Application of Box-Behnken Design for Validation of High-Per- Formance Thin-Layer Chromatography/Densitometry Method for Robustness Determination of Apremilast in Bulk and inhouse Tablets

Chaudhari and Shirkhedkar: Application of Box-Behnken Design for Validation of High-Per Formance Thin-Layer Chromatography/Densitometry Method for Robustness Determination of Apremilast in Bulk and in-house Tablets



Psoriasis is classified in a group of chronic inflammatory disease troubling approximately 2 – 3% of the wide-reaching people. In psoriasis, the augmented the levels of pro-inflammatory mediators, such as, tumor necrosis factor, interleukin IL-17 and IL-23, and decreases the level of anti-inflammatory mediators, viz. IL-10. Phosphodiesterase 4 (PDE4) a principle enzyme dominant in immune cells and modulates the production of this cytokinines.1

Apremilast (APL), [N-[2-[(1S)-1-(3-ethoxy-4-methoxyphenyl)-2-methylsulfonylethyl]-1,3dioxoisoindol - 4- yl] acetamide] (Figure 1) is an orally existing, small molecule inhibitor of phosphodiesterase-4 (PDE-4) and an immunomodulating agent which is used for management of refractory psoriatic arthritis. Apremilast has been linked to a low rate of serum enzyme elevations during therapy, but has not been given in cases of clinically apparent acute liver injury.2 Literature revealed few methods for analysis of apremilast in bulk, pharmaceutical formulations and biological fluids which includes UPLC–MS/MS,1 High-Performance Liquid -Chromatography method for quantification of impurities of Apremilast3 and stability-indicating UV- Spectrophotometric method.4 Robustness can be interpreted as the capability to reproduce the (analytical) method in diverse laboratories or under different conditions without the occurrence of unexpected differences in the obtained result(s), and a robustness test as an experimental set-up to evaluate the robustness of a method.5 The implication of use of ‘Quality by Design’ (QbD) or ‘Design of Experiments’ (DoE) approach suggested achieving these goals.6 The robustness test ensures the possible causes of changeability in one or a number of responses of the method.7,8 To assess likely sources of changeability, a number of factors are selected from the working procedure and examined in an interval that to some extent exceeds the differences which can be anticipated when a method is conveyed from one instrument to another or from one laboratory to another, these factor are observed in an experimental design and the effect of the factors on the response(s) of the method is assess in this way the factors that possibly will harm the method performance are discovered. The analysts subsequently know that such factors should be more stringently forbidden throughout the implementation of the method.9

Figure 1

Chemical structure of Apremilast.

Various design methodologies to assess robustness of method such as; full factorial design, fractional factorial designs, Asymmetrical Factorial Designs (AFD),10 Central Composite Design (CCD) either as Circumscribed or Face-centred, Doehlert Designs, Box–Behnken Design (BBD), Plackett–Burman Design (PBD), Star Designs;11 supported with graphical methods of interpretation such as Normal probability plot, half-normal probability plot, bar plot with or without limit value, counter plot, standardized pareto chart and response surface method (RSM). 12

The Box-Behnken is an excellent design for response surface methodology because it permits: (I) estimation of the parameters of the quadratic model; (II) structure of sequential designs; (III) recognition of lack of fit of the model; (IV) use of blocks. None of these methods have been found to be cost-effective.13 However, to our endeavor no HPTLC method has been studied so for the estimation of apremilast in bulk material and in-house tablets formulation. The present research illustrates a simple, sensitive, effective and economical Normal Phase (NP) - HPTLC method for estimation of apremilast in bulk material and in-house tablets formulation giving emphasis on application of DoE approach to evaluation of robustness of method.



Chemicals and Reagents

Pharmaceutical grade Apremilast working standards were obtained as generous gifts from Intas Pharmaceuticals Ltd, Ahmedabad, India. Methanol (HPLC Grade) and aluminium backed TLC plates pre-coated with silica gel 60 F254 (0.2 mm thick) were purchased from E. Merck Ltd., Mumbai (India).

