Goals of and Indications for Renal Replacement Therapy

The treatment of AKI with RRT has several interrelated goals: (1) to maintain fluid and electrolyte, acid–base, and solute homeostasis; (2) to prevent complications of uremia; and (3) to allow other supportive measures (e.g., antibiotics, nutrition support) to proceed without limitation. Ideally, therapeutic interventions should be designed to achieve those goals while minimizing the risk for further insults to the kidney and facilitating recovery of kidney function. In practice, the use of RRT is commonly based on physician preferences and experience. No evidence-based criteria have been established to guide modality choice, thereby making comparisons among centers or strategies at the same or different institutions difficult. An important consideration is to recognize that patients with AKI are distinct from those with end-stage renal disease (ESRD). The rapid decline of kidney function in AKI does not permit the adaptive responses that characterize the course of the patient with chronic kidney disease (CKD). Consequently, the traditional indications for renal replacement, developed for patients with advanced CKD, are not necessarily valid in this context ( Box 49.1 ). For instance, massive volume overload, resulting from volume resuscitation, a common strategy used for MOF, may be an indication for dialysis, even in the absence of significant elevations in blood urea nitrogen (BUN) or plasma creatinine. In this instance, it may be more appropriate to consider RRT in the ICU patient as a form of renal support as opposed to “replacement.” Indeed, some of the traditional indications for dialysis (e.g., uremic pericarditis, pleuritis, or other serositis) would be considered “rescue indications” suggesting undue delay in initiation of RRT.

Modalities of Renal Replacement Therapy

The most common RRT modalities used for AKI include IHD, prolonged intermittent renal replacement therapy (PIRRT), continuous renal replacement therapy (CRRT), and peritoneal dialysis (PD) ( Tables 49.1 and 49.2 ). IHD, PIRRT, and CRRT are extracorporeal therapies that require vascular access in the form of a large-bore, double-lumen central venous catheter, whereas PD requires the placement of an intraabdominal dialysis catheter.

The various modalities of RRT use varying proportions of diffusion and convection for solute transport and ultrafiltration, driven either by hydrostatic or osmotic gradients, for volume management. With extracorporeal therapies, some solutes may also be removed by adsorption onto the membrane. The relative contributions of diffusion, convection, and adsorption depend on the solute, the membrane, the geometry of the dialyzer, and operating conditions such as blood and dialysate flow rates and ultrafiltration rate. Ultrafiltration, which is used for volume management and which drives convective clearance, occurs by application of a hydrostatic pressure across the membrane from the blood to effluent side in IHD, PIRRT, and CRRT and by an osmotic gradient in PD. In diffusion, blood is exposed to dialysate across a semipermeable membrane and solute transfer across the membrane is driven by the concentration gradient between blood and dialysate. With extracorporeal therapies, the concentration gradient is maximized and maintained throughout the length of the membrane by running the dialysate countercurrent to the blood flow. Solute clearance is dependent on molecular size, membrane characteristics such as pore size, charge, water permeability, and extracorporeal circuit factors such as dialysate and blood flow rates. Small molecular weight solutes (<500 Daltons) are cleared efficiently by diffusion because their mobility in solution is high. As molecular weight increases, the mobility of solutes in solution diminish and diffusive clearance decreases. Convective clearance occurs when application of a hydrostatic pressure or osmotic gradient across a semipermeable membrane forces solvent (i.e., plasma water) across the membrane by ultrafiltration and ‘‘drags’’ with it small- and middle-molecular-weight solutes by bulk flow. The size of solutes that can be removed by convection is determined primarily by the size selectivity and charge characteristics of the membrane. For solutes that are smaller than the size cutoff of the membrane, the major determinant of convective clearance is the rate at which ultrafiltration occurs.

