Abstract
This is a visual representation of the abstract.
Background
Various methods have been reported for improving the water-insoluble drugs in oral administration formulations. Among them, amorphization has been attracting attention and developed as a method for solubilizing API (active pharmaceutical ingredient)s by changing their physicochemical properties. Molecular complexation is also known as a method for solubilizing APIs by synthesizing cocrystals, etc. Co-amorphization, which achieves both molecular complexation and amorphization, is effective and has attracted attention. Thus, co-amorphization has been proven to be an effective approach to solubilization.
Objective
This study aims to improve the solubility of lumefantrine, used here as a model compound, through co-amorphization with deoxycholic acid.
Significance
The physicochemical properties are an important factor in developing pharmaceutical ingredients. Hydrogen-bonded co-amorphization has gained attention as a method to enhance the physicochemical properties of hydrophobic drugs.
Methods
The co-amorphous Lumefantrine-deoxycholic acid system was prepared using a mechanochemical synthesis method based on ball milling. The synthesis process was monitored by powder X-ray diffraction and near-infrared spectroscopy. The products and materials were analyzed by thermal analysis.
Results
Spectroscopic analysis revealed that the two molecules were complexed through intermolecular hydrogen bonding interactions. The produced co-amorphous has no melting point was found by thermal analysis.
Conclusions
Process monitoring also indicated the presence of a metastable crystalline Lumefantrine (LMF) intermediate.
Introduction
Various methods have been reported for improving the water-insoluble drugs in oral administration formulations. 1 Among them, amorphization has been attracting attention and developed as a method for solubilizing API (active pharmaceutical ingredient)s by changing their physicochemical properties. 2 Molecular complexation is also known as a method for solubilizing APIs by synthesizing cocrystals, 3 etc. Co-amorphization, which achieves both molecular complexation and amorphization, is effective and has attracted attention. 4 Thus, co-amorphization has been proven to be an effective approach to solubilization. 5 It has also been reported that solubilization leads to improved bioavailability. Lumefantrine (LMF) is a representative example of a poorly soluble drug, and there have been many reports on its low solubility in water. 6 Sonal et al. prepared a solid dispersion of LMF using hydroxy propyl methyl cellulose phthalate, cellulose acetate phthalate and spray drying to improve solubility. 7 Kanojiya et al. reported that amorphous dispersions were synthesized and investigated using Soluplus in combination with LMF. 8
The demand for process monitoring in pharmaceutical manufacturing is increasing. Phase transitions and chemical changes in the manufacturing process are important events that affect the quality of the final product. At-line monitoring is effective for investigating whether the material has become amorphous in the crystalline phase. NIR (near-infrared) is easy to use for monitoring because it requires a short investigation time and is a nondestructive and non-contact method. 9 However, its spectra are complex and overlapping; machine learning analysis is required. Monitoring of mechanochemical synthesis using NIR spectra and machine learning was used to evaluate solid dispersions of curcumin.10,11 It has also been applied to the evaluation of solid dispersions of rebamipide. 12 SVD (singular value decomposition) has been shown to be a useful computational method for spectral analysis as machine learning. A study of rebamipide showed that the correlation between crystal planes and functional groups can be evaluated by using multiple spectral data sets combining X-ray diffraction (XRD) and NIR and SVD. 12
Deoxycholic acid is a type of bile acid, and its use as a reagent is a biomimetic application. It is soluble in alcohol and acetic acid and also functions as a surfactant, making it a potential agent to enhance solubility.
In this study, LMF, a poorly soluble drug model, was processed into a co-amorphous form with deoxycholic acid using a mechanochemical method. The co-amorphous synthesis process was evaluated through at-line monitoring, using the correlation between XRD and NIR measurements.
Materials and methods
Materials
LMF was purchased from Tokyo Chemical Industry Co, Ltd (Tokyo, Japan). Deoxycholic acid (DEO) was purchased from Wako Pure Chemical Industries, Ltd (Osaka, Japan).
Methods
LMF-DEO co-amorphous preparation by mechanochemical synthesis
LMF (1.27 g) and DEO (1.72 g) were mixed to obtain a 1:1 co-amorphous LMF-DEO mixture. The sample was sealed in an 80 mL alumina jar along with ten balls, each with a diameter of 1.0 cm. The mixture was milled using a planetary ball mill (Fritsch) at 500 rpm for 1 h. Small amounts of the sample were taken every 10 min for measurement.
