Perioperative fluid management in kidney transplantation: a black box

The definition of DGF is not consistent in the literature. At least 18 different heterogeneous criteria were identified in a systematic review [20]. Delayed graft function is most commonly used to describe the failure of the transplanted kidney to function promptly after transplantation, leading to dialysis within 1 week after transplant [20, 21].

The reported incidence of DGF varies from 2 to 70% in deceased kidney transplants [22, 23], not surprising given the wide range of definition. Outcomes are best in living donor transplantation, in which rates are 4–10% [23]. Despite improvement in transplantation techniques, the incidence of DGF has not decreased. A possible explanation may be the increasing number of grafts from expanded criteria donors and nonheart-beating donations [23, 24].

The underlying pathophysiology of DGF can be donor, recipient or surgeon related and can be due to hemodynamic (ischemic–reperfusion) or immunological (particularly T-lymphocyte) processes. A key factor is occult imbalance between oxygen delivery and consumption in the graft [25].

Acute tubular necrosis (ATN) is the most common cause of DGF. Acute tubular necrosis can be already present at the time of organ procurement if the deceased donor’s kidney is injured. Prolonged warm and cold ischemia time, as well as the manner of preservation, also can lead to ATN [23]. Acute tubular necrosis of the transplanted kidney can be diagnosed initially by nuclear renal scan which shows gradual uptake but minimal or no excretion of the tracer [26]. The hemodynamic status of the recipient during renal transplantation, the time following kidney implantation and completion of the anastomosis, and reperfusion injury all can effect graft perfusion [27] and the development of DGF [4, 5, 28]. Furthermore, the transplanted kidney is denervated and lacks neurogenic regulation of renal blood flow [29]. The vasodilation caused by mediators that accumulate during the ischemia period, and the increase in vascular permeability, might contribute to the invariable decline in CVP, from the operating room to the postanesthesia care unit [30]. Continuous monitoring is therefore highly recommended in postoperative care.

Several epidemiological studies in different patient populations have confirmed an association between stages of AKI and short-term and long-term outcomes [31‐33] as well as the independent association of AKI with higher mortality risk [34]. This is likely also true in kidney transplantation. Once the donated kidney has reached the implantation unit, donor and graft characteristics are no longer modifiable, and hemodynamic management becomes an adjustable portion of the process (Fig. 1).

In summary, DGF is predominantly caused by ATN. Adequate perfusion of the transplanted kidney is required, but to date no specific therapeutic intervention has reduced the incidence of ATN and recovery [35]. The role of health care professionals involved in perioperative fluid management of kidney transplantation is to identify the perfect balance of fluid therapy.

The assessment of a patient’s intravascular volume for achievement of appropriate fluid management is a major challenge for the clinician and has been shown to improve postoperative outcomes [12]. Understanding of the following four definitions is essential in perioperative care:

Despite a clear definition, there is no gold standard for the diagnosis and treatment of hypovolemia. Both acute and chronic conditions interact with fluid homeostasis, and medical history must always be considered. Progressive loss of renal function results in various adaptive and compensatory changes to maintain homeostasis with glomerular filtration rates below 10 ml/min, and these abnormalities typically have clinical consequences. A patient with end-stage renal disease has a disturbed acid–base and electrolyte balance and compromised hemodynamic autoregulation. Inflammatory mediators can alter renal hemodynamics and cause glomerular, tubular and interstitial damage, if uncontrolled [36].

In addition to the uncertainty associated with the patient’s volume state, appropriate targets remain unclear. To determine which monitoring method provides the most accurate information in the context of kidney transplantation, and which target to rely on, we will review previously implemented methods, their advantages and disadvantages, and their relative utility in predicting fluid responsiveness and tissue perfusion. Methods range from conventional techniques such as clinical history and physical examination, measurement of blood pressure (BP), heart rate (HR), urine output (UOP), CVP and pulmonary artery pressure (PAP), to novel technologies such as echocardiography, cardiorespiratory interactions and dynamic analyses of flow parameters.

Clinical experience and use of traditional parameters such as BP, HR, UOP, CVP [37] and PAP to guide perioperative fluid therapy in kidney transplantation have been shown to be unreliable.

Hypovolemia in the perioperative setting often can be misdiagnosed clinically [38] due to a lack of obvious fluid loss in the context of altered systemic vascular resistance associated with capillary leak. An observational study in healthy volunteers during progressive hemorrhage showed that blood loss of 25% can appear with minimal clinical impacts on BP and HR, despite worsening microcirculatory flow [39]. Hypotension is not present in all patients in shock. If the physician waits for hypotension to start treating hypovolemia, tissue hypoxia will already be present [40].

