HPLC) technology over time. In this installment, we discuss changes in technologies for mobile-phase degassing, silica-based stationary phases, and models for reversed-phase selectivity.