Determination of Hardness of a Pharmaceutical Oral Jelly by Using T₂ Relaxation Behavior Measured by Time-Domain NMR

Abstract

Hardness is a critical quality characteristic of pharmaceutical oral jelly. In this study, the hardness was determined by using the T₂ relaxation curves measured by time-domain NMR. For sample preparation, kappa- and iota-carrageenans, and locust bean gum, were used as gel-forming agents. Ten test jellies with different gel-forming agent composition were prepared, and their hardness and T₂ relaxation curves were measured by a texture analyzer and time-domain NMR (TD-NMR). A negative correlation between T₂ relaxation time (T₂) and hardness was observed; however, it was difficult to determine the hardness directly from the T₂ value. That is probably because the T₂ relaxation curve contains information about molecular states, not only of water but also of the solute, and T₂ values calculated by single-exponential curve fitting only express one property of the test jelly. By considering this issue, partial least squares (PLS) regression analysis was performed on the T₂ relaxation curves for hardness determination of the test jellies. According to the analysis, an accurate and reliable PLS model was created that enabled accurate assessment of the hardness of the test jellies. TD-NMR enables the measurement of samples nondestructively and rapidly with low cost, and so could be a promising method for evaluation of the hardness of pharmaceutical oral jellies.

Introduction

Pharmaceutical oral jelly is relatively new oral dosage form. It is considered a highly acceptable dosage form for patients with swallowing difficulties especially for those who suffer from dysphasia. This dosage form has advantages including the ability to take the medication without water and a lower risk of accidental ingestion.^1,2) In Japan, jelly for oral administration has been listed in the Japanese Pharmacopeia since 2011 (16th edition, JPXVI). Accordingly, pharmaceutical oral jellies containing alendronate sodium, acyclovir, cilostazol, granisetron, and donepezil hydrochloride have been released on the market.

Texture is a critical quality attribute of jelly products that enables the development of good jelly products. In general, texture analyzers^3–5) and rheometers^6–8) have been used to evaluate the texture of jelly formulations. They can measure hardness, viscosity, and viscoelastic properties (i.e., storage modulus and loss modulus). However, these measurements are destructive and time-consuming, and are therefore difficult to use as quality control tools for manufacturing products.

The present study evaluated the hardness of sample jellies using NMR relaxation parameters. The spin–lattice relaxation time, T₁, is the time constant with which the net magnetization of the spin ensemble returns to equilibrium by rearranging with the external magnetic field, whereas the spin–spin relaxation time, T₂, reflects the fact that the transverse relaxation is the time constant with which the transverse polarization of the spin ensemble returns to its zero equilibrium value after excitation by a radiofrequency pulse.^9–12) These NMR parameters represent the molecular mobility of the measured substances. In particular, it is known that T₂ becomes monotonically shorter with molecular mobility restriction. Furthermore, the measurement time for the T₂ relaxation time is much shorter than for T₁.¹²⁾ To date, numerous technical reports show that the T₂ values of jellies and hydrogels are negatively correlated with their hardness. The tested samples contained hydrolysate-polysaccharide gel,⁸⁾ gelatin hydrogels cross-linked by cellulose nanowhiskers,⁶⁾ myofibrillar gels,^3,5) and polyvinyl alcohol cryogel.¹³⁾

The aim of this study was to find a new method for hardness evaluation of jelly formulations. The key element of this method is the measurement of T₂ relaxation behavior by using time-domain NMR (TD-NMR).¹²⁾ TD-NMR is a low-field (20 MHz) benchtop ¹H-NMR instrument specialized in the investigation of NMR relaxometry. It enables rapid and easy measurement of these NMR parameters, irrespective of the physical state of the sample (i.e., the measurement is applicable to both liquid and solid samples). TD-NMR has been used in various fields of research and industry to investigate the physicochemical properties of samples in terms of molecular mobility. The applications have also expanded into the fields of chemistry,^14,15) food,^16–18) and plants.^19,20) We applied TD-NMR analysis to evaluate characteristics of pharmaceuticals such as the crystalline state of active pharmaceutical ingredients in solid dosage forms,^12,21) the state of water incorporated in wet granules,²²⁾ the dispersion state of pharmaceutical nanosuspensions,^23,24) and the hygroscopic behavior of disintegrants.²⁵⁾

