Near-infrared spectroscopy (NIRS) is a technique widely used for rapid and nondestructive analysis of solid samples. A method for simultaneous analysis of the two components of paracetamol and caffeine from powder blends has been developed by using chemometry with near-infrared spectroscopy (NIRS). The method development was performed on samples containing 80, 90, 100, 110, and 120% active pharmaceutical ingredients, and near-infrared spectroscopy (NIRS) spectra of prepared powder blends were recorded and analyzed in order to develop models for the prediction of drug content. Many calibration models were applied in order to perform quantitative determination of drug content in powder, and choosing the appropriate number of factors (principal components) proved to be of highly importance for a PLS chemometric calibration. Once the methods were developed, they were validated in terms of trueness, precision, and accuracy. The results obtained by NIRS methods were compared with those obtained by HPLC reference method, and no significant differences were found. Therefore, the NIR chemometry methods were proved to be a suitable tool for predicting chemical properties of powder blends and for simultaneous determination of active pharmaceutical ingredients.
Near-infrared spectroscopy (NIRS) has been proved to remain a powerful analytical tool for analyzing a vast variety of samples from petrochemical, food, agricultural, and pharmaceutical industries [
In recent years, NIRS methods are starting to become popular in the pharmaceutical area. NIRS needs chemometric analysis of data in order to be used as quantitative technique. Chemometric methods such as three linear regression modeling methods, principal component regression (PCR), partial least squares (PLS) [
In order to increase patient compliance, the combination of two or more active compounds in the same commercial preparation may be used. The manufacturing process of tablets containing two or more active compounds, such us fixed-dose combination tablets, involves the following unit operations: dispersing, granulation, mixing, tableting, and coating. Each of the unit operations may have huge influence on the quality of the final product. For example, powder mixing is an essential unit operation for manufacturing fixed-dose combination, because inadequate mixing process conducts to poor quality of the final product due to low blend uniformity that is critical to ensure compliant content uniformity per united dose [
Paracetamol (N-acetyl-p-aminophenol, acetaminophen) is a long-established substance, being one of the most extensively employed drugs in the world. For patients with sensitivity to aspirin, it is a noncarcinogenic and effective substitute and it is accepted to be a suitable drug for the relief of pain and fever in adults and children [
Caffeine (3,7-dihydro-1,3,7-trimethyl-1H-purine-2,6-dione) is an alkaloid N-methyl derivative of xanthine that is broadly distributed in natural products, commonly used in beverages. Its consumption has many physiological effects, such as gastric acid secretion, diuresis, and stimulation of the central nervous system [
HPLC methods are widely used for powder blend uniformity evaluation, due to good selectivity, specificity, and linear range. However, this technique requires sample preparation and chromatographic separation of the analytes, so it takes hours and therefore can be done only offline. Monitoring the blend uniformity in the mixing steps of tablet manufacturing is considered to be an important goal for PAT concept and can be done only by a direct analysis. Once calibrated and validated, NIRS methods seem to be the best analytical options, due to short analysis time and low cost per analysis in contrast with HPLC analysis methods.
In this paper, we explored the applications of chemometry on near-infrared spectroscopy (NIRS) for the quantitative analysis of paracetamol and caffeine, in powder blend for tableting, predicting their concentrations simultaneously, without any processing of the sample.
All the raw material powders including paracetamol (Novacyl, France) and anhydrous caffeine (BASF, Germany) as active compounds and lactose (Meggle, Germany), cornstarch (Roquette, France), colloidal silicon dioxide—Aerosil (Rohm Pharma Polymers, Germany), polyvinyl-pyrrolidone (BASF, Germany), talcum (IMERYS Luzenac, France), and magnesium stearate (Union Derivan S.A, Spain) as excipients were pharmaceutical grade.
Pharmaceutical industries frequently use wet granulation in order to convert fine cohesive powders into dense and round granules. The granules are produced by vigorous mixing of a wet-powdered mixture composed of active compounds, some excipients, and binder. The overall purpose of this operation is to obtain a final product with improved characteristics, such as better flowability and compressibility. Other benefits were obtained using wet granulation; the distribution of the drug in the final product, as well as the dissolution properties of tablets may be improved.
For calibration and validation purpose, powder blends for tablets were prepared as presented in Table
Composition of calibration and validation samples.
