Systematic Development and Validation of a Thin-Layer Densitometric Bioanalytical Method for Estimation of Mangiferin Employing Analytical Quality by Design (AQbD) Approach

Abstract

The present work aims at the systematic development of a simple, rapid and highly sensitive densitometry-based thin-layer chromatographic method for the quantification of mangiferin in bioanalytical samples. Initially, the quality target method profile was defined and critical analytical attributes (CAAs) earmarked, namely, retardation factor (R_f), peak height, capacity factor, theoretical plates and separation number. Face-centered cubic design was selected for optimization of volume loaded and plate dimensions as the critical method parameters selected from screening studies employing D-optimal and Plackett–Burman design studies, followed by evaluating their effect on the CAAs. The mobile phase containing a mixture of ethyl acetate : acetic acid : formic acid : water in a 7 : 1 : 1 : 1 (v/v/v/v) ratio was finally selected as the optimized solvent for apt chromatographic separation of mangiferin at 262 nm with R_f 0.68 ± 0.02 and all other parameters within the acceptance limits. Method validation studies revealed high linearity in the concentration range of 50–800 ng/band for mangiferin. The developed method showed high accuracy, precision, ruggedness, robustness, specificity, sensitivity, selectivity and recovery. In a nutshell, the bioanalytical method for analysis of mangiferin in plasma revealed the presence of well-resolved peaks and high recovery of mangiferin.

Introduction

Mangiferin, a polyphenolic C-glucosylxanthone, primarily exists as a principal phytoconstituent in the leaves and stem bark of Mangifera indica (family Anacardiaceae). Chemically, 2-C-β-d-gluco-pyranosyl-1,3,6,7-tetrahydroxyxanthone (Figure 1) exhibits moderate aqueous solubility, i.e., 1.44 mg/mL and low lipophilicity (log P) of −0.59 (1, 2) predicted by in silico simulations performed on mangiferin using ADMET Predictor (Version 7.1.0013; Simulations Plus, Inc., USA) and GastroPlus Simulation software (Version 8.6; Simulations Plus, Inc.). Presence of the phenolic xanthone moiety owes to the powerful antioxidant activity of mangiferin for scavenging free radicals and protection against ROS-induced oxidative stress (3, 4). Other potential therapeutic applications of mangiferin include being used as an antidiabetic (5), antiobesitic, antiosteoclastogenic, antiasthmatic (6), antidiarrhoeal (7), immunomodulator, analgesic, antiallergic, antibacterial (5, 8), antimicrobial (4), antiviral (9, 10) and anticancer (3, 11–13) agent. Considering the number of its therapeutic benefits, mangiferin has proved to have promising utility in clinical treatment and management of multiple disorders.

Figure 1.

Chemical structure of mangiferin.

Development of a simple, rapid and highly sensitive bioanalytical method is highly desirable for quantification and routine chromatographic analysis of mangiferin in pharmaceutical formulations, and bioanalytical and toxicological samples. Several literature reports have been documented for quantification of mangiferin employing techniques like UV–Vis spectroscopy (1, 14), spectrofluorimetry (15), thin-layer chromatography (TLC) (16, 17), high-performance liquid chromatography (HPLC) (18–20), high-performance thin-layer chromatography (HPTLC) (21), liquid chromatography–mass spectrometry (LC–MS) (22, 23), tandem mass spectrometry (LC–MS-MS) (24, 25) in the bulk drug (26–30) as well as in bioanalytical samples like rat plasma (31, 32), urine (33) and aqueous humor (34, 35). However, none of these methods have been considered to be highly satisfactory owing to the involvement of complex aqueous and organic solvent mixtures, and monitoring of the complex chromatographic conditions.

Consequently, in order to obviate the tedious, expensive and prolonged sample preparation, use of intricate mobile phase compositions and optimization of instrumental method variables associated with the conventional HPLC procedures, HPTLC and TLC techniques have been considered highly useful for ease of quantification and faster analysis of drugs, requiring minimal expenditure of developmental effort, time and resources. Use of a densitometry-based TLC method during analytical method development offers multiple advantages over HPLC, like lower mobile phase consumption, higher scanning speed, shorter analysis time, reduced sample cleanup, lower operational and maintenance costs, and, above all, ease of detection and optimization of critical method variables, thus facilitating faster analysis of drug by saving a great deal of time and resources (36). Also, it provides the leverage of avoiding tedious sample preparation, liquid–liquid extraction and filtration steps encountered during HPLC analysis. Besides, other stellar merits of TLC techniques encompass the possibility of selecting corrosive solvents with high pH range, minimal chances of sample contamination with the previous ones, high sensitivity from the nanogram to pictogram level and highly reproducibility of the densitometric method for identifying the concentration of the sample (37).