HPTLC instrumentation and Chromatographic conditions


HPTLC system : Camag TLC system (Muttenz, Switzerland)

Sample applicator : Linomat 5

Scanner : TLC scanner 3

Data processor : winCATS (version 1.3.0)

Development chamber : Camag twin trough chamber (20 x 10 cm)

Syringe for application : Hamilton syringe (100 μL)

Ultrasonicator : ENERTECH Electronics Pvt. Ltd., India

Chromatographic conditions

Stationary phase : Aluminium backed precoated silica gel 60-F254 (20 x 10 cm)

Mobile phase : Toluene : Methanol (8:2 v/v)

Development distance : 8 cm

Saturation time : 25 mins

Scanning wavelength : 230 nm

Densitometry scanning mode : Absorbance–Reflectance.

Preparation of Stock Standard Solution and study of linearity curve

Stock standard solution was prepared by weighing 10 mg of apremilast. Weighed powder was transferred into volumetric flask of 10 mL and dissolved and diluted to mark with methanol to obtain concentration 1 mg/mL.

An appropriate volume of 0.5 – 3 mL from stock standard solution was transferred with the help of previously calibrated pipette into series of 10 mL volumetric flasks. A fixed volume of 5 μL was applied on the HPTLC plates to obtain concentrations 250, 500, 750, 1000, 1250 and 1500 ng/band of APL, respectively.

Preparation of Sample

As the pharmaceutical formulation of apremilast is not available in the local Indian market; therefore, in-house tablets were prepared with 30 mg of apremilast and common excipients. The sample solution was prepared from in-house formulated apremilast tablets.

Accurately weighed power drug equivalent to 30 mg apremilast was quantitatively transferred into 100 mL volumetric flask dissolved and diluted volume with methanol. The resulting solution was filtered through a 0.45 μm filter (Millifilter, Milford, MA, USA). An appropriate volume of 4 μL from filtrate was applied on HPTLC plates were subjected to proposed method for further analysis.

Statistical tools

Experimental design for robustness study was performed using Design Expert® (Version, Stat-Ease Inc., Minneapolis, MN, USA statistical software. The rest of the calculations for the analysis were performed by use of Microsoft Excel 2007 software (Microsoft, USA).

Validation of HPTLC Method

The anticipated method was validated as to ensure it for precision, accuracy, sensitivity and robustness as per recommendations of International Conference on Harmonization (ICH) guidelines.14

Precision and Accuracy

Method precision was performed as repeatability, intra-day and inter-day deviation. Repeatability was accessed at concentration of 1000 ng/band of APL using six replicates. Intra-day deviation was studied using concentration 500, 700 and 1000 ng/band of APL; analyzed it for three times in the same day while it was analyzed for the three different days over a period of week for inter-day studies.

Accuracy of the method was estimated by spiking the drug standard in pre-determined laboratory mixture solution at concentration levels of 80 %, 100 % and 120 % and determined as percent recovery studies.


The sensitivity of proposed methods was estimated in terms of Detection Limit (DL) and Quantification Limit (QL) determinations for both specified methods were based on the standard deviations of the responses and slopes of constructed calibration curves (n = 3) as described by the International Conference for Harmonization guidelines Q2(R1). For the determination of DL and QL during HPTLC method validation; APL solutions of, 250, 300, 350, 400,450 and 500 ng /band were applied on HPTLC plates. The DL and QL were calculated using equations DL = 3.3·N/B and QL = 10·N/B; where, ‘N’ is standard deviation of peak areas of the drug (n = 3) taken as a measure of noise and ‘B’ is the slope of corresponding calibration curve.

Experimental design methodology for robustness

A Box-Behnken statistical screening design was used to optimize the compositional parameters and to evaluate quadratic effects of the mobile phase composition, saturation time, development distance and activation time of plate on the retention factor (Rf) and peak area. The linear polynomial equations generated from ANOVA are in the form, depicted below.


