Abstract

Background Continuous renal replacement therapy (CRRT) remains a crucial treatment for critically ill patients with acute kidney injury (AKI), although the timing of its initiation is still a matter of contention. Furosemide stress testing (FST) may be a practical and benefcial prediction instrument. This research was meant to examine if FST can be used to identify high-risk patients for CRRT.

Methods This study is a double-blind, prospective interventional cohort study. For patients with AKI receiving intensive care unit (ICU) income, FST was selected with furosemide 1 mg/kg intravenous (1.5 mg/kg intravenous if a loop diuretic was received within 7 days). Urinary volume more than 200 ml at 2 h after FST was FST-responsive, less than 200 ml was FST-nonresponsive. The FST results are kept strictly confdential from the clinician, who decides whether to initiate CRRT based on laboratory testing and clinical symptoms other than the FST data. The FST data are concealed from both the patients and the clinician.

Results FST was delivered to 187 of 241 patients who satisfed the inclusion and exclusion criteria, with 48 patients responding to the test and 139 patients not responding. 18/48 (37.5%) of the FST-responsive patients received CRRT, while 124/139 (89.2%) of the FST-nonresponsive patients received CRRT. There was no signifcant diference between the CRRT and non-CRRT groups in terms of general health and medical history (P>0.05). Urine volume after 2 h of FST was considerably lower in the CRRT group than in the non-CRRT group (35 ml, IQR5-143.75 versus 400 ml, IQR210-890; P=0.000). FST non-responders were 2.379 times more likely to initiate CRRT than FST responders (95% CI 1.644–3.443, P=0.000). The area under the curve (AUC) for initiating CRRT was 0.966 (cutof of 156 ml, sensitivity of 94.85%, specifcity of 98.04%, P<0.001).

Conclusion This study demonstrated that FST is a safe and practical approach for predicting the initiation of CRRT in critically ill AKI patients.

Background

Acute kidney injury (AKI) in critically ill patients has long been a major complication for intensive care unit (ICU) physicians due to its morbidity of 57.3% and mortality of 26.9% [1]. Continuity renal replacement therapy (CRRT) is the standard treatment for severe AKI and its usage for life-threatening complications (e.g., severe hyperkalemia), but the results of major randomized studies on the date of CRRT initiation remain unclear [2–5]. Te AKIKI trial categorized AKI stage 3 as the early group compared to patients with urgent reasons to commence CRRT and did not discover better mortality in the early group, most likely because it was too late to start CRRT [2]. Te ELAIN trial identifed phase 2 of AKI as an early group, demonstrating a substantial beneft over commencing CRRT in phase 3. Obviously, the addition of NGAL also provides us with fresh ideas [3]. Since the AKIKI and ELAIN investigations fnished abruptly, the STARRT-AKI study included both ICU patients with renal insufciency and AKI stage 3; nonetheless, no CRRT improved patient outcome [4], and Bayesian secondary analysis did not demonstrate superiority for early beginning of CRRT [5].

As is the case with the design of these massive research, the terms "early" and "delayed" have distinct meanings. Te basis on which clinicians determine whether to commence CRRT is also fundamentally diverse, and in the same patient, various physicians may decide diferently whether to start CRRT. Te reason for this is that, to far, no comparable troponin marker has an extraordinarily high specifcity and sensitivity for identifying myocardial infarct. It has been established that furosemide stress testing (FST) may predict AKI development [6], and it has also outperformed various new markers in predicting bad outcomes [7]. Terefore, FST may be appropriate for risk classifcation of AKI patients to guide the choice to commence CRRT. Based on the preceding principles, we performed a prospective, double-blind, interventional cohort study to assess the predictive usefulness of FST for CRRT initiation.

Method

Trial design and ethics

The double-blind, prospective, interventional cohort study was conducted in the general intensive care unit of the Fourth Hospital of Hebei Medical University, China, between February 2021 and August 2022, collecting patients who met the study’s inclusion and exclusion criteria. Te study was registered on the China Clinical Trial Registry website: www.chictr.org.cn (ChiCTR1800015734) and authorized by the Fourth Hospital of Hebei Medical University’s Ethics Committee. Orally and in writing, the investigator notifed the patient or their authorized representative about the experiment.

