Measuring Renal Blood Flow with the Intraoperative Transesophageal Echocardiography Probe

In 2007, an international organization endorsed by the American Society of Nephrology and the Society of Critical Care Medicine convened to draft consensus statements regarding the pathophysiology and treatment of acute kidney injury (AKI) in cardiac surgery.^1,2 Six general pathophysiological processes were concluded to contribute to AKI: exogenous and endogenous toxins, metabolic factors, ischemia-reperfusion, neurohormonal activation, inflammation, and oxidative stress.¹ It was proposed that they are most likely interrelated and probably synergistic. Isolation of a single factor in the clinical setting was not considered feasible, rendering proof of causation unlikely. Given the complexity of cardiac surgery-associated AKI and a current lack of preemptive interventions or proven treatment strategies, only limited recommendations could be made in the consensus statements. Of particular interest to readers of the journal, one of the recommendations was “maintenance of adequate renal perfusion.”²

Therein lies the difficulty for the clinician participating in the care of patients undergoing cardiac surgery. Renal autoregulation is incomplete during anesthesia in surgical patients^3,4 and is complicated by the addition of hemodilution and hypothermia.^5–7 In general surgery patients, renal plasma flow, as measured by standard clearance methods, is significantly reduced after the induction of anesthesia. This occurs despite a stable mean arterial blood pressure³ or cardiac output,⁴ decreasing by as much as 40%.³ With the addition of normovolemic hemodilution, renal blood flow (RBF) may be even further reduced.⁵ In the rat model, initially there is an increase in RBF at a hematocrit of 25% due to the reduced viscosity, but this is accompanied by a halving of the baseline parenchymal oxygen tension. Hematocrits below 15% are associated with a significant decrease in RBF and an even more precipitous decrease in oxygen tension.⁵ Although animal models demonstrate that hypothermia is associated with a 50% reduction of RBF,⁶ invasive measurements in cardiac surgery patients suggest that the transition from normothermic to hypothermic bypass at the same flow rate is not accompanied by a reduced RBF.⁷ Nevertheless, RBF passively follows pump flow during cardiopulmonary bypass and is on average 60% of the prebypass level in the presence of typical mean arterial blood pressures and pump flows encountered during hypothermic bypass.⁷

Promoting “maintenance of adequate renal perfusion” would imply that we should be able to measure it. There is no practical method of accomplishing this intraoperatively. Investigations during cardiac surgery have been difficult to conduct. Techniques have included insertion of a renal vein retrograde thermodilution catheter guided by fluoroscopy and iodinated contrast venography⁷ or the measurement of effective renal plasma flow with ¹²⁵I-hippuran clearance,⁸ which is problematic in patients with significant tubular dysfunction or during hypothermia.⁹

Outside the operating room, transabdominal renal Doppler ultrasonography is a primary modality for assessing RBF. With this technique, two-dimensional scanning of the main renal arteries is followed by analysis of renal artery velocities by Doppler. An anterior approach (patient supine, transducer applied to the midline of the abdomen) or anterolateral (flank) approach (patient in the lateral decubitus position, transducer applied to the nondependent flank) allows visualization of both renal arteries. However, owing to factors such as bowel gas interposition or the variable anatomy of the left renal artery, complete examination of both renal arteries can be achieved in only 50% in some patient series but in up to 90% of patients in others.¹⁰ The segmental and intrarenal arteries are evaluated by means of a translumbar approach (patient in lateral decubitus position, transducer applied posteriorly). Using this approach, technical failure occurs in only 0%–2% of kidneys examined.¹⁰ The increased success rate is due to the closer proximity of the target vessel and transducer in the lumbar position, lack of air tissue interfaces and a greater likelihood of a favorable orientation of a segmental vessel with respect to the ultrasound beam because they radiate out from the main renal artery in several directions. (For a review of technical details and anatomy the reader is referred to http://www.gehealthcare.com/usen/ultrasound/education/products/cme_ren_art.html#4). Ultrasound imaging of the kidneys and renal arteries from either the transabdominal or translumbar position is now so routine that the majority of studies do not even describe the location of the transducer; the term “transcutaneous” will be used if the transducer position is not reported.