Timing of Initiation of Renal Replacement Therapy

Although whether or not to provide dialytic support with RRT and when to start are two of the most fundamental questions facing nephrologists and intensivists in caring for patients with AKI, the optimal timing of acute RRT for AKI is not defined. An improvement in survival associated with early initiation of RRT was first suggested by case series with historical controls conducted in the 1960s and 1970s. However, given that BUN concentrations at the start of dialysis in the “early” treatment groups in these studies were high by modern standards, their relevance to current practice is limited. Moreover, over the last 30 years, the severity of illness of hospitalized patients continues to increase, especially among the critically ill population, and life support technologies have rapidly progressed, allowing for improving survival, which further limits comparisons of distantly historical studies. More recent single-center observational studies that were restricted to AKI after trauma and coronary artery bypass surgery suggested a benefit to dialysis initiation at lower BUN concentrations. A prospective multicenter observational cohort study performed by the Program to Improve Care in Acute Renal Disease (PICARD) analyzed dialysis initiation—as inferred by BUN concentration—in 243 patients from five geographically and ethnically diverse clinical sites. Survival rates were slightly lower for patients who started dialysis at higher BUN concentrations, despite a lesser burden of organ system failure. Adjusting for age, hepatic failure, sepsis, thrombocytopenia, and serum creatinine and stratified by site and initial dialysis modality, initiation of dialysis at higher BUN was associated with an 85% increased relative risk (RR) for death (RR = 1.85; 95% CI, 1.16 to 2.96).

Although the maintenance of BUN concentrations below arbitrarily set levels is usually a reference for starting dialysis treatment, BUN reflects factors not directly associated with kidney function, such as catabolic rate, nutritional intake (especially protein intake), and volume status. Plasma creatinine is influenced by age, race, muscle mass, and catabolic rate, and its concentration is also affected by changes in extracellular volume status (decreasing as total body volume increases). In a prospective multicenter observational study conducted at 54 ICUs in 23 countries, timing of RRT was stratified into “early” or “late” by median urea at the time RRT started and also categorized temporally from ICU admission into early (less than 2 days), delayed (between 2 and 5 days), or late (more than 5 days). Timing by serum urea indicated no significant difference in mortality (63.4% for urea ≤ 24.2 mmol/L vs. 61.4% for urea > 24.2 mmol/L). However, when timing was analyzed relative to ICU admission, late RRT was associated with greater crude mortality (72.8% late vs. 62.3% delayed vs. 59% early; P = 0.001) and covariate-adjusted mortality (OR, 1.95; 95% CI, 1.30 to 2.92; P = 0.001). Overall, late RRT was associated with a longer duration of RRT and a longer stay in hospital and greater dialysis dependence, which was attributed to accumulation of volume, worsening acidemia, and/or life-threatening electrolyte derangements. In addition, timing of initiation of RRT relative to ICU admission may also reflect differences with regard to whether AKI was present at ICU admission or developed as a complication during ICU admission, which may also influence these outcomes.

An important methodological flaw underlies these observational studies and limits their interpretation. By only including patients who actually received RRT rather than all patients in whom AKI was present and might have been considered as candidates for early RRT, these observational cohorts excluded patients with early AKI who either recovered kidney function or died without ever receiving RRT. Because the clinical question is not early versus late initiation of RRT, but early initiation of RRT versus a delayed approach to management, failure to include these patients who do not receive RRT may provide a biased assessment of the benefits of early initiation of RRT.

Several randomized controlled trials have attempted to address this issue ( Table 49.3 ). In an older study of critically ill patients, Bouman et al. randomly assigned 106 critically ill patients with AKI to early or late initiation of dialysis. The early initiation group started CVVH within 12 hours of developing a low urine output (less than 30 mL/h for 6 hours, not responding to diuretics or hemodynamic optimization) or a creatinine clearance less than 20 mL/min. CVVH was started in the late initiation group when classic indications were met. Although underpowered to detect survival differences, the study did not find differences in ICU or hospital mortality between the early and late groups or in renal recovery among survivors.

Two more recent clinical trials—Artificial Kidney Initiation in Kidney Injury (AKIKI) and Early Versus Late Initiation of Renal Replacement Therapy in Critically Ill Patients with Acute Kidney Injury (ELAIN) —reached divergent conclusions. The AKIKI trial was a multicenter randomized controlled trial conducted in France that assigned 620 patients who reached Kidney Disease: Improving Global Outcomes (KDIGO) stage 3 AKI to either immediate initiation of RRT or a delayed initiation strategy in which RRT was not begun until specific metabolic parameters or life-threatening indications developed. Ninety-eight percent of the patients in the early arm received RRT compared with only 51% of the patients randomized to the delayed therapy arm. The median time from development of stage 3 AKI to initiation of RRT was 4.3 hours in the early treatment arm and 57 hours in the delayed arm; the predominant modality of RRT used was IHD. There was no difference in survival at 60 days between the early and delayed RRT groups (48.5% vs. 49.7%; P = 0.79). Comparing the patients who received RRT, early RRT was associated with lower mortality compared with the patients initiated late (48.5% vs. 61.8% at 60 days); however, those patients in the delayed initiation arm who did not receive RRT had a substantially lower mortality (37.1% at 60 days) than those patients who received RRT. Although these patients represented a lower risk population, with lower severity of illness at randomization, the inability to prospectively identify patients who ultimately do not receive RRT illustrates the need to include them in analyses of early versus delayed strategies of initiation.