Of late, the use of Quality by Design (QbD) paradigms has been permeating the industrial, academic and federal domains at an alarming pace with the objective to gain a holistic process and product understanding. Implementation of the Analytical Quality by Design (AQbD) approach, in particular, has been highly popularized for developing robust methods by specifically understanding the critical method parameters (CMPs) influencing method performance (38, 39). Considered as a science and risk-based approach, AQbD provides rational comprehension of the CMPs affecting the critical analytical attributes (CAAs) of the bioanalytical method by risk assessment and factor screening studies, followed by unearthing the plausible interactions among them and optimizing them by employing Design of Experiments for enhanced method performance (40). The AQbD approach of method development primarily involves defining the quality target method profile (QTMP) and CAAs, identifying those CMPs that are significantly influencing the CAAs using prioritization and screening studies, method optimization using suitable experimental designs, modelization and optimum search through response surface methodology to embark upon the analytical design space and postulation of control strategy for continuous improvement in method performance (41). In the past few years, several literature reports have successfully demonstrated the immense utility of the AQbD approach for developing efficient and cost-efficient liquid chromatographic methods for estimating different analytes in pharmaceutical formulations and bioanalytical samples (42–48). Application of AQbD specifically in the current thin-layer densitometry method development envisioned for identifying method variables plausibly associated with high degree of variability and exhibiting maximal influence on method performance.

The present studies were accordingly undertaken to develop an AQbD-enabled, simple, ultra-fast, robust, sensitive, effective and economical densitometry-based TLC method for quantification of mangiferin in the bioanalytical samples.

Materials and methods

Standards and reagents

Mangiferin was purchased from M/s Sigma-Aldrich Co., Mumbai, India, and was used as working standard in the concentration of 1 µg/mL. Analytical grade ethyl acetate, acetic acid, formic acid and methanol were purchased from M/s Merck Ltd., Mumbai, India, and were used as received without further purification.

Preparation of standard solutions

The stock solution was prepared by weighing accurately 10 mg of mangiferin in a 10 mL volumetric flask followed by dilution in methanol to obtain the concentration of 1,000 µg/mL. This stock solution was further appropriately diluted with methanol in a 10 mL volumetric flask to prepare the final concentration of 1,000 ng/mL.

Biological matrix

Fresh human blood was procured ex gratis from the Rotary Blood Bank, Chandigarh, India. Harvested plasma samples were further employed for the entire method development and validation studies. For preparation of calibration standards and quality control samples, human plasma was extracted by centrifugation of blood treated with ethylenediaminetetraacetic acid at 10,000 r.p.m. (5,590 × g) using a centrifuge (Remi, Mumbai, India).

Instrumentation and chromatographic conditions

Bioanalytical method development for mangiferin was carried out using a Camag HPTLC system equipped with Linomat-V semiautomatic sample applicator. Suitable volumes of standard and sample solutions were applied in triplicate using a Hamilton syringe (100 mL) with slit dimensions of 6 mm × 0.30 mm, scanning speed of 20 mm/s and data resolution of 100 µm/step, equipped with an optical filter (second order) and filter factor (Savitzky golay 7). Chromatographic separation was achieved on a precoated silica gel aluminum TLC plate 60 (10 × 10 cm; Lot No. HX 130,195; Product Catalog No. 1.05554.0007; E. Merck, Darmstadt, Germany) using a Camag [CAT No. 022.7400, Ser No. 130,513, Muttenz, Switzerland (Anchrom Enterprises (I) Pvt. Ltd., Mumbai, India)]) flat bottom and twin-trough TLC developing chamber. The position of sample application was fixed at 10 mm from the bottom and 10 mm from the side edges with 10 mm as the distance between tracks in the form of bands, with each band of length 6 mm on a precoated silica gel aluminum plate 60 (10 × 10 cm, 100 µm thickness) using a Camag Linomat sample applicator.

The suitable mobile phase composition was employed consisting of a mixture of organic and aqueous solvents like ethyl acetate : acetic acid : formic acid : water. Mobile phase components were mixed prior to use and the development chamber was left to saturate with mobile phase vapors for suitable time before each run. The development of the plate was carried out by the ascending technique to a migration distance of 80 mm. After development, aluminum-backed silica gel TLC 60 plates were dried in a hot air oven at 25°C. Densitometric scanning of mangiferin was performed at 262 nm (deuterium lamp) with a Camag TLC Scanner III in the remission/absorption mode operated by a user-friendly WinCATS software (version 1.4.2) (Muttenz, Switzerland).

Defining the QTMP and CAAs

As per the AQbD approach for the systematic development of a bioanalytical method of mangiferin, the QTMP was defined encompassing the summary of the quality characteristics of the targeted method. Table I enlists the QTMP elements setup for developing an efficient thin-layer densitometric method with enhanced chromatographic separation of mangiferin. In order to meet the QTMP, various CAAs were earmarked, namely R_f, peak height, capacity factor, theoretical plates and separation number. Table II enumerates the list of CAAs considered during method development along with rational justification for each of them.

Table I.

QTMP for Bioanalytical Thin-Layer Densitometry Method of Mangiferin

QTMP elements	Target	Justification
Mangiferin	Active ingredient/antioxidant	For quantitative estimation in diverse biological samples from pharmacokinetic, clinical and toxicological studies.
Method type	Liquid densitometry	Liquid densitometric method is primarily desirable for attaining chromatographic separation of mangiferin owing to its hydrophobic nature. Due to the partitioning phenomenon, the drug tends to retain easily on the stationary phase composed of silica, while the latter tends to provide an adsorbent effect for efficient separation owing to its narrow particle size, small pore volume and high surface area.
Instrument requirement	HPTLC system equipped with a scanner	It facilitates measuring the absorbance or fluorescence from the test substances at a very wide wavelength ranging between 200 and 800 nm. It allows detection of 31 wavelengths following integration of densitometric data for simultaneous estimation.
Sample characteristic	Liquid	It is mandatory to prepare the samples in liquid form for application on the stationary phase matrix, which tends to migrate on it for quantitative estimation of the sample concentration based on the densitometric evaluation.
Biological sample preparation	Handling, weighing, sampling, admixing with solvents	Plasma sample preparation is carried out by spiking the accurate volume of bioactive compound each time and mixing with a fixed amount of plasma to obtain the stock followed by ultracentrifugation and separation of the supernatant.

