肾脏超声联合指标对非脓毒症危重患者急性肾损伤的预测价值

目的:探讨肾动脉阻力指数（renal resistive index，RRI）和肾能量多普勒超声（power Doppler ultrasound，PDU）半定量评分联合指标对入住重症监护室（intensive care unit，ICU）的非脓毒症患者发生急性肾损伤（acute kidney injury，AKI）的预测价值。方法:采用前瞻性观察性研究的方法，纳入2018年1月至2019年8月期间于沧州市中心医院急诊ICU住院的非脓毒症危重患者作为研究对象。记录一般资料；于入ICU 6 h内应用医学超声仪完成RRI和PDU半定量评分测量。入ICU第5天依据改善全球肾脏病预后组织（KDIGO）标准评估肾功能，按肾功能情况分为AKI 3期组（入ICU 5 d内进展为AKI 3期）和AKI 0~2期组（未发生AKI或发生AKI 1或2期）。分别在非脓毒症和急性心力衰竭患者中比较不同AKI分期两组间各指标的差异。计量资料两组间比较采用独立样本 t检验或Mann-Whiney秩和检验。计数资料两组间比较采用卡方检验。绘制受试者工作特征曲线（receiver operator characteristic，ROC）分析RRI、PDU评分、RRI-RDU/10、RRI/PDU和RRI+PDU对AKI 3期的预测价值。使用Delong检验方法比较每个预测因子之间ROC曲线下面积的差异。 结果:共纳入110例非脓毒症危重患者（无AKI 51例，AKI 1期21例，AKI 2期11例，AKI 3期27例），其中急性心力衰竭患者63例（无AKI 21例，AKI 1期15例，AKI 2期7例，AKI 3期20例）。在非脓毒症患者及急性心力衰竭患者中，AKI 3期患者的急性生理学与慢性健康状况（APACHEⅡ）评分、序贯器官衰竭（sequential organ failure assessment，SOFA）评分、动脉乳酸水平、机械通气比例、血管活性药物比例、28 d病死率、肌酐、RRI、RRI-PDU/10、RRI/PDU、RRI+PDU及连续性肾脏替代治疗（continuous renal replacement therapy，CRRT）比例均明显高于AKI 0~2期患者（ P<0.05）；而尿量和PDU评分明显低于AKI 0~2期患者（ P<0.05）。非脓毒症患者中，RRI/PDU曲线下面积(AUC)=0.915，95%可信区间( CI)：0.846~0.959， P<0.01）及RRI+PDU（AUC=0.914，95% CI：0.845~0.959， P<0.01）对AKI 3期的预测价值最高，且两者与RRI（AUC=0.804，95% CI：0.718~0.874， P<0.01）和PDU评分（AUC=0.868，95% CI：0.791~0.925， P<0.01），差异均有统计学意义（均 P<0.05）；RRI/PDU预测AKI 3期的最佳临界值为0.355（灵敏度92.6%，特异度81.9%，约登指数0.745）；RRI-PDU/10（AUC=0.899，95% CI：0.827~0.948， P<0.01）对AKI 3期的预测价值亦优于RRI和PDU评分，但较RRI/PDU和RRI+PDU略差，仅RRI与RRI-PDU/10之间差异有统计学意义（ P<0.05）。在急性心力衰竭患者中，RRI/PDU（AUC=0.962，95% CI：0.880~0.994， P<0.01）及RRI+PDU（AUC=0.962，95% CI：0.880~0.994， P<0.01）对AKI 3期的预测价值亦最高，且两者与RRI（AUC=0.845，95% CI：0.731~0.924， P<0.01）和PDU评分（AUC=0.913，95% CI：0.814~0.969， P<0.01）两两间均差异有统计学意义（均 P<0.05）；RRI/PDU预测AKI 3期的最佳临界值为0.360（灵敏度95.0%，特异度90.7%，约登指数0.857）；RRI-PDU/10（AUC=0.950，95% CI:0.864~0.989， P<0.01）对AKI 3期的预测价值亦优于RRI和PDU评分，但较RRI/PDU和RRI+PDU略差，仅RRI与RRI-PDU/10之间差异有统计学意义（ P<0.05）。 结论:RRI和PDU评分的联合指标可有效预测非脓毒症患者发生AKI 3期，尤其在急性心力衰竭患者中表现更优。RRI与PDU评分的比值对AKI 3期的预测价值以及实用价值最好，建议临床推广应用。

