Optimization of Processing of Tiebangchui with Highland Barley Wine Based on the Box-Behnken Design Combined with the Entropy Method

Summary

The present protocol describes an efficient method for optimization of the processing technology of Tiebangchui processed with highland barley wine based on a Box-Behnken design response surface combined with the entropy method.

Abstract

The processing of toxic ethnomedicines is of great significance for their safe clinical application. Thus, the limitations of traditional processing should be addressed, and the processing method of ethnomedicines should be standardized using modern research methods. In this study, the processing technology of a commonly used Tibetan medicine Tiebangchui (TBC), the dried root of Aconitum pendulum Busch, processed with highland barley wine was optimized. Diester-diterpenoid alkaloid (DDA) (aconitine, 3-deoxyaconitine, 3-acetylaconitine) and monoester-diterpenoid alkaloid (MDA) (benzoylaconine) content were used as evaluation indicators, and the weight coefficient of each evaluation index was determined by the entropy method.

The single factor test and Box-Behnken design were used in investigating the influence of the ratio between highland barley wine and TBC, slice thickness of TBC, and processing time. Comprehensive scoring was performed according to the objective weight of each index determined by the entropy method. The optimal processing conditions of TBC with highland barley wine were as follows: the amount of highland barley wine is five times that of TBC, a soaking time of 24 h, and a TBC thickness of 1.5 cm. The results showed that the relative standard deviation between the verification test and predicted value was less than 2.55% and the optimized processing technology of TBC processed with highland barley wine is simple, feasible, and stable, and so can provide a reference for industrial production.

Introduction

Tiebangchui (TBC), the dried root of Aconitum pendulum Busch, is a well-known Tibetan medicine and was initially recorded in the classic Tibetan medical book "Four Medical Tantra"^1,2. According to "Drug Standards of the Ministry of Health of the People's Republic of China (Tibetan Medicine)", TBC is effective in expelling cold, relieving pain, dispelling wind, and calming shock, and is commonly used to treat rheumatoid arthritis in clinics^3,4,5.

TBC mainly contains alkaloids, including highly toxic diester-diterpenoid alkaloids (DDAs), and the moderately toxic monoester-diterpenoid alkaloids (MDAs)^6,7,8. These chemical components are active ingredients with medicinal effects but are toxic. One of the most famous active and toxic ingredients, aconitine, causes poisoning when it exceeds 1 mg⁹. Therefore, improper or excessive use of TBC might result in poisoning and even death, and the toxicity attenuation and efficacy reservation of TBC is crucial for its safe clinical application^10,11.

Processing is an effective method for detoxifying TBC. According to ancient Tibetan medicine books, processing with highland barley wine is an efficient way to attenuate toxicity and preserve the efficacy of TBC. TBC is soaked in highland barley wine, stored for one night, dried, and added to medicines¹². However, the specific processing technology and potential influencing factors are rarely reported, and the traditional processing process often relies on experience and lacks standardized methods. Hence, modern scientific and technological methods for optimizing and standardizing the processing process are needed.

The Box-Behnken design method is used in investigating interactions among different factors and their influence on comprehensive scoring through quadratic polynomial fitting. This design allows the intuitive observation of optimal conditions and has been widely used in the field of pharmacy¹³. For example, the Box-Behnken design method, based on the entropy method, successfully optimized the processing technology of stir-frying with vinegar of Curcuma Longa Radix¹⁴. In this study, the Box-Behnken response surface experimental design combined with the entropy method was used in optimizing the processing technology of TBC processed with highland barley wine. The optimized processing technology is expected to ensure quality control and safe clinical use.

Protocol

In this study, the processing technology of TBC processed with highland barley wine was optimized with a Box-Behnken design combined with the entropy method. DDA and MDA content were used as evaluation indicators, and the weight coefficient of each evaluation index was determined by the entropy method.

