Proteomics Unit, IMIBIC, Hospital Universitario Reina Sofía, Universidad de Córdoba, Cordoba, Spain.
Universidad de Córdoba, Córdoba, Spain.
J Proteomics. 2018 Mar 20;175:105-113. doi: 10.1016/j.jprot.2018.01.003. Epub 2018 Jan 6.
A typical clinical manifestation of growth hormone deficiency (GHD) is a short stature resulting from delayed growth, but GHD affects bone health, cardiovascular function and metabolic profile and therefore quality of life. Although early GH treatment during childhood has been shown to improve outcomes, no single biochemical parameter is currently available for the accurate diagnosis of GHD in children. There is hence a need for non-invasive biomarkers. In this study, the relative abundance of serum proteins from GHD children and healthy controls was measured by next-generation proteomics SWATH-MS technology. The data generated was analysed by machine-learning feature-selection algorithms in order to discover the minimum number of protein biomarkers that best discriminate between both groups. The analysis of serum proteins by a SWATH-MS approach yielded a useful method for discovering potential biomarkers of GHD in children. A total of 263 proteins were confidently detected and quantified in each sample. Pathway analysis indicated an effect on tissue/organ structure and morphogenesis. The top ten serum protein biomarker candidates were identified after applying feature-selection data analysis. The combination of three proteins - apolipoprotein A-IV, complement factor H-related protein 4 and platelet basic protein - demonstrated the best classification performance for our data. In addition, the apolipoprotein group resulted in strong over-representation, thus highlighting these proteins as an additional promising biomarker panel.
Currently there is no single biochemical parameter available for the accurate diagnosis of growth hormone (GH) deficiency (GHD) in children. Simple GH measurements are not an option: because GH is released in a pulsatile action, its blood levels fluctuate throughout the day and remain nearly undetectable for most of that time. This makes measurements of GH in a single blood sample useless for assessing GH deficiency. Actually, the diagnosis of GHD includes a combination of direct and indirect non-accurate measurements, such as taking several body measurements, testing GH levels in multiple blood samples after provocative tests (GH peak <7.3ng/mL, using radioimmunoassay), and conducting magnetic resonance imaging (MRI), among others. Therefore, there is a need for simple, non-invasive, accurate and cost-effective biomarkers. Here we report a case-control study, where relative abundance of serum proteins were measured by next-generation proteomics SWATH-MS technology in 15 GHD children and 15healthy controls matched by age, sex, and not receiving any treatment. Data generated was analysed by machine learning feature selection algorithms. 263 proteins could be confidently detected and quantified on each sample. The top 10 serum protein biomarker candidates could be identified after applying a feature selection data analysis. The combination of three proteins, apolipoprotein A-IV, complement factor H-related protein 4 and platelet basic protein, showed the best classification performance for our data. In addition, the fact that the pathway and GO analysis we performed pointed to the apolipoproteins as over-represented highlights this protein group as an additional promising biomarker panel for the diagnosis of GHD and for treatment evaluation.
生长激素缺乏症（GHD）的典型临床表现为生长迟缓导致的身材矮小，但 GHD 还会影响骨骼健康、心血管功能和代谢特征，从而影响生活质量。虽然儿童时期早期的 GH 治疗已被证明可以改善结局，但目前尚无用于准确诊断儿童 GHD 的单一生化参数。因此，需要非侵入性的生物标志物。在本研究中，使用下一代蛋白质组学 SWATH-MS 技术测量了 GHD 儿童和健康对照组的血清蛋白相对丰度。通过机器学习特征选择算法分析生成的数据，以发现最佳区分两组的最小数量的蛋白生物标志物。SWATH-MS 方法分析血清蛋白可提供一种有用的方法来发现儿童 GHD 的潜在生物标志物。在每个样本中均可靠地检测到并定量了 263 种蛋白。途径分析表明对组织/器官结构和形态发生有影响。应用特征选择数据分析后，确定了 10 种血清蛋白生物标志物候选物。三种蛋白（载脂蛋白 A-IV、补体因子 H 相关蛋白 4 和血小板碱性蛋白）的组合对我们的数据具有最佳的分类性能。此外，载脂蛋白组表现出强烈的过度表达，这突出表明这些蛋白是另一个有前途的生物标志物组合。
目前，尚无用于准确诊断儿童生长激素（GH）缺乏症（GHD）的单一生化参数。简单的 GH 测量不是一种选择：因为 GH 呈脉冲式释放，其血液水平在一天中波动，并且在大部分时间内几乎无法检测到。这使得单次血液样本中的 GH 测量对评估 GH 缺乏症毫无用处。实际上，GHD 的诊断包括直接和间接的非准确测量的组合，例如进行多次身体测量，在激发试验后在多个血液样本中测试 GH 水平（GH 峰值<7.3ng/mL，使用放射免疫测定法），并进行磁共振成像（MRI）等。因此，需要简单、非侵入性、准确且具有成本效益的生物标志物。在这里，我们报告了一项病例对照研究，其中使用下一代蛋白质组学 SWATH-MS 技术在 15 名 GHD 儿童和 15 名健康对照组中测量了血清蛋白的相对丰度，这些对照组通过年龄、性别和不接受任何治疗进行匹配。生成的数据通过机器学习特征选择算法进行分析。在每个样本上都可以可靠地检测到并定量 263 种蛋白。应用特征选择数据分析后，可以确定 10 种血清蛋白生物标志物候选物。三种蛋白（载脂蛋白 A-IV、补体因子 H 相关蛋白 4 和血小板碱性蛋白）的组合对我们的数据具有最佳的分类性能。此外，我们进行的途径和 GO 分析表明，载脂蛋白过度表达，这突出表明该蛋白组是诊断 GHD 和评估治疗效果的另一个有前途的生物标志物组合。