In the case of hypovolemia and hypoperfusion, physiological compensatory mechanisms can maintain effective vascular volume and perfusion by increasing the vascular retention of an infused volume and decreasing vascular capacitance. Plasma levels of renin and aldosterone increase and atrial natriuretic factor decreases, followed by arterioconstriction which decreases precapillary hydrostatic pressure and filtration [41, 42]. A 30–50% reduction in glomerular filtration, and an increase in fluid reabsorption from the interstitial space, can also help avoid tissue edema [42, 43]. These compensatory mechanisms are highly compromised in end-stage renal disease and in the postischemic kidney.

Blood pressure is the most commonly used vital parameter for organ perfusion, and functions as an indirect measurement of renal perfusion in clinical practice. Although classic physiological studies suggest that autoregulation of renal blood flow is maintained at lower levels of mean arterial pressure (MAP 50–60 mmHg) [44, 45], retrospective data suggest that intraoperative MAP < 55 mmHg is associated with AKI and myocardial injury [46]. In a prospective randomized controlled trial in which MAP targets of 65–70 mmHg were compared to MAP targets of 80–85 mmHg in septic patients, there were no significant differences in mortality [47]. Renal replacement therapy was less commonly required in patients with chronic hypertension whose MAP was maintained at values of ≥ 80 mmHg [47]. However, raising MAP by increasing the systemic vascular resistance does not necessarily improve flow and potentially also not renal perfusion.

Given the expected large oscillations in hemodynamic parameters during kidney transplantation, invasive arterial BP monitoring is commonly used. In a recent study of the biochemical outcomes of renal transplant recipients, no difference was observed in the creatinine levels of patients with a MAP between 95 and 131 mmHg, but MAP was continuously maintained at > 95 mmHg via the judicious use of fluids and dopamine [27]. Consistent with other studies, Campos et al. [4] observed that MAP < 93 mmHg is associated with poor graft function. In an earlier study, Tóth et al. [11] observed stable creatinine levels in patients with a MAP of 80–100 mmHg, but an increase in creatinine in patients with MAP < 80 mmHg. However, these associations do not imply that raising pressure would restore function. The optimal measurement to evaluate renal microcirculation is unclear, as well as it being unknown which individual MAP could prevent AKI during different medical conditions.

Earlier studies have indicated that the magnitude of UO is one of the most clinically used predictors for graft function immediately after reperfusion [48] and proposed the use of CVP and PAP values to predict diuresis [8, 10]. Although aggressive volume expansion to a target CVP of 10–15 mmHg [49] and PAP > 20 mmHg [10] at the time of reperfusion were associated with improved renal blood flow and better graft outcomes, targeting these values was not recommended by other authors. A twofold greater risk of kidney dysfunction was observed with CVP ≥ 11 mmHg [4]. Overinfusion can lead to tissue edema with decreased tissue oxygenation [50] and subjects vulnerable patients to complications such as pulmonary edema, infections, myocardial ischemia, ileus, compromised renal flow, renal injury and even increased mortality [4, 12, 51‐53].

A decrease in CVP in the post-transplant period is observed in a majority of kidney recipients, independent of their intravascular fluid status [30], but this does not mean the patient needs fluid. The etiology is unclear, but might be related to increased vascular permeability or the release of vasodilator mediators during the ischemic period. An increase in vascular capacitance also could be a factor [54]. Overemphasis on CVP as a goal can lead to excess infusion of fluid and increased capillary leakage, which can promote further fluid administration.

Use of pulmonary artery catheters has fallen out of favor due to concerns about complications such as pulmonary embolism among high-risk patients [55], pulmonary artery rupture [56] and cardiac perforation [57], and a frequent misinterpretation of the waveforms [58].

Static cardiac filling pressures such as CVP and PAP correlate poorly with the intravascular volume and are notoriously unreliable in predicting accurate fluid responsiveness compared to dynamic parameters [59, 60]. In a recent meta-analysis of the correlation between CVP and changes in cardiac performance, Marik and Cavallazzi [15] showed that only 57% of high-risk patients in the intensive care unit and operating room who appeared to be hypovolemic based on CVP were fluid responders, whereas the other half were unnecessarily loaded with fluids. However, a high CVP could potentially compromise renal flow, particularly when the arterial pressure is low.

Thus, the CVP should not be used alone to guide fluid responsiveness [15] and should not be a target for clinical decisions regarding fluid management [61]. Although the conventional parameters discussed previously are relevant to the hemodynamic assessment, none are good predictors of fluid responsiveness [59], which has prompted a search for alternative methods. There currently is increasing use of measurement of CO and SV with the aid of less invasive and easier to use devices. Use of these novel technologies to guide therapy is becoming vital in modern medical practice.