1. Experimental preparation

1. Prepare highland barley wine¹⁵.
   1. Take 500.00 g of black highland barley rice and add five times the amount of water. Cook the rice until the remaining water is absorbed (~2 h). Pour it out, wait until the temperature falls to 37 °C, add 4 g of Jiuqu (see Table of Materials), mix well, seal the can, wrap the container with cotton wool, and let it stew for 7 days.
   2. Add 300 mL of water on the 7th day and seal again. On the 8th day, begin removing the wine and replace with 300 mL of water afterward. Seal and ferment for 1 day, take the wine, and add 300 mL of water again. Repeat this procedure three times and combine the liquors.
   3. Bring to a boil, then reduce the heat to a simmer, and continue cooking until the remaining water is absorbed.
2. To prepare processed products, accurately weigh the TBC in a container, add highland barley wine, and soak for 1 day. Then, dry in a constant temperature electric drying oven.
   NOTE: The drying temperature should be less than 40 °C to avoid changes in the alkaloid composition.
3. Prepare test sample solution.
   1. Accurately weigh TBC processed product powder (2 g) in a conical flask, add 40% ammonia solution, and perform ultrasound-assisted extraction with isopropanol-ethyl acetate (1:1) mixed solvents (50 mL) (power: 200 W; frequency: 40 kHz; temperature: 40 °C) for 30 min.
      NOTE: To prepare 40% ammonia solution, transfer 40 mL of ammonia to a 100 mL volumetric flask and then dilute with pure water.
   2. Adjust the extracted solution to the original weight by adding an isopropanol-ethyl acetate mixture (1:1 v/v).
   3. Accurately transfer the extracted solution (25 mL) to a round-bottom flask for the recovery of the solvent under reduced pressure until dry.
   4. Finally, transfer 0.05% hydrochloric acid-methanol solution to dissolve the residue from step 1.3.3 in a 5 mL volumetric flask and dilute with 0.05% methanol hydrochloride solution. Filter the solution through a 0.22 µm microporous membrane filter prior to injection into the high-performance liquid chromatography (HPLC) systems.
      NOTE: Prepare 0.05% methanol hydrochloride acid by adding 0.05 mL of hydrochloric acid to a 100 mL volumetric flask, then dilute with methanol.
4. Prepare a standard solution by weighing 5.18 mg of benzoylaconine, 13.13 mg of aconitine, 10.05 mg of 3-deoxyaconitine, and 10.09 mg of 3-acetylaconitine accurately, and then place the solids in a 5 mL volumetric flask individually. Dilute with 0.05% methanol hydrochloride solution.

2. Chromatographic condition

1. Set up the chromatographic conditions as shown in Table 1 for HPLC. Details of the instruments used are provided in the Table of Materials.

3. System adaptability test

1. Range of linearity
   NOTE: First, we used HPLC to determine the peak areas of benzoylaconitine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine in the sample, and then randomly determined the peak area of one known concentration of standard solution. Next, we compared the difference between two peak areas (sample solution and standard solution) to estimate the concentration of benzoylaconitine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine in different samples, and then adjusted the standard solution into a linear range to include the concentration of the sample in the curve. The standard curve concentrations are shown in Table 2.
   1. Prepare benzoylaconitine reference solutions containing 1.036 mg/mL, 0.518 mg/mL, 0.2072 mg/mL, 0.1036 mg/mL, and 0.0518 mg/mL.
   2. Prepare aconitine reference solutions containing 1.313 mg/mL, 0.5252 mg/mL, 0.2626 mg/mL, 0.1313 mg/mL, and 0.05252 mg/mL.
   3. Prepare 3-deoxyaconitine reference solutions containing 1.005 mg/mL, 0.5025 mg/mL, 0.201 mg/mL, 0.1005 mg/mL, and 0.402 mg/mL.
   4. Prepare 3-acetylaconitine reference solutions containing 0.2018 mg/mL, 0.1009 mg/mL, 0.04036 mg/mL, 0.02018 mg/mL, and 0.01009 mg/mL.
   5. Investigate the linearity of each compound by plotting the peak area versus injection concentration.
2. To perform the precision test, inject 10 µL of each reference solution into the HPLC system six times daily and employ the same HPLC conditions described in step 2.1 to run the samples Record the peak area of each component.
3. Perform intraday stability testing by injecting 10 µL of the prepared sample solution via step 1.3 and determine the peak areas after 0 h, 2 h, 4 h, 8 h, 14 h, 12 h, and 24 h¹⁶.
4. Perform a reproducibility test by taking six samples of the same batch of TBC to prepare the test sample solution, according to step 1.3. Inject 10 µL of each sample into the HPLC system and run the samples as described in step 2.1.
5. Perform the recovery test to evaluate the accuracy of the method. Add 100% of the standard solution of each index component (benzoylaconitine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine) in the test solution to calculate the recovery rate, respectively. For example, as the content of benzoylaconitine is 0.1524 mg/mL in the TBC sample, accurately weigh 0.1524 mg of benzoylaconitine standards and add to the TBC sample, then prepare the test sample solution according to step 1.3. Run these samples with the same HPLC conditions described in step 2.1. Calculate the recovery rate using Equation (1):
       (1)
   Here, A is the amount of component (benzoylaconitine, aconitine, 3-deoxyaconitine, or 3-acetylaconitine) to be measured in the sample solution, B is the amount of standard added (benzoylaconitine, aconitine, 3-deoxyaconitine, or 3-acetylaconitine), and C is the measured value of the solution containing the standard solution and the sample solution (see Table 3). Refer to step 2.1 for the chromatographic conditions to perform the above steps. The recovery rate reflects the degree of loss of the target component (benzoylaconitine, aconitine, 3-deoxyaconitine, or 3-acetylaconitine) during the sample analysis; the higher the recovery rate, the lower the loss of the target component.

