Liquid Chromatography–Electrospray Ionization Tandem Mass Spectrometry Estimation of Quercetin-Loaded Nanoemulsion in Rabbit Plasma: In Vivo–In Silico Pharmacokinetic Analysis Using GastroPlus

In the present study, we developed and validated a rapid, specific, sensitive, and reproducible liquid chromatography–electrospray ionization tandem mass spectrometry method for quantifying quercetin (QT) in rabbit plasma using hydrochlorothiazide as the internal standard. Animals were orally administered with optimized QT-loaded nanoemulsion (QTNE) and QT suspension (QTS), equivalent to 30 mg/kg, to the test and control group, respectively. The blood samples were collected at pre-determined time points up to 48 h. The linearity range was from 5 to 5000 ng mL–1 with R2 = 0.995. Further, we analyzed the various pharmacokinetic parameters and established the in vitro–in vivo correlation (IVIVC) of QTNE using GastroPlus software. The method was successfully developed and validated, and when applied for the determination of QT in rabbit plasma, it exhibited an increase in Cmax from 122.56 ng mL–1 (QTS) to 286.51 ng mL–1 (QTNE) (2.34-fold) and AUC0–48 from 976 ng h mL–1 (QTS) to 4249 ng h mL–1 (QTNE) (4.35-fold), indicating improved oral bioavailability QT when administered as QTNE. Statistical analysis revealed that the Loo–Riegelman method (two-compartmental method) best fitted the deconvolution approach (R2 = 0.998, SEP = 4.537, MAE = 2.759, and AIC = 42.38) for establishing the IVIVC. In conclusion, the established bioanalytical method and IVIVC studies revealed that QTNE is a potential carrier for the effective delivery of QT with enhanced oral bioavailability.


INTRODUCTION
Quercetin (QT, 3,3,4,5,7-pentahydroxyflavone) is a polyphenolic flavonoid that is abundantly found in various plant sources, including vegetables (e.g., onion and broccoli), fruits (e.g., apple and blueberry), and also in herbs. 1,2 QT exhibits potential antioxidant properties, leading to its diverse biological actions, including anticancer, antimicrobial, antiproliferative, anti-inflammatory, and neuroprotective effects and others. 3 Although QT exhibits various pharmacological actions, its applications are restricted due to its low mucosal permeability leading to low oral bioavailability (<17% in rats and ∼1% in humans). 4 The low bioavailability of QT could be due to its lipophilic behavior and its high affinity for both the Cyp450 (CYP3A) and the intestinal efflux pumps (e.g., P-gp and MRP2), abundantly found in the GIT epithelium. 5 Therefore, to improve the oral bioavailability of QT, it becomes essential to incorporate it within a stable nanocarrierbased delivery system with negligible or no systemic toxicity.
Several nanotechnology-mediated strategies such as nanoemulsions, liposomes, polymeric nanoparticles, and micelles for QT have been established to improve the solubility, permeability, and bioavailability. 6 Lipid-based nanotechnological strategies have shown potential effects in improving the therapeutic effects of encapsulated polyphenolic compounds, including QT. 6 Nanoemulsions (NEs), a type of lipid-based nanocarrier, are formed from the dispersion of two immiscible liquid phases (oil and water) that forms oil-in-water (o/w) or water-in-oil (w/o nanodroplets) systems stabilized with the amphiphilic surfactant or cosurfactant. 4 The compatibility between oil and surfactant and the encapsulated drug moiety plays a crucial role in maintaining the stability of the NE system. 7 The estimation of possible interactions (either negative or positive) between the drug and excipients exhibits a vital role during preformulation studies for the development of stable NEs. Moreover, the interplay between drugs and excipients significantly improves the drug's pharmacokinetic parameters by improving the pharmaceuticals. 8 Pharmacokinetic characteristics of drug molecules play a crucial role in understanding their in vivo performance and mechanism of action. Usually, the flavonoids are consumed orally and may be bio-transformed via intestinal microbiota and metabolized within the liver. 9−11 During this process, metabolites (in plasma) may produce similar, more potent, or weaker effects than those with the parent drug moiety. 12,13 Various analytical approaches have been developed and implemented to understand the drug release mechanism of QT in biological samples using several analytical methods, including high-performance liquid chromatography (HPLC) with UV detection, 14,15 fluorescence detection, 16 electrochemical detection, 17 and liquid chromatography−electrospray ionization tandem mass spectrometry (LC-MS/MS) analysis. 18−20 In another study, the researchers compared the pharmacokinetic parameters of QT with QT-3-O-β-glucuronide after oral administration (100 mg/kg dose) in rats using the UHPLC-MS/MS method. 21 In a recent study, the QTchitosan oligosaccharide-based amorphous dispersions exhibited improved oral bioavailability (1.64−2.25 times) compared to pure QT-treated male Sprague Dawley rats, evaluated by the HPLC method. 14 Herein, we developed and validated a rapid, specific, sensitive, and reproducible liquid chromatography−electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) method for quantifying QT in rabbit plasma after oral administration of quercetin-loaded nanoemulsion and quercetin suspension. A comparative analysis of pharmacokinetic parameters of QT from nanoemulsion and suspension and the IVIVC studies for QTNE was performed using the GastroPlus (Simulations Plus Inc., Lancaster, CA, USA) software.