The primary goal in the perioperative setting is to avoid tissue hypoxia, the major cause of organ dysfunction. Traditional indicators can be normal in the presence of tissue hypoxia and cannot be used to predict an imbalance between oxygen demand and consumption, particularly if they are not interpreted in the context of perfusion markers such as CO, lactates and central venous saturation [13‐15, 61].

Sufficient oxygenation and tissue perfusion are crucial for all metabolic needs of cells and are based on hemoglobin concentration, arterial oxygen content and CO. Therefore, the percent change in any one of these three variables equally alters oxygen delivery, with CO usually having the largest changes [62]. As CO is the product of SV and HR, it can also be influenced by changes in preload, contractility and afterload. A mismatch between tissue needs and oxygen delivery leads to tachycardia and hypotension in most patients. Thus, an increase in preload with fluid therapy is only one part of treatment. An increase in contractility with inotropes remains the only therapeutic option when cardiac function is volume limited [62].

According to Frank–Starling’s law of the heart, administering fluids can increase the preload and thereby increase SV and CO. Under physiological conditions, both ventricles work on the ascending portion of the Frank–Starling curve, and afterload, contractility and HR are assumed to remain constant during the volume infusion [63]. This mechanism represents the functional preload reserve for the heart under stress conditions. If the patient is a fluid responder, this mechanism translates to a shift to the right of the venous return curve relative to the cardiac function curve and intersects it on the steep part of the cardiac function at a higher value [64]. A patient is considered a fluid responder when SV increases by at least 10–15% in response to volume expansion [65]. However, when the venous return intersects the flat part of the cardiac function curve there is no increase in CO and there even can be a decrease in ventricular performance and an increase in tissue edema and consequent tissue dysoxia. In nonresponders, CO only can rise by an increase in contractility or HR. The change in CVP in response to a fluid bolus can help interpret what happened with the fluid bolus when used in combination with the CO response.

Among the many devices and monitors currently available for measuring cardiac function and output, we will review the intraoperative use of transesophageal echocardiography (TEE) and emphasize the utility of noninvasive devices that calculate reliable trends in CO and dynamic indices based on simple physiology.

Transesophageal echocardiography can be a useful diagnostic tool for evaluating right and left heart function, potential outflow tract obstruction and the presence of a pericardial effusion. It also can be used to obtain Doppler measurements of CO. However, TEE is operator dependent, and errors in the measurement of the diameter of the left ventricular outflow tract skew results. Presence of an irregular HR also affects the final calculation of CO.

Use of intraoperative TEE on renal allograft outcomes has only been evaluated in one study, in which living donor renal transplants were examined. Guiding fluid therapy in 110 patients by corrected flow time obtained with continuous transesophageal Doppler (TED) monitoring did not improve immediate graft function compared to fluid administration guided by CVP. However, significantly less fluid was used in the TED-monitored group, and the incidence of postoperative complications related to fluid overload was reduced [66].

Transesophageal eechocardiography does not permit continuous monitoring for instantaneous evaluation of fluid responsiveness. Transesophageal echocardiography is also time consuming and expensive, and requires intensive operator training. Careful attention must be given to the limitations of the assessment of CO by TEE, which may be beyond the level of expertise of a physician with only moderate experience. New studies with more robust data are required to support the use of TEE for fluid management in this patient population.

Cardiac output alone can be a poor predictor of fluid responsiveness, although its response to a preload-modifying maneuver such as mechanical ventilation can provide useful information about fluid responsiveness without the administration of fluid [67].

Currently there are several noninvasive CO monitoring technologies including pulse contour analysis (PCA), pulse wave transit time, thoracic electrical bioimpedance/bioreactance and CO₂ rebreathing [68]. We will focus on parameters derived from technology using PCA to calculate CO. The SV is derived from the arterial waveform signal using the area under the systolic part of the arterial pressure curve. There are few devices using PCA technology: FloTrac/Vigileo/EV1000 system (Edwards Lifesciences, Irvine, CA, USA), LiDCOrapid (LiDCO, Cambridge, UK), MostCare (Vygon, Vytech, Padua, Italy) and ProAQT/PulsioFlex (Pulsion/MAQUET, Rastatt, Germany). Although the absolute CO values may not be as accurate as those derived by pulmonary artery catheter, these devices are very useful in providing acceptable trending abilities to follow changes in CO and SV. More studies are warranted to determine the effectiveness of these devices in improving patient outcomes [69].