Concentration level | 1 | 2 | 3 | 4 | 5 |
---|---|---|---|---|---|
80% | 90% | 100% | 110% | 120% | |
Paracetamol | 38.71 | 43.55 | 48.39 | 53.23 | 58.06 |
Caffeine | 3.84 | 4.35 | 4.84 | 5.32 | 5.81 |
|
|||||
Tablets composition (mg/tablet) | |||||
Paracetamol | 240.0 | 270.0 | 300.0 | 330.0 | 360.0 |
Caffeine | 24.0 | 27.0 | 30.0 | 33.0 | 36.0 |
Lactose | 146.0 | 113.0 | 80.0 | 47.0 | 14.0 |
Cornstarch | 210.0 | 210.0 | 210.0 | 210.0 | 210.0 |
Colloidal silicon dioxide | |||||
PVP | |||||
Talcum | |||||
Magnesium stearate | |||||
|
|||||
Tablet weight | 620.0 | 620.0 | 620.0 | 620.0 | 620.0 |
Calibration samples: levels 1, 2, 3, 4, and 5. Validation samples: levels 2, 3, and 4.
Paracetamol, caffeine, lactose, some cornstarch, and PVP (which was previously dissolved in 4.8 ml distilled water) were mixed. The wet mixture was passed through a sieve. It was left to dry at room temperature until the next day; when it was weighted, the amount of remaining excipients to be added was adjusted according to the weighted mixture. The powder blend was passed through a 800
The mixture composition was designed for a tablet weight of approximately 620 mg and a usual amount of active ingredients of 300 mg paracetamol (48.38%, w/w) and 30 mg caffeine (4.84%, w/w). This formulation will be further considered as 100% active content formulation.
A calibration set was built using a full factorial experimental design with two factors and two levels, using Modde 11.0 Software (Umetrics, Sweden) to build and analyze the experimental plans. Each sample in the calibration set contained two components (paracetamol, caffeine); each component was taken at five concentration levels (80, 90, 100, 110, and 120% reported to the theoretical amount). The composition of calibration set samples is presented in a full factorial matrix of experimental plan, in Table
Experimental design matrix for calibration set.
Exp |
Run |
X1 | X2 | Exp |
Run |
X1 | X2 |
---|---|---|---|---|---|---|---|
N1 | 9 | 38.71 | 3.87 | N15 | 13 | 58.06 | 4.84 |
N2 | 26 | 43.55 | 3.87 | N16 | 25 | 38.71 | 5.32 |
N3 | 16 | 48.39 | 3.87 | N17 | 24 | 43.55 | 5.32 |
N4 | 21 | 53.23 | 3.87 | N18 | 17 | 48.39 | 5.32 |
N5 | 11 | 58.06 | 3.87 |
|
|
|
|
N6 | 15 | 38.71 | 4.35 | N20 | 7 | 58.06 | 5.32 |
|
|
|
|
N21 | 28 | 38.71 | 5.81 |
N8 | 6 | 48.39 | 4.35 | N22 | 1 | 43.55 | 5.81 |
N9 | 27 | 53.23 | 4.35 | N23 | 3 | 48.39 | 5.81 |
N10 | 23 | 58.06 | 4.35 | N24 | 18 | 53.23 | 5.81 |
N11 | 14 | 38.71 | 4.84 | N25 | 22 | 58.06 | 5.81 |
N12 | 20 | 43.55 | 4.84 | N26 | 12 | 48.39 | 4.84 |
|
|
|
|
N27 | 19 | 48.39 | 4.84 |
N14 | 10 | 53.23 | 4.84 | N28 | 5 | 58.06 | 4.84 |
X1—paracetamol concentration (mg/tablet); X2—caffeine concentration (mg/tablet).
Near-infrared spectra of powder blends were recorded using a Fourier-transform NIRS analyzer (Antaris II, ThermoElectron Scientific, USA) in reflectance sampling configuration, equipped with an indium gallium arsenide (InGaAs) detector. Since the powder samples are not homogeneous, the device is equipped with a system for the rotation of samples during the measurements, so that obtained spectrum is representative for the sample and ensures the reproducibility of measurements. Each reflectance spectrum was recorded using OMNIC software (Termo Scientific, USA) by integrating 32 scans, over the range of 11000 to 4000 cm−1, with a resolution of 8 cm−1.
After the collection of all NIR spectra from each individual powder blend, high-performance liquid chromatography (HPLC) analysis was performed for reference. Weighted powder samples were dissolved in methanol, in a 25 ml volumetric flask. The flask was vibrated in an ultrasonic bath (Transsonic T700, Germany) until complete dissolution. 5 ml were transferred to a centrifuging tube and were centrifuged (Sigma 2-16, Sartorius, Germany) for 5 min at 5000 rpm. One milliliter of the resulting supernatant was pipetted into a 10 ml volumetric flask and diluted to volume with water-acetonitrile (75 : 25, v/v).
Separately, 10 mg of paracetamol and 10 mg of caffeine were accurately weighted (using 0,01 mg analytical balance) into a 10 ml volumetric flask, and the same operation as described above was carried out to prepare the standard solution for calibration curve.
