Anti-tumor composition and preparation method thereof

Abstract

The invention provides an anti-tumor composition, which is characterized in that the anti-tumor composition is prepared from the following raw materials: ganodronic acid DM, ganodronic B, 6-O-methylethylcerevistol, 3, 5-dihydroxyergost-7, 22-dien-6-one and ganodronic acid S1, and the preparation method of the anti-tumor composition comprises the following steps of: preparing the ganodronic acid DM, the ganodronic B, the 6-O-methylethylcerevistol, the 3, 5-dihydroxyergost-7, the 22-dien-6-one and the ganodronic acid S1. According to the anti-tumor composition disclosed by the invention, 15 triterpenes and sterol compounds from ganoderma lucidum are optimized by adopting a pharmaceutical combinatorial optimization framework (IFER) combining feedback system control (FCS) and elastic network algorithm (EN), so that the ganoderma lucidum anti-tumor composition is prepared. The anti-tumor composition provided by the invention has specific activity of inhibiting tumor cell proliferation, can inhibit synthesis of cholesterol, has lower toxicity to normal cells compared with any single compound in the composition, and has a wide application prospect.

Description

Technical Field

[0001] This invention relates to the field of pharmaceutical preparations, and more specifically to an antitumor composition and its preparation method. Background Technology

[0002] A tumor is a new growth formed when cells in the body experience genetic problems due to various carcinogenic factors (radiation, viruses, and chemical carcinogens, etc.), resulting in the loss of normal regulation of their growth and the eventual clonal abnormal proliferation. The prevention and treatment of tumors has always been a key focus of medical research and a strategic issue related to national health.

[0003] In recent years, the development of anti-tumor drugs has increasingly emphasized targeting, meaning that while killing tumor cells, drugs should minimize toxic side effects and avoid killing normal cells. Traditional Chinese medicine (TCM), a culmination of thousands of years of Chinese wisdom, is an important source for finding superior active drugs. Discovering active compounds from traditional Chinese medicine and modifying their structures to enhance efficacy and reduce toxicity has long been a primary approach in drug development. In recent years, some scholars have attempted to develop modern TCM based on the principles of TCM compatibility, using scientific methods and following traditional compatibility theories and principles. This approach, based on component-activity relationships, optimizes the design and screens for highly effective component TCMs, offering a better strategy than single-drug treatment for complex diseases.

[0004] However, current component drug screening methods suffer from problems such as high workload and high experimental costs, necessitating the search for more efficient screening methods. Summary of the Invention

[0005] The primary objective of this invention is to provide an antitumor composition comprising the following five compounds:

[0006] ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one and ganoderic acid S1;

[0007] All five compounds are triterpenoids or sterols derived from Ganoderma lucidum, and their molar concentration (μM) ratio among the five compounds in the antitumor composition is (70-72):(5-6):(5-6):(18-20):(4-5);

[0008] The preferred molar concentration (μM) ratio among the five compounds in the antitumor composition is 71.12:5.68:5.26:18.69:4.61.

[0009] The above-mentioned antitumor composition can be used to prepare drugs for the prevention or treatment of tumors. It can be compounded with pharmaceutically acceptable carriers to prepare any pharmaceutically acceptable drug dosage form, such as injection, tablet, powder, granule, pill, capsule, oral liquid, ointment, for the prevention or treatment of tumors, including liver cancer, lung cancer, breast cancer, myeloma, acute / chronic leukemia, brain cancer, etc.

[0010] The present invention also provides a method for preparing the above-mentioned antitumor composition, which is obtained by optimizing 15 cytotoxic triterpenoids and sterol compounds from Ganoderma lucidum using an Iterative Framework via Elasticnet Regression (IFER) that combines Feedback Control System (FCS) and ElasticNet (EN) algorithm.