The jellies tested in this study were prepared with different compositions using ternary gel-forming agents. The T₂ relaxation curves and the hardness of the test jellies were measured by TD-NMR and texture analysis. The relationships between these properties were also evaluated. The T₂ relaxation curves were analyzed by partial least squares (PLS) regression, which is a common chemometric method. The PLS regression enabled us to determine precisely the hardness of the tested jellies from the T₂ relaxation curves.

Experimental

Materials

Kappa carrageenan (κ-CG), locust bean gum (LBG), and D-sorbitol were purchased from FUJIFILM Wako (Osaka, Japan). Iota carrageenan (ι-CG) was purchased from Tokyo Chemical Industry (Tokyo, Japan). All other chemicals were of analytical grade and commercially available.

Preparation of Sample Oral Jelly

Ten different jellies with different compositions of gel-forming agents (i.e., κ-CG, ι-CG, and LBG) were prepared as a sample (Fig. 1). The designated amounts of the gel-forming agents were dissolved in purified water at 90 °C. Next, D-sorbitol and KCl were added to the solution and stirred until these additives were completely dissolved. The resultant solution was poured into a glass vial, which was cooled at room temperature to obtain the jelly. The final concentrations of the gel-forming agent, D-sorbitol, and KCl were adjusted to 1.0, 45.0, and 0.3%, respectively.

Fig. 1. Composition of Gel-Forming Agents for the Preparation of Test Jellies

The final concentration of the gel-forming agent in the jelly was fixed at 1.0%. The maximum proportions of κ-CG, ι-CG, and LBG were 100, 75, and 75% of the total amount of gel-forming agents, respectively. Ten test jellies with different gel-forming agent compositions were prepared.

Hardness

The hardness of the prepared jellies was measured using a texture analyzer (EZ-SX; Shimadzu, Kyoto, Japan) equipped with a 5 N load cell. The probe used in this study had a round-shaped flat surface with a 10 mm diameter. The probe moved downward at 1 mm/s to a depth of 5 mm, and the maximum stress (N) on the probe surface was recorded as the hardness of the sample.

T₂ Measured by TD-NMR

The ¹H T₂ relaxation behavior of the samples was measured by TD-NMR using a Bruker minispec mq20 (Bruker BioSpin, Billerica, MA, U.S.A.) at a ¹H frequency of 20 MHz at 25 °C. For the T₂ measurement, the Carr–Purcell–Meiboom–Gill (CPMG) pulse sequence was used. The acquisition parameters were as follows: number of scans, 8; time between each pulse (τ spacing), 0.25 ms; recycle delay, 10 s; echo number, 8000. The NMR signal was monitored for approx. 4 s. T₂ was also calculated using the TD-NMR Analyze software. The single- and biexponential curve fittings were conducted using Eqs. (1) and (2), respectively:

  

(1)

  

(2)

where I(t) and I₀ are the transverse magnetization at times t and 0 with exponential decay, t is the acquisition time, and T₂ is the T₂ relaxation time. T₂(short) and T₂(long) are the T₂ relaxation times of components having fast and slow relaxation rates. P_short is the proportion of protons corresponding to T₂(short).

Additionally, the relaxation curve was fitted to a multiexponential curve using CONTIN equipped with the TD-NMR Analyze software (Bruker BioSpin). CONTIN is based on the inverse Laplace transform algorithm, which transforms CPMG data to distributions of T₂. The mathematical formulation of CONTIN is described elsewhere.²⁶⁾

NMR Spectroscopy

After the preparation of the test jellies, for NMR experiments, designated amounts of D₂O were added to each sample. The final concentration of D₂O was set at 10%. All NMR experiments were performed at 25 °C on a Bruker Avance 800 MHz spectrometer equipped with a cryoprobe.

