目的 通过算法识别与实验验证相结合，高效识别抗胰腺导管腺癌（pancreatic ductal adenocarcinoma，PDAC）中药活性成分，并评价其抗肿瘤活性。方法 利用公共数据库的PDAC表达谱数据及检索国内外相关文献，通过单样本基因集富集分析（single sample gene set enrichment analysis，ssGSEA）方法验证其驱动基因，计算驱动基因之间的最大连通分量（largest connected component，LCC）并构建疾病网络，运用网络邻近度算法量化中药活性成分靶点与疾病网络在全人类蛋白质-蛋白质相互作用网络中的距离预测其治疗潜力，并对候选中药活性成分进行实验验证。结果ssGSEA分析显示，PDAC驱动基因在肿瘤组的富集得分显著高于正常组（P≤0.001）。LCC分析表明，驱动基因形成了高度紧密的相互作用模块（P＝0.021）。运用网络邻近度算法对中药活性成分进行快速筛选，识别出地肤子皂苷Ic为潜在的抗PDAC候选药物。进一步实验验证表明，在人胰腺癌PANC-1和人胰腺导管癌MIA PaCa-2细胞中，地肤子皂苷Ic在8.5～9.5 μmol/L浓度下显著抑制肿瘤细胞的增殖、迁移和集落形成（P＜0.05），并诱导细胞凋亡。GSEA富集分析显示，缺氧反应因子-1（hypoxia-inducible factor-1，HIF-1）信号通路活性显著下调（normalized enrichment score，NES＝−0.83），生存分析揭示HIF1A的高表达与生存期呈负相关。KEGG富集分析显示地肤子皂苷Ic主要富集于丝裂原活化蛋白激酶（mitogen-activated protein kinase，MAPK）、磷脂酰肌醇-3-羟激酶（phosphatidylinositol-3-hydroxykinase，PI3K）-蛋白激酶B（protein kinase B，Akt）及血管内皮生长因子（vascular endothelial growth factor，VEGF）等信号通路。结论 基于网络邻近度算法，提出了一种中药活性成分高效识别策略，成功识别出地肤子皂苷Ic为PDAC潜在治疗药物，为中药新药开发提供了新思路。

Objective Efficient identification of active ingredients from traditional Chinese medicine (TCM) with anti-pancreatic ductal adenocarcinoma (PDAC) properties is achieved through a combination of algorithm identification and experimental validation, followed by an evaluation of their antitumor activity. Methods PDAC expression profile data from public databases and relevant literature were analyzed. Single sample gene set enrichment analysis (ssGSEA) was used to validate PDAC driver genes. The largest connected component (LCC) of driver genes was calculated, and a disease network was constructed. A network proximity algorithm was applied to quantify the distance between TCM active ingredient targets and the disease network within the human protein-protein interaction network, predicting their therapeutic potential. Candidate TCM active ingredients were experimentally validated. Results The ssGSEA analysis revealed that the enrichment scores of PDAC driver genes were significantly higher in the tumor group than in the normal group (P ≤ 0.001). LCC analysis indicated that the driver genes formed a tightly connected interaction module (P = 0.021). Using the network proximity algorithm to rapidly screen the active components of TCM, momordin Ic was identified as a potential anti-PDAC candidate drug. Further experimental validation demonstrated that momordin Ic significantly inhibited tumor cell proliferation, migration, and colony formation in PANC-1 and MIA PaCa-2 cells at concentrations of 8.5—9.5 μmol/L (P < 0.05) and induced apoptosis. GSEA enrichment analysis showed that the hypoxia-inducible factor-1 (HIF-1) signaling pathway activity was significantly downregulated (normalized enrichment score, NES = −0.83). Survival analysis revealed a negative correlation between high HIF1A expression and survival period. KEGG enrichment analysis indicated that momordin Ic primarily affected the MAPK, PI3K-Akt, and VEGF signaling pathways. Conclusion This study proposes an efficient strategy for identifying active ingredients in traditional Chinese medicine based on a network proximity algorithm, successfully identifying momordin Ic as a potential therapeutic agent for PDAC, offering new insights for TCM drug development.

TP18;R285

国家重点研发计划（2022YFC3502000）；国家自然科学基金重点项目（82430119）；上海市晨光计划（23CGA45）