透過您的圖書館登入
IP:44.222.249.19
  • 學位論文

以孟德爾隨機分派試驗法探討鎂營養與心血管代謝疾病及失智症之因果關係

Mendelian Randomization Studies to Investigate The Causal Role of Magnesium Status on Cardio-metabolic Diseases and Dementia

指導教授 : 潘文涵
本文將於2025/08/11開放下載。若您希望在開放下載時收到通知,可將文章加入收藏

摘要


背景: 鎂是人體細胞內第二多的陽離子、參與超過600個酵素反應。鎂狀態也在體內扮演多個重要角色,包含神經傳導、肌肉收縮、血壓調節、血糖代謝等。過去的動物實驗、流行病學研究指出較高的鎂狀態對心血管代謝疾病、失智症等都有保護作用,然而其因果關係仍有待證實。 材料與方法: 我們使用英國人體生物資料庫的基因與臨床診斷資料執行孟德爾隨機分派研究,並從過往全基因體關聯分析(genome-wide association studies)找到影響鎂濃度之單核苷酸多型性(Single Nucleotide Polymorphism)作為工具變數代表鎂營養狀態,建立一個遺傳鎂傾向分數,由ICD-10診斷紀錄、HbA1c測量數值定義失智症、第二型糖尿病、缺血性心臟病、腦血管疾病四個疾病組別。本研究使用通過品質管理檢查的269169筆英國白人數據。 我們將所有樣本根據基因風險分數分成三組(T1, T2, T3),並針對四個疾病組別使用Cox比例風險模型(Cox proportional hazard model)進行存活分析,我們先將資料隨機分成三子群、重複三次發現-驗證(2:1)之分析流程,並將T1(0), T2(1), T3(2)視為連續變數並求得p for trend。針對失智症的分析進一步將干擾因子放入模型校正。此外,我們也將T1視為參考組,評估T2和T3得病風險,並將失智症分類後再作進一步分析。 結果: 將資料隨機分三子群並重複三次發現-驗證(2:1)之分析流程,發現鎂對失智症的保護趨勢,但是並沒有在每個子群中都顯著。然而則未觀察到遺傳鎂傾向分數和第二型糖尿病、缺血性心臟病及腦血管疾病有顯著關聯。我們發現較高的鎂狀態對失智症有保護作用,每增加1三分位單位,風險平均下降10% (HR=0.903, 95%CI: 0.839-0.973, p=0.007),且這樣的因果關係在控制了一些潛在干擾因子後更加顯著(HR=0.882, 95%CI: 0.815-0.955, p=0.0019)。失智罹患風險在遺傳鎂傾向分數在第2三分位和第3三分位分別比第1三分位低了15%和18% (T2: HR=0.853, 95%CI: 0.739-0.985, p=0.0306; T3: HR=0.82, 95%CI: 0.708-0.951, p=0.0084)。 結論: 我們觀察到遺傳預測的血清鎂較高者失智症發生之風險較低,然而三分驗證的結果可能由於工具變數解釋度偏低而稍不穩定,未來須進一步修正鎂營養狀態之工具變數,同時在不同群體、種族做研究以確認鎂對失智症之保護作用。

並列摘要


Background: Magnesium is the second most abundant intracellular cation in human body, which involves in over six hundred enzymatic reactions. Magnesium plays critical roles in human body, involving neurotransmission, muscular contraction, blood pressure regulation, and glucose metabolism, just to name a few. Previous epidemiological studies have shown inverse associations of dietary and/or serum magnesium with cardio-metabolic diseases and neurologic disorders. However, it is not clear whether these associations are causal. Materials and Methods: We utilized genetic and clinical data from UK Biobank to conduct a Mendelian randomization study, in which genetic variants discovered from previous GWAS were used to construct genetic propensity score (GPS) as an instrumental variable of magnesium status. Dementia, type 2 diabetes, coronary artery disease, and cerebrovascular accident were defined by their ICD-10 diagnosis records as well as baseline HbA1c values. A total of 269,129 Caucasian data which passed the quality control check were included in the statistical analyses. Discovery-validation procedure was conducted by randomly splitting the data to three parts and checked if the relationship remains in the 1st part and in the 2nd and 3rd parts combined in 3 different ways. We divided the entire studied samples into 3 tertiles (T1, T2, T3) according to their GPSs. We then conducted survival analysis with Cox proportional hazard model on each of the 4 disease groups, treating T1(0), T2(1), T3(2) grouping as continuous variable and obtained p-value for trend with or without adjusting potential confounders. For dementia disease group, we further treated T1 as the reference group and estimated the hazard ratios of T2/T1 and T3/T1, respectively. We also analyzed dementia subgroups as endpoints. Results: We discovered a trend of protective effect on dementia, although not always significant when discovery-validation procedure was carried out with three random splitting data. However, no significant associations were found between GPS and the other three disease groups. We found higher level of magnesium-GPS was inversely associated with a lower risk of developing dementia. The hazard decreased by 10% when shifting up 1 tertile of GPS (HR=0.903, 95%CI: 0.839-0.973, p=0.007). This effect become more significant when potential confounders were adjusted in the model (HR=0.882, 95%CI: 0.815-0.955, p=0.0019). Also, the hazard is 15% and 18% lower in T2 and in T3 respectively, compared to T1. (T2: HR=0.853, 95%CI: 0.739-0.985, p=0.0306; T3: HR=0.82, 95%CI: 0.708-0.951, p=0.0084). Conclusion: We observed lower risk of dementia development when genetically determined serum magnesium level increases. However, the validation result of three random splitting data was unstable. This is probably because the instrumental variables only explained a very small percentage of the serum magnesium variance. Improving instrumental variables for magnesium nutrition is crucial for future research. It is warranted to carry out further confirmation studies in different populations or ethnic groups on the causal relation between magnesium and dementia.

參考文獻


1. de Baaij, J.H., J.G. Hoenderop, and R.J. Bindels, Magnesium in man: implications for health and disease. Physiol Rev, 2015. 95(1): p. 1-46.
2. Grober, U., J. Schmidt, and K. Kisters, Magnesium in Prevention and Therapy. Nutrients, 2015. 7(9): p. 8199-8226.
3. Jahnen-Dechent, W. and M. Ketteler, Magnesium basics. Clin Kidney J, 2012. 5(Suppl 1): i3-i14.
4. Alexander, R.T., J.G. Hoenderop, and R.J. Bindels, Molecular determinants of magnesium homeostasis: insights from human disease. J Am Soc Nephrol, 2008. 19(8): p. 1451-1458.
5. Vormann, J., Magnesium: nutrition and metabolism. Molecular aspects of medicine, 2003. 24(1-3): p. 27-37.

延伸閱讀