Rapid Screening and Quantitative Determination of Illegal Phosphodiesterase Type 5 Inhibitors (PDE-5i) in Herbal Dietary Supplements

Phosphodiesterase type 5 inhibitors (PDE-5i) are the first-line medication for oral erectile dysfunction, which are used according to the prescription of doctors. However, these substances have been found illegally in supplementary foods. The quality and safety of dietary supplements for enhancing male sexual performance have been questioned, raising the need for continual development of analytical methods. Liquid chromatography coupled with high-resolution mass spectrometry has become one of the most effective methods to identify and measure PDE-5i concentration. In this research, we focused on (i) developing and validating an effective screening and quantitation method for more than 53 PDE-5i in ingredients and supplementary products using LC-Q-Exactive after a simple sample extraction and (ii) assessing PDE-5i content in natural-based supplementary products available in Vietnam market. The extraction method used a small amount of organic solvent, which makes it more environmentally friendly (greener). The developed method has a limit of detection of 0.4 mg/kg, a limit of quantitation of 1.2 mg/kg, recoveries from 80 to 110%, and repeatability lower than 15%. Ninety-two herbal supplementary foods and ingredients used for enhancement of male sexual performance available in Vietnamese markets were collected. Fourteen PDE-5i including conventional and novel analogous were detected and measured in eighteen food supplements and two formulation ingredient samples.


Introduction
Phosphodiesterase type 5 (PDE-5) is an enzyme responsible for the breakdown of cyclic guanosine monophosphate (cGMP) in the corporal smooth muscle [1]. us, PDE-5i are considered the first-line medication for oral erectile dysfunction (ED) therapies [2]. After the approval of sildenafil, several PDE-5i have been approved and demonstrated wellestablished efficacy in patients with ED such as tadalafil, vardenafil, avanafil, mirodenafil, undenafil, and lodenafil. Additionally, PDE-5i proved their great potential in the treatment of neuroinflammation, neurodegeneration, cognition (Alzheimer's disease), cancer therapeutics, diabetic peripheral neuropathy, renoprotection, etc. [3]. Although the safety of PDE-5i was proven, the use of PDE-5i has some adverse effects such as ataxia caused by acetildenafil and its analogs and symptoms of giddiness, headache, shortness of breath, and backache [4]. PDE-5i have an interaction effect with other medicines such as nitrates. From 2007, the US FDA announced that a warning of the potential risk of visual and auditory impairment related to nonarteritic anterior ischemic optic neuropathy and sudden sensorineural hearing loss would be added to drug labels of PDE-5i [3]. Seriously, the first known fatal case related to desmethyl carbodenafil, an unapproved PDE-5i, on a 34-year old male was reported in 2017 [5]. erefore, the use of PDE-5i in therapy should strictly obey the advice of pharmacists.
PDE-5i, namely sildenafil, tadalafil, vardenafil, and their analogs, have been added in supplementary foods, which were supposed to be made of natural ingredients. e presence of PDE-5i in these supplement foods without labeling is consumer deception. In 2013, J.H. Lee et al. reported more than 46 PDE-5i analogs in various forms of health food products in the online and offline market of Korea [6]. A case study in the Czech market revealed that 10 over 64 natural herbal-based supplements for ED treatment contain both registered or unregistered synthetic PDE-5i [7]. In the summary of the Min-Yong Low research group, Asia reported the highest number of PDE-5i as adulterants in dietary supplements and was followed by Europe and North America [4]. Research of the Malaysian market showed 82% tested unregistered products and 14% of the registered products were adulterated with PDE-5i or their analogs [8].
e regulation of PDE-5i has been complicated because of the increasing number of novel synthetic PDE-5i analogs.
us, the current situation raises the need for continual development of analytical methods to quickly detect PDE-5i analogs in these products. e rapid and accurate identification and measurement of popular and unknown PDE-5i have been improved by numerous analytical techniques including high-performance liquid chromatography (HPLC) [9,10], gas chromatographymass spectrometry (GC-MS), nuclear magnetic resonance (NMR) spectroscopy [11], vibrational spectroscopy, liquid chromatography-Fourier transform ion cyclotron resonance mass spectrometry (LC-FT-ICR-MS), etc. [12]. e most effective approach for the identification of PDE-5i in supplements is HLPC-MS [6,13,14], although some publications had been done by HPLC with a UV detector [9] or photodiode array detector [14]. In 2015, the AOAC International published an official method for screening and identification of PDE-5i in dietary ingredients and supplements [15]. e samples were simply extracted with a mixture of solvent (methanol, acetonitrile, water), then diluted, filtered, and analyzed by LC quadrupole-orbital ion trap MS. Identification of targeted and nontargeted analytes was conducted based on retention time, accurate mass, and isotopic pattern of precursors ions and product ions using an in-house database. Recently, Hong et al. have reported a screening and classification method of PDE-5i by GC-MS [16]. Specific common ions according to structural after the trimethylsilyl derivatization characteristics of four PDE-5i classes were found. e development of high-resolution mass spectrometry (HRMS) techniques has made screening applications more selective than conventional MS techniques. Currently, timeof-flight mass spectrometry (TOF) or Orbitrap mass spectrometry techniques can achieve high mass accuracy (below 5 ppm). In particular, Orbitrap mass spectrometry can perform high-resolution MS/MS allows both the screening of unknown compounds and the quantification of target substances. erefore, liquid chromatography coupled with HRMS has become one of the most effective methods to identify and measure PDE-5i concentration. Our goals are (i) developing and validating a screening and quantitation method for PDE-5i in supplementary products and (ii) assessing PDE-5i content in natural-based supplementary products and ingredients available in the Vietnam market.
Each solid standard was accurately weighed about 10 mg and diluted by 10 mL methanol to obtain a 1000 μg/mL standard solution.
e stock solutions were kept in dark bottles at 4°C and to use for 1 year. e working solutions were prepared by diluting the stock solutions with methanol into the concentration of 0.01, 0.1, 0.2, 0.5, and 1 μg/mL.