Table II.

CAAs for Bioanalytical Thin-Layer Densitometric Method of Mangiferin

CAAs	Target	Justification
Retention/retardation factor (Rf)	Optimum (between 0.5 and 0.7)	Rf is a function of the partition coefficient and is considered to be constant for a drug substance under a given set of chromatographic conditions. Usually, Rf is always less than unity. Larger Rf reveals the non-polar nature of the compound, thus shortening the time required for separation. Thus, intermediate Rf value is desirable for attaining optimal separation and, hence, deemed as highly critical.
Peak height	High	For most chromatographic analysis, peak height is particularly employed for quantitative estimation of the analyte. It is more vital as peaks may undergo tailing owing to diverse factors, which cause variation in the area count leading to weird results. In this context, peak height was considered as highly critical for estimation of the target compound.
Capacity factor	High	Capacity factor is best described as a measure of the retentivity of the analyte, which can be used to assess the column's efficiency. The longer a component is retained by the column, the greater is the capacity factor. Hence, it was assumed to be highly critical for the target analytical method.
Theoretical plates	High	Theoretical plate count measures the efficiency of both the method and column. The more the number of theoretical plates the higher is efficiency of the method. Thus, it is considered as highly critical for the chromatographic analysis.
Separation number	High	Separation number provides the basis for evaluating the separation capacity of a chromatographic method. It describes the number of zones that can be separated with a resolution of 4 s. It must be as high as possible to depict the separation ability of the method for a mixture of analytes. Hence, it indicates high method efficiency, and is considered as one of the critical parameters.

Preparation of bioanalytical samples

To prepare the standard calibration curve and quality control samples, an aliquot of 200 µL of the blank plasma was added to different diluted working solutions (200 µL) in a centrifuge tube. The contents of the tube were subjected to vortex mixing for 30 s in order to ensure complete mixing.

Preliminary screening studies

Preliminary screening studies were carried out for identifying suitable solvent systems for apt chromatographic separation of the mangiferin. As reported in the literature, diverse combinations of organic and aqueous solvents were explored as the potential mobile phase(s), and analytical trials were performed. After thoroughly scrutinizing various reports in the literature, rational combinations of ethyl acetate : methanol (49), toluene : acetone : glacial acetic acid (50), ethyl acetate : glacial acetic acid : formic acid : water, methanol : ethyl acetate : glacial acetic acid, ethyl acetate : methanol : toluene, ethyl acetate : methanol : chloroform and hexane : methanol : ethyl acetate : glacial acetic acid (51) were identified as the suitable mobile phases for estimating mangiferin in plasma. At a fixed concentration of the analyte (1,000 ng/band), all the mobile phases were visually examined for band migration distance, peak tailing, R_f, migration distance of the solvent front for ruling out the less pragmatic mobile phases and drawing a final inference to select the best solvent system with higher resolution for mangiferin.

AQbD-based method development and optimization studies

As per the AQbD approach of method development, endeavor was made to carry out primary and secondary screening studies for identifying the CMPs critically influencing the method CAAs, and subsequently optimizing the same employing suitable experimental designs.

Primary parameter selection

Selection of apposite combinations of the solvents for the mobile phase is highly desirable for efficient chromatographic analysis and migration of the analyte in the stationary phase. The solvent system exhibiting higher affinity for separation of the analyte was used to identify the optimum ratio of the solvents for the selected mobile phase. In the present studies, a D-optimal design was employed for selecting the apt combination of the ratios of ethyl acetate, glacial acetic acid and formic acid as the mobile phase mixture for separating mangiferin, and evaluating the method CAAs like R_f and peak height. Table III illustrates the design matrix depicting the experimental runs obtained as per the D-optimal design along with the actual and coded values of the levels of CMPs.

Table III.

Design Matrix as per the D-Optimal Response Surface Design Illustrating the Experimental Runs

Experimental run	Ethyl acetate	Acetic acid	Formic acid	Water
1	6	1	1	2
2	6	1	2	1
3	6	2	1	1
4	7	1	1	1
5	8	0	1	1
6	6.5	0.5	1	2
7	6	2	1	1
8	7	0	2	1
9	6.25	1.25	1.25	1.25
10	6	0	3	1
11	7	0	1	2
12	7	0	2	1
13	7.5	0	1	1.5
14	6.5	0	1.5	2
15	6	1	2	1
16	6	0.5	1.5	2
17	7	1	1	1
18	6	0	3	1
19	6.25	0.25	2.25	1.25
20	6	0	2	2
Name of the factors	Low		High
Translation of coded levels in actual units
Ethyl acetate	6		8
Acetic acid	0		2
Formic acid	1		3
Water	1		2

Secondary parameter selection

Following primary parameter selection, the secondary parameters influencing method performance were selected. An 11-factor 12-run Plackett–Burman design was employed for screening studies for identifying the “vital few” CMPs influencing the method CAAs viz. R_f, peak height, capacity factor, theoretical plates and separation number. Table IV illustrates the design matrix enlisting the studied factors along with the translation of their coded low and high levels into actual units. Model development was carried out by fitting the experimental data to the linear polynomial equations by obviating the interaction term(s) (52). The quantitative factor effects and their statistical levels were analyzed employing the Pareto charts, based on the principle of “factor sparsity” (53, 54).