Response Surface

Based on the experimental data, a correlation model between compositions of gel-forming agents and responses (hardness and T₂) was constructed using response surface methodology (RSM). For RSM, the dataNESIA software, version 3.0 (Azbil Corp., Tokyo, Japan) was used, with multivariate spline interpolation (RSM-S).

PLS Regression

The experimental data were analyzed with the statistical software JMP^® 15 Pro (SAS Institute, Cary, NC, U.S.A.). A PLS regression²⁷⁾ was used to build up a regression model between input variables (NMR signal intensity at time t) and response variables (hardness). This study employed the data at 80 time points ranging from 0.522 to 2014 ms as input variables. The PLS models were evaluated using a leave-one-out cross-validation (LOOCV) method. In the LOOCV method, the PLS model is built with n–1 data items, and the model is then validated by the one datum remaining. The modeling and validation process are repeated n times, changing the one datum left sequentially. As judgment statistics for the optimal PLS model (to determine the number of latent factors), the root mean-predicted residual error sum of squares (PRESS) was employed, which is defined as the square root of the average of PRESS values across all output variables. This analysis was also performed by the statistical software JMP^® 15 Pro. The variable importance in projection (VIP)^28–30) is a measure of the importance of each input variable in PLS modeling. The VIP score of variable j is given by Eq. (3)³⁰⁾:

  

(3)

where w_jk is the weight value for the j variable and k-th component, SSY_k is the sum of squares of explained variance for the k-th component, J is the number of X variables, TSSY is the total sum of squares for the dependent variables, and F is the total number of components. The term w_jk² gives the importance of the j variable in each k-th component and VIP_j is a measure of the global contribution of the j variable to the PLS model.

Results and Discussion

Evaluation of T₂ and Hardness of the Test Jellies

The hardness was measured first and the values ranged from 3.22 ± 0.73 to 12.2 ± 0.46 N (Fig. 2). The T₂ relaxation curves of the samples were also acquired (Fig. 3a) and then their T₂ values were calculated from Eq. (1). The initial signal intensity of each T₂ relaxation curve was normalized as 1 (Fig. 3a). As shown, slight but obvious differences were observed from the T₂ relaxation curves. The calculated T₂ values are presented in Fig. 3b and Table 1. The shortest T₂, 453.4 ± 4.2 ms, was observed for Sample#8, and the longest T₂, 543.2 ± 1.3 ms, was observed for Sample#4.

Fig. 2. Hardness of the Test Jellies

Each experimental value represents the mean ± S.D. (n = 4).

Fig. 3. T₂ Relaxation Curves (a) and T₂ Relaxation Times (b) of the Test Jellies

T₂ relaxation curves were measured by CPMG. The NMR signal intensity of each T₂ relaxation curve was normalized so that the initial values were set at “1.” The T₂ relaxation times were calculated according to Eq. (1). Each experimental value represents the mean ± S.D. (n = 4).

Table 1. T₂ Relaxation Times of the Test Jellies

Sample#	Composition (A/B/C)	T2 (ms)	Biexponential curve-fitting analysis
			T2(short) (ms)	T2(long) (ms)	Pshort
#1	100/0/0	456.8 ± 16.3	208.0 ± 16.8	561.0 ± 23.2	0.325 ± 0.014
#2	75/25/0	465.5 ± 23.4	195.7 ± 15.0	567.6 ± 27.4	0.310 ± 0.012
#3	50/50/0	498.2 ± 15.7	196.6 ± 10.0	598.2 ± 19.4	0.288 ± 0.006
#4	25/75/0	543.2 ± 1.3	207.5 ± 1.2	650.1 ± 0.7	0.283 ± 0.003
#5	75/0/25	467.7 ± 7.3	218.1 ± 4.9	582.9 ± 7.6	0.344 ± 0.008
#6	50/25/25	478.6 ± 9.9	213.3 ± 8.9	585.8 ± 14.2	0.319 ± 0.006
#7	25/50/25	507.9 ± 1.4	209.6 ± 2.8	608.3 ± 3.2	0.287 ± 0.005
#8	50/0/50	453.4 ± 4.2	218.0 ± 8.5	564.2 ± 10.4	0.346 ± 0.016
#9	25/25/50	517.9 ± 2.7	211.4 ± 1.8	619.0 ± 3.3	0.284 ± 0.001
#10	25/0/75	520.0 ± 1.2	213.4 ± 2.5	617.0 ± 2.8	0.275 ± 0.003

