By Samuel Nadler, MD, PhD
Critical Care, Pulmonary Medicine, The Polyclinic Madison Center, Seattle; Clinical Instructor, University of Washington, Seattle
Dr. Nadler reports no financial relationships relevant to this field of study.
Acute kidney injury (AKI) is common in patients admitted to the ICU and is associated with high mortality. The decision to initiate renal replacement therapy (RRT) for these patients is complicated. Not only is the optimal time for starting therapy unclear, decisions to transfer patients to higher levels of care often are driven by the perceived need to initiate hemodialysis or continuous RRT.
Criteria For Acute Kidney Injury
Multiple systems to grade AKI have been developed. These include the RIFLE criteria (Risk, Injury, Failure, Loss, End-stage kidney disease), KDIGO criteria (Kidney Disease: Improving Global Outcomes), and AKIN criteria (Acute Kidney Injury Network).1-3 (See Table 1.) Investigators have studied the ability of each of these criteria to predict mortality. The authors of a multicenter, prospective analysis of 3,107 patients admitted to the ICU demonstrated rates of AKI varying from 38.4% with AKIN to 46.9% with RIFLE to 51% with KDIGO.4 The areas under the curve for each receiver operator characteristic (ROC) for in-hospital mortality were 0.746 for AKIN, 0.738 for RIFLE, and 0.757 for KDIGO. Similarly, the authors of a retrospective, observational study of 49,518 patients admitted to the hospital reported ROC for hospital mortality of 0.69, 0.77, and 0.78 for AKIN, RIFLE, and KDIGO, respectively.5 Specifically in patients admitted to the ICU with sepsis, the ROCs for in-hospital mortality with each criterion were similar for each system (RIFLE, 0.652; AKIN, 0.686; KDIGO, 0.658).6
These models of AKI predict mortality, not the need for RRT. This prompted the development of biomarkers that may be more predictive of the need for RRT. Beyond serum creatinine (sCr), these markers include serum cystatin C (sCysC) and urinary neutrophil gelatinase-associated lipocalin (uNGAL), as well as others. The utility of these remains an area of continuing study. The authors of a prospective study of 310 patients receiving mechanical ventilation (MV) demonstrated ROCs for RRT for uNGAL of 0.727 and sCysC of 0.764, compared with sCr of 0.785 and Acute Physiology and Chronic Health Evaluation (APACHE) II severity of 0.728.7 Combining these markers improved their predictive accuracy but they did not perform much better than the sCr and APACHE II scores alone. Thus, predicting which patients will require RRT remains difficult.
Clinical Trials Examining Timing Of RRT
With the uncertainty in predicting which patients will need RRT, the last few years have seen the authors of several randomized, controlled trials evaluate the best initiation approach. These include the ELAIN and AKIKI trials in 2016 and the IDEAL-ICU study in 2018.8-10 Other studies dating back to 2002 were included in a systematic review and meta-analysis in 2017 that did not include the IDEAL-ICU trial.11
The ELAIN trial was a single-center, parallel group trial of 231 patients admitted to the ICU.8 It included patients aged 18-90 years with KDIGO stage 2 and a marker of disease severity (sepsis, vasopressor requirement, refractory fluid overload, Sequential Organ Failure Assessment score ≥ 2). The criteria also included plasma NGAL > 150 ng/mL as a predictor of the need for RRT. It is notable that most patients were admitted with surgical indications. Exclusion criteria included chronic kidney disease (glomerular filtration rate [GFR] less than 30 mL/min), previous RRT, occlusion of the renal artery, glomerulonephritis, interstitial nephritis, vasculitis, post-renal obstruction, or thrombotic thrombocytopenic purpura/hemolytic uremic syndrome. Patients were randomized to early or delayed start of RRT. Early initiation occurred within eight hours of KDIGO stage 2, while delayed initiation occurred within 12 hours of KDIGO stage 3 AKI. Patients in the delayed initiation group could start early if any of the absolute indications for RRT were met (blood urea nitrogen [BUN] > 100 mg/dL, potassium > 6 mmol/L, or ECG changes, magnesium > 8 mEq/L, or urine production < 200 mL per 12 hours). Protocolized delivery of RRT ensured all patients received similar continuous venovenous hemodiafiltration. In this study, early initiation was associated with reduced 90-day mortality (hazard ratio [HR], 0.66; P = 0.03), the primary outcome. However, 28- and 60-day mortality rates in the early and delayed groups were not statistically different. Patients in the early initiation arm had shorter hospital length of stay (LOS) and shorter duration of MV.