4. Single factor test of TBC processed with highland barley wine

NOTE: The ratio between highland barley wine and TBC, slice thickness of TBC, and soaking time will affect the dissolution of more toxic components (aconitine, 3-deoxyaconitine, and 3-acetylaconitine) in TBC during the TBC processed with highland barley wine¹⁷. The single factor test and Box-Behnken design were used to investigate the influence of the ratio of highland barley wine to TBC, slice thickness of TBC, and soaking time.

1. Perform the highland barley wine addition test (A) by setting up five groups of tests, each with 30 g of TBC, where the amount of highland barley wine is two, three, four, five, and six times the amount of TBC in the recipe. The soaking time is 12 h, and the slices are 1.0 cm thick¹⁸.
   NOTE: Each group of the same condition test should be processed in three parallel groups.
2. Perform the soaking time test (B) by setting up five groups of tests, each with 30 g of TBC. The soaking times are 12 h, 24 h, 36 h, and 48 h. The amount of highland barley wine is five times that of TBC, and the slices are 1.0 cm thick¹⁹.
   NOTE: Each group of the same condition experiment should be processed in three parallel groups.
3. Perform the slicing thickness test (C) by setting up five groups of tests, each with 30 g of TBC. The slices are 0.5, 1.0, 1.5, 2.0, and 2.5 cm thick, the soaking time is 24 h, and the amount of highland barley wine is five times that of TBC²⁰.
   NOTE: Each group of the same condition experiment should be processed in three parallel groups.
4. Accurately weigh processed products for each test group to prepare test sample solution according to step 1.3. Determine the peak area of each sample by HPLC and use the standard curve to estimate the amounts of MDAs and DDAs. In the standard curve, y is the peak area and x is the content. The content of MDAs is benzoyl aconitine, and the content of DDAs is the sum of aconitine, 3-deoxyaconitine, and 3-acetylaconitine.
5. Use the total content of DDAs and the content of MDAs as evaluation indicators, and determine the weight coefficient of each evaluation index and the comprehensive scoring via the entropy method (section 5).
   CAUTION: TBC is toxic, and thus protective measures should be taken during processing.

5. Entropy method to calculate the comprehensive scoring

NOTE: We use the experimental data of the slicing thickness test in the single factor test as an example to illustrate the calculation process in detail. We use the peak area of the components in each sample in Supplementary Table S1 and the standard curve in Table 2 to calculate the content of MDAs and DDAs (see Supplementary Table S2). In the linear equation, y is the peak area and x is the content. In this study, the moderately toxic MDA (benzoylaconitine) was used as the positive indicator, and the total content of DDAs (aconitine, 3-deoxyaconitine, and 3-acetylaconitine) with high toxicity was used as the negative indicator. The content of MDAs is benzoyl aconitine, and the content of DDAs is the sum of aconitine, 3-deoxyaconitine, and 3-acetylaconitine. Each sample has two evaluation indicators: i = 1,2,…,n, and j = 1,2,…m²¹.

