Ando Tatsuya, Suguro Miyuki, Hanai Taizo, Kobayashi Takeshi, Honda Hiroyuki, Seto Masao
Department of Biotechnology, School of Engineering, Nagoya University, Chikusa-ku, Nagoya 464-8603, Japan.
Jpn J Cancer Res. 2002 Nov;93(11):1207-12. doi: 10.1111/j.1349-7006.2002.tb01225.x.
Diffuse large B-cell lymphoma (DLBCL) is the largest category of aggressive lymphomas. Less than 50% of patients can be cured by combination chemotherapy. Microarray technologies have recently shown that the response to chemotherapy reflects the molecular heterogeneity in DLBCL. On the basis of published microarray data, we attempted to develop a long-overdue method for the precise and simple prediction of survival of DLBCL patients. We developed a fuzzy neural network (FNN) model to analyze gene expression profiling data for DLBCL. From data on 5857 genes, this model identified four genes (CD10, AA807551, AA805611 and IRF-4) that could be used to predict prognosis with 93% accuracy. FNNs are powerful tools for extracting significant biological markers affecting prognosis, and are applicable to various kinds of expression profiling data for any malignancy.
弥漫性大B细胞淋巴瘤（DLBCL）是侵袭性淋巴瘤中最大的类别。不到50%的患者可通过联合化疗治愈。微阵列技术最近表明，对化疗的反应反映了DLBCL中的分子异质性。基于已发表的微阵列数据，我们试图开发一种早就该有的方法，用于精确且简单地预测DLBCL患者的生存期。我们开发了一个模糊神经网络（FNN）模型来分析DLBCL的基因表达谱数据。从5857个基因的数据中，该模型识别出四个基因（CD10、AA807551、AA805611和IRF-4），它们可用于以93%的准确率预测预后。模糊神经网络是提取影响预后的重要生物标志物的强大工具，适用于任何恶性肿瘤的各种表达谱数据。