The AKIKI trial was a multicenter, randomized trial of patients with KDIGO stage 3 disease who required MV, vasopressors, or both. Patients were excluded if they had severe metabolic abnormalities, including BUN greater than 112 mg/dL, potassium > 6 mmol/L, pH < 7.15, or acute pulmonary edema requiring > 5 L/min O2. In the early initiation group, researchers started RRT within six hours of KDIGO stage 3. Meanwhile, in the delayed group, RRT was started if patients progressed to the above-mentioned exclusion criteria or if they became anuric for more than 72 hours after randomization. The choice of RRT was left to each study site; thus, there was some variability in the mode of RRT. The primary outcome was 60-day mortality. Most patients presented with severe sepsis or septic shock (> 70%). Unlike the ELAIN trial, the AKIKI trial investigators demonstrated no significant difference in 60-day mortality between the early and delayed groups. Although there were more patients in the early initiation arm who received RRT, the only other secondary outcomes that showed a difference was a higher incidence of catheter-related bloodstream infections in the early initiation arm (10% vs. 5%; P = 0.03) and higher incidence of hypophosphatemia.
More recently, the IDEAL-ICU trial was a multicenter, randomized, controlled trial comparing early vs. late initiation of RRT for patients with AKI and sepsis.10 Inclusion criteria included patients ≥ 18 years of age, admission to the ICU with sepsis within 48 hours of the start of vasopressor therapy, and AKI defined by “failure” under the RIFLE criteria. The early intervention group started RRT within 12 hours of meeting inclusion criteria, while the delayed group were monitored for the development of criteria for emergency RRT, including potassium 6.5 mmol/L, pH < 7.15, or fluid overload refractory to diuretics. If any of these criteria were met, RRT was started as soon as possible. If none of these criteria were met, RRT was initiated 48 hours after the diagnosis of AKI (unless spontaneous renal recovery occurred, defined by spontaneous urine output > 1,000 mL per 24 hours or > 2,000 mL per 24 hours with diuretics). The choice of RRT was at the discretion of each site. The primary outcome was 90-day mortality. After the second planned interim analysis, the authors ended the trial for futility. There was no difference in 90-day mortality between groups.
Prior to the IDEAL-ICU trial, a systematic review and meta-analysis of both the ELAIN and AKIKI trials (as well as seven other randomized, controlled trials for a total of 1,636 patients) was published.11 The definitions for AKI and early and delayed initiation varied. For overall mortality, there was no difference in early vs. late initiation of RRT (HR, 0.98; 95% confidence interval, 0.78-1.23). Similarly, ICU LOS and hospital LOS were similar in each group. There was no difference in renal recovery, dependence on RRT, duration of RRT, or time to renal recovery. However, there was a high degree of heterogeneity, and the authors warned these conclusions must be interpreted cautiously.
What guidance can we take from these mostly negative trials of early vs. late initiation of RRT? A few lessons can be learned from the similarities and differences in each trial. The most recent trials contained absolute criteria for the initiation of RRT. (See Table 2.) In general, patients with serum potassium > 6-6.5 mmol/L or with ECG changes, BUN > 100-112 mg/dL, or pH < 7.15 should be dialyzed emergently regardless of RIFLE or KDIGO stage. However, each trial had different RRT modalities, and it remains unclear whether continuous RRT or intermittent RRT is most appropriate.
The ELAIN trial was the largest positive trial. Several features of this trial deserve special attention. First, most patients were admitted for surgical indications. Of the 231 patients randomized, 108 were admitted for cardiac surgery, 78 for abdominal surgery, and 28 for trauma. In contrast, AKIKI primarily included medical patients, with only 20% of patients admitted for surgical indications. IDEAL-ICU focused on patients who met criteria for sepsis. Second, the inclusion criteria for ELAIN included elevated NGAL. This was intended to enrich the population for patients who would require RRT. Indeed, if one examines the proportion of patients in each study who underwent RRT, ELAIN was the highest, with 100% of the early arm and 91% of the late arm. (See Table 3.) In the AKIKI and IDEAL-ICU trials, only 51% and 62%, respectively, of the delayed arms ever underwent RRT, implying one-half to one-third of patients had spontaneous renal recovery. Third, the timing of initiation of RRT was much earlier in ELAIN than in the other trials. (See Table 4.)
