The silica-based sorbents for the mixed-mode extraction of risdiplam and its metabolite from serum samples

Abstract

Analysis of risdiplam and its main metabolite is critical for pharmacokinetic studies and therapeutic drug monitoring in spinal muscular atrophy. However, current analytical methods face challenges due to low analyte stability, limited selectivity, and pronounced matrix effects, with only four studies published to date. In this work, a comprehensive analytical strategy was developed for the first time, combining optimized chromatographic separation with a novel sample preparation approach. The effects of methanol content, salt concentration, and pH on retention, peak symmetry, and resolution were systematically investigated. Chromatographic conditions were optimized using a central composite design, enabling rapid selection of mobile phase and achieving efficient separation of both compounds within 3.5 min. General trends in mobile phase composition were identified, indicating optimal performance at either low salt concentration and high pH or high salt concentration and low pH. For sample preparation, dispersive solid-phase extraction was introduced using newly synthesized hydrophobic and mixed-mode sorbents designed to provide hydrophobic, π–π, electrostatic, and hydrogen-bonding interactions. The extraction procedure was extensively optimized with respect to sorbent type and elution conditions (solvent composition and the addition of salt, formic acid, or ammonia). Compared with conventional non-polar sorbents, mixed-mode materials containing alkylamide, benzoic acid, cholesterol, and residual aminopropyl groups demonstrated superior recoveries and reproducibility. Finally, an alkylamide sorbent with acidified methanol-water elution was selected. The method enabled direct extraction from serum without protein precipitation, yielding recoveries of 86% and 53% for risdiplam and its metabolite, respectively. The proposed strategy provides a robust, selective, and efficient approach for risdiplam bioanalysis.