Introduction

Acute kidney injury (AKI) occurs commonly in cancer patients and independently increases morbidity and mortality. ^, Despite impressive gains in the understanding of the basic pathophysiologic principles underlying kidney injury, there are no therapeutic options to prevent or ameliorate AKI; treatment consists of supportive care and avoidance of nephrotoxic agents, such as radiocontrast and nonsteroidal antiinflammatory agents.

Patients with cancer are at risk for the development of AKI from causes similar to all hospitalized patients, such as radiocontrast administration, hypotension, antibiotics, infections, and sepsis. Unique causes of AKI in these patients include chemotherapy exposure, tumor lysis syndrome, hematopoietic stem cell transplantation, irradiation, and direct effects of malignancy. In a cross-sectional analysis of prospectively collected data on 3358 patients admitted to a large United States cancer center, 12% developed AKI based on the RIFLE (Risk, Injury, Failure, Loss of kidney function, and End-stage renal disease [ESRD]) creatinine criteria, of whom 4% received dialysis. ^, In a multivariate analysis, the development of AKI was associated with a significantly increased odds of death (odds ratio [OR] 4.72; 95% confidence interval [CI], 3.3–6.7).

At a certain point in the course of AKI, use of renal replacement therapy (RRT) may be considered. Although the literature is sparse and somewhat contradictory, most studies show that in patients started on RRT for AKI, survival is significantly lower for those with cancer compared with those without cancer; survival is particularly poor with hematologic malignancies. In a retrospective study of 309 cancer patients with AKI (based on criteria proposed by Bellomo et al.), an increased risk of mortality was associated with age greater than 60 years, dysfunction of more than two organs, impaired performance status, and uncontrolled cancer. ^, In another analysis, the risk of mortality in 345 patients with hematologic cancer and AKI (based on RIFLE criteria) was significantly associated with septic shock, mechanical ventilation, and allogeneic stem cell transplantation. Patients with all three risk factors had a mortality rate of 86%. Based on such findings, it is somewhat controversial whether initiating RRT in patients with multiple organ dysfunction and uncontrolled cancer is appropriate because RRT is unlikely to change the ultimate outcome. This concern has triggered an interest in “palliative nephrology.” For such patients, it may be reasonable to offer a time limited trial of RRT and discontinue treatment if there is no significant improvement in clinical status. Currently, there are no strict guidelines that define what is a reasonable time limit. The Renal Physicians Association, however, has published general guidelines on shared decision making in the appropriate initiation and withdrawal from dialysis.

Although AKI occurs in patients in all hospital units, the highest frequency is in intensive care units (ICUs) and the majority of research described in this chapter relates to critically ill patients. Several critical issues regarding the use of RRT remain controversial and are outlined in Box 32.1 .

Initiation of renal replacement therapy (timing)

The classic “indications” for initiating RRT in a patient with AKI are listed in Box 32.2 . It is misleading to refer to these clinical conditions as indications because it implies that RRT should only be started when such criteria are present. Reliance on these criteria could delay appropriate therapy, resulting in serious adverse events in patients. Rather, the conditions listed should necessitate emergent RRT, unless palliative care measures are planned.

In the case of lesser degrees of kidney injury, the timing of RRT remains a contentious issue. On the one hand, early initiation would avoid the development of any serious complication of AKI; however, the early use of RRT could expose patients to the potential harm of RRT, when it might have been unnecessary ( Box 32.3 ).

Two retrospective studies divided patients into “early versus late” groups based on the median blood urea nitrogen (BUN) concentration when RRT was started and found a survival advantage in the early dialysis group. ^, In addition, a metaanalysis also reported a benefit to earlier initiation of RRT. Unfortunately, the overall data quality is poor and does not determine when RRT should actually be started.