Thermal analysis
TG/DTA curves were obtained by the thermal analyzer (TG-DTA; Rigau Co. Ltd, Themo plus evo TMA 8310, Japan) at temperatures ranging from 30 to 200 °C at a heating rate of 10 °C/min.
XDR analysis
XRD patterns were obtained using an Ultima IV X-ray diffractometer (Rigaku Co, Ltd Japan) under the following conditions: Cu-Kα radiation, current: 40 mA, voltage: 40 kV, range: 2θ = 3–60°.
Attenuated total reflection (ATR)–FT infrared (IR) spectroscopy
An ATR-IR spectrometer (FT/IR-4200, JASCO Co, Tokyo, Japan) was used to obtain the spectra from 4000 to 400 cm−1 with a resolution of 1 cm−1. The air spectrum was selected as the background spectrum.
NIR spectra
NIR spectra of the samples were measured using DLP NIR scan™ NIR-S-G1 (InnoSpectra Co., Hsinchu, Taiwan). The wavelength was in the range of 900 to 1600 nm with a resolution of 4 nm.
Multiple dataset principal component analysis
Combined NIR spectra and XRD patterns of the prepared samples were constructed via SNV normalization. The combined spectra dataset was analyzed using SVD. The details of the SVD calculation are described elsewhere.13–15
Results
Figure 1 shows the thermal analysis results of the prepared sample and bulk LMF and DEO. The melting point of LMF and DEO were reported as 136 °C 16 and 174 °C, 17 respectively. Thermal analysis data in this study agree with previous reports. The prepared 60 min ground samples have no melting point from 30 to 200 °C.

Tg/DTA thermal analysis of the prepared samples: (a) LMF (b) DEO (c) co-ground LMF-DEO.
XRD was used to evaluate the crystal planes and clarify the three-dimensional periodicity of the compound molecules. Figure 2 shows the effects of grinding time on the LMF-DEO mixture samples. LMF has unique peaks at 10.72 and 16.04 degrees. DEO has pronounced peaks at 5.52 and 23.10 degrees. The peak at 5.52 degrees was observed significantly in the 1-min sample, however disappeared after 40 min of grinding. In addition, it was revealed that a broad peak was generated between 10 and 30 degrees between 20 and 60 min. Figure 2(b) also shows crystal DEO concentration changes and it decreased from the beginning of the synthesis to the end. XRD can show the presence or absence of periodic structure, and these papers then used vibrational spectroscopy to determine whether molecular complexation occurred. 18

Effect of mechanochemical synthesis on the LMF-DEO mixture samples (a) X-ray diffractograms of the prepared bulk materials and grounded samples by ball milling (b) intensity time course of 5.52 degree.
Two techniques, NIR spectroscopy and FTIR spectroscopy, were utilized to characterize the LMF-DEO mixed amorphous samples. Figure 3 shows the FTIR spectra of bulk LMF and DEO, ground for 1 min and 60 min. LMF had characteristic peaks at 3396, 2956, 1634, 1071, and 872 cm−1, which were derived from OH str, CH str, CH aroma, CH def, and CH def, respectively. 19 DEO had characteristic peaks at 3558, 2927, and 1716 cm−1, which were derived from OH str, CH str, and -COOH str, respectively. 20 Comparing the 1 min sample and the 60 min ground sample, the disappearance of OH str of DEO at 3396 cm−1, the change in the peak shape of the carboxylic acid of DEO at 1714 cm−1, and the disappearance of aroma CH of LMF at 873 cm−1 were revealed.

FT-IR spectra of the prepared samples and bulk materials. 1 min: ground 1 min sample; 60 min: ground 60 min sample of LMF-DEO mixture.