Transcutaneous ultrasonography has been well established as a diagnostic tool for detecting abnormal RBF patterns in many disease states,¹¹ and the characteristics of blood flow in the main renal artery and intrarenal arteries are well described.¹² Doppler velocities correlate with clearance studies in volunteers, where mean RBF correlates best with mean renal artery velocity.¹³

The use of transcutaneous ultrasonography to measure RBF may be applicable in the intensive care setting or even before and after surgery in the operating room but is not practical once surgery has started. This inevitably leads to the question of whether a transesophageal echocardiography (TEE) probe can image the renal vasculature during surgery. The first case report of TEE interrogation of the renal vasculature was by Chouinard et al.¹⁴ in 1991. Using a TEE probe, the aorta, renal, and mesenteric vessels of a patient were imaged after abdominal aneurysm resection, the transabdominal windows being obscured by dressings. This was followed in 1995 by an abstract presented at the American Society of Echocardiography which described 8 of 10 patients undergoing TEE during surgery yielded Doppler velocities of the renal artery.¹⁵ The following year the first study of the use of TEE to interrogate the renal arteries during surgery was published.¹⁶ Its goal was to identify aortic pathology and renal artery stenosis in 12 patients during open abdominal aortic surgery and to compare the findings with preoperative transabdominal ultrasonography. Once the peritoneal cavity was opened, the TEE probe was advanced into the stomach, grasped by the surgeon through the anterior wall of the stomach and “steered manually” to obtain images of the aorta and branches. The origin and proximal segments of 23 of the possible 24 renal arteries could be visualized, compared with 6 of 24 by transabdominal ultrasonography. Transabdominal Doppler ultrasound identified four of five angiographically proven renal artery stenoses, whereas TEE not only correctly identified all five but also included a sixth false positive. Of note, using the TEE, a 0° angle of interrogation was possible for Doppler measurements of renal artery flow versus up to 60° in the transabdominal approach. In 2001, Garwood et al.¹⁷ measured intrarenal blood flow using intraoperative TEE in nine patients undergoing cardiac surgery, without the requirement of the surgeon directing the probe. Velocities and calculated resistances were consistent with published transcutaneous data and followed the changes expected during an infusion of low-dose dopamine (2 μg · kg⁻¹ · min⁻¹). In this issue of the journal, Yang et al.¹⁸ report on the measurement of RBF with TEE during cardiac surgery, assess the feasibility of the procedure, and present volumetric estimates of RBF.

Yang et al. studied 60 consecutive elective cardiac surgery patients (mean weight 56 kg and mean body surface area 1.56 m²) with normal renal function; dopamine was administered on separation from bypass at the rate of approximately 4–8 μg · kg⁻¹ · min⁻¹ as part of the routine institutional standard of care. Using a 5 MHz TEE transducer, the authors were more successful at visualizing the left kidney. This is in contrast to Garwood et al.¹⁷ who were better able to differentiate the left kidney from the liver than the right kidney from the spleen. Yang et al. identified the left kidney “by location.” Starting from the midesophageal four-chamber view, the TEE probe was turned 90°–180° counterclockwise to visualize the descending aorta and then advanced 10–20 cm until the left kidney was seen. The main renal artery and vein were then identified and interrogated with Doppler after induction of anesthesia (baseline), during and after bypass. RBF was calculated from the product of vessel cross-sectional area, time velocity integral of the renal artery and heart rate. Given the lack of pulsatile flow, it was calculated during bypass as the product of the cross-sectional area of the vessel times the mean velocity.

The authors do not comment on the learning curve associated with this technique or how many patients they had to practice on before the study could begin. Garwood et al.¹⁷ state that it was a steep learning curve, suggesting that a significant effort was required to become acquainted with the location of the kidney. Unlike the transabdominal approach, it is rarely possible to follow the aorta distally with the TEE until the origin of the renal arteries. This may be a pitfall of the technique as described by Yang et al. One or two accessory renal arteries are present in approximately 30% of the population, more commonly on the left side, and failure to recognize this in a patient will significantly underestimate RBF. Also, 15% of the population exhibit early branching of the renal artery and, again, misinterpretation of a branch artery would underestimate RBF.

