磺醯尿素類對第二型糖尿病人心臟血管風險之探討: 以全民健康保險資料庫為材料之藥物流行病學研究

Abstract

研究背景 心肌梗塞等大血管病變是第二型糖尿病患的主要死因。1970年代的Uiversity Group Diabetes Program (UGDP) 研究發現，第一代磺醯尿素類藥物(sulfonylureas，SU)，tolbutamide，有增加心血管死亡率之疑慮。雖然在體外實驗與動物實驗已發現，SU藥物可能會抑制心血管系統之K(ATP) channel，影響心臟的缺血性調適功能(ischemic preconditioning)，但是自從UGDP研究之後，並沒有針對此議題進行探討的大型臨床試驗。而各國以資料庫針對此議題執行的觀察性研究之結果並不一致。有鑑於SU藥物在第二型糖尿病的治療上佔有重要地位，而目前國內尚沒有探討此議題的全國性藥物流行病學研究，因此希望本研究能提供國內臨床醫療人員關於藥物選擇的參考。 研究目的 針對門診或住院期間新開方糖尿病藥物的第二型糖尿病患，探討SU藥物相較於其他糖尿病藥物，發生心血管相關住院事件的風險。此外，也針對不同種類的SU藥物在臨床上的使用，進行比較分析，以探討這些藥物對於心血管住院事件的影響。 研究方法 本研究為一項以台灣全民健保資料庫為材料執行的回溯性世代研究，利用健保資料庫2005年一百萬人承保抽樣歸人檔(LHID 2005)，收集2001年1月1日至2007年12月31日期間，於門診或住院期間新開方糖尿病藥物且大於18歲的糖尿病患為研究對象。本研究共分為2部分，第一部份，根據研究期間內使用過SU藥物(SU組)與未使用過SU藥物(Non-SU組)將研究族群分為兩組，進行描述性統計分析，探討兩組病人在人口統計學、臨床用藥分布狀況之異同。另外，根據SU組病人在研究期間內有否使用過glyburide，進一步分為兩個次族群分組(GLY組與Non-GLY組)，進行上述描述性統計分析。第二部份，利用time dependent Cox proportional hazard model進行以下三個研究終點事件的多變項迴歸存活分析：急性心肌梗塞與心臟血管手術住院事件(MI事件)、心衰竭住院事件(HF事件)、缺血性中風住院事件(IS事件)，以探討SU藥物在短期暴露(30天)與長期暴露(365天)兩種狀況下，發生這三個研究終點事件之風險。另外，將各個SU藥物獨立分為各單項後再進行分析，以探討個別sulfonylureas藥物對於研究終點的影響。而我們也針對口服降血糖藥物的累積暴露劑量，進行上述迴歸模式的分析。 研究結果 經過篩選之後，2001年至2007年間，符合本研究篩選條件的病患共37290位，平均年齡為55.9±14.4歲(mean±SD)；51.2%為男性。其中SU組為28340人，Non-SU組為8950人。第一部份的研究結果，在平均年齡(mean±SD)方面，SU組與Non-SU組之平均年齡分別為56.6±13.17歲與53.7±17.4歲；性別方面，男性在這兩組分別佔54.2%與41.7%。SU組病人在進入本世代研究後，使用各種糖尿病與血管藥物的人數比例均顯著高於Non-SU組。而GLY組進入世代研究後在心血管與糖尿病藥物之使用人數比例上，均顯著高於Non-GLY組。第二部份研究結果，針對MI事件，在校正相關的干擾因子後，顯示MI事件前30天有否使用過SU藥物，對於MI事件的發生沒有顯著影響。至於MI事件前365天內曾暴露過SU藥物的病人，相較於未暴露過SU藥物者發生MI事件之hazard ratio (HR)為1.15 (95% CI：0.99-1.32)，僅存在邊際顯著意義(p值=0.0598)。在同一模式下，若將SU藥物獨立開來分析，則發現gliclazide之HR為1.2 (95% CI：1.04-1.39)，達到統計顯著意義(p值=0.012)。而HF事件方面，在HF事件發生前30天有否使用過SU藥物對於HF事件之發生沒有顯著影響。而在HF事件發生前365天曾暴露過SU藥物，對發生HF事件之風險並未較高。但在同一模式下，若將SU藥物獨立開來分析，則顯示gliquidone之HR為2.08 (95% CI：1.11-3.38，p值=0.022)。至於IS事件方面，在IS事件前30天曾使用過SU藥物，對於IS事件之HR顯著較低，其中以gliclazide與glimepiride之效果較為顯著。而IS事件前365天曾使用過SU藥物，對於IS事件之HR則沒有顯著影響。且將SU藥物獨立開分析的結果顯示，在IS事件前365天曾使用過glyburide可能有較高之風險。 結論 研究期間使用過SU藥物者與未使用者，在族群基本特性以及心血管、糖尿病藥物的使用等背景資料存在差異，包括SU組年齡較高、使用心血管及糖尿病藥物人數比例較高等。而使用過glyburide與未使用過glyburide兩個SU藥物暴露族群，在藥物開方型態上亦有所不同。SU藥物對於急性心肌梗塞與心臟血管手術事件，在短期(30天)與長期(365天)的暴露下雖未發現顯著的影響，但其中在心肌梗塞事件前一年內，使用過gliclazide者可能具有較高的風險。對於心衰竭事件，只有在前一年內使用過gliquidone，可能和發生心衰竭事件具有較高相關性。而在缺血性中風方面，SU藥物在30天與365天內的暴露具有保護性的效果，以glimepiride與gliclazide較顯著，但其365天內的保護效果可能會受到glyburide的影響。本研究結果顯示各種SU藥物對於心血管系統的影響並不一致，因此後續的研究或許可以由個別藥物的角度進行更深入之探討。

