Research Article
Print
Research Article
Development of amlodipine and enalapril combined tablets based on quality by design and artificial neural network for confirming of qualitative composition
expand article infoNataliia Behei, Oksana Tryhubchak, Bogdan Pryymak§
‡ R&D Department, Farmak JSC, Kyiv, Ukraine
§ National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Kyiv, Ukraine
Open Access

Abstract

Using approaches of Quality by design, namely dispersion analysis, random balance method, regression analysis and neural networks, the composition and technology tablets based on amlodipine besylate and enalapril maleate have been developed. Using dispersion analysis was determined the effect of 30 excipients on 10 pharmaco-technological parameters of tablets. With the help of the desirability function, the leaders of the excipients were selected from 6 functional groups. It was confirmed that the composition with the best pharmaco-technological parameters, determined by using statistical methods, coincides with the results studies the synthesized Feed-forward neural network. The quantities of preferable excipients at 3 levels were identified by the random balance method. The relationship between the studied factors and the quality of the tablets was described by regression equations. Based on the placement of equal output lines, the optimal composition of amlodipine tablets with enalapril was established.

Keywords

amlodipine and enalapril tablets, artificial intelligence, pharmaceutical development, quality by design

Introduction

Every year 17 million people worldwide die from the cardiovascular disease, including heart attacks and strokes. Cardiovascular disease is the leading cause of death and disability in most countries. By 2030, more than 23 million people are expected to die from these diseases. They will be the leading cause of death on the planet (Virani et al. 2020). One of the first among this pathology is hypertension, that is a major risk factor for stroke, myocardial infarction, heart, and kidney failure. Therefore, the primary task of the pharmaceutical industry is the development and implementation in the production of highly effective cardiovascular medications (Kjeldsen 2018).

In order to increase the effectiveness of pharmacotherapeutic benefit and rise safety combined drugs are obtained. The main advantage of combined antihypertensive therapy is the merging of active pharmaceutical ingredients with different mechanisms of action to achieve additional antihypertensive activities and reduce side effects (Lopatowska et al. 2018). Usage of scientifically based drugs from different pharmacological groups in one medicine is the most beneficial. Herewith, the safety and effectiveness of antihypertensive therapy is increased. Moreover, the cost of treatment is reduced and patient compliance is enhanced by single tablet regiment (Guerrero-García and Rubio-Guerra 2018).

Angiotensin-converting enzyme inhibitors (enalapril) and calcium antagonists (amlodipine) are the most often compounded in combination therapy (Pongpanich et al. 2018). The ACCOMPLISH study showed that mortality and morbidity from cardiovascular disease were reduced by amlodipine in combination with an ACE inhibitor. The progression of kidney disease was also decreased. (Okuda et al. 2018). Since the release of amlodipine in the early 1990s, it remains one of the world’s top five pharmaceutical products with sales of $ 1 billion. Amlodipine is the best agent for both monotherapy and combination therapies to reduce the severity of cardiovascular disease worldwide (Khan et al. 2021). Combinations of active pharmaceutical ingredients and their concentrations are an important part of research when choosing a treatment for arterial hypertension.

The quality and bioavailability of tablets depends on many factors. The main are the physico-chemical properties of the active pharmaceutical ingredients, correctly selected excipients and obtaining tablets with a given quality and the stability of the drug in the future. The quality of the medicinal product is ensured primarily by compliance with the pharmaco-technological parameters and their indicators for each dosage form.

According to the pharmacopoeial requirements, the main pharmaco-technological indicators of powders are bulk density and tapped density. These indicators characterize the ability of raw materials to compact. They will allow further calculation of such indicators as the loading volume of technological equipment. The criterion for evaluating the flowability of the tableting blend is flowability. The flowability study will enables to investigate the ability to fill the appropriate press form. The angle of angle of repose is an additional characteristic of the flowability of the tableting blend or granulate. For well-flowing materials, its value is 25–40° (Council of Europe 2021).

The main pharmaco-technological indicators of tablet quality are: uniformity of mass, resistance of tablets to crushing, friability of uncoated tablets, disintegration. Uniformity of mass shows the deviation of each of the 20 tablets separately from the average weight and should not exceed 5%. The resistance of tablets to crushing characterizes their strength and is defined as the average value for 10 tablets. The pharmacopoeia describes the acceptance criteria for tablets with different diameters (for tablets with a diameter of 8 mm - at least 25 N). Friability of uncoated tablets is carried out in order to find out the resistance of tablets to the action of mechanical impact, or abrasion. Friability should not exceed 1% and tablets should not have chips or cracks. Disintegration allows you to determine the time of disintegration of tablets in a liquid environment, usually water. Disintegration requirements for each dosage form are different. Uncoated tablets should disintegrate as quickly as 15 minutes (Council of Europe 2021). In addition, the appearance of the tablets was investigated which is decisive for consumer compliance.

