Centre for International Child Health, Department of Paediatrics, Imperial College London, St. Mary's Hospital, Praed Street, London, W2 1NY, United Kingdom.
Division of Computational and Systems Medicine, Department of Surgery and Cancer, Imperial College London, Sir Alexander Fleming Building, South Kensington, London, SW7 2AZ, United Kingdom.
Sci Rep. 2020 Apr 29;10(1):7302. doi: 10.1038/s41598-020-64413-6.
We applied a metabonomic strategy to identify host biomarkers in serum to diagnose paediatric tuberculosis (TB) disease. 112 symptomatic children with presumptive TB were recruited in The Gambia and classified as bacteriologically-confirmed TB, clinically diagnosed TB, or other diseases. Sera were analysed using H nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS). Multivariate data analysis was used to distinguish patients with TB from other diseases. Diagnostic accuracy was evaluated using Receiver Operating Characteristic (ROC) curves. Model performance was tested in a validation cohort of 36 children from the UK. Data acquired using H NMR demonstrated a sensitivity, specificity and Area Under the Curve (AUC) of 69% (95% confidence interval [CI], 56-73%), 83% (95% CI, 73-93%), and 0.78 respectively, and correctly classified 20% of the validation cohort from the UK. The most discriminatory MS data showed a sensitivity of 67% (95% CI, 60-71%), specificity of 86% (95% CI, 75-93%) and an AUC of 0.78, correctly classifying 83% of the validation cohort. Amongst children with presumptive TB, metabolic profiling of sera distinguished bacteriologically-confirmed and clinical TB from other diseases. This novel approach yielded a diagnostic performance for paediatric TB comparable to that of Xpert MTB/RIF and interferon gamma release assays.
我们应用代谢组学策略来鉴定血清中的宿主生物标志物，以诊断小儿结核病（TB）。在冈比亚招募了 112 名有疑似结核病症状的儿童，并将其分为细菌学确诊的结核病、临床诊断的结核病或其他疾病。使用 H 核磁共振（NMR）光谱和质谱（MS）分析血清。使用多元数据分析来区分 TB 患者和其他疾病。使用Receiver Operating Characteristic (ROC) 曲线评估诊断准确性。在来自英国的 36 名儿童的验证队列中测试模型性能。使用 H NMR 获得的数据的灵敏度、特异性和曲线下面积（AUC）分别为 69%（95%置信区间[CI]，56-73%）、83%（95% CI，73-93%）和 0.78，正确分类了 20%的来自英国的验证队列。最具区分力的 MS 数据的灵敏度为 67%（95% CI，60-71%），特异性为 86%（95% CI，75-93%），AUC 为 0.78，正确分类了 83%的验证队列。在疑似结核病的儿童中，血清代谢组学分析可区分细菌学确诊的结核病和临床结核病与其他疾病。这种新方法在诊断小儿结核病方面的表现与 Xpert MTB/RIF 和干扰素γ释放试验相当。