Table IV.

Design Matrix as per the 11-Factor 12-Run Plackett–Burman Design for Screening of Method Variables and Process Parameters at Their Respective Low and High Levels

Runs	Mobile phase ratio	Saturation time	Volume -loaded	Developmental phase	Scanning speed	Slit dimension	Distance between tracks	Migration distance	Application position	Stationary phase	Plate dimension
1.	1	1	1	−1	−1	−1	1	−1	1	1	−1
2.	−1	1	1	1	−1	−1	−1	1	−1	1	1
3.	−1	1	−1	1	1	−1	1	1	1	−1	−1
4.	1	1	−1	1	1	1	−1	−1	−1	1	−1
5.	1	1	−1	−1	−1	1	−1	1	1	−1	1
6.	−1	−1	−1	−1	−1	−1	−1	−1	−1	−1	−1
7.	−1	1	1	−1	1	1	1	−1	−1	−1	1
8.	−1	−1	−1	1	−1	1	1	−1	1	1	1
9.	1	−1	−1	−1	1	−1	1	1	−1	1	1
10.	1	−1	1	1	−1	1	1	1	−1	−1	−1
11.	1	−1	1	1	1	−1	−1	−1	1	−1	1
12.	−1	−1	1	−1	1	1	−1	1	1	1	−1
Name of the factors						Levels
						Low (−1)			High (+1)
Mobile phase ratio						6 : 1.5 : 1.5 : 1			8 : 0.5 : 0.5 : 1
Saturation time (min)						2			4
Volume loaded (µL)						2			6
Developmental phase (mL)						10			15
Scanning speed (mm/sec)						10			15
Slit dimension (mm)						5			6
Distance between tracks (mm)						5			10
Migration distance (mm)						70			80
Application position (mm)						5			10
Stationary phase						Silica-coated aluminum-based 60 plates			Silica-coated aluminum-based 60F254 plates
Plate dimension (cm)						10 × 10			20 × 10

Method optimization studies

Based on the preliminary screening studies, followed by primary and secondary parameter selection, the CMPs actually affecting method performance were embarked upon for method optimization. A face-centered cubic design (FCCD) with degree of rotatability (i.e., α = 1) was employed for optimization of the CMPs, viz. volume loaded (X₁) and plate dimension (X₂), at three different levels, i.e., low (−1), intermediate (0) and high (+1) levels. Table V outlines the design matrix as per the FCCD with 13 experimental runs including quintuplicate studies at the center point runs (0, 0). The standard concentrations of 200, 400 and 600 ng/band were used for all the experimental runs, for the selected CAAs as per the design.

Table V.

Design Matrix as per the Face-Centered Cubic Design (FCCD) Employed for Thin-Layer Densitometry Method Optimization

Trial no.	Coded factor levels
	Factor 1	Factor 2
1.	1	−1
2.	1	1
3.	0	0
4.	0	0
5.	0	0
6.	−1	1
7.	0	−1
8.	−1	−1
9.	1	0
10.	0	0
11.	0	1
12.	−1	0
13.	0	0
Coded level	−1	0	1
Translation of coded levels in actual units
Volume loaded (µL)	2	4	6
Plate dimension (cm)	10 × 10	15 × 10	20 × 10

Optimization data analysis and validation studies

The optimization data analysis was carried out by multiple linear regression analysis using Design Expert 9.0 software (M/s Stat-Ease Inc., Minneapolis, USA) to fit the experimental data into the second-order quadratic polynomial model with added interaction terms, followed by establishing the relationship among the studied CMPs with the chosen CAAs. Only the coefficients with high statistical significance (P < 0.001) as per analysis of variance were considered in framing the polynomial equations for the CAAs, while evaluating the model aptness by lack of fit analysis, coefficient of correlation (r) and predicted error sum of squares (PRESS). Response surface analysis was carried out employing 3D response surface plots and 2D contour plots to discern the factor–response relationship and plausible interaction(s) among them. Search for the optimum solution was carried out using a numerical optimization desirability function by “trading-off” various CAAs as per the acceptance criteria, i.e., optimum R_f, and maximization of peak height, capacity factor, theoretical plates and separation number, to obtain efficient method performance. On the heels of numerical optimization, graphical optimization was also carried out to embark upon the analytical design space for locating the optimized solution.

Method validation

It is imperative to validate a bioanalytical method meant for analysis of drug molecules in the biological samples for ensuring efficient chromatographic separation and recovery. Thus, the developed method was validated as per FDA guidance for validation of a bioanalytical method (55–58).

Preparation of calibration standards and linearity range

The calibration samples were prepared by spiking mangiferin with a fixed amount of plasma. Extraction of mangiferin was established upon the principle of liquid–liquid extraction, i.e., relative solubility of the solute in the mobile phase, usually a mixture of organic solvents and water. Linearity of the developed method was determined by analyzing serial dilutions of mangiferin between the concentration range of 50 and 1,000 ng/band, and plotting the peak height versus concentration to obtain a linear correlation plot. Further, linearity of the method was confirmed by least square regression analysis on the obtained data by comparing the predicted and observed responses at 95% confidence intervals.