Where, ‘y’ is the measured response (dependent variable) associated with the each factor level combination; ‘b0’ represents the polynomial equation intercept representing average arithmetic mean of all quantitative outcomes of twenty-nine runs and ‘b1 – b44’ is regression coefficients computed from the observed experimental values of ‘y’ and ‘y1’ ‘x1’, ‘x2’ ,‘x3’ and ‘x4’ represent the coded levels of independent variables; Where, x1: development distance, x2: saturation time, x3: activation time of plate and x4: proportion of methanol in mobile phase; ranges selected for independent variables during determination of method robustness were 7 to 9 cm, 20 – 30 minutes, 8-10 minutes and 1.80 – 2.20 mL, respectively for x1, x2, x3 and x4. The x1x2, x1x3 and x1x4 represent the interaction terms. Polynomial terms x21, x22, x23 and x24 are including investigating the type of model. The considered responses were retention factor (y1) and peak area (y2).


Development of optimum mobile phase

To obtain high resolution and reproducible peaks, various mobile phase compositions were experimented. The essential parameters were found optimum with use of toluene - methanol (8:2 v/v) as mobile phase. The wavelength of 230 nm was selected to be optimal for the highest sensitivity. A sharp and well resolved peak was obtained for APL at Rf of 0.64 ± 0.05 when the chamber was saturated with mobile phase for 25 min at room temperature.

Linearity and calibration curve

The calibration curve constructed was assessed by its correlation coefficient. A fixed volume in the range of 0.5 - 3 mL was transferred from stock solution into series of 10 mL volumetric flasks and volumes were adjusted up to mark with methanol. From each volumetric flask, 5 μL of solution was applied on HPTLC plate to get concentration in the range of 250 – 1500 ng/band. After evaporation of solvents at room temperature for 25 min, chromatography was performed as described above. The linearity of APL was shown in (Figure 2). Calibration curve was developed by plotting Peak-area against drug quantity per band. Calibration equations were determined by use of linear regression analysis and correlation coefficients (r2) were calculated. No significant difference was observed in the slope of standard curve. The HPTLC chromatogram of standard for APL was shown in (Figure 3).


Precision and Accuracy

Repeatability, Intra-day and inter-day precisions were perceived using six repetitive measurements in target concentration level. The precision of developed method was evaluated in terms of % RSD.

For Repeatability, intra-day and inter-day precision % RSD values were found to be in range of 1.35, 0.19 – 0.29 and 0.28 - 0.85, respectively. Results for the precision studies are represented in (Table 1).

Accuracy study was executed by standard addition method using three different levels. Recovery experiment was evaluated by over spotting the drug standard at 80 %, 100 % and 120 % to the pre- analyzed sample and the results were re-analyzed by proposed HPTLC method; shown in (Table 2).

Detection limit (DL) and Quantification limit (QL)

The determination of DL and QL was based on the standard deviations of the responses and slopes of constructed calibration curves (n = 3) as described by ICH guidelines Q2 (R1). The DL and QL values found were 0.35 ng and 1.068 ng, respectively.

Robustness and design analysis for robustness

A Box-Behnken Design (BBD) comprising a total twenty-nine experiment runs obtained from the design matrix were subjected to experiment in order to generate the response variables (y1 and y2) shows in (Table 3). All experimental runs were performed in randomized order to minimize the effects of uncontrolled factors that may introduce biased responses. Rather than analysis of single coefficient whole model equation was used and for response surface analysis; crucial focus was given to factors whose responses are with or without significance and are considered too.

Table 1

Precision studies for Apremilast.

Drug Conc. (ng/bandIntra-dayInter-day

% Amount found (ng/band)% RSD% Amount found (ng/band)% RSD

n- number of determinations

Table 2

Recovery studies.

Initial AmountAmount of drug added (%)Amount recovered ± SD (ng/band) n=3% Recovery% RSD
50080907.07 ± 1.82101.760.44
5001001000.60 ± 0.33100.120.66
5001201111.01 ± 8.34101.831.36

n- number of determinations

Figure 2

Linearity curve of Apremilast.
Table 4

The estimates of BBD regression analysis and statistical parameters of ANOVA for robustness determination of APL.