The components were as follows: A, kappa carrageenan (κ-CG) (%); B, iota carrageenan (ι-CG) (%); C, locust bean gum (LBG) (%). Each experimental value represents the mean ± standard deviation (n = 4).

To visualize the relationships between T₂ and the hardness of the jellies with different gel-forming agent compositions, their response surfaces were generated by the RSM-S (Figs. 4a, b), which is a nonlinear response surface method integrating a multivariate spline interpolation to generate the response surface.³¹⁾ RSM-S can estimate nonlinear relationships between factors and characteristics with high accuracy. Further details of this technique are described elsewhere.³¹⁾ The response surfaces are expressed in a color scale: a red or blue region indicates a large or small value of the response, respectively. The regions of the top, left-base and right-base vertices correspond to κ-CG-, ι-CG-, and LBG-enriched jellies, respectively. According to these response surfaces, the characteristics appeared to negatively correlate with each other to a certain extent. For example, the ι-CG-enriched jelly (around the left-base vertex) showed longer T₂ values and lower hardness, whereas the κ-CG-enriched jelly (around the top vertex) showed shorter T₂ and higher hardness values. Scatterplots of the T₂ values versus hardness are shown in Fig. 4c. The determination coefficient (R²) was 0.505. This level was lower than we expected, and it was difficult to determine the hardness from the T₂ values.

Fig. 4. Correlation between Hardness and T₂ Relaxation Time Evaluated by the Response Surface Method (RSM)

Response surfaces for hardness (a) and T₂ relaxation time (b) were created by RSM with multivariate spline interpolation (RSM-S). (c) Scatterplot of the experimental data between hardness and T₂ relaxation time.

CGs are sulphated linear polysaccharides extracted from different species of Rhodophyta (red seaweed): Gigartina, Chondrus crispus, Eucheuma, and Hypnea.³²⁾ There are three commercially popular CGs, namely ι-, κ- and lambda (λ)-CGs, which are classified according to the degree of substitution on their free hydroxyl groups. Among commercial CGs, κ- and ι- are gel-forming CGs, whereas λ-CG is characterized only as a thickener agent. In particular, κ-CG is known to be a strong gel-forming agent.³³⁾ Accordingly, in the present study, the hardness of the test jellies increased with a higher weight ratio of κ-CG (Fig. 4a). LBG is a galactomannan obtained from seed endosperm of the carob tree (Ceratonia siliqua).³⁴⁾ It is widely used as an additive in various industries including food, pharmaceuticals, paper, textile, oil well drilling, and cosmetics because of its ability to form hydrogen bonding with water molecules. LBG is used in combination with other hydrophilic polymers including CGs and xanthan gum.^34–36) Martins et al. reported the synergistic action of LBG and κ-CG in enhancing the elasticity and strength of κ-CG films³⁵⁾; based on Fourier transform infrared spectroscopy analysis, they showed that improvement in the mechanical strength of the film is due to the hydrogen bond interaction between LBG and κ-CG. The present study also showed the synergistic action between LBG and κ-CG. For example, at lower κ-CG concentrations, the hardness values of LBG-enriched jellies (around the right-base vertex) were clearly higher than those of ι-CG enriched jellies (around the left-base vertex).

