預測抗結核藥物治療期間肝毒性之評分系統的建立
中文摘要
結核病 (Tuberculosis or TB) 是目前全球性公共衛生的重要議題， 而目前結核病需要至少 6 個月密 集且多重藥物合併的療程，此種治療雖然可以獲得極佳的臨床療效，但是治療期間病患發生肝毒性 的比例卻相當高，然而肝毒性產生的機轉及與危險因子的關聯性尚未完全確立，也缺乏相關研究以 藉由高危險族群預測其肝毒性的發生，因此，藉由進一步確認兩者的關聯性，將有助於臨床工作。
本研究利用回溯性研究分析臨床實例，研究對象為民國九十年七月至民國九十一年七月，曾於台北
市立萬芳醫院診斷為結核病且處方結核藥物的病患，首先分析抗結核藥物治療後產生肝毒性的發生 率，本研究共收入 245 位病人進行研究，研究對象中共計 122 位 (49.8%) 發生肝毒性，其中肝功能 指數異常者為 81 位 (33.1%) ，肝損傷者為 41 位 (16.7%) ，若再將肝損傷細分，則以肝細胞型肝損 傷 (35 位 , 14.3%) 佔多數，膽汁鬱積型及混和型則各為 5 位 (2.0%) 及 1 位 (0.4%) 。
同時分別就病患基本資料、藥物使用及生化檢驗值三部分析相關危險因子對抗結核藥物治療後所產
生肝毒性的影響，結果發現男性、年齡大於 35 歲、累積合併疾病數大於 ( 含 ) 兩個、單一合併症高 血壓、單一合併症氣喘、單一合併症慢性阻塞性肺病、單一合併症腸胃道相關疾病、抽煙、飲酒、
酗酒、併服其他具肝毒性藥物、合併使用多種藥物、生化檢驗值 AST 異常及 ALP 異常等因子與肝 毒性的發生具統計學上的意義。
接著利用研究所得具統計學意義的危險因子，建構預測服用抗結核藥物後產生肝毒性的評分系統，
之後利用獨立變數變異的方法，以 3-fold validation 反覆驗證這一套系統的 internal validity ，最後選 用 9 個危險因子組合的一套系統，可得到最佳的評估結果，完成 Tuberculosis Therapy Hepatotoxicit y Scale (TTH) ，此評分系統對肝毒性族群的平均正確預測率為 78.5% ，整體平均正確率為 63.2% 。
藉由本研究所建立的評分系統，可快速於病患處方抗結核藥物前給予適當的評估，以加強對服用抗
結核藥物後可能發生肝毒性族群的監測。
A Method of Predicting Hepatotoxicity during Antituberculosis Therapy
英文摘要
Advanced clarify the relationship between hepatotoxicity during tuberculosis therapy and risk factors is an important issue to clinicians.
Patients who had prescribed with anti-tuberculosis medications from July 2001 through July 2002 at Taipei Municipal Wan-Fang Hospital were enrolled in present study. The medical charts of patients in enrollment group were retrospectively reviewed.
245 patients with a high percentage (122, 48.9%) of patients developed hepatotoxicity in present study. A mong patients developed hepatotoxicity, 81 (33.1%) had abnormal liver function and 41 (16.7%) had liver injury.
Through data colletion from clinical cases, male, age older than 35 years old, concommitant more than 2 ot her diseases, hypertension, asthma, COPD, GI-related disease, smoking, severe drinking, cocurrent use of o ther hepatotoxic medications, multiple medication use, abnormal baseline liver function test in AST and A LP were the idependent risk factors related to hepatotoxicity.
Using significant risk factors found in present study, a risk score draft were constructed to predict hepatoto xicity during antituberculosis therapy. Then use three-fold validation to train and test the internal validity o f the score through independent variable variation. After repeatly validation, the best score composed of 9 r isk factors was found, named Tuberculosis Therapy Hepatotoxicity Scale, with the average accuracy rate of 78.5% for hepatotoxicity group and average accuracy rate of 63.2 % for all.
By using of the risk score developed in present study, clinicians could evaluate patients quickly and easier before prescribing medications to improve monitoring in patients who would develop hepatotoxicity durin g their tuberculosis therapy.