1. Use Equation (2) to standardize the content of MDAs²².
        (2)
   Thus,
   NOTE: Xij is the value of the j-th indicator of the i-th sample. Xij* is the standardized value of Xij. For example, i = 3 and j = 1, X₃₁ represents the value of the first indicator of the third sample, and is the standardized value of the first indicator in the third sample. are shown in Supplementary Table S3.
2. Use Equation (3) to standardize the total content of the DDAs²³.
        (3)
   
   NOTE: Here, i = 3, j = 2, represents the second indicator of the third sample. is the standardized value of the second indicator in the third sample. are shown in Supplementary Table S3.
3. Use Equations (4) and (5) to define the entropy value (Hj) of each indicator²³.
   1. Calculate the probability of the j-th trial under the i-th evaluation indicator P_ij using equation (4).
           (4)
      For number 3,
      
      
      ​NOTE: The probability values for the first indicator and second indicator of the third sample are 0.2374 and 0.2812, respectively. are shown in Supplementary Table S3.
   2. Calculate the information entropy H_j.
           (5)
      
      
      NOTE: H₁ is the entropy of the first indicator (MDAs) and H₂ is the entropy of the second indicator (DDAs) in the slicing thickness test.
4. Use Equation (6) to calculate the indicator weights (Wj)²³.
   (6)
      = 33.3%
      = 66.7%
   NOTE: W_j is the weight coefficient of each indicator. In the slicing thickness test, the weight coefficient of the positive indicator (MDAs) and negative indicator (DDAs) are 33.3% and 66.7%, respectively.
5. Use Equation (7) to calculate the comprehensive scoring of the indicators²³.
   (7)
   For number 3,
   
   NOTE: S_i is the comprehensive scoring of each sample. We need to obtain the highest score as the central point in the Box-Behnken design. S₁, S₂, S₃, S₄, and S₅ are show in Supplementary Table S3.

6. Box-Behnken design

1. Through the single factor test, use the condition with the highest comprehensive scoring (see Table 4, Table 5, Table 6, and Figure 2) as the center point of the response surface. Use the amount of highland barley wine (A), soaking time (B), and slice thickness of TBC (C) as the influencing factors and the comprehensive scoring as the response value²⁴.
   NOTE: Based on the single factor data in Table 4, Table 5, and Table 6, the highest comprehensive scoring is calculated by equations (2), (3), (4), (5), (6), and (7) in section 5, and the best point is obtained. The amount of highland barley wine was five times that of TBC, the soaking time was 36 h, and the slicing thickness was 1.0 cm.

7. Box-Behnken design software operation steps

1. Open the software (see Table of Materials) and select New Design | Box-Behnken Design (see step 5.1; Supplementary File 1).
   1. Input the number of influencing factors and input the level information (three-level-three-factor; see Table 7). The Box-Behnken design is composed of 17 experiments in this study. Finally, click Continue (see step 5.2; Supplementary File 1).
   2. Set the comprehensive scoring (Y) by equations (2), (3), (4), (5), (6), and (7) in section 5 as the response. Input the number of response values (image shows only one response value) and click Finish (see step 5.3; Supplementary File 1).
   3. Process the TBC with highland barley wine according to the design results and complete the experiment based on the 17 scenarios designed for the response surface.
   4. Prepare the sample solutions by following step 1.3 and calculate the total content of the MDAs and DDAs by the HPLC system.
   5. Calculate the comprehensive scoring for each group by equations (2), (3), (4), (5), (6), and (7) in step 5, and input the score results (see step 5.4; Supplementary File 1).
2. Click analyze to analyze the date and model information (see step 5.4.1; Supplementary File 1).
   1. Perform statistical validation of polynomial equations and response surface analysis plotted in 3D model plots obtained by the software.
   2. Click on ANOVA in the top menu and observe the results table.
3. Click Optimization to view the predicted optimal process conditions (see step 5.4.2; Supplementary File 1).

8. Validation test

1. According to the results predicted from the Box-Behnken response surface design, in step 7.3, identify the optimal processing condition of TBC. Here, it is as follows: TBC is soaked for 24 h in five times the amount of highland barley wine, and the thickness of the TBC is 1.5 cm. Take the optimal level of influencing factors as processing conditions and set up three parallel sets of experiments to verify the stability of the processing technology.