Inclusion and exclusion criteria

All AKI patients admitted to the ICU who satisfed the following criteria were assessed for FST. Inclusion criteria: (1) patients admitted to the ICU meeting the AKI diagnostic criteria for Kidney Disease Improving Global Outcomes (KDIGO) guidelines [8]; (2) appropriate blood volume and central venous pressure (CVP)≥6  mmHg; and (3) urine output≤0.5 ml/kg/h for 6 h. Exclusion criteria: (1) indications for emergency CRRT: hyperkalemia, potassium of blood≥6.5  mmol/L; metabolic acidosis, PH≤7.15; acute pulmonary edema due to fuid over-load; developed uremia-related complications, such as pericarditis, bleeding, etc.; (2) age<18  years; (3) during pregnancy or lactation; (4) chronic kidney disease or having received renal replacement therapy 30  days prior to inclusion; (5) patients treated with ECMO; (6) pulmonary embolism; (7) presence of postrenal obstruction factors; (8) the primary disease is irreversible and/or is expected to die within 24 h. If all inclusion criteria were satisfed and there were no exclusion factors, we believed this patient would be eligible for inclusion.

Furosemide stress test

The urine volume was measured 2 h following administration of a 1 mg/kg intravenous infusion of furosemide (1.5 mg/kg if loop diuretics were used during the previous 7  days) and appropriate fuid resuscitation. FSTresponsive if the urine volume was more than 200 mL 2 h after the FST; FST-nonresponsive otherwise [6]. Te FST results are recorded and kept secret from the responsible clinician, who selects whether to initiate the CRRT based on laboratory testing and clinical performance other than the FST data. Te fndings of the FST were concealed from both the subjects and the responsible clinician.

Study protocol

We established an independent study group to screen patients admitted to the ICU for AKI and collect specimens and perform FST on patients who satisfed the inclusion criteria but did not match the exclusion criteria.

The responsible physician and the patient were both blinded to the 2-h urine volume following FST. Patients are not notifed of any changes in urine volume following an FST. Te study team blinded the physicians by covering the urine bag with an opaque paper bag for 2 h, recording the 2-h urine volume, and then discarding the urine.When the blinding was revealed, the urine volume 2 h after FST was withheld from the critical care record sheet until the patient initiated CRRT or was transferred out of the ICU, at which point the urine volume 2 h after FST was added to the record sheet.

Biomarkers

We believe that biomarkers may serve as reliable predictors of AKI development and risk stratifcation. After determining that a patient was eligible, 5 ml of blood was extracted from a vein and spun at 4000 rpm for 10 min. Te supernatant was frozen in a refrigerator at − 80 °C, and the neutrophil gelatinase-associated lipocalin (NGAL) concentration was determined using a double-antibody sandwich enzyme linked immunosorbent assay.

Kidney ultrasound

Kidney ultrasound was then performed, followed by the measurement of the renal resistance index (RRI): locating intrarenal vessels using a convex array probe in two-dimensional ultrasound mode and color Doppler, selecting the interlobe artery or arcuate artery, and meas-uring renal RRI. RRI=(PSV-EDV)/PSV, where PSV is the peak systolic velocity and EDV is the end-diastolic velocity. Te RRI is calculated using the average of three readings [9].

CRRT parameter setting

The initial CRRT mode is continuous venous-venous hemofltration (CVVH) using an integrated machine (Te Prismafex® system, Gambro, Sweden or the Aquarius™ system, Nikkiso, Japan) flter (AN69) with a prescribed dose set to 25–30  mL/kg/h. Te blood fow target was 150–200 mL/min. Local anticoagulant citrate is the frstline anticoagulation method. Patients with citrate anticoagulation contraindications opt for heparin for general anticoagulation, whereas patients with citrate anticoagu-lation contraindications and a propensity to bleed do not employ an anticoagulation strategy.

Statistical analysis

The SPSS 26.0 statistical program was utilized for processing and analysis. Te Chi-square test or Fisher’s exact test is utilized to evaluate categorical data between treatment groups. Continuous variables are presented as means (with standard deviations (SD)) or medians (with inter quartile ranges (IQR)) and compared across groups using the independent t-test for normally distributed data or the Wilcoxon rank sum test for non-normal data. Multivariate analysis was performed using logistic regression.

The receiver operating characteristic (ROC) curve was drawn using MEDCALC 15.2.2 software to evaluate the predictive value of CRRT at 2 h after FST. Te area under curve (AUC) indicates the predictive value and calculates the sensitivity and specifcity of the optimal cut-of value. Te P<0.05 was considered as a statistically signifcant diference.

