LSS and QSRR combined modelling for mechanistic elucidation of phenolic compound retention under diverse...

Phenolic compounds are a structurally diverse and analytically challenging class of bioactive molecules, encompassing numerous subclasses with many compounds exhibiting only subtle structural variations. Their heterogeneous nature complicates chromatographic separation, often requiring tailored workflows for accurate quantification and characterization. In reversed-phase HPLC, optimizing retention behavior is particularly difficult due to differences in column selectivity and solvent-dependent effects. To address these challenges, the present study integrates Linear Solvent Strength (LSS) theory with Quantitative Structure-Retention Relationship (QSRR) modelling to achieve both mechanistic understanding and predictive capability of phenolic retention across multiple chromatographic systems.

A set of fifty phenolic standards, including flavonoids, hydroxybenzoic, and hydroxycinnamic acid derivatives, was analyzed under six reversed-phase chromatographic conditions, defined by three stationary phases (alkyl diol, diisopropyl-cyanopropylsilane, and pentafluorophenyl-octadecylsilica) and two organic modifiers (acetonitrile and methanol). All systems were pressure-normalized to ensure comparability of retention factors. Molecular descriptors, calculated from density functional theory (DFT)-optimized structures, were subjected to a multi-output Genetic Algorithm-Partial Least Squares-2 (GA-PLS2) approach, with gradient retention factors as dependent variables. PLS2 models consistently highlighted lipophilicity (ALOGP2) and solubility (ESOL) as key predictors of retention, alongside descriptors linked to polarizability and molecular symmetry. Predicted retention factors and times were validated through LSS modelling, demonstrating strong agreement with experimental data. Overall, these results confirm that the integration of LSS modelling into the QSRR framework can provides reliable, system-wide insights into the retention behaviour of phenolic compounds.

This study shows how combining LSS with QSRR offers a dual advantage of predictive performance and mechanistic interpretation for reversed-phase HPLC of phenolic compounds. By quantitatively linking molecular properties to chromatographic retention across diverse column and solvent conditions, the approach enhances understanding of key interactions governing separation. The resulting framework provides a practical tool for improving method development, particularly in complex phenolic mixtures and high-throughput analytical workflows.