ORIGINAL ARTICLES Year : 2022  Volume : 11  Issue : 1  Page : 4150 Chemometricsassisted spectrophotometric method development and validation for simultaneous estimation of emtricitabine, tenofovir alafenamide fumarate, and dolutegravir sodium in dosage form Sapna M Rathod^{1}, Paresh U Patel^{2}, ^{1} Research Scholar, Ganpat University, Kherva, Gujarat, India ^{2} Department of Pharmaceutical Chemistry and Quality Assurance, S.K. Patel College of Pharmaceutical Education and Research, Ganpat University, Kherva, Gujarat, India Correspondence Address: Aim: This study aims on the development of a chemometricassisted spectroscopic method for the analysis of combined dosage form of emtricitabine (EMT), tenofovir alafenamide fumarate (TEN), and dolutegravir sodium (DOL). The use of a multivariate algorithm to analyse spectrophotometric data is a novel approach to estimating drug concentrations in formulations. Materials and Methods: The quantitative estimation of EMT, TEN, and DOL in tablets was carried out using four chemometric approaches: Classical least square (CLS), inverse least square, partial least square, and principal component regression. Thirtytwo ternary mixtures of calibration sets and 16 mixtures of validation sets were prepared. The absorbance data matrix was attained by calculating absorbance at 25 different wavelengths in a range of 240–336 nm (Δλ = 4 nm). The chemometric calculations were performed using Matlab2018a and Minitab software. The developed methods were validated. Results: The great accuracy of the current study was justified by the nearperfect recovery values (100%) and low standard deviation. For chemometrics approaches, the root mean square error of calibration (RMSEC), root mean square error of prediction (RMSEP), and root mean square error of crossvalidation (RMSECV) outcomes display decent accuracy and precision. Conclusion: The CLS approach yielded the lowest predicted residual error sum of squares, RMSEC, RMSEP, and RMSECV scores. As a result, CLS might be regarded as the best chemometric approach among all techniques utilized. The label claim determined is in excellent accordance with the mean recoveries for EMT, TEN, and DOL. So, it can be used in quality control laboratories.
List of Abbreviations Emtricitabine; EMT, Tenofovir Alafenamide Fumarate; TEN, Dolutegravir Sodium; DOL, Classical Least Square; CLS, Inverse Least Square; ILS, Partial Least Square; PLS, and Principal Component Regression; PCR, Standard deviation; SD, Root mean square error of Calibration; RMSEC, Root mean square error of Prediction; RMSEP, Root mean square error of Cross Validation; RMSECV, Predicted residual error sum of squares; PRESS, Analytical Reagent; A.R., Leave one out; LOO, correlation coefficient; R2, Figure of merits; FOM, Sensitivity; SEN, Limit of Detection; LOD, limit of quantitation; LOQ. Introduction In chemometrics, there are two types of data set namely, calibration and validation data set. The findings of calibration data set were utilized to determine the component concentrations in unknown sample.[1] For both classical statistics and chemometric approaches, there is currently a considerable amount of computer software readily available.[2] Chemometrics methods are particularly useful approaches for analyzing many compounds at the same time, in which the overlap of the active compounds’ spectra generates an interference that makes determining the amounts of each component impossible.[3] In addition, chemometric calibration methods are simple since they can evaluate a large number of samples in a short amount of time more accurately and precisely compared with other methods. It can be described as the use of mathematical and statistical approaches to create and/or optimize measurement procedures, as well as the analysis of pertinent data to offer chemical information.[4] Multivariate calibrations such as classical least square (CLS), inverse least square (ILS), principle component regression (PCR), and partial least square (PLS) have been widely used in quantitative spectrum analysis in recent years to extract selective information from unselective data.[5],[6] These approaches are commonly used since they produce the greatest outcomes when it comes to resolving complex mixtures.[7] These approaches can be used to estimate medications in pharmaceutical formulations containing two or more drug components using simultaneous spectrophotometric methods.[8] CLS and ILS are two of the most basic approaches, both based on Beer’s principle and using a multivariate least square procedure. Factor analysis methods such as PCR and PLS are used to establish a link between chemical data matrices.