The ELAIN trial was a single-center study that randomly assigned 231 patients with KDIGO stage 2 AKI to either immediate RRT or delayed initiation until they reached KDIGO stage 3 AKI or developed life-threatening indications. CVVHDF was the modality of RRT used in all patients and only 9 of 118 in the delayed group did not receive RRT. The median time from reaching KDIGO stage 2 AKI to initiation of RRT was 6.0 hours in the early RRT arm compared with 25.5 hours in the late arm. Mortality at 90 days was 39.3% in the early RRT group versus 54.7% in the delayed group ( P = 0.03) with a hazard ratio (HR) for death of 0.66 (95% CI, 0.45 to 0.97) in the early group. Two additional multicenter trials (Standard Versus Accelerated Initiation of Renal Replacement Therapy in AKI [STARRT-AKI; ClinicalTrial.gov #NCT01557361 ] and Initiation of Dialysis Early Versus Delayed in Intensive Care Unit [IDEAL-ICU; ClinicalTrial.gov #NCT01682590 ]) are ongoing. Pilot data from 100 participants enrolled in the pilot phase of the STARRT-AKI trial indicated a separation in time from eligibility in the two arms of approximately 24 hours with no significant difference in mortality between the two treatment groups (38% in the accelerated arm vs. 37% in the standard arm).

The role of volume overload as an independent criterion for initiation of RRT in critically ill patients with AKI should be noted. Over the last 10 to 15 years, multiple studies in critically ill patients have demonstrated that volume overload is detrimental to critically ill patients both with and without AKI, contributing to mortality and to other morbidities such as duration of ICU length of stay, time on mechanical ventilation, and intraabdominal hypertension. Almost all patients requiring acute RRT, especially in the ICU, have some degree of volume accumulation or frank volume overload, and there is a strong association between volume overload at time of RRT initiation and increased risk for death in both adults and pediatric populations. Furthermore, retrospective data suggest that the excess hospital mortality seen in AKI seems primarily driven by fluid overload, hyperkalemia, and metabolic acidosis rather than by azotemia.

Much of the uncertainty regarding timing of initiation of RRT rests with difficulty that is often present in assessing the trajectory of AKI in an individual patient. Will oliguria be transient or prolonged? Will kidney function begin to recover in a matter of hours or days, or will the AKI be prolonged? Novel biomarkers of kidney injury such as neutrophil gelatinase-associated lipocalin (NGAL), tissue inhibitor of metalloproteinase-2 (TIMP-2), and insulin-like growth factor–binding protein 7 (IGFBP7) have been proposed as prognostic markers; however, further validation is required for widespread adoption in clinical decision-making. The role of functional testing, such as the furosemide stress test, also has potential but requires further validation.

In considering the optimal timing of initiation of RRT in AKI, safety concerns associated with RRT need to be considered. Whether blood based or peritoneal membrane based, acute RRT requires access insertion and maintenance, both of which can cause visceral and vascular injury, including bleeding, vascular laceration, acute arteriovenous fistula formation, pneumothorax, and abdominal wall or mesenteric injury, as well as infection, although the risks associated with vascular catheter insertion have been reduced with the adoption of bedside ultrasound guidance for insertion. In addition, RRT is associated with risks of hypotension and leukocyte and platelet activation from exposure to the extracorporeal circuit, both of which may delay recovery of kidney function.

