摘要目的分析乳腺叶状肿瘤的常规形态学特点及其与预后的相关性，为我国乳腺叶状肿瘤患者的分级诊断提供辅助信息。方法回顾性分析就诊于1999至2013年的83例乳腺叶状肿瘤患者资料，按2012年WHO乳腺肿瘤病理学和遗传学分类标准提供的6个组织学指标进行分级，采用Cox回归模型分析乳腺叶状肿瘤的组织学分级指标与复发转移的相关性。结果乳腺良性、交界性和恶性叶状肿瘤分别为48例（57.8％）、24例（28.9％）和11例（13.3％）。72例随访病例的局部复发率为20.8％（15／72），中位复发时间为24个月。良性、交界性、恶性叶状肿瘤的局部复发率分别为17.5％（7／40）、22.7％（5／22）和3／10，差异无统计学意义（P＞0.05）。1例（1／22）乳腺交界性叶状肿瘤和3例（3／10）恶性叶状肿瘤均于术后5年内发生肺或骨转移。6个组织学指标中，仅间叶细胞核分裂象、间叶性肿瘤细胞异型性与叶状肿瘤复发明显相关（P＝0.001和P＝0.006）。 Cox多因素生存分析结果提示，间叶性肿瘤细胞异型性可作为叶状肿瘤无复发生存期的独立预测因子，相对危险度为6.40（P＝0.018，95％CI＝1.378～29.732）。结论 WHO乳腺肿瘤病理学和遗传学分类标准的6个组织学指标对于提示乳腺叶状肿瘤局部复发的重要性并不相同，其中间叶细胞核分裂象增多或间叶性肿瘤细胞重度异型性的乳腺叶状肿瘤更易发生局部复发。因此在对乳腺叶状肿瘤进行分级时，应更关注这2个组织学指标的意义。

abstractsObjective To study the relationship between morphological characteristics , grading, diagnosis and prognosis in phyllodes tumors ( PT ) of the breast .Methods A retrospective study was carried out on 83 PTs diagnosed between 1999 and 2003 that were classified semi-quantitatively according to the WHO recommendation .Follow-up data was available for some cases , and Cox regression analysis was used to evaluate factors affecting metastasis and recurrence .Results All cases were classified into the benign ( 57.8%) , borderline ( 28.9%) and malignant ( 13.3%) .The overall recurrence rate for the 72 cases with follow-up data was 20.8%(15/72), and was 17.5%(7/40) in benign, 22.7%(5/22) in borderline and 3/10 in malignant PT , respectively , with no significant difference ( P>0.05 ) .The median interval between the initial diagnosis and the first recurrence was 24 months.Lung or bone metastases occurred in 1/22 borderline and 3/10 malignant PT patients 5 years post-surgery.The mitotic count and the degree of stromal cell atypia were significantly correlated with recurrence ( P =0.001 and P =0.006 ) . Multivariate analysis showed that severe stromal cell atypia was an independent predictor of recurrence -free survival in PT [HR=6.40 (95%CI=1.378 to 29.732), P=0.018].Conclusions Each parameter in the histological grading of PT may have different prognostic value , and markedly increased mitotic count and were predictive of relapse .