Result

Cohort characteristics and baseline data

This study examined 241 AKI patients in the ICU; 187 patients got FST; the reasons for not administering FST were as follows: emergency CRRT (18), ICU stay less than 24  h (12), chronic kidney disease (11), pulmonary embolism (9), age less than 18 years (2), and ECMO (2). 18/48 (37.5%) of the FST-responsive patients received CRRT, while 124/139 (89.2%) of the FST-nonresponsive patients received CRRT (Fig. 1).

Baseline data included: general condition, reason for ICU admission, previous medical history, laboratory examination, stage of acute kidney injury, SOFA score, and APACHE II score. Te average age of the 187 individuals who had FST was 71 years, and 119 (63.63%) of them were male. 141 (75.4%) patients were admitted to the ICU due to respiratory and circulatory failure. Tere were 49 (26.20%) cases of AKI in stage 1, 36 (19.25%) in stage 2, and 102 (54.55%) in stage 3. Nearly half of the patients had a previous history of hypertension (47.05%), and 77 (41.18%) had a malignancy. Laboratory examination: white blood cells (15.58× 109 /L, IQR 10–23.18) and procalcitonin (6.85  ng/ml, IQR 1.5–17.03) levels were generally increased, and the mean serum creatinine was 201.4  mmol/L (IQR 100.53–267.4). We tested the mean blood NGAL concentration of 190.66 ng/ml (IQR 138.93–318.77), and ultrasound evaluated the patient with RRI 0.66 (SD 0.08). Te proportions of 126 (67.38%) patients with mechanical ventilation and 107 (57.22%) patients with sepsis were relatively high. Te mean time from FST to initiation of CRRT was 8.35 (SD 5.75) hours. Te SOFA score was 8.58 (SD 0.55), and the APACHE II score was 17.5 (IQR 15–25) (Table 1).

Univariate analysis of the CRRT group and the non‑CRRT group

The comparison CRRT and non-CRRT groups did not difer signifcantly in terms of general condition and previous medical history (P>0.05). Te platelets were lower in the CRRT group (P=0.04), and the PCT, blood creatinine, blood urea nitrogen, and blood NGAL were all higher than those in the non-CRRT group (P<0.05). Tere were more acidosis patients in CRRT group (P<0.05). In terms of the stage of acute kidney injury, there were a greater number of stage 2 and stage 3 patients in the CRRT group and a greater number of stage 1 patients in the non-CRRT group. We discovered that urine output after 2 h of FST was signifcantly lower in the CRRT group than in the non-CRRT group (35 ml, IQR 5–143.75 versus 400  ml, IQR 210–890; P=0.000). Te SOFA score and APACHE II score were higher in the CRRT group (P<0.05) (Table 2).

The RR of CRRT initiation

This study aimed to determine if FST, as an exposure factor, infuences doctors’ decisions to initiate CRRT. To do this, the relative risk (RR) of the FST not responding to the initiation of CRRT was established. Patients who were FST-nonresponsive were 2.379 times more likely to initiate CRRT than those who were FSTresponsive (95% confdence interval [CI], 1.644–3.443, P=0.000).

Multivariate analysis of the predicted CRRT initiation

As is well known, there are several factors that infuence doctors’ decisions on the initiation of CRRT. As a result, we included variables that could afect decision-making, omitted confounding variables, and conducted a multivariate analysis. Urinary output 2  h after FST was a predictor of CRRT initiation (P=0.032) (Table 3).

The ROC of CRRT initiation

As we found FST to be an independent and valuable predictor, we investigated a urine output cutof of 2  h after FST to better direct clinical work. By drawing the receiver operating characteristic (ROC) curve, the area under curve (AUC) was 0.966 (cutof 156 ml, sensitivity 94.85%, specifcity 98.04%, P<0.001). We also analyzed the AUC of serum creatinine and SOFA and discovered that urine production at 2 h after FST had a substantially greater predictive value than the former (Fig. 2) (Table 4).

Discussion

In recent years, there has been increasing interest in using loop diuretics as trials to assess proximal renal tubular cell injury (e.g., FST) and to predict the progression of AKI. In a follow-up study, urine output at 2 h after FST was superior to the individual urinary biomarker in predicting the progression of stages 1 and 2 AKI to stage 3 AKI [7]. In addition, FST was also used in patients undergoing kidney transplantation to predict the delayed recovery of kidney graft function, defned as receiving RRT within 7 days after kidney transplantation [10]. Therefore, the change in urine volume after FST partly refects the reserve situation of renal function, especially to avoid the delayed initiation of CRRT [11].

