A strategy to identify and quantify closely related adulterant herbal materials by mass spectrometry-based partial least squares regression

- An approach to quantifying adulterant herbal materials by MS techniques and partial least squares regression was developed.
- Good linearity and high predictive accuracy were demonstrated.
- The performance of different MS techniques in modeling PLSR were firstly reported.

In this study, a new strategy combining mass spectrometric (MS) techniques with partial least squares regression (PLSR) was proposed to identify and quantify closely related adulterant herbal materials. This strategy involved preparation of adulterated samples, data acquisition and establishment of PLSR model. The approach was accurate, sensitive, durable and universal, and validation of the model was done by detecting the presence of Fritillaria Ussuriensis Bulbus in the adulteration of the bulbs of Fritillaria unibracteata. Herein, three different MS techniques, namely wooden-tip electrospray ionization mass spectrometry (wooden-tip ESI/MS), ultra-performance liquid chromatography quadrupole time-of-flight mass spectrometry (UPLC-QTOF/MS) and UPLC-triple quadrupole tandem mass spectrometry (UPLC-TQ/MS), were applied to obtain MS profiles for establishing PLSR models. All three models afforded good linearity and good accuracy of prediction, with correlation coefficient of prediction (rp2) of 0.9072, 0.9922 and 0.9904, respectively, and root mean square error of prediction (RMSEP) of 0.1004, 0.0290 and 0.0323, respectively. Thus, this strategy is very promising in tracking the supply chain of herb-based pharmaceutical industry, especially for identifying adulteration of medicinal materials from their closely related herbal species.

231