[9],[10] Emtricitabine (EMT), also known as 2′,3′dideoxy5fluoro3′thiacytidine (FTC), is a synthetic nucleoside reverse transcriptase inhibitor (NRTI) that is taken once a day orally.[11] Emtricitabine 5′ triphosphate, an active metabolite formed by intracellular kinases phosphorylating emtricitabine, inhibits HIV reverse transcriptase by competing for entrance into the HIV DNA chain with the endogenous substrate 2′deoxycitidine 5′triphosphate.[12] Because emtricitabine 5′triphosphate lacks a hydroxyl group in the 3′ position of the sugar moiety, it causes chain termination when it is incorporated into the HIV DNA chain.[13],[14] Tenofovir alafenamide fumarate (TEN), chemically propan2yl(2S)2{[(S)({[(2R)1(6amino9Hpurin9yl)propan2yl]oxy}methyl)(phenoxy)phosphoryl]amino}propanoate.[15] TEN has a molecular weight of 476.47 g/mol and a chemical formula of C21H29N6O5P. TEN is a NRTI and tenofovir prodrug is used to treat HIV1 infection. The drug has a solubility in water, methanol, and dimethyl sulfoxide.[16],[17] Dolutegravir sodium (DOL) is integrase strand transfer inhibitor. The drug prevents the viral genome from being integrated into the host cell by blocking the strand transfer stage. It is chemically sodium; (3S,7R)13[(2,4difluorophenyl) methylcarbamoyl] 7methyl9,12dioxo4oxa1,8diazatricyclo[8.4.0.03,8]tetradeca10,13dien11olate.[18],[19] No reported analytical methods were found for estimating EMT, TEN, and DOL in bulk and in their combination dosage form. As a result, the current study was attempted to design and validate multivariate approaches for resolving complex drug spectra. Instrumentation and Software To test the absorbance of all the solutions, a shimadzu model 1700 (Japan) double beam UV/Visible spectrophotometer with a spectral width of 2 nm, wavelength accuracy of 0.5 nm, and a pair of 10 mm matched quartz cells was employed. The spectra of various calibration and validation sets were recorded using UV probe software. Chemometric calculations were performed using MATLABR2018a Software, Minitab 16.1.1, and Microsoft Excel 2010. MVC1toolbox (with MATLAB) was used to estimate figures of merit for multivariate calibration models. Materials and Methods TEN was kindly gifted by Bulat Pharmaceuticals, Hyderabad, Andhra Pradesh. EMT was provided as gift samples from Amneal Pharmaceuticals, Ahmedabad, Gujarat, and DOL were provided as gift samples from Cipla Pharmaceuticals, Mumbai, Maharashtra. The local market provided the commercial combination tablet (SPEGRA). Methanol (A.R.) grade (SD fine grade chemicals Ltd.), distilled water, and other chemicals used were of analytical grade. No. 41 Whatman filter paper was utilized in the study. Preparation of solutions Each drug was accurately weighed and transported to a separate volumetric flask and dissolved using methanol, and the volume was brought up to the mark using methanol (1000 µg/mL). Aliquot each drug’s standard stock solution to a separate volumetric flask to produce a working standard solution of 100 µg/mL, using distilled water as diluent. Preparation of calibration set and validation set A data set of calibration samples was created using fractional factorial design. A total 32 ternary mixture solutions were prepared by mixing known amount of drugs under study in varied proportions. Validation set consisting of 16 samples was prepared from the working solutions in the same manner as that of calibration set. The composition of calibration as well as validation set were represented in [Table 1].{Table 1} Optimization and selection of method parameters Sample solutions were analyzed across 200–400 nm for calibration and validation datasets, and zeroorder spectra were obtained [Figure 1]. The absorbance data from the spectrum regions of 200–220 nm with noise and 350–400 nm with zero reading were excluded because they were not essential for the chemometric approach. Wavelength in the range of 240–336 nm was chosen to produce minimal root mean square error of calibration (RMSEC) and root mean square error of crossvalidation (RMSECV) values.