[0011] Specifically, the preparation method of the above-mentioned antitumor composition provided by the present invention includes the following steps:

[0012] (1) Selecting the experimental concentration of the candidate compounds;

[0013] Based on preliminary measurements of the cell proliferation inhibition rates of 15 triterpenoids and sterols derived from Ganoderma lucidum, different concentrations of the compounds were obtained through dilution and experimental verification to identify the IC50 value of each compound for inhibiting the proliferation of K562 leukemia cells. 30 The effective concentration is below the value;

[0014] (2) Optimization framework for combinatorial compounds IFER

[0015] The drug combination optimization framework IFER, which combines a feedback system-controlled FCS and an elastic network algorithm EN, optimizes the selection of 15 compounds.

[0016] The Elastic Network Algorithm (EN) combines lasso regression and ridge regression. The lasso regression model uses an L1 regularization term as a penalty term, while the ridge regression model uses an L2 regularization term as a penalty term. The Elastic Network Algorithm (EN) combines the two and uses the parameter α to adjust the proportion of each, which can maintain a balance between variable preservation and model accuracy.

[0017] Drugs were screened iteratively, and the weight of the L1 regularization term in the lasso regression was increased, with the α parameter set to 0.75. When performing model regression, the parameter λ was also introduced to determine the number of terms to retain in the model, and the value of λ was determined by cross-validation.

[0018] The drug combination optimization framework IFER sets two optimization indicators: the inhibition rate of proliferation of tumor cells K562 and normal cells MSC; and designs an optimization objective function formula:

[0019] T = I K562 -I MSC -β|I MSC | Formula (1)

[0020] In formula (1), I K562 I MSC The values ​​represent the inhibition rates of the compound on the proliferation of K562 cells and MSC cells, respectively, β|I MSC | indicates a penalty term, with the β value set to 0.5;

[0021] Three rounds of combination optimization experiments were conducted on 15 compounds based on the drug combination optimization framework IFER:

[0022] First round of combination optimization:

[0023] The goal of the first round of optimization was compound screening. Based on the balance of a large number of compounds and the aim of achieving accurate screening, 48 combinations were selected for experiments according to the principle of uniform experimental design, and the average number of compounds in each combination was controlled to be 7.5. According to the experimental data of the K562 and MSC cell lines, among the 48 combinations, the difference in the maximum inhibition rate between the best combination of K562 and MSC was only 24.6%, and a second round of combination optimization is still needed.

[0024] The inhibition rates of K562 and MSC were modeled and analyzed using the drug combination optimization framework IFER. The λ value was determined through cross-validation, and λ = 10 was selected for both cell lines. -2 The model was used; the screening target of the first round of combinatorial optimization experiment was set as 8 compounds. The K562 and MSC inhibition rates of all possible combinations containing ≤8 compounds were predicted, and the objective function value was calculated according to formula (1); all predicted combinations were sorted by objective function value, and the occurrence frequency of each drug in the top 100 combinations was counted. The top eight compounds were selected as ganoderic acid T, ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, ganoderic acid S1, ganoderic acid TN and ganoderic acid S;

[0025] The eight compounds selected above were then subjected to a second round of combination optimization:

[0026] Based on the principle of uniform experimental design, 32 combined compounds were selected for experimental verification in the second round of optimization, and the average number of compounds in the combination was reduced to 6.

[0027] The experimental results show that the difference in inhibition rates between K562 and MSC in the 32 combinations in the second round has increased to a maximum of 45.7%. Modeling analysis was performed on the results of the second round of experiments; combinations containing 3, 4, 5 and 6 compounds were predicted according to the objective function of formula (1). It was found that the modeling results still obtained a high R value (>0.9), indicating that the model has good predictive ability; in order to facilitate drug development, the number of drugs in the combination is best to be 3 to 6. Therefore, the optimal three-drug, four-drug, five-drug and six-drug combinations were predicted according to the objective function of formula (1), with 4 of each type, for a total of 16 combinations for the next round of experiments.

[0028] Third round of combinatorial optimization and verification:

[0029] The main purpose of this round of combination optimization is to verify the activity prediction model established in the second round of experiments, and at the same time find the optimal combination compound; according to the objective function formula (1), the objective function values ​​of all possible 3-6 combination compounds are calculated, and the objective function values ​​of combination compounds 3-10 are the highest; pharmacodynamic experiments were conducted on 16 combinations, and the above antitumor composition was obtained based on the comparison between its predicted value and experimental results.

