Renal function impairment is a frequent condition during septic shock with an incidence of acute kidney injury (AKI) ranging from 55 to 73% [
]. It is independently associated with mortality [
]. Renal perfusion alterations, especially at the microvascular level, appear to play a crucial role in AKI occurrence [
]. However, renal perfusion status remains poorly studied at the bedside during septic shock and macrovascular parameters (i.e. mean arterial pressure (MAP) and systemic oxygen delivery) mostly guide hemodynamic resuscitation in the intensive care unit (ICU) without renal microcirculation monitoring, which is often unavailable at the bedside in daily practice. Thus, despite optimization of macrovascular hemodynamic parameters according to Surviving Sepsis Campaign guidelines [
], occult renal hypoperfusion may persist with an early onset of renal dysfunction [
]. On the other hand, after initial hemodynamic resuscitation, renal blood flow (RBF) was also reported to be normal or even increased during septic shock with AKI developing concurrently [
]. Since renal function impairment may also occur despite a preserved global RBF, the study of intrarenal blood flow distribution becomes an important step in understanding the pathophysiology of AKI during septic shock. Hence, whether alterations of microvascular RBF are due to insufficient hemodynamic optimization or to specific microvascular injuries induced by sepsis (endothelial dysfunction, increased leukocyte adhesion, rheological abnormalities, glycocalyx alterations and functional shunting), we crucially need a tool to measure renal perfusion in daily practice [
Several tools were proposed to investigate renal microcirculatory perfusion at the bedside. Renal Doppler ultrasound measures the variations in intrarenal resistances and was proposed to assess a modification in renal perfusion after fluid loading [
] or vasopressor administration [
]. However, it does not directly quantify renal microcirculatory perfusion. Renal contrast-enhanced ultrasound (CEUS) was recently proposed to quantify renal cortical perfusion at the bedside in various conditions like chronic kidney disease, angiotensin intravenous infusion or captopril oral intake [
]. During CEUS, the use of intravenous contrast product increases blood echogenicity and enhances the visualization of microcirculatory renal perfusion when using ultrasonography. Good correlation has been demonstrated between CEUS measurements of renal perfusion and gold standard renal blood flow measurements (para-aminohippurate clearance) [
]. Experimental data on isolated pig and dog kidney reveals that renal cortical CEUS differentiates between macro and microcirculation [
]. This promising tool has been used in ICU patients in two pilot studies [
Since we lack clinical data on renal microcirculatory perfusion during septic shock, we intended to measure renal cortical perfusion with CEUS during the first 3 days in patients with septic shock.
The aims of our study were:
To compare cortical renal perfusion between patients with septic shock and ICU patients without sepsis
To evaluate renal cortical perfusion in patients with septic shock who develop or do not develop AKI
To test correlation between renal cortical perfusion and macrohemodynamic and microhemodynamic data
This study was approved by the comité d’Evaluation de l’Ethique des projets de Recherche Biomédicale (CEERB) du Groupe Hospitalier Universitaire (GHU) Nord (Institutional Review Board (IRB) of Paris North Hospitals, Paris 7 University, AP-HP) with the number 11-065. The institutional Review Board waived the need for informed consent. Patients (or relatives) were informed (explanatory note) and we obtained confirmation of non-objection to data use.
Patients were recruited in the 28-bed surgical ICU of the Bicêtre University Hospital between January 2014 and July 2015. Inclusion criteria were septic shock defined by the presence of two or more diagnostic criteria for systemic inflammatory response syndrome: elevated heart rate (> 90/min), tachypnea > 20/min, hyperthermia/hypothermia (> 38.3 °C/< 36 °C), leucocytosis or leukopenia (> 12,000/mm
3 or < 4000/mm
3), proven infection and vasopressors (norepinephrine or epinephrine) at a minimum rate of 0.1 μg.kg
Exclusion criteria were pregnancy, age < 18 years, urologic sepsis, left ventricular ejection fraction < 30% (measured with transthoracic echocardiography), systolic pulmonary arterial pressure > 90 mmHg (measured with transthoracic echocardiography), recent ischemic heart disease (within < 7 days) and chronic hemodialysis.
A control (CL) group constituted 10 ICU patients with brain damage (traumatic brain injury, stroke) without septic shock with an expected ICU length of stay > 72 h. These control patients were enrolled during the same period as patients with septic shock.