Preparation of Quercetin-Loaded Nanoemulsion.
The optimized quercetin-loaded nanoemulsion (QTNE) was prepared as mentioned in our earlier published work. 4 The QTNE comprises QT (drug, 10 mg), CAP MCM NF (oil, 250 mg), and CR RH 40 (surfactant, 250 mg). The QTNE formulation was prepared by mixing the aqueous phase (surfactant + Milli Q water, 10 mL) into the oil phase (drug + oil) followed by stirring (IKA C-MAG HS 7, Germany) at 500 rpm. Further, the system was homogenized (IKA T25, ULTRA TURRAX, Germany) at 11,000 rpm for 20 min.  standard; Figure 1c,d), with a scan time of 200 ms per transition.

Preparation and Calibration of Standard Stock
Solutions. The stock solutions (100 μg mL −1 ) of QT and HCZ (internal standard) were prepared by dissolving 2.0 mg of each in methanol (20.0 mL). The standard calibration curves of QT in plasma samples were plotted in the 5−5000 ng mL −1 concentration range. Further, the quality control (QC) samples were prepared by diluting QT in the plasma and were represented as 11.356 ng mL −1 (low concentration; LQC), 1880.966 ng mL −1 (medium concentration; MQC), and 3761.932 ng mL −1 (high concentration; HQC).

Method Validation.
A total of seven calibration standards in a concentration range of 5−5000 ng mL −1 were selected for evaluating the linearity. The calibration curves were plotted for the y axis (peak ratio of QT and HCZ) vs the x axis (concentration of QT) using a weighted factor (1/x). The LLOQ represents the lowest concentration over the calibration curve and should have an accuracy (within ±20%) and precision (<20%).
For three independent days, accuracy and precision were investigated for the drug's LQC, MQC, and HQC concentrations. The inter-and intra-day precision were measured by percent relative standard deviation (%RSD). The accuracy was determined using eq 1: 19 where C m is the mean observed concentration and C s is the spiked concentration.
The extraction recovery of QT was estimated by correlating the QT samples' (LQC, MQC, and HQC) responses with the analyte's responses from the post-extracted plasma samples at equivalent concentrations. The recovery studies for QT were achieved at three concentration levels (11.2560, 1880.9660, and 3761.9320 ng mL −1 ) in the rabbit plasma. The matrix effects were studied at LQC, MQC, and HQC by relating the peak area of the treated plasma samples and spiked later to acquire three QC level concentrations (A P ) with the peak area of the standard solutions (A S ). Equation 2 was used to determine the matrix effect.
QT's freeze−thaw stability in the rabbit plasma was studied for three QC level concentrations (LQC, MQC, and HQC) after three freeze (−20°C)−thaw cycles. Analysis and comparison of the freeze-thawed samples and freshly prepared QC samples were made by monitoring the peak areas of the respective samples.

Pharmacokinetic Studies.
To perform the pharmacokinetic studies of QTNE and QTS, adult male New Zealand white rabbits (1.3 ± 0.8 kg body weight) were used. The animals were acclimatized for 7 days under standard conditions (25 ± 2°C, 60% RH) before conducting the study, and the necessary diet and water were provided ad libitum. The rabbits were fasted overnight (before experimentation) and were arbitrarily divided into two groups (n = 3): QTNE-treated animals (group I) and QTS-treated animals (group II). QTS was prepared by mixing an adequate amount of QT in an aqueous suspension comprising 0.5% w/v Na-CMC. The samples were orally administered to all animals with a 30 mg/ kg QT dose. The protocols for the study were followed as per the CPCSEA (Committee for Control and Supervision of Experiments on Animals) guidelines after approval from the Institutional Animal Ethics Committee, Department of Pharmaceutical Sciences and Technology, BIT, Mesra, Ranchi (1972/PH/BIT/42/18/IAEC*). The blood samples were withdrawn from the marginal ear veins from treated animals at time points of 0.5, 1, 2, 8, 12, 24, and 48 h, collected in heparinized tubes, and centrifuged at 3500 rpm for 10 min, and the corresponding plasma samples were stored at −20°C. The samples were later analyzed using the LC-ESI-MS/MS technique, and the process is illustrated in Figure 2.

