Metformin 於晚期慢性腎臟病的預後及影響： 系統性回顧、統合分析及臨床試驗計畫書

Abstract

背景： 慢性腎臟病（chronic kidney disease, CKD）常見於第二型糖尿病患者，並與末期腎臟病及死亡風險增加密切相關。雖然二甲雙胍為第一線治療藥物，但基於乳酸中毒的疑慮，當估算腎絲球過濾率（estimated glomerular filtration rate, eGFR）低於 30 mL/min/1.73 m² 時通常建議停用。近期證據顯示，經審慎調整劑量後，二甲雙胍可能仍可使晚期 CKD 患者獲益，但其安全性及對死亡與腎病進展的影響仍未明確。 目的： 本系統性回顧與統合分析旨在評估二甲雙胍對晚期 CKD 患者之全因死亡、心血管事件、酸中毒及進展至末期腎臟病（end-stage renal disease, ESRD）風險的影響，作為臨床決策參考。 方法： 本研究依循 PRISMA 指引，系統性檢索 PubMed、Embase 及 Cochrane Library，納入研究對象之估算腎絲球過濾率（eGFR）低於 30 mL/min/1.73 m² 的觀察性研究。研究品質以 Newcastle–Ottawa Scale（NOS）評估，並以 GRADE 評定各結局的證據確定性。主要分析採隨機效應模型合併校正後風險比；事後次群組、統合迴歸與試驗序列分析均視為探索性分析。 結果： 在初步篩選的 1,777 篇文獻中，最終納入 15 項觀察性研究。研究地區涵蓋亞洲、北美洲及歐洲，追蹤期間介於 1 至 11 年，各研究納入的二甲雙胍使用者人數為 72 至 26,086 人。全因死亡分析共納入 10 項研究；合併結果顯示，二甲雙胍使用者的死亡風險呈下降趨勢，但未達統計顯著性（風險比：0.86；95%信賴區間：0.72–1.03；p = 0.112），且研究間異質性偏高（I² = 88.7%）。 二甲雙胍使用與重大不良心血管事件（major adverse cardiovascular events, MACE）風險降低 20% 顯著相關（風險比：0.80；95%信賴區間：0.69–0.91；p = 0.001；I² = 48.8%），亦與進展至 ESRD 的風險降低 37% 顯著相關（風險比：0.63；95%信賴區間：0.49–0.80；p < 0.001；I² = 90.8%）。然而，二甲雙胍使用同時與酸中毒風險增加 93% 顯著相關（風險比：1.93；95%信賴區間：1.07–3.48；p = 0.028；I² = 84.9%）。 2021 年後發表的研究一致顯示二甲雙胍具有保護性關聯，且研究間無明顯異質性（風險比：0.81；95%信賴區間：0.79–0.84；p < 0.001；I² = 0.0%）。全因死亡的試驗序列分析顯示，累積 Z 曲線未跨越校正後的序列監測界值。統合迴歸分析則顯示，追蹤時間較長之研究，其全因死亡 風險比 較低（p = 0.004）。 結論： 在晚期慢性腎臟病患者中，使用二甲雙胍與較低的重大不良心血管事件（MACE）風險及較低的末期腎臟病進展風險相關，但同時與較高的酸中毒相關事件風險相關；其與全因死亡率之間則未見顯著關聯。追蹤時間較長之研究，其死亡風險估計值較低。然而，這些結果無法證實因果關係，亦不足以支持在低於目前建議之 eGFR 閾值時使用二甲雙胍。未來仍需進行前瞻性試驗，並審慎調整劑量及監測安全性。

Background Chronic kidney disease (CKD) is common in patients with type 2 diabetes mellitus and is associated with substantial risks of end-stage renal disease and mortality. Although metformin is recommended as first-line therapy, its use is generally discontinued when eGFR falls below 30 mL/min/1.73 m² because of concerns about lactic acidosis. Emerging evidence suggests that carefully dose-adjusted metformin may remain beneficial in advanced CKD, but its safety and effects on mortality and kidney disease progression remain uncertain. Objective This systematic review and meta-analysis evaluated metformin's impact on all-cause mortality, cardiovascular events, acidosis risk, and ESRD progression in advanced CKD patients to inform clinical practice. Methods Following PRISMA guidelines, we searched PubMed, Embase, and Cochrane Library for observational studies in patients with eGFR <30 mL/min/1.73 m². Case reports, animal experiments, and studies lacking statistical effect sizes were excluded. Risk of bias was assessed using the Newcastle-Ottawa Scale (NOS). Adjusted hazard ratios (HR) were pooled using random-effects models; subgroup analyses, meta-regression, and trial sequential analysis were exploratory. Results From 1,777 articles screened, 15 observational studies were included, covering Asia, North America, and Europe, with follow-up periods of 1–11 years and sample sizes ranging from 72 to 26,086 metformin users. All-cause mortality analysis (10 studies) showed a trend toward reduced death risk (HR: 0.86; 95% confidence interval [CI]: 0.72–1.03; p = 0.112) with high heterogeneity (I² = 88.7%). Metformin use was significantly associated with a 20% reduction in major adverse cardiovascular events (MACE) (HR: 0.80; 95% CI: 0.69–0.91; p = 0.001; I² = 48.8%) and a 37% reduction in ESRD progression risk (HR: 0.63; 95% CI: 0.49–0.80; p < 0.001; I² = 90.8%), but was also associated with a 93% increase in acidosis risk (HR: 1.93; 95% CI: 1.07–3.48; p = 0.028; I² = 84.9%). Studies published after 2021 consistently showed protective effects with low heterogeneity (HR: 0.81; 95% CI: 0.79–0.84; p < 0.001; I² = 0.0%). Trial sequential analysis of all-cause mortality showed that the cumulative Z-curve did not cross the adjusted monitoring boundary. Meta-regression using each study’s mean follow-up duration as a study-level covariate showed that studies with longer follow-up reported lower hazard ratios for all-cause mortality (p = 0.004). Conclusion Metformin use in advanced CKD was associated with lower risks of MACE and progression to end-stage kidney disease, but with a higher risk of acidosis-related outcomes; no significant association with all-cause mortality was observed. Given the observational design, substantial heterogeneity, and very low-to-low certainty of evidence, these findings neither establish causality nor support initiating or continuing metformin below the recommended eGFR threshold. Prospective trials with prespecified estimands, dose adjustment, and rigorous safety monitoring are needed.