Sample Preparation.
Ninety-two supplement foods and ingredients specific for the enhancement of male sexual performance were collected in local stories in Vietnam.
Samples in the form of tablets were crushed into a fine powder. In samples in the form of hard-shelled capsules or soft-gel capsules, the capsules had been removed and the inside content only has been homogenized. Each homogeneous sample was weighed 0.10 g in a 15 mL centrifuge tube by an analytical balance. en, 4.0 mL of acetonitrile: water (1 : 1, v/v) was added and mixed well before being sonicated for 30 minutes. e extract was centrifuged for 5 minutes at a speed of 6000 rpm. e solution was separated from the residue and filtered through a polytetrafluoroethylene filter (0.2 μm) before being injected into a liquid chromatography high-resolution tandem mass spectrometry (LC-HRMS). For the soft-gel capsule samples, 1.0 mL of n-hexane was added to the solution after centrifuging and mixed well to clean the oily components; only the aqueous layer was used for LC-HRMS analysis. For samples containing PDE-5i over the calibration curves, we have to reanalyze with a proper dilution factor.   5 , isolation window 1 m/z, normalized collision energy 40%, 70%, 100%. Full spectral information was utilized for identification and quantification. For data collection and analysis, the screening PDE-5i process was conducted by Compound discoverer 3.1 software ( ermo Fisher), and the quantitation process was conducted by TraceFinder 4.1 software ( ermo Fisher). Mass spectrometric information, including m/z of precursor and product ions of analytes, was shown in Table 1. It can be seen that the analogs of sildenafil produced the common ions at m/z 283. It is in line with previous studies that the ion is the result of cleavage of the C-S bond and loss of the ethyl group on the ethoxy substituent on the phenyl ring. For the tadalafil group, the ions at m/z 169 (pyridine-indole ring) and 135 were always recorded. e ion at m/z 344 is characterized for vardenafil and its analogs [7,17]. iosildenafil group often produces ion at m/z 299 corresponding to the cleavage of C-N bond and loss of the ethyl group on the ethoxy substituent on the phenyl ring [18].