With respect to the methodology used by Yang et al. to measure RBF, it is important to emphasize its significant failure rate. The left kidney was not visualized in four patients (includes one case in which TEE was deferred after a difficult intubation, 6.7% overall). However, the main objective of the Yang et al. study was to measure RBF, and because a Doppler angle of <30° was not acquired in a further 20 patients (33%), only 36 patients (60%) yielded acceptable Doppler signals that could be used to calculate RBF.

The failure to acquire Doppler signals of the main renal artery within 30° of flow by Yang et al. is a problem. The Doppler Quantification Task Force section of the Standards Committee of the American Society of Echocardiography recommend that the Doppler sound beam should be oriented as parallel as possible to flow and that angle correction should not be used.¹⁹ Although a specific upper limit for the angle between flow and the Doppler beam is not given, the Task Force states that small (<20°) deviations in angle produce mild (<10%) errors in velocity measurements, presumably suggesting that the threshold of acceptability is somewhere in that region. If the Doppler angle is unacceptable, accurate measurement is not possible. So, is trying to measure RBF by TEE a nonstarter? There are other options: use segmental or other intrarenal vessels or angle independent Doppler measures of RBF. Segmental and other intrarenal vessels radiate out from the hilum of the kidney like the spokes of a wheel and increase the chance of finding blood flow parallel to the interrogating Doppler beam. It is not possible to calculate total RBF from a branch artery, but two indices that have been shown to correlate well with RBF and function can be assessed: resistive index (RI) and pulsatility index (Fig. 1).

Figure 1.: Doppler velocity profile of an intrarenal artery. Panel A: Doppler spectral display with gray scale two-dimensional ultrasound reference frame. Panel B: Line drawing of Doppler spectral display. PS = peak systolic velocity (cm/s); ED = end diastolic velocity (cm/s). Resistive index (RI) = (PS − ED)/ED; pulsatility index (PI) = (PS − ED)/mean velocity. Mean velocity is provided by proprietary software after tracing one Doppler velocity profile. Reproduced from Ref. ¹⁷ with permission.

These indices are angle independent by virtue of the fact that they are calculated as ratios of velocities rather than absolute velocities (Fig. 1). The error caused by the cosine of the angle between flow and the Doppler beam appears in both the numerator and denominator of the ratio, thus cancelling out.

The use of these angle-independent indices is less intuitive than measuring RBF as they actually reflect downstream resistance rather than flow. Nevertheless, just as in systemic vascular resistance and pulmonary vascular resistance, the bigger the number the higher the resistance and the lower the flow (Fig. 2). These indices are used in clinical practice in transcutaneous ultrasound studies of the kidneys and have been shown to correlate with measured renal vascular resistance, serum creatinine, and excretion of urinary markers of AKI in humans and vary inversely with effective renal plasma flow and creatinine clearance.^20–23 They can distinguish between acute tubular necrosis (ATN) and prerenal azotemia because of the profound renal vasoconstriction in ATN.²² RI is also highly correlated with the severity of abnormal histopathology at biopsy and clinical progression of renal diseases.²⁴ Postoperative transcutaneous pulsatility index was higher in children who developed ATN after cardiac surgery compared with those who did not, decreasing toward normal at the onset of recovery.²⁵

Figure 2.: Doppler velocity profiles of a renal artery in a pig hemorrhage model. Animals were bled at the rate of 1 mL/min and transabdominal Doppler ultrasonography of the main renal artery, interlobar, and arcuate arteries performed at A = 13 min, B = 35 min, C = 41 min, and D = 58 min. In (E) shed blood was reinfused. Resistive indices (RI) are shown at each time point demonstrating progressive vasoconstriction as cardiac output and renal blood flow (RBF) decrease. After restoration of the shed blood, RI returned toward normal. In (D), the end diastolic velocity is negative, producing a RI >1. Reproduced with permission from Clancy MJ, Alderman J, Case C, Taylor KJ. Resuscitation, 1995, 30, 161–7.