Quality by design (QbD) is a sound approach to pharmaceutical implementation. It is taken by the developers as a basis from the beginning of the formation of research objectives till the release of final drugs (Kadam et al. 2017). This is especially important when creating and researching tablet drugs because the process of developing the optimal composition and technology of tablets is multi-stage. The application of QbD principles in drug development ensures the quality of medicines for patients and doctors, as well as reducing risks at the production stages (Singh et al. 2018). It is possible to improve the quality of the drug and its production process (Schlindwein 2018) with the help of qualitative planning of the experiment, collection of preliminary data on the studied object, inclusion of all factors and control of the drug, taking into account all pharmacological indicators (Zhang and Mao 2017). QbD-based planning brings results that minimize the amount of research to develop the optimal composition and technology of the drug.

As a general rule, the practical implementation of QbD in the development of new pharmaceutical products can go through the following steps:

  1. Define the desired performances of the product and identify the Quality Target Product Profile: including dosage form, delivery systems, dosage strength(s), etc;
  2. Identification of the Critical quality attributes: including physical, chemical, biological, or microbiological properties or characteristics of an output material including finished drug product;
  3. Identification of possible Critical Material Attributes: including physical, chemical, biological, or microbiological properties or characteristics of an input material and Critical Process Parameters: parameters monitored before or in process that influence the appearance, impurity, and yield of final product significantly;
  4. Setup and execution of Design of experiment to link Critical Material Attributes and Critical Process Parameters to Critical quality attributes and get enough information of how these parameters impact Quality Target Product Profile. Thereafter, a process of Design Space should be defined, leading to an end product with desired Quality Target Product Profile;
  5. Identify and control the sources of variability from the raw materials and the manufacturing process;
  6. Continually monitor and improve the manufacturing process to assure consistent product quality (Zhang and Mao 2017).

The principles of QbD are enshrined in the regulations of pharmaceutical development (ICH Q8 (R2)), quality risk management (ICH Q9) and pharmaceutical quality system (ICH Q10) (Sangshetti et al. 2017). Successful use of QbD has been demonstrated in the development of quetiapine fumarate prolonged-release matrix tablets (Gavan et al. 2017), during the simultaneous quantification of telmisartan, amlodipine and chlorthalidone (Palakurthi et al. 2020), and in the use of a 3D printer for printing immediate-release tablets (Than and Titapiwatanakun 2021).

One of the modern methods of medicines manufacturing is an area of artificial intelligence known as artificial neural networks. Artificial neural networks use personalized knowledge and learn from experimental data to solve complex problems. Technologies involving artificial intelligence have become universal tools that can be used everywhere at different stages of drug development, such as the identification and verification of target medicine, developing of new drugs, drug reprofiling, improving the research and development efficiency, aggregating, and analyzing biomedical information and refining the decision-making process for involving patients in clinical research (Das et al. 2016). This potential use of artificial intelligence makes it possible to counteract the inefficiencies and uncertainties that arise in classical drug development methods, minimize bias and human interference in the process (Mak and Pichika 2019; Pu et al. 2019). Artificial intelligence was used to simulate the release of a drug from tablets containing lignin as an excipient (Pishnamazi et al. 2019) and for forecasting of disintegrating oral tablets (Han et al. 2018). It was applied in the manufacture of optimized mucoadhesive buccal tablet containing flurbiprofen and lidocaine for the treatment of toothache (Hussain et al. 2016), in the development of a floating tablet containing rosiglitazone maleate (Güler et al. 2017) and in optimization of the formulation of a self-emulsifying drug delivery system containing rosuvastatin (Vu et al. 2020).

In connection with the above, the creation of medicine based on QbD for the treatment of hypertension, namely the combined tablets based on antihypertensive substances of different pharmacological groups is relevant for pharmaceutical science and practice.

The aim of the work was to develop the qualitative and quantitative composition of amlodipine and enalapril combined tablets based on QbD using an artificial neural network.

Materials and methods

Materials

Enalapril maleate from Zhejiang Huahai Pharmaceutical Co., Ltd, China and Amlodipine besylate from Anek Prayog, India were used as APIs for the development of the drug.

For the study, excipients were grouped into 6 factors by functional purpose (Table 1).

Table 1.

Factors and their levels studied in the development of amlodipine and enalapril combined tablets.

Factor Factor level
A – filler a1 – microcrystalline cellulose 101
a2 – sucrose
a3 – calcium hydrogen phosphate
a4 – corn starch
a5 – pregelatinized starch
B – disintegrant b1 – croscarmellose sodium
b2 – povidone К17
b3 – crospovidone XL-10
b4 – sodium starch glycolate
b5 – potato starch
C – binder c1 – pregelatinized starch
c2 – macrogol 20
c3 – povidone К17
c4 – povidone К30
c5 – hypromellose Е5
D – glidant d1 – aerosil 200
d2 – aeroperl 300
d3 – talc
d4 – neusilin US-2
d5 – aerosil 200 + talc (1:1)
E – lubricant e1 – magnesium stearate
e2 – calcium stearate
e3 – stearic acid
e4 – sodium stearyl fumarate
e5 – polyethylene glycol 4000
F – stabilizing f1 – sodium bicarbonate
f2 – maleic acid
f3– citric acid monohydrate
f4 – magnesium carbonate
f5 – lactic acid

Statistical methods

A generalized scheme of research on the development of amlodipine and enalapril combined tablets is shown in Fig. 1.