Numerous reports show that T₂ is correlated with the hardness of jellies and gels. Chen et al. investigated the relationship between water molecular mobility and the rheological property of a hydrolysate-polysaccharide composite gel consisting of konjac glucomannan and κ-CG (GSH-P gel).⁸⁾ The gels were used before or after autoclaving treatment (121 °C, 10 min) as a test sample and the viscoelastic modulus was strengthened significantly by the autoclaving treatment. Accordingly, a shorter T₂ was observed from the GSH-P gel after the autoclaving treatment, indicating that the water molecular mobility of the gel was more tightly restricted. Dash et al. assessed the effect of cross-linking on the gel properties of gelatin hydrogel cross-linked by cellulose nanowhiskers.⁶⁾ Because the T₂ relaxation behavior is sensitive to local chain dynamics of the polymer and can determine the relative amount of rigid and mobile components, they were successful in understanding the cross-linking structure of hydrogel based on T₂ measurements. Namely, they identified that T₂ values of the hydrogel were shortened with increasing degree of cross-linking. In addition, the resultant hydrogel with a low T₂ possessed rigid and less swellable properties. Wu et al. examined the effect of addition of starch and lipid to the properties of myofibrillar protein composite gels,³⁾ and reported that adding esterified potato starch and emulsified lipids (lard and peanut oil) to composite gels caused shorter T₂ values, higher storage moduli, higher hardness, and higher water-holding capacity compared with the control hydrogel. Chu and Rutt studied the effect of freeze–thaw cycles on the gel properties of polyvinyl alcohol cryogels.¹³⁾ The rigid cryogels appeared to have shorter T₁ and T₂ values because these NMR parameters constantly decreased with repeated freeze–thaw cycles.

Based on these studies, we expected the T₂ values of the test jellies to be highly correlated with the hardness; however, the determination coefficient was lower than our expectation. The possible reason for this was the more complicated than expected molecular status of the test jellies. In other words, Eq. (1) could not express the complete information that the T₂ relaxation curves possessed about the molecular status of the test jellies. To clarify this issue, the measured T₂ relaxation curves were analyzed by multicomponent curve fitting and then the distribution of T₂ of protons in the test jelly was identified (Fig. 5). Accordingly, all test jellies showed two main peaks at about 200 and 500–700 ms, indicating that the test jellies have two major components showing distinct ¹H T₂ values. NMR spectra were also acquired and are indicated in Fig. 6. The NMR spectra also showed two major components, including water (4.7 ppm) and D-sorbitol (3.3–3.7 ppm). Taken together, it can be suggested that the two major components observed in Fig. 5 were derived from water and D-sorbitol. Based on these findings, the T₂ relaxation curves were reanalyzed by biexponential curve fitting (Eq. (2)), and the calculated parameters are listed in Table 1. The T₂(short) and T₂(long) values were fully consistent with those observed from the multicomponent curve fitting (Fig. 5). We further focused on P_short, which is the proton fraction having T₂(short). The calculated P_short values ranged from 0.275 ± 0.003% to 0.346 ± 0.016%. The P_short values were equivalent to the proportions of the peak area of D-sorbitol calculated from NMR spectra (Fig. 6) to the entire peak area. Therefore, we concluded that the T₂(short) values corresponded to the T₂ of D-sorbitol, and the T₂(long) to the T₂ of the water contained in the test jellies. In addition, scatterplots of T₂ vs. T₂(short) and T₂(long) are presented in Figs. 7a and b, respectively. The values of T₂(long) were highly correlated with T₂ (r = 0.977), but the correlation coefficient between T₂(short) and T₂ (r = 0.150) was very low, indicating that T₂ values of the test jellies calculated from Eq.(1) largely express the molecular mobility of their contained water.

Fig. 5. T₂ Relaxation Time Profiles of the Test Jellies

Fig. 6. ¹H-NMR Spectrum of the Test Jelly (Sample#1)

Fig. 7. Scatterplot of T₂ vs. T₂(Short) or T₂(Long)

T₂ values were calculated by single-exponential curve-fitting according to Eq. (1), whereas the values of T₂(short) and T₂ (long) were calculated by biexponential curve fitting according to Eq. (2).

