Development and Validation of a Rapid RP-HPLC Method for the Determination of Venlafaxine Hydrochloride in Pharmaceutical Dosage forms using Experimental Design

The objective of the current study was to develop a simple, accurate, precise and rapid reversed-phase HPLC method and subsequent validation as per ICH guidelines for the determination of venlafaxine hydrochloride in pharmaceutical dosage forms. The proposed RP-HPLC method utilizes a 5 μm Varian MicrosorbMV 100 C18 column (250 mm x 4.6 mm) at ambient temperature. A 2 3 factorial design consisting of 3 factors at 2 levels was set up to standardize the chromatographic conditions. A numerical optimization technique employing the desirability approach was used to locate the optimum chromatographic conditions. The optimum mobile phase consisted of acetonitrile, 0.04 M potassium dihydrogen phosphate buffer and methanol (45:25:30, v/v), with pH adjusted to 5.5 using 10% phosphoric acid solution. The mobile phase was delivered isocratically at a flow rate of 1 mL/min with UV detection at 224 nm. The calibration plots constructed using the optimized chromatographic conditions displayed good linear relationship in the concentration range of 1-50 μg/mL with r=0.9992. The method was validated for precision, accuracy, robustness and recovery. The minimum detectable and minimum quantifiable amounts were found to be 0.568 and 1.72 μg/mL, respectively and the method was found to be reproducible from the statistical data generated. Venlafaxine hydrochloride was eluted at 3.43 min.

Chemical structure of venlafaxine hydrochloride.Hence, in the present study, an attempt has been made to develop an accurate, rapid and precise HPLC method with isocratic elution for determination of venlafaxine in pharmaceutical formulations.The developed method was validated and used for the assay of venlafaxine hydrochloride in capsules and tablets.

Experimental
Venlafaxine hydrochloride was a gift sample from Lupin Ltd., Pune, India.VENLIFT OD-37.5 and VENLOR-XR 75 were marketed samples of Torrent Pharmaceuticals Ltd., India and Protec, India, respectively.Acetonitrile, methanol, potassium dihydrogen phosphate and sodium hydroxide of Rankem used were of HPLC grade.Ortho-phosphoric acid used was of analytical reagent grade and was purchased from S.D. Fine Chemicals Ltd., Mumbai, India.Experiments were conducted with ultra pure water obtained from Milli-Q academic system (Millipore Pvt. Ltd., Bangalore, India).

HPLC instrumentation and conditions
The HPLC system consisted of Hitachi pump L-7110, Rheodyne universal injector 7725 and Hitachi L-7400 UV-visible detector.The chromatographic studies were performed using Varian ® Microsorb-MV 100 C 18 , 5 µm, 250 mm x 4.6 mm i.d.column, at ambient temperature and eluted with mobile phase at the flow rate of 1.0 mL/min.The mobile phase consisted of acetonitrile, 0.04 M potassium dihydrogen phosphate buffer and methanol (45:25:30, v/v), with pH adjusted to 5.5 with 10% phosphoric acid solution and was delivered isocratically at a flow rate of 1 mL/min.The mobile phase was filtered through 0.45 µm nylon filter and degassed in ultrasonic bath prior to use.Absorption maximum was detected by scanning standard solution of the drug over 200 to 400 nm wavelengths in Shimadzu model 1700 double beam UV-visible spectrophotometer with a pair of 10mm matched quartz cells.Measurements were made with injection volume of 20 µL and at a wavelength of 224 nm using an ultraviolet (UV) detector.

Experimental design
An 8-run, 2 3 factorial design consisting of 3 factors at 2 levels was set up to standardize the chromatographic conditions which are likely to be employed.Percentage of methanol in the organic phase (X 1 ), proportion of buffer in the mobile phase (X 2 ) and pH of the mobile phase (X3) as per 2 3 factorial design are represented in Table 1.By varying X 1 , X 2 and X 3 , a total of eight runs were performed as represented in Table 2.