Renal perfusion was measured using CEUS at study inclusion, which was considered to be day 0 (within the first 24 h of ICU hospitalization), and at day 1 and day 3. All the measurements were acquired in patients with stable hemodynamics (i.e. MAP ≥ 65 mmHg with a stable norepinephrine infusion over 2 h). Demographic data and Simplified Acute Physiology Score (SAPS) II were reported at admission. The Sequential Organ Failure Assessment (SOFA) score, heart rate and MAP were reported. Cardiac index was measured by transthoracic echocardiography (2–4 MHz probe) using the Aplio 500 ultrasound machine (Toshiba Medical Systems Co. Ltd., Otawara, Japan) as described elsewhere [
]. Serum lactate, blood gases, norepinephrine infusion rate, serum creatinine, urine creatinine, serum urea, 24 h diuresis and 8 h diuresis (4 h before and 4 h after each CEUS measurement) were reported at day 0, 1 and 3. The severity of AKI was evaluated by the Kidney Disease Improving Global Outcomes (KDIGO) criteria. The renal resistivity index (RI) was measured in the interlobar arteries using Doppler ultrasound (Aplio 500, Toshiba Medical Systems Co. Ltd., Otawara, Japon) with a 6–11 MHz transducer. The Doppler spectrum was considered optimal when three consecutive waveforms were obtained. The RI was calculated as follows:
(Systolic velocity – End diastolic velocity)/Systolic velocity.
Contrast enhanced ultrasound measurements
The Sonovue® (Bracco, Milano, Italy) contrast ultrasound agent was used for this study. After reconstitution, the contrast ultrasound agent was diluted with an equal amount of NaCl 0.9%. It was infused through a central venous catheter with a dedicated Vueject® syringe pump (Bracco, Milano, Italy) at a rate of 1 mL.min
− 1 all along the sequence acquisition. The left kidney was visualized in B mode in its long axis view with an abdominal convex probe (6–11 MHz, Toshiba Medical Systems Co. Ltd., Otawara, Japon). Synchronous images were acquired in contrast ultrasound mode. The same echo parameters were applied for each acquisition to allow for contrast intensity comparison:
Mechanical index = 0.07 – Depth = 12 cm – Gain = 73 – Frame rate = 10 s
After 2 min of continuous contrast ultrasound agent administration (steady state), three consecutive destruction/replenishment sequences (with 15 s replenishment) were acquired. The destruction phase consisted of an ultrasound flash of high mechanical index (mechanical index = 1) that instantaneously destroyed the contrast ultrasound agent in the ultrasound plane observed with the probe. The replenishment phase corresponds to the progressive contrast enhancement of the kidney ultrasound plane after destruction.
CEUS sequences analysis
CEUS was performed at day 0, day 1 and day 3 in each patient. The ultrasound sequence was exported in digital imaging and communication in medicine (DICOM) format. Renal perfusion was analyzed using VueBox v 4.3 software (Bracco, Research, Geneva, Switzerland). The region of interest (ROI) was the cortical zone of the kidney parenchyma. Movement compensation (because of respiratory movements of the kidneys) was automatically applied. DICOM data on the ROI were converted into echo power data that are proportional to the concentration of the ultrasound contrast product in the parenchyma. A time-intensity curve (intensity of renal parenchyma replenishment over time) of the ROI was generated and three analysis parameters were calculated from this curve:
mTT (mean transit time, seconds) is the time needed after contrast agent destruction to reach 50% of the maximal intensity signal
rBV (renal blood volume, a.u.) is a measure of the ROI maximal intensity after complete replenishment
PI (perfusion index) is calculated by dividing rBV by mTT
The values obtained for the three destruction/replenishment sequences were averaged for each time measurement (day 0, day 1 and day 3).
Data are expressed as mean ± standard deviation or median (interquartile range) unless otherwise specified. Categorical data are expressed as number (percentage). Comparisons of general characteristics between septic shock group and CL at baseline (day 0) were performed using the Student t test or the Mann Whitney test according to the type of data. Comparison of proportions were done using Chi-squared test.
The evolution of CEUS parameters (mTT, PI) over time in septic shock group was evaluated with a one factor ANOVA on ranks (factor time: day 0, day 1, day 3). A
post hoc analysis with a Bonferroni correction was done between days if the global ANOVA was significant.
Evolution of CEUS parameters (mTT, PI, rBV) over time was compared between the septic shock group and CL or between patients with severe AKI (AKI stage 2 or 3) or no AKI (no AKI or AKI stage 1) during the first 72 h in patients with septic shock or between ICU survivors and ICU non survivors in patients with septic shock with a two way ANOVA analysis on ranks: a factor group and a factor time (day 0, day 1, day 3).