These findings demonstrated that the T₂ relaxation curves obtained from test jellies contained a wide range of molecular state information that was related not only to water but also to D-sorbitol. We assume that the hardness of the jelly was significantly affected by the molecular state of the solute (i.e., D-sorbitol) as well as that of water, resulting in the lower correlation between T₂ and hardness.

PLS Regression to Precisely Quantify Jelly Hardness from T₂ Relaxation Curves

Precise determination of hardness of the test jellies can likely be achieved by considering the complete molecular state information contained in the T₂ relaxation curve. We therefore applied PLS regression analysis to estimate the hardness of the test jelly by analyzing the T₂ relaxation curves. PLS regression is a method for establishing a prediction model between two variable matrices: the X matrix for the input variables and the Y matrix for the response variables.²⁷⁾ It is a particularly powerful method to analyze data with strongly collinear (correlated) noisy and numerous input variables. The PLS method has already been applied in the pharmaceutical industry.^27,37–39)

The measurements of T₂ relaxation curves were performed four times for each formulation, and these data were divided into three calibration and one validation data sets. The PLS model was created from the calibration data, and then the accuracy of the estimation was evaluated by using the validation data. The resultant optimal PLS model consisted of three factors. The PLS regression model explained 70.5% of the variance of the hardness data; the percent of variation was explained by the PLS factors: factors 1, 2, and 3 were 47.9, 15.1, and 7.6%, respectively (Fig. 8a). Scatterplots of experimental and predicted hardness are shown in Fig. 8b. The PLS model showed acceptable R² and root mean square error (RMSE) values: the R² values for the calibration and validation data were 0.706 and 0.640, and their RMSE values were 1.509 and 1.708, respectively. Thus, the molecular status of the test jelly components was significantly affected by the hardness and a precise determination of the hardness can be performed by considering these molecular states.

Fig. 8. PLS Regression Analysis for Determination of Hardness of the Test Jellies

(a) Contribution of PLS factors to the variance in hardness data. (b) Scatterplots of experimental vs. predicted values for calibration and validation data sets. VIP values (c) and loading scores (d) of input variables of the PLS models for hardness determination.

The constructed PLS regression model was further investigated to establish the crucial phase of the T₂ relaxation curve for hardness determination. Given the VIP values, the mode of contribution of input variables to the PLS model can be understood. VIP values >0.8 are recognized as meaningful.²⁸⁾ As shown in Fig. 8c, the VIP values were >0.8 throughout the monitoring period, indicating that all the input data were meaningful; in particular, the values at the initial phase of the relaxation curves were clearly high. We further examined the loading score plots for PLS factors (Fig. 8d). A large absolute value of the loading score means that the corresponding input variables are more important for the PLS factors. For PLS factor 1, which is the most significant factor in the PLS model, the loading score appeared to be constant. We believe that PLS factor 1 mainly expresses the contribution of the molecular state of water. Undoubtedly, water is the dominant component of the test jellies. Furthermore, we have already confirmed that the T₂ values calculated by single-exponential curve fitting expressing an overall change in the T₂ relaxation curve were entirely consistent with T₂(long), which corresponds to the molecular state of water (Fig. 7b). For PLS factor 2, higher absolute values were observed only in the initial phase, from which it can be interpreted that PLS factor 2 is derived from the component having shorter T₂ values; thus, PLS factor 2 likely indicates the effect of the molecular state of D-sorbitol. For PLS factor 3, although the precise reason for this behavior remains unclear and needs further investigation, one possible reason is that this factor express the effect of the gel-forming agents. In addition, a part of water and D-sorbitol might show distinct molecular states that differ from those corresponding to T₂(long) or T₂(short) by the interaction with the gel-forming agent. Therefore, although the PLS model was constructed by largely considering the molecular state of water, the contribution of the molecular state of the other components is also considerable.

