Establishment of the prediction model and biological mechanism exploration for secondary imatinib-resistant in gastrointestinal stromal tumor

A gastrointestinal stromal tumor (GIST) is mostly driven by the auto-activated, mutant <i>KIT</i> receptor tyrosine kinase gene or by the platelet-derived growth factor receptor alpha. Inhibition of <i>KIT</i>-signaling is the primary molecular target therapy for GIST, which is performed by the drug imatinib clinically. However, more than half of advanced or metastatic GIST develop secondary resistance to imatinib within 2 years after initiation of treatment, and the mechanism of acquired imatinib-resistant in GIST remains unclear. Therefore, we designed the present study, and firstly analyzed the gene expression profile of imatinib-resistant and sensitive GIST from GEO DataSet and identified 44 differential expressed genes. Then, a model including nine genes with their expressed coefficients was identified as a risk score to predict imatinib-resistant GIST. Internal and external validation of the prediction model was performed through the ROC curve, and the area under the curve was 0.967 (95%CI 0.901–1.000) and 0.917 (95%CI 0.753–1.000), separately. Lastly, the effect of immune, m⁶A, pyroptosis, and ferroptosis-related genes on imatinib-resistant GIST was also assessed because DNA replication was the most enriched biological function of DEGs after functional annotation, pathway enrichment, and protein-protein interaction network analyses. In conclusion, the present study established a novel model to predict secondary imatinib-resistant GIST. Meanwhile, the bioinformatic mining results provided potential and promising targets for imatinib-resistant therapy.

胃肠道间质瘤（gastrointestinal stromal tumor, GIST）大多由自发激活的突变型KIT受体酪氨酸激酶基因，或血小板衍生生长因子受体α驱动。KIT信号通路抑制是GIST的一线分子靶向治疗策略，临床中以伊马替尼作为该治疗的常用药物。然而，超过半数的晚期或转移性GIST患者在接受治疗后的2年内会出现伊马替尼继发耐药，而GIST获得性伊马替尼耐药的具体机制仍未阐明。为此，本研究设计并开展了本项工作：首先从GEO数据集（GEO DataSet）中分析伊马替尼耐药与敏感型GIST的基因表达谱，筛选并鉴定得到44个差异表达基因；随后构建了包含9个基因及其表达系数的风险评分模型，用于预测伊马替尼耐药型GIST。通过ROC曲线对该预测模型进行内部与外部验证，其曲线下面积分别为0.967（95%置信区间：0.901–1.000）与0.917（95%置信区间：0.753–1.000）。最后，经功能注释、通路富集及蛋白质相互作用网络分析发现，差异表达基因最显著富集的生物学功能为DNA复制，因此本研究还评估了免疫、m⁶A、细胞焦亡（pyroptosis）及铁死亡（ferroptosis）相关基因对伊马替尼耐药GIST的影响。综上，本研究构建了一种全新的伊马替尼继发耐药GIST预测模型，本研究的生物信息学挖掘结果同时为伊马替尼耐药的临床治疗提供了潜在且极具前景的靶点。