1. Patient characteristics.
The patients were selected for our study as following criteria: 18≤ age≤75；the total of survival time≥6 months; diagnosed by pathological examination or clinical diagnosis. The major exclusive criteria as follows: concurrent other malignant tumors; comorbidities included severe diseases (severe cardiac kidney or pulmonary disease; acute upper gastrointestinal bleeding leading to death)；incomplete medical record; loss to follow-up. Informed consent was waived off due to the retrospective design of the study.
All the patients were received routine treatments. Based on their wishes, the patients choose to take CM or not. According to the syndrome differentiation, all the patients were administered individually, wherein the formula was administered orally two times daily 30 minutes after meals.
3. Identification of core drugs
The prescriptions in CM group were collected. From the date of receiving CM treatment to the patient's death or up to the time of the research, the patient prescriptions of CM group were collected. Refer to the 2015 edition of Chinese Pharmacopoeia to unify the name of CM to their official name. According to this criteria, “chaihu(local fried or bran fried)” is equal to “chaihu”. “shengshaishen” is dubbed for “renshen”. In order to decrease the bias, the CM not included in the Pharmacopoeia, such as “baihuasheshecao”, is uniformly named with reference to textbooks or literatures. Then the prescriptions were input TCMISS by professional worker and checked two times by another two workers to assure its correctness, accuracy and reliability. Based on the threshold of "support degree" to 300 and the “confidence level” to 0.95, “Data analysis” module in TCMISS was employed to acquire core drugs (HXCF).
4. Screening the active ingredients of HXCF
All of the ingredients in HXCF were retrieved from the relevant database, including Traditional Chinese Medicine Systems Platform (TCMSP, http://tcmspw.com/tcmsp.php, Version: 2.3, accessed March 2020), Natural Product Activity & Species Source Database(NPASS, http://bidd2.nus.edu.sg/NPASS , accessed March 2020), A Bioinformatics Analysis Tool for Molecular Mechanism of Traditional Chinese Medicine (BATMAN-TCM http://bionet.ncpsb.org/batman-tcm/ , accessed March 2020). Finally, 1125 chemical ingredients were collected. Then, the structure of the chemical ingredients were acquired from NCBI PubChem database (https://www.ncbi.nlm.nih.gov/, accessed March 2020). Then, the ingredients were screened as following criteria.
Oral bioavailability (OB) is named as the relative amount of unmodified drug absorbed into the circulatory system after extravascular administration. It is a significant index to evaluate the efficacy of drug absorption. Components with high OB are often regarded as the potential drugs being developed. Therefore , chemicals with OB≥30% were selected as the threshold value for screening potential ingredients.
Drug-Likeness (DL) is a criterion to evaluate the pharmacokinetic characteristics of chemicals in human body, comprehensively. DL is used to filter excellent ingredients with undesirable properties. The DL is calculated as the following formula:
In the formula, X represents the chemical properties of herbal ingredients based on
Dragon soft molecular descriptors, and Y represents the average drug-likeness index of all 6511 molecules in Drug-Bank database ( http://www.drugbank.ca/ accessed March 2020). In this study, the criterion, DL≥0.18, was defined to screen out potential active ingredients.
The drug half-life (HL), commonly named as the biological half-life, refers to the time it takes for the concentration of the drug in the blood or the amount of the drug to be decreased to half in the body. In this study, HL≥4 was defined as the criteria to select potential active ingredients.
5. Identification of the chemical targets of HXCF
The chemical targets of HXCF were acquired from TCMSP database (http://tcmspw.com/tcmsp.php, Version: 2.3, accessed March 2020) DrugBank and related literatures. Then the resulted targets were normalized their corresponding nomenclatures by UniPort Database (https://www.uniprot.org/, accessed March 2020).
6. PLC Related Targets
The differential expressed genes (DEG) of PLC patients were acquired from GEO database (https://www.ncbi.nlm.nih.gov/geo/ Series: GSE45267, Samples: GSM1100370 to GSM1100456). Genes with a P value < 0.005 and |log 2| > 1 were deemed as PLC related targets.
7. Construction of Network
To elucidate the pharmacological mechanism of HXCF in treatment of PLC, the active ingredient-target system of HXCF was generated by Cytoscape software (Version 3.7.2 Boston, MA, USA). Plugin Bisogenet of Cytoscape was used to build Protein-Protein Interaction (PPI) from Human Protein Reference Database (HPRD), Database of Interacting Proteins (DIP™), Biological General Repository for Interaction Datasets (BioGRID), biomolecular interaction network database (BIND), IntAct Molecular Interaction Database (IntAct) and Molecular INTeraction database (MINT). All network was visualized by Cytoscape software (Version 3.7.2 Boston, MA, USA).
8. Network Merge
The PPI network of HXCF was merged by Cytoscape software. And the nodes with topological importance in the interaction network were screened by calculating Betweenness Centrality (BC), Degree Centrality (DC), Eigenvector Centrality (EC) and Closeness Centrality (CC), which represent the topological significance. What’s more, the parameters also have been reported about their definitions and computational formulas and used in network pharmacology and systems pharmacology. Local average connectivity-based method (LAC), and Network Centrality (NC) with the Cytoscape plugin CytoNCA.
9. Bioinformatic Analysis
The Database for Annotation, Visualization and Integrated Discovery (DAVID https://david.ncifcrf.gov/ version 6.8) is a website providing a comprehensive set of functional annotation tools to understand biological meaning behind large list of genes and carrying out GO enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis.In this study, top 10 of GO analysis in molecular function (MF), biological process (BP) and cellular component (CC) as well as KEGG were listed. Eventually, we used GOplot package and cytoscape for visualization.
10. Molecule Docking
The most maximum betweenness centrality molecule and corresponding ligands were chose to dock. The crystal structure of active molecules were draw by Protein Data Bank[21,22] (PDB: 2gmx, 1p9m). Through Chemical Book (https://www.rcsb.org/ accessed May 2020) database, the small molecules were saved as mol files uniformly. The spatial structure convert it to pdb format after checked in the PyMol software (https://pymol.org/version version 2.3). For increasing accuracy, Auto Tools, as well as Autodock Vina (http://vina.scripps.edu/index.html version 1.2), was employed to conduct docking. After pretreatment of the excess protein chain and ligands deletion, the water molecules removement as well as Gasteiger charge calculation from molecule ligand complex, the Autodock Vina was employed to dock between small molecules with proteins. The dominant conformation was analyzed and plotted with Schrodinger software (https://www.schrodinger.com accessed May 2020).