Figure 1. 

Research algorithm for the development of qualitative and quantitative composition of amlodipine and enalapril combined tablets.

To implement the experiment mathematical and statistical methods of planning the experiment and processing the results of the study were used.

Method of dispersion analysis

Method of dispersion analysis based on the second-order hyper-Greek-Latin square. Analysis of variance is a statistical method used to divide the total sum of squares of observations into components due to the influence of various factors, their interactions and random variables. Dispersion analysis was used to be able to statistically identify the influence of various factors on the variability of the studied feature (Hroshovyi et al. 2008). The second order hyper-Greek-Latin square made it possible to reduce the number of studies from 7776 to 27. The plan of the experiment is shown in Table 2.

Table 2.

Experimental plan based on the second-order hyper-Greek-Latin square.

Factor / Batch A B C D E
1 a1 b1 c1 d1 e1
2 a1 b2 c2 d2 e2
3 a1 b3 c3 d3 e3
4 a1 b4 c4 d4 e4
5 a1 b5 c5 d5 e5
6 a2 b1 c2 d3 e4
7 a2 b2 c3 d4 e5
8 a2 b3 c4 d5 e1
9 a2 b4 c5 d1 e2
10 a2 b5 c1 d2 e3
11 a3 b1 c3 d5 e2
12 a3 b2 c4 d1 e3
13 a3 b3 c5 d2 e4
14 a3 b4 c1 d3 e5
15 a3 b5 c2 d4 e1
16 a4 b1 c4 d2 e5
17 a4 b2 c5 d3 e1
18 a4 b3 c1 d4 e2
19 a4 b4 c2 d5 e3
20 a4 b5 c3 d1 e4
21 a5 b1 c5 d4 e3
22 a5 b2 c1 d5 e4
23 a5 b3 c2 d1 e5
24 a5 b4 c3 d2 e1
25 a5 b5 c4 d3 e2

Statistical processing of the results of experimental studies of intermediates and tablets was performed by the method of dispersion analysis. Ranked batches of advantages were built for each parameter using Duncan’s criterion. It is one of the multiple comparison procedures that are used in statistical analysis. Duncan’s multiple range test or Duncan’s test, or Duncan’s new multiple range test, provides significance levels for the difference between any pair of means (Bewick et al. 2004), therefore every indicator was studied twice. Based on them, the influence of excipients on the studied pharmaco-technological parameters was determined.

Desirability function

The leaders in all respects were selected from each group of excipients for further studies on the selection of the optimal quality composition of amlodipine and enalapril combined tablets. The choice of the best combinations of excipients in the production of tablets was carried out using a generalized quality indicator such as the desirability function. The research results were converted into dimensionless values from 0 to 1 by using the scale shown in Fig. 2. The 10th root of the product of the obtained values (D, Dʼ in Table 7) was subjected to variance analysis.

Figure 2. 

Desirability function.

The root of the tenth degree of the product of the obtained values was subjected to dispersion analysis. Based on these results, ranked batches of benefits for each functional group of excipients were built. That allowed to identify leaders among them for the introduction of tablets.

Principles of the artificial neural networks

In applied fields of science, the problem of approximating experimental data often arises. For a qualitative solution to this problem, feed-forward neural networks (multilayer perceptron). They are often called universal approximators. It is mathematically proven that with a sufficiently large number of neurons in the hidden layer, the feed-forward neural network is able to approximate any vector nonlinear function with a limited number of breakpoints with a given accuracy. The minimum required number of neurons in the hidden layer is determined heuristically based on the permissible approximation error.

The ability of neural network to high-quality approximation of complex dependencies is achieved due to the segmentation of the input space into subspaces by the action of non-linear functions of neuron activation and the subsequent interpolation of neuron data on each segment. The analysis of the work of neural network proves that in some cases input signals, some neurons work and others are passive, however, other neurons are already working for other input signals.

In this work the set task is to design a feed-forward neural network that approximates the complex nonlinear dependence of ten pharmaco-technological indicators of tablets on their six-factors qualitative composition.

Design of the feedforward neural network

The design of the artificial neural network was carried out using the commercially available Matlab software product the Neural Network Toolbox through the following five main steps: 1 - training data set collection, 2 - data processing, 3 - selection of network architecture, 4 - network training, 5 - evaluation of the quality of learning.

The training set of data was the result of 25 experiments for different level combinations of 6 qualitative factors - A, B, C, D, E, F. Each of factors had 5 levels - 1, 2, 3, 4, 5. As a result of each experiment, 10 pharmaco-technological indicators of tablets y1 ... y10 were determined. So, the set of data for training the network includes an input array of size 6×25=150 and an output array of size 10×25=250. Such a set of data is not large. However, it is representative since it was obtained based on the mathematical theory of experiment planning. As is well known the representativeness of the training data set is fundamentally necessary for the successful synthesis of a neural network.