Screening and Quantification of Real Samples.
Real samples were first screened PDE-5i as the scheme in Figure 1. Most of the detected PDE-5i were listed in our mass spectrometry library unless the new suspected compounds were extracted and purified, and then the structure was determined by infrared spectroscopy. For the quantitative purpose, the concentration of PDE-5i in the samples was calculated by matrix match calibration curves.

Optimization of LC-HRMS Condition.
First of all, parameters for detecting PDE-5i in Q-exactive mass spectrometer should be set up before further optimization. All PDE-5i have chemical structures suitable for being ionized by electrospray ionization source in positive mode. A 500 μL mixture of 1 μg/mL standard solution of PDE-5i was injected into the Q-exactive mass spectrometer to optimize ionization and detection conditions such as capillary voltage, the temperature of HESI, the temperature of ion transfer tube, S-lens level, maximum injection time, and automatic gain control. After that, the MS/MS data were recorded: full-scan mode for precursor ions and dd-MS 2 (data-dependent MS 2 ) for product ions (All ion fragmentation-AIF). e normalized collision energy (NCE) was 40, 70, 100%. e MS/MS data were compared with mzCloud Mass Spectral Library ( ermo Fisher Scientific) and the mass accuracy was less than 5 ppm, which meets the requirements of AOAC International. e MS/MS data of fifty-three standard PED-5i for identification was presented in Table 1. For the detection of PDE-5i without standard solutions, MS/MS information (Table S1) in the mzCloud Mass Spectral Library can be used.
PDE-5i are less polar compounds, so they can be analyzed by the C18 base chromatography column. Because of their similar structures, PDE-5i should be separated by a chromatography column that has a small particle size. erefore, we chose BEH C18 (100 mm × 1.7 μm × 2.1 mm, Waters, Milford, Massachusetts, USA). Commonly, mobile phases for PDE-5i analysis are acetonitrile: water and acetonitrile: methanol (1 : 1, v/v), adding additives such as formic acid, ammonium format, or both of them. We investigated and chose the mobile phase system including mobile phase A: 10 mM ammonium formate and 0.1% formic acid in water, and mobile phase B: 10 mM ammonium format and 0.1% formic acid in acetonitrile: methanol (1 : 1, v/v). e use of both ammonium formate and formic acid additives is important to gain the sensitivity of some PDE-5i and is consistent with AOAC 2015.12 method [15]. en, the gradient was optimized and lasted 29.0 minutes to separate some isomeric PDE-5i such as Carbodenafil and Noracetildenafil, Benzamidenafil, and Tadalafil. e flow rate was 0.3 mL/min. is slow and long gradient is similar to that of the reference methods published by AOAC International and US USP [19]. e retention time of each analyte was shown in Table 1. Extracted chromatograms of PDE-5i were shown in the supplementary document ( Figures S1, S2).

Optimization of Extraction.
Referring to previous studies [7,20], five extraction solutions were selected to examine extraction efficiency when extracting spiked samples at the concentration of 4 mg/kg in samples. e results of four representative compounds were presented in Figure 2. A oneway ANOVA test was conducted to compare the intensity of four compounds. e results (P value from 1.5E-10-4.9E-6 < 0.05) indicated that signal intensity changed significantly with different solvent extraction, and the mean comparison showed that the mixture of acetonitrile: water (1 : 1, v/v) gave significantly higher intensities of analytes compared to the other tested solvents. Comparing to the mixture of methanol: water (70 : 30, v:v) used by Lee et al. [17], Jeong et al. [21], or methanol used by Ren et al. [22], this method uses less organic solvent for a greener sample preparation. us, it was chosen to extract real samples.
For the soft-gel capsule sample, however, we added a second solvent to remove oily components of samples before injecting the extraction into the LC-HRMS system. ree organic solvents including n-hexane, diethyl ether, and ethyl acetate were examined, and the result was shown in Figure 3.