The samples were then analyzed by HPLC with UV detection. The HPLC system was a 1100 series model (Agilent Technologies, USA) consisted in a binary pump, an autosampler, a column thermostat, and a UV detector. Separation was carried out on a Gemini C18 (100 × 3.00 mm; 3
NIR spectra recorded for multivariate calibration models were previously processed using several established methods: first derivative, second derivative, standard normal variate (SNV), multiplicative scattering correction (MSC), straight line subtraction (SLS), minimum maxim normalization (MMN), in order to construct the calibration models. The partial least square (PLS) regression was conducted using multivariate analysis Opus Quant software (Bruker Optics, Germany).
This software allows validation of the chemometric multivariate calibration by the “full cross-validation.” According to this procedure, iterative calibrations were performed by removing in turn each standard from the training set and then predicting the excluded sample with that calibration [
Once a calibration is developed and favourable predictions are expected, they must be validated to be accepted for routine use. Independent sets of samples are needed for external validation. There are several validation parameters that must be determined in order to be consistent with the recommendations of the International Conference of Harmonization (ICH) and with other guidelines: accuracy, precision (repeatability and intermediate precision), linearity, and range of application. The validation was performed according to the strategy proposed by Hubert et al., [
Validation of NIR methods for both paracetamol and caffeine assays was performed considering 90%, 100%, and 110% active compound content (formulations N7, N13, and N23). Four replicates were prepared for each formulation, in three different days, resulting a 36-sample validation set. In order to see which model fits the best, linearity and accuracy profiles were computed and compared, considering ICH Q2 guideline requirements.
Development of chemometric multivariate calibration models means to calculate the calibration parameters of the obtained data after the analysis of the NIR spectral calibration set. To do this, various methods of pretreatment of spectra in combination with selecting different spectral regions and different methods of regression analysis may be used. The entire spectrum and selecting certain spectral regions containing strong absorption bands in combination with different pretreatment methods of single spectra as standard normal variate (SNV), first derivative (FD), multiplicative scattering correction (MSC), straight line subtraction (SLS), minimum maximum normalization (MMN), or combined (FD + SNV, FD + MSC, and FD + SLS) were tested for building a calibration model. Once the calibration model has been developed, the capacity of prediction was tested on the test samples used during development.
The NIR reflectance spectrum of the calibration set is presented in Figure
Reflectance spectrum of powder mixture recorded at a resolution of 8 cm−1 for a calibration set. The highlighted blue area defines the spectral ranges selected for the paracetamol, and the green area defines the spectral ranges selected for the caffeine.
Many calibration models were applied in order to perform quantitative determination of drug content in powder. Choosing the appropriate number of factors (principal components) is highly important for a PLS chemometric calibration. The number of factors for the experimental data obtained in calibration has to be chosen so that to avoid “over fitting.” There were different methods proposed in order to select the optimum number of factors [
Statistical parameters and number of PLS factors for different models proposed for paracetamol and caffeine assays in powder blends.
|
|||||||||||
Model | a | b | c |
|
e | f | g | h |
|
j | k |
Pretreatment | None | COE | SLS |
|
mMN | MSC | FD | SD | FS+SLS | FD+SVN | FD+MSC |
Spectral range |
9000-4000 | ||||||||||
Number of PLS factors | 2 | 5 | 6 |
|
5 | 5 | 5 | 5 | 5 | 3 | 3 |
|
0.978 | 0.997 | 0.997 |
|
0.998 | 0.998 | 0.997 | 0.997 | 0.997 | 0.998 | 0.997 |
RMSECV (%) | 0.966 | 0.334 | 0.311 |
|
0.261 | 0.259 | 0.328 | 0.363 | 0.326 | 0.302 | 0.337 |
Bias (%) | −0.0075 | 0.00088 | 0.00055 |
|
−0.0029 | 0.00047 | −0.0093 | 0.0010 | −0.0039 | 0.0098 | 0.0048 |
|
|||||||||||
|
|||||||||||
Model | a | b |
|
d | e | f | g | h |
|
J | k |
Pretreatment | None | COE |
|
SNV | mMN | MSC | FD | SD | FS+SLS | FD+SVN | FD+MSC |
Spectral range |
6200-4100 | ||||||||||
Number of PLS factors | 14 | 14 |
|
11 | 11 | 11 | 10 | 8 | 11 | 9 | 9 |
|
0.958 | 0.96 |
|
0.958 | 0.941 | 0.945 | 0.946 | 0.881 | 0.952 | 0.939 | 0.940 |
RMSEP | 0.137 | 0.132 |
|
0.137 | 0.163 | 0.156 | 0.155 | 0.23 | 0.146 | 0.165 | 0.164 |
Bias (%) | −0.0019 | −0.0020 |
|
−0.0036 | −0.0061 | −0.0054 | −0.0046 | −0.0029 | −0.0029 | −0.00407 | −0.0045 |
Validation results of NIR method for paracetamol and caffeine assays in powder blends.