NIR spectra can be easily measured within seconds without sample preparation and have been reported to be useful for process monitoring. 21 Figure 4 shows the effect of co-grinding on the SNV-normalized NIR spectrum of the mixed powder of LMF and DEO. A peak of the hydrocarbon chain was observed around 1200 nm. The peak at 1150 nm was suggested to be the aroma CH peak unique to LMF. At 1437 nm, it was suggested to be the OH peak of the carboxylic acid of DEO. The peak at 1437 nm was observed in the 1 min sample. After 60 min of grinding, it shifted to 1417 nm due to a blue shift. This 20 nm peak shift was indicated to be due to the formation of hydrogen bonds with -COOH. The formation of hydrogen bonds in -CONH2 of ethenzamide has been reported in a previous report, 22 where a blue shift of about 20 nm was observed from 1502 nm to 1484 nm. It was suggested that the change in the characteristic vibration due to the formation of hydrogen bonds is directly shown in the spectrum.

SNV-NIR spectra of the prepared samples. The time min shows grinding times.
Discussion
The thermal analysis results suggested that the ground sample was transformed into nonperiodic phases. In thermal analysis of amorphous compounds, a positive crystallization peak may or may not be observed with increasing temperature. In this sample, neither crystallization nor a clear glass transition was observed; only a baseline shift in the endothermic signal was detected. This suggests that the material is an energetically stable amorphous form and is therefore likely to exhibit good storage stability. Co-amorphous systems, such as naproxen and cimetidine, have been similarly reported using thermal analysis. 23
It was revealed that both the LMF and DEO crystals were transformed into amorphous phases without periodic structure after 60 min. Han et al. reported a powder grinding method for beta-azelnidipine and maleic acid co-amorphous. 24 The maleic acid and beta-azelnidipine mixture was put in the vibrational disk mill for 150 min. The formation was evaluated by the wet method with the assistance of ethanol and the dry method without organic solvent. Mitsu et al. reported the co-amorphization of ketoconazole and dicarboxylic acids characterized by FTIR, Raman, XRD, and thermal analysis.
For the amorphous sample of LMF-DEO, it was necessary to investigate the intermolecular interactions between the two molecules to determine whether molecular complexes could be formed. In the case of other co-amorphous, 25 the intermolecular interactions of the co-amorphous formed from valsartan and nicotinamide show a strong influence of hydrogen bonds, which is explained based on the peak shifts of various spectroscopic spectra. The carbonyl peak of valsartan and the amide of nicotinamide are mainly important functional groups for hydrogen bonds. Thus, spectroscopic analysis is ideal for the analysis of amorphous samples.
The spectroscopy analysis data indicates that carboxylic acid interacts with LMF, which suggests that molecular complexation occurs in the amorphous phase. The disappearance of the aromatic CH of LMF suggests that stacking on the ring plane may improve the hydrophobic surface of the molecule. The fact that no crystallization was observed in the thermal analysis suggests that stacking between rings is an important indicator of crystallization, but that the planar structure of DEO makes crystallization unstable. There was no change in the amine peak at 1634 cm−1. It has been reported that the peak shifts from 1634 cm−1 to 1622 cm−1 in the lumefantrine maleic acid co-amorphous sample due to amorphization. 26 Analysis of the FTIR spectrum showed that intermolecular interactions between LMF and DEO were at work, resulting in complexation.
We have successfully performed process monitoring of co-crystallization 22 and co-amorphization 12 during ball mill grinding. We succeeded in evaluating the interactions of functional groups in pharmaceutical molecules that cannot be identified by XRD alone and showed that hydrogen-bonded complexes can be formed depending on the grinding time. NIR spectral measurements of lumefantrine are used to identify counterfeit drugs. 27 This paper shows that the spectra are complex and that clustering using multivariate analysis is effective. In our previous work, we focused on the combination of data sets and achieved the unification of XRD data and NIR spectral data, reporting a new evaluation method for solid dispersions. 12 By applying PCA to the combined data set, the crystal face information based on XRD and the intermolecular interactions and functional group information of NIR can be expressed using collinearity.
