Abstract：
OBJECTIVE: Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is a life-saving procedure for supporting patients with cardiogenic shock after cardiac surgery. This work aimed to analyse the impact of changes in blood lactate levels on the survival of patients on post-cardiotomy ECMO (PC-ECMO) and whether lactate clearance (LC) performs better than absolute lactate levels.
RESULTS: We retrospectively analysed the data of adult patients who received PC-ECMO at our centre between 2016 and 2022. The primary outcome was the in-hospital mortality rate. Arterial lactate levels were measured at ECMO initiation, peak and 12 and 24 h after VA-ECMO support. LC was calculated at 12 and 24 h. Out of 2368 patients who received cardiac surgeries, 152 (median age, 48 years; 57.9% of them were men) received PC-ECMO. Of them, 48 (31.6%) survived and were discharged, while 104 (68.4%) died during the index hospitalization. Non-survivors had higher frequencies of atrial fibrillation (41.35% vs. 12.5%, P < 0.001), chronic kidney disease (26.9% vs. 6.3%, P = 0.004), prolonged cardiopulmonary bypass (237 vs. 192 min, P = 0.016) and aortic cross-clamping times (160 vs. 124 min, P = 0.04) than survivors. Non-survivors had a significantly higher median Sequential Organ Failure Assessment (SOFA) score at ECMO initiation (13.5 vs. 9, P < 0.001) and a lower median Survival After Veno-arterial ECMO (SAVE) score (-3 vs. 3, P < 0.001) with higher SAVE classes (P < 0.001) than survivors. After 12 h of VA-ECMO support, the blood lactate level was negatively correlated with LC in survivors (r = -0.755, P < 0.001) and non-survivors (r = -0.601, P < 0.001). After 24 h, the same negative correlation was identified between survivors (r = -0.764, P < 0.001) and non-survivors (r = -0.847, P < 0.001). Blood lactate levels measured at 12 h to determine hospital mortality [>8.2 mmol/L, area under the receiver operating characteristic curve (AUROC): 0.868] and 24 h (>2.6 mmol/L, AUROC: 0.896) had the best performance, followed by LC-T12 (<21.94%, AUROC: 0.807), LC-T24 (<40.3%, AUROC: 0.839) and peak blood lactate (>14.35 mmol/L, AUROC: 0.828). The initial pre-ECMO blood lactate (>6.25 mmol/L, AUROC: 0.731) had an acceptable ability to discriminate mortality but was less than the following measurements and clearance. Kaplan-Meier curves demonstrated that LC of <21.94% at T12 h and <40.3% at T24 h was associated with decreased survival (log-rank P < 0.001). Cox proportional hazards regression analysis for mortality revealed that LC of <21.94% at T12 h had an adjusted hazard ratio (HR) of 2.73 [95% confidence interval (CI): 1.64-5.762, P < 0.001] and LC of <40.3% at T24 h had an adjusted HR of 1.98 (95% CI: 1.46-4.173, P < 0.001). The predictors of hospital mortality after PC-ECMO were the lactate level at 12 h [odds ratio (OR): 1.67, 95% CI: 1.121-2.181, P = 0.001], initial SOFA score (OR: 1.593, 95% CI: 1.15-2.73, P < 0.001), initial blood lactate (OR: 1.21, 95% CI: 1.016-1.721, P = 0.032) and atrial fibrillation (OR: 6.17, 95% CI: 2.37-57.214, P = 0.003). Bivariate models using lactate levels and clearance at the same points revealed that blood lactate levels performed better than the clearance percentage.
CONCLUSIONS: Serial measurements of arterial blood lactate and LC help in obtaining early prognostic guidance in adult patients supported by VA-ECMO after cardiac surgery. Absolute lactate levels, compared with LC at the same time points, demonstrated better performance in differentiating mortality.