Statistical ParametersY1 Retention factorY2 Peak Area
Coefficient of regression (r2)0.83010.7578
Adjusted coefficient of regression (r2 adj.)0.66030.5155
Standard deviation (SD)5.623.56
Degree of freedom (DF)1414
Sum of squares (SS)0.0924.189
Mean of square (MS)6.600E-0032.992E + 005
Fischer’s ratio (F-ratio)4.893.13
Table 5

Summary of Regression, validation and laboratory mixture assay parameters for NP-HPTLC

Validation ParametersResults
Regression coefficient0.998
Linearity range (ng/band)250 – 1500
Intra-day precision (n= 3, RSD, %)0.19 - 0.29
Inter-day precision (n= 3, RSD, %)0.28 – 0.85
Repeatability (n= 3, RSD, %)1.35
Accuracy100.23 %
DL0.35 ng
QL1.068 ng
Analysts I (n= 3, RSD, %)98.80, 1.49
Analysts II (n= 3, RSD, %)100.52, 1.72
in-house tablet assay99.13 ± 0.75
Figure 3

Standard of APL.
Figure 4

3D response surface plots for impact of mobile ratio, development distance, activation time and saturation time on retention factor and peak area.
Table 3

Box-Behnken Design BBD consisting of twenty-nine experiment runs.

RunsDevelopment DistanceSaturation TimeActivation TimeMobile Phase RatioRetention FactorPeak area
Figure 5

Peak - purity spectrum of Apremilast.

Evaluation of experimental fault and quantity of validity of polynomial models (lack of fit) were obtained through replication of experimental points (optimized level of variables). ANOVA was applied to obtain regression lack of fit models and the models were provided with adequate representation of data. Considering the degrees of freedom, it was indicated that the data is well fitted to regression models as depicted in (Table 4). The values of coefficients from the polynomial models (Eqs. (2) And (3)) and their signs indicates that, x1 (development distance) has negative effect on the responses y1 (retention factor) and y2 (peak area) while x2 (saturation time) x3 (activation time) and x4 (mobile phase ratio) has positive effect on the retention factor and peak area. Response surfaces from DoE showed that quadratic model suggested for the entire four variable x1, x2, x3 and x4 and depicted least influence of the saturation time, activation time and mobile phase ratio. Whereas development distances show the highest influence on peak area and retention time, respectively. When the saturation time, activation time and mobile phase ratio was kept constant and increases in the development distance to increases in the peak area (y2) and retention time (y1).

The 3D response surface plots (Figure 4) - (a - a1, b – b1, c – c1, d – d1, e - e1 and f - f1) shows the impact of mobile ratio, development distance, activation time and saturation time on retention factor and peak area.


Ruggedness of HPTLC method was performed at a concentration of 1000 ng/band. Methods were found to be rugged when analysis was performed by two different analysts under the same experimental and environmental conditions.

Assay of in-house tablets of APL

A distinct peak at Rf of 0.64 ± 0.05 was observed in the chromatogram for APL throughout HPTLC analysis. There was no obstruction observed from the excipients used in the in-house tablets of APL. The drug content ± SD found for HPTLC analysis was found to 99.13 ± 0.75.


A typical absorption spectrum of APL was shown in (Figure 5) the peak - purity of APL was hardened by correlating the spectra of APL added to laboratory at the peak-start (S), peak - apex (A) and at the peak - end (E) positions. Correlation between these spectra indicated purity of APL peak {correlation r (S, M) = 0.9995, r (M, E) = 0.9806}.

The summary of regression, validation and laboratory mixture assay parameters is represented in (Table 5), for developed NP-HPTLC analysis.


The method was successfully developed and robustness determination through DOE. Application of Box-Behnken design was used for evaluation of robustness of the method exhibited slight changes in different factors such as development distance, saturation time, activation time and mobile phase ratio had a exactly effect on a peak area and retention factor. Accordingly, specific consideration required for stringent control of this factors during the analysis of APL in chromatography. Validated method was simple, precise and rugged. Further, the method is found to be accurate and sensitive. The developed method can be used for regular analysis of APL in bulk and in pharmaceutical formulation.



  • Apremilast is immunomodulating agent which is used for management of refractory psoriatic arthritis.

  • Simple, robust, specific and rapid HPTLC method developed for has been established for the determination of Apremilast in bulk and in-house tablets.

  • Design of experiment (DoE) successfully applied for determination of robustness studies.

  • The methods were validated as per International Conference on Harmonization (ICH) guidelines.


