In this study, we used a complex network to identify the coupled herbs of ZGXFD in the treatment of hypertension. A CNN was established to summarize the dose laws of herbs. We mapped and summarized the herb targets, hypertension and their symptom-related genes on the PPIN; analysed the topological relationships in the network; and obtained their biological characteristics. After multimodel comparison, the best efficacy KNN framework was established, which was used to summarize the clinical and biometric information of the coupled herbs. Finally, we summarized 13 herbs commonly coupled with ZGXFD that regulate hypertension and cardiovascular risk factors.
We found that ZGXFD mainly targets the inflammatory-related pathway and its downstream pathway, in which ZGXFD can inhibit the three NF-κB pathways and simultaneously activate the MAPK signalling pathway. Regarding its downstream pathway, ZGXFD not only inhibits insulin resistance induced by inflammatory cytokines, such as TNF-α, but also regulates glucose and lipid metabolism by directly or indirectly increasing IRS-1 activity.
In the progression of hypertension, it is also necessary to control cardiovascular risk factors. The kidney is the first target organ that develops pathological changes in hypertension. When hypertension causes pathological damage to the kidney, inflammatory factors stimulate fibroblasts to lay extracellular matrix, resulting in tubulointerstitial fibrosis[30, 31]. These haemodynamic changes also activate the renin–angiotensin–aldosterone system (RAAS) and aggravate vascular endothelial hyperplasia, abnormal glomerular metabolism and podocyte damage. After excessive activation of the RAAS, the secretion of angiotensin II (Ang II) increases, resulting in the upregulation of the expression of inflammatory cytokines, such as TNF-α, IL-6, and IL-8. In terms of kidneys, type I and IV collagen mRNA and protein levels increased [32–34]. In addition, the overexpression of Ang II receptors on podocytes could promote the disappearance of the foot process[35], induce the reorganization of the podocyte cytoskeleton and lead to proteinuria[36].
The hypertension-related inflammatory pathway regulated by ZGXFD is mainly the NF-κB signalling pathway. ZGXFD inhibits the excessive activation of the NF-κB signalling pathway by indirectly or directly regulating the activities of TNF-α, IL-1β, IL-6 and other inflammatory factors. ZGXFD further inhibits the inflammatory response caused by these inflammatory cytokines as well as the upregulation of the MAPK signalling pathway caused by MAPK8 and TNF and interstitial fibrosis caused by increased serine[37, 38].
ZGXFD regulates insulin resistance and the PPAR signalling pathway through specific action on INS, PPARG, ADIPOQ and other targets; improves the utilization of lipids and sugars, especially the fat metabolism regulated by insulin; and enhances the utilization efficiency of sugars and lipids[39, 40]. ZGXFD regulates and inhibits cytoskeletal changes, reduces glial cell migration and improves vascular endothelial function by interfering with VEGFA, PRKCB and JNK in the focal adhesion pathway to inhibit podocyte injury and protect nephrons[41, 42].
In the past, few studies have analysed the clinical dose application characteristics of herbs, especially the dose changes after herb coupling.
In this study, several dimensions of clinical application, including frequency, dose information and clinical symptoms of hypertension, were analysed with target-based herb biological characteristics in the KNN model. TCM attaches great importance to changing the dose and type of herbs according to the characteristics of the disease. Compared with the existing methods of complete target gene crossover in network pharmacology, our method can better reflect the prescription law of "jun-chen-zuo-shi" in TCM.
However, this study did not assess the relationship between herbal compounds and diseases, which brings confusion to further drug development in the future. Thus, we will prioritize the screening of effective compounds in our future work. The challenge of this study is primarily derived from the noise removal of data sources. For EMR data, the baseline matching of patients was based on propensity scores, but it is difficult to determine the herb combination with poor curative effects during treatment. Another source of noise is the collection of herb effective targets and disease genes. As a slow disease, hypertension has a long course and easily causes target organ damage in the heart, brain, and kidney. It is difficult to collect related genes and screen effective herbs in different stages of hypertension.