Gong Ting-Ting, He Xin-Hui, Gao Song, Wu Qi-Jun
Department of Obstetrics and Gynecology, Shengjing Hospital of China Medical University, Shenyang, China.
Department of Clinical Epidemiology, Shengjing Hospital of China Medical University, Shenyang, China.
J Cancer. 2021 Mar 15;12(10):2877-2885. doi: 10.7150/jca.46621. eCollection 2021.
Ovarian cancer (OC) has the highest mortality among gynecological malignancies, and resistance to chemotherapy drugs is common. We aim to develop a machine learning approach based on gut microbiota to predict the chemotherapy resistance of OC. The study included patients diagnosed with OC by pathology and treated with platinum and paclitaxel in Shengjing Hospital of China Medical University between 2017 and 2018. Fecal samples were collected from patients, and 16S rRNA sequencing was used to analyze the differences in gut microbiota between OC patients with and without chemotherapy resistance. Nine machine learning classifiers were used to derive the chemotherapy resistance of OC from gut microbiota. A total of 77 chemoresistant OC patients and 97 chemosensitive OC patients were enrolled. The gut microbiota diversity was higher in OC patients with chemotherapy resistance. There were statistically significant differences between the two groups in Shannon indexes (P <0.05) and Simpson indexes (P <0.05). Machine learning techniques can predict the chemoresistance of OC, and the random forest showed the best performance among all models. The area under the ROC curve for RF model was 0.909. The diversity of gut microbiota was higher in OC patients with chemotherapy resistance. Further studies are warranted to validate our findings based on machine learning techniques.
卵巢癌（OC）在妇科恶性肿瘤中死亡率最高，且对化疗药物产生耐药性很常见。我们旨在开发一种基于肠道微生物群的机器学习方法来预测OC的化疗耐药性。该研究纳入了2017年至2018年间在中国医科大学附属盛京医院经病理诊断为OC并接受铂类和紫杉醇治疗的患者。从患者身上采集粪便样本，采用16S rRNA测序分析有化疗耐药性和无化疗耐药性的OC患者肠道微生物群的差异。使用九个机器学习分类器从肠道微生物群中推导OC的化疗耐药性。共纳入77例化疗耐药的OC患者和97例化疗敏感的OC患者。有化疗耐药性的OC患者肠道微生物群多样性更高。两组在香农指数（P<0.05）和辛普森指数（P<0.05）方面存在统计学显著差异。机器学习技术可以预测OC的化疗耐药性，随机森林在所有模型中表现最佳。RF模型的ROC曲线下面积为0.909。有化疗耐药性的OC患者肠道微生物群多样性更高。有必要进行进一步研究以基于机器学习技术验证我们的发现。