Accuracy

Accuracy of the developed method was estimated as mean percentage of recovery from four different quality control samples containing 100 ng/band of mangiferin and a fixed amount of plasma (200 µL), followed by spiking with 50% lower quality control (LQC), 100% medium quality control (MQC) and 150% higher quality control (HQC) of additional amount of mangiferin. Samples were subjected to chromatographic analysis to estimate the peak height, the values of recovery and % Relative Standard Deviation (RSD) to validate whether the accuracy of data is falling within the acceptance limits.

Precision

Precision was assessed by analyzing three different concentrations of mangiferin each (LQC: 100, MQC: 150 and HQC: 200 ng/band) at different time intervals on the same day (i.e., intra-day precision or repeatability), and repetition on the next day (i.e., inter-day or intermediate precision). All the samples were subjected to analysis and the peak height was noted for calculating the standard deviation (SD) and %RSD to investigate the accuracy of data within the acceptance limit.

Limit of detection and limit of quantification

Limit of detection (LOD) and limit of quantification (LOQ) were determined from the slope (S) of the linear calibration plot and the SD of the response to the blank sample (σ), as per the following formulae:

$$\text{LOD}=3.3\times \frac{\sigma }{S},\text{LOQ}=10\times \frac{\sigma }{S}$$ (1)

Specificity

The specificity of the method was determined in relation to confirm the absence of any interference of the components of the biological matrix with the chromatogram of the drug substance. For this purpose, a comparison of the chromatographic peaks of the standard sample tracks (i.e., at LOQ: 100 ng/band) and the tracks representing mangiferin in plasma preparations were performed and compared to check the prevalence of any interference (59).

Repeatability

Studies were performed to investigate the influence of different plates, accuracy of the applicator, effect of the developing phase and operator on the chromatographic separation of mangiferin performed employing different quality control samples. These samples at four different concentration levels (50, 100, 200 and 400 ng/band) were applied in triplicate in 12 tracks on 3 different TLC plates on the same day. The repeatability of the method was considered to be acceptable if all the bioactive compounds on each plate were identical with respect to position, intensity and formation of parallel lines on the chromatogram. It was seen that the R_f value for each of these bioactive compounds on three plates did not vary by more than 0.02 (59).

System suitability

System suitability was assessed by hexaplicate analyses of standard concentration of mangiferin 1,000 ng/band, followed by estimation of the SD and %RSD for peak height and retention time.

Robustness

Robustness of the method was evaluated by analyzing the system suitability parameters after alteration in the volume of the mobile phase (±0.5%v/v), scanning speed (±10%), volume loaded (±0.2%) and saturation time (±10 s). A working standard of 1,000 ng/band was used during the experimentation for assessing the values of mean percentage of recovery, and %RSD.

Ruggedness

The ruggedness of the method was determined by chromatographic analysis of the three different concentrations of the mangiferin (i.e., LQC: 100 ng/band, MQC: 150 ng/band and HQC: 200 ng/band) by two different analysts. The obtained data were analyzed for calculating the mean percentage of recovery and %RSD to corroborate the method ruggedness.

Matrix effect

The matrix effect was investigated by comparing the heights of the analyte in spiked quality control samples with or without biological matrix, at three different concentrations (i.e., LQC: 100 ng/band, MQC: 150 ng/band and HQC: 200 ng/band) by fixing the concentrations of the biological matrix. The corresponding peak heights of the analyte in spiked QC samples in plasma (A) were compared with those of the aqueous standards in the mobile phase (B) at equivalent concentrations. The matrix effect was calculated as per the following equation:

$$\text{Matrix}\text{effect}=\frac{A}{B}\times 100$$ (2)

Results

Preliminary screening studies

The preliminary screening studies were aimed at identifying the suitable mobile phase composition for efficient chromatographic separation of mangiferin. In this regard, various combinations of solvent systems were explored based on extensive literature survey. At a fixed concentration of the analyte (100 ng/band), the peaks obtained at all the explored mobile phases were visually examined for band migration distance, peak tailing, R_f, migration distance of the solvent front, etc. Table VI enlists various mobile phase combinations employed for obtaining “the best possible” resolution of mangiferin and the observations are made therein. Among the eight mobile phase combinations studied, ethyl acetate : glacial acetic acid : formic acid : water was found to be ideal one for efficient chromatographic separation of mangiferin. From the findings, it was very well deduced that being polar in nature, both precoated silica plates as well as mangiferin, require more polar composition of the mobile phase to resolve mangiferin better. Thus, the selected mobile phase composition was finalized for the entire method development study.

Table VI.