The architecture of a neural network includes the number of inputs and outputs of the network, the number of neurons in the hidden and output layers, and the type of their activation function. The inputs to the network are 6 factors, so it will have 6 inputs. The output values of the network are 10 indicators of the tablets, so it will have 10 outputs. Accordingly, there will be 10 neurons in the output layer. To date, there is no theoretically justified method for choosing the number of neurons in the hidden layer of the S1 network. Some practical recommendations are known, but they are rather inaccurate. In general, the choice of S1 must satisfy 2 conditions. First, S1 should be small enough to avoid the phenomenon of over-learning, when the network remembers the training data but does not generalize it. Second, S1 must be large enough for the network to be able to approximate complex functional dependence.

Taking into account the specified conditions, the number of neurons in the hidden layer S1=11 was empirically selected with an activation function of the hyperbolic tangent (“tansig”) type. Neurons of the output layer usually have a linear activation function. However, we applied the “tansig” activation function because the network had higher accuracy in this case. The scheme of the synthesized feed-forward neural network with the architecture of the 6-11-10 type is shown in Fig. 3.

Figure 3. 

Scheme of the synthesized neural network.

The learning quality of the designed neural network was evaluated by regression analysis. The results are shown in Fig. 4, where along the abscissa axis of the graphs, the experimental values of the indicators of the tablets are plotted, and along the ordinate axis - the estimates of these indicators, which are formed by the neural network. These graphs show that the training quality of the network is not perfect when the regression coefficient R=1 is reached. However, the obtained training results can be qualified as pretty good.

Figure 4. 

Regression assessment of the quality of neural network training with S1 = 19.

With the help of the synthesized neural network, estimates of indicators of tablets for all possible combinations of 5 levels of 6 factors by quantity 56 = 15625 were obtained. From the obtained data, variants of the combination of investigated excipients that ensure finding of all 10 pharmaco-technological indicators in the established ranges were selected.

Method of random balance

The method of random balance was used in the study of many quantitative factors that significantly affect the object of research. The use of this method makes it possible to reduce the number of test subjects and to make a research plan to optimize the processes of tablet technology. Construction of the experimental plan was performed by random mixing of complete factor plans. Significant factors were determined using scatter plots. The significance of the selected factors was checked using the t-test (Hroshovyi et al. 2008).

Based on the results of previous studies, the excipients that showed the best results were allocated in 7 factors, that were studied at the lower «-», basic «0» upper «+» according to Table 3. The list of studied quantitative factors is given in Table 3.

Table 3.

Quantitative factors and their levels studied in the development of amlodipine and enalapril combined tablets.

Factor Factor level
lower «-» basic «0» upper «+»
х1 – amount of calcium hydrogen phosphate, g 0.14 0.15 0.16
х2 – amount of croscarmellose sodium, g 0.006 0.008 0.010
х3 – amount of povidone К17, g 0.004 0.005 0.006
х4 – amount of aerosil 200, g 0.001 0.0015 0.002
х5 – amount of talc, g 0.001 0.0015 0.002
х6 – amount of sodium stearyl fumarate, g 0.001 0.0015 0.002
х7 – amount of citric acid monohydrate, g 0.004 0.005 0.006

Using the method of random balance, an experimental plan was drawn up to study seven factors (Table 4).

Table 4.

Experimental plan for the development of amlodipine and enalapril combined tablets.

Batch х1 х2 х3 х4 х5 х6 х7
26 + + +
27 + + +
28 + +
29 + + + +
30 + + +
31 + + + +
32 + + + +
33 + + + + +
34 0 0 0 0 0 0 0
35 0 0 0 0 0 0 0

10 batches of experiments were performed, that differed in the quantitative ratio of excipients. The active substances were added to all batches in equal amounts, the amount of sucrose was added to obtain an average weight of 0.2 g per 1 tablet.

Method of regression analysis

The method of regression analysis was used to analyze and process experimental data when influencing the response of only quantitative factors. Regression analysis allows you to get a mathematical model of the process in the form of a regression equation and analyze this equation.

For a detailed study of the impact of excipients, croscarmellose sodium, povidone K17 and anhydrous citric acid were selected. A list of three quantitative factors, each of which was studied at five levels, is given in Table 5.

Table 5.

Factors and their levels studied in the development of amlodipine and enalapril combined tablets.

Factor Lower star point «-α» Lower level
«-1»
Basic level «0» Upper level «+1» Upper star point «+α»
х1 – amount of croscarmellose sodium, mg 4.318 5 6 7 7.682
х2 – amount of povidone К17, mg 4.318 5 6 7 7.682
х3 – amount of citric Acid monohydrate, mg 3.659 4 4.5 5 5.341

A symmetrical rotatable uniforms composite plan with second order was used for the study. The planning matrix of the experiment of amlodipine and enalapril combined tablets is shown in Table 6.

Table 6.

Experiment planning matrix based on symmetrical rotatable uniforms composite plan with second order.

Batch х1 х2 х3
36 + + +
37 + +
38 + +
39 +
40 + +
41 +
42 +
43
44 0 0
45 –α 0 0
46 0 0
47 0 –α 0
48 0 0
49 0 0 –α
50 0 0 0
51 0 0 0
52 0 0 0
53 0 0 0
54 0 0 0
55 0 0 0

The relationship between the studied factors and the quality of amlodipine and enalapril combined tablets was described by regression equations. After checking the statistical significance of the coefficients, taking into account the Student’s test (t5 = 2.571; p = 0.05), the adequacy of the models was checked using the F-test (F0.05; 10; 5 = 4.74). The regression equations were adequate if Fexperimental <Ftabular.