More recently, two randomized controlled trials on the timing of RRT initiation have been published. In the multicenter trial reported by the AKIKI (Artificial Kidney Initiation in Kidney Injury) study group, 620 patients with Kidney Disease Improving Global Outcomes (KDIGO) stage 3 AKI (≥ threefold increase in baseline serum creatinine or urinary output < 0.3 mL/kg/h for ≥ 24 hours) were randomized to early RRT (at enrollment) or delayed RRT (development of hyperkalemia, severe metabolic acidosis, hypervolemia refractory to diuretics, BUN ≥ 112 mg/dL, or oliguria ≥ 72 hours). ^, There was no difference between the early and delayed groups in the primary endpoint of 60 day mortality (48.5% vs. 49.7%, respectively). Importantly, 49% of the delayed group never received RRT, and diuresis, a marker of improved kidney function, occurred significantly earlier. The rate of catheter-related blood stream infections was significantly higher in the early strategy group.

The ELAIN study (Effect of Early vs. Delayed Initiation of Renal Replacement Therapy on Mortality in Critically Ill Patients With Acute Kidney Injury) randomized 231 patients at a single center with KDIGO stage 2 AKI (≥ twofold increase in baseline serum creatinine or urinary output < 0.5 mL/kg/h for ≥ 12 hours) to either early RRT (within 8 hours of diagnosis of KDIGO stage 2 AKI) or delayed RRT (within 12 hours of developing KDIGO stage 3 AKI). Patients in the early group compared with the delayed group had improved survival at 90 days (39.3% vs. 54.7%, respectively). In addition, more patients in the early group recovered kidney function at 90 days compared with the delayed group.

These discordant results are not unexpected. Both studies had a small number of patients, numerous confounders, enrolled patients at different stages of kidney dysfunction, and relied on arbitrary indications for RRT. The two trials also differed in the population studied, where most patients in AKIKI were medical as opposed to those in ELAIN, who were mainly surgical. Whether this difference makes one trial more applicable to cancer patients than the other is unknown. Therefore initiation of RRT should be individualized to each patient, taking into consideration several factors, including fluid balance, severity of multiple organ dysfunction, urinary output, age, and comorbid conditions.

Discontinuation of renal replacement therapy

Similar to the circumstance of initiating RRT, there are little published data in regards to stopping RRT. Given the complexity of patients with AKI, especially those who are critically ill, it is difficult to imagine devising a generalizable “weaning” plan similar to those used in mechanically ventilated patients. It is also probable that it is easier to determine when RRT can be successfully terminated in patients on intermittent therapy because trends in creatinine levels can be followed.

A retrospective observational case-controlled study was conducted in 304 postoperative patients with AKI receiving intermittent RRT. A total of 94 patients (30.9%) were successfully weaned from RRT for more than 5 days and 64 (21.1%) were free of RRT for at least 30 days. Independent predictors for resuming RRT within 30 days were: (1) longer duration of RRT; (2) higher severity of illness scores on the day of stopping RRT; (3) oliguria (< 100 mL/8 hours) 1 day after stopping RRT; and (4) age over 65 years.

A post-hoc analysis of a prospective observational trial of 529 patients on continuous RRT for AKI was performed to identify variables associated with successful discontinuation. Success was defined as no further RRT for at least 7 days. Multivariate analysis identified increased urine output and decreased creatinine levels as predictors of successful discontinuation. However, the predictability of urine output was significantly reduced by the use of diuretics.

In an anonymous electronic survey of practicing nephrologists treating AKI, the most common criteria used to stop RRT was urine output (51%). Other factors included resolution of hypervolemia (29%), decreased creatinine levels (27%), and correction of hyperkalemia (21%). Overall, however, there was considerable variability in the survey regarding what criteria was used to stop RRT.

It appears that the majority of physicians rely on improvement in urinary output as an indication for stopping dialysis. In those patients with nonoliguric AKI, decreasing creatinine levels and improvement in overall clinical status appears to influence the decision to discontinue RRT.

Modalities

The major objectives of RRT are correction of electrolyte and acid-base disorders, solute clearance, and ultrafiltration (volume removal.) There are two basic methods of solute removal from the blood with RRT: diffusion via dialysis, and convection, or solvent drag, using hemofiltration.

Methods

Once the decision has been made on when to start RRT and what form of clearance will be used, it must be determined what type of delivery method will be used ( Table 32.1 ). Typically, RRT is divided into three major categories: intermittent (IRRT); continuous (CRRT); and prolonged intermittent (PIRRT).

