浆液性卵巢癌预后评分模型的建立和预后相关基因的筛选

目的寻找与卵巢癌预后相关的临床因素和基因。方法用数学统计方法对104例浆液性卵巢癌患者进行生存分析，筛选预后相关的临床因素，建立卵巢癌预后评分模型。通过对22例浆液性卵巢癌基因表达谱芯片进行生物信息学分析，筛选卵巢癌预后相关的基因群，并进行基因调控网络分析。结果根据对预后的影响程度，卵巢癌预后相关的临床因素依次包括手术病理分期、化疗，术后残余灶，组织学分级和淋巴结转移；根据各因素及其内部分层因素对预后影响的风险系数，对各因素及其内部因素进行量化，结合生存函数对患者生存概率的估计，建立了卵巢癌预后评分模型。通过生物信息学分析，筛选得到卵巢癌预后相关的基因群，共236个基因。筛选得到调控基因数超过20个的基因共112个。结论通过该模型可以用直观的数字反映各因素对预后的影响程度和用患者各项因素总评分来判断患者的3年和5年生存概率，使综合判断患者预后的工作得以简化，对于临床工作有一定意义。利用生物学信息学分析卵巢癌基因表达谱芯片，为未来医学研究提供了从基因组水平进行研究的思路和方法学上的摸索。

Construction of prognostic scoring model and screening the prognosis associated genes of ovarian serous adenocarcinoma

Objective To explore the clinical and molecular factors associated with prognosis of ovarian cancer. Methods We retrospectively analyzed 104 cases with ovarian serous adenocarcinoma and screened the prognosis associated clinical factors by statistical methods, and constructed a prognostic scoring model. We analyzed the gene expression profiles of 22 patients with ovarian cancer by cDNA microarray and screened the prognosis related genes by bioinformatics, and analysed the relationship between every two genes, according to the principle of gene network. Results In order of the influence on prognosis, the clinical prognostic factors of ovarian serous adenocarcinoma were FIGO stage, the chemotherapy regiments, histological grade, metastasis of lymph node and residual disease after primary surgery. The hazard coefficient of each variable was analyzed by Cox multivariable analysis and transformed to scores according to its effect on prognosis. According to the survival probability from Proportional Hazard Model, a prognostic model by linking total score with survival probability was set up. We selected the prognostic associated gene clusters by bioinformtics analysis and got 236 genes in total, related to FIGO stage, histological grade, metastasis of lymph node, residual disease and the reactivity to chemotherapy respectively, 112 genes of which controled more than 20 other genes. Conclusions Using this model, the 3 - year and 5 - year survival probability of ovarian cancer patients could be estimated with a score, integrating all the factors. Analyzing the gene expression profile of ovarian serous adenocarcinoma by bioinformatics is a beneficial exploration on the genomic level for the future medical study.