Representative Results

In this study, the precision, stability, repeatability, and sample recovery of TBC indicated that the method is feasible. The four index components in TBC had a good linear relationship within a specific concentration range. Typical chromatograms are shown in Figure 1. The precision test results (Table 8) showed that the relative standard deviation (RSD) of the peak areas were 2.56%, 1.49%, and 2.03% for benzoylaconine, aconitine, and 3-deoxyaconitine, respectively, and 0.21% for 3-acetylaconitine, indicating that the precision of the instrument was good. The study of stability performed for 24 h (n = 6) indicated relative standard deviation values of 2.76%, 2.21%, 2.98%, and 2.31% for benzoylaconine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine, respectively (Table 9), suggesting that the sample solution was stable for 24 h. The repeatability test results (Table 10) showed that the RSDs of the peak areas of benzoylaconine, aconitine, 3-deoxyaconitine, and 3-deoxyaconitine were 2.80%, 2.92%, 2.92%, and 2.07%, respectively, showing that the repeatability of this method was good. The recovery experiment results indicated that the average recovery rates of benzoylaconine, aconitine, 3-deoxyaconitine, and 3-deoxyaconitine were 99.7%, 100.84%, 103.27%, and 100.92%, respectively.

The single factor test of TBC processed with highland barley wine revealed that the amount of highland barley wine was five times that of TBC, the soaking time was 36 h, and the slicing thickness was 1.0 cm (Figure 2). The experimental design and results of the response surface model are shown in Table 11. The results of the experimental ANOVA are shown in Table 12. The factors are fitted by regression to obtain a quadratic multinomial regression equation (8). A: Highland barley wine addition; B: soaking time; C: slicing thickness. The results showed that the model was well fitted and was able to predict the relationship between the comprehensive scoring of highland barley wine addition, soaking time, and slice thickness. The order of the factors by the strength of the effects was highland barley wine addition > slice thickness of medicinal herbs > soaking time.

     (8)

According to Equation (8), Design-Expert 8.0.6 analysis software is used to plot a 3D curve via step 7.2.1 (Figure 3). A steeper slope of the response surface indicates a stronger horizontal interaction of factors, and a gentler slope is the opposite. The p-value (p < 0.0001) of the model in Table 12 shows that the model is significant, with an R² of 0.9754 and a non-significant misfit term (p = 0.7253), indicating that the model is a good fit and better reflects the relationship among the highland barley wine addition, soaking time, slice thickness of medicinal herbs, and overall score.

According to the objective weight of each index determined by the analytic entropy method, comprehensive scoring was performed, and the optimal processing condition of TBC was determined as follows: TBC is soaked for 24 h in five times the amount of highland barley wine, and the thickness of the TBC is 1.5 cm. The validation test results showed that the total DDAs were 0.6963, 0.6793, and 0.7023 mg/g, respectively, and the MDA content in three sets of parallel tests was 0.2096, 0.2237, and 0.2109 mg/g. The average comprehensive scoring was 83. The RSD between the verification test and the predicted value was less than 1.8%, indicating that the optimized processing technology of TBC processed with highland barley wine is simple, feasible, and stable, providing a reference for industrial production.

Figure 1: Representative chromatograms of the four characteristic components after setting the chromatographic conditions mentioned in step 2.1 (n = 1). (A) Typical chromatograms of the reference solution. Peak 1 is benzoylaconine, peak 2 is aconitine, peak 3 is 3-deoxyaconitine, and peak 4 is 3-acetylaconitine. (B) Typical chromatograms of the sample solution. Peak 1 is benzoylaconine, peak 2 is aconitine, peak 3 is 3-deoxyaconitine, and peak 4 is 3-acetylaconitine. Please click here to view a larger version of this figure.

Figure 2: The comprehensive scoring of the single factor test of TBC processed with highland barley wine (n = 3). (A) Amount of highland barley wine (times); (B) Soaking time (h); (C) Slice thickness (cm). Please click here to view a larger version of this figure.

Figure 3: 3D response surface map of the effect of interaction of various factors on comprehensive scoring. Please click here to view a larger version of this figure.