The nature of the influence of the studied factors is determined by the values and signs of regression coefficients. The second-order model for three factors is:

y=b0x0+b1x1+ b2x2+ b3x3+ b12x1x2+ b13x1x3+ b23x2x3+ b11x12+ b22x22+ b33x32 (1),

where y is the value of the response (indicator);

b0 – zero regression equation coefficient;

b1, b2, b3 – factorial regression equation coefficients;

b12, b13, b23 – interaction coefficients of independent factors;

b11, b22, b33 – quadratic regression equation coefficients;

x0 – zero factor;

x1, x2, x3 - independent factors.

The magnitude of the coefficients and the signs in front of them indicate the nature and strength of the influence of the studied factors (Hroshovyi et al. 2008).

To be able to obtain information about the interaction between factors and to obtain the optimal composition of amlodipine and enalapril combined tablets, it is necessary to determine whether there is an extremum. If so, to find its coordinates. For this regression equation, obtained by the results of the experiment, was led to the canonical (standard) expression. The canonical transformation consists in choosing a coordinate system in which the geometric analysis of the equation is greatly facilitated. When making decisions on the model of the second order, the regression equation is transformed into a model for two factors with the stabilization of others at the optimal levels for the study area. Instead of the factor corresponding to the condition bii>0 і |bi|-∑|bij|>2|bii| the optimal value was entered, and the equation was converted into an expression with two variables. Based on the transformed regression equations, equal exit lines were constructed. According to the location of the lines, the optimal amounts of the other two studied factors were chosen.

Technology and methods of analysis

Amlodipine and enalapril combined tablets were made by wet granulation. This method of obtaining tablets includes the following stages: 1) mixing powders of active pharmaceutical ingredients with excipients from groups A and B; 2) moistening the mixture of powders with solutions of binders and stabilizing substances (groups C and F) to obtain a mass that sticks to the lump, but does not stick to the fingers; 3) obtaining wet granules, i.e. wiping the wet mass through the perforated plates; drying of wet granules; 4) obtaining dry granules, for which the dry mass is wiped through perforated plates to destroy lumps and obtain homogeneous granules; 5) dusting of dry granules with substances from groups D and E; 6) compression of tablets.

The obtained granulate was investigated for loss of drying (Council of Europe 2021). The tablet mass was tested by bulk density and tapped density of powders, flowability, angle of repose according to pharmacopoeial methods. The following pharmaco-technological indicators were determined for tablets: uniformity of mass, resistance to crushing of tablets, friability of uncoated tablets and disintegration. Research methods are described in the European Pharmacopoeia.

Results

The results of the study of pharmaco-technological parameters of intermediates and tablets, and data of the desirability function, obtained on the basis of the dispersion analysis are shown in Suppl. material 1: The results of the study of pharmaco-technological parameters of intermediates and amlodipine tablets with enalapril, data of the desirability functions. By comparative analysis, it was confirmed that obtained by statistical methods of pharmaco-technological indicators of tablets are among the 10 best results (Table 7), which were determined by the synthesized feed-forward neural network.

Table 7.

Top 10 best formulations and corresponding pharmaco-technological indicators of tablets determined by means of an artificial neural network.

Batch у1 у2 у3 у4 у5 у6 у7 у8 у9 у10
a2b1c3d5e4f3 0.6231 0.7027 0.7854 8.1976 40.9032 4.7546 0.9079 215.3673 0.0689 1.1836
a2b1c1d3e4f1 0.5992 0.7145 0.7942 8.2304 38.7178 4.0668 1.1343 208.9395 0.0436 1.3524
a2b1c1d4e4f1 0.6164 0.6072 0.8266 8.8173 37.1463 3.9409 1.2119 176.8341 0.0332 2.4224
a2b1c2d3e4f1 0.6085 0.6980 0.7948 8.2771 43.4693 4.3995 1.2394 218.2921 0.0505 2.1201
a2b1c4d1e5f2 0.7954 0.7412 0.7512 7.6007 40.8676 4.7813 0.6644 236.1660 0.0674 2.0053
a2b2c4d1e5f2 0.7917 0.6909 0.7545 7.8050 39.6801 4.5692 0.6215 224.3214 0.0596 2.4497
a2b3c2d2e5f2 0.6685 0.7295 0.8038 8.5070 37.6365 4.9435 0.7308 193.9313 0.0669 2.8729
a1b1c1d2e4f1 0.6021 0.6917 0.8049 8.3470 41.9503 4.1000 0.9881 212.3585 0.0699 2.5884
a1b2c2d1e5f2 0.6679 0.7069 0.7687 7.8222 38.3734 4.9176 0.6228 229.7394 0.0512 1.9746
a1b3c1d1e5f2 0.8319 0.7510 0.7294 8.1947 37.2146 5.1709 0.6183 218.0585 0.0573 2.4652

The results of the study of the amounts of excipients by random balance in the development of amlodipine and enalapril combined tablets are shown in Table 8.

