智能每日多次注射系统对高HbA1c 1型糖尿病患者的随机对照研究
本研究在HbA1c≥8%的1型糖尿病患者中，比较智能每日多次注射系统与每日多次注射联合持续葡萄糖监测的疗效和安全性。6个月时两组葡萄糖目标范围内时间差值为0.19个百分点，HbA1c差值为-0.13个百分点，均达到非劣效但未显示优效。事后分析显示对警报响应率高于28.3%的患者血糖改善更显著，并在真实世界队列中得到验证。
要点
- 智能MDI系统在TIR和HbA1c上非劣效于MDI联合CGM
- 对警报响应率超过28.3%的患者TIR和HbA1c改善更明显
- 研究中仅报告1例糖尿病酮症酸中毒和1例严重低血糖事件
核心结论
- 智能每日多次注射系统在TIR和HbA1c方面非劣效于每日多次注射联合持续葡萄糖监测，且安全性可接受。
- 对警报响应率高于中位水平28.3%的患者，TIR升高和HbA1c降低均优于低响应者。
- 对警报的响应行为是驱动血糖改善的关键因素，而非单纯佩戴设备。
研究设计与样本
- 这是一项随机对照试验，143名HbA1c≥8%的1型糖尿病患者经3周导入期后随机分配至智能MDI组或MDI联合CGM组。
- 主要终点为6个月时两组TIR的组间差异，调整基线血糖，非劣效界值为7.5个百分点。
- 次要终点包括HbA1c的非劣效性检验（界值0.4个百分点）以及TIR和HbA1c的优效性分析。
- 真实世界验证队列纳入5507名使用者，用于验证警报响应性与血糖结局的关联。
主要发现
- 6个月时两组TIR差值为0.19个百分点（95%置信区间：-4.9至5.3），达到非劣效性标准。
- HbA1c差值为-0.13个百分点（95%置信区间：-0.46至0.19），同样达到非劣效性但未显示优效。
- 高响应者（警报响应率高于中位28.3%，n=31）较低响应者（n=28）TIR增益高8.12个百分点（p=0.04）。
- 高响应者较低响应者HbA1c多降低0.6个百分点（p=0.02）。
- 安全性方面：导入期发生1例糖尿病酮症酸中毒，对照组发生1例严重低血糖事件。
关键数值与推荐
- 入组标准为1型糖尿病且HbA1c≥8%。
- TIR非劣效界值为7.5个百分点，HbA1c非劣效界值为0.4个百分点。
- 警报响应率中位数为28.3%，作为高响应与低响应的分组阈值。
- 6个月TIR组间差值为0.19个百分点（95%置信区间：-4.9至5.3）。
- 6个月HbA1c组间差值为-0.13个百分点（95%置信区间：-0.46至0.19）。
临床怎么用
- 智能MDI系统可作为HbA1c≥8%的1型糖尿病患者中MDI联合CGM的替代选择。
- 临床中应关注并提升患者对警报的响应行为，而非仅关注设备佩戴。
- 对警报响应率达到约28%即可观察到有临床意义的血糖改善。
局限性
- 警报响应性分析为事后分析，高响应组与低响应组样本量较小（n=31和n=28）。
- 主要终点仅达到非劣效性，未显示优效性。
以上解读依据论文摘要（全文未开放获取或许可不允许翻译转载）。
原文
ABSTRACT Aims The aim was to assess the efficacy and safety of the smart multiple daily injections (MDI) system versus MDI plus continuous glucose monitor (CGM) in people with Type 1 diabetes and HbA1c ≥ 8%. Material and Methods After a 3-week run-in, 143 participants were randomized to Smart MDI or MDI + CGM. The primary endpoint was between-group difference (Δ) in time-in-range (TIR) at 6 months, adjusted for baseline glycaemia and assessed for non-inferiority (margin 7.5%). Secondary endpoints included HbA1c (non-inferiority margin 0.4%) and superiority analyses for TIR and HbA1c. Smart MDI alerts prompt insulin bolus actions (in contrast to traditional CGM alerts), and post hoc analyses were performed to evaluate ‘alert-responsiveness’ on glycaemia. Alert-responsiveness findings were validated in a real-world cohort of 5507 users. Safety endpoints were assessed. Results At 6 months, Δ TIR between groups was 0.19 percentage points (95% confidence interval: −4.9 to 5.3), thus meeting non-inferiority; HbA1c also met non-inferiority (Δ: −0.13 percentage points, −0.46 to 0.19) but neither showed superiority. Post hoc analyses showed higher responders to alerts (those responding more than the median response rate of 28.3%, n = 31) had greater TIR gains than lower responders (n = 28) (+8.12 percentage points; p = 0.04) and HbA1c reduction (−0.6 percentage points; p = 0.02). TIR benefits in higher responders were confirmed in the real-world cohort. One episode of diabetic ketoacidosis occurred (run-in) and one severe hypoglycaemic event occurred (control arm). Conclusions Smart MDI is safe and non-inferior, achieving non-inferiority to MDI + CGM regarding TIR and HbA1c. Responding to as few as 28% of alerts resulted in glycaemic improvements.
引用
- Diabetes, Obesity and Metabolism：The Smart MDI System in People With Type 1 Diabetes and High HBA1C: A Randomized Controlled Trial Demonstrating That Response to Alerts Drives Outcomes
供执业医师工作参考，不替代临床判断；数值与分级以指南原文为准。