Comparative Preliminary Screening of the Various Mobile Phase Mixtures for Chromatographic Separation of Mangiferin

Developmental phase	Ratio (v/v)	Rf	Inference drawn
Ethyl acetate : methanol	4 : 6	0.12	Poor resolution of bands with a high solvent front
Ethyl acetate : methanol (with four drops of glacial acetic acid)	4 : 6	0.15	Solvent front decreased with no significant improvement in the resolution
Toluene : acetone : glacial acetic acid	6 : 3 : 1	0.14	Poor resolution of the bands
Methanol : ethyl acetate : glacial acetic acid	5 : 4 : 1	0.46	Better resolution of the bands with presence of tailing
Ethyl acetate : methanol : toluene	6 : 3 : 1	0.15	No apt resolution of the bands
Ethyl acetate : methanol : chloroform	6 : 3 : 1	0.17	No apt resolution of the bands
Hexane : methanol : ethyl acetate : glacial acetic acid	2 : 2 : 5 : 1	0.21	Low band resolution with tailing
Ethyl acetate : glacial acetic acid : formic acid : water	5 : 2 : 2 : 1	0.56	Efficient band resolution with improved Rf

Selection of primary parameters

After identifying the ideal mobile phase, the next step was to optimize the suitable ratio of the solvents in the mobile phase employing the D-optimal design for attaining superior chromatographic separation of mangiferin. A total of 20 experimental runs carried out as per the selected design were fitted to the polynomial equations for each of the CAAs. A quadratic model was employed for analysis of the individual responses, followed by framing the polynomial model by considering the model terms with highly significant statistical difference (P < 0.001). Evaluation of the suitability of the selected model was confirmed from higher values of coefficient of correlation (ranging between 0.8006 and 0.9362), insignificant values for the lack of fit (ranging between 0.3863 and 0.8978) and lower values of PRESS (ranging between 0.091 and 0.17; 1.239E + 007).

The rational understanding of the factor–response relationship was established by the response surface analysis. Figure 2 illustrates the 3D response surface plot for the CAAs, R_f and peak height. Among the studied combinations of the mobile phase solvents, higher influence of the concentrations of ethyl acetate was observed on the R_f, while both the concentrations of ethyl acetate and acetic acid exhibited higher influence on the peak height. Based on the above observations, the search for the optimum mobile phase combination was identified employing numerical optimization to meet the desired objectives for the CAAs, i.e., optimum value of R_f and maximizing the peak height. The optimized chromatographic solution was observed using ethyl acetate : acetic acid : formic acid : water in the ratio of 7 : 1 : 1 : 1.

Figure 2.

3D response surface plot using D-optimal design depicting the influence of the mobile phase ratio on CAAs, i.e., R_f and peak height, where (A) ethyl acetate, (B) acetic acid, (C) formic acid and (D) water. This figure is available in black and white in print and in color at JCS online.

Selection of secondary parameters

Factor screening studies were carried out for selecting secondary parameters using the Plackett–Burman design. The first-order polynomial models for estimating the main effect(s) was performed by analysis of coefficients as per the following equation, where β₁ to β_n represent the coefficients of the model terms, and β₀ represents the intercept term.

$$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+{\beta }_3{X}_3+{\beta }_4{X}_4+{\beta }_5{X}_5+{\beta }_6{X}_6+{\beta }_7{X}_7$$ (3)

The evaluation of polynomial equations generated for each CAA selected during screening studies indicated the absence of any significant interaction effect(s) among the factors. Table VII enlists the frequency of influence of various studied factors identified from the Pareto charts on the method CAAs. The factors, i.e., volume loaded and plate dimension, showed statistically significant influence on all the studied CAAs, as these were found to be above the t-value limit and Bonferroni limit with maximum frequency. On the contrary, other factors studied in the design revealed miniscule/less influence, and thus were fixed as constant for further method development studies. Based on the critical observations on the highly influential factors obtained from the Pareto charts, the volume loaded and plate dimension were finally selected as the CMPs for optimizing the method performance.

Table VII.

Frequency of Various Studied Factors Falling Above t Value and Depicting Their Influence on the Method CAAs

CAAs	CAAs
	Rf	Peak height	Capacity factor	Theoretical plates	Separation time
Mobile-phase concentration	–	✓	–	–	–
Saturation time	–	✓	–	–	–
Volume loaded	–	✓	✓	–	✓
Developmental phase	–	✓	–	–	–
Scanning speed	–	✓	–	–	–
Slit dimension	–	–	–	✓	✓
Distance between tracks	–	–	–	✓	✓
Migration distance	–	–	–	–	–
Application position	–	✓	–	–	–
Stationary phase	–	–	–	–	–
Plate dimension	✓	✓	✓	✓	✓

Method optimization studies and response surface analysis

Based on the highly influential factors selected from the factor screening studies, the method optimization studies were conducted employing FCCD. After conducting the experiments as per the suggested design, the optimization data analysis was carried out by fitting the data to the second-order quadratic polynomial model for detecting the main effects as well as the interaction effects. Table VIII illustrates the coefficients of the polynomial model equations, indicating highly significant model terms (P < 0.001) and higher values of r² (ranging between 0.8960 and 0.9777), thus ratifying the apt selection of model terms and high goodness of fit of the data to the selected model. The coefficient analysis was performed by analyzing the quadratic polynomial model equations generated as per the following equation for each CAA.

$$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+{\beta }_3{X}_1^2+{\beta }_4{X}_2^2+{\beta }_5{X}_1{X}_2+{\beta }_6{X}_1^2{X}_2+{\beta }_7{X}_1{X}_2^2$$ (4)

Response surface analysis was carried out employing 3D response surface plots and 2D contour plots for each of the CAAs, i.e., R_f, peak height, capacity factor, theoretical plates and separation time, as shown in Figure 3A–H.

Table VIII.