Table 8.

The results of the study of the amounts of excipients by random balance.

Batch y1 у2 y3 y4 y5 y6 y7 y8 y9 y10
26 2.03 0.720 0.787 9.3 40.4 4.5 1.48 157 0.043 10.8
27 2.62 0.704 0.790 9.6 42.6 4.5 2.05 125 0.003 8.6
28 2.81 0.643 0.693 9.2 39.3 4 0.86 104 0.006 6.3
29 2.61 0.659 0.807 8.8 40.2 5 1.17 103 0.126 6.2
30 1.73 0.721 0.831 8.6 41.0 4 1.15 135 0.009 5.2
31 2.65 0.658 0.750 9.1 37.9 5 1.26 119 0.011 7.3
32 2.30 0.692 0.764 9.5 40.6 4.5 0.73 123 0.004 6.0
33 1.30 0.646 0.778 11.1 42.0 5 1.49 60 0.100 5.5
34 2.16 0.666 0.788 10.4 42.0 5 1.42 126 0.009 7.8
35 2.16 0.711 0.807 10.3 42.3 5 1.51 123 0.028 8.0

The results of the study of the amounts of excipients by regression analysis in the development of amlodipine and enalapril combined tablets are shown in Table 9.

Table 9.

The results of the study of the amounts of excipients by regression analysis.

Batch у1 у2 у3 у4 у5 у6 у7 у8 у9 у10
36 1.44 0.642 0.768 11.6 40.6 5 1.13 146 0.139 8.05
37 0.90 0.687 0.801 10.6 41.2 5 1.21 151 0.163 8.88
38 2.43 0.626 0.767 14.4 41.0 5 0.79 142 0.165 6.48
39 2.92 0.677 0.792 14.8 39.0 5 0.79 177 0.163 6.23
40 2.07 0.630 0.793 18.6 41.8 5 1.16 125 0.173 6.60
41 1.16 0.668 0.781 13.4 41.6 5 0.92 153 0.161 8.23
42 2.33 0.639 0.819 18.5 41.0 5 0.86 121 0.197 5.62
43 1.75 0.665 0.774 13.8 42.0 5 0.92 152 0.172 7.38
44 1.87 0.655 0.786 19.5 42.8 5 0.86 127 0.172 5.93
45 1.67 0.655 0.791 14.8 41.0 5 1.29 140 0.167 7.42
46 2.04 0.687 0.793 13.7 40.8 5 1.11 142 0.153 7.18
47 1.34 0.653 0.818 18.2 42.2 5 1.17 104 0.219 5.10
48 1.27 0.642 0.768 14.7 42.6 5 0.82 128 0.139 6.78
49 1.06 0.635 0.775 16.8 41.3 5 1.31 119 0.181 6.22
50 1.75 0.639 0.799 14.9 41.0 5 1.19 118 0.210 5.83
51 1.17 0.649 0.777 16.2 41.3 5 0.81 131 0.180 6.15
52 1.08 0.643 0.790 12.0 40.5 5 0.99 133 0.181 6.53
53 1.58 0.646 0.781 15.2 41.9 5 1.30 118 0.219 6.10
54 1.50 0.646 0.774 14.4 41.2 5 0.77 135 0.146 6.95
55 1.71 0.661 0.780 13.2 41.4 5 0.90 136 0.162 6.17

Discussion

Based on the dispersion analysis of experimental data, the significance of the studied factors on the pharmaco-technological properties of powder masses and tablets was determined (responses). An analysis of these reviews revealed that the same excipient may improve one response but worsen another response at the same time. For example, when one of the studied excipients (povidone XL-10) has the best effect on the resistance of tablets to crushing but gives the worst result of the abrasion of tablets.

Statistical processing of the summary results obtained by the desirability function shows that the factors in the study sequence have the greatest influence on the studied indicators: A > E > D > F > C > B. According to the results of the desirability function among calcium fillers anhydrous dihydrogen phosphate (a3) ​​and sucrose (a2) have the same effect. Unequivocal leader among lubricants is sodium stearyl fumarate (e4). Aerosil 200 + talc (1:1) (d5) was preferred in the group of glidants. Among the stabilizing substances, citric acid predominates (f3). Of the selected binders, povidone K17 (c3) had the greatest effect on amlodipine and enalapril combined tablets. Analyzing studies on the use of these leavening agents, croscarmellose sodium is preferred (b1).

Among the top 10 combinations with optimal pharmaco-technological parameters that were shown by the artificial neural network, the composition of a2b1c3d5e4f3 coincides with the results of previous studies using statistical methods of dispersion analysis and the desirability function. The obtained data confirm that for the selection of the qualitative composition of the drug statistical processing of the experimental results can be analyzed using statistical methods of dispersion analysis and desirability function or as an alternative using artificial neural networks.

The method of random balance made it possible to reduce the number of experimental batches to 10. Using scattering diagrams, the dependence of the studied quality indicators on the change in the quantities of excipients is shown and significant factors are selected. In order to select the amounts of excipients that provided indicators in accordance with pharmacopoeial requirements, Table 10 is constructed.

