Breslow N E, Palmer N F, Hill L R, Buring J, D'Angio G J
Cancer. 1978 Apr;41(4):1577-89. doi: 10.1002/1097-0142(197804)41:4<1577::aid-cncr2820410448>3.0.co;2-7.
Multivariate statistical analyses are used to evaluate a wide range of factors in predicting relapse and survival for 429 children enrolled in the National Wilms' Tumor Study. Anaplastic or sarcomatous histology, specimen weight over 250 grams, positive regional lymph nodes, treatment with only a single drug and age over two years are the most important predictors of relapse. The first three factors also predict mortality. Laterality, capsular penetration, intrarenal vascular invasion, direct regional extension and operative spillage have lesser effects which do not contribute significantly to the multivariate prediction equations, while sex, race and the presence of a tumor thrombus in the renal vein have essentially no effect. Treatment with combination chemotherapy is confirmed to be efficacious regardless of what prognostic factors are operating. A statistical model allows prediction of relapse and death for each of 24 patient subgroups, with the estimates of percentage relapsed ranging from 3% to 100%.
多变量统计分析用于评估参加国家肾母细胞瘤研究的429名儿童复发和生存预测中的广泛因素。间变性或肉瘤样组织学、标本重量超过250克、区域淋巴结阳性、仅用单一药物治疗以及年龄超过两岁是复发的最重要预测因素。前三个因素也可预测死亡率。侧别、包膜侵犯、肾内血管侵犯、直接区域扩展和手术溢出的影响较小，对多变量预测方程没有显著贡献，而性别、种族和肾静脉内肿瘤血栓的存在基本没有影响。无论存在何种预后因素，联合化疗被证实是有效的。一个统计模型可以预测24个患者亚组中每组的复发和死亡情况，复发百分比估计范围为3%至100%。
