Towards Consensus in Timing of Kidney Replacement Therapy for Acute Kidney Injury?

Abstract

Commentary on STARRT-AKI Investigators; Canadian Critical Care Trials Group, the Australian and New Zealand Intensive Care Society Clinical Trials Group, the United Kingdom Critical Care Research Group, the Canadian Nephrology Trials Network, and the Irish Critical Care Trials Group, Bagshaw SM, et al. Timing of initiation of renal-replacement therapy in acute kidney injury. N Engl J Med. 2020;383(3):240-251. Commentary on STARRT-AKI Investigators; Canadian Critical Care Trials Group, the Australian and New Zealand Intensive Care Society Clinical Trials Group, the United Kingdom Critical Care Research Group, the Canadian Nephrology Trials Network, and the Irish Critical Care Trials Group, Bagshaw SM, et al. Timing of initiation of renal-replacement therapy in acute kidney injury. N Engl J Med. 2020;383(3):240-251. Acute kidney injury (AKI) affects >50% of critically ill patients and is associated with considerable morbidity and mortality.1Hoste E.A. Bagshaw S.M. Bellomo R. et al.Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study.Intensive Care Med. 2015; 41: 1411-1423Crossref PubMed Scopus (1053) Google Scholar Initiation of kidney replacement therapy (KRT) is the standard of care when AKI leads to life-threatening complications such as severe hyperkalemia, metabolic acidosis, or diuretic-resistant volume overload. However, the optimal timing of KRT initiation remains unclear. On the one hand, an early initiation strategy may theoretically maintain physiologic balance and avoid harm from volume overload and metabolic derangements. On the other hand, KRT initiation represents a considerable escalation in care that may unnecessarily expose patients to invasive procedures with risk for harm. The KDIGO AKI guideline advises to “consider the broader clinical context, the presence or absence of conditions that can be modified with [KRT], and the trends of laboratory tests…when making the decision to start [KRT]”2Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work GroupKDIGO clinical practice guidelines for acute kidney injury.Kidney Inter., Suppl. 2012; 2: 1-138Abstract Full Text Full Text PDF Scopus (1662) Google Scholar and therefore leaves considerable room for clinician judgment. In recent years, a few randomized controlled trials have attempted to address this issue with mixed results,3Zarbock A. Kellum J.A. Schmidt C. 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: 2190-2199Crossref PubMed Scopus (542) Google Scholar, 4Gaudry S. Hajage D. Schortgen F. et al.Initiation strategies for renal-replacement therapy in the intensive care unit.N Engl J Med. 2016; 375: 122-133Crossref PubMed Scopus (539) Google Scholar, 5Barbar S.D. Clere-Jehl R. Bourredjem A. et al.Timing of renal-replacement therapy in patients with acute kidney injury and sepsis.N Engl J Med. 2018; 379: 1431-1442Crossref PubMed Scopus (223) Google Scholar and therefore the optimal timing of KRT in critically ill patients with AKI remains the subject of significant debate. The recently published STARRT-AKI trial6STARRT-AKI Investigators; Canadian Critical Care Trials Group, the Australian and New Zealand Intensive Care Society Clinical Trials Group, the United Kingdom Critical Care Research Group, the Canadian Nephrology Trials Network, and the Irish Critical Care Trials Group Bagshaw S.M. et al.Timing of initiation of renal-replacement therapy in acute kidney injury [published correction appears in N Engl J Med. 2020].N Engl J Med. 2020; 383: 240-251Crossref PubMed Scopus (120) Google Scholar is the largest study to date on this topic and provides the most definitive answers regarding the controversy of timing of KRT in AKI. STARRT-AKI was a multinational multicenter trial that randomly assigned 3,019 critically ill adults with KDIGO stage 2 or 3 AKI to an accelerated or standard KRT initiation strategy. Exclusion criteria included