Two-Step Screening Using Mid-Infrared and HPLC-UV Fingerprints to Detect Undeclared Herbal Ingredients in Weight-Loss and Erectile Dysfunction Food Supplements

Reviewed: Deconinck E, Vanhamme M, Bothy JL, Courselle P. A strategy based on fingerprinting and chemometrics for the detection of regulated plant food supplements from the Belgian market. J Pharm Biomed Anal. 2019;166:189-196.

Keywords: Adulteration, erectile dysfunction, weight loss, Aristolochia fangchi, Epimedium spp., Ilex paraguariensis, Pausinystalia yohimbe, Tribulus terrestris, fingerprint, infrared spectroscopy, HPLC-UV, chemometrics

Food supplements in Belgium are regulated under a Royal Decree issued in 1997, which was amended in 2012. The Decree includes three lists: (1) herbs that are forbidden in herbal supplements, (2) all edible mushrooms, and (3) plants that are permitted as ingredients in food supplements, but have a maximum daily limit, and need to be notified to the Belgian Federal Public Service Health, Food Chain Safety and Environment. Examples of plants on list 1 are aristolochia (Aristolochia spp., Aristolochiaceae), aconite (Aconitum spp., Ranunculaceae), arnica (Arnica montana, Asteraceae), and yohimbe (Pausinystalia johimbe, syn. Corynanthe johimbe, Rubiaceae). List 3 includes the majority of commonly used medicinal plants such as echinacea (Echinacea spp., Asteraceae), ginkgo (Ginkgo biloba, Ginkgoaceae), or cranberry (Vaccinium macrocarpon, Ericaceae).

The approach by the authors of this paper involved the development of a model to identify five plants, Aristolochia fangchi, epimedium (Epimedium spp., Berberidaceae), maté (Ilex paraguariensis, Aquifoliaceae), tribulus (Tribulus terrestris, Zygophyllaceae), and yohimbe, using a two-step identification process based on fingerprints from infrared (IR) spectra and high-performance liquid chromatography-ultraviolet detection (HPLC-UV) chromatograms. For the IR spectra, a chemometric model was built based on serial dilutions of authentic plant materials with lactose. For the HPLC-UV model, 50% aqueous methanol extracts and, for erectile dysfunction (ED) products, an additional extract made with 7.4% HCl were used. The same LC gradient was used to obtain the epimedium, tribulus, and yohimbe fingerprints; however, specific conditions had to be developed for both aristolochia and maté. Different algorithms (soft-independent modeling of class analogy [SIMCA], partial least squares [PLS], and k-Nearest neighbors [k-NN]) were used to build the chemometric models. The strategy was then applied to 35 weight-loss and 34 erectile dysfunction food supplements in tablet or capsule form. The supplements were obtained by the researchers from the Belgian Federal Agency for the Safety of the Food Chain (FASFC), which had previously investigated them based on the suspicion of containing active pharmaceutical ingredients.

After developing appropriate chemometric models for IR and HPLC-UV using the authenticated plant materials, the commercial samples were assessed using the same conditions. In addition, samples that tested positive for one of four plants in the screening (presence of epimedium could not be confirmed due to a lack of a characteristic chromatogram), were submitted to HPLC-MS/MS analysis for confirmation. In the weight-loss category, 16 of 35 samples (46%) were correctly assigned by IR, while predictions with HPLC-UV were 100% accurate. For ED supplements, the target plants were identified in 14 of 34 commercial products (41%) with IR but for all target plants with HPLC-UV. The final results showed that one weight-loss product contained undeclared aristolochia, and 11 included undeclared maté. In the ED category undeclared epimedium, tribulus, and yohimbe were detected in one, one and three samples, respectively. Curiously, no tribulus was found in four commercial ED products claiming to contain this ingredient.

Comment: While ultimately successful, the proposed approach using chemometric models based on IR spectra and HPLC chromatograms to predict the presence/absence of certain herbal ingredients may not be the best option for a forensic laboratory. The advantage of IR and HPLC-UV is that the instruments are common, and the sample preparation is relatively easy. However, herbal food or dietary supplements often include excipients and possibly other active ingredients. This is especially the case in the weight-loss and ED categories, where multi-ingredient products are common, and manufacturing processes tend to concentrate or remove certain phytochemicals. Therefore, fingerprints of commercial samples are likely to be very different from those of the original crude raw material that was selectively prepared, making predictions based on chemometric models prone to errors. An assessment using several (ideally characteristic) marker compounds/ions, as in the HPLC-MS/MS method used to confirm the screening results, provides a more robust identification method.