Thus the clinical decision to initiate RRT must balance both uncertainty regarding risks and benefits for the individual patient. The KDIGO Clinical Practice Guidelines for AKI have encapsulated this uncertainty in the ungraded recommendations to “initiate RRT emergently when life-threatening changes in fluid, electrolyte, and acid-base balance exist;” but to also “consider the broader clinical context, the presence of conditions that can be modified with RRT, and trends of laboratory tests—rather than single BUN and creatinine thresholds alone—when making the decision to start RRT.”

Selection of Modality of Renal Replacement Therapy

Multiple factors need to be taken into consideration in the selection of modality of RRT, including clinical setting, hemodynamic status, volume status, medication dosing, nutritional needs and the nursing and medical expertise and resources and available. Based on randomized controlled trials and metaanalyses, CRRT and IHD are generally considered to be complementary therapies with no clear evidence that either modality provides a survival advantage. However, CRRT may be preferable to IHD in specific clinical settings. In the setting of significant acute brain injury or fulminant hepatic failure, CRRT may better protect cerebral perfusion by avoiding the rapid shifts in blood osmolality which are associated with the more rapid solute clearance during IHD and may lead to increased intracranial pressure and decreased cerebral perfusion pressure. Furthermore, CRRT has been associated with less need for escalation of vasopressors and with achievement of greater net negative fluid balance. Therefore CRRT may be better suited for some critically ill AKI patients who cannot not tolerate fluid shifts and fluctuations in metabolic parameters.

The effect of modality of RRT on recovery of kidney function has been debated. Although observational studies have found CRRT to be associated with higher rates of renal recovery, the evidence is insufficient and most studies only evaluated renal recovery in patients who survived. Studies analyzing combined mortality and nonrecovery of renal function between groups have found no difference in recovery of kidney function. In a retrospective propensity matched cohort, Wald et al. compared CRRT with IHD as the initial RRT modality in critically ill adults with AKI and found that CRRT was associated with a lower likelihood of chronic dialysis (HR 0.75; 95% CI, 0.65 to 0.87). However, a subsequent retrospective cohort study of 638 critically ill patients who received RRT for AKI and survived to hospital discharge or 90 days found no difference in renal recovery based on initial RRT modality. Multiple randomized controlled trials have analyzed recovery of kidney function as a secondary endpoint and have found no difference in recovery of kidney function, based on dialysis independence, as a function of modality of RRT. A metaanalysis of predominantly observational studies reported CRRT as the initial RRT modality to be associated with higher rates of renal recovery among survivors; however, when the analysis was limited to randomized controlled trials, no benefit with regard to renal recovery was observed.

It has been postulated that continuous hemofiltration might have immunomodulatory benefits based on removal of proinflammatory mediators such as interleukin (IL)-6, IL-8, IL-1, and tumor necrosis factor by convection or by adsorption to the membrane. Despite some encouraging results, the clinical benefit of conventional CRRT in sepsis has been disappointing and the preferential use of convective as opposed to diffusive modalities cannot be supported by the data. A number of factors may contribute to this absence of benefit, including concomitant removal of both pro- and antiinflammatory mediators and relatively low extracorporeal cytokine clearance compared with endogenous clearance rates.

Efforts at further augmenting cytokine clearance have used higher hemofiltration volumes, high-cutoff membranes (with a molecular weight range of up to 60 kDa), and coupled plasma filtration and adsorption (CPFA). In a pilot study using a high-cutoff membrane for hemofiltration in septic AKI, clearance rates for IL-6 and IL-1 were found to be augmented and vasopressor requirements decreased ; however, benefits with regard to patient survival could not be demonstrated. Similarly, compared with conventional CRRT, CPFA in septic shock improves hemodynamics and decreases need for vasopressor support but has not been found to reduce mortality or other important clinical outcomes. Similarly, the High Volume in Intensive Care (IVOIRE) and HEROICS trials were unable to demonstrate benefit to high-volume (70 to 80 mL/kg/h) hemofiltration in either sepsis-associated AKI or post–cardiac surgery cardiogenic shock compared with conventional management. Thus use of these novel approaches to CRRT cannot be recommended for clinical practice.

There are limited data comparing other modalities of RRT in AKI. The few studies comparing IHD with PIRRT have not found a clear modality benefit. Studies comparing PIRRT with CRRT have found that PIRRT provides comparable hemodynamic stability and solute control to CRRT with no significant differences in mortality. A metaanalysis of studies comparing CRRT and PIRRT indicated no difference in mortality, recovery of kidney function, fluid removal, days in the intensive care unit, or solute control.