ACS Omega
http://pubs.acs.org/journal/acsodf Article Further, the pharmacokinetic parameters including C max , t max , AUC 0−t , AUC 0−∞ , AUMC, t 1/2 , MRT, K a , K e , and microconstants were measured using the PKPlus module of GastroPlus software. Furthermore, different pharmacokinetic compartment models, including 1C, 2C, and 3C, were statistically analyzed, and the best-fit model was selected considering high R 2 and low SC and AIC values. 22 2.6. In Vitro−In Vivo Correlation Determination. The foremost goal behind the in vitro−in vivo correlation (IVIVC) study is to act as a substitute for in vivo studies. IVIVC studies impart potential benefits for the pharmaceutical industries as it saves time, improves cost efficiency, and assists in formulation optimization. IVIVC studies were established through several steps, including (i) generation of in vitro cumulative drug release data (in vitro); (ii) development of PDCT profiles (in vivo dosing); and (iii) generation of an in vivo dissolution data through numerical deconvolution of the PDCT profiles. Finally, the dissolution profile (in vivo) was compared with the experimental dissolution data (in vitro). 23 The IVIVC module of GastroPlus software was used to establish the correlation between in vitro dissolution study data, as reported in our previously published work, 4 and in vivo plasma concentration data for QTNE and QTS. The PDCT profile was studied by 1C, 2C, and 3C methods using the PKPlus module of GastroPlus software, and the relevant pharmacokinetic parameters were analyzed. Further, deconvolution methods were employed for establishing correlations, and based on the values of R 2 , SEP, and MAE, the best-fit correlation model was selected for establishing IVIVC. Additionally, convolution was performed and was assessed based on the statistical findings of MAPPE for two prime pharmacokinetic parameters, i.e., C max and AUC.  The calibration curve was linear in the 5−5000 ng mL −1 concentration range. It was plotted by keeping the peak area ratio of QT to HCZ (Y axis) vs QT concentration (X axis) with the calibration equation y = 0.0063x + 0.0592 (R 2 = 0.995). The LC-MS/MS representative chromatograms of plasma containing standard QT (Figure 4a) and HCZ ( Figure  4b) in rabbit plasma are illustrated. Further, the chromatogram of QT in plasma that was obtained after 2 h of oraladministered QTNE (30 mg/kg) is illustrated in Figure 4c and HCZ during sample analysis is illustrated in Figure 4d. The respective chromatograms exhibited no interference by endogenic plasma constituents. The LLOQ for QT was found to be 5.1440 ng mL −1 ; at the LLOQ, the precision and accuracy values were 2.38% and 100.32%, respectively. The inter-and intra-day results for the precision and accuracy are tabulated in Table 1. The accuracy (intra-day) of QT was in the range of 93.80−100.27%, with precision (%RSD) ≤4.96%. The accuracy (inter-day) of QT was in the range of 93.14− 99.90%, with precision (%RSD) ≤5.62%. The extraction recoveries of QT were estimated at three, 11.2560, 1880.9660, and 3761.9320 ng mL −1 , and the absolute mean extraction efficiency (%) of QT was in the range of 87.99− 90.62%, with %RSD ≤ 9.87 (

