Smell-led metabolomics analyses for rapid, early diagnosis and stratification of tuberculosis infection

2017年有100万儿童（<14岁）感染结核病（TB）[1]。虽然到2035年消除结核病传播的全球努力的重点是预防和治疗结核病，但为了降低疾病传播的风险，在早期阶段诊断儿童结核病感染非常重要。目前对儿童结核病的诊断测试包括对痰或胃吸出物进行细菌学测试（培养、分子测试、显微镜检查）。虽然痰液是非侵入性的，但儿童通常难以产生足够量的痰液。延误诊断将增加疾病传播给接触者的风险，尽管在发展中世界的许多高流行地区，甚至无法获得培养设备和专门知识。因此，需要一种非侵入性、快速且经济实惠的检测方法，能够在结核病发病早期准确检测出结核病。在我与曼彻斯特大学的Perdita Barran教授和坦桑尼亚非营利组织Apopo正在进行的一项试点研究中，超级嗅觉者Joy Milne描述了痰中与结核病相关的独特气味。乔伊已经成功地证明了生物液体的气味可以与一个人的疾病状态联系起来[3]。该项目将研究痰沿着呼吸和皮脂，以确定皮肤上结核病的早期诊断生物标志物。皮脂是用于疾病诊断的未开发的生物流体。我们假设皮脂是气味化合物的汇集处，这些气味化合物可能是许多疾病的标志，例如帕金森病[4]和结核病。这些以非侵入性方式获得的有气味的分子将是早期诊断测试的理想候选者，因为我们已经证明，在帕金森病等疾病出现临床症状之前，体味会发生变化[5]。将对疑似结核病、确诊结核病和健康参与者的样本进行纵向研究。使用色谱联用高分辨率质谱法，将沿着皮肤微生物组的定性分析生成痰液、呼吸和皮脂中的代谢组学和挥发物组学特征。先进的化学计量学和机器学习方法将用于建立数据驱动的模型，用于结核病的分类和预测，从而实现早期诊断。将微生物组和代谢组数据沿着患者信息、元数据和其他临床观察结果进行整合，以进行TB分层。

One million children (<14 yrs) contracted tuberculosis (TB) in 2017[1]. While the global efforts to eliminate the spread of TB by 2035 are focused on prevention and treatment of TB, it is important to diagnose TB infections in children, in very early stages in order to decrease the risk of spread of the disease. Current diagnostic tests for TB in children include bacteriological tests (culture, molecular tests, microscopy) on sputum or gastric aspirates. While sputum is non-invasively available, children often have difficulties in producing sufficient quantity of sputum. Delay in diagnosis will increase the risk of spread of disease to contacts, although in many high-prevalence areas in the developing world, access to equipment and expertise for culture is not even available. Thus, a non-invasive, rapid and affordable test that can accurately detect TB early in its onset is required.In an ongoing pilot study with my collaborators Prof Perdita Barran (The University of Manchester) and Apopo, a non-profit in Tanzania, Joy Milne, a super smeller [2] has described a unique smell associated with TB in sputum. Joy has successfully demonstrated that smell of biofluids can be linked to one's disease state [3]. This project will investigate sputum along with breath and sebum to identify early diagnostic biomarkers of TB on skin. Sebum is an unexplored bio-fluid for disease diagnostics. We have hypothesised sebum as a sink to odourous compounds that can be a hallmark of many diseases such as Parkinson's disease [4] and also TB. These odorous molecules obtained non-invasively will be ideal candidates for early diagnostic test since we have demonstrated that body odour changes before presentation of clinical symptoms in disease such as Parkinson's [5]. A longitudinal study will be performed on samples from adults with suspected TB, confirmed TB and healthy participants. Using chromatography hyphenated to high resolution mass spectrometry, metabolomics and volatilomics profiles in sputum, breath and sebum will be generated along with qualitative analysis of skin microbiome. Advanced chemometrics and machine learning approaches will be employed to build data driven models for classification and prediction of TB leading to its early diagnosis. Integration of microbiome and metabolome data along with patient information, metadata and other clinical observations will be performed for stratification of TB.