Trichosanthis Fructus, the dried ripe fruits of Trichosanthes kirilowii
Maxim. or T. rosthornii
Harms (Fam. Cucurbitaceae), has been commonly used in Traditional Chinese Medicine (TCM) for the treatment of cough with lung heat, sticky phlegm, constipation, thoracic obstruction, and cardiodynia [1
]. In modern clinical practice, Trichosanthis Fructus and its TCM prescriptions played a very important role in treating cardiovascular diseases including angina, cardiac failure, myocardial cinfarction, arrhythmia during acute myocardial infarction reperfusion, pulmonary heart disease, and cerebral ischaemic disease [2
]. Because of its high medicinal value and good economic benefit, T. kirilowii
has been cultivated widely in China [5
], such as Shandong, Shanxi, Henan, and Hebei Provinces [6
]. Specially, Trichosanthis Fructus produced from Shandong Province, was considered to be genuine since it showed the highest curative effect in traditional clinical use and active constituent content [6
]. However, it was not easy to discriminate the geographical origin by visual inspection. There were a few studies on the discrimination of Trichosanthis Fructus from different cultivars or geographical origins using analytical methods including high pressure liquid chromatography (HPLC) fingerprint [9
], seed protein electrophoresis [10
], scanning electron microscope [11
], and random amplified polymorphic DNA (RAPD), internal transcribed spacer (ITS), and sequence-related amplified polymorphism (SRAP) molecular markers [12
]. However, these methods were time-consuming, costly, and destructive. Therefore, a fast, accurate, and non-destructive analytical method was established to discriminate the geographical origins of Trichosanthis Fructus in this work.
Near infrared (NIR) spectroscopy is a fast, accurate, and nondestructive technique requiring minimal sample processing before analysis. Coupled with chemometric techniques, it appears to be an effective and powerful analytical tool widely used in different fields, such as agricultural food [14
], petrochemical [16
], pharmaceutical [17
], environment [18
], metabolomic profiling [19
], etc. The NIR region spans the wavelength range between 780 and 2500 nm. The absorption bands in this region correspond mainly to combinations and overtones of the fundamental vibrations of O-H, C-H, S-H, and N-H bonds, which are the primary structural components of organic chemical constituents [20
]. As the environment factors including light, climate, water, soil, planting methods, etc. have great influences on the growth quality of the medicinal plants, some chemical constituents of the same crude drug from different geographical origins vary in content. Therefore, NIR spectroscopy has been also used to determine the geographical origins of TCM, such as Radix Pseudostellariae [21
], Herba Epimedii [22
], and Gastrodiae Rhizoma [23
] in recent years. However, there has not been any reports until now on the use of NIR spectroscopy for the discrimination of Trichosanthis Fructus from different geographical origins, and the discrimination between crude drugs and prepared slices. It has been found that the concentrations of total saponins, amino acids and total flavonoids were different in different geographic origins [7
], and the concentrations of 5-hydroxymethylfurfural, vanillic acid, quercetin, luteolin, and sugar in prepared slices showed significant changes compared with crude drugs, especially the concentration of 5-hydroxymethylfurfural increased to nearly 26 times as much as crude drugs of Trichosanthis Fructus (p
< 0.05 or p
< 0.01) [25
]. All of the above laid the foundation for the feasibility of the following experiment.
In this study, four chemometric techniques including K-nearest neighbors (KNN), soft independent modeling of class analogy (SIMCA), partial least squares discriminant analysis (PLS-DA), and support vector machine discrimination analysis (SVM-DA) were attempted to discriminate Trichosanthis Fructus from different geographical origins. Among them, KNN, SIMCA, and PLS-DA were three linear methods, while SVM-DA was a non-linear method. Principal component analysis (PCA) was conducted on the NIR data to extract some principal components (PCs) as the inputs of the supervised pattern classification models. The number of PCs was optimized by cross-validation.
This study sufficiently demonstrated that NIR spectroscopy coupled with chemometric techniques had high potential to distinguish the crude drugs of Trichosanthis Fructus from different geographical origins and to discriminate the crude drugs and prepared slices in an accurate and non-destructive way. The successful discrimination using chemometric analysis was based on their differences in NIR spectra, which mainly correlated with the differences in their chemical compositions. The differences of crude drugs from different geographic origins might be caused by soil, climate, light, planting methods, and other factors. Light affected the synthesis and accumulation of carbohydrates and nitrogen metabolism of plants, soil affected the absorption of mineral elements in plants, and climate affected the growth cycle of plants resulting in the inconsistency of fruit maturity. All these led to changes in the types and contents of the constituents in Trichosanthis Fructus of different geographic origins. The variations between crude drugs and prepared slices might lie in that the former was only dried from the fresh fruits in the air, while the latter also needed to be steamed through, pressed, shredded, and sun-cured after dried fruits.
Four chemometric techniques (KNN, SIMCA, PLS-DA, and SVM-DA) were applied comparatively to construct the classification models. Among the four classification models, SVM-DA as a non-linear classification method showed superior performance over the linear ones of KNN, SIMCA, and PLS-DA after preprocessing with MSC. The classification accuracy of the calibration set and prediction set were both 100% when C = 100, γ = 0.00316, and PCs = 6. Generally, the non-linear model performed better than the linear models.
The genuineness of herbal medicine depends mostly on its geographical origins. It can be concluded that NIR spectroscopy coupled with chemometric techniques will have more application on the discrimination of TCM according to different geographical origins similarly, which is essential for quality control and traceability management. It also has a promising future to distinguish crude drugs and prepared slices, authenticity, adulteration, and storage period of TCM, all of which cannot be easily recognized by simple visual inspection. Therefore, more representative TCM samples need to be collected and experimented to develop more robust models for prediction in further studies.