Background Macrovascular complication is one of the main causes of death in patients with type 2 diabetes. In 1970s, the large clinical trial, University Group Diabetes Program (UGDP) study, pointed out that the first generation sulfonylureas (SU), tolbutamide, could be associated with higher cardiovascular mortality. Although it had been found that sulfonylureas would inhibit the K(ATP) channel in cardiovascular system to prevent the ischemic preconditioning, there has not been any large clinical trial investigating the controversial issue. The results of observational studies based on databse were not consistent with each other. Since sulfonylureas remain the mainstay in the pharmacotherapy of diabetes and to our knowledge, there has not been any pharmacoepidemiological study designed for this issue in Taiwan, so we conducted the present study to provide a reference for clinical medical staffs to select medications. Objective The study aimed to assess the association between the exposure of sulfonylureas and the risk of cardiovascular events on new users of antidiabetic medications, and furthermore, to compare the effects of individual sulfonylureas on the cardiovascular events. Mehtod This was a retrospective cohort study based on the National Health Insurance Research Database (NHIRD) in Taiwan. We established a cohort of patients with diabetes and aged older than 18 years, who initiated any antidiabetic medications between January 2001 and December 2007. Descriptive analysis. We divided our cohort into 2 groups according to the exposure of sulfonylureas. The SU group was composed of the patients who had ever used sulfonylureas, and the Non-SU group included the ones who had never used sulfonylureas. We compared the demographic characteristics and the patterns of the usage of antidiabetic and cardiovascular medications between the 2 groups. Besides, we divided the SU group into 2 sub-groups: the GLY group was the patients who had ever used glyburide, and the Non-GLY group was the ones who had never used. And the above comparison was conducted for the 2 sub-groups. Survival analysis. There were 3 end points in the present study: hospitalization for acute myocardial infarction and coronary revascularization (MI event), hospitalization for congestive heart failure (HF event), and hospitalization for ischemic stroke (IS event). We fitted 2 multivariate time dependent Cox proportional hazard models to evaluate the adjusted hazard ratio (HR) of all sulfonylureas and individual sulfonylureas for each end point. One of the models was constructed to estimate the HR for patients who exposed sulfonylureas during 30 days before the event, and the other one was similar but to assess the exposure of sulfonylureas during 365 days before the event. We also analyzed the effects of the accumulative dosage of oral hypoglycemic agents (OHAs) in the above 2 models. Results A total of 37290 eligible patients were included in the cohort during 2001 to 2007. The mean age of the cohort was 55.9±14.4 years (mean±SD); 51.2% were male. There were 28340 patients in SU group and 8950 in Non-SU group. The mean age of both groups was 56.6±13.17 years (mean±SD) and 53.7±17.4 years respectively. The male patients were 54.2% in SU group and 41.7% in Non-SU group. After the index date, the proportion of prescription of antidiabetic and CV medications in SU group was significant higher than Non-SU group. And the similar condition was also found in GLY group and Non-GLY group. Aftter adjusting the related confounding factors, the exposure of sulfonylureas during 30 days before MI events was not associated with the occurance of MI events. And there was a trend of increased risk for the patients who exposed sulfonylureas during 365 days before MI events (HR: 1.15, 95%CI: 0.99-1.32, p-value: 0.0598). If we analyzed individual sulfonylurea in the 365-day model, it was found that the HR of gliclazide was 1.2 (95% CI: 1.04-1.39, p-value:0.012). In the analysis of HF events, the variable of sulfonylureas was not significant in the model of 30 days before HF events, and the results were similar in the 365-day model. When seperating the individual sulfonylureas, the HR of gliquidone was 2.08 (95% CI: 1.11-3.38, p-vaule: 0.022). In IS events, the exposure of sulfonylureas during 30 days before IS event was associated with decreased risk, especially for gliclazide and glimepiride. But the protective effect did not exist in the 365-day model. Interestingly, the exposure of glyburide during 365 days before IS events could increase the risk of IS events. Conclusion In our findings, the demographic characteristics and the prescription of antidiabetic and CV durgs were significant different in SU group and Non-SU group. And the prescription patterns in GLY group and Non-GLY group was different. Moreover, patients who exposed sulfonylureas during 30 days or 365 days were not at elevated risk of MI event. But we found the individual one, gliclazide, would be associated with higher risk of MI events in the past 1 year. In HF event, we only found the association between gliquidone and increased risk of hospitalization for CHF when exposing 1 year before. Lastly, sulfonylureas would provide protective effects on ischemic stroke in short-term or long-term exposure, especially gliclazide and glimepiride. But the protection in the period as long as 365 days might be balanced by glyburide. The results in the present study indicated that the impact on CV system of individual sulfonylureas were different, we recommend that the further studies might be designed to explore the issue more deeply.