Dietary supplement mislabelling: case study on selected slimming products by developing a green isocratic HPLC method for their quality control

Nowadays, a huge population consumes Dietary supplements for losing weight. Products are often claimed as botanical blends, yet they aren't necessarily safe. Misleading labels are also very common. Thus, validated analytical methods for a wide range of slimming compounds are highly needed. Herein, we present a simple HPLC/PDA method for the quantitation of seven popular slimming ingredients. Studied compounds were Caffeine, Raspberry Ketone, trans-Resveratrol, p-Synephrine, p-Octopamine, p-Hordenine and 2-phenethylamine. After optimization, separation was carried out on a C18 column and mobile phase was a mixture of Acetonitrile:Water containing 0.1% phosphoric acid (50:50, %v/v). The last compound was eluted at 9.76 min. Separation was efficient showing baseline- separated symmetric peaks, without using any gradient programs, organic mobile phase modifiers or modified stationary phases. Method validation was done following ICH guidelines. Calibration curves were linear over wide concentration ranges and calculated LOD values were in the range 0.02–0.09 µg/mL. Method greenness was assessed using Analytical Eco-scale, GAPI and AGREE metric tools. Further, four random sample products purchased from online supplement stores were assayed. Results proved some mislabeling actions. To support our findings, standard addition was carried out and average % recoveries were 96.67 – 101.44% with standard deviation ≤ 2.83 between measurements.

Instrumentation and chromatographic conditions. Analysis was carried out using HPLC Waters Alliance e2695 separating module (Waters Co., MA, USA) equipped with a quaternary solvent system, in-line vacuum degasser, 100 µl injection loop; autosampler, heated column compartment and photo-diode array (PDA) detector. Analytical column was RP Spherisorb® ODS-2; 250 × 4.6 mm with 5 µm particle size (Waters Co., MA, USA). Empower-3 chromatography data software was used for data collection and acquisition. Column compartment and sample tray were kept at 25 °C. Mobile phase was a mixture of Acetonitrile: 0.1% phosphoric acid in ratio 50:50%v/v. 20 µl of samples were injected in triplicates and isocratically eluted with flow rate 0.9 mL/min. Total run time was 11 min and column was equilibrated with mobile phase for 5 min between single injections. PDA detection was carried out at 205 nm for Caffeine and 2-Phenylethylamine, 225 nm for Raspberry Ketone, p-Octopamine, p-Synephrine and p-Hordenine and 305 nm for trans-Resveratrol. Separation efficiency was verified by system suitability results. Sample preparation. Emptied capsule contents or finely powdered tablets equivalent to one serving size were accurately weighted, dissolved in water (samples 1 and 2) or acetonitrile: water (1: 9, %v/v) (samples 3 and 4), sonicated for 10 min, filtered and stock solutions of each sample were prepared. Appropriate dilutions of each sample were prepared in mobile phase to fall within the calibration range and analysed under the previously mentioned chromatographic separation conditions. Extraction efficiency was further evaluated using standard addition technique. Known amounts of pure standards were added to commercial samples; those mixtures were then subjected to the entire sample preparation and quantitation steps. Recoveries were calculated by subtracting the concentration of the non-spiked samples from the total concentration of the spiked samples.