Journal of Analytical Methods in Chemistry
We also compared the intensity of analytes by one-way ANOVA test. All three cleaning ways improved the intensity of analytes. e intensity of analytes in oily samples washed by n-hexane was significantly higher than that of the other solvents. erefore, n-hexane was chosen for cleaning oily samples to reduce unwanted compounds injected into the LC-HRMS system. e use of n-hexane has not been reported before. It helps protect the ion source from fat contamination.

Method Validation.
e developed method had been validated before applied to real samples analysis. e specificity of the method was proved by mass accuracy of precursor ions and productions and comparison between blank samples and standard materials. e validation parameters were showed in Table S2. Calibration curves of PDE-5i were constructed from 10 to 1000 ng/mL, and the regression coefficients were larger than 0.995, and relative standard deviations were less than 15%. e limit of detection and limit of quantification were 0.4         e method met the AOAC International requirement as recoveries were in the range of 80-110%, and the relative standard deviation was from 2.81 to 12.6%. e matrix effect (ME) of the method was assessed by comparing the slope of the calibration curve in solution (A) and one in the matrix (A′) as follows: All the compounds showed ME less than 10%. us, calibration curves in the standard solution can be used to calculate the concentration of PED-5i in real samples.

Analysis of Real Samples.
For screening purposes, 92 real samples were analyzed by the developed method (n � 3); the screening process was conducted by Compound Discoverer 3.1 software. e spectrum was compared with the online mzCloud mass spectrometry library, Chemspider library, Mzvault library, and predicted structure. e accuracy of the process was presented by the matching index (>80%) with each library. Among 92 collected samples, twenty samples were detected containing PDE-5i; the others were not detected. In the positive samples, we identified thirteen PDE-5i already existing in the used libraries and one compound nonexisting in the used libraries. is compound was discovered as N-hydroxyethyl dithio-desethyl carbodenafil in a previous study [23]. e number of detected samples and identified PDE-5i was shown in Table 2. Ten of fourteen PDE-5i (71%) were sildenafil analogs, which is higher than the value (62%) reported by Kee et al. [4]. ere were three analogs of tadalafil (21%) and only one analog of vardenafil (7.1%).
After screening, positive samples were confirmed and quantified. e concentration of PDE-5i in these samples was calculated by calibration curves and presented in Table 3. It can be seen that most of the positive samples contain one or two PDE-5i at high concentration (>1 mg/g) and some other PDE-5i at low concentration. We suppose that high concentration PDE-5i ingredients were added intentionally to the sample, and the low concentration PDE-5i may be side products in the production of the main PDE-5i ingredients. Nortadalafil, Tadalafil, and Sildenafil were often detected in real samples as the main active compounds. On the other hand, few samples (S 14 and 15) detected some PDE-5i at low concentration (much lower than the dosage using in ED treatment). e origin of PDE-5i in these samples was not clear and needed to be studied further.

Conclusion
In this study, we have developed and validated a rapid screening and quantitation method using LC-HRMS for more than 53 PDE-5i in ingredients and supplementary products for enhancing male sexual performance. e validation parameters of this method, such as LODs, LOQs, recoveries, and regression coefficients, were acceptable according to the requirement of AOAC for an analytical method. e success of this method demonstrated the utilization of the fragmentation mass spectra library for analytes confirmation. e developed method was applied to analyze 92 natural-based ingredients and supplementary products available in the Vietnam market. We had screened and detected 14 PDE-5i. e results of real samples analysis implied that the manufacturers had deceived customers by not declaring PDE-5i on the label but adding these ingredients in the products. us, our study provides a warning on the quality control of supplementary foods to avoid any health risks to the community.

Data Availability
e data used to support the findings of this study are available within the article, the support information in word form, and from the corresponding author upon request.

Supplementary Materials
Supplementary information is provided in word form.