Condition	Parameter
Chromatographic column	Ultimate ODS-3 C18 (4.6 mm x 250 mm, 5 μm)
Mobile phase	Acetonitrile (A) - 0.04 mol/L ammonium acetate solution (B) pH= 8.5 ± 0.5
Gradient elution	0–10 min, 0%-70% A; 10–15 min, 70–50% A; 15–30 min, 50–40% A; 30–38min, 40–15% A; 38–45 min, 15–15% A; 45–55 min, 15–70% A
Flow rate	1 mL/min
Column temperature	30 °C
Detecting wavelength	235 nm
Sample volume	10 μL

Table 1: The chromatographic conditions set in this experiment. Details about the chromatographic column, the mobile phase, the gradient elution, the flow rate, the column temperature, the detection wavelength, and the sample volume.

Index components	Linear equation	Range of linearity (mg/mL)	R2
Benzoylaconine	y=11,658,706.1677x +19,717.0872	1.036-0.0518	0.9995
Aconitine	y=11,199,784.3030x -67,641.2429	1.313-0.05252	0.9999
3-Deoxyaconitine	y=11,214,550.3140x +59,795.9119	1.005-0.0402	0.9999
3-Acetylaconitine	y=9,887,511.9074x +26,713.6359	0.2018-0.01009	0.9994

Table 2: The linear relationship of the index components in TBC. The four index components in TBC had a good linear relationship in a specific concentration range.

Index components	Known content (mg)	Adding quantity (mg)	Measuring quantity (mg)	Recoveries (%)	Average recoveries (%)	RSD (%)
Benzoylaconine	0.1558	0.1295	0.2901	96.4	99.7	3.14
	0.1574	0.1295	0.2849	98.46
	0.156	0.1295	0.2871	101.24
	0.1574	0.1295	0.2923	104.95
	0.1449	0.1295	0.2736	99.38
	0.1566	0.1295	0.2839	98.3
Aconitine	0.3099	0.3283	0.645	102.07	100.84	2.02
	0.3153	0.3283	0.6371	98.02
	0.2928	0.3283	0.6314	103.14
	0.2969	0.3283	0.6325	102.23
	0.3035	0.3283	0.6343	100.76
	0.3094	0.3283	0.6339	98.84
3-Deoxyaconitine	0.1789	0.201	0.3788	99.45	103.27	2.65
	0.1793	0.201	0.3845	102.09
	0.1741	0.201	0.3774	101.14
	0.1635	0.201	0.3753	105.37
	0.1708	0.201	0.383	105.57
	0.1653	0.201	0.3783	105.97
3-Acetylaconitine	0.0169	0.02	0.0374	102.5	100.92	1.15
	0.0168	0.02	0.037	101
	0.0166	0.02	0.0366	100
	0.0161	0.02	0.0365	102
	0.017	0.02	0.0369	99.5
	0.0171	0.02	0.0372	100.5

Table 3: The results of the sample recovery rate measurement. The RSD of the recovery rate of benzoylaconine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine were 3.14%, 2.02%, 2.65%, and 1.15%, respectively.

Number	Highland barley wine addition test (times)	Content of MDAs (mg/g)	Content of DDAs (mg/g)	Comprehensive scoring/points
1	2	0.1875	0.8254	58.98421777
2	3	0.1099	0.9847	0.056898711
3	4	0.2296	0.8487	71.12048666
4	5	0.2161	0.6894	94.6966946
5	6	0.2006	0.7472	78.22537224

Table 4: The results of the single factor test of the ratio between highland barley wine and TBC.

Number	Soaking time test (h)	Content of MDAs (mg/g)	Content of DDAs (mg/g)	Comprehensive scoring/points
1	6	0.236292609	1.047811476	59.67501032
2	12	0.193880685	1.164420534	23.10718817
3	24	0.229606225	0.848736346	53.86313899
4	36	0.151447388	0.701045217	79.15664943
5	48	0.193311963	0.767427412	68.88872066

Table 5: The results of the single factor test of the soaking time.

Number	slicing thickness test (cm)	Content of MDAs (mg/g)	Content of DDAs (mg/g)	Comprehensive scoring/points
1	0.5	0.1043	0.6190	66.96
2	1	0.1709	0.6992	75.05
3	1.5	0.1507	0.6954	66.23
4	2	0.1459	0.8347	20.66
5	2.5	0.1451	0.8298	21.79

Table 6: The results of the single factor test of the slice thickness of TBC.