[0030] The antitumor composition provided by this invention can selectively enhance the inhibitory effect on K562 tumor cells, while significantly reducing the inhibitory effect on normal MSC cells. Experiments have confirmed that this composition exerts its antitumor effect by downregulating key proteins FDFT1 and CYP51A1 in the cellular cholesterol synthesis pathway, thereby inhibiting cellular cholesterol synthesis. The screening and preparation of the antitumor composition of this invention is based on an iterative optimization framework that combines iteration and model compression algorithms. By modeling the results of each round of experiments, the number of drugs in the combination is continuously reduced, thereby improving the predictive accuracy of the model. This optimization method based on engineering feedback control theory can maximize optimization efficiency while maintaining optimization accuracy.

[0031] Reishi mushroom is a traditional Chinese medicine. Modern research has found that various terpenoids and steroids in Reishi mushroom have significant cytotoxic effects. To develop these active compounds to exert antitumor effects while reducing their toxicity to normal cells, this invention screened 15 compounds with strong cytotoxicity from Reishi mushroom. Furthermore, by combining feedback system control with elastic network algorithms, an optimization framework for feedback system control was established. This resulted in the first optimization of a highly efficient and low-toxicity antitumor combination compound consisting of five compounds, providing a new approach for the development of Reishi mushroom antitumor drugs.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] This invention provides a Ganoderma lucidum antitumor composition, namely ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one and ganoderic acid S1, for pharmaceutical use in preparing drugs for the prevention or treatment of tumors, and has no toxic side effects on normal cells, thus having good market application value. Attached Figure Description

[0034] Figure 1 Candidate compound information

[0035] Figure 2 Compound Combination Optimization Flowchart

[0036] Figure 3 The first round of in vitro proliferation experiments showed inhibition results.

[0037] Figure 4 Results of the first round of K562 cell line modeling

[0038] Figure 5 Results of the first round of MSC cell line modeling

[0039] Figure 6 The results of the second round of in vitro proliferation experiments showed inhibition.

[0040] Figure 7 Results of the second round of K562 cell line modeling

[0041] Figure 8 Results of the second round of MSC cell line modeling

[0042] Figure 9 Comparison of predicted and measured efficacy values ​​of the third round of combined drug therapy

[0043] Figure 10 Effects of combined compounds on the K562 cell cycle

[0044] Figure 11 Western Blot Detection of Differential Protein Expression and Gray-Scale Analysis

[0045] Appendix Explanation

[0046] Table 1. Concentration determination of 15 candidate compounds

[0047] Table 2. First Round Experimental Combinations

[0048] Table 3. Second Round Experimental Combinations

[0049] Table 4. Third Round Experimental Combination Table

[0050] Table 5. Differential protein expression in the cholesterol synthesis pathway Detailed Implementation

[0051] Unless otherwise specified, the experimental methods used in the following examples are conventional methods.

[0052] The human chronic myeloid leukemia cells K-562 and mesenchymal stem cells MSCs used in the following examples were purchased from the Cell Bank of the Chinese Academy of Sciences.

[0053] All 15 compounds provided in this invention were prepared in-house [preparation methods are described in the literature: Guo Xiaoyu et al., Preparation and isolation of moderately polar triterpenes from Ganoderma lucidum fruiting bodies. Acta Mycologica Sinica, 2023, 42(12):2454-2469. Yue Yawen et al., An active ingredient of a sporozoite-free Ganoderma lucidum mycelium. Acta Mycologica Sinica, 2020, 39(1):128-136]. The purity of all compounds is ≥98%, and the structural formulas of the 15 compounds are shown in the [reference needed]. Figure 1 .

[0054] Unless otherwise specified, other preparations or materials are commercially available products.

[0055] Example 1: Screening method for antitumor compositions

[0056] (1) Selection of experimental concentration of candidate compounds

[0057] The effects of 15 compounds at different concentrations on the proliferation of K-562 and MSC cells were determined using the Alamar Blue assay, with the anticancer drug pentafluorouracil (5FU) used as a positive control.