{Figure 1} Classical least square CLS is also known as K matrix. Basically, it involves the usage of multiple linear regression to represent the Beer–Lambert law of spectroscopy in a classical way. [INLINE:1] Calibration set comprised of concentration matrix, C, and an absorbance matrix, A for known sets of samples is constructed to generate calibration using CLS. In MATLAB2018a software, the CLS model was developed by adding absorbance (A) and concentration matrix (C) data. The calculated K can be used to forecast the concentration of an unknown sample, Cunk, based on its measured spectrum, and it can be stored as an absorbance matrix, Aunk. There are mainly two subclasses of CLS namely, direct CLS and indirect CLS. The K matrix is calculated in direct CLS by measuring the spectra of the pure component, either neat or in a nonabsorbing solvent. In the indirect CLS technique, pure spectra are calculated from mixture spectra rather than being measured directly. Absorbance matrix A is comprised of zeroorder spectra at 4 nm intervals between 240 and 336 nm, that is, absorbances at 25 wavelength points. The developed model comprised absorbance values of samples at 25 various wavelength points, and quantities of EMT, TEN, and DOL in the validation data set as well in tablet formulations were predicted. Inverse least square It is also called as P – matrix calibration as it originally requires the use of multiple linear regression to calculate the inverse expression of the Beer–Lambert equation of spectroscopy. [INLINE:2] where, C = concentration matrix, P = calibration coefficient, and A = absorbance matrix. To determine P, a training set containing a concentration matrix, C, and an absorbance matrix, A, is used to create a calibration using ILS. ILS differs from the classical technique, which involves fitting a linear mixture of pure spectra to an unknown spectrum. This distinction provides ILS with several advantages. When all of the system’s components aren’t explicitly evaluated, CLS fails to provide accurate predictions. The software MATLAB2018a was used to construct the approach. The samples’ absorbance values were inputted into the calibrations at 25 various wavelength points in the spectral area in a range of 240–336 nm. The concentrations of EMT, TEN, and DOL in validation set as well in tablets were predicted. Partial least square and principal component regression These are the most widely used methods in the multivariate calibration approach. The inverse calibration methodology is used in both procedures. The PCR is a method that operates on the principle of lowering the original data’s dimensionality. The original variables are replaced with linear combinations of the variables in both PLS and PCR to solve the inversion problem (factors). The PCR employs the wellknown singular value decomposition method. Using the converted data as input, this function was used to fit a PCR model. When fitting the PCR model, the leaveoneout (LOO) crossvalidation approach was utilized. The optimal principal components (or eigenvectors) corresponding to the large eigenvalues are identified using crossvalidation in the calibration step. The PLS employs a nonlinear iterative partial least square algorithm to generate the model. The independent and dependent variables are simultaneously compressed and decomposed, resulting in latent variables, in the PLS calibration using the orthogonalized PLS technique. In Minitab 16.1.1, the A and C data matrix were incorporated in PCR and PLS models. The concentration of EMT, TEN, and DOL in the validation set and formulation were predicted. PLS and PCR calibrations were constructed by using the nonlinear iterative partial least squares (NIPALS) algorithm and standard singular value decomposition (SVD) algorithm, respectively. For PCR and PLS calibrations, an adequate number of principal components or factors must be chosen. Validation of developed methods Precision Intraday and interday precision study was established by triplicate analysis of ternary mixture containing different proportions of EMT, TEN, and DOL (5/5/10 μg/mL, 10/20/5 μg/mL, and 10/20/10 μg/mL) on one day and on three successive days, respectively. The absorbance data of the ternary mixture obtained after scanning in UV spectrophotometer were incorporated in respective equations and concentrations were calculated. The results were expressed as a percent recovery ± standard deviation (SD). Accuracy The method’s accuracy was determined by applying the analytical approach to fabricated blends of drug product components (placebo) to which known proportions of the drug ingredient to be analyzed were incorporated. Accuracy of the method was studied in triplicate at three various levels (80%, 100%, and 120%). The known amounts of standard solutions containing EMT (25.6, 32.0, and 38.4 μg/mL), TEN (3.2, 4.0, and 4.8 μg/mL), and DOL (6.4, 8.0, and 9.6 μg/mL), to achieve the various levels, were added to placebo sample solutions. The absorbance