patients with potassium level > 5.5 mEq/L or bicarbonate level < 15 mEq/L at time of screening, kidney transplantation in the preceding year, advanced chronic kidney disease (defined as estimated glomerular filtration rate < 20 mL/min/1.73 m2), presence of overdose or dialyzable toxin as indication for KRT, or strong clinical suspicion of kidney obstruction, vasculitis, thrombotic microangiopathy, or acute interstitial nephritis. After satisfying inclusion and exclusion criteria, clinical equipoise with respect to the 2 initiation strategies was determined by the primary treating clinician(s). Those meeting clinical equipoise were subsequently randomly assigned. In the accelerated group, clinicians were urged to initiate KRT as soon as possible and within 12 hours after random assignment. Conversely, in the standard group, clinicians had the discretion to initiate KRT at any time based on their clinical judgment. They were also provided general guidelines for initiation of KRT, including: (1) potassium level ≥ 6 mEq/L, (2) pH ≤ 7.20, (3) bicarbonate level ≤ 12 mEq/L, (4) Pao2/fraction of inspired oxygen ≤ 200 mm Hg along with clinical perception of volume overload, or (5) persistence of kidney injury 72 hours after randomization. Of those randomly assigned, 2,927 (97%) were included in a modified intention-to-treat analysis (1,465 accelerated, 1,462 standard). Both groups shared similar baseline characteristics. Among those receiving KRT, therapy was initiated at a median time of approximately 6 hours (from meeting eligibility criteria) in the accelerated group and 31 hours in the standard group. Notably, 38.2% of patients in the standard group never initiated KRT compared with 3.2% in the accelerated group. The groups were similar with regard to initial KRT modality, dose, and anticoagulation strategy. There was no difference in the primary outcome of 90-day mortality with respect to initiation strategy (43.9% in accelerated vs 43.7% in standard; P = 0.92). This finding persisted across subgroup analyses by sepsis, estimated glomerular filtration rate, type of admission (medical vs surgical), Simplified Acute Physiology Score II, and geographic region. In terms of secondary outcomes, survivors of the accelerated versus standard groups experienced greater KRT dependence at 90 days (10.4% vs 6.0%; relative risk [RR], 1.74 [95% CI, 1.24-2.43]) but no difference in the composite of major adverse kidney events at 90 days. Survivors in the accelerated group experienced shorter intensive care unit length of stay on average by approximately 1.6 (95% CI, 0.3-2.9) days, but there was no significant difference in ventilator-free days or overall hospital length of stay. STARRT-AKI found more adverse events with the accelerated compared with standard approach (23.0% vs 16.5%; P < 0.001). Hypotension and severe hypophosphatemia were the most commonly observed complications. Serious adverse events were rare and similar between the groups (1.0% vs 0.5%; P = 0.15). The main strengths of this trial include its large size with near-complete follow-up, as well as multinational and varied (medical and surgical) treatment settings, thereby providing generalizability of the results. Additionally, the inclusion of equipoise as a requirement prevented enrollment of patients in which the benefit of immediate initiation or deferral of KRT is already certain. A limitation of this trial is the heterogeneity in timing of KRT initiation in the standard-strategy group. Among patients initiated on KRT in the standard-initiation arm, about one-third did not meet at least one of the suggested criteria and therefore the decision was at the discretion of the treating clinician. A number of small single-center observational studies in cardiac surgery patients have suggested mortality or morbidity benefit from “early” initiation of dialysis.7Sugahara S. Suzuki H. Early start on continuous hemodialysis therapy improves survival rate in