Level		Factor
	A (amount of highland barley wine, times)	B (soaking time, h)	C (slice thickness, cm)
1.0000	4.0000	24.0000	0.5000
2.0000	5.0000	36.0000	1.0000
3.0000	6.0000	48.0000	1.5000

Table 7: Box-Behnken design response surface level factor table.

Peak area in index components	1	2	3	4	5	6	RSD (%)
Benzoylaconine	1281252	1290912	1198912	1256056	1256704	1266738	2.56%
Aconitine	2861208	2881686	2785022	2790990	2859024	2799395	1.50%
3-Deoxyaconitine	2356317	2328383	2429059	2350987	2406114	2450374	2.04%
3-Acetylaconitine	2008110	2021560	2014519	2015881	2015209	2012529	0.22%

Table 8: The results of the precision measurement. The RSD of the peak areas of benzoylaconine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine were 2.56%, 1.49%, 2.03%, and 0.22%, respectively (n = 6).

Peak area in index components	0	2	4	8	12	24	RSD (%)
Benzoylaconine	191657	189590	193934	205135	196159	195954	2.76
Aconitine	312259	310240	294331	309104	312199	305360	2.22
3-Deoxyaconitine	230174	246787	239760	249302	248806	243396	2.98
3-Acetylaconitine	17086	16953	16826	16914	16979	17896	2.31

Table 9: The results of the stability test. The RSD of the peak areas of benzoylaconine, aconitine, 3-deoxyaconitine and 3-acetylaconitine were 2.76%, 2.21%, 2.98%, and 2.31%, respectively (n = 6).

Peak area of index components	1	2	3	4	5	6	RSD (%)
Benzoylaconine	191067	192795	191058	192907	179103	192008	2.79
Aconitine	308142	313754	290487	294740	301515	307654	2.92
3-Deoxyaconitine	249021	249456	243963	232781	240524	234661	2.92
3-Acetylaconitine	17465	17451	17247	16691	17608	17686	2.07

Table 10: The results of the reproducibility test. The RSD of the peak areas of benzoylaconine, aconitine, 3-deoxyaconitine, and 3-acetylaconitine were 2.79%, 2.92%, 2.92%, and 2.07%, respectively (n = 6).

Number	A (Highland barley wine addition, times)	B (Soaking time, h)	C (Slicing thickness, cm)	Content of MDAs (mg/g)	Content of DDAs (mg/g)	Comprehensive scoring/points
1	4	36	0.5	0.1032	0.6882	28.2
2	5	48	1.5	0.1688	0.6588	56.49
3	6	24	1	0.1236	0.6535	33.02
4	5	24	1.5	0.2201	0.692	87.23
5	5	36	1	0.2094	0.6199	70.71
6	5	24	0.5	0.1809	0.5689	48.56
7	4	24	1	0.2016	0.7744	90.74
8	5	36	1	0.2169	0.6889	85.15
9	5	36	1	0.2103	0.6802	80.5
10	6	36	0.5	0.1036	0.5072	0.36
11	6	36	1.5	0.1089	0.5062	2.86
12	4	48	1	0.1789	0.6789	64.6
13	6	48	1	0.1036	0.5536	7.55
14	5	36	1	0.2062	0.6084	67.33
15	4	36	1.5	0.1832	0.6954	69.31
16	5	48	0.5	0.1759	0.5569	44.21
17	5	36	1	0.2161	0.6894	84.82

Table 11: Design and results of the response surface design test.

Source	Sum of Squares	df	Mean Square	F -Value	P-Value
Model	14403.27	9	1600.36	30.8	<0.0001
A	5463.26	1	5463.26	105.15	<0.0001
B	939.61	1	939.61	18.08	0.0038
C	1117.7	1	1117.7	21.51	0.0024
AB	0.11	1	0.11	0.00216	0.9642
AC	372.68	1	372.68	7.17	0.0316
BC	174.11	1	174.11	3.35	0.1099
A2	4133.52	1	4133.52	79.55	<0.0001
B2	28.63	1	28.63	0.55	0.482
C2	1890.1	1	1890.1	36.38	0.0005
Residual	363.71	7	51.96
Lack of Fit	93.28	3	31.09	0.46	0.7253
Pure Error	270.43	4	67.61
Cor Total	14766.99	16

Table 12: ANOVA for the regression model.

Supplementary File 1: A detailed guide of the Box-Behnken design software. Please click here to download this File.