[0058] like Figure 1 Fifteen triterpenoids and sterols with strong cytotoxicity were identified. Based on preliminary measurements of their cell proliferation inhibition rates, different concentrations of these compounds were diluted and experimentally verified to determine the IC50 of each compound on K562 leukemia cells. 30 The concentrations below the effective value (Table 1) are used as high doses (represented by "2" in Tables 2, 3, and 4), and the concentrations after a 4-fold dilution are used as low doses (represented by "1" in Tables 2, 3, and 4) for subsequent combination optimization.

[0059] Table 1. Concentration determination of 15 compounds

[0060]

[0061]

[0062] (2) Optimization framework for combinatorial compounds IFER

[0063] All experimental points are uniformly distributed in the combinatorial space, satisfying the Uniform Design of Experiments (DOE). A drug combinatorial optimization method based on polynomial fitting modeling is adopted. To address the problem that increasing the number of drugs leads to an increase in polynomial coefficients and thus an increase in the number of experiments, an Iterative Framework via Elasticnet Regression (IFER) is designed, combining a Feedback Control System (FCS) and an Elastic Net (EN) algorithm.

[0064] The Elastic Network algorithm combines Lasso Regression and Ridge Regression. Lasso Regression uses an L1 regularization term as a penalty, which can compress coefficients with small absolute values ​​to zero, effectively selecting variables, but this can also be interpreted as loss of variable information. Ridge Regression uses an L2 regularization term as a penalty, which does not turn non-zero coefficients to zero and does not lose variable information, but it does reduce model accuracy. The Elastic Network combines the two, using a parameter α to adjust their proportions, thus maintaining a balance between variable preservation and model accuracy. Considering the need for iterative drug screening, the weight of the L1 regularization term in Lasso Regression is increased, with α set to 0.75.

[0065] When using elastic networks for model regression, an additional parameter λ is used to determine the number of terms retained in the model, affecting the model's accuracy and predictive ability. Cross-validation was used to determine the value of λ.

[0066] This framework sets two optimization metrics: the inhibition rate of K562 tumor cells and the inhibition rate of normal MSC cells. Given the current situation where drugs with good anti-tumor activity also exhibit strong cytotoxicity to normal cells, the optimization objective of this embodiment is to maximize the inhibition of K562 cell proliferation while minimizing the inhibition of MSC cells, aiming for a proliferation inhibition rate near 0%. Based on this, the optimization objective function formula is designed as follows:

[0067] T = I K562 -I MSC -β|I MSC | Formula (1)

[0068] In formula (1), I K562 I MSC The values ​​represent the inhibition rates of the compound on the proliferation of K562 cells and MSC cells, respectively. β|I MSCThe β value represents a penalty term, which aims to ensure that the compound has no inhibitory effect on MSC cells, nor promotes their proliferation. Considering the practical purpose and the fact that excessively pursuing a 0% inhibition rate of MSC cell proliferation is not very meaningful, the β value is set to 0.5 here.

[0069] For the detailed process of compound combinatorial optimization, please refer to Figure 2 .

[0070] The first round of combination optimization (the screening target for 15 compounds was set at 8 compounds)

[0071] The primary objective of the first round of combinatorial optimization was compound screening. Based on the principle of balancing a large number of compounds with the goal of accurate screening, 48 combinatorial combinations were selected for experimentation, adhering to the uniform design principle. Simultaneously, to avoid overly complex interactions between compounds in the combinations, the average number of compounds in each combination was controlled to approximately 7.5, which is half the total number of compounds. Specific combinations are shown in Table 2.

[0072] Table 2. First Round Experimental Combinations

[0073]

[0074]

[0075] Note: "1" indicates low concentration, and "2" indicates high concentration. The same applies below.

[0076] The inhibitory effects of 48 combined compounds are as follows Figure 3 As shown, among the 48 combinations, the combination compounds (combinations 1-22) consisting of ganoderic acid DM, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, ganoderic acid S1, and ganoderic acid T1) showed the best effect, with a difference of 24.59% in the inhibition rate of K562 and MSC proliferation.