Predictive value of renal ultrasound joint indicators to acute kidney injury in non-septic critically ill patients

Objective:To explore the predictive value of renal resistive index (RRI) joint with semiquantitative power Doppler ultrasound (PDU) score to acute kidney injury (AKI) in non-septic critically ill patients.Methods:This prospective observational study enrolled non-septic critically ill patients admitted to the Emergency Intensive Care Unit of Cangzhou Central Hospital from January 2018 to August 2019. In addition to general data, RRI and PDU scores were measured with medical ultrasonic instrument within 6 h after admission. Renal function was assessed on the 5th day in accordance with kidney disease: Improving Global Outcomes criteria. The patients who progressed to AKI stage 3 within 5 days after admission were classified into the AKI 3 group, and the rest were classified into the AKI 0-2 group. The difference of each index was compared between the two groups in non-septic critically ill patients and patients with acute heart failure (AHF). Normal distributed continuous variables were compared using independent sample t-tests, whereas Mann-Whitney U tests were used to examine the differences in variables without a normal distribution. Categorical data were compared with the Chi-square test. Receiver operator characteristic curves were plotted to examine the values of RRI, PDU score, RRI-RDU/10 (subtraction of RRI and 1/10 of PDU score), RRI/PDU (the ratio of RRI to PDU score), and RRI+PDU (the prediction probability of the combination of RRI and PDU score for AKI stage 3 obtained by logistic regression analysis) in predicting AKI 3. Delong's test was used to compare the area under the curve (AUC) between predictors. Results:A total of 110 non-septic critically ill patients (51 patients with no AKI, 21 with AKI stage 1, 11 with AKI stage 2, and 27 with AKI stage 3) were recruited. Among them, there were 63 patients with AHF (21 patients with no AKI, 15 with AKI stage 1, 7 with AKI stage 2, and 20 with AKI stage 3). Among the non-septic critically ill patients as well as its subgroup of AHF, compared with the AKI 0-2 group, acute physiology and chronic health evaluation-Ⅱ score, sequential organ failure assessment score, arterial lactate concentration, mechanical ventilation rate, proportion of vasoactive drugs, 28-day mortality, serum creatinine, RRI, RRI-RDU/10, RRI/PDU, RRI+PDU, and rate of continuous renal replacement therapy were higher in the AKI 3 group, and urine output and PDU score were lower ( all P<0.05). As for non-septic critically ill patients, RRI/PDU AUC=0.915, 95% confidence interval ( CI): 0.846-0.959, P<0.01] and RRI+PDU (AUC=0.914, 95% CI: 0.845-0.959, P<0.01) performed best in predicting AKI 3, and the AUCs were higher than RRI (AUC=0.804, 95% CI: 0.718-0.874, P<0.01) and PDU score (AUC=0.868, 95% CI: 0.791-0.925, P<0.01). The optimal cutoff for RRI/PDU was > 0.355 (sensitivity 92.6%, specificity 81.9%, Youden index 0.745). The predictive value of RRI-RDU/10 for AKI 3 (AUC=0.899, 95% CI: 0.827-0.948, P<0.01) was also better than RRI and PDU scores, but slightly worse than RRI/PDU and RRI+PDU, with statistically difference only between RRI and RRI-RDU/10 ( P<0.05). As for patients with AHF, RRI/PDU (AUC=0.962, 95% CI: 0.880-0.994, P<0.01) and RRI+PDU (AUC=0.962, 95% CI: 0.880-0.994, P<0.01) also performed best in predicting AKI 3, and the AUCs were higher than RRI (AUC=0.845, 95% CI: 0.731-0.924, P<0.01) and PDU score (AUC=0.913, 95% CI: 0.814-0.969, P<0.01) with statistically differences (all P<0.05). The optimal cutoff for RRI/PDU was > 0.360 (sensitivity 95.0%, specificity 90.7%, Youden index 0.857). The predictive value of RRI-RDU/10 for AKI 3 (AUC=0.950, 95% CI: 0.864-0.989, P<0.01) was also better than RRI and PDU score, but slightly worse than RRI/PDU and RRI+PDU, with statistically difference only between RRI and RRI-RDU/10 ( P<0.05). Conclusions:The combination of RRI and PDU score could effectively predict AKI 3 in non-septic critically ill patients, especially in patients with AHF. The ratio of RRI to PDU score is recommended for clinical application because of its excellent predictive value for AKI and its practicability.