Peritoneal dialysis (PD), on the other hand, is not routinely prescribed in the acute setting. In the past, PD was widely used because of its easy availability and low cost. With the advent of improved technology in IRRT and CRRT, PD fell out of favor. Decline in the use of PD was also coupled with the concerns of risk of peritonitis, impaired ventilation from abdominal distension, variable volume control, and hyperglycemia from dextrose-based PD fluid. More recently, the use of PD in the treatment of AKI has experienced a renaissance. Renewed interest results from data, mostly reported from countries dependent on PD, as the only method of RRT available to critically ill patients, which demonstrate the effectiveness of acute PD in treating AKI. Several studies in acutely ill patients with AKI have shown that acute PD is not inferior to both daily IRRT and CRRT in terms of metabolic and volume control, and survival. However, such studies are small and require verification in larger trials. Nevertheless, these results led to the International Society of Peritoneal Dialysis guideline that PD should be considered as a suitable method of RRT in patients with AKI. Whether the use of PD in the setting of AKI will increase substantially likely depends on the local expertise of treating physicians.

CRRT is performed as either continuous veno-venous hemofiltration or hemodialysis (CVVH and CVVHD, respectively). Although CRRT on a minute by minute basis is less efficient than IRRT because of lower flow rates, it provides excellent volume and solute removal because of its continuous application.

It was thought that CRRT would be superior to IRRT as a modality in critically ill patients because of improved hemodynamic stability, safer volume removal, better acid-base and electrolyte balance, and the ability to give more nutritional supplementation. However, despite these apparent advantages, several randomized controlled trials were unable to demonstrate any benefit to CRRT. It has been argued that this lack of superiority was caused by study design, where the sickest patients were excluded from participation, thereby creating bias. On the other hand, another interpretation is a failure to recognize the potential negative effects from CRRT that could negate its positive attributes as listed in Box 32.4 . In a study from the Cleveland Clinic, 27% of patients on CRRT experienced severe hypophosphatemia and its development was associated with a significantly increased risk of respiratory failure necessitating tracheostomy. Furthermore, the most common cause of AKI in critically ill cancer patients is sepsis, and little is currently known about the appropriate dosing of antibiotics in patients receiving CRRT. In a study of 70 patients on CRRT given intravenous vancomycin, drug trough levels were below the recommended threshold in 50% of patients. This finding was independent of location (medical vs. surgical intensive care) and prescriber (physician vs. pharmacist). Because vancomycin levels are easily available, this finding suggests many patients on CRRT may be receiving inadequate doses of other antibiotics where drug levels are not available.

PIRRT, as shown in Table 32.1 , is a hybrid therapy of CRRT and intermittent hemodialysis (IHD). The major advantage of PIRRT is freeing the patient for competing procedures, such as imaging studies and surgical interventions, while providing adequate clearance and hemodynamic stability; it also reduces nursing workload. A metaanalysis of seven randomized and 10 observational studies demonstrated that PIRRT is not inferior to CRRT in terms of mortality, fluid removal, and kidney recovery. The choice of PIRRT or CRRT is usually based on local logistics; some view PIRRT as a transition step between CRRT and IHD.

For now, the debate on CRRT/PIRRT versus IRRT continues, although CRRT/PIRRT certainly has a role in the care of a select group of patients. Reasonable guidelines for selecting CRRT/PIRRT are listed in Box 32.5 . Patients especially suited for CRRT/PIRRT are those with cerebral edema associated with hepatic encephalopathy or neurogenic edema. ^, Rapid solute removal with IRRT in these situations is associated with increased intracranial pressure and reduced cerebral perfusion.

Intensity of dialysis

Other than control of metabolic and volume disturbances in patients with AKI, there is the issue of how much dialysis does a patient need, or the concept of dialysis dose. An early study of dialysis intensity in stable outpatient IRRT patients showed that the amount of solute clearance (as measured by the percentage decline in the initial BUN concentration) was more predictive of morbidity and mortality than duration of dialysis. This landmark study led to the concept of urea kinetic modeling as a means to assess “adequate dialysis.” In essence, it showed that simply looking at the BUN concentration as a marker of “good” dialysis was severely flawed, because BUN levels are affected by numerous nonkidney factors, such as protein intake, catabolic rate, and medications. What mattered was the percent reduction in the BUN concentration. Adequate dialysis is a greater than 65% reduction in the BUN level at the end of treatment regardless of the initial value. Lesser amounts of reduction in stable outpatient IRRT patients are associated with significantly higher morbidity and mortality rates. In fact, urea reduction ratio (URR) is mandatorily followed in dialysis clinics, as a measure of quality of care.