[0077] Based on experimental data from the K562 and MSC cell lines, the inhibition rates of K562 and MSC were modeled and analyzed using the IFER drug combination optimization framework. Cross-validation was used to determine the value of λ. Considering the randomness inherent in cross-validation and the consistency of the two cell line models, λ = 10 was ultimately selected for both cell lines. -2 The model was used to determine the λ value. After determining the λ value, the inhibition rates of K562 and MSC were modeled and analyzed using the IFER drug combination optimization framework. The model fitting results for the two cell types are shown below. Figure 4 and Figure 5 As shown.

[0078] Both cell lines achieved high R-values ​​(>0.9) in their data modeling results. Two models were used to predict the efficacy of the drug combinations. Considering that the number of components in a combination should generally be limited, and to improve prediction accuracy, the number of compounds in the combination should be as small as possible, the screening target for the first round of combination optimization experiments was set at 8 compounds. Next, the proliferation inhibition rates of all possible combinations containing ≤8 compounds against K562 and MSC were predicted, and the objective function value was calculated according to formula (1). The objective function values ​​of all predicted combinations were ranked, and the top eight compounds in terms of frequency were ganoderic acid T, ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, ganoderic acid S1, ganoderic acid TN, and ganoderic acid S. These selected compounds will proceed to the second round of combination optimization experiments.

[0079] Second round of combinatorial optimization

[0080] The second round of experiments, still based on the principle of uniform experimental design, identified 32 combined compounds (Table 3), reducing the average number of compounds in each combination to 6.

[0081] Table 3. Second Round Experimental Combinations

[0082]

[0083]

[0084] The results of the second round of in vitro antitumor activity assays of 32 combined compounds are as follows: Figure 6 As shown in the figure. Compared with the results of the first round of experiments, among the 32 combinations in the second round, the optimal combination showed the largest difference in proliferation inhibition rate between K562 and MSC, reaching 45.69%. This optimal combination consisted of six compounds: ganoderic acid T, ganoderic acid DM, gandoelectol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, and ganoderic acid TN. Next, modeling and analysis of the results of the second round of experiments will be performed.

[0085] Modeling and analysis were performed on the results of the second round of experiments. Relevant modeling information and results are as follows: Figure 7 and Figure 8As shown, although the number of combinations in the second round of experiments decreased, the modeling results still maintained a high R-value and degrees of freedom, demonstrating the model's predictive power.

[0086] To facilitate drug development, the number of compounds in the combination is preferably 3 to 6. Therefore, combinations containing 3, 4, 5, and 6 compounds were predicted based on the objective function of formula (1), with 4 compounds in each type, for a total of 16 combinations (Table 4). The third round of experiments will verify the modeling results of the second round of experiments and find the minimum number of compounds required to achieve the optimal drug efficacy.

[0087] Third round of combinatorial optimization

[0088] The main objective of this round of experiments was to validate the activity prediction model established in the second round of experiments, and to identify the optimal combination compounds. The 16 combinations screened in the second round of experiments were subjected to in vitro antitumor activity assays, and a comparison between the model predictions and the actual activity experimental results is shown below. Figure 9 In combinations containing three compounds, the objective function values ​​were generally low. Combinations containing four compounds (combinations 3-5), five compounds (combinations 3-10 and 3-11), and six compounds (combinations 3-13 and 3-16) all had objective function values ​​exceeding 35%. Among them, combination 3-10, consisting of five compounds—ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, and ganoderic acid S1 (molar concentration ratio 71.12:5.68:5.26:18.69:4.61)—had the highest objective function value, reaching 46.27%, with inhibition rates of 50.20% on K562 cells and 2.62% on MSC cells. This essentially achieved the goal of achieving a 50% inhibition rate on tumor cells K562 while having almost no effect on normal MSC cells. The objective function values ​​for all possible combinations of 3-6 compounds were calculated using the objective function formula (1). The results showed that no combination of compounds had an objective function value higher than that of combination 3-10 in this round of experiments. This indicates that the optimization process has reached its endpoint and can be terminated; that is, combination 3-10 is the optimal combination of 3 to 6 compounds. Next, the antitumor mechanism of this combination of compounds will be further investigated.