Values of the Coefficients of the Polynomial Equations and r² for Various CAAs

Coefficient code	Polynomial coefficient values for response variables
	Rf	Peak height	Capacity factor	Theoretical plates	Separation number
β0	+0.80	+5732.17	+0.41	+43677.94	+103.02
β1	−0.045	+11693.45	−0.025	−1882.10	−3.52
β2	+3.333E−003	−1678.52	−0.033	+539.50	−0.95
β3	−0.015	−3656.53	−0.069	+812.70	+9.88
β4	−0.11	+8150.01	__	+1159.59	+1.42
β5	−0.013	−1930.49	__	−8235.63	−15.78
R2	0.8960	0.9571	0.9777	0.9422	0.9747

Figure 3.

3D response surface plot depicting the interactions among chosen CMPs, i.e., volume loaded and plate dimension on CAAs, i.e., (A) R_f, (B) peak height, (C) capacity factor, (D) theoretical plates and (E) separation number. This figure is available in black and white in print and in color at JCS online.

Figure 3A depicting the 3D response surface plot for R_f reveals higher influence of volume loaded than the plate dimension. At higher levels of volume loaded, an initial increase in R_f was observed up to the intermediate levels, followed by declining values for R_f. The higher values of R_f are obtained at high levels of plate dimension and intermediate levels of volume loaded, respectively.

Like R_f, the 3D response surface plots for peak height again show significant influence of the volume loaded when compared with the plate dimension (Figure 3B). An escalating trend for peak height is observed at all the levels of volume loaded, while plate dimension shows lack of any significant influence on the peak height. The highest peak height has been observed at higher levels of volume loaded, thus indicating better densitometric separation.

The response surface plot depicted in Figure 3C shows significant influence of the plate dimension on the capacity factor when compared with the volume loaded. The 3D plot reveals a sharp declining trend in the peak height with increase in the levels of plate dimension; in contrast increase in the levels of volume loaded indicates miniscule change in the capacity factor.

Figure 3D and E portraying the 3D response surface plot for theoretical plate and separation number reveals almost an analogous trend for influence of both volume loaded and plate dimension on the respective CAAs. The plate dimension shows higher influence on both the CAAs from low to intermediate levels, followed by a dip. However, the volume loaded exhibits a negative influence on both the CAAs.

Search for the optimum solution

The optimum solution was determined by numerical optimization while “trading-off” various CAAs for attaining optimum values of R_f, peak height, separation number, theoretical plates and capacity factor, all indicative of the efficient densitometric separation and enhanced resolution of the mangiferin. The numerical optimization suggested CMPs with volume loaded (i.e., 5.9 µL) and plate dimension (i.e., 10 × 13.503 cm) as the optimized solution, exhibiting a desirability value of 1 and values of the CAAs, i.e., R_f of 0.65, peak height of 26876.2, capacity factor of 0.4127, theoretical plates of 41811.3 and separation number of 96.9. The obtained optimal solution was demarcated in the analytical design space employing the graphical optimization method (Figure 4). A typical densitogram of mangiferin obtained at optimal mobile phase composition with optimal method parameters is illustrated in Figure 5.

Figure 4.

Overlay contour plot depicting the optimal analytical design space region. This figure is available in black and white in print and in color at JCS online.

Figure 5.

Representative densitogram of mangiferin obtained employing chromatographic conditions from design space. This figure is available in black and white in print and in color at JCS online.

Method validation

Calibration curves

The linear regression data for the calibration curves (n = 3) depict a good linear relationship over the concentration range of 50–800 ng/band with respect to height, as shown in Figure 6. No significant difference was observed in the slopes of standard curves. Table IX depicts various parameters of the linear calibration plot of mangiferin in human plasma in the developed densitometric method with the value of coefficient of correlation being 0.996 ± 0.0012 (P < 0.001). The responses with a percent bias of ±5% were taken into consideration for validating the linearity range, with none of the points observed as outliers (60). This revealed that all the responses are present within the desirable statistical limits of 95% confidence intervals, confirming high data reliability and degree of closeness of the predicted data with those of the observed ones.

Figure 6.

Thin-layer densitometry and the corresponding chromatogram profiles obtained by the HPTLC instrument, with linearity graph (50–800 ng/band) of mangiferin in triplicate depicting (A) calibration bands on aluminum-based silica plate and (B) corresponding 3D densitogram tracks. This figure is available in black and white in print and in color at JCS online.

Table IX.

Linear Regression Data for Calibration Curve of Mangiferin in Human Plasma (n = 3)

Parameters	Values
Linearity range (ng/band)	50–800
Regressed equation	Y = 3.046x + 1,066
Correlation coefficient	0.996 ± 0.0012
Slope ± SD	3.046 ± 0.557
LOD	12.06 ng/mL
LOQ	36.55 ng/mL

Accuracy

Accuracy data for the standard concentrations of mangiferin, i.e., LQC (50 ng/band), MQC (100 ng/band) and HQC (150 ng/band) indicated good percentage of recovery, ranging between 96.75 and 99.47% with %RSD value <2%, respectively. This unequivocally vouches for a high degree of accuracy of the developed method. Table X explicitly enlists the accuracy data for various quality control samples of mangiferin, while Figure 7 shows the thin-layer densitometry curves and their corresponding profiles obtained through TLC.

Table X.

