利用全外显子测序检测复发性前列腺癌全新突变并构建前列腺癌复发预测模型

Whole exome sequencing was used to identify novel mutations in recurrent prostate cancer and to construct a prediction model for recurrence of prostate cancer

目的利用全外显子测序检测复发性前列腺癌全新突变,并构建前列腺癌复发预测模型.方法利用外显子测序分析了30例复发的和44例未复发的前列腺癌患者,测序的平均覆盖率和平均深度分别达99.4%和44.8X,并利用随机森林模型,建立了一个包含5个突变的预测前列腺癌复发的模型.应用随机森林包的函数rfcv(Random Forest Cross-Validation for Feature Selection,随机森林交叉验证特征选择)(www.rdocumentation.org/packages/randomForest/versions/4.6-14/topics/rfcv)来显示模型的交叉验证预测性能,通过嵌套交叉验证顺序减少预测变量的数量(按变量重要性排序)程序.结果鉴定了复发前列腺癌患者的72个特异性体细胞单核苷酸变异(SNVs),发现了17个有害的和10个保护性的种系(SNVs).前列腺癌复发模型的敏感性和特异性分别为83.3%和88.9%.结论本研究可以为独立的更大临床样本研究验证前列腺癌预后和复发预测的标志物群提供基础.

Objective Prostate cancer (PCa) is a top common cancer among men with great differences between prostate cancer that progresses rapidly and recurs and that progresses very slowly and causes little harm during the lifespan of the individual. A robust diagnostics that can predict disease recurrence after initial treatment, for example, with radical prostatectomy, is the key for better management of PCa patients. Although we made some progresses in recent years with the discovery of the androgen receptor gene (AR) splice events, the TMPRSS2: EGR gene fusion, long noncoding RNA MALAT-1 and SCHLAP1 for PCa prognosis, but there still lack good prognostic markers that can stratify PCa into those recur early and those do not. Methods We conducted a whole exome sequencing of 30 recurrent and 44 nonrecurrent PCa patients with an average sequencing coverage and depth of 99.4% and 44.8x, respectively. By random forest model, we established a model using the 5 most informative SNVs that were able to predict recurrence of prostate cancer. Results We identified 72 specific somatic single nucleotide variations (SNVs) in the recurrent PCa and further identified 17 harmful and 10 protective germline SNVs. The prostate cancer recurrence model could predict the recurrence of prostate cancer with a sensitivity and a specificity of 83.3% and 88.9% respectively. Conclusion Our findings warrant further research with independent and larger clinical samples so as to confirm the signature for PCa prognosis and recurrence prediction.