As a result, we designed this prospective interventional cohort study with a double-blind design to prevent FST results from interfering with clinicians initiating CRRT decisions and to ensure that both clinicians and patients were unaware of the FST results. From the results of the study, the feasibility of FST in judging the timing of CRRT initiation is proven. Te FST standardization study conducted by Chawla and colleagues showed that the use of 1–1.5 mg/kg of furosemide is safe in patients with AKI [6]. Pharmacologically, furosemide is a chemosynthetic loop diuretic commonly used in clinical practice. Te main mechanism of action is to act as by inhibiting the transmembrane domains 11 and 12 of the Na–K–2Cl cotransporter located in the thick wall segment of the renal tubular pulp loop [12]. Inhibiting sodium reabsorption creates a concentration gradient in the renal medulla, which is associated with the main driver of increased water and sodium excretion [13]. We often worry about whether a one-time infusion of furosemide at 1–1.5 mg/kg will burden the kidney and thus further worsen the kidney injury. Indeed, Mariano et al. found that furosemide infusion at 500 mg/24 h was safe with a urine volume greater than 500  ml/24  h and that the diuretic efect was linear with dose. Notably, when the urine volume per 24  h was less than 500  mL, the quantitative-efect relationship vanished and the toxic efect was observed [14]. Similarly, any FST-related safety issues were not observed in the conduct of our study.

Comparing the CRRT group and the non-CRRT group showed a huge diference in urine output at 2 h after FST, which was actually our expected result. Te furosemide response has been used to predict renal recovery and stop of RRT in critically ill patients recovering from AKI [15]. Terefore, we designed such a study to validate our idea. Currently, relatively few studies of the FST predict the timing of CRRT initiation; the best known "the FST trial" [16]. Te FST was performed on 162 patients with AKI in the ICU, with 118 patients remaining unresponsive. Te investigators randomized these 118 people into the early RRT group and the standard RRT group for treatment. Forty-four patients with an FST response were excluded from the randomized cohort, and only 6 of 44 (13.6%) received RRT. Furthermore, patients with non-responders to the FST were highly predictive of requiring RRT, and 45 out of the 60 patients in the standard treat-ment group ultimately received RRT. FST sorts out severe AKI cases from all AKI cases with high demand for RRT and avoids initiating RRT in low-risk patients. Due to the randomization, the relative risk of initiating CRRT in FST-responsive and FST-nonresponsive patients cannot be compared. Our study, cleverly using a double-blind setting, analyzed the efect of FST as an exposure factor on the outcome of whether to initiate CRRT or not. Ultimately, we found that patients with FST non-responses initiated CRRT 2.379 times more than patients with FST responses. Matsuura et  al. [17] reported an OR as high as 11.3, but the 95% CI range was 2.0–65.9. Te reasons for this diference may result from the severity of the included cases and the diferences in sample sizes. However, our results all suggest that a higher proportion of patients without FST initiate CRRT.

Although FST plays a large role in CRRT initiation and risk stratifcation in early AKI, some studies have found that it still has appropriate predictive value in AKI stage 3 patients. A retrospective study [18] with a large sample, which included 687 AKI stage 3 patients who underwent FST, observed the predictive value of urine output at 2 h and 6 h after FST for initiating CRRT, with an AUC 0.67 versus 0.71, P=0.02, respectively. Te cut-of value at 6 h after the FST was 600 ml, with a sensitivity of 80.9% and a specifcity of 50.5%. Te proportion of patients with early AKI (stages 1 and 2) and late AKI (stage 3), respectively, was 45.45%, 45%, and 54.55%, and 75.94% of patients received CRRT, indicating that some patients with early AKI received CRRT. Terefore, we plotted the receiver operating characteristic curve to analyze the predictive value and cutof value of urine volume for CRRT ini-tiation at 2 h after FST. Surprisingly, the area under the curve (AUC) reached 0.966 (cut-of value 156 ml, sensitivity 94.85%, specifcity 98.04%, P<0.001). By identifying potential AKI progression, we can initiate early intervention before life-threatening complications occur. Furthermore, several emerging AKI biomarkers, including tissue inhibitor of metalloproteinases-2 and insulin-like growth factor binding protein-7 (TIMP-2×IGFBP-7), interleukin-18 (IL-18), and plasma NGAL, have been shown to predict the progression of AKI [19–21]. In particular, TIMP-2×IGFBP-7 was confrmed for early AKI risk stratifcation in critically ill patients in multicenter studies and have been approved by the US FDA for clinical use [22, 23]. Tese cell cycle arrest biomarkers are expected to facilitate the early detection of patients at risk for AKI in a variety of clinical settings. Terefore, plasma NGAL was also detected in our study, and although there were statistically signifcant diferences between the CRRT and non-CRRT groups (P<0.05), plasma NGAL did not show superiority when we performed a multivariate analysis afecting the timing of CRRT initiation.