The authors are thankful to Dr. S. J. Surana, Principal, R. C. Patel Institute of Pharmaceutical Education and Research, Shirpur, Dist. Dhule (MS) India for the facilities provided to carry out this research work.


Mr. Suraj Rajendra Chaudhari: is working as Assistant Professor, Department of Pharmaceutical Chemistry at R.C. Patel Institute of Pharmaceutical Education and Research, Shirpur, Dist. Dhule (MS), India.

Dr. Atul Arun Shirkhedkar: Is working as Vice-Principal and Head of Department of Pharmaceutical Chemistry at R.C. Patel Institute of Pharmaceutical Education and Research, Shirpur, Dist. Dhule (MS), India. His scientific contribution to the field of drug analysis is globally recognized. He has published more than hundred research articles in international and national peer reviewed journals. He also authored few books on topics related to pharmacy field. He has more than 19 years of research experience in the field of drug analysis. So far 45 students have completed their M. Pharm. Thesis and supervising 4 students for doctoral program. He has also organized more than 5 national conferences as a convener.


[3] Conflicts of interest CONFLICT OF INTEREST Authors has no conflict of interest





Design of Experiment


High - Performance Thin - Layer Chromatography


Box - Behnken Design



Iqbal M, Ezzeldin E, Al-Rashood ST, Imam F, Al-Rashood KA , authors. Determination of apremilast in rat plasma by UPLC–MS/MS in ESI-negative mode to avoid adduct ions formation. Bioanalysis. 2016;8(14):1499–508


Mease PJ, Armstrong AW , authors. Managing patients with psoriatic disease: the diagnosis and pharmacologic treatment of psoriatic arthritis in patients with psoriasis. Drugs. 2014;74(4):423–41


Xiong K, Ma X, Cao N, Liu L, Sun L, Zou Q, et al. , authors. Identification, characterization and HPLC quantification of impurities in apremilast. Analytical Methods. 2016;8(8):1889–97


Panchumarthy R, Sulthana MS, Babu PS , authors. Development and validation of stability-indicating UV spectrophotometric method for determination of Apremilast in bulk and pharmaceutical dosage form. Ind. J. Res. Pharm. and Biotech. 2017;5(1):147


Kumar L, Reddy MS, Managuli RS, Pai G , authors. Full factorial design for optimization, development and validation of HPLC method to determine valsartan in nanoparticles. Saudi Pharmaceutical Journal. 2015;23(5):549–55


Walter PK , author. Top 10 changes in FDA’s process validation guidance. BioProcess. Int. 2011;9:72


Cano CB, Felsner ML, Bruns RE, Matos JR, Almeida-Muradian LB , authors. Optimization of mobile phase for separation of carbohydrates in honey by high performance liquid chromatography using a mixture design. Journal of the Brazilian Chemical Society. 2006;17(3):588–93


Almeida AA, Scarminio IS , authors. Statistical mixture design optimization of extraction media and mobile phase compositions for the characterization of green tea. Journal of Separation Science. 2007;30(3):414–20


Heyden YV, Nijhuis A, Smeyers-Verbeke J, Vandeginste BGM, Massart DL , authors. 2001. Guidance for robustness/ruggedness tests in method validation. J Pharma bio Ana. 2001;24(5):723–53


Iriarte G, Ferreiros N, Ibarrondo I, Alonso RM, Maguregi MI, Gonzalez L, et al. , authors. Optimization via experimental design of an SPE-HPLC-UV-fluorescence method for the determination of valsartan and its metabolite in human plasma samples. J Sep sci. 2006;29(15):2265–83


Goupy J , author. What kind of experimental design for finding and checking robustness of analytical methods? Analytica Chimica acta. 2005;544(1):184–90


Dejaegher B, Heyden VY , authors. Ruggedness and robustness testing. Journal of Chromatography A. 2007;1158(1):138–57


Dejaegher B, Capron X, Smeyers-Verbeke J, Heyden VY , authors. Randomization tests to identify significant effects in experimental designs for robustness testing. Analytica Chimica acta. 2006;564(2):184–200


ICH (Q2R1), Draft Guidelines on Validation of Analytical Procedures: Text and Methodology. IFPMA; Switzerland: 1995