Concentration level (paracetamol) | Mean paracetamol |
Trueness | Precision | Accuracy | |||
---|---|---|---|---|---|---|---|
Relative |
Recovery |
Repeatability |
Intermediate precision |
Relative |
Tolerance limits (mg/tablet) | ||
|
|||||||
43.55 | 43.90 | 0.805 | 100.8 | 0.384 | 0.393 | [−0.11, 1.72] | [43.50, 44.30] |
48.39 | 48.44 | 0.101 | 100.1 | 0.188 | 0.185 | [−0.32, 0.52] | [48.64, 48.64] |
53.23 | 53.07 | −0.307 | 99.69 | 0.187 | 0.163 | [−0.68, 0.06] | [53.26, 53.26] |
|
|||||||
4.355 | 4.437 | 1.880 | 101.88 | 1.34 | 3.19 | [−6.26, 9.92] | [4.08, 4.82] |
4.839 | 4.808 | −0.659 | 99.34 | 2.49 | 2.79 | [−7.53, 6.21] | [4.48, 5.14] |
5.323 | 5.137 | −1.592 | 98.41 | 1.83 | 1.63 | [−5.28, 2.01] | [4.95, 5.33] |
In case that PLS algorithms are used for the development of methods, it is known that the method of spectral data pretreatment and the number of factors (components) are critical parameters. Selecting the optimal number of factors was performed using the criterion of Haaland and Thomas [
As presented in Table
Validation of NIR methods for both paracetamol and caffeine assays was performed considering 90%, 100%, and 110% active compound content (formulations N7, N13, and N23). The linear profile, as seen in Figure
Linearity profile of the NIR—chemometic methods for paracetamol and caffeine determination. (a) paracetamol; (b) caffeine.
The
Accuracy profile of the NIR—chemometic methods for paracetamol and caffeine determination. (a) paracetamol; (b) caffeine.
The precision of the methods was evaluated by calculating two parameters: repeatability and intermediate precision at three concentration levels. Both parameters had satisfactory values for all concentration levels. The best recovery was obtained for the 100% paracetamol content, while best intermediate precision was obtained for the 110% paracetamol content. As for caffeine, the best results in terms of precision were obtained for the 90% active compound. In terms of accuracy, the
According to data presented in Figures
Once the NIR methods for the determination of paracetamol and caffeine were validated, they were applied for the active content determination in six control powder blends containing 48.39% w/w paracetamol and 4.84% w/w caffeine. The reference HPLC method has been also used for the active content assay in the same samples. The results obtained by the NIR method were compared with the values obtained by the reference HPLC method, in terms of the active content recovery, as shown in Table
Paracetamol and caffeine determination by NIR-validated method and HPLC reference method.
API | Paracetamol | Caffeine | ||
---|---|---|---|---|
Method | NIR | HPLC | NIR | HPLC |
Theoretic concentration (mg) | 48.39 | 48.39 | 4.839 | 4.839 |
Found (mg) | 48.43 | 48.82 | 4.810 | 4.674 |
Recovery (%) | 100.08 | 100.89 | 99.40 | 96.59 |
CV | 0.17 | 1.19 | 2.15 | 4.53 |
|
1.186 | 1.865 | ||
|
0.236 | 0.0915 |
As can be seen, the recovery was quite similar for both methods. Also, Student’s
The use of chemometry on near-infrared spectroscopy (NIRS) was explored for nondestructive quantitative analysis of two components in powder blends. The two components were determined simultaneously using pretreated spectra together with chemometry and PLS multivariate calibration. The models were validated in terms of trueness, precision, and accuracy, for an active content of 90, 100, and 110% for paracetamol and caffeine. Good statistical indicators were obtained. Furthermore, it was proved that the proposed methods are suitable for active pharmaceutical ingredient determination, as the results obtained are similar with those obtained by HPLC, which are used as the reference method.
Considering the results presented in this work, the NIR chemometry methods proved to be a suitable tool for predicting the chemical concentration of powder blends during preparation of fixed-dose combination tablets with paracetamol and caffeine. These methods can be used for on-line, in-line, or at-line monitoring of the blend uniformity in the mixing steps of the manufacturing process of tablets, with considerable saving in time and money in comparison with HPLC analysis.
The authors declare that there is no conflict of interest regarding the publication of this paper.
This work was supported by a grant of the Romanian National Authority for Scientific Research and Innovation, CNCS-UEFISCDI, Project no. PN-III-P2-2.1-BG-2016-0201.