Dawany Noor, Showe Louise C, Kossenkov Andrew V, Chang Celia, Ive Prudence, Conradie Francesca, Stevens Wendy, Sanne Ian, Azzoni Livio, Montaner Luis J
Center for Systems and Computational Biology, The Wistar Institute, Philadelphia, Pennsylvania, United States of America.
Genomics Facility, The Wistar Institute, Philadelphia, Pennsylvania, United States of America.
PLoS One. 2014 Feb 25;9(2):e89925. doi: 10.1371/journal.pone.0089925. eCollection 2014.
Co-infection with tuberculosis (TB) is the leading cause of death in HIV-infected individuals. However, diagnosis of TB, especially in the presence of an HIV co-infection, can be limiting due to the high inaccuracy associated with the use of conventional diagnostic methods. Here we report a gene signature that can identify a tuberculosis infection in patients co-infected with HIV as well as in the absence of HIV.
We analyzed global gene expression data from peripheral blood mononuclear cell (PBMC) samples of patients that were either mono-infected with HIV or co-infected with HIV/TB and used support vector machines to identify a gene signature that can distinguish between the two classes. We then validated our results using publically available gene expression data from patients mono-infected with TB.
Our analysis successfully identified a 251-gene signature that accurately distinguishes patients co-infected with HIV/TB from those infected with HIV only, with an overall accuracy of 81.4% (sensitivity = 76.2%, specificity = 86.4%). Furthermore, we show that our 251-gene signature can also accurately distinguish patients with active TB in the absence of an HIV infection from both patients with a latent TB infection and healthy controls (88.9-94.7% accuracy; 69.2-90% sensitivity and 90.3-100% specificity). We also demonstrate that the expression levels of the 251-gene signature diminish as a correlate of the length of TB treatment.
A 251-gene signature is described to (a) detect TB in the presence or absence of an HIV co-infection, and (b) assess response to treatment following anti-TB therapy.
结核病(TB)合并感染是HIV感染者的主要死因。然而,由于传统诊断方法存在高度不准确性,结核病的诊断,尤其是在合并HIV感染的情况下,可能会受到限制。在此,我们报告一种基因特征,它可以识别合并HIV感染的患者以及未感染HIV患者中的结核感染。
我们分析了HIV单感染或HIV/TB合并感染患者外周血单个核细胞(PBMC)样本的全局基因表达数据,并使用支持向量机来识别能够区分这两类患者的基因特征。然后,我们使用来自TB单感染患者的公开可用基因表达数据验证了我们的结果。
我们的分析成功识别出一个由251个基因组成的特征,该特征能够准确区分HIV/TB合并感染患者与仅感染HIV的患者,总体准确率为81.4%(敏感性=76.2%,特异性=86.4%)。此外,我们表明,我们的251个基因特征还能够准确区分无HIV感染的活动性TB患者与潜伏性TB感染患者和健康对照(准确率为88.9 - 94.7%;敏感性为69.2 - 90%,特异性为90.3 - 100%)。我们还证明,251个基因特征的表达水平会随着TB治疗时间的延长而降低。
描述了一种251个基因的特征,用于(a)在有或无HIV合并感染的情况下检测TB,以及(b)评估抗结核治疗后的治疗反应。