ObjectiveTo explore the potential molecular mechanism of Qizhu Kang'ai Formula （芪术抗癌方， QZKAF） for the treatment of colorectal cancer （CRC）. MethodsNetwork pharmacology was used to analyze the active ingredients and targets of QZKAF for CRC， and analyze the key targets of QZKAF for the treatment of CRC by gene function annotation （GO） and Kyoto Encyclopedia of Genomes （KEGG） pathway enrichment analysis. Molecular docking was applied to predict the binding activity of the core active ingredients to the key targets. A orthotopic transplantation tumor mice model of CRC was established to validate the key targets of QZKAF for CRC obtained from network pharmacology analysis. Forty-eight mice were randomly divided into the sham operation group， the model group， the 5-fluorouracil （5-Fu） group， and the QZKAF low-， medium-， and high-dose groups， with 8 mice in each group. Except for the sham operation group， the remaining groups underwent colon cancer orthotopic transplantation tumor modeling. The 5-Fu group was given 30 mg/kg of 5-Fu by intraperitoneal injection once every 3 days on the alternate day after modeling， while the QZKAF low-， medium-， and high-dose groups were given 2.925， 5.85， and 11.7 g/（kg·d） of QZKAF by gastric gavage， respectively， and the sham-operation group and the model group were gavaged with 0.1 ml/10 g of normal saline every day， all for 21 days. The in situ tumors mass and the number of liver metastases were compared between the groups. The pathological changes of colon tumor tissues were observed by HE staining， and the protein expression of protein tyrosine phosphatase nonreceptor type 1 （PTPN1）， vinculin， integrin subunit αν， integrin subunit β3， and E-cadherin were detected in colon tumor tissues by Western blot. ResultsNetwork pharmacology screening yielded that the top six core active ingredients of QZKAF intervening in CRC were quercetin， kaempferol， apigenin， luteolin， baicalein and ursolic acid. There were 212 targets of action， and the ranked top three were prostaglandin endoperoxide synthase 1 （PTGS1）， prostaglandin endoperoxide synthase 2 （PTGS2）， and PTPN1， which may be the key targets of QZKAF in the treatment of CRC. These key targets were significantly enriched mainly in phosphatidylinositol 3-kinase/protein kinase B （PI3K-Akt） signaling pathway， focal adhesion and adhesion junction. Molecular docking results： except for PTGS1 with better binding activity to quercetin， kaempferol， and apigenin （binding energy ≥-7.0 kcal/mol）， PTGS1 showed strong binding activity to lignans， baicalein， ursolic acid， as well as PTGS2 and PTPN1 to the six core active ingredients （binding energy ＜-7.0 kcal/mol）. Experimental validation results： the protein expression of PTPN1， vinculin， integrin subunit αν， integrin subunit β3 in the colon tumor tissues of mice in the model group significantly increased， and the expression of E-cadherin significantly decreased compared to those in the sham operation group （P＜0.05）. Compared to those in the model group， the mass of the in situ tumors was reduced， and the number of hepatic metastasis nodules decreased in the high- and medium-dose QZKAF groups （P＜0.05）； the expression levels of PTNP1， vinculin， integrin subunit αν， integrin subunit β3 and E-cadherin in all QZKAF groups and 5-Fu group showed different degrees of retracement， and the changes of the indicators in all QZKAF groups showed a certain degree of dose-dependence （P＜0.05）. HE staining showed that the nuclei of tumor cells in the model group were mostly schizophrenic， and there were different degrees of nuclear fragmentation of tumor cells in all QZKAF groups with more in the medium- and high-dose groups. ConclusionQZKAF could inhibit the growth of in situ tumors and liver metastasis of CRC. Its mechanism might be related to the regulation of tumor cell-cell and tumor cell-extracellular matrix adhesion by PTPN1.