Several small randomized controlled trials have compared PD with other RRT modalities in AKI. In a prospective study performed in Vietnam, 70 patients with AKI caused by either malaria or sepsis (48 and 22 individuals, respectively) were randomly assigned to either PD or CVVH. PD was associated with significantly increased risk for death (47% vs. 15%, OR 5.1, 95% CI, 1.6 to 16). In this study the mortality rate for patients on CVVH was unusually low and PD was not performed using the most recent technological advances such as use of bicarbonate-based dialysate and soft catheters. In contrast, a prospective randomized controlled trial performed in Brazil compared the effect of HVPD and daily hemodialysis (DHD) on AKI patient survival. A total of 120 patients with acute tubular necrosis (ATN) were assigned to HVPD or DHD. The delivered dose, assessed as weekly Kt/V, was 3.67 in HVPD and 4.77 in DHD ( P = 0.01). Metabolic control, mortality rate (58% and 53%), and renal function recovery (28% and 26%) were similar in both groups, whereas HVPD was associated with a significantly shorter time to the recovery of renal function. Another randomized controlled trial comparing high-volume PD to PIRRT found no difference in mortality. A metaanalysis of seven cohort studies and four randomized trials studies comparing PD with extracorporeal therapies in the setting of AKI suggested there were no significant differences in outcomes between patients treated with PD and IHD or hemodiafiltration (HDF).

Because studies comparing modalities of RRT for AKI have not demonstrated superiority of any individual modality with regard to survival or recovery of renal function, selection of RRT modality should be tailored to specific needs of the patient. Most clinicians choose IHD for AKI patients who are hemodynamically stable and CRRT or PIRRT for AKI patients who are hemodynamically unstable, are fluid overloaded, and/or have sepsis and multiorgan failure. Data suggest that given slower rates but more prolonged duration of fluid removal, CRRT may provide more effective volume management than conventional IHD, whereas volume control with CRRT and PIRRT are comparable. Similarly, the slower solute and volume removal with CRRT provides better preservation of cerebral perfusion, making it the preferred treatment compared with IHD in patients with fulminant hepatic failure or cerebral edema. In contradistinction, IHD is favored in patients who need rapid solute removal, such as patients with marked hyperkalemia or drug intoxications. Pragmatically, CRRT and PIRRT require more continuous patient observation and are generally restricted to patients in critical care units. Given the continuous nature of CRRT, treatment interruptions are necessary if the patient needs to be moved for diagnostic imaging or surgical procedures. Breaks in time on treatment with PIRRT and IHD may also facilitate earlier initiation of physical therapy and minimize periods of prolonged immobilization. Transitions between modalities of therapy are common, reflecting the changing needs of patients during their hospital course. Finally, choice of RRT modality must also reflect the modalities available and physician and nursing expertise at the individual hospital as well as considerations of resources, cost, and provider preference.

Dose of Renal Replacement Therapy

Assessment of dose of acute RRT is complex and can be considered as a multidimensional construct encompassing clearance of small solutes (most often modeled based on urea clearance), larger solutes (so-called “middle” molecules), ultrafiltration, treatment duration, and treatment frequency. The relative significance of these factors varies based on the modality of RRT. For example, although treatment duration and frequency are important parameters for intermittent modalities such as IHD and PIRRT, their significance only relates to unplanned interruptions of therapy in the continuous therapies. There are also important interrelationships between duration and frequency of treatment and solute clearance and ultrafiltration: When an intermittent treatment is provided more frequently (e.g., daily IHD), less solute and volume removal during each treatment is required to achieve the same overall dose of therapy than if provided on a less frequent schedule (e.g., every other day or three times per week). In addition, when solute clearance is primarily provided by diffusion, clearance of larger solutes is more dependent on duration and membrane surface area and can be dissociated from clearance of smaller solutes such as urea. Similarly, ultrafiltration may be completely dissociated from solute clearance. Despite this multidimensional complexity, the majority of studies evaluating the relationship between delivered dose of therapy and outcomes have focused primarily on small-solute clearance, primarily modeled as urea clearance ( Table 49.4 ).