Dynamic variation of arterial waveform-derived parameters (i.e., systolic pressure variation (SPV), pulse pressure variation (PPV) and stroke volume variation (SVV)) in mechanically ventilated patients provides a precise indication of fluid responsiveness, particularly when compared to static indices [59]. However, SPV, PPV and SVV only can predict variations in the stroke volume index and cardiac index when the patient receives controlled ventilation and has no spontaneous respiratory efforts (Fig. 2). The percentages by which both parameters are likely to increase can also be predicted.

The fundamentals underlying these parameters are based on simple physiology and are well represented in the Frank–Starling curve (Fig. 3). The cyclic positive-pressure ventilation intermittently increases pleural pressure and alters the filling conditions of the left and right ventricles. The increase in pleural pressure decreases venous return and right ventricle (RV) preload. Additionally, the RV afterload can increase due to an inspiratory rise in transpulmonary pressure. The decrease in the RV preload and the increase in the RV afterload decrease the RV SV, with minimum values reached end-inspiratory. The inspiratory reduction in the RV ejection will in a few beats decrease left ventricle (LV) preload and SV. The respiratory changes in the LV SV indicate a biventricular preload dependence and only occur in the ascending portion of the Frank–Starling curve [70].

Dynamic analyses of flow parameters need to be restricted to patients under controlled mechanical ventilation with tidal volumes between 8 and 10 ml/kg ideal body weight because heart–lung interactions are more complicated with spontaneous breathing efforts with or without a ventilator [71]. When right ventricular insufficiency is present, an experimental study of anesthetized pigs indicated that PPV and SVV only can be used to assess volume responsiveness during a volume-loading maneuver [72].

The use of PPV may be preferable to SPV and SVV because it is measured directly from the arterial pressure tracing and uses more advanced digital software [59]. Despite limitations in the use of dynamic indices and the inability to assess global ventricular function, SPV, PPV and SVV are currently the most precise predictors of fluid responsiveness [59], and their use does not require any specific training or the physical presence of the clinician during measurements.

In a recent study from Japan, SVV was compared with CVP and pulmonary artery diastolic pressure as estimates of the right and left ventricular preload in patients undergoing renal transplantation. As expected, SVV better predicted volume responsiveness [60]. A retrospective study from 2014 questioned whether SVV can be an alternative to CVP in kidney transplant recipients [73].

An overview of studies monitoring and targeting fluid therapy in kidney transplantation and their main outcomes is presented in Table 1. Most trials have been retrospective cohort studies. Errors due to bias and confounding variables are common and likely affect the results obtained using this study design. More robust clinical trials are necessary to better understand fluid management in kidney recipients.

Accumulating evidence supports the concept that fluid therapy should be individualized and based on dynamic indices of the intravascular volume. Conventional monitoring does not provide sufficient information for this approach, but novel technologies may allow titration of therapeutic interventions to ensure that oxygen delivery matches tissue need. One approach using dynamic indices has been goal-directed therapy (GDT). Goal-directed therapy decreased postoperative complications, hospital length of stay, mortality and hospital costs in high-risk surgical patients [17‐19]. However, Pearse et al. [74] found no decrease in mortality after use of a CO-guided hemodynamic therapy algorithm compared with usual care. Goal-directed therapy protocols are often associated with an increased infusion volume, and often include significantly more colloids than crystalloids [75]. A factor that needs to be discussed in fluid management is not only the total amount of fluid infused but how, when and to which patient the volume is administered. There are no studies of GDT in kidney transplantation.

In contrast to GDT, which is guided by “goals”, flow-directed therapy dictates fluid administration based on “triggers” [62]. The triggers indicate that the clinical state has fallen below an acceptable level and may benefit from increased volume with an increase in CO. It does not aim for a higher than normal target, but rather is based on the clinical impression that the flow is inadequate for tissue needs and asks whether a given volume increased flow [62]. The change in CO is used as an indicator of the effectiveness of the therapy and not as the primary objective [62]. Thus, when the patient reaches the plateau of the Frank–Starling curve and the SV does not show a change after a fluid bolus, volume expansion is no longer the therapy of choice. Disregard for the status of fluid responsiveness before administering fluids can lead to the unnecessary delivery of fluids, even in GDT protocols. We advocate a flow-directed responsive protocol to be used in the management of perioperative fluid in kidney transplantation. Noninvasive CO monitoring is used to measure perfusion and assess fluid responsiveness when clinically indicated. After a volume is given (500 ml of crystalloid) the response of CO is checked. A 15% increase in CO with a CVP rise of at least 2 mmHg constitutes a positive response. When there is no appropriate response other therapies should be chosen (i.e., vasopressor and/or inotropic therapy to treat hypotension).