Pharmacokinetics Study.
The drug carriers usually move into the organs, in the initial phase, rather than going to the targeted sites, and further, they reach the targeted organs through the systemic circulation. The process of drug absorption from the site of administration plays a crucial role before its passage into systemic circulation. 22 The extent and rate of drug absorption depend upon various factors, including the route of administration, physiological conditions of the drug absorption site, and the mechanism of drug absorption. In oral administration, before moving into systemic circulation, the drug passes through various layers of epithelial cells of the GIT, leading to difficulties in the absorption of the drug. 24 In addition, poor solubility and/or low permeability of drugs leads to the problem of poor absorption and low bioavailability, and these limitations can be overcome by encapsulating these drugs within specific nanocarrier systems. The validated bioanalytical method was effectively employed to quantify the amount of QT in rabbit plasma. The pharmacokinetic parameters followed by oral administration of QTNE and QTS were determined using GastroPlus software. Finally, the results for the relative PDCT profile of QTNE and QTS are illustrated in Figure 5.
Additionally, the results of the plasma-drug concentration time profile of QTNE and QT suspension were evaluated for non-compartmental and compartmental analysis (1C, 2C, and 3C) using the PkPlus module ( Table 3).
The non-compartmental method does not undertake any specific compartmental model and provides precise results; thus, it is considered more versatile and is widely applied in bioequivalence studies. 25 The pharmacokinetic compartment modeling produces necessary information associated with the fate of a drug to time. The compartment modeling exhibits advantages compared to NCA in predicting the drug concentration at any specified time. The 1C pharmacokinetic model is the simplest compartmental pharmacokinetic model in which the whole organism is considered as a single compartment where the distribution of the drug is homogeneous and instantaneous ( Figure 6A). 26 On the other hand, in the 2C pharmacokinetic model (Figure 6B), the body is divided into two compartments: the central and peripheral. The first compartment (central compartment) comprises the plasma and tissues where the drug distributes almost instantaneously. The second compartment (peripheral compartment) comprises tissues where the drug distributes slowly. 27 In the 3C pharmacokinetic model, the body is divided mainly into central and two peripheral compartments.   The central compartment (compartment 1) is connected with the second and third compartments (peripheral compartments) where the drug distributes slowly as compared to the central compartment 28 ( Figure 6C). The correlation and fitness of the superposition among the true and simulated PDC (1C, 2C, and 3C models) and the corresponding compartment models (added as inserts) are illustrated in Figure 6. Generally, after absorption, the drug moieties vacate the administration site to enter the central compartment and are then released into the peripheral compartment followed by distribution and finally eliminated irreversibly. This approach to drug movement from one compartment to another is considered by transfer rate constants, termed micro-constants. 29 An upsurge in C max from 122.56 ng mL −1 (QTS) to 286.51 ng mL −1 (QTNE) (2.34-fold), AUC 0−48 from 976 ng h mL −1 (QTS) to 4249 ng h mL −1 (QTNE) (4.35-fold), and AUC 0−∞ from 992 ng h mL −1 (QTS) to 4337 ng h mL −1 (QTNE) (4.37-fold) confirmed a significant improvement in the oral bioavailability of QT when delivered using the NE system. In a pharmacokinetic study performed with male albino Wistar rats using the HPLC technique, C max of QTNE was 3.64-fold higher than pure QT; also, the QTNE exhibited a delayed t max compared to pure QT. 30 The results signified that QTNE improved pure QT's bioavailability and sustained release properties. In another study, a single dose of QTNE and pure QT (dispersed in 0.3% sodium carboxymethyl cellulose) was administered orally (40 mg/kg) to mouse models. The results showed that QTNE exhibited improved solubility and permeability with C max (10.04 ± 2.010 μg mL −1 ) values, which were 28.7-fold higher than the pure QT. 31 Studies have shown that the encapsulation of QT within polymeric coats has also helped increase its oral bioavailability. Penalva et al. reported that the value of C max of QT-loaded zein nanoparticles (Q-ZNP) was 2.5 times higher than the QT-PEG400/water solution studied in Wistar rats (25 mg/kg) using the HPLC method, ensuing an improved oral bioavailability of QT. 32 In a recent study, researchers reported that QT's oral bioavailability was significantly higher when encapsulated within the shell of zein-caseinate NPs (QZCNs) than caseinate-chitosan double layering (QZCCNs). The oral bioavailability of QT, studied in the rat model using the HPLC method, was enhanced in both QZCN-and QZCCN-treated groups by 2.34 and 1.89 times, respectively. 15 Shen and coresearchers analyzed pharmacokinetic parameters of QT hybrid nanocrystals (QHNs; 90 mg/kg) with varying sizes (280 and 550 nm) in plasma and tissue homogenates of treated male Sprague−Dawley (SD) rats using the HPLC technique. Both types of QHNs showed improved oral bioavailability of QT; however, QHNs-280 exhibited higher AUC 0−t (82.40 h μg mL −1 ) and C max (3.70 mg mL −1 ) than QHNs-550 with AUC 0−t (50.32 h μg mL −1 ) and C max (2.05 mg mL −1 ), showing higher oral bioavailability of QHNs-280. 33 The NE system mainly comprises oil and/or surfactant, which assists in enhancing the drug solubility and permeability, decreasing gastric degradation, and overcoming the issues of first-pass metabolism. 34 Also, the smaller size of NE allows them to permeate deep within the tissues and prolong their circulation, leading to improved bioavailability of the encapsulated drug/s. 35 In addition, NE protects the encapsulated drug/s from hydrolysis, oxidation, and volatilization. 36 Drug delivery is affected by various factors, including solubility, enzymes, pH of GIT, ionic strength, consumed food constituents, dissolution rate, absorption window, and residence time. 37 Sha et al. stated that the possible mechanism for the increased oral absorption of the QT-loaded NE was due to the surfactants used in NE formulation, which assisted the opening of tight junctions (paracellular pathway), whereas aqueous dispersion of QT showed precipitation in the cell monolayer. 38 Furthermore, pancreatic lipases specifically digest the oil/lipid, major compositions of NE, on the apical margins of enterocytes. Later, the solubilized QT may permeate through the epithelial cells of the intestine via passive diffusion. NEs (∼20 nm) might permeate directly through the intestinal membrane or be captivated into the enterocytes through caveola-or clathrin-facilitated macropinocytosis and enterocytosis. 39,40 Our study used CR RH 40 (ethoxylated hydrogenated castor oil) as a surfactant for preparing the NE system. 3,4 It is well