Preparation of standard solutions
Stock standard solution of venlafaxine hydrochloride was prepared by dissolving appropriate amounts in methanol to give a final concentration of 1000 µg/mL.Standard solutions of venlafaxine hydrochloride (1.0, 2.5, 5.0, 10.0, 20.0, 30.0, 40.0, 50.0 µg/mL) were prepared by subsequent dilution with mobile phase.

Regression analysis
The targeted response parameters were statistically analyzed by applying one-way ANOVA at 0.05 levels in Design-Expert 7.1 demo version software (Stat-Ease Inc., Minneapolis, MN, USA).The individual parameters were evaluated using the F test and mathematical models of the form indicated in Eq (1) were generated for each response: where, Y is the level of the measured response, β 0 is the intercept, β 1 to β 7 are the regression coefficients, X 1 , X 2 and X 3 stand for the main effects, X 1 X 2 , X 2 X 3 and X 1 X 3 are the two-way interactions between the main effects and X 1 X 2 X 3 is the three-way interaction between the main effects.The polynomial models containing only the significant terms (P<0.05) were generated for each response parameter using multiple linear regression analysis (MLRA) and analysis of variance (ANOVA).Since the ratio of the highest and lowest value for the Tailing Factor was more than 3, a log transformation was performed as indicated by the Box-cox diagnostic plot available in the software.The models generated were used to construct the 3-dimensional graphs in which response parameter Y was represented as a function of X.The effect of independent variables on each response was also visualized from the contour plots.

Verification of the mathematical models and optimization
A numerical optimization technique employing the desirability approach was used to locate the optimum chromatographic conditions for the method developed and hence to verify the mathematical models generated.Various feasibility and grid searches were executed to establish the optimum chromatographic conditions for the method employed.Constraints set on each factor to locate the optimum chromatographic conditions are listed in Table 1.The optimized conditions as per Table 3 were employed and evaluated for the responses.The experimental values of retention time and tailing factor were compared with those predicted by the mathematical models.

Validation
Three series of standard calibration solutions in the range of 1.0 -50.0 µg/mL were prepared and analyzed as described above.Calibration curves were constructed by plotting the measured peak area of venlafaxine hydrochloride versus concentrations of standard samples.
To establish the accuracy and intra-day and inter-day precision of the method, three replicate standard solutions at three different concentrations (1.0, 10.0 and 50.0 µg/mL) were assayed on single day and three separate days.

Analysis of dosage forms
The content of 20 capsules were combined and weighed.An amount of powder equivalent to 100 mg of venlafaxine hydrochloride was accurately weighed, transferred to a 100 mL volumetric flask, made up to volume with methanol and placed in an ultrasonic bath for 20 min.After filtration through a 0.45 µm membrane filter, the solution was diluted with mobile phase to obtain a concentration of 20 µg/mL.The drug concentrations of three parallels were determined by HPLC using the calibration curve.