Results and discussion
In view of the poor DS regulation, the existence of analytical methods to simultaneously identify and quantify different constituents in these products is very important to the field of health surveillance. This study sheds the light on seven popular slimming compounds; namely: Caffeine, trans-Resveratrol, Raspberry Ketone, p-Octopamine, p-Synephrine, p-Hordenine and 2-Phenethylamine. A simple, ecological and cheap RP-HPLC/PDA method was developed and validated for their simultaneous determination. To date, analytical methods are published for quantitation of a single compound or combination of few of them. Moreover, these methods employ long gradient programs and propose minimal validation data 5,22,30,33 . Thus, this proposal can be put forward as a general green, simple and valid method for the quality control of a wide variety of commercially available DS products.
Method optimization. On our chemical investigation of all mentioned compounds, we aimed to develop a green method that shows best resolution, peak shape and sensitivity using a conventional ODS column in a very short run time. Chromatogram showing all separated compounds is presented in Fig. 1. A standard C18 column was preferred as it is the most popular among quality control labs. However, all analytes were basic hydrophilic and separation of the basic PEAs on unmodified reversed phase silica was challenging. At neutral  www.nature.com/scientificreports/ water pH, basic analytes are strongly retained on silanol residues and so peaks didn't appear in the relevant chromatograms. At low pH, unionized silanols and ionized basic analytes limit ion-exchange and let compounds be eluted 47 . Thus, methods in literature always used acidic buffers or ion pairing agents (e.g.: heptane sulphonate or sodium lauryl sulphate) through gradient mobile phase composition and/or gradient flow rate programs in long run times 31,48 . These mobile phase additives shorten the column lifetime and tedious gradient procedures aren't feasible for quality control practices. Only two methods were published in literature presenting a low concentration phosphoric acid in mobile phase instead of using organic acid modifiers. Insufficient resolution was observed between Octopamine and Synephrine peaks despite that both used gradient mobile phase composition and/or flow rate gradient elution 36,37 . The proposed method was optimized regarding mobile phase composition, flow rate, detection wavelength and injection volume. Mixtures of methanol: 0.1% phosphoric acid and acetonitrile: 0.1% phosphoric acid were tested in different proportions and flow rates. Acetonitrile was more promising than methanol due to its lower viscosity and UV cutoff. Methanol chromatograms were noisy and this reduced method sensitivity 49 . Acetonitrile: 0.1% phosphoric acid mixture was tested in the ratios, 30:70, 50:50 and 70:30, %v/v. In low acetonitrile ratio, Octopamine and Synephrine peaks co-eluted. Increasing acidified water proportion caused relative increase in protonation of bases and decreased their retention on column. In high acetonitrile ratio, last compound was eluted at 30 min along with co-elution of Caffeine and trans-Resveratrol peaks.
The best baseline separation of the seven compounds was observed with isocratic elution of acetonitrile: 0.1% phosphoric acid mixture (50:50, %v/v). Isocratic elution was sufficient to provide satisfactory baseline separation with resolution between peaks more than 2, eliminating the need for the previously published laborious gradient programs and organic modifiers in the mobile phase. Blank solvent samples were injected and no carryover was detected between runs. Different pump flow rates were tested, a flow rate 0.9 mL/min showed the best resolution between peaks along with the least possible run time with the last compound eluted at 9.76 min. Although Raspberry Ketone and PEAs showed nearly similar UV spectra, they were differently eluted from the column under optimized conditions. Each analyte was detected at its λ max ; 205 nm for Caffeine and 2-Phenethylamine, 225 nm for Raspberry Ketone, p-Synephrine, p-Octopamine and p-Hordenine and 305 nm for trans-Resveratrol. Optimum injection volume was 20 µl, above which peak shapes were distorted. Hence, both detection wavelengths and injection volume attributed to the method sensitivity.
Method validation. All validation parameters for our proposed method were calculated as per the ICH Q2 (R1) guidelines 50 . Linearity. Five to seven point calibration curves were constructed for the studied drugs, correlating average peak area at the selected detection wavelength to the corresponding concentration and regression equations were computed. All linearity ranges and regression parameters are listed in Table 1. All drugs were linear over acceptably wide concentration ranges with correlation coefficient values higher than 0.999.

Limit of detection (LOD) and limit of quantitation (LOQ).
Method sensitivity was assayed in terms of LOD and LOQ values. Determination of LOD and LOQ was based on calibration data using the residual standard deviation of regression line in the range of detection limit (σ) and the slope of the calibration curve (s), applying the formula LOD = 3.3 × σ/s and LOQ = 10 × σ/s. Results are presented in Table 1.
Accuracy and precision. Accuracy was evaluated by analyzing different concentrations of pure drug solutions across the linear ranges. Accuracy results were expressed as average percentage recoveries. The percentage relative standard deviation (%RSD) was calculated between a set of triplicate measurements of three different concentrations of each drug performed on the same day and on three consecutive days for expressing both Intra-day and Inter-day precision; respectively. Calculated results, expressed in Table 1, prove that our method is accurate and precise as indicated by percentage recoveries close to 100% and %RSD values less than 2%.
System suitability. System suitability testing evaluates the whole components of an analytical system and efficiency of separation. During method optimization, System suitability parameters were calculated using Empower® software as described in FDA guidance for industry on Validation of Chromatographic Methods and USP-NF General Chapter <621> Chromatography-system suitability 51 . Under the selected conditions, results ( Table 2) were within the acceptable values verifying that the chromatographic system is efficient showing well resolved symmetric peaks. Thus, the proposed method is adequately valid for its intended purpose.
Robustness. Robustness testing is a measure of the degree of method reliability to small but deliberate variation in operating conditions. It also highlights and set limits for critical method parameters. Multiple samples at the same concentration level were analyzed after varying pump flow rate (± 0.05 mL/min) and mobile phase composition (± 2%) and %RSD of assay was calculated at each typical variation. Results (   www.nature.com/scientificreports/ Analysis of real samples. After validation, the viability of the proposed method was endorsed. Four different DS samples were purchased from online retailers and analyzed using our mentioned procedures. Extraction was simply done using water, except for samples claimed to contain Resveratrol, that 10% acetonitrile in water was used. Compounds were identified on basis of relevant retention times, UV spectra and spiking samples with known amounts of pure standards for spectrally similar compounds. Representative chromatograms (Fig. 3) showed no interfering peaks in the relevant regions. Moreover, peak purity testing was performed using Empower® PDA software to exclude that possible impurities or degradates were co-eluted with studied compounds. Evaluation involved comparing a purity angle to threshold angle for each peak. Purity angle, the degree of spectral homogeneity, compares spectra across each point in the peak to the spectrum at the peak apex. Threshold angle is the largest possible purity angle attributed to co-elution, noise, solvent and detector error. An impurity is detected within a peak when its purity angle exceeds its threshold angle. In all tested DS products, peaks corresponding to studied drugs showed purity angle lower than threshold angle (Table 4) demonstrating spectral homogeneity, therefore, confirming the purity of eluting peaks 52 .  Table 3. Robustness results of the proposed HPLC method expressed as %RSD of the drug assay.  www.nature.com/scientificreports/ Contents in samples were measured and compared relevant to label claims. Then samples were spiked with known amounts of pure standards to further identify compounds and evaluate the extraction efficiency. Assay and standard addition results are presented in Table 4.