data of the ternary mixture obtained after scanning in UV spectrophotometer were incorporated in respective equations and concentrations were calculated. The results were expressed as a percent recovery ± SD. Assay of formulation Weigh 10 spegra tablets and determine the average content of blend. The tablet powder equivalent to 100 mg EMT, 26.3 mg DOL (≈ 25 mg dolutegravir), and 15.6 mg TEN (≈ 12.5 mg tenofovir alafenamide) was transferred to volumetric flask and dissolved in methanol by sonication for 20 min and the volume was made up to the mark with methanol. Filter paper No. 41 (Whatman) was used to filter the solution. Aliquot required amount from the above solution to achieve 32 μg/mL EMT, 8.4 μg/mL DOL, and 5 μg/mL TEN. At the specified wavelengths, the absorbance of the sample solutions was recorded, and the amount of individual component was measured. Results and Dıscussıon Classical least square The value of calibration coefficient can be calculated by using the equation: [INLINE:3] where, pinv(c) is the pseudo inverse of concentration matrix and A is matrix of absorbance of mixture. [INLINE:4] where, pinv(K) is pseudo inverse of K matrix concentration of unknown: [INLINE:5] Spectra of solutions containing unknown concentrations of drugs were recorded in the optimized range of wavelength and absorbance matrix A were generated. Using the calibration coefficient matrix K, the concentration was computed. [INLINE:6] #For representation of matrix conveniently, Kcal values are shown in transposed form. Where A is the absorbance values at 25 points corresponding to the 240–336 nm spectral range at an interval of 4 nm. CEMT, CTEN, and CDOL represent the concentrations of EMT, TEN, and DOL, respectively. Inverse least square The value of calibration coefficient can be calculated by using the following equation: [INLINE:7] Where, P is the matrix of the unknown calibration coefficients relating the concentrations to the spectral intensities. Spectra of solutions containing unknown concentrations of drugs mixture were recorded in the optimized range of wavelength and absorbance matrix A was generated. Using the calibration coefficient matrix P, the concentration was computed using the equation: [INLINE:8] [INLINE:9] Where, A is the absorbance values at 25 points corresponding to the 240–336 nm spectral range at interval of 4 nm. CEMT, CTEN, and CDOL represent the concentrations of EMT, TEN, and DOL, respectively. Partial least square and principle component regression The component number for the experimental data should be chosen in such a way that overfitting is avoided. The number of principal components determined using the following approaches: For PCR, two PCs were selected based on retaining components with eigenvalues greater than 1 (and retain components that cumulatively explain 90% of the variance.) and it was confirmed using scree plot [Figure 2]. Scree plot shows steep curve up to three PCs, followed by a bend and then a flat line. A number of PCs in PLS selected using a model selection plot; scatterplot of the crossvalidated R2 and fitted R2 values as a function of the number of components [Figure 2]. Three numbers of components were selected based on retaining components with identical R2 values of validated R2 and fitted R2. The components selected are also assessed using score plot. Here, the first two components make up the majority of the variance in the data, and there are no outliers in this data set; the points are spread randomly around zero.{Figure 2} The equations for the PLS method were obtained as: CEMT = 0.107 + 9.485 × A1 + 28.863 × A2 + 18.46 × A3 − 11.686 × A4 −11.264 × A5 − 18.456 × A6 −10.271 × A7 + 3.293 × A8 − 11.557 × A9 − 9.722 × A10 − 3.309 × A11 − 4.031 × A12 − 8.275 × A13 − 5.134 × A14 + 52.76 × A15 − 6.275 × A16 − 64.879 × A17 + 126.418 × A18 − 55.385× A19 − 26.903 × A20 + 33.011 × A21 − 33.954× A22 − 0.849× A23 – 109.399 × A24 + 89.03× A25 CTEN = 0.035 – 36.553 × A1 − 47.925× A2 − 10.068× A3 + 58.874× A4 + 44.355× A5 + 30.864× A6 + 2.532 × A7 − 29.872× A8 − 32.414× A9 + 11.905× A10 +9.001× A11 + 10.04× A12 + 29.664× A13 + 44.557× A14 − 95.969× A15 − 15.952× A16 + 102.497× A17 − 211.513 × A18 + 80.741× A19 − 61.178 × A20 + 6.754× A21 + 155.712× A22 − 20.279 × A23 – 111.113 × A24 − 260.844 × A25 CDOL = − 0.172 + 19.94 × A1 − 10.77× A2 − 27.725× A3 − 9.676× A4 − 3.541× A5 + 24.839 × A6 + 11.394 × A7 − 9.1× A8 + 1.65 × A9 + 7.97 × A10 + 0.778 × A11 − 4.315 × A12 + 4.747 × A13 + 24.486 × A14 + 0.708 × A15 − 71.69 × A16 + 150.532 × A17 − 267.055× A18 + 53.179 × A19 + 155.129 × A20 − 23.697 × A21 − 168.71 × A22 − 30.047× A23 + 201.696× A24 − 1.725× A25 The equations for the PCR method were obtained as: CEMT = − 1.128 + 1.519 × A1 + 0.937 × A2 − 0.215 × A3 − 1.501 × A4 −2.545 × A5 − 2.937 × A6 −2.406 × A7 – 0.914 × A8 + 1.271 × A9 + 3.656 × A10 + 5.860 × A11 + 7.234 × A12 + 7.563 × A13 + 7.241 × A14 + 6.397 × A15 + 4.780 × A16 + 2.080 × A17 − 0.865 × A18 − 2.814 × A19 − 3.660 × A20  3.981 × A21 − 4.131× A22 − 4.072 × A23 – 3.995 × A24 − 3.966 × A25 CTEN = − 0.021 + 0.660 × A1 + 2.985× A2 + 6.505 × A3 + 9.676× A4 + 11.730 × A5 + 12.157 × A6 + 10.802 × A7 + 7.162 × A8 + 2.310 × A9 − 3.218 × A10 − 8.297 × A11 − 11.407 × A12 − 12.227 × A13 − 11.941 × A14 − 11.261 × A15 − 9.945 × A16 − 8.157 × A17 − 6.689 × A18 − 5.560× A19 − 5.667 × A20 − 6.437× A21 − 7.693 × A22 − 9.922 × A23 – 12.885 × A24 − 15.807 × A25 CDOL = 2.687 − 0.298 × A1 − 0.962 × A2 − 1.717 × A3 − 2.172× A4 − 2.274× A5 − 2.136 × A6 − 1.924 × A7 − 1.417 × A8 − 0.927 × A9 − 0.285 × A10 + 0.283 × A11 + 0.599 × A12 + 0.713 × A13 + 0.874 × A14 + 1.317 × A15 − 2.154 × A16 + 3.740 × A17 + 5.659 × A18 + 6.827 × A19 + 7.627 × A20 + 8.297 × A21 + 9.060 × A22 + 10.125× A23 + 11.545× A24 + 12.980 × A25 Where, A is the absorbance values at 25 points corresponding to the 240–336 nm spectral range at intervals of 4 nm and CEMT,CTEN, and CDOL are the concentrations of EMT, TEN, and DOL, respectively. Method validation Precision Method reproducibility for each title ingredient was demonstrated by repeatability and intermediate precision measurements. The obtained results within and between days trials are represented in [Table 2] and [Table 3]. The recovery values were close to 100% with low SD justified the good precision of the proposed methods.{Table 2} {Table 3} Accuracy The mean percent recoveries for EMT, TEN, and DOL are reported in [Table 4]. The remarkable accuracy of the recommended approaches was justified by the near100% recovery values with low SD.{Table 4} Chemometric methods LOO approach was employed for cross validation technique using calibration set of 32 mixtures. Each calibration sample’s predicted concentrations were tested to the known concentrations of compounds. RMSECV and RMSEP were calculated to validate the model. For a given model, these values must be as low as possible. For assessing the inaccuracies in the predicted concentrations, the RMSECV value was utilized as a screening test. It denotes the precision as well as the accuracy of predictions. The prediction capabilities of developed methods (CLS, ILS, PCR, and PLS) are evaluated by using two different methods. Plotting the known concentration against the predicted concentration was the first approach used. The aforesaid chemometric procedures [Table 5][Table 6][Table 7] yielded a reasonable correlation coefficient (R2) value for each drug, and the second way was the calculation of RMSECV and RMSEP.{Table 5} {Table 6} {Table 7} The analytical figure of merits (FOM) is critical for quantifying the quality of an approach or comparing methods. Several FOM has been observed in multivariate calibration, including sensitivity (SEN), analytical sensitivity, the limit of detection (LOD), and limit of quantitation (LOQ) [Table 5][Table 6][Table 7]. Assay of formulation EMT, TEN, and DOL in tablet formulations were assessed using the proposed chemometric approach. The results were satisfactory and in line with the label claim. The assay results [Table 8] show that the approach is acceptable for simultaneous quantification of EMT, TEN, and DOL without intervention from common excipients.{Table 8} Conclusion The chemometric method is more accurate and precise than conventional methods as the total absorbance of the ternary mixture was measured. The developed method holds an acceptable degree of precision and accuracy in accordance with international guidelines. With great recoveries and precision, the proposed approach was successfully used to the assay of formulation. As a result, the current method can be used to estimate EMT, TEN, and DOL in formulation simultaneously. Acknowledgement The authors express gratitude towards Bulat Pharmaceuticals, Amneal Pharmaceuticals, Cipla Pharmaceuticals for gifting drug samples. The authors also acknowledge APMC College of Pharmaceutical Education and Research, Himatnagar (Managed by Himatnagar Kelvani Mandal) for providing facilities to conduct this research. Financial support and sponsorship Nil. Conflicts of interest I/we certify that no actual or potential conflict of interest in relation to this article exists. References


