Gene expression profiling in association with interferon-alpha-sensitivity for the personalized treatment of renal cell carcinoma

背景资料。我们分析干扰素-α应答与基因表达谱的相关性，以预测干扰素-α的敏感性，并找出调节肾细胞癌干扰素-α应答的关键分子。采用3,840个克隆芯片进行基因表达谱分析，筛选干扰素α应答相关基因，根据干扰素α应答水平对肾癌患者进行评估。用四甲基偶氮唑盐比色法检测细胞对干扰素α的反应，加入300~10000IU/ml干扰素α。根据干扰素α的敏感性，对筛选出的基因进行基因芯片，然后进行有监督的层次聚类分析。为了解干扰素α诱导肾癌细胞基因表达的变化，采用基因芯片技术检测干扰素α作用后3、6、12h肾癌细胞的基因表达水平。根据基因表达谱与干扰素α敏感性的统计相关性，建立干扰素α应答的预测模型。根据14例肾癌患者的基因芯片分析，干扰素α应答者可能由所选择的基因集来识别。基因芯片通过对干扰素α敏感株和干扰素α耐药株的聚类分析，最终筛选出7个基因：ADFP、MITF、MTUS1、Pme-1、MLLT3、MLF1和TNNT1作为肾细胞癌α敏感相关基因的候选基因。我们进一步利用多元线性回归分析(系数=0.948，p=0.0291)建立了ADFP、MITF、MTUS1和TNNT1四种分子对肿瘤抑制作用的预测模型，并用原代培养的肾癌细胞对模型进行了验证。联合选择的基因的表达水平可以为肾细胞癌的干扰素α反应提供预测性信息。此外，干扰素对肾癌的α反应可能是通过调节这些分子的表达水平来调节的。

Background. We analyzed the correlation between interferon-α (IFNα) response and gene expression profiles to predict IFNα sensitivity and identified key molecules regulating the IFNα response in renal cell carcinoma (RCC).Methods. To evaluate patients with RCC according to IFNα-response, gene expression profiling was done using 3,840 clones microarray, and then IFNα-responder related genes were selected by clustering analysis. Basically, IFNα-response of RCC cell lines was evaluated by MTT assay with 300 to 10,000 IU/ml of IFNα treatment. Microarray, followed by supervised hierarchical clustering analysis, was applied to selected genes according to IFNα sensitivity. In order to find alteration of expression profiles induced by IFNα, sequential microarray analyses were performed at 3,6, and 12 hrs after IFNα treatment of RCC cell lines and mRNA expression level was confirmed using quantitative RT-PCR. A model for prediction of IFNα response was created according to the statistical correlation between gene expression profiles and IFNα-sensitivity.Results. According to microarray analysis in 14 patients with RCC,IFNα-responder might be identified by the selected gene sets. Microarray followed clustering analysis between IFNα-sensitive and IFNα-resistant line, seven genes, ADFP,MITF,MTUS1,PME-1,MLLT3,MLF1, and TNNT1, were eventually selected as candidates for IFNα-sensitivity-related genes in RCC cell lines. We further developed a model to predict tumor-inhibition with four molecules, i.e., ADFP,MITF,MTUS1, and TNNT1, using multiple linear regression analysis (coefficient=0.948,p=0.0291) and validated the model using primary cultured RCC cells.Conclusions. The expression levels of the combined selected genes can provide predictive information on the IFNα response in RCC. Furthermore, the IFNα-response to RCC might be modulated by regulation of the expression level of these molecules.