第二型糖尿病患者服用Pioglitazone罹患膀胱癌、腎盂癌及輸尿管癌風險：以傾向分數配對模型比較

背景: 美國食品及藥物管理局(Food and Drug Administration，FDA)於2010年9月發出聲明表示，第二型糖尿病口服降血糖藥物pioglitazone可能潛在提高罹患膀胱癌之風險，並有進一步研究發現使用pioglitazone超過一年的時間或累積暴露量越高，可能造成罹患膀胱癌的風險提高。綜觀國內許多關於糖尿病議題的研究，大多聚焦於糖尿病流行病學探討、自我照顧行為與成效、重複用藥情況、心血管疾病預防等，鮮少觀察特定降血糖藥物對於罹患特定疾病之影響，且亞洲國家關於pioglitazone藥物使用研究較歐美國家少許多，再加上生活習慣、種族及環境之差異，外國之研究發現不見得符合台灣糖尿病患者用藥後之狀況。膀胱癌、腎盂癌及輸尿管癌皆屬尿路上皮癌，三者間有相似危險因子，在台灣膀胱癌的發生率較歐美國家低，為確保研究暴露組有足夠罹癌人數進行分析，本研究將三者皆納入為研究結果做觀察；故本研究欲了解台灣糖尿病患者服用pioglitazone藥物後之狀況，及其與罹患膀胱癌、腎盂癌及輸尿管癌之相關性探討。研究目的: 探討台灣地區第二型糖尿病患者族群，使用pioglitazone藥物與罹患膀胱癌、腎盂癌及輸尿管癌之關連性。研究方法: 採用次級資料進行分析，資料來源為2002-2010年全民健康保險資料庫兩百萬歸人檔，研究對象包含2003年至2010年使用口服降血糖藥物之第二型糖尿病患，追蹤病患是否罹癌至2010年12月，利用傾向分數配對(propensity score matching)和Cox regression analysis分析暴露組與對照組在罹癌狀況之差異。研究結果: 男性有較高的膀胱癌、腎盂癌及輸尿管癌罹癌風險，風險比值為女性的1.8倍，此外罹患膀胱癌、腎盂癌及輸尿管癌之風險也隨者年齡增長而提高；使用傾向分數配對後發現，pioglitazone與提高罹患膀胱癌、腎盂癌及輸尿管癌之風險無顯著相關性存在；是否使用pioglitazone的分組中，病患的存活曲線亦無顯著差異，但暴露組罹癌者有罹癌時間提早之趨勢；罹癌pioglitazone用藥時間與罹患尿路上皮癌同樣無顯著差異性存在，而pioglitazone累積暴露量1-10,500mg則會提高膀胱癌、腎盂癌及輸尿管癌罹癌風險(HR=1.77, p=0.05) ，但p值正好位於統計臨界值0.05，顯示配對後之差異性並不是相當顯著。結論: 本研究認為是否使用pioglitazone與罹患膀胱癌、腎盂癌及輸尿管癌並無顯著差異存在，在原樣本中雖有顯著提高罹癌之現象，但在配對後則轉為不顯著，顯示過去研究可能高估pioglitazone危險性；但pioglitazone暴露可能導致罹癌徵狀提早出現，在少量暴露下立即出現膀胱癌、腎盂癌及輸尿管癌診斷。

關鍵字

糖尿病患者、pioglitazone、膀胱癌、腎盂癌、輸尿管癌、傾向分數配對

並列摘要

Background:In September 2010, the U.S. Food and Drug Administration (FDA) issued a statement indicating that pioglitazone, one of the type II diabetes oral hypoglycemic agents, may potentially increase the risk of developing bladder cancer. Further studies state that the use of pioglitazone for over a year or high cumulative exposure may cause increased risk of bladder cancer. In Taiwan, most of the research based on diabetes focused on the epidemiological study of diabetes, self-care and its effectiveness, repeated drug abuse and cardiovascular disease prevention. Only a few studies were published discussing the relationship between hypoglycemic agents and diseases. Studies about pioglitazone drug use in Asia is relatively less compared with Europe and the United States. In addition, due to the differences in race, lifestyle and environment, statements concluded from foreign study may not necessarily conform to the situation in Taiwan. The incidence rate of bladder cancer is relatively low in Taiwan, in order to ensure there is enough patients in the exposure group, renal pelvis tumor and ureter cancer, which both has similar risk factors as bladder cancer and are also categorized as urothelial carcinoma, were included in this study. The word bladder cancer appeared in the following text refers to all three cancers which were taken into account in this study, including bladder cancer, renal pelvis tumor and ureter cancer.This study evaluated the health status of diabetes patients after taking pioglitazone and the relationship between pioglitazone and bladder cancer. Aim: Examine the relationship between pioglitazone and incidence rate of bladder cancer, renal pelvis tumor and ureter cancer among type 2 diabetes patients in Taiwan Methods: Secondary data obtained from General Practice Research Database (GPRD) between 2002 and 2010 has been analyzed. Type 2 diabetes patients that were exposed to oral hypoglycemic agents between 2003 and 2010 were studied and followed up until December, 2010. Propensity score and Cox regression analysis were carried out to analyze the different incidence rate of bladder cancer between treatment group and control group. Results: An increase bladder cancer risk is observed for men compared to women, with a hazard ratio of 1.8. In addition, bladder cancer risk also increases with increasing age. With propensity score matching, no significant correlation was found between the risk of bladder cancer and pioglitazone using. There was no significant difference in survival curves between ever users and never users of pioglitazone. However, there is a trend of getting bladder cancer earlier for ever users compared to never users. No significant correlation was found between the duration of using pioglitazone and bladder cancer. Pioglitazone users with cumulative dose 1-10500 mg is observed to have a higher risk of getting bladder cancer (HR = 1.77, p = 0.05), nevertheless, the difference after pairing seems insignificant with a threshold-reaching p-value 0.05. Conclusion: There was no significant difference in bladder cancer risks between ever users and never users of pioglitazone. A significant increase of bladder cancer risk is observed for ever users in the original sample, but the increase risk diminishes after adjusting with propensity score matching, indicating that the potential hazard of pioglitazone may be overestimated. A significant increase of bladder cancer risk is observed in low cumulative pioglitazone users, however the interpretation is controversial since the p-value reaches the threshold of 0.05.