Condition Caffeine resveratrol Raspberry Ketone Octopamine Synephrine Hordenine 2-phenethyl amine
For samples (1) and (2), measured amounts were in fairly good agreement with those on the label. Sample (1) contained a total intake of p-Synephrine and p-Octopamine higher than the safe levels, in addition to the high amount of Caffeine 2,53 . This probably explains the manufacturer's intention to mislead the consumer about the active agent content, that the mixture was labeled as a proprietary blend without stating the exact amount of each component. Despite that assay results of sample (2) matched with label claim, it is crucial to alert that it contains a very high level of the prohibited synthetic 2-Phenethylamine and yet is still available for sale to the public. For the other tested samples, results showed that the actual contents largely varied from the label claims. No resveratrol was detected above method LOD in sample (3), although it was labeled to contain 100 mg of grape extract. Moreover, it contained nearly half of its labeled Raspberry Ketone content. Similarly, Sample (4) contained approximately 3% of the labeled Raspberry Ketone, although it was declared as the main active agent in the product. To support the accuracy of our findings, standard addition was applied to the faulty samples and average recoveries varied between 96.67 and 103.33%.
Green assessment. Analytical eco-scale assessment. Method greenness was assessed using the ''Eco-Scale'' tool. In evaluation, it considers hazard and amount of used re-agents, energy use, occupational hazard and waste generation. Each item is given a number of penalty points (PPs) and the total PPs are subtracted from an ideal total score of 100. Eco-scale scores >75, >50 or < 50 describe excellent, acceptable or inadequate greenness; respectively. Eco-scale score, calculated as described by Namiesnik et al. 54 , was 90 proving that our method is an excellent green one (Table 5).
Green analytical procedure index (GAPI). GAPI is a semi-quantitative tool that evaluates all stages of an analytical procedure from sample collection to preparation to final analysis. It also shows whether a method is qualitative or quantitative. Assessment involves a three colored symbol that consists of five pentagrams. Each section is colored green, yellow or red for high, mid and low environmental friendliness; respectively. This metric visually highlights the variations among different analytical methodologies 55,56 . GAPI assessment of our proposed analytical method was calculated using ComplexGAPI® software 57 and results are shown in Fig. 3a.
Analytical GREEnness metric approach (AGREE). AGREE is an automated metric that considers the 12 principles of Green Analytical Chemistry (SIGNIFICANCE) during evaluation. Each criterion is given a score from 0 to 1. Final result is a pictogram divided into 12 segments. Each segment is colored green, yellow or red in different color intensities, according to its greenness. The width of each segment corresponds to its weight assigned by user. Finally, a total score of the whole analytical methodology is presented in the middle of the pictogram with a color corresponding to this score, where dark green is associated with values close to 1.This method is considered unique for its flexibility and that it not only considers the hazards of an analytical method but its output as well 58 . AGREE assessment of our proposed analytical method was calculated using AGREE® software and results are shown in Fig. 3b.

Conclusion
In this manuscript, the problem of DS mislabeling in market was assessed. For the assay of weight loss DS, an isocratic RP-HPLC method with diode array detection was developed for the separation and quantitation of seven popular lipolytic agents in weight loss DS. The proposed method is environment-friendly in which only acetonitrile and acidified water are enrolled for complete separation in an adequately short run time. Separation was efficient showing well resolved Gaussian shaped peaks. Validation was done in accordance to ICH guidelines and results proved that the method is sensitive, accurate, precise and robust across a wide concentration range. Further, it was applied to the assay of four DS products randomly purchased from online stores. Our results revealed the discrepancy between labeled and actual contents in two of the analyzed samples. The other two samples agreed with label claims. However, it's worth alerting that these two samples contained higher than safe levels of caffeine and PEAs that are already on monitoring/Prohibited WADA list and yet still available to the public. Finally, we believe that this simple, green, efficient and cost-effective procedure will contribute to enhancement of DS regulatory practices.

Data availability
The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.