We plotted a receiver operating characteristic (ROC) curve of sensitivity versus (1-specificity) for various threshold values of a variable (mTT, PI, cardiac index, Doppler RI) at day 0 and calculated the area under this ROC curve to evaluate whether these variables may discriminate patients according to the occurrence of severe AKI (KDIGO stage 2 or 3).
Three CEUS measurements were obtained in the 30 patients. Thus, 90 measurements should have been done; however, 6 measurements were not be obtained in 4 patients because they died before day 3, and 2 other measurements were not obtained in a further 2 patients because of ultrasound system unavailability. The missing values in these six patients would have modified the strength of the statistical analysis. A value of zero could have been attributed to the missing data on the patients who died; however, it would have increased the difference between groups. That is the reason why we chose a more conservative approach by attributing the same value as the previous measure to missing data. We used the same approach with the other two patients who had missing data because of ultrasound system unavailability.
value <0.05 was considered to be statistically significant. The statistical analysis was conducted with software Prism (GraphPad software) and R 3.1.1 (
We performed ultrasound assessment of cortical renal perfusion during the first 72 h of septic shock resuscitation, using contrast technology for the first time for this indication. The first finding of this study was the highly variable cortical renal perfusion in patients with septic shock, which was high, intermediate or low despite satisfactory macrocirculatory hemodynamics. Furthermore, we showed that perfusion of the renal cortex is significantly reduced in patients with septic shock developing severe AKI (KDIGO stages 2 or 3) compared to those who do not develop AKI and compared to a control group without septic shock. Finally, a gradual improvement in renal cortical perfusion was observed over the first 72 h in patients with septic shock.
In this study, we quantified the cortical renal perfusion with CEUS for the first time and reported a decrease in renal cortical perfusion in patients with septic shock compared to ICU patients without septic shock. This decrease in cortical renal perfusion was the greatest in the first 24 h and progressively improved over the next 3 days. Renal hypoperfusion can lead to a decrease in renal function and was even reported to induce early renal injuries despite a complete correction of hypoperfusion after fluid resuscitation [
]. Studies conducted in patients with septic shock, using magnetic resonance imaging (MRI) and in patients with vasoplegic shock after cardiac surgery with renal venous thermodilution have also demonstrated a decrease in average renal blood flow [
]. On the contrary, recent experimental studies conducted in animal models of septic shock have demonstrated preserved, or even increased renal blood flow with a concomitant decrease in GFR [
]. Maldistribution of blood flow inside the renal cortex and the renal medulla, which subsequently leads to a drop in renal medulla partial pressure of oxygen (PO
), could explain this renal function impairment despite preserved blood flow [
]. In our study, we report a result that is consistent with the divergent results of these studies, as cortical renal perfusion ranged from almost no perfusion to hyperemic perfusion (even above the cortical renal perfusion observed in the control group) despite the fact that MAP was in the recommended target level in septic shock (> 65 mmHg) and cardiac output was comparable to that of ICU patients without septic shock. As a consequence, based on the macrohemodynamic data, the level of cortical renal perfusion cannot be predicted. Lactate level was higher in patients with septic shock with AKI than in those with no AKI or mild AKI, suggesting that renal microcirculatory impairment is likely to occur in patients with the most severe metabolic consequences of septic shock. Because renal perfusion patterns are heterogeneous in patients with septic shock, renal CEUS could thus be an interesting tool for early detection of those patients with severe renal hypoperfusion. The introduction of this potential new noninvasive research tool would help to monitor the effect of interventions designed to prevent or treat early renal dysfunction and AKI. This obviously needs to be investigated in clinical practice.
The evolution of renal CEUS parameters was not statistically significantly different between survivors and non survivors, though a tendency was observed. This could be due to the small number of patients in our study.
A decrease in cortical renal perfusion measured with CEUS (increase in mTT) was associated with severe AKI occurrence in patients with septic shock , and an increase in renal resistivity index (RI), as reported in previous studies [
]. Moreover, the strong and independent correlation between mTT and RI and between PI and RI suggests that RI changes are associated with cortical renal perfusion variations. This is consistent with previous Doppler ultrasound studies (measuring RI) that have suggested renal microcirculatory dysfunction in critically ill patients [
]. However, in this study, mTT was a better predictor of severe AKI (KDIGO stage 2 or 3) at day 0 than RI. The evaluation of cortical renal perfusion with CEUS (mTT) may be more accurate than Doppler ultrasound (RI) without contrast in detecting renal perfusion impairment.