patients with acute renal failure following coronary bypass surgery.Hemodial Int. 2004; 8: 320-325Crossref PubMed Scopus (96) Google Scholar, 8Manche A. Casha A. Rychter J. Farrugia E. Debono M. Early dialysis in acute kidney injury after cardiac surgery.Interact Cardiovasc Thorac Surg. 2008; 7: 829-832Crossref PubMed Scopus (29) Google Scholar, 9Demirkilic U. Kuralay E. Yenicesu M. et al.Timing of replacement therapy for acute renal failure after cardiac surgery.J Card Surg. 2004; 19: 17-20Crossref PubMed Scopus (170) Google Scholar Larger observational cohort studies in the intensive care unit setting have similarly suggested a mortality benefit with early dialysis.10Liu K.D. Himmelfarb J. Paganini E. et al.Timing of initiation of dialysis in critically ill patients with acute kidney injury.Clin J Am Soc Nephrol. 2006; 1: 915-919Crossref PubMed Scopus (243) Google Scholar,11Bagshaw S.M. Uchino S. Bellomo R. et al.Timing of renal replacement therapy and clinical outcomes in critically ill patients with severe acute kidney injury.J Crit Care. 2009; 24: 129-140Crossref PubMed Scopus (241) Google Scholar Studies varied widely in definitions of “early” dialysis and include arbitrary cutoffs for values for serum creatinine, urea, urine output, time from intensive care unit admission, and duration of AKI. Fundamentally, all these studies only included patients receiving KRT and assigned timing retrospectively without consideration of patients who did not require KRT. Nonetheless, the potential promise of these studies helped lay the foundation for 4 recent randomized controlled clinical trials: ELAIN,3Zarbock A. Kellum J.A. Schmidt C. 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: 2190-2199Crossref PubMed Scopus (542) Google Scholar AKIKI,4Gaudry S. Hajage D. Schortgen F. et al.Initiation strategies for renal-replacement therapy in the intensive care unit.N Engl J Med. 2016; 375: 122-133Crossref PubMed Scopus (539) Google Scholar IDEAL-ICU,5Barbar S.D. Clere-Jehl R. Bourredjem A. et al.Timing of renal-replacement therapy in patients with acute kidney injury and sepsis.N Engl J Med. 2018; 379: 1431-1442Crossref PubMed Scopus (223) Google Scholar and STARRT-AKI.6STARRT-AKI Investigators; Canadian Critical Care Trials Group, the Australian and New Zealand Intensive Care Society Clinical Trials Group, the United Kingdom Critical Care Research Group, the Canadian Nephrology Trials Network, and the Irish Critical Care Trials Group Bagshaw S.M. et al.Timing of initiation of renal-replacement therapy in acute kidney injury [published correction appears in N Engl J Med. 2020].N Engl J Med. 2020; 383: 240-251Crossref PubMed Scopus (120) Google Scholar Table 1 provides a comparison of key characteristics and findings from these trials.Table 1Comparison of Recent Randomized Clinical Trials of Timing of KRT in AKIVariableaAll comparisons are early versus delayed/standard groups.ELAIN (N = 231)AKIKI (N = 620)IDEAL-ICU (N = 488)STARRT-AKI (N = 3,019)Study designSingle-center RCTMulticenter RCTMulticenter RCTMulticenter multinational RCTEarly groupWithin 8 h of stage 2 AKI and NGAL > 150 ng/mL; median: 6.0 hWithin 6 h of stage 3 AKI; median: 2 hWithin 12 h of stage 3 AKI; median: 7.6 hWithin 12 h of stage 2 or 3 AKI; median: 6.1 hDelayed/standard group (time from eligibility)Within 12 h of stage 3 AKI or clinical indication; median: 25.5 hAfter 72 h in stage 3 AKI or clinical indication; median: 57 hAfter 48 h in stage 3 AKI or clinical indication; median: 51.5 hAfter 72 h in AKI or clinical indication; median: 31.1 hKey exclusionseGFR < 30Severe lab abnormalitiesEmergent need for dialysiseGFR < 20KRT protocolPrescribed protocol using CVVHDFDiscretion of site providers, guideline basedDiscretion of site providers, guideline basedRecommendations provided based on guidelinesPrimary outcome90-d mortality: 39.3% vs 54.7% (P = 0.03)60-d mortality: 48.5% vs 49.7% (P = 0.79)90-d mortality: 57.7% vs 53.8% (P = 0.38)90-d mortality: 