The PLS regression analysis on the T₂ relaxation curves enabled an accurate and reliable determination of the test jelly hardness values. In the course of the experiments, we found that the T₂ relaxation curves contain molecular state information of various components of the test jellies and these molecular states significantly affect the hardness. PLS is a common chemometric method that has already been applied to process analytical technology for manufacturing pharmaceuticals.^27,37–39) By integrating PLS regression analysis and T₂ measurements, TD-NMR is a promising method for quality control for manufacturing pharmaceutical oral jelly products. We recently demonstrated that the TD-NMR method is applicable for the determination of water content in granules for use in manufacturing tablets.⁴⁰⁾

Conclusion

We report a novel method for hardness estimation of pharmaceutical oral jelly. The key technology for this is evaluation of the molecular status of the jelly by measuring the T₂ relaxation curves using TD-NMR. In the initial phase, we demonstrated a certain correlation between T₂ and hardness, although it was not sufficient to determine the hardness directly from the values of T₂. To address this issue, PLS regression was applied to the analysis of T₂ relaxation curves and a reliable and accurate PLS model for hardness estimation of the test jellies was obtained. TD-NMR enables the measurement of samples nondestructively and rapidly with low cost, and thus shows promise for evaluation of the hardness of pharmaceutical oral jellies.

Acknowledgments

This study was supported by a Grant-in-Aid for Scientific Research from Toyama Pharmaceutical Valley Development Consortium.

Conflict of Interest

The authors declare no conflict of interest. The Laboratory of Pharmaceutical Technology, University of Toyama is an endowed department, supported by an unrestricted grant from the Nichi-Iko Pharmaceutical Co. Ltd. (Toyama, Japan).

