Department of Surgery, Breast Cancer Center, Gil Medical Center, Gachon University College of Medicine, Incheon, Republic of Korea.
Department of Surgery, Breast Cancer Center, Catholic University Saint Mary's Hospital, Incheon, Republic of Korea.
PLoS One. 2018 Jul 6;13(7):e0200100. doi: 10.1371/journal.pone.0200100. eCollection 2018.
Personalized treatment for cancer patients is a hot topic of debate, particularly the decision to initiate chemotherapy in patients with Estrogen receptor (ER)-positive, HER2-negative tumors in the early stages of breast cancer (BC). Owing to significant advancements in information technology (IT) and genomics, clinicians are increasingly attaining therapeutic goals rapidly and safely by effectively differentiating patient subsets that require chemotherapy. IBM Watson for Oncology (WFO) is a cognitive computing system employed by clinicians to provide evidence-based treatment options for cancer. WFO aids in clinical diagnosis, with claims that it may be superior in performance to human clinicians. The current study was based on the hypothesis that WFO alone cannot effectively determine whether or not chemotherapy is essential for the subset of ER-positive, HER2-negative BC patients.
From December 2015 to July 2017, 95 patients with ER-positive, HER2- negative BC subjected to treatment were retrospectively examined using WFO, and outputs compared to real clinical practice. Treatment options were suggested by WFO, and WFO recommendations calculated both with and without data from the gene expression assay (GEA).
WFO without GEA was unable to determine the groups of patients that did not require chemotherapy. Concordant therapeutic recommendations between real clinical practice and WFO without GEA were obtained for 23.2% of the patient group. On the other hand, the results of WFO with GEA showed good clinical applicability. Sensitivity, specificity, positive predictive and negative predictive values of WFO with GEA were 100%, 80%, 61% and 100%, respectively.
Our collective findings indicate that WFO without the gene expression assay has limited clinical utility.
针对癌症患者的个体化治疗是一个热门的讨论话题，尤其是在乳腺癌（BC）早期雌激素受体（ER）阳性、HER2 阴性肿瘤患者中启动化疗的决策。由于信息技术（IT）和基因组学的显著进步，临床医生通过有效区分需要化疗的患者亚组，正在越来越快、越来越安全地实现治疗目标。IBM Watson for Oncology（WFO）是一种认知计算系统，供临床医生用于为癌症提供基于证据的治疗选择。WFO 有助于临床诊断，据称其性能可能优于人类临床医生。本研究基于这样一种假设，即单独使用 WFO 无法有效确定 ER 阳性、HER2 阴性 BC 患者亚组是否需要化疗。
2015 年 12 月至 2017 年 7 月，回顾性分析了 95 例接受治疗的 ER 阳性、HER2 阴性 BC 患者，使用 WFO 进行检查，并将结果与实际临床实践进行比较。WFO 提供了治疗选择建议，并计算了有无基因表达分析（GEA）数据的 WFO 建议。
没有 GEA 的 WFO 无法确定不需要化疗的患者群体。无 GEA 的 WFO 与真实临床实践之间的治疗建议一致，占患者组的 23.2%。另一方面，有 GEA 的 WFO 结果具有良好的临床适用性。有 GEA 的 WFO 的敏感性、特异性、阳性预测值和阴性预测值分别为 100%、80%、61%和 100%。
我们的综合研究结果表明，没有基因表达分析的 WFO 临床应用价值有限。