[0089] Table 4. Third Round Experimental Combination Table

[0090]

[0091] Example 2: Effect of the combined compounds screened in Example 1 on the K562 cell cycle

[0092] Since the optimal combination 3-10 exhibits good in vitro inhibitory activity against K562 cells, flow cytometry was used to investigate whether this optimal combination has any effect on the cell cycle of tumor cells. K562 cells were treated with three doses of the combined compound—medium dose (concentration in Table 1), low dose (half the concentration in Table 1), and high dose (twice the concentration in Table 1)—for 48 hours, and the cell cycle distribution was analyzed. The results are as follows: Figure 10 As shown in the figure, compared with the control group, the proportion of K562 cells in the G0 / G1 phase significantly increased after treatment with compound 3-10, from 37.91% in the untreated group to 50.66% in the high-dose treatment group. The proportion of S phase decreased slightly, and the proportion of G2 / M phase decreased by 1.74 times compared with the control group. This indicates that the compound inhibits cell proliferation by arresting the cell cycle of K562 cells.

[0093] Example 3: Proteomic analysis of the inhibitory effect of combined compounds on K562 cell proliferation

[0094] K562 cells in good growth condition were collected and treated with a high dose (twice the original dose) of compound 3-10 in quadruplicate, and cultured in a CO2 incubator. Cells were collected at 0 h, 24 h, 48 h, and 72 h, washed with PBS, centrifuged, and lysed. After determining the protein concentration, the protein was precipitated with acetone, followed by reductive methylation, enzymatic digestion, and desalting. The protein sample was then resuspended in 0.1% FA aqueous solution and quantified using NanoDrop to achieve a final peptide concentration of 0.5 μg / μL. After centrifugation at 12000g for 8 min, 10 μL of the supernatant was added to a sample vial for detection using liquid chromatography-tandem mass spectrometry (LC-MS / MS).

[0095] Based on the experimental results, GO function enrichment, KEGG signaling pathway enrichment, and protein network interaction analysis were performed. The results showed that the combined compound 3-10 mainly interfered with the cholesterol synthesis pathway in K562 cells, thereby inducing cell cycle arrest and ultimately leading to cell death. After treatment with the combined compound, the expression of key proteins in the cholesterol synthesis pathway, HMGCS1, MVK, FDFT1, and CYP51A1, was downregulated, and the downregulation fold increased with time. Specifically, the downregulation fold of FDFT1 and CYP51A1 proteins at 48 h was approximately twice that at 24 h (Table 5), indicating that the combined compound's effect on K562 proteins has a certain time effect.

[0096] Table 5. Differential protein expression in the cholesterol synthesis pathway

[0097]

[0098] Example 4: Western Blot Experiment Verification

[0099] Based on the results of proteomics data analysis and relevant literature reports, key differentially expressed proteins in the cholesterol synthesis pathway, CYP51A1 and FDFT1, as well as the cell cycle-related differentially expressed protein SKP2, were selected and validated using Western blot experiments. The results are shown below. Figure 11 The proteins CYP51A1 and FDFT1, which are related to cholesterol synthesis, were downregulated within the first 24 hours, while the protein SKP2, which is related to the cell cycle, was not significantly downregulated at 24 hours but was significantly downregulated at 48 hours. Western blot results confirmed the inhibitory effect of compound 3-10 on cholesterol synthesis and cell cycle arrest in K562 cells.

Claims

1. An antitumor composition, characterized in that... The antitumor composition consists of ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one and ganoderic acid S1.

2. The antitumor composition according to claim 1, wherein the molar concentration ratio of ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one and ganoderic acid S1 is 70-72:5-6:5-6:18-20:4-5.

3. The antitumor composition according to claim 2, wherein the preferred molar ratio of ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one and ganoderic acid S1 is 71.12:5.68:5.26:18.69:4.

61.

4. An antitumor drug, characterized in that... It consists of the antitumor composition of any one of claims 1-3, and a pharmaceutically acceptable carrier.