Chromatographic conditions
Potassium dihydrogen phosphate buffer in the concentration range 0.025 to 0.075 M was studied and 0.04 M was selected as it gave shorter retention time with minimum tailing factor.
A 2 3 factorial design was employed to prepare the mobile phase using potassium dihydrogen phosphate buffer (0.04M).A set of preliminary studies were undertaken to establish the range of each variable with the aim of fine-tuning the chromatographic peaks.Since lower pH values have a deteriorating effect on silica based columns 18 and pH values above 5.5 resulted in broad peaks with significant tailing, the lower and higher levels of pH during the runs were maintained at 4.0 and 5.5.Decreasing proportion of buffer in mobile phase to less than 25% decreased retention time to a value such that the drug peak merged with the solvent peak, while increasing buffer proportion to more than 75% substantially increased retention time making the analysis time too long.Hence the lower and higher levels of proportion of buffer in mobile phase were maintained at 25 and 75%v/v.Decreasing methanol percentage in organic phase to less than 20% increased tailing drastically while 100% methanol in organic phase increased retention time.Hence the lower and higher levels of percentage of methanol in the organic phase were maintained at 20 and 80%v/v respectively.
The results of analysis of variance and mathematical models generated by regressional analysis are represented in Table 4.The Fisher F test indicated that all the mathematical models generated for the response parameters were found to be significant (P > F less than 0.05).The primary statistical tool for identifying the need for transformations and for pinpointing which one works best is the Box-cox plot 19 .The transformation was performed as indicated by the diagnostic plot.The predictor model generated for the Tailing Factor upon transformation was found to be significant.The polynomial models constitute the coefficients for the intercept, first order main effects and interaction effects.The sign and magnitude of the main effects signify the relative influence of each factor on the response.Regression equations of the fitted model: The Fisher F test with a very low probability value (P > F = 0.0004) demonstrated a very high significance of the predictor model generated for the Retention Time (Y 1 ).The goodness of fit of the model was checked by the adjusted determination coefficient (adjusted R 2 ).The determination coefficient (R 2 ) is a measure of the amount of reduction of variability of Y obtained using the regressor variables X 1 , X 2 and X 3 .If insignificant terms were to be included in the model the adjusted R 2 value tends to decrease 20 .The adjusted R 2 value approached unity as the model was refined by removing the insignificant variables.The value of the determination coefficient (R 2 = 0.9998) was in reasonable agreement with the adjusted determination coefficient (adjusted R 2 = 0.9995), which confirmed the high significance of the model.A relatively low value of relative standard deviation (%RSD = 0.24) confirmed the improved precision and reliability of the conducted trials 20 .The polynomial model generated for the Retention Time indicated that the percentage of methanol in the organic phase (P > F = 0.0002), the proportion of buffer in the mobile phase (P > F = 0.0005) and the pH of the mobile phase (P > F = 0.0052), had significant positive influence on Y 1 .In addition to the main effects, the interaction terms X 1 X 2 (P > F = 0.0002) and X 2 X 3 (P > F = 0.0025) also had significant negative influence on Y 1 .
The combined influence of X 1 and X 2 on Y 1 is shown by the 3-dimensional plot represented in Figure 2A.The pattern of the contours (Figure 2B) reflects a significant interaction between the two variables analyzed, particularly at higher buffer proportions in the mobile phase.The plots clearly indicate that a retention time within the experimental constraints can be obtained using lower and intermediate percentage of methanol in the organic phase.The combined influence of X 2 and X 3 on Y 1 is shown by the 3-dimensional plot represented in Figure 3A.The pattern of the contours (Figure 3B) reflects a significant interaction between the two variables analyzed, particularly at higher pH values.The plots clearly indicate that the goals set for the retention time can be achieved at any point in the experimental domain.On the log 10 (Tailing Factor) (log 10 Y 2 ), the F test with a very low probability (P > F = 0.0028) demonstrated a high significance for the polynomial model.The value of the determination coefficient (R 2 = 0.9989) was found to be a good match with the adjusted determination coefficient (adjusted R 2 = 0.9960), which established the high significance of the model.A relatively low value of relative standard deviation (%RSD = 4.50) suggested improved precision and reliability of the experiments performed.The predictor model generated for log 10 (Tailing Factor) indicated that the proportion of buffer in the mobile phase (P > F = 0.0013), the percentage of methanol in the organic phase (P > F = 0.0174) and the pH of the mobile phase (P > F = 0.0024), all had significant positive influence on Log 10 Y 2 .Other than the main effects, the interaction terms X 1 X 2 (P > F = 0.0037) and X 2 X 3 (P > F = 0.0040) illustrated a significant negative influence on log 10 Y 2 .