ACS Omega
http://pubs.acs.org/journal/acsodf Article known for its effective inhibitory effects against the P-gp efflux pump, leading to enhanced absorption of the encapsulated drug. 41 MRT is described as the arithmetic mean of the total time taken by a drug to remain in the body before elimination.
It is an important pharmacokinetic parameter as it provides exact information regarding the existence of a drug molecule within the body, i.e., some drug molecules last for a concise period, and others last longer. 42 In our study, the MRTs of QTNE and QTS were found to be 12.36 and 8.79 h, respectively (  the correlation between dosage forms in vitro drug release profile and significant in vivo response. It is an important design component for modified-release dosage forms. 43 A dosage form's in vitro and in vivo properties are correlated and represented mathematically through various linear or/and nonlinear methods. 23 Generally, the linear method is represented mathematically by using the convolution or/and deconvolution approaches. The numerical convolution or/and deconvolution approaches are usually considered because they do not create any pharmacokinetic model presumptions. Furthermore, the K a (in vivo) can be estimated using a compartmental approach if the pharmacokinetic parameters of the drug are known. 44 In the present study, the IVIVC model was established using the in vitro dissolution (reported in our previously published work) 4 and in vivo pharmacokinetic profile of QTNE and was analyzed using GastroPlus software. The correlation between the parameters on the Y axis (IVIVC fit, fraction bioavailability, AUC, and plasma concentration) and X axis (fraction in vitro release) for QTNE is illustrated in Figure 7. The IVIVC data were fitted using the third-order polynomial (eq 3) where x and y signify the fraction in vitro release and fraction absolute bioavailability of the drug, respectively. Furthermore, the correlation function is related to the Loo− Riegelman (2C model) with percent prediction error (%PE) between the observed (Obs.) and predicted (Pred.) values of C max and AUC 0−t . The statistical analysis of reconstructed PDCT profiles from the convolution data is represented as R 2 , SEP, MAE, and AIC. The %PE was calculated as per eq 4 22 The observed and predicted values for C max of QTNE were 287 and 256 ng mL −1 , respectively, with a %PE of 10.80. A similar method was used to determine the observed and predicted values for AUC 0−48 of QTNE, and they were found to be 4020 and 3897 ng h mL −1 , respectively, with a %PE of 3.06%. Based on the statistical analysis of reconstructed PDCT profiles indicating a high value of R 2 (0.998) and low values of SEP (4.537), MAE (2.759), and AIC (42.38), the Loo− Riegelman method (2C model) was selected as the best-fit deconvolution model. The IVIVC exhibited a similar approach as that of the pharmacokinetic profile of QTNE, showing the 2C model as the best-fit model.

CONCLUSIONS
A rapid, specific, sensitive, and reproducible LC-ESI-MS/MS method was developed and validated to determine QT in rabbit plasma. HCZ was employed as an internal standard, and the linearity was determined in a concentration range of 5− 5000 ng mL −1 , with R 2 = 0.995. The higher levels of C max and AUC 0−t of the QTNE signified an improved oral bioavailability (2.34-fold) of QT compared to QTS, probably due to enhanced solubility, absorption, and residence time of QT. GastroPlus simulation software showed superior prediction accuracy used to estimate various pharmacokinetic parameters (compartmental and non-compartmental methods) of the developed QTNE. IVIVC was established using the IVIVC module of GastroPlus software. Analysis of the statistics generated from various IVIVC approaches revealed that Loo− Riegelman (2C model) was the best-fit model for the IVIVC of QTNE with R 2 = 0.998. The study revealed that the QTNE is a potential nanocarrier for effectively delivering QT with improved oral bioavailability. The established bioanalytical method can be efficiently employed in preclinical/clinical studies to quantify QT in animal plasma.