Studies measuring renal perfusion with CEUS [
] or MRI [
] have not explored the association between renal perfusion and renal function. For the first time, we report an association between cortical renal perfusion and the occurrence of severe AKI in patients with septic shock. The greater the alteration in mTT (increase in mTT) the higher the risk of severe AKI. The decrease in renal cortical perfusion could have favored severe AKI in our cohort. However, on the contrary, evolving or established AKI may also have led to intrarenal microcirculatory disorders that further altered mTT. Interestingly, in a recent review reporting the renal histologic changes in experimental studies in septic AKI over 8 years (1059 animals), acute tubular necrosis was reported in 17% of animals and was only observed in the presence of decreased renal blood flow, and preserved renal blood flow was not associated with acute tubular necrosis [
]. Though renal function was altered either in the case of preserved or decreased renal blood flow in experimental septic shock, it seems that these two conditions have different effects on renal integrity.
Several parameters are available when studying renal perfusion with CEUS. mTT deals with the time needed to enhance renal parenchyma after contrast product destruction like “a renal refill time”. Alteration in dynamic perfusion tests (arm occlusion test with near-infrared spectroscopy) were reported to be associated with microvascular dysfunction in septic shock [
]. We reported an independent association between mTT and GFR and renal RI, whereas MAP and CI were not statistically associated with mTT. Thus, mTT seems to be mostly linked to intrarenal hemodynamics. The perfusion index (rBV/mTT) includes rBV, a measure of maximum CEUS signal intensity that is more variable than time to replenishment. Indeed, rBV depends on (1) the amount of circulating microbubbles (the concentration of which may vary between vials according to a ratio up to 1.5 (manufacturer data)), (2) the distance between the ultrasound probe and the kidney, which varies according to patients’ fluid balance and abdominal thickness and (3) renal cortical perfusion. These multiple factors are likely to explain why rBV failed to give relevant information on kidney perfusion. PI was independently associated with RI and CI but with MAP and GFR. Thus, PI may provide less information about intrarenal hemodynamics than mTT. This could explain why mTT and PI may differ though they are both considered renal perfusion parameters. In the end, mTT could be a more accurate parameter for evaluating intrarenal hemodynamics.
There are several limitations in our study. There was a heterogeneity in the renal CEUS parameters that were measured in the patients of the control group who were devoid of AKI. Renal CEUS quantification was reported to be reproducible in healthy volunteers [
]. Thus, the heterogeneity between individuals should be due to the acquisition conditions in the ICU: the depth of renal cortex according to the patient’s body mass index, the respiratory movements that may modify the acquisition ultrasound plane and the thickness and echogenicity of tissue that are modified over time because of fluid balance increase. This heterogeneity was reported in previous studies of ICU patients; however, changes in cortical renal perfusion were easily detected despite this heterogeneity [
]. Another explanation for measurement heterogeneity may come from patient enrollment at different phases of AKI, which probably correspond to different phases of microcirculatory alterations in the course of septic shock. A second limit is that we only measured cortical perfusion while medullary perfusion would be relevant and complementary [
]. Further studies are needed to investigate medullary perfusion during septic shock. This could be achieved in a future protocol with the injection of a higher dose of contrast because the renal medulla is less enhanced than the cortex at baseline. A measure of renal blood flow could have been relevant to distinguish between a decrease in global renal perfusion and an alteration in blood distribution inside the kidney. However, renal venous thermodilution would have been difficult to maintain for 3 days, with subsequent risk of renal venous thrombosis. This is a single-center study including surgical patients with septic shock, therefore further studies are needed to extend the conclusions to other ICUs. This study began before the new definitions of septic shock were published [
]. That is the reason why we used the former definition; 14 patients out of 20 would have satisfied the criteria for septic shock with the new definition. Patients with septic shock with AKI were older and had worse baseline renal function than those without AKI. Thus, it cannot be excluded that CEUS alterations were present because of older age or decreased baseline renal function in patients with AKI. Further investigations are necessary to extract the data on the changes observed by CEUS that are due to age, basal renal function and sepsis. We acknowledge that the sample size is small and subsequently implies the need for validation in a larger population. A larger cohort of patients would have allowed us to separate patients into four categories (no AKI, stage 1, stage 2, stage 3 AKI) to investigate if increasing severity of AKI is associated with increasing disorders in renal perfusion.