RESULTS: We retrospectively analysed the data of adult patients who received PC-ECMO at our centre between 2016 and 2022. The primary outcome was the in-hospital mortality rate. Arterial lactate levels were measured at ECMO initiation, peak and 12 and 24 h after VA-ECMO support. LC was calculated at 12 and 24 h. Out of 2368 patients who received cardiac surgeries, 152 (median age, 48 years; 57.9% of them were men) received PC-ECMO. Of them, 48 (31.6%) survived and were discharged, while 104 (68.4%) died during the index hospitalization. Non-survivors had higher frequencies of atrial fibrillation (41.35% vs. 12.5%, P < 0.001), chronic kidney disease (26.9% vs. 6.3%, P = 0.004), prolonged cardiopulmonary bypass (237 vs. 192 min, P = 0.016) and aortic cross-clamping times (160 vs. 124 min, P = 0.04) than survivors. Non-survivors had a significantly higher median Sequential Organ Failure Assessment (SOFA) score at ECMO initiation (13.5 vs. 9, P < 0.001) and a lower median Survival After Veno-arterial ECMO (SAVE) score (-3 vs. 3, P < 0.001) with higher SAVE classes (P < 0.001) than survivors. After 12 h of VA-ECMO support, the blood lactate level was negatively correlated with LC in survivors (r = -0.755, P < 0.001) and non-survivors (r = -0.601, P < 0.001). After 24 h, the same negative correlation was identified between survivors (r = -0.764, P < 0.001) and non-survivors (r = -0.847, P < 0.001). Blood lactate levels measured at 12 h to determine hospital mortality [>8.2 mmol/L, area under the receiver operating characteristic curve (AUROC): 0.868] and 24 h (>2.6 mmol/L, AUROC: 0.896) had the best performance, followed by LC-T12 (<21.94%, AUROC: 0.807), LC-T24 (<40.3%, AUROC: 0.839) and peak blood lactate (>14.35 mmol/L, AUROC: 0.828). The initial pre-ECMO blood lactate (>6.25 mmol/L, AUROC: 0.731) had an acceptable ability to discriminate mortality but was less than the following measurements and clearance. Kaplan-Meier curves demonstrated that LC of <21.94% at T12 h and <40.3% at T24 h was associated with decreased survival (log-rank P < 0.001). Cox proportional hazards regression analysis for mortality revealed that LC of <21.94% at T12 h had an adjusted hazard ratio (HR) of 2.73 [95% confidence interval (CI): 1.64-5.762, P < 0.001] and LC of <40.3% at T24 h had an adjusted HR of 1.98 (95% CI: 1.46-4.173, P < 0.001). The predictors of hospital mortality after PC-ECMO were the lactate level at 12 h [odds ratio (OR): 1.67, 95% CI: 1.121-2.181, P = 0.001], initial SOFA score (OR: 1.593, 95% CI: 1.15-2.73, P < 0.001), initial blood lactate (OR: 1.21, 95% CI: 1.016-1.721, P = 0.032) and atrial fibrillation (OR: 6.17, 95% CI: 2.37-57.214, P = 0.003). Bivariate models using lactate levels and clearance at the same points revealed that blood lactate levels performed better than the clearance percentage.