RI is reported as a secondary outcome variable in the study by Yang et al., but only for those patients in whom the Doppler angle was <30°. If this angle-independent index was calculated for all patients in whom the left kidney could be visualized, the success rate would have been 56 of 60 (93%) rather than 60%.

The main outcome variables of the Yang et al. study as stated were the feasibility and reproducibility of measuring RBF with intraoperative TEE. Their results are within the range of published values, increasing from 200 mL/min prebypass to 400 mL/min postbypass, which is consistent with the reported increased cardiac output and use of dopamine. RBF during bypass remained more or less at the same level as the post induction baseline, findings which are similar to a previous report.⁷ Nevertheless, this technique was only successful in less than two thirds of their patients. Therefore, the technique as described is feasible in some but not all patients. With respect to reproducibility, their calculations from stored images are reproducible with fairly low inter- and intraobserver variability. The reproducibility of technique across different patient populations or by different practitioners in the same patients has not been addressed.

In terms of the accuracy of the technique, no comparison was made to accepted methodology. Yang et al. report that an increased RBF after bypass was accompanied by an increased RI. Transcutaneous studies have shown that a normal renal RI is 0.58–0.64.^21,26–28 In general, most sonographers consider 0.70 to be the upper threshold of the normal RI in adults; higher values signify an increased renal vascular resistance. Yang et al. are therefore reporting that the renal vascular resistance was normal before bypass (0.68 ± 0.11) but was increased after bypass (0.77 ± 0.10), demonstrating an increased renal vascular resistance. However, hemodynamic measurements before and after bypass suggest otherwise. Calculated RBF increased after bypass, whereas mean arterial blood pressure remained the same, implying a decreased renal vascular resistance. Renal vasodilation during the dopamine infusion after bypass would be more consistent with the generally held notion that dopamine is a renal vasodilator. In a study of transabdominal renal ultrasonography in volunteers, Yura et al.¹³ demonstrated that 5 μg · kg⁻¹ · min⁻¹ dopamine decreased RI, which correlated with decreased renal vascular resistance as measured by standard clearance methods. In another study of patients undergoing coronary angiography, Manoharan et al.²⁹ placed a Doppler flow wire and pressure catheter in the main renal artery to investigate RBF during a dopamine infusion. At 5 μg · kg⁻¹ · min⁻¹, Manoharan et al. confirmed an increase in RBF velocities accompanied by a decrease in RI. At higher infusion rates, however, no further decline in RI was noted and at very high rates (>10 μg · kg⁻¹ · min⁻¹) α-adrenergic effects predominated with a trend toward increased RI. Increased RBF velocities at higher rates of dopamine infusion were driven by increased systemic blood pressure rather than by renal vasodilation. The findings of the biphasic effect of dopamine on renal vascular resistance by Manoharan et al. may account for the discrepancies in findings of increased RBF with increased RI in the study by Yang et al.

There are two other possible reasons for the reported increased RBF and simultaneously increased RI in the Yang et al. study which should be entertained: 1) the relationship between RI and actual renal vascular resistance does not hold after bypass, and 2) there are measurement errors in the study. With respect to the first possibility, the relationship between actual renal vascular resistance and RI may be more complex than previously thought. In a series of in vitro experiments, the importance of vascular compliance (rate of change of volume as a function of pressure) in RI analysis was demonstrated.³⁰ RI was found to be dependent on vascular resistance and compliance, becoming less and less dependent on resistance as compliance decreased until it became completely independent of vascular resistance when compliance was zero.³⁰ It may be possible that renal artery compliance decreased after bypass, changing the relationship of renal vascular resistance and RI. This possibility can neither be confirmed nor refuted from the data presented.