5. A method for preparing the antitumor composition according to any one of claims 1-3, characterized in that... Includes the following steps: (1) Selecting the experimental concentration of the candidate compounds; Based on preliminary measurements of the cell proliferation inhibition rates of 15 triterpenoids and sterols derived from Ganoderma lucidum, different concentrations of the compounds were obtained through dilution and experimental verification to identify the IC50 value of each compound for inhibiting the proliferation of K562 leukemia cells. 30 The effective concentration is below the value; (2) IFER optimization framework for combinatorial compounds The drug combination optimization framework IFER, which combines a feedback system-controlled FCS and an elastic network algorithm EN, optimized the selection of 15 compounds. The Elastic Network Algorithm (EN) combines lasso regression and ridge regression. The lasso regression model uses an L1 regularization term as a penalty term, while the ridge regression model uses an L2 regularization term as a penalty term. The Elastic Network Algorithm (EN) combines the two and uses the parameter α to adjust the proportion of each, which can maintain a balance between variable preservation and model accuracy. Drugs were screened iteratively, and the weight of the L1 regularization term in the lasso regression was increased, with the α parameter set to 0.

75. When performing model regression, the parameter λ was also introduced to determine the number of terms to retain in the model, and the value of λ was determined by cross-validation. The drug combination optimization framework IFER sets two optimization indicators: the inhibition rate of proliferation of tumor cells K562 and normal cells MSC; and designs an optimization objective function formula: T = I K562 -I MSC -β|I MSC | Equation (1) In formula (1), I K562 I MSC The values ​​represent the inhibition rates of the compound on the proliferation of K562 cells and MSC cells, respectively, β|I MSC | indicates a penalty term, with the β value set to 0.5; Three rounds of combination optimization experiments were conducted on 15 compounds based on the drug combination optimization framework IFER: First round of combination optimization: The goal of the first round of optimization was compound screening. Based on the balance of a large number of compounds and the aim of achieving accurate screening, 48 combinations were selected for experiments according to the principle of uniform experimental design, and the average number of compounds in each combination was controlled to be 7.

5. According to the experimental data of the K562 and MSC cell lines, among the 48 combinations, the difference in the maximum inhibition rate between the best combination of K562 and MSC was only 24.6%, and a second round of combination optimization is still needed. The inhibition rates of K562 and MSC were modeled and analyzed using the drug combination optimization framework IFER. The λ value was determined through cross-validation, and λ = 10 was selected for both cell lines. -2 The model was used; the screening target of the first round of combinatorial optimization experiment was set as 8 compounds. The K562 and MSC inhibition rates of all possible combinations containing ≤8 compounds were predicted, and the objective function value was calculated according to formula (1); all predicted combinations were sorted by objective function value, and the occurrence frequency of each drug in the top 100 combinations was counted. The top eight compounds were selected as ganodericacid T, ganoderic acid DM, ganoderol B, 6-O-methylcerevisterol, 3,5-dihydroxyergosta-7,22-dien-6-one, ganoderic acid S1, ganoderic acid TN and ganoderic acid S; The eight compounds selected above were then subjected to a second round of combination optimization: Based on the principle of uniform experimental design, 32 combined compounds were selected for experimental verification in the second round of optimization, and the average number of compounds in the combination was reduced to 6. The experimental results show that the difference in inhibition rates between K562 and MSC in the 32 combinations in the second round has increased to a maximum of 45.7%. Modeling analysis was performed on the results of the second round of experiments; combinations containing 3, 4, 5 and 6 compounds were predicted according to the objective function of formula (1). It was found that the modeling results still obtained a high R value (>0.9), indicating that the model has good predictive ability; in order to facilitate drug development, the number of drugs in the combination is best to be 3 to 6. Therefore, the optimal three-drug, four-drug, five-drug and six-drug combinations were predicted according to the objective function of formula (1), with 4 of each type, for a total of 16 combinations for the next round of experiments. Third round of combinatorial optimization and verification: The main purpose of this round of combination optimization is to verify the activity prediction model established in the second round of experiments, and at the same time find the optimal combination compound; according to the objective function formula (1), the objective function values ​​of all possible 3-6 combination compounds are calculated, and the objective function values ​​of combination compounds 3-10 are the highest; pharmacodynamic experiments were conducted on 16 combinations, and the antitumor composition according to any one of claims 1-3 is obtained based on the comparison between its predicted value and the experimental results.

Images