摘要目的:探讨原发灶手术（PLR）对首次诊断为Ⅳ期乳腺癌的患者的预后价值，并分析从PLR中可能获益的人群特征。方法:从SEER数据库中提取2010—2018年初诊Ⅳ期乳腺癌的20 433例病例。按照是否进行PLR分为手术组和非手术组，并通过倾向性评分匹配法平衡混杂偏倚。采用 Cox比例风险模型寻找影响Ⅳ期乳腺癌患者预后的独立影响因素，绘制 Kaplan-Meier生存曲线，并对其进行 Log-rank检验以评价生存差异。比较不同转移状况时PLR对预后的影响。 结果:年龄≥55岁，黑种人，除已婚外的婚姻状况，较高的T分期、N分期，组织学分级高，雌激素受体、孕激素受体阴性表达，Luminal A型肿瘤，骨、脑、肺、肝转移，均与患者的不良预后相关（均 P<0.05）。PLR（ HR=0.58，95% CI为0.55~0.61， P<0.001）、化疗（ HR=0.64，95% CI为0.60~0.68， P<0.001）及远处转移部位手术（ HR=0.88，95% CI为0.79~0.99， P=0.035）为保护因素。 结论:PLR对不同T分期、N分期、组织学分级Luminal分型的Ⅳ期乳腺癌患者的预后均有显著改善。

abstractsObjective:To investigate the prognostic value of primary lesion resection (PLR) for patients with initially-diagosed stage Ⅳ breast cancer, and to analyze the characteristics of individuals who may benefit from PLR.Methods:20 433 cases initially diagnosed with stage Ⅳ breast cancer from 2010 to 2018 were extracted from SEER database. Patients were divided into surgery group and non-surgery group based on whether PLR was performed, using propensity score matching method to balance confounding biases. Cox proportional hazards model was utilized to analyze independent prognostic factors affecting the prognosis of patients with stage Ⅳ breast cancer. Kaplan-Meier survival curve was plotted, and Log-rank test was conducted to evaluate the difference in survival. The impact of PLR on prognosis was compared under different metastatic conditions. Results:Age ≥55 years, black race, marital status other than married, higher T stage and N stage, high histological grade, negative expression of estrogen and progesterone receptors, Luminal A subtype tumors, and presence of bone, brain, lung, or liver metastases were associated with poor prognosis in the patients (all P<0.05). PLR ( HR=0.58, 95% CI: 0.55-0.61, P<0.001), chemotherapy ( HR=0.64, 95% CI: 0.60-0.68, P<0.001) and distant metastasis surgery ( HR=0.88, 95% CI: 0.79-0.99, P=0.035) were protective factors. Conclusion:PLR significantly improves the prognosis of patients with stage Ⅳ breast cancer of different T stages, N stages, histologic grades, and Luminal subtypes.