Classification of Spanish Paprika by Linear Discriminant Analysis
Spain is one of the main producers of paprika (Capsicum annuum, Solanaceae) fruit. There are three main types of commercial paprika: sweet, bittersweet, and spicy. Phenolic compounds are ubiquitous in plants, and to date, >8000 phenolic compounds have been identified. Phenolic profiling can be used to classify products and identify adulteration. According to the authors, phenolic profiling and high-performance liquid chromatography-ultraviolet (HPLC-UV) fingerprinting of paprika has not been previously conducted as a method for detecting adulteration. Hence, the purpose of this study was to develop a simple and inexpensive HPLC-UV method to classify, characterize, and authenticate Spanish paprika samples in a way that can be easily conducted in any food control laboratory.
The authors purchased 96 paprika products from Spain. A total of 72 sweet (n = 26), bittersweet (n = 23), and spicy (n = 23) samples were from La Vera protected designation of origin (PDO) region, and a total of 24 sweet (n = 12) and spicy (n = 12) samples were from Murcia PDO region. HPLC-UV chromatographic fingerprints were obtained at 280 nm wavelength. Sonication/centrifugation and determination of phenolic compounds were carried out in reversed-phase chromatography using a C18 column.
The authors started with 17 major phenolic compounds that have been previously identified in paprika. The total chromatographic elution profile was 30 min. To evaluate the performance of the proposed HPLC-UV method, run-to-run (intra-day) and day-to-day (inter-day) precision of compounds quantification were evaluated. The run-to-run precision values (0.5%) were better than the day-to-day precision values (2-10%). This difference was expected. The run-to-run trueness was compared with standard known phenolic compounds, and the results were “very good” with relative errors of 0.04-8.9% for 14 of 17 compounds.
The next step was to test whether the proposed HPLC-UV method could produce discriminant UV fingerprinting profiles for the classification and characterization of the Spanish paprika samples. The authors used three paprika samples from one producer in La Vera PDO but of different varieties (sweet, bittersweet and spicy). The HPLC-UV chromatograms were compared, and the signal profiles and intensities varied depending on the paprika variety.
Rather than trying to match the chromatogram peaks to a specific compound and analyze their areas, the authors decided to process the whole chromatogram after its compression with fast Fourier transform (FFT) as a unique profile and analyze it with pattern recognition methods such as principal component analysis (PCA) and linear discriminant analysis (LDA) for the classification of paprika samples. This pattern approach eliminates the need for marker compounds. The authors “trained” the system on 71 samples, and then tested the predictive capability on 25 samples (five of each variety). This method prevents bias because the testing and training was done with different samples. The method was able to successfully authenticate the paprika samples.
The authors conclude that “the combination of liquid chromatography with chemometric methods is a suitable approach for the authentication of paprika samples.” Besides this, “the proposed HPLC-UV method is, in general, very satisfactory in terms of sensitivity, precision, and trueness for the determination of phenolic compounds.” The advantage of this method is that it is a simple and inexpensive extraction procedure; also, the approach is versatile and quick. Since this method does not require marker compounds, after optimization of chromatographic conditions for the analysis of a proper set of samples to build the chemometric model, it may be useful for other products with unknown or questionable markers giving relevance to the presence of phenolic compounds. The authors declare no conflicts of interest.
Ceto X, Serrano N, Arago M, et al. Determination of HPLC-UV fingerprints of Spanish paprika (Capsicum annuum L.) for its classification by linear discriminant analysis. Sensors. December 18, 2018; 18(12). PII: E4479. DOI: 10.3390/s18124479.