References

• 1) Imai K., Clin. Interv. Aging, 8, 681–688 (2013).
• 2) Kim K. H., Jun M., Lee M. K., Pharmaceutics, 12, 1073 (2020).
• 3) Wu M., Hu J., Gu X., Wang Q., Wei R., Wang J., Li Z., Liu R., Ge Q., Yu H., J. Sci. Food Agric., 102, 2731–2740 (2022).
• 4) Goetz L., Foston M., Mathew A. P., Oksman K., Ragauskas A. J., Biomacromolecules, 11, 2660–2666 (2010).
• 5) Gravelle A. J., Barbut S., Marangoni A. G., Sci Rep, 7, 11544 (2017).
• 6) Dash R., Foston M., Ragauskas A. J., Carbohydr. Polym., 91, 638–645 (2013).
• 7) Agarwal D., MacNaughtan W., Foster T. J., Carbohydr. Polym., 185, 112–119 (2018).
• 8) Chen L., Liu X., Yang Y., Xue C., Xue Y., J. Texture Stud., 51, 323–332 (2020).
• 9) Schumacher S. U., Rothenhausler B., Willmann A., Thun J., Moog R., Kuentz M., J. Pharm. Biomed. Anal., 137, 96–103 (2017).
• 10) Stueber D., Jehle S., J. Pharm. Sci., 106, 1828–1838 (2017).
• 11) Cooper C. L., Cosgrove T., van Duijneveldt J. S., Murray M., Prescott S. W., Soft Matter, 9, 7211–7228 (2013).
• 12) Chiba Y., Okada K., Hayashi Y., Leong K. H., Kumada S., Onuki Y., Chem. Pharm. Bull., 70, 162–168 (2022).
• 13) Chu K. C., Rutt B. K., Magn. Reson. Med., 37, 314–319 (1997).
• 14) Kim Y. R., Yoo B. S., Cornillon P., Lim S. T., Carbohydr. Polym., 55, 27–36 (2004).
• 15) Kimoto H., Tanaka C., Yaginuma M., Shinohara E., Asano A., Kurotsu T., Anal. Sci., 24, 915–920 (2008).
• 16) Abraham A., Elkassabany O., Krause M. E., Ott A., Magn. Reson. Chem., 57, 873–877 (2019).
• 17) Chen L., Tian Y., Sun B., Wang J., Tong Q., Jin Z., Food Chem., 233, 525–529 (2017).
• 18) Colnago L. A., Azeredo R. B., Marchi Netto A., Andrade F. D., Venancio T., Magn. Reson. Chem., 49 (Suppl. 1), S113–S120 (2011).
• 19) Niu L., Li J., Chen M. S., Xu Z. F., Ind. Crops Prod., 56, 187–190 (2014).
• 20) Rolletschek H., Fuchs J., Friedel S., Borner A., Todt H., Jakob P. M., Borisjuk L., Plant Biotechnol. J., 13, 188–199 (2015).
• 21) Okada K., Hayashi Y., Kumada S., Onuki Y., J. Pharm. Sci., 109, 2577–2584 (2020).
• 22) Ito T., Okada K., Leong K. H., Hirai D., Hayashi Y., Kumada S., Kosugi A., Onuki Y., Chem. Pharm. Bull., 67, 271–276 (2019).
• 23) Ito T., Kida C., Okada K., Ikeda J., Otomura K., Hayashi Y., Obata Y., Takayama K., Onuki Y., Langmuir, 34, 12093–12099 (2018).
• 24) Okada K., Hayashi Y., Kumada S., Onuki Y., Eur. J. Pharm. Sci., 164, 105908 (2021).
• 25) Ougi K., Okada K., Leong K. H., Hayashi Y., Kumada S., Onuki Y., Eur. J. Pharm. Sci., 154, 105502 (2020).
• 26) Provencher S. W., Comput. Phys. Commun., 27, 213–227 (1982).
• 27) Wold S., Sjöström M., Eriksson L., Intell. Lab. Syst., 58, 109–130 (2001).
• 28) Akarachantachote N., Chadcham S., Saithanu K., Int. J. Pure. Appl. Math., 94, 307–322 (2014).
• 29) Farrés M., Platikanov S., Tsakovski S., Tauler R., J. Chemometr., 29, 528–536 (2015).
• 30) Takayama K., Sato T., Sato K., Todo H., Obata Y., Sugibayashi K., J. Drug. Deliv. Sci. Technol., 52, 1021–1031 (2019).
• 31) Takayama K., Obata Y., Morishita M., Nagai T., Pharmazie, 59, 392–395 (2004).
• 32) Campo V. L., Kawano D. F., Silva D. B., Caevahlo D. J., Carbohydr. Polym., 77, 167–180 (2009).
• 33) Vigani B., Rossi S., Gentile M., Sandri G., Bonferoni M. C., Cavalloro V., Martino E., Collina S., Ferrari F., Mar. Drugs, 17, 153 (2019).
• 34) Barak S., Mudgil D., Int. J. Biol. Macromol., 66, 74–80 (2014).
• 35) Martins J. M., Cerqueira M. A., Bourbon A. I., Pinheiro A. C., Souza B. W. S., Vicente A. A., Food Hydrocoll., 29, 280–289 (2012).
• 36) Goycoolea F. M., Morris E. R., Gidley M. J., Carbohydr. Polym., 27, 69–71 (1995).
• 37) Dunko A., Dovletoglou A., J. Pharm. Biomed. Anal., 28, 145–154 (2002).
• 38) Mantanus J., Ziémons E., Lebrun P., Rozet E., Klinkenberg R., Streel B., Evrard B., Hubert P., Anal. Chim. Acta, 642, 186–192 (2009).
• 39) Nieuwmeyer F. J., Damen M., Gerich A., Rusmini F., van der Voort Maarschalk K., Vromans H., Pharm. Res., 24, 1854–1861 (2007).
• 40) Ohgi K., Hayashi Y., Tsuji T., Ito T., Leong K. H., Usuda S., Kumada S., Okada K., Onuki Y., Int. J. Pharm., 604, 120770 (2021).