Table 10.

The results of the analysis of scattering diagrams are summarized.

Factor / Indicator х1 х2 х3 х4 х5 х6 х7
у1 +* + + +
у2 –* –* +* 0 + –*
у3 + + +
у4 + +
у5 + + +* –* +* +
у6 + +* + +* +* 0
у7 + –* + + –* +
у8 –* –* + +* + + –*
у9 –* –* + +* –*
у10 +* +* –* +*
general + + 0 0

As a result of the research conducted by the method of random balance, the amount of investigated excipients in general for all quality indicators was determined.

Croscarmellose sodium, povidone K17 and anhydrous citric acid were selected for detailed study of the effect of excipients. The relationship between the studied factors and the loss in mass during drying of the granulate (y1) is described by the regression equation (FExperimental = 1.95): у1=1.45-0.20х2-0.27х2х3.

The regression equation describing the relationship between the studied factors and the bulk density of the tableting blend (y2) is as follows (FExperimental = 2.19): у2=0.647-0.012x1+0.006x2+0.08x22.

The nature of the influence of the quantities of the studied factors on the tapped density of the tableting blend (y3) is expressed by the regression equation (FExperimental = 0.50): у3=0.783-0.014х1х3+0.007х22.

The flowability of the tableting blend (y4) depends on the amount of test excipients as follows (FExperimental = 0.80): у4=14.4+1.3х1-1.1х2-1.2х3.

The regression equation describing the angle of repose of the tableting blend (y5) has the form: у5 = 41.2 (FExperimental = 2.66).

The appearance of the tablets on a 5-point rating system was excellent in all batches studied.

The regression equation for uniformity of mass (FExperimental = 0.63) has the form: y7 = 1.00. Therefore, this indicator is not affected by the studied factors, and the average value of uniformity of mass is 1.00%.

The influence of the studied factors on the resistance of tablets to crushing (y8) illustrates the regression equation (FExperimental = 2.47): у8 = 127.828-8.850x1+5.868x3+5.838x12.

The studied factors do not have a significant effect on the friability of uncoated tablets index (y9), as y9 = 0.18 (FExperimental = 0.32).

The relationship between the studied factors and the disintegration (y10) is described by the following regression equation: у10= 6.262-0.474x1+0.699x2+0.292x12.

To translate the regression equations to the canonical standard expression instead of x2 in the model enter +1. We are building new models. On the basis of the transformed models, equal exit lines were built in the x1x3 coordinate system (Fig. 5).

Figure 5. 

Lines of equal exit in the coordinate system x1x3.

Taking into account the results of placement of equal output lines, the optimum point is set at x1 = -α and x3 = + α is established. This allowed us to calculate the optimal composition of amlodipine and enalapril combined tablets (Table 11).

Table 11.

Optimal composition of tablets with enalapril and amlodipine.

Ingredient Amount
Enalapril maleate 6.950 mg 3.48%
Amlodipine besylate 5.010 mg 2.51%
Croscarmellose Sodium 4.318 mg 2.16%
Calcium Hydrogen Phosphate 166.381 mg 83.19%
Povidone К17 7.000 mg 3.50%
Citric Acid Monohydrate 5.341 mg 2.67%
Aerosil 200 2.000 mg 1.00%
Talc 1.500 mg 0.75%
Sodium Stearyl Fumarate 1.500 mg 0.75%
Total 200.000 mg 100.00%

The proposed composition was tested experimentally and the following results were obtained: loss of drying of the granules 0.81%, bulk density of the tableting blend of 0.753 g / ml, tapped density of the tableting blend of 0.807 g / ml, flowability of the tableting blend of 10.3 s / 100 g, angle of repose of the tableting blend of 41.3°, appearance of the tablets 5 points, uniformity of mass of 1.18%, resistance of tablets to crushing 142.77 N, friability of uncoated tablets of 0.17%, disintegration of 7 minutes 27 seconds.

Conclusions

The composition and technology of amlodipine and enalapril combined tablets by wet granulation were developed with the help of QbD. The effect of 25 excipients on the pharmaco-technological parameters of amlodipine tablets with enalapril was studied. The desirability function was used to select the best combinations of excipients in the tablets. Qualitative composition is confirmed by an artificial neural network. The choice of the quantities of excipients is made by the method of random balance. Using the method of regression analysis, the optimal amount of excipients for tablets was determined: amlodipine besylate - 5.010 mg, enalapril maleate - 6.950 mg, croscarmellose sodium - 4.318, calcium dihydrogen phosphate - 166.381 mg, K17 - 7 mg, citric acid anhydrous - 5 mg aerosil 200 – 2.000 mg, talc – 1.500 mg, sodium stearyl fumarate – 1.500 mg.

The authors report there are no competing interests to declare.