43.9% vs 43.7%; RR, 1.00 (95% CI, 0.93-1.09)Proportion not requiring KRT0% vs 9.2%1.9% vs 49.0%2.8% vs 38.4%3.2% vs 38.2%Dialysis dependence (among survivors)At 90 d: 13.4% vs 15.1% (P = 0.80)At 60 d: 1.9% vs 5.2% (P = 0.12)At 90 d: 2.0% vs 2.7% (P = 1.00)At 90 d: 10.4% vs 6.0%; RR, 1.74 (95% CI, 1.24-2.43)Adverse eventsNo significant differencesCatheter-related bloodstream infections: 10.0% vs 5.2%; P = 0.03Hyperkalemia: 0% vs 4.1%; P = 0.03Any: 23.0% vs 16.5% (P < 0.001); no difference in serious adverse eventsAbbreviations: AKI, acute kidney injury; CVVHDF, continuous venovenous hemodiafiltration; eGFR, estimated glomerular filtration (in mL/min/1.73 m2); KRT, kidney replacement therapy; lab, laboratory; NGAL, neutrophil gelatinase-associated lipocalin; RCT, randomized controlled trial; RR, relative risk.a All comparisons are early versus delayed/standard groups. Open table in a new tab Abbreviations: AKI, acute kidney injury; CVVHDF, continuous venovenous hemodiafiltration; eGFR, estimated glomerular filtration (in mL/min/1.73 m2); KRT, kidney replacement therapy; lab, laboratory; NGAL, neutrophil gelatinase-associated lipocalin; RCT, randomized controlled trial; RR, relative risk. The ELAIN trial is the only one of the 4 randomized trials to demonstrate mortality benefit. An important criticism of this study was that it had a fragility index of 3 (ie, a shift of only 3 patients from the event to nonevent group would have changed the results to statistically insignificant).3Zarbock A. Kellum J.A. Schmidt C. 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: 2190-2199Crossref PubMed Scopus (542) Google Scholar The ELAIN trial also enrolled primarily a surgical cohort of patients, and questions have arisen as to whether this group differs importantly from medical populations. Comparatively, STARRT-AKI had a more heterogeneous population, including more surgical patients overall than ELAIN, with no mortality benefit in the surgical subgroup. Similar to ELAIN (and in contrast to some of the other trials), STARRT-AKI enrolled patients starting at stage 2 AKI but did not require biomarker criteria for kidney injury. In the AKIKI trial, there was no difference in the primary outcome of 60-day mortality, but the delayed group had a significantly lower incidence of catheter-related infections and 49% did not require KRT.4Gaudry S. Hajage D. Schortgen F. et al.Initiation strategies for renal-replacement therapy in the intensive care unit.N Engl J Med. 2016; 375: 122-133Crossref PubMed Scopus (539) Google Scholar Of note, the early group in AKIKI was most similar to the delayed group in ELAIN (ie, stage 3 AKI), so one might speculate as to whether KRT was not initiated early enough to make a difference in AKIKI. However, the STARRT-AKI trial addresses this more definitively by including stage 2 AKI as entry criteria, similar to ELAIN. The IDEAL-ICU trial focused on patients with stage 3 AKI and septic shock and thus reported higher mortalities than the other studies. The trial was stopped early due to futility with no statistically significant difference in the primary outcome of 90-day mortality.5Barbar S.D. Clere-Jehl R. Bourredjem A. et al.Timing of renal-replacement therapy in patients with acute kidney injury and sepsis.N Engl J Med. 2018; 379: 1431-1442Crossref PubMed Scopus (223) Google Scholar The AKIKI and IDEAL-ICU trials shared similar definitions of early versus delayed KRT initiation and the consistent findings should be considered somewhat confirmatory. STARRT-AKI added to these findings by extending to an earlier period (stage 2 AKI) and still showed no apparent benefit from early KRT initiation. The STARRT-AKI study is the most definitive study to date regarding timing of KRT initiation in AKI. The results are fairly unequivocal: no mortality benefit from early initiation and potential signals for harm (increased KRT dependence and adverse events). These findings contradict the positive