This study has some limitations: frst, this is a singlecenter prospective cohort study with a double-blind design that was limited by a small single-center sample size and an ofset in sample characteristics (for example, 41.18% of participants had a malignancy). As a result, the study’s fndings must be validated in a larger sample size and multicenter study. Second, although we designed the RRI, the two groups did not show signifcant diferences. Tere are also studies showing a limited predictive value of RRI in AKI progression [24, 25]. Limited by conven-tional ultrasound resolution, which does not show very well for changes in renal cortical microcirculation, perhaps we should further evaluate renal perfusion in using microbubble contrast ultrasound imaging [26]. From a pathophysiological point of view, the renal blood fow should change [27].

Conclusion

This study shows that FST is a safe and practical way to fgure out when CRRT should be started in critically ill AKI patients.

References

1. Hoste EA, Bagshaw SM, Bellomo R, Cely CM, Colman R, Cruz DN, Edipidis K, Forni LG, Gomersall CD, Govil D, et al. Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study. Intensive Care Med. 2015;41(8):1411–23.

2. Gaudry S, Hajage D, Schortgen F, Martin-Lefevre L, Pons B, Boulet E, Boyer A, Chevrel G, Lerolle N, Carpentier D, et al. Initiation strategies for renal-replacement therapy in the intensive care unit. N Engl J Med. 2016;375(2):122–33.

3. Zarbock A, Kellum JA, Schmidt C, Van Aken H, Wempe C, Pavenstadt H, Boanta A, Gerss J, Meersch M. Efect of early vs delayed initiation of renal replacement therapy on mortality in critically Ill patients with acute kidney injury: the ELAIN randomized clinical trial. JAMA. 2016;315(20):2190–9.

4. Investigators STARRT-AKI, Canadian Critical Care Trials Group, Australian, New Zealand Intensive Care Society Clinical Trials Group, United Kingdom Critical Care Research Group Canadian Nephrology Trials Network, Irish Critical Care TrialsGroup, Bagshaw SM, Wald R, Adhikari NKJ, et al. Timing

5. Zampieri FG, da Costa BR, Vaara ST, Lamontagne F, Rochwerg B, Nichol AD, McGuinness S, McAuley DF, Ostermann M, Wald R, et al. A Bayesian reanalysis of the standard versus accelerated initiation of renalreplacement therapy in acute kidney injury (STARRT-AKI) trial. Crit Care. 2022;26(1):255.

6. Chawla LS, Davison DL, Brasha-Mitchell E, Koyner JL, Arthur JM, Shaw AD, Tumlin JA, Trevino SA, Kimmel PL, Senef MG. Development and standardization of a furosemide stress test to predict the severity of acute kidney injury. Crit Care. 2013;17(5):R207.

7. Koyner JL, Davison DL, Brasha-Mitchell E, Chalikonda DM, Arthur JM, Shaw AD, Tumlin JA, Trevino SA, Bennett MR, Kimmel PL, et al. Furosemide stress test and biomarkers for the prediction of AKI severity. J Am Soc Nephrol. 2015;26(8):2023–31.

8. Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract. 2012;120(4):c179-184.

9. Schnell D, Darmon M. Bedside Doppler ultrasound for the assessment of renal perfusion in the ICU: advantages and limitations of the available techniques. Crit Ultrasound J. 2015;7(1):24.

10. McMahon BA, Koyner JL, Novick T, Menez S, Moran RA, Lonze BE, Desai N, Alasfar S, Borja M, Merritt WT, et al. The prognostic value of the furosemide stress test in predicting delayed graft function following deceased donor kidney transplantation. Biomarkers. 2018;23(1):61–9.

11. McMahon BA, Chawla LS. The furosemide stress test: current use and future potential. Ren Fail. 2021;43(1):830–9.

12. O’Grady SM, Palfrey HC, Field M. Characteristics and functions of Na-K-Cl cotransport in epithelial tissues. Am J Physiol. 1987;253(2 Pt 1):C177-192.

