CIBER de Enfermedades Respiratorias (CIBERES), Madrid, Spain.
Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain.
Shock. 2018 Nov;50(5):504-510. doi: 10.1097/SHK.0000000000001099.
The integrated analysis of changes in the metabolic profile could be critical for the discovery of biomarkers of lung injury, and also for generating new pathophysiological hypotheses and designing novel therapeutic targets for the acute respiratory distress syndrome (ARDS). This study aimed at developing a nuclear magnetic resonance (NMR)-based approach for the identification of the metabolomic profile of ARDS in patients with H1N1 influenza virus pneumonia.
Serum samples from 30 patients (derivation set) diagnosed of H1N1 influenza virus pneumonia were analyzed by unsupervised principal component analysis to identify metabolic differences between patients with and without ARDS by NMR spectroscopy. A predictive model of partial least squares discriminant analysis (PLS-DA) was developed for the identification of ARDS. PLS-DA was trained with the derivation set and tested in another set of samples from 26 patients also diagnosed of H1N1 influenza virus pneumonia (validation set).
Decreased serum glucose, alanine, glutamine, methylhistidine and fatty acids concentrations, and elevated serum phenylalanine and methylguanidine concentrations, discriminated patients with ARDS versus patients without ARDS. PLS-DA model successfully identified the presence of ARDS in the validation set with a success rate of 92% (sensitivity 100% and specificity 91%). The classification functions showed a good correlation with the Sequential Organ Failure Assessment score (R = 0.74, P < 0.0001) and the PaO2/FiO2 ratio (R = 0.41, P = 0.03).
The serum metabolomic profile is sensitive and specific to identify ARDS in patients with H1N1 influenza A pneumonia. Future studies are needed to determine the role of NMR spectroscopy as a biomarker of ARDS.
代谢谱的综合分析对于发现肺损伤的生物标志物至关重要，也有助于产生新的病理生理学假说，并为急性呼吸窘迫综合征（ARDS）设计新的治疗靶点。本研究旨在开发一种基于核磁共振（NMR）的方法，以确定 H1N1 流感病毒肺炎患者 ARDS 的代谢组学特征。
对 30 例（推导组）诊断为 H1N1 流感病毒肺炎的患者血清样本进行无监督主成分分析，通过 NMR 光谱鉴定 ARDS 患者与非 ARDS 患者之间的代谢差异。建立偏最小二乘判别分析（PLS-DA）的预测模型，以识别 ARDS。PLS-DA 用推导组进行训练，并用另一组 26 例同样诊断为 H1N1 流感病毒肺炎的患者样本（验证组）进行测试。
血清葡萄糖、丙氨酸、谷氨酰胺、甲基组氨酸和脂肪酸浓度降低，苯丙氨酸和甲基胍浓度升高，可区分 ARDS 患者与非 ARDS 患者。PLS-DA 模型在验证组中成功识别 ARDS 的存在，成功率为 92%（灵敏度 100%，特异性 91%）。分类函数与序贯器官衰竭评估评分（R=0.74，P<0.0001）和 PaO2/FiO2 比值（R=0.41，P=0.03）具有良好的相关性。
血清代谢组学特征可敏感且特异地识别 H1N1 流感 A 肺炎患者的 ARDS。需要进一步研究来确定 NMR 光谱作为 ARDS 生物标志物的作用。