Supplementary Table S1: Sample peak area of the slicing thickness test by HPLC. Please click here to download this File.

Supplementary Table S2: Content of MDAs (benzoylaconine) and DDAs (aconitine, 3-deoxyaconitine, and 3-acetylaconitine) in the slicing thickness test. Please click here to download this File.

Supplementary Table S3: Comprehensive scoring of the slicing thickness test. Please click here to download this File.

Discussion

As a commonly used Tibetan medicine with toxic effects, the toxicity-attenuating effect of processing is extremely important for TBC's clinical application²⁵. In this study, the processing technology of TBC processed with highland barley wine was optimized. By reviewing the main active ingredients and relating the pharmacological effects of TBC, we found that TBC alkaloids have anti-inflammatory and analgesic effects and can be used to treat rheumatoid arthritis. In this study, a moderate toxic MDA was used as the positive indicator. The total content of DDAs was used as the negative indicator. The entropy method was used in calculating the index weights and optimizing the processing technology²⁶.

During the experiment, two points should be particularly noted. First, three parallel sets of experiments are required for each condition of a single factor to improve the accuracy of subsequent results. Second, in the calculation of the comprehensive scores, the standardized values of each indicator are multiplied by the weighting factor instead of the raw data. This procedure ensures the accuracy of calculation results. In addition, the R² value in ANOVA should be as close as possible to one; otherwise, the levels of factors are extremely close, having little impact on results; the difference between the levels should be moderately wide.

Although the use of multi-index comprehensive scoring combined with the response surface method ensures the precise prediction of the processing process of TBC, it does have limitations. First, when laboratories use HPLC to measure indicated components in medicinal materials, human error may occur due to the small scale of experiments. More convincing results can be obtained if pilot evaluation is carried out in a large-size herb processing factory. Second, the Box-Behnken design method is not suitable for optimization of the whole process. After importing the experimental data into the response surface software, the model term needs to be significant (p < 0.05), and the lack of fit needs to be nonsignificant (p > 0.05) in the ANOVA data. If the result is reversed, the process is unsuitable for optimization by this method.

In conclusion, compared with the commonly used single factor test, uniform design, orthogonal design, and star point design, the Box-Behnken response surface approach is an experimental optimization design method that uses multivariate linear and quadratic term model fitting²⁷. The model predicted by the Behnken response surface method has continuity and high experimental accuracy, and predicts optimal points^28,29. In the present experiment, the preferred process of TBC was determined by comprehensive scoring based on the single factor test, and the final validation test did not deviate much from the predicted value, indicating that the selected model is reasonable and can provide references for the processing technology of TBC processed with highland barley wine. The optimized processing technology can provide information and guidance for the study of the toxicity-attenuating effect of processing other toxic ethnic medicines.

Materials

Name	Company	Catalog Number	Comments
Aconitine	Chengdu Push Biotechnology Co.,Ltd	PS000905
3-Acetylaconitine	Chengdu Push Biotechnology Co.,Ltd	PS010552
3-Deoxyaconitine	Chengdu Push Biotechnology Co.,Ltd	PS011258
Benzoylaconine	Chengdu Push Biotechnology Co.,Ltd	PS010300
Circulating water vacuum pump	Gongyi City Yuhua Instrument Co., Ltd	SHZ-DIII
Design-Expert	State-East Corporation	8.0.6
Electric constant temperature drying oven	Shanghai Yuejin Medical Equipment Co., Ltd	101-3-BS
Electronic analytical balance	Shanghai Liangping Instruments Co., Ltd.	FA1004
High performance liquid chromatography	Shimadzu Enterprise Management (China) Co., Ltd	shimadzu 2030
Highland barley rice	Kangding City, Ganzi Tibetan Autonomous Prefecture, Sichuan Province	20221015
Millipore filter	Tianjin Jinteng Experimental Equipment Co., Ltd	φ13 0.22 Nylon66
Rotary evaporator	Shanghai Yarong Biochemical Instrument Factory	RE-2000A
Starter of liquor-making	Angel Yeast CO., Ltd	BJ22-104
Ultra pure water systemic	Merck Millipore Ltd.	Milli-Q
Ultrasonic cleansing machine	Ningbo Xinyi Ultrasonic Equipment Co., Ltd	SB-8200 DTS