The authors of these trials all started RRT in the early arm within four to eight hours of meeting criteria. However, the delayed arm in ELAIN started at a median of 25 hours, while AKIKI and IDEAL-ICU started at a median of 57 and 51.5 hours, respectively. The final key difference was the modality of RRT. ELAIN specified continuous venovenous hemodiafiltration, while the authors of the other studies allowed each participating site to make that decision.
With these features in mind, certain populations might benefit from early initiation of RRT. First, surgical patients without other medical complexities might perform better with earlier RRT. Ninety-day mortality in the early arm of the ELAIN trial was 39.3%, significantly lower than the early arms of the other trials (58% in IDEAL-ICU; 48.5% 60-day mortality in AKIKI). These patients simply may tolerate RRT better. Second, early initiation likely works best for patients who will need RRT so that the benefits outweigh the risks. In ELAIN, 100% and 91% of the early and late arms, respectively, received RRT. In contrast, in the late arm of AKIKI and IDEAL-ICU, only 51% and 62% required RRT, respectively. Thus, many patients in the early arms of these two trials may never have required RRT and, thus, no benefit was achieved. Third, early interventions for AKI may prevent organ dysfunction. ELAIN initiated RRT much earlier than AKIKI and IDEAL-ICU. (See Table 4.) These latter trials may have missed the window. Initiating RRT 57 hours (AKIKI) and 52.5 hours (IDEAL-ICU) after meeting criteria for renal failure may havemay have been too late to achieve benefit. Finally, it is notable that the largest positive trial clearly specified continuous venovenous hemodiafiltration at very specific flows and pump rates, while the negative trials were more heterogeneous.
The optimal time for initiating RRT in patients admitted to the ICU remains uncertain. For most patients, there is no clear difference in mortality between early and delayed initiation. However, patients admitted for strictly surgical indications did demonstrate a benefit for earlier RRT. As we become better at predicting which patients will need RRT, early initiation may be beneficial for those patients specifically. Otherwise, there does not seem to be harm in delaying RRT until absolute indications for RRT are met.
- Bellomo R, et al. Acute renal failure – definition, outcome measures, animal models, fluid therapy and information technology needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8:R204-R212.
- KDIGO Workgroup. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl 2012;2:124-138.
- Mehta RL, et al. Acute kidney injury network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31.
- Luo X, et al. A comparison of different diagnostic criteria of acute kidney injury in critically ill patients. Crit Care 2014;18:R144.
- Fujii T, et al. Validation of the Kidney Disease Improving Global Outcomes criteria for AKI and comparison of three criteria in hospitalized patients. Clin J Am Soc Nephrol 2014;9:848-854.
- Pereira M, et al. Acute kidney injury in patients with severe sepsis or septic shock: A comparison between the ‘Risk, Injury, Failure, Loss of kidney function, End-stage kidney disease’ (RIFLE), Acute Kidney Injury Network (AKIN) and Kidney Disease: Improving Global Outcomes (KDIGO) classifications. Clin Kidney J 2017;10:332-340.
- Pipili C, et al. Prediction of the renal replacement therapy requirement in mechanically ventilated critically ill patients by combining biomarkers for glomerular filtration and tubular damage. J Crit Care 2014;29:692.e7-13.
- Zarbock A, et al. Effect of early vs delayed initiation of renal replacement therapy on mortality in critically ill patients with acute kidney injury: The ELAIN randomized clinical trial. JAMA 2016;315:2190-2199.
- Gaudry S, et al. Initiation strategies for renal-replacement therapy in the intensive care unit. N Engl J Med 2016;375:122-133.
- Barbar SD, et al. Timing of renal-replacement therapy in patients with acute kidney injury and sepsis. N Engl J Med 2018;379:1431-1442.
- Yang X, et al. A comparison of early versus late initiation of renal replacement therapy for acute kidney injury in critically ill patients: An updated systematic review and meta-analysis of randomized controlled trials. BMC Nephrol 2017;18:264.