findings from the small ELAIN trial and should diminish enthusiasm for early initiation of KRT based solely on meeting biochemical definitions of AKI. Specifically, pre-emptive KRT initiation at stage 2 AKI (in absence of other clinical indications) should be discouraged. However, some important questions remain. What if we could predict which patients will need KRT? An important finding in STARRT-AKI was that nearly 40% of patients in the delayed group did not require KRT, similar to that observed in the IDEAL-ICU (38%) and AKIKI (49%) trials. Most nephrologists would agree that avoiding KRT is better than unnecessarily starting KRT, and this is supported by the increased complications seen with early initiation strategies. However, if we knew which patients with AKI were not going to recover, earlier KRT initiation could theoretically still prove beneficial while reducing the risk for unnecessary KRT exposure. The furosemide stress test has been shown to predict subsequent KRT need in AKI with reasonable sensitivity and specificity.12Chen J.J. Chang C.H. Huang Y.T. Kuo G. Furosemide stress test as a predictive marker of acute kidney injury progression or renal replacement therapy: a systemic review and meta-analysis.Crit Care. 2020; 24: 202Crossref PubMed Scopus (12) Google Scholar More recently, a new biomarker (CCL14 [C-C motif chemokine ligand 14]) has shown promise in predicting persistence of severe AKI for 72 hours or longer.13Hoste E. Bihorac A. Al-Khafaji A. et al.Identification and validation of biomarkers of persistent acute kidney injury: the RUBY study.Intensive Care Med. 2020; 46: 943-953Crossref PubMed Scopus (42) Google Scholar Future clinical trials are needed to evaluate whether incorporating these tools into KRT decision making may be beneficial. What clinical judgment parameters were used in the STARRT-AKI standard group? Do the reported mean laboratory values at KRT initiation in this group (serum creatinine, 4.9 mg/dL; serum urea nitrogen, 85 mg/dL; potassium, 4.6 mmol/L; and bicarbonate, 19.5 mmol/L) reflect most clinical practice? Notably, nearly one-third of patients started on KRT in the standard arm did not meet one of the prespecified indications. It will be interesting to see whether subsequent publications from this study delineate the clinical decision making and help inform clinical practices. In addition, the ongoing AKIKI-2 trial will help answer whether delaying KRT even longer might still be equivalent with even less requirement for KRT.14Gaudry S. Hajage D. Martin-Lefevre L. et al.The Artificial Kidney Initiation in Kidney Injury 2 (AKIKI2): study protocol for a randomized controlled trial.Trials. 2019; 20: 726Crossref PubMed Scopus (6) Google Scholar Should we consider alternative definitions for timing of KRT? Clinically, most nephrologists do not rely on serum creatinine values to determine KRT initiation, yet that is what clinical trials have relied on (ie, AKI stage). Fluid overload has been implicated as an important prognostic factor in AKI and critical illness and may be a more physiologic way to define timing for KRT. Notably, the standard group in STARRT-AKI had markedly greater fluid accumulation compared with the early group (5,893 vs 2,714 mL) at time of KRT initiation, but still no benefit was seen. The STARRT-AKI trial brings welcome clarity to the long-standing controversy around timing of KRT initiation in AKI. Although some questions remain to be answered by future trials, it is clear that initial careful observation for kidney function recovery versus persistence/worsening of AKI should remain the current standard when deciding on KRT timing. Ryann Sohaney, DO, Lenar T. Yessayan, MD, and Michael Heung, MD, MS. None. The authors declare that they have no relevant financial interests. Received August 17, 2020, in response to an invitation from the journal. Accepted August 19, 2020, after editorial review by an Associate Editor and a Deputy Editor.