13. Burg MB. Tubular chloride transport and the mode of action of some diuretics. Kidney Int. 1976;9(2):189–97.

14. Mariano F, Leporati M, Carignano P, Stella M, Vincenti M, Biancone L. Urine volume as a predicting factor for furosemide clearance during continuous infusion in AKI septic shock patients on hemodiafltration. J Nephrol. 2018;31(6):889–97.

15. van der Voort PH, Boerma EC, Koopmans M, Zandberg M, de Ruiter J, Gerritsen RT, Egbers PH, Kingma WP, Kuiper MA. Furosemide does not improve renal recovery after hemofltration for acute renal failure in critically ill patients: a double blind randomized controlled trial. Crit Care Med. 2009;37(2):533–8.

16. Lumlertgul N, Peerapornratana S, Trakarnvanich T, Pongsittisak W, Surasit K, Chuasuwan A, Tankee P, Tiranathanagul K, Praditpornsilpa K, Tungsanga K, et al. Early versus standard initiation of renal replacement therapy in furosemide stress test non-responsive acute kidney injury patients (the FST trial). Crit Care. 2018;22(1):101.

17. Matsuura R, Komaru Y, Miyamoto Y, Yoshida T, Yoshimoto K, Isshiki R, Mayumi K, Yamashita T, Hamasaki Y, Nangaku M, et al. Response to different furosemide doses predicts AKI progression in ICU patients with elevated plasma NGAL levels. Ann Intensive Care. 2018;8(1):8.

18. Sakhuja A, Bandak G, Barreto EF, Vallabhajosyula S, Jentzer J, Albright R, Kashani KB. Role of loop diuretic challenge in stage 3 acute kidney injury. Mayo Clin Proc. 2019;94(8):1509–15.

19. Pan HC, Yang SY, Chiou TT, Shiao CC, Wu CH, Huang CT, Wang TJ, Chen JY, Liao HW, Chen SY, et al. Comparative accuracy of biomarkers for the prediction of hospital-acquired acute kidney injury: a systematic review and meta-analysis. Crit Care. 2022;26(1):349.

20. Piedrafta A, Siwy J, Klein J, Akkari A, Amaya-Garrido A, Mebazaa A, Sanz AB, Breuil B, Montero Herrero L, Marcheix B, et al. A universal predictive and mechanistic urinary peptide signature in acute kidney injury. Crit Care. 2022;26(1):344.

21. Marakala V. Neutrophil gelatinase-associated lipocalin (NGAL) in kidney injury—a systematic review. Clin Chim Acta. 2022;536:135–41.

22. Yang HS, Hur M, Lee KR, Kim H, Kim HY, Kim JW, Chua MT, Kuan WS, Chua HR, Kitiyakara C, et al. Biomarker rule-in or rule-out in patients with acute diseases for validation of acute kidney injury in the emergency department (BRAVA): a multicenter study evaluating urinary TIMP-2/IGFBP7. Ann Lab Med. 2022;42(2):178–87.

23. Hoste EA, McCullough PA, Kashani K, Chawla LS, Joannidis M, Shaw AD, Feldkamp T, Uettwiller-Geiger DL, McCarthy P, Shi J, et al. Derivation and validation of cutofs for clinical use of cell cycle arrest biomarkers. Nephrol Dial Transplant. 2014;29(11):2054–61.

24. Fu Y, He C, Jia L, Ge C, Long L, Bai Y, Zhang N, Du Q, Shen L, Zhao H. Performance of the renal resistive index and usual clinical indicators in predicting persistent AKI. Ren Fail. 2022;44(1):2028–38.

25. Darmon M, Bourmaud A, Reynaud M, Rouleau S, Meziani F, Boivin A, Benyamina M, Vincent F, Lautrette A, Leroy C, et al. Performance of Doppler-based resistive index and semi-quantitative renal perfusion in predicting persistent AKI: results of a prospective multicenter study. Intensive Care Med. 2018;44(11):1904–13.

26. Lima A, van Rooij T, Ergin B, Sorelli M, Ince Y, Specht PAC, Mik EG, Bocchi L, Kooiman K, de Jong N, et al. Dynamic contrast-enhanced ultrasound identifes microcirculatory alterations in sepsis-induced acute kidney injury. Crit Care Med. 2018;46(8):1284–92.

27. Harrois A, Grillot N, Figueiredo S, Duranteau J. Acute kidney injury is associated with a decrease in cortical renal perfusion during septic shock. Crit Care. 2018;22(1):161.