Accuracy of the Bioanalytical Thin-Layer Densitometric Method of Mangiferin

Standard concentration	Levels (%)	Concentration (ng/band)	Amount recovered (ng/band) ± SD	Recovery (%)	RSD (%)
Mangiferin (100 ng/band)	LQC: 50	150	146.13 ± 2.52	96.75	1.72
	MQC: 100	200	194.69 ± 3.51	97.34	1.80
	HQC: 150	250	248.69 ± 3.51	99.47	1.41

Figure 7.

Thin-layer densitometry and the corresponding chromatogram profiles obtained by HPTLC instrument, for accuracy of mangiferin at LOQ, MOQ and HOQ at 100, 150 and 200 ng/band, respectively, in triplicate depicting (A) calibration bands on aluminum-based silica plate and (B) corresponding three-dimensional densitogram tracks. This figure is available in black and white in print and in color at JCS online.

Precision

Intra-day and inter-day precision studies indicated higher values of percentage of recovery of mangiferin, ranging between 92.1 and 97.9%, respectively. Further, the %RSD values for mangiferin as per repeatability and intermediate precision were found to be well within the stipulated limit of <2%. These results confirmed the high degree of precision of the developed method. Table XI illustrates the intra- and inter-day precision data for various quality control samples of mangiferin.

Table XI.

Intra- and Inter-Day Precision Data of Bioanalytical Thin-Layer Densitometric Method of Mangiferin Intra-Day Precision (Within Day)

Standard concentration (ng/band)	Amount recovered (ng/band) ± SD	Recovery (%)	RSD (%)
Intra-day precision (within day)
LQC: 100	95.06 ± 1.10	95.06	1.15
MQC: 150	146.92 ± 1.53	97.94	1.04
HQC: 200	195.10 ± 1.00	97.55	1.02
Inter-day precision (between day)
LQC: 100	92.06 ± 1.10	92.06	1.19
MQC: 150	144.92 ± 1.53	96.13	1.59
HQC: 200	193.10 ± 1.00	96.55	1.03

LOD and LOQ

The method showed the LOD and the LOQ to be 12.1 and 36.6 ng/band, respectively, thus indicating quite high sensitivity of the developed method for quantification of mangiferin.

System suitability

The system suitability results confirmed lack of significant difference in the peak height, R_f, capacity factor, theoretical plates and separation number of mangiferin following triplicate injections. The values of RSD and standard error of the mean (SEM) were found to be <1%, thus corroborating the high degree of reliability of the HPTLC instrument.

Robustness

Among varied operable conditions like alteration in the composition of mobile phase, saturation time, volume loaded and volume of the developmental phase, no appreciable difference was perceived in the observed parameters like peak height, R_f, capacity factor, theoretical plates and separation number. The magnitudes of RSD and SEM were also found to be well within the specification limits.

Ruggedness

As per the ruggedness studies carried out employing two different analysts on the HPTLC instrument, no significant change in the values of the mean percentage of recovery was observed in three QC samples (i.e., LQC, MQC and HQC), thus ratifying through high degree of ruggedness of the developed densitometry method.

Matrix effect

The matrix effect, calculated using three different QC samples, i.e., 50, 100 and 150 ng/band showed the mean percentage of recovery values of 87.00, 93.03 and 96.75%, respectively. Higher values of the percentage of recovery, i.e., >80%, corroborated efficient densitometric separation of the mangiferin from the biological matrix. This also vouched for the aptness and robustness of the sample preparation and extraction technique employed throughout the bioanalytical method development and validation studies.

Discussion

The present studies describe the application of QbD principles for establishment of a bioanalytical thin-layer densitometric method of mangiferin in human plasma. Initially, the QTMP and CAAs of the TLC method were defined and CMVs identified by extensive risk assessment followed by factor screening studies. Further, the method was optimized by response surface optimization methodology for identifying the optimal solution followed by critical understanding of the factor–response relationship. Validation studies demonstrated a high degree of linearity, sensitivity, selectivity and specificity of the obtained method. The superiority of the current method can also be demonstrated on the basis of its high sensitivity, which is evident from much lower values of the LOQ, i.e., 12.1 ng/band, over the existing liquid chromatographic methods showing LOQ of 480 ng/mL and 3 µg/mL, respectively (61, 62). In a nutshell, the studies vouch for the high degree of utility of the developed method for estimation of mangiferin in bioanalytical samples.

Conclusion

A simple, rapid, sensitive and economical bioanalytical thin-layer densitometric method has been successfully developed employing the AQbD approach for quantification of mangiferin in bioanalytical samples, rat plasma per se. The systematic AQbD tools like risk assessment and factor screening exercise helped in identifying the “prominent few” method variables associated with high degree of variability from “plausible many” and facilitated optimization of them for attaining superior method performance. Overall, the volume loaded and plate dimension showed the maximum influence on all the vital response variables investigated like R_f, peak height, capacity factor, theoretical plates and separation number. Method validation studies corroborated the excellent linearity, accuracy, precision, high sensitivity, system suitability, robustness and ruggedness of the developed thin-layer densitometric method. Further, stability studies of the drug in plasma revealed lack of any interaction of the mangiferin with the biological matrix leading to the formation of any unwanted peak(s). In a nutshell, the studies indisputably vouch for the applicability and utility of the science and risk-based AQbD approach for developing a bioanalytical method of mangiferin employing thin-layer densitometry, a technique that can be extrapolated to other lipophilic drug substance(s) too.