基於親水作用液相層析質譜儀的代謝體研究 - 化學計算方法開發與腹膜透析引流液分析

Abstract

代謝體學是經由全面性的分析生物體樣品內代謝體的種類與含量，表達出生物系統的表現型，有助於了解個體的差異與疾病的機轉。代謝體學的資料主要是來自核磁共振質譜儀或液相層析串聯質譜儀 (LC-MS) 產生的高通透性大量資料，需靠種種的生物資訊工具來分析。本論文針對代謝體學的研究介紹了資料處理的步驟與多變數分析的應用。 代謝體學所研究的生物體樣品常來自體液如血漿及尿液，而腹膜透析治療尿毒症時產生的腹膜透析引流液也是其中一種。生物體液內含有許多小分子的代謝體，且多半是具有分子極性，在眾多的可分離代謝體的液相層析柱中，親水作用層析柱 (HILIC) 可互補傳統上常用但較傾向於分離非極性分子的碳18逆向 層析柱在分離極性代謝體的不足。經由簡單步驟的電腦算法，我們針對親水性層析柱在四種不同的移動相組成下比較了71種極性標準品得到的檢測資料，以多種演算法經電腦快速比較各個標準品的萃取離子層析圖 (XIC) 出現的波峰訊號開發了適合代謝體學研究的最佳化移動相。也就是從各種移動相下測得的標準品數量與皆有測得的20種標準品的波形對稱值比較後以評分來決定最佳化移動相組成。然後，基於HILIC LC-MS代謝體分析平台我們開發了代謝體參考標準品資料庫。我們選用了571種標準品，層析部分是不同於之前使用的等位流析而改採梯度流析方式進行。而經由 HILIC LC-MS 平台檢測產生出的大量數據我們藉由設計出的自動化化學計量工作流程的步驟，以電腦算法過濾篩選標準品測定資料，達到自動篩選並完成測定出的標準品參考滯留時間的數據庫建立，這資料庫共獲得291個氫化陽離子與241個去氫化陰離子標準品的比對數據。為驗證參考滯留時間的應用性，我們將測得波峰的標準品以高濃度和低濃度添加入 人體血漿並以自動化工作流程偵測在HILIC LC-MS平台下的偵出與量度，結果有很高的偵測率 ，計有286個在正 電壓模式與235個在負電壓模式的血漿內添加的標準品訊號強度以電腦工作流程可測得。結論是以電腦演算代謝體學分析的大量資料可有效比對出血漿內的目標代謝體。 腹膜透析引流液的代謝體學研究目的是分析臨床上不同腹膜透析效率分類組別間的代謝體差異，接受腹膜透析治療之20名病人，從常規檢體中採集腹膜透析引流液，依臨床上腹膜對肌酸酐的通透率(腹膜平衡試驗)分成了高、平均高、平均低與低四組每組5名，收集了留置2小時與4小時的腹膜透析引流液，以HILIC LC-MS收集代謝體的質譜儀資料，再以PITracer的R統計軟體套件比對我們的HILIC LC-MS代謝體的荷質比與滯留時間資料庫找出有出現的代謝體並將波峰資料做表以多變量分析比較四組的差異。經由非監督法的主成分(PCA)分析後我們發現高與低通透率組有群組差異，接著再以偏最小二乘法判别分析 (PLS-DA) 的變量重要性指標 (VIP) 找出高與低通透率兩組間的濃度有統計差異的代謝體，結果發現一種可能與腹膜透析包囊性腹膜硬化症有正相關的代謝體 Hippurate 濃度在高通透組中呈現偏高，另外也發現低通透組的四小時引流液內的前列腺素E2 (PGE2) 濃度反較高通透組高，可能是因為低通透組腹腔內仍有較高濃度的糖分導致腹膜細胞分泌PGE2的現象。其他有濃度差異的代謝體也包括了幾種胺基酸，胺基酸衍生物及醣類，因此最後再以代謝體集富集分析 (MSEA) 找出了兩組別有關醣類與胺基酸等代謝途徑上的統計差異。由於臨床上高通透組的預後一般要比低通透組較差，我們藉由代謝體學的分析提供了相關的代謝體標誌與相關差異的代謝途徑。

Through holistic metabolite profiling and quantitation in biosystems, metabolomics conveys phenotype information and provides understanding difference among living systems and the mechanisms of diseases. Metabolomic datasets come from state-of-art high-throughput chemical analytical instruments such as nuclear magnetic resonance (NMR) or liquid chromatography hyphenated to mass spectrometry. The data size is huge and requires bioinformatic tools to facilitate analysis. We introduced for the study of metabolomics the procedures of data management and pertinent multivariate analysis. Biosamples used in metabolomics study may come from biofluids such as plasma, urine or others, such as peritoneal effluent disposed from peritoneal dialysis. Hydrophilic interaction liquid chromatography (HILIC) hyphenated to mass spectrometry (LC-MS) is widely used in the study of metabolomics, particularly for polar metabolites. However, determining an optimized mobile phase and developing a proper HILIC method tend to be laborious, time-consuming, and experience-dependent tasks. In this study, we developed a chemometric workflow to quickly determine the optimized mobile phase and to objectively construct a HILIC LC-MS reference library database. All chromatograms in the process were baseline corrected, smoothed and noise filtered. A mass chromatographic quality value, an asymmetric factor, and local maxima of the extracted ion chromatogram were calculated to determine the number of peaks and peak retention time. The optimal mobile phase can be quickly determined by selecting the mobile phase that produces the largest number of resolved peaks. Moreover, the workflow enables one to automatically process the repeats and to determine the retention time of large numbers of standards. This method was successfully applied to construct a reference library of 571 metabolites for HILIC LC-MS platform based metabolomics study . We used metabolomic approach to study peritoneal effluent with HILIC LC-MS platform. The function of peritoneal membrane is critical to uremic patients receiving peritoneal dialysis (PD). Clinically, the result of peritoneal equilibration test (PET) defines peritoneal transport status. Metabolite profiles in peritoneal effluents can imply dialytic efficiency of peritoneal dialysis and elucidate differences among peritoneal transport status. We collected peritoneal effluents from 20 PD patients whose PET were evenly distributed among four peritoneal transport status groups (low, low average, high average and high). We analyze metabolomic differences in PD effluents among PET groups of peritoneal transport status. HILIC LC-MS was implemented for metabolite detection. With data processing, PITracer peak detection algorithm and multivariate analysis we successfully recognized metabolite profiling patterns between PET defined high and low transport groups. The difference can be detected at 2 hour and 4 hour dialysate dwell by PCA and PLS-DA. Using variable important projection (VIP) values from PLS-DA, we found several metabolites significantly differentiated between Low and High transport groups, at 2 and/or at 4 hour dwell. Hippurate concentration was higher in High transport groups which had been suspected a risk metabolites in peritoneal effluent for encapsulating peritoneal sclerosis, a severe peritoneal dialysis complication. Prostaglandin E2, a peritoneal cellular reactive substance to high osmolar dialysate was found in higher concentration to Low transport group. A possible response biomarker indicating the higher osmolar gap from related lower flux of dialysate sugar. To find related metabolic pathways distinguished between Low and High transport group. Metabolic pathways related to amino acid, sugar and fatty acid were found. As the prognosis of peritoneal dialysis is deemed poorer in High transporter than in Low transporter. We provided possible biomarkers and involved metabolic pathways from metabolomics study to peritoneal creatinine equilibration defined clinical transport status. In conclusion, metabolomics study for clinically easily feasible peritoneal dialysis effluents provides promising study for peritoneal membrane pathophysiological changes.