R e te n tio n time
The influence of X 1 and X 2 on log 10 Y 2 is illustrated by the 3-dimensional plot represented in Figure 4A.The pattern of lines in the contour plot (Figure 4B) reveals a significant interaction between the two variables investigated particularly at higher proportions of buffer in the mobile phase.The plots clearly show that the log 10 (Tailing Factor) can be minimized using lower proportion of buffer in the mobile phase and lower percentage of methanol in the organic phase.The influence of X 2 and X 3 on log 10 Y 2 is illustrated by the 3-dimensional plot represented in Figure 5A.The pattern of lines in the contour plot Figure 5B reveals a significant interaction between the two variables particularly at higher proportions of buffer in the mobile phase.The plots clearly show that the log 10 (Tailing Factor) can be minimized using lower proportion of buffer in the mobile phase at higher pH value.The optimum chromatographic conditions were predicted using the desirability function.Extensive grid searches performed within the experimental domain suggested a few checkpoints.The first few with a predicted overall desirability function (D) of 0.996 to 0.977 were found to have almost the same composition as that of the experimental run R5 The next checkpoint (D=0.949) that differed in composition from the model experimental runs was considered for further method validation.The chromatographic conditions as well as the predicted and experimental values of the response parameters are represented in Table 5.The experimental values recorded for the optimum chromatographic conditions were found to agree with those predicted by the mathematical model.The %prediction error for the response parameters was found to range from -8.53 to 0.58 which prove the validity of the mathematical models generated by ANOVA and regressional analysis 21 .

Linearity
Calibration curves were constructed using three series of standard venlafaxine hydrochloride solutions in the range of 1.0-50.0µg/mL.The equation of linear regression and statistical data are presented in Table 6.The linearity of the calibration curve was validated by the high value of the correlation coefficient.

Limit of detection (LOD) and limit of quantification
In the present study, the LOD and LOQ were calculated according to the 3.3 σ/s and 10 σ/s criterions, respectively; where σ is the standard deviation of y-intercepts of regression lines and s is the slope of the corresponding calibration curve 22 .The LOD and LOQ values were found to be 0.568 and 1.72 µg/mL respectively.

Accuracy and precision
The precision was determined by analyzing three samples of venlafaxine hydrochloride at 1.0, 10.0 and 50.0 µg/mL on three separate days.Intra-day and inter-day data are given in Table 7.The intra-day and inter-day variability showed RSD values less than 1.4% in all three selected concentrations indicating good repeatability over the entire concentration range, which revealed that the proposed method was precise.The accuracy of the method was checked by recovery study using standard addition method at three different concentration levels, i.e., multilevel recovery study.The preanalyzed samples were spiked with extra 50 and 100% of the standard venlafaxine hydrochloride and the mixtures were analyzed by proposed method.Recovery of standard drugs added was found to be 99.4 -100.4% with the value of %RSD less than 1.0 indicating that the proposed method was accurate.Results of recovery study are shown in Table 8.

Robustness
Influences of small changes in the mobile phase composition (±10%) and flow rate (±10%) were studied to determine robustness of the method.Peak areas and retention time changes were observed.Peak area values and retention time values are varied by less than ± 1.39% and < 0.73% respectively.Despite the changes in retention time there was no problem for quantification.The results are summarized in Tables 9 and 10.

Specificity
The specificity test of the proposed method demonstrated that the excipients from marketed capsules and tablets prepared in the laboratory did not interfere in the drug peak.Furthermore, well-resolved peaks indicate the specificity of the method Figure 6.

Assay of venlafaxine hydrochloride in dosage forms
The developed method was applied to assay venlafaxine hydrochloride in marketed capsules and tablets prepared in the laboratory.The content was calculated as an average of five determinations and the experimental results are given in Table 11.The results were very close to each other as well as to the label value of commercial capsules and prepared tablets.Recoveries were very close to 100%, which proved the suitability and accuracy of the proposed method.

Conclusion
The proposed HPLC method developed using a 2 3 factorial design and optimized by employing the desirability approach provides a simple, accurate and reproducible quantitative method for routine analysis of venlafaxine hydrochloride in bulk and pharmaceutical formulations.The major advantage of this method is the quick sample analysis without prior separation or purification.Sample preparation procedure was simple with a short chromatographic time making the method suitable for processing multiple samples in a limited period of time.Finally, no pharmacopeial method has been reported yet though a couple of published methods for determination of venlafaxine hydrochloride in pharmaceutical dosage forms have been reported.It can be concluded that the proposed method is useful and suitable for routine quality control tests such as content uniformity of commercial formulations of venlafaxine hydrochloride.

