Development of a high-performance thin-layer chromatography (HPTLC) method coupled with chemometric fingerprinting analysis for rapid identification and quantification of soyasaponins

Introduction Saponins, produced by plants, are tensioactives secondary metabolites. Those molecules provide emulsifying and stabilizing properties that have been exploited in the pharmaceutical and food industries. However, quantification of saponins is considered challenging because they do not have chromophores in their structure which restricts their detection in UV light. Current analytical strategies commonly rely on liquid chromatography coupled with evaporative light-scattering or mass-spectrometric detection. Methods Herein, we report the development of a high-performance thin-layer chromatography (HPTLC) method for the detection and determination of soyasaponins in six different soy-derived samples, combined with chemometric fingerprint analysis to improve interpretation of closely migrating chromatographic signals. Samples were applied onto precoated silica gel 60 F254 plates, developed using an optimized mobile phase, and derivatized with anisaldehyde/H 2 SO 4 . Results Detection at 366 nm produced higher chromatographic responses and more similar calibration behavior among soyasaponins Aa, Ba, and Bb than detection at 550 nm. The residual error of the pooled calibration decreased from 52.76% at 550 nm to 12.06% at 366 nm, supporting the use of a simplified calibration approach under the selected conditions. In addition to conventional peak-area analysis, chromatographic fingerprints were processed using a reference-guided strategy in which experimental (Rf) values of Aa, Ba, and Bb were used to define reference-associated regions. The complete chromatographic profiles were further converted into regional descriptors and evaluated by hierarchical cluster analysis. This approach allowed similarities and differences among sample fingerprints to be visualized and helped identify chromatographic regions that contributed most strongly to sample differentiation, even when the chemical identity of every detected signal was not known. The method showed an experimental detection limit of 20.3 ng and good intraday repeatability, with RSD values below 1% for (Rf) and below 10% for peak area. Discussion The method combines HPTLC separation, simplified calibration, reference-guided signal assignment, and chemometric fingerprint analysis in a practical strategy for soyasaponin analysis. The chemometric treatment extends the information obtained from conventional HPTLC by facilitating comparison of complex samples and highlighting chromatographic regions for subsequent targeted characterization.