To analyze the correlation and collinearity between XRD and NIR, the normalized XRD and NIR data were integrated. The integrated data set was analyzed by SVD. Figures 5 and 6 shows the XRD and NIR loadings and scores in SVD. PC1 and PC2 were 72% and 9% explanatory variable percentages. The sum of the two, 81%, was considered reasonable for the two axes considering the consistency with the number of compositions in the sample. There are studies that analyze only the number of variables without assuming the number of main components present in the sample, but we do not agree. We note that the number of compositions that may exist should be kept in mind, considering the experimental system. PC1 in loadings showed the main XRD peak of LMF at 5.54 degrees, along with NIR peaks at 1417 nm and 1440 nm. In PC2, peaks at 14.96 degrees and 16.04 degrees were shown along with peaks at 1414 nm and 1440 nm. The change in score shows the chemical composition assigned in loadings. In other words, PC1 shows the amount of reduction due to the production of DEO, while PC2 shows the remaining intermediate of microcrystalline DEO. The peaks at 14.96 degree and 16.04 degree remain and are correlated with 1440 nm, suggesting that these crystal faces show intermolecular hydrogen bonds between deoxycholic acid dimers. It was suggested that crystal faces with hydrogen bonds require more activation energy to make the crystals amorphous. It is reasonable to think that this is because the hydrogen bonds between DEO are once peeled off and the complex with LMF is achieved.

Scores of SVD analysis based on the multiple XRD-NIR spectra dataset.

Loadings of SVD analysis based on the multiple XRD-NIR spectra dataset. (a) XRD patterns (b) NIR spectra.
Co-amorphization was demonstrated by grinding synthesis process monitoring. In the grinding process, collisions between particle interfaces generated a new phase, an amorphous phase. Many amorphous synthesis methods using grinding with polymers have been reported for lumefantrine.6,28 Co-amorphization has been reported in many studies to improve the water solubility of other compounds. The combination of Flubendazole with l-phenylalanine and l-tryptophan was achieved by co-grinding to improve water solubility. 29 The combination of ibuprofen and paracetamol has also been reported to improve solubility and co-amorphize. 30 It has been suggested that hot melt extrusion methods are also useful for synthesis. 31 The combination of Griseofulvin and L-leucine has been reported to construct co-amorphous structures with an extruder. 32 Co-amorphization is known as one of the methods to improve solubility. It was concluded that NIR is useful for industrial process analysis with quality control.
The complexation and amorphization technology with DEO obtained in this study shows great potential for improving the in vivo pharmacokinetics of low-solubility compounds such as lumefantrine. In general, poorly soluble APIs are known to have a bottleneck in their absorption in the body due to the rate-limiting dissolution in water. In this method, the amorphous phase improves the dissolution rate in the dissolution rate-limiting model, and as a result, the maximum blood concentration and AUC are expected to increase.
Furthermore, by establishing a correlation between the dissolution profile obtained in vitro and the in vivo pharmacokinetic data, the prediction accuracy of the administration design at the formulation development stage can be improved, and the efficiency of the clinical trial phase can be improved. This will clarify the safety margin based on comparison with conventional formulations, and make it possible to optimize dose setting and administration intervals.
Another major advantage of this approach is that it can be applied not only to lumefantrine, but also to other compounds with solubility issues. If a solubilization platform using DEO is established, it is believed that the commercialization of various candidate compounds will be accelerated, and ultimately contribute to improving the QOL of patients. In the future, it is anticipated that the absorption rate, exposure level, and side effect profile will be evaluated in detail in actual animal models and clinical trials to verify the usefulness and versatility of this technology.
Conclusion
This study demonstrated that near-infrared spectroscopy is effective for monitoring the synthesis of co-amorphous materials. It highlighted the technique's usefulness in mechanochemical synthesis, offering ease of use and effective time resolution for at-line monitoring. Lumefantrine was successfully amorphized, as confirmed by XRD, which showed an amorphous pattern, and FTIR, which indicated changes in intermolecular interactions. The NIR spectrum revealed an increase in the amorphous fraction over time, along with a shift in the -COOH peak. Machine learning analysis demonstrated that changes in absorbance enabled a quantitative comparison of material amounts. It is presumed that co-amorphization was achieved through the formation of hydrogen bonds between the -COOH functional groups, which increased the activation energy for the amorphous-to-crystalline phase transition of lumefantrine, thus preventing recrystallization by creating a barrier-like effect.
Footnotes
Author contributions
Y Otsuka: Investigation, calculation, original write manuscript.
K Naitou: Review and edit the manuscript.
T Miyata Review and edit the manuscript, conception.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was conducted with partial support from the Kagoshima University Research Associate Professorship system.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
No data was used for the research described in the article.
