OBJECTIVETo establish the chromatography-efficacy relation method for analyzing the anti-inflammatory activity of Qizhi Weitong particles, in order to lay a foundation for quality control and pharmacodynamic evaluation of traditional Chinese medicine compounds.

METHODOn the basis of a full-time multi-wavelength fusion fingerprint of Qizhi Weitong particles, the latin hypercube sampling was used to divide six herbs in Qizhi Weitong particles into groups of different proportions to determine their inhibition ratios of TNF-alpha, IL-6 and NO released by LPS-induced RAW264. 7 cells. Pharmaeodynamic data and chemical information of HPLC fingerprints of each group were analyzed with the gray correlation method to get the anti-inflammatory effect of each chromatographic peak, and then fitted with BP neural network to establish the chromatography-efficacy relation.

RESULTThere were 25 peaks closely related to the anti-inflammatory activity. With the 25 peaks as input items, the 3-BP network was adopted to establish the neural network model for anti-inflammatory effect of Qizhi Weitong particles.

CONCLUSIONWith an error of less than 7%, the model could better fit with the complicated non-linear relation of the compound, and applied in studying the chromatography-efficacy relation. In this study on the chromatography-efficacy relation, a new method is established to evaluate the anti-inflammatory activity of Qizhi Weitong particles. It is of practical significance as an effective approach for controlling quality and exploring the material basis for efficacy of traditional Chinese medicine compounds.