| 摘要: | 背景：口腔癌好發於台灣男性族群，在男性十大癌症死因中排名第四名。DNA甲基化（methylation）被認為是促使腫瘤的形成和成長的重要因素；DNA過度甲基化發生在啓動子區域上，促使基因沈默進而造成基因失去功能。本研究全面性鑑定807個癌症相關基因，評估1,505個CpG位點之甲基化狀態，希望找出檢測口腔癌病人之診斷與篩檢的生物標記。
研究方法：本研究採用病例對照研究法。樣本來自中國醫藥大學附設醫院組織庫，樣本來源為男性個案之口腔組織。病例組為口腔癌病人之口腔腫瘤組織（n=40）；對照組共15個正常的口腔組織樣本，包括正常口腔組織（n=5），以及來自病例組之腫瘤鄰近正常組織（n=10）。利用錯誤發現率方法來校正多重檢定的問題。評估每個CpG位點之敏感度、特異度以及接收操作特徵（Receiver Operating Characteristic，ROC）曲線下面積（area under curve，AUC）；並運用5-疊交叉驗證來評估模型的準確性。使用集群分析將CpG位點群集與研究樣本以熱圖呈現。
結果：挑選34個CpG位點作為群集，其敏感度及特異度皆大於或等於70%；亦進行排列組合來提高敏感度，從群集中至多考慮三個CpG位點來進行組合標記，只要組合中有一個CpG位點有甲基化則視為該組合有甲基化。排名前20名的標記是由七個基因組合而成（ASCL1、FGF3、FLT4、GAS7、KDR、TERT和TFPI2）；表現最好為ASCL1和FLT4的組合標記，特異度高達100%，敏感度為90%和AUC值為95%；另外，FLT4是唯一單一位點有進入前20名者，其特異度和敏感度分別為100%和82.5%，而AUC值為91.3%。為了找出預測早期口腔癌病人之生物指標，定義病理分期為I、II期的樣本為早期口腔癌（n=16）並進行敏感度分析。排名前20名的早期標記是由八個基因組成（ADCYAP1、EPHA7、FLT4、GSTM2、KDR、MT1A、NPY和TFPI2）；在早期口腔癌中表現最好的組合標記，敏感度高達100%，特異度為86.7%以及AUC為93.3%。
結論：我們研究結果發現十二個基因的CpG位點之甲基化狀態，在檢測口腔癌有好的表現，可以作為診斷口腔癌的生物指標；也提供我們未來研究的候選基因，評估使用非侵略性樣本（例如：唾液或口腔細胞），提供臨床上早期診斷與篩檢之價值。
Background: Oral squamous cell carcinoma (OSCC) is the fourth most common cancer in men in Taiwan. DNA methylation plays an important role in cancer progression and development. DNA hypermethylation in promoter regions has been described as a mechanism of gene silencing. The aim of this study is to perform a genowide methylation profile of 1,505 CpG sites of 807 cancer-associated genes and search for a diagnosis and screening biomarker for OSCC.
Study design: We conducted a case-control study to obtain tissue samples from the tissue bank of China Medical University Hospital. The tissue samples were obtained from oral cavity tissues of male subjects. The case group was comprised of samples from 40 OSCC males. A total of 15 samples composed the control group which included normal tissues (n=5) and adjacent normal tissues from OSCC cases (n=10). The control of false discovery rate (FDR) was used to correct for multiple comparisons. Specificity, sensitivity, and the area under the Receiver Operating Characteristic (ROC) curve (AUC) were calculated for each CpG site. We also applied 5-fold cross validation to evaluate the accuracy of a predictive model. The Hierarchical Clustering with a heat map was used to cluster CpG sites classifier and samples.
Results: Thirty-four single CpG sites with both the sensitivity and specificity greater than 70% were selected as the classifier. In order to improve the sensitivity, we also considered marker panels which were all possible combinations of up to 3 CpG sites in the clasidfier. The marker panel was defined to be methylated if at least one CpG site of the panel being methylated. The top 20 marker panels with the highest AUCs composed of seven genes, including ASCL1, FGF3, FLT4, GAS7, KDR, TERT, and TFPI2. The panel of genes ASCL1 and FLT4 represented the best combination with the specificity as high as 100%, 90% of sensitivity, and the AUC=95%. In addition, FLT4 was the only panel among the top 20s consisted of a single marker which had 100% of specificity, 82.5% of sensitivity, and the AUC=91.3%. In order to discover biomarker of the detection for the early stages of OSCC, we defined pathologic stage I and II as the early OSCC (n=16) to performing a sensitivity analysis. The top 20 marker panels of the early OSCC composed of eight genes including ADCYAP1, EPHA7, FLT4, GSTM2, KDR, MT1A, NPY, and TFPI2. The best marker panel for the early OSCC had the sensitivity as high as 100% with 86.7% of specificity and the AUC=93.3%.
Conclusions: In the present study we found the methylation status of CpG sites of twelve genes might be with great potential as the diagnostic biomarker for OSCC. Our study also suggests theses candidate genes for further study, especially the evaluation of using non-invasive samples, e.g. saliva or buccal cells, for the early diagnosis and screening of OSCC. |