If the URR is a good measure of adequate dialysis in ESRD patients, what about unstable or critically ill patients with AKI needing RRT? In other words, if “dose” matters in ESRD, does it matter in AKI and how do you measure it? Inherent in the issue is that the URR was only validated in the ESRD population and not patients with AKI.

With this as a background, in the early 2000s, there arose great interest in assessing the “dose” of RRT in critically ill patients with AKI. In a trial by Ronco et al., patients receiving CVVH for ICU acquired AKI were randomized to low-dose (20 mL/kg/h) or high-dose (35–45 mL/kg/h) replacement fluid rates. Patients in the higher-dose group had significantly better survival rates. In another trial of patients receiving IRRT for AKI, Schiffl et al. showed that patients receiving daily dialysis had better survival rates compared with those who received dialysis on an every-other day basis. Based on these findings and other supportive retrospective studies, higher doses of RRT for critically ill patients were strongly encouraged. However, there were several problems with this recommendation: (1) control groups may have been “underdialyzed;” for example, in the Schiffl study, the mean URR during each treatment was below 60%; (2) demographics of the study patients were not reflective of those usually seen in the ICU; in the Ronco trial, 85% of patients were surgical and only 15% had sepsis; (3) volume control was not standardized; and 4) most studies had small numbers of patients and were underpowered.

Based on clinical equipoise, the Veterans Affairs/National Institutes of Health (VA/NIH) Consortium embarked on the ambitious study Acute Renal Failure Trial Network Study (ATN) to address the question of RRT adequacy in ICU patients with AKI. Patients were randomized to either high dose or usual dose dialysis until death, recovery, discharge, or day 30 of hospitalization. Furthermore, modality (CRRT or IRRT) was determined by the cardiac sequential organ failure assessment (SOFA) score. If patients were considered hemodynamically unstable (cardiac SOFA score of 3 or 4), they received CRRT. Otherwise IRRT was performed. Patients switched between modalities as their cardiac SOFA score changed; however, they remained in the same dosing arm. High-dose CRRT was 35 mL/kg/h of dialysate/replacement fluid, whereas usual dose was 20 mL/kg/h. High-dose IRRT was six treatments weekly and usual dose was three weekly treatments. Each treatment was required to achieve a URR of more than 65%. The study randomized more than 1000 patients with a 90% power to detect a 10% absolute reduction in mortality rate with an expected mortality rate of 55%. The study found there was no survival benefit to intensive dialysis in either the entire group or in any predefined subgroup of patients. Note that all patients achieved a URR greater than 65% during IRRT treatments and the CRRT dose (defined as either hours on machine or quantity of used fluids) was achieved in 90% of cases. Therefore “underdosing” of dialysis did not occur.

Likewise, in the RENAL (Randomized Evaluation of Normal vs. Augmented Levels of Renal Replacement) trial with more than 1500 AKI patients in the ICU with similar demographics to the ATN study, randomization to high (40 mL/kg/h) versus usual (25 mL/kg/h) dose CRRT did not confer any survival benefit.

These two trials clearly demonstrate that if an adequate dose of dialysis is delivered to critically ill patients with AKI, higher doses are unnecessary. It is vital to remember that adequate dosing is not what is prescribed but what is achieved. Barriers to achieving the prescribed dose include poor catheter function, filter/blood line clotting, competing procedures (abdominal washouts, radiology procedures), and morbid obesity. Therefore patients with ICU-associated AKI can be safely treated with thrice weekly IRRT (as long as the URR is measured and a target of > 65% is achieved) or CRRT with a dose of 20 to 25 mL/kg/h, as long as they receive at least 22 hours (90% of dose) of therapy per day.