CONCLUSIONS: Serial measurements of arterial blood lactate and LC help in obtaining early prognostic guidance in adult patients supported by VA-ECMO after cardiac surgery. Absolute lactate levels, compared with LC at the same time points, demonstrated better performance in differentiating mortality.
摘要：
目的：静脉-动脉体外膜肺氧合（VA-ECMO）是支持心脏手术后心源性休克患者的一种挽救生命的方法。这项工作旨在分析血液乳酸水平变化对心脏手术后ECMO（PC-ECMO）患者生存的影响，以及乳酸清除率（LC）是否优于绝对乳酸水平。
结果：我们回顾性分析了2016年至2022年在我们中心接受PC-ECMO治疗的成年患者的数据。主要结果是住院死亡率。在ECMO开始时测量动脉血乳酸水平，峰值以及VA-ECMO支持后12和24小时。在12和24小时计算LC。在2368例接受心脏手术的患者中，152(平均年龄，48岁；其中57.9%是男性）接受PC-ECMO。其中,48人(31.6%)存活出院,104人(68.4%)在住院期间死亡。非幸存者的房颤发生率较高（41.35％vs.12.5%,P<0.001)，慢性肾脏病（26.9%vs.6.3%,P=0.004),延长体外循环(237vs.192分钟,P=0.016)和主动脉交叉钳夹时间(160vs.124分钟,P=0.04）比幸存者。非幸存者在开始ECMO时序贯器官衰竭评估（SOFA）评分中位数明显较高（13.5vs.9，P<0.001）和静脉动脉ECMO（SAVE）评分后中位生存率较低（-3vs.3，P<0.001），SAVE等级高于幸存者（P<0.001）。经过12小时的VA-ECMO支持，存活者(r=-0.755,P<0.001)和非存活者(r=-0.601,P<0.001)血乳酸水平与LC呈负相关。24小时后，存活者(r=-0.764,P<0.001)和非存活者(r=-0.847,P<0.001)之间存在相同的负相关。在12小时测量血乳酸水平以确定医院死亡率[>8.2mmol/L，接收器工作特征曲线下面积（AUROC）：0.868]和24h（>2.6mmol/L，AUROC：0.896）具有最佳性能，其次是LC-T12（<21.94%，AUROC：0.807），LC-T24(<40.3%,AUROC：0.839）和峰值血乳酸（>14.35mmol/L，AUROC：0.828）。最初的前ECMO血乳酸（>6.25mmol/L，AUROC：0.731）具有可接受的区分死亡率的能力，但低于以下测量值和清除率。Kaplan-Meier曲线表明，T12h时LC<21.94%，T24h时<40.3%与生存率降低相关（log-rankP<0.001）。死亡率的Cox比例风险回归分析显示，T12h时<21.94％的LC调整后的风险比（HR）为2.73[95％置信区间（CI）：1.64-5.762，P<0.001]，T24h时<40.3％的LC调整后的HR为1.98（95％CI：1.46-4.173，P<0.001）。PC-ECMO后住院死亡率的预测因素是12h时的乳酸水平[比值比（OR）：1.67，95％CI：1.121-2.181，P=0.001]。初始SOFA评分（OR：1.593，95%CI：1.15-2.73，P<0.001），初始血乳酸(OR:1.21,95%CI:1.016-1.721,P=0.032)和心房颤动(OR:6.17,95%CI:2.37-57.214,P=0.003)。在相同点使用乳酸水平和清除率的双变量模型显示，血液乳酸水平优于清除率。
结论：连续测量动脉血乳酸和LC有助于获得心脏手术后VA-ECMO支持的成年患者的早期预后指导。绝对乳酸水平，与LC在同一时间点相比，在区分死亡率方面表现更好。
结果：我们回顾性分析了2016年至2022年在我们中心接受PC-ECMO治疗的成年患者的数据。主要结果是住院死亡率。在ECMO开始时测量动脉血乳酸水平，峰值以及VA-ECMO支持后12和24小时。在12和24小时计算LC。在2368例接受心脏手术的患者中，152(平均年龄，48岁；其中57.9%是男性）接受PC-ECMO。其中,48人(31.6%)存活出院,104人(68.4%)在住院期间死亡。非幸存者的房颤发生率较高（41.35％vs.12.5%,P<0.001)，慢性肾脏病（26.9%vs.6.3%,P=0.004),延长体外循环(237vs.192分钟,P=0.016)和主动脉交叉钳夹时间(160vs.124分钟,P=0.04）比幸存者。非幸存者在开始ECMO时序贯器官衰竭评估（SOFA）评分中位数明显较高（13.5vs.9，P<0.001）和静脉动脉ECMO（SAVE）评分后中位生存率较低（-3vs.3，P<0.001），SAVE等级高于幸存者（P<0.001）。经过12小时的VA-ECMO支持，存活者(r=-0.755,P<0.001)和非存活者(r=-0.601,P<0.001)血乳酸水平与LC呈负相关。24小时后，存活者(r=-0.764,P<0.001)和非存活者(r=-0.847,P<0.001)之间存在相同的负相关。在12小时测量血乳酸水平以确定医院死亡率[>8.2mmol/L，接收器工作特征曲线下面积（AUROC）：0.868]和24h（>2.6mmol/L，AUROC：0.896）具有最佳性能，其次是LC-T12（<21.94%，AUROC：0.807），LC-T24(<40.3%,AUROC：0.839）和峰值血乳酸（>14.35mmol/L，AUROC：0.828）。最初的前ECMO血乳酸（>6.25mmol/L，AUROC：0.731）具有可接受的区分死亡率的能力，但低于以下测量值和清除率。Kaplan-Meier曲线表明，T12h时LC<21.94%，T24h时<40.3%与生存率降低相关（log-rankP<0.001）。死亡率的Cox比例风险回归分析显示，T12h时<21.94％的LC调整后的风险比（HR）为2.73[95％置信区间（CI）：1.64-5.762，P<0.001]，T24h时<40.3％的LC调整后的HR为1.98（95％CI：1.46-4.173，P<0.001）。PC-ECMO后住院死亡率的预测因素是12h时的乳酸水平[比值比（OR）：1.67，95％CI：1.121-2.181，P=0.001]。初始SOFA评分（OR：1.593，95%CI：1.15-2.73，P<0.001），初始血乳酸(OR:1.21,95%CI:1.016-1.721,P=0.032)和心房颤动(OR:6.17,95%CI:2.37-57.214,P=0.003)。在相同点使用乳酸水平和清除率的双变量模型显示，血液乳酸水平优于清除率。
结论：连续测量动脉血乳酸和LC有助于获得心脏手术后VA-ECMO支持的成年患者的早期预后指导。绝对乳酸水平，与LC在同一时间点相比，在区分死亡率方面表现更好。