So, are there measurement errors in the study? If so, then either RBF is incorrect or RI is incorrect. A review of the literature would suggest that Doppler ultrasound is not used to calculate actual RBF; renal flow velocities are reported out rather than volumetric RBF. The calculation for volumetric blood flow from Doppler readings involves multiplying a cross-sectional area of the conduit by the time velocity integral of the Doppler wave form. The renal artery is a pulsatile structure with flow occurring in systole and diastole, and the diameter does not remain constant throughout the cardiac cycle. This is in contrast to measuring intracardiac flows with Doppler, where flow occurs in either systole or diastole and data are acquired at an orifice which remains relatively constant during that period, e.g., a valve orifice. The changing diameter of the renal artery would create a measurement error, which is subsequently squared in calculating the cross-sectional area. Using the diameter measurement made at the peak of the R-wave of the electrocardiogram would overestimate the RBF; it is advised that such diameter variation should be taken into account when calculating flow in arteries by ultrasound.¹² This can become quite complex. In a study of patients undergoing coronary catheterization, Horita et al.³¹ placed an IV ultrasound catheter and a Doppler flow wire into the renal artery to calculate RBF. To calculate cross-sectional area of the renal arteries, renal artery planimetry was performed using the IV ultrasound and custom-designed software then averaged over 15–30 heart beats to correct for changes related to the cardiac cycle. To confound things even more, respiratory variations also need to be considered. Mockel et al.³² found respiratory variations of both RBF (measured by an extravascular renal artery flowprobe) and renal artery velocity (measured by an intraarterial Doppler flow wire) in a pig model. And finally, with respect to calculating volumetric RBF from Doppler velocities, there is uncertainty about the shape of the profile of RBF velocities in the renal artery.^12,29 The exact position of the Doppler sampling volume along the diameter of the vessel introduces approximations that might lead to large and uncontrolled inaccuracies in flow calculations³³ (Fig. 3). With respect to errors in the renal resistive indices, RI was introduced and validated as a measure of parenchymal vessel characteristics, typically being measured no more proximally than the segmental arteries,³⁴ not the main renal artery.

Figure 3.: Idealized examples of different possible velocity profiles in an artery with laminar flow. Both profiles have the same maximum velocity but different means velocities. Note how placing a sampling volume at the same distance from the wall of the vessel may result in different velocities. Adapted from Ref. ¹², with permission.

Final thoughts on the Yang et al. study? Volumetric evaluation of RBF by Doppler is not recommended; literature concerned with this topic typically reports peak systolic or mean velocity as a surrogate for RBF. The correlation coefficient between measured RBF and renal artery Doppler velocities is excellent,^13,32 achieving a value of 1.0 in the study by Mockel et al.³² for both beat-to-beat measurements and timed flow. Doppler velocities were measured in 60% of the Yang et al. patients but 60% is not good enough for clinical application; using angle-independent indices could have produced a 93% success rate. The use of dopamine confuses the results by virtue of the reported biphasic effect at higher doses. And finally, the ability to visualize the kidneys and measure Doppler velocities in the typical body habitus of Western cardiac surgery patients has yet to be determined. Garwood et al.¹⁷ studied American cardiac surgery patients, but patient size and weight were not reported and the sample size was small (nine patients).

Being able to measure RBF during cardiac surgery would be a major step forward in demystifying the pathophysiology of cardiac surgery associated AKI. Yang et al. have shown that the kidneys can be visualized with TEE in the majority of patients at least in their patient population, and Doppler profiles of intrarenal vessels could probably have been determined. The next step is to compare TEE renal Doppler with accepted methodology for measuring RBF. A number of study designs could address that comparing TEE Doppler with transcutaneous Doppler, standard clearance methods, flowprobe measurements during open surgeries or to a renal artery Doppler flow wire during cardiac catheterization are all possibilities. If TEE Doppler evaluation of RBF proves to be a valid method of assessing renal perfusion, then it would be possible to manipulate RBF and establish its role in the complex process of cardiac surgery AKI. Whether it turns out to be the smoking gun is yet to be determined.