不同果糖比例誘發代謝症候群統計量之比較

代謝症候群是指一群容易導致心血管疾病的危險因子合併在一個人身上出現之徵狀的總稱。常見的心血管疾病有中風、高血壓、心臟病、冠狀動脈阻塞等。根據先前的研究指出餵飼大鼠60% 高果糖飼料可誘發大鼠肥胖、高血糖、高血壓、高胰島素血症、高三酸甘油酯、氧化壓力改變等症狀。亦觀察到大鼠代謝作用上和人類是相似的，有多重代謝失調如葡萄糖不耐，脂肪組織肥大和脂質在肝臟中堆積等，因此常用於代謝症候群相關藥物或保健食品的開發上。不過，前述模式的飼料中60% 為碳水化合物，且碳水化合物全部都是以果糖為主，在此種情形下若保健食材含有碳水化合物，則會因顧慮調整果糖比例後，代謝症候群是否會被誘發成功而有使用上的疑慮。鑑於此，中原大學生科系顏肇均的碩士論文為探討此一問題，其實驗將雄性大鼠分為四組，為一般飼料(控制組,C)，高果糖組果糖含量40%(F40)、50%(F50)、60%(F60)，餵飼13 週。期間每週觀察體重和血壓變化，每兩週採集禁食血液，測量血漿葡萄糖、胰島素、三酸甘油酯的變化，每四週進行一次口服葡萄糖耐受性試驗，實驗結束，將動物犧牲，採集血液，分析生化指標並取肝臟測量肝中三酸甘油酯、膽固醇、氧化壓力。其實驗結果顯示飼料含40% 果糖時，大鼠的血壓、葡萄糖不耐性、血糖、胰島素、三酸甘油酯、氧化壓力皆顯著上升，代表誘發出了代謝症候群的現象，在果糖50%、果糖60% 時亦有相同的情況。不過，細究顏的數據，發現資料有些涵蓋較多離群值和資料不為常態分配，故本論文使用One-Way Anova 和無母數統計Kruskal-Wallis test，比較這兩種的統計方法分析的結果與無母數統計是否可改善資料不為常態分配和有較多離群值的因素。結果顯示使用無母數分析發現有些結果與平均值分析的結果不一樣，差別為用無母數分析時，果糖為50% 時OGTT 葡萄糖和血漿胰島素當顯著高於控制組，果糖為40% 時血漿三酸甘油酯分別顯著高於F50、F60，當果糖40% 與50% 時肝臟三酸甘油酯分別顯著高於控制組，果糖40% 時肝臟氧化壓力顯著低於F50，皆從無顯著差異變成有顯著差異。結論為資料中若含有離群值或不為常態分配時，使用無母數統計方法可改善這兩個缺點，使分析更為精確。

關鍵字

代謝症候群；高果糖；胰島素阻抗；單因子變異數分析；無母數統計

並列摘要

Metabolic syndrome is a general term for the symptoms of a group of risk factors that cause cardiovascular disease to merge into a person. Common cardiovascular diseases include stroke, hypertension, heart disease, obstruction of coronary artery syndrome, etc. According to previous studies, rats were fed with 60% high fructose diets could induced obesity, hyperglycemia, hypertension, hyperinsulinemia, hypertriglyceridemia, and changes in oxidative stress. It is also observed that rats are similar in metabolism to humans and have multiple metabolic disorders such as glucose intolerance, adipose tissue hypertrophy, and accumulation of lipids in the liver. Therefore, it is often used in the development of metabolic syndrome-related drugs or health foods. However, 60% of the feeds in the aforementioned model are carbohydrates, and all carbohydrates are mainly fructose. In this case, if the health food contains carbohydrates, the metabolic syndrome will be induced success and affected use doubts, after adjusting the fructose ratio. In view of this, the master’s thesis of Chao-Jium Yen, a department of Chung Yuan Chirstian University Bioscience Technology student, was to explore this issue. The experiments were performed in male Sprague-Dawley rats that assigned to four groups: in control group (C), rats were maintained on standard chow and, in F40, F50, F60 group, rats were fed a 40%, 50%, 60% high-fructose diet respectively during 13 weeks. Body weight and blood pressure were measured per week. Fasting blood was collected biweekly, and plasma glucose, insulin, triglyceride were examined. The oral glucose tolerance test were performed per month. At the end of the experiment, rats were killed by decapitation. Plasma was collected for further bioassay. The liver triglyceride, cholesterol, and TBARS were assayed as well. The experimental results showed that the rats’ blood pressure, glucose intolerance, blood glucose, insulin, triglyceride, and oxidative stress all increased significantly when the diet contained 40% fructose, which represented a phenomenon that induced metabolic syndrome, the same situation occurs at 50% fructose and 60% fructose. However, when we examined Yen’s experimental data, there were some outliers and non-normal distribution including the data. Therefore, comparing two statistical analysis results and whether nonparametric statistics can prove the data that non-normal distribution and have more outliers by using One-Way ANOVA and nonparametric statistics in this paper. The results shown that, the outcomes of some factors were different in nonparametric statistics and mean of analysis. They were different that OGTT was significantly higher than control group when 50% fructose, plasma triglycerides were significantly higher than group F50 and group F60, respectively, when fructose was 40%, liver triglyceride was significantly higher than control group when fructose 40% and 50% respectively, the liver TBARS was significantly lower than F50 when fructose was 40%. These outcomes changed from no significant differences to significant differences. The conclusion is nonparametric statistic can improve these two defects and make the analysis more accurate if there were outliers in the data or non-normal distribution of the data.