Figure 2 (
Figure 2(A).3-D plot and (B) Contour plot showing the influence of percentage of methanol in the organic phase and proportion of buffer in the mobile phase on the retention timeThe combined influence of X 2 and X 3 on Y 1 is shown by the 3-dimensional plot represented in Figure3A.The pattern of the contours (Figure3B) reflects a significant interaction between the two variables analyzed, particularly at higher pH values.The plots clearly indicate that the goals set for the retention time can be achieved at any point in the experimental domain.

Figure 3 (
Figure 3(A).3-D plot and (B) Contour plot showing the influence of proportion of buffer in the mobile phase and the pH of the mobile phase on the retention timeOn the log 10 (Tailing Factor) (log 10 Y 2 ), the F test with a very low probability (P > F = 0.0028) demonstrated a high significance for the polynomial model.The value of the determination coefficient (R 2 = 0.9989) was found to be a good match with the adjusted determination coefficient (adjusted R 2 = 0.9960), which established the high significance of the model.A relatively low value of relative standard deviation (%RSD = 4.50) suggested improved precision and reliability of the experiments performed.The predictor model generated for log 10 (Tailing Factor) indicated that the proportion of buffer in the mobile phase (P > F = 0.0013), the percentage of methanol in the organic phase (P > F = 0.0174) and the pH of the mobile phase (P > F = 0.0024), all had significant positive influence on Log 10 Y 2 .Other than the main effects, the interaction terms X 1 X 2 (P > F = 0.0037) and X 2 X 3 (P > F = 0.0040) illustrated a significant negative influence on log 10 Y 2 .The influence of X 1 and X 2 on log 10 Y 2 is illustrated by the 3-dimensional plot represented in Figure4A.The pattern of lines in the contour plot (Figure4B) reveals a significant interaction between the two variables investigated particularly at higher proportions of buffer in the mobile phase.The plots clearly show that the log 10 (Tailing Factor) can be minimized using lower proportion of buffer in the mobile phase and lower percentage of methanol in the organic phase.

Figure 4 (
Figure 4(A).3-D plot and (B) Contour plot showing the influence of percentage of methanol in the organic phase and proportion of buffer in the mobile phase on the log 10 (Tailing Factor).The influence of X 2 and X 3 on log 10 Y 2 is illustrated by the 3-dimensional plot represented in Figure5A.The pattern of lines in the contour plot Figure5Breveals a significant interaction between the two variables particularly at higher proportions of buffer in the mobile phase.The plots clearly show that the log 10 (Tailing Factor) can be minimized using lower proportion of buffer in the mobile phase at higher pH value.

Figure 5 (
Figure 5(A).3-D plot and (B) Contour plot showing the influence of proportion of buffer in the mobile phase and the pH of the mobile phase on the Log 10 (Tailing Factor).

Table 1 .
Factors and their corresponding levels as per 2 3 factorial design.

Table 2 .
3hromatographic conditions employed as per 23factorial design

Table 3 .
Experimental values for the retention time and the tailing factor.

Table 4 .
3ummary of ANOVA for the response parameters of the model mobile phases prepared as per 23factorial design

Table 5 .
Optimum chromatographic conditions and comparison of the experimental and predicted values of the responses

Table 6 .
Statistical data of calibration curves of venlafaxine hydrochloride.

Table 8 .
Recovery data for the proposed RP-HPLC method (n=5).

Table 9 .
The influence of small changes in mobile-phase composition (Method Robustness)

Table 10 .
The influence of small changes in flow rate (Method Robustness).

Table 11 .
Analysis of venlafaxine hydrochloride from pharmaceutical formulations by proposed method.