References

  • Council of Europe (2021) European Pharmacopoeia 10th Edn., Strasbourg, France: Council of Europe.
  • Das PJ, Preuss C, Mazumder B (2016) Artificial neural network as helping tool for drug formulation and drug administration strategies. In: Artificial Neural Network for Drug Design, Delivery and Disposition. Elsevier, 263–276. https://doi.org/10.1016/B978-0-12-801559-9.00013-2
  • Gavan A, Porfire A, Marina C, Tomuta I (2017) Original research paper. Formulation and pharmaceutical development of quetiapine fumarate sustained release matrix tablets using a QbD approach. Acta pharmaceutica 67: 53–70. https://doi.org/10.1515/acph-2017-0009
  • Güler GK, Eroğlu H, Öner L (2017) Development and formulation of floating tablet formulation containing rosiglitazone maleate using Artificial Neural Network. Journal of Drug Delivery Science And Technology 39: 385–397. https://doi.org/10.1016/j.jddst.2017.04.029
  • Han R, Yang Y, Li X, Ouyang D (2018) Predicting oral disintegrating tablet formulations by neural network techniques. Asian Journal Of Pharmaceutical Sciences 13: 336–342. https://doi.org/10.1016/j.ajps.2018.01.003
  • Hroshovyi TA, Martseniuk VP, Kucherenko LI, Vronska LP, Hureieva SM (2008) Matematychne planuvannia eksperymentu pry provedenni naukovykh doslidzhen v farmatsii. [The mathematical planning of experiment is during realization of scientific researches in pharmacya.]
  • Hussain A, Syed MA, Abbas N, Hanif S, Arshad MS, Bukhari NI, Hussain K, Akhlaq M, Ahmad Z (2016) Development of an ANN optimized mucoadhesive buccal tablet containing flurbiprofen and lidocaine for dental pain. Acta pharmaceutica 66: 245–256. https://doi.org/10.1515/acph-2016-0020
  • Kjeldsen SE (2018) Hypertension and cardiovascular risk: General aspects. Pharmacological Research: The Official Journal Of The Italian Pharmacological Society 129: 95–99. https://doi.org/10.1016/j.phrs.2017.11.003
  • Lopatowska P, Mlodawska E, Tomaszuk-Kazberuk A, Banach M, Malyszko J (2018) Adhering to the principles of clinical pharmacology-the correct fixed combinations of antihypertensive drugs. Expert Review Of Clinical Pharmacology 11: 165–170. https://doi.org/10.1080/17512433.2018.1412826
  • Okuda T, Okamura K, Shirai K, Urata H (2018) Effect of angiotensin-converting enzyme inhibitor/calcium antagonist combination therapy on renal function in hypertensive patients with chronic kidney disease: Chikushi Anti-Hypertension Trial-Benidipine and Perindopril. Journal of Clinical Medicine Research 10(2): 117–124. https://doi.org/10.14740/jocmr3253w
  • Palakurthi AK, Dongala T, Yalavarthi RK, Anireddy J (2020) QbD‐based development of an extraction procedure for simultaneous quantification of telmisartan, amlodipine besylate and chlorthalidone in combination complex matrix formulation. Biomedical chromatography 34(2): e4755. https://doi.org/10.1002/bmc.4755
  • Pishnamazi M, Ismail HY, Shirazian S, Iqbal J, Walker GM, Collins MN (2019) Application of lignin in controlled release: development of predictive model based on artificial neural network for API release. Cellulose 26: 6165–6178. https://doi.org/10.1007/s10570-019-02522-w
  • Pongpanich P, Pitakpaiboonkul P, Takkavatakarn K, Praditpornsilpa K, Eiam-Ong S, Susantitaphong P (2018) The benefits of angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers combined with calcium channel blockers on metabolic, renal, and cardiovascular outcomes in hypertensive patients: a meta-analysis. International Urology And Nephrology 50: 2261–2278. https://doi.org/10.1007/s11255-018-1991-x
  • Pu Q, Li Y, Zhang H, Yao H, Zhang B, Hou B, Li L, Zhao Y, Zhao L (2019) Screen efficiency comparisons of decision tree and neural network algorithms in machine learning assisted drug design. Science China Chemistry 62: 506–514. https://doi.org/10.1007/s11426-018-9412-6
  • Than YM, Titapiwatanakun V (2021) Tailoring immediate release FDM 3D printed tablets using a quality by design (QbD) approach. International Journal Of Pharmaceutics 599: 120402. https://doi.org/10.1016/j.ijpharm.2021.120402
  • Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP (2020) American Heart Association Council on Epidemiology and Prevention Statistics. Committee and Stroke Statistics Subcommittee. American Heart Association. Circulation 141: e139–e596. https://doi.org/10.1161/CIR.0000000000000757
  • Vu GTT, Phan NT, Nguyen HT, Nguyen HC, Tran YTH, Pham TB (2020) Application of the artificial neural network to optimize the formulation of self-nanoemulsifying drug delivery system containing rosuvastatin. Journal of Applied Pharmaceutical Science 10: 1–011.

Supplementary material

Supplementary material 1 

The results of the study of pharmaco-technological parameters of intermediates and amlodipine tablets with enalapril, data of the functions of desirability

Nataliia Behei

Data type: pharmaco-technological parameters in the table

Explanation note: Results of the study of pharmaco-technological parameters of intermediates and amlodipine tablets with enalapril.

This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0). The Open Database License (ODbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited.
Download file (46.04 kb)
login to comment