Diagnosis, evaluation, and management of acute kidney injury: a KDIGO summary (Part 1)

Acute kidney injury (AKI) is a common and serious problem affecting millions and causing death and disability for many. In 2012, Kidney Disease: Improving Global Outcomes completed the first ever, international, multidisciplinary, clinical practice guideline for AKI. The guideline is based on evidence review and appraisal, and covers AKI definition, risk assessment, evaluation, prevention, and treatment. In this review we summarize key aspects of the guideline including definition and staging of AKI, as well as evaluation and nondialytic management. Contrast-induced AKI and management of renal replacement therapy will be addressed in a separate review. Treatment recommendations are based on systematic reviews of relevant trials. Appraisal of the quality of the evidence and the strength of recommendations followed the Grading of Recommendations Assessment, Development and Evaluation approach. Limitations of the evidence are discussed and a detailed rationale for each recommendation is provided.

weight ing the evidence across diff erent outcomes, we selected as the crucial outcome that which weighed most heavily in the assessment of the overall quality of evidence. Values and preferences articulated by the Work Group included: a desire to be inclusive in terms of meeting criteria for AKI; a progressive approach to risk and cost such that, as severity increased, the group put greater value on possible eff ectiveness of strategies, but maintained high value for avoidance of harm; and intent to guide practice but not limit future research.

Grading the quality of evidence and the strength of recommendations
Th e grading approach followed in this guideline and the wording of each recommendation are adopted from the Grading of Recommendations Assessment, Development and Evaluation system [4,5]. Th e strength of each recommendation is rated as level 1 (strong) or level 2 (weak or discretionary). In addition, each statement is assigned a grade for the quality of the supporting evidence: A (high), B (moderate), C (low), or D (very low). Furthermore, on topics that cannot be subjected to systematic evidence review, the Work Group issued statements that are not graded which hopefully will provide general guidance that is based on clinical experience.
Th e Grading of Recommendations Assessment, Develop ment and Evaluation system is best suited to evaluate evidence on comparative eff ectiveness. Some of our most important guideline topics involve diagnosis and staging of AKI, and here the Work Group chose to provide un graded statements. Th ese statements are indirectly sup ported by evidence on risk relationships and resulted from unanimous consensus of the Work Group and should not be viewed as weaker than graded recommendations.

A. Defi nition and staging of AKI
AKI is defi ned by an abrupt decrease in kidney function that includes, but is not limited to, acute renal failure. AKI is a broad clinical syndrome encompassing various etiologies, including pre-renal azotemia, acute tubular necrosis, acute interstitial nephritis, acute glomerular and vasculitic renal diseases, and acute postrenal obstruc tive nephropathy. More than one of these conditions may coexist in the same patient and epidemiological evidence supports the notion that even mild, reversible AKI has important clinical consequences, including increased risk of death [6,7]. AKI can thus be considered more like acute lung injury or acute coronary syndrome. Furthermore, because the manifestations and clinical consequences of AKI can be quite similar (even indistinguishable) regard less of whether the etiology is predominantly within the kidney or predominantly from outside stresses on the kidney, the syndrome of AKI encompasses both direct injury to the kidney as well as acute impairment of function.
Th e Acute Dialysis Quality Initiative group developed the Risk, Injury, Failure; Loss and End-stage kidney disease (RIFLE) system for diagnosis and classifi cation of a broad range of acute impairment of kidney function through a broad consensus of experts [8]. Studies totaling over 0.5 million patients from around the world have shown that AKI defi ned by RIFLE is associated with decreased survival and that increasing RIFLE stage leads to increased risk of death [9][10][11][12][13][14].
More recently, the Acute Kidney Injury Network endorsed the RIFLE criteria with a modifi cation to include small changes in serum creatinine (SCr) (≥0.3 mg/dl or 26.5 μmol/l) when they occur within a 48-hour period [15]. Two recent studies examining large databases in the USA [12] and Europe [13] validated these modifi ed criteria. Th akar and colleagues found that increased severity of AKI was associated with an increased risk of death independent of comorbidity [12]. Patients with stage 1 AKI (≥0.3 mg/dl or 26.5 μmol/l increase in SCr but less than a twofold increase) had an odds ratio of 2.2; in patients with stage 2 AKI (corresponding to RIFLE-I) there was an odds ratio of 6.1; and in stage 3 AKI patients (RIFLE-F) the odds ratio was 8.6 for hospital mortality. An additional modifi cation to the RIFLE criteria has been proposed for pediatric patients in order to better classify small children with acute-on-chronic disease [16].
Unfortunately, the existing criteria -while useful and widely validated -are still limited. First, despite eff orts to standardize the defi nition and classifi cation of AKI, there is still inconsistency in application [10,11]. A minority of studies have included urinary output criteria despite their apparent ability to identify additional cases [13,17] and many studies have excluded patients whose initial SCr is already elevated. Preliminary data suggest that roughly one-third of AKI cases are community acquired [18] and many cases may be missed by limiting analysis to documented increases in SCr. Indeed, the majority of cases of AKI in the developing world are likely to be community acquired. Few studies can thus provide accurate incidence data. An additional problem relates to the limitations of SCr and urine output for detecting AKI. In the future, biomarkers of renal cell injury may identify additional patients with AKI and may identify the majority of patients at an earlier stage. Th ese concerns notwithstanding, and in view of the available evidence, the Work Group accepted the existing criteria for the diagnosis and staging of AKI and proposed a single defi nition of AKI that should be useful for practice, research, and public health. A1: AKI is defi ned as any of the following (not graded): • increase in SCr by ≥0.3 mg/dl (≥26.5 μmol/l) within 48 hours; or • increase in SCr to ≥1.5 times baseline, which is known or presumed to have occurred within the prior 7 days; or • urine volume <0.5 ml/kg/hour for 6 hours. A2: AKI is staged for severity according to the criteria presented in Table 1 (not graded). A3: Th e cause of AKI should be determined whenever possible (not graded).

B. Risk assessment
Th ere are many types of exposure that may cause AKI. However, the chances of developing AKI after exposure to the same insult depend on a number of susceptibility factors that vary widely from individual to individual. Our understanding of susceptibility factors is based on many observational studies that address diff erent settings with regards to the type, severity, duration, and multiplicity of insults. While this heterogeneity provides insight into some susceptibility factors that are common across various populations, the generalizability of results from one particular setting to the next is uncertain.
It is important to screen patients who have undergone an exposure (Table 2) and to continue monitoring highrisk patients until the risk has subsided. Exact intervals for checking SCr and for which individuals' urine output should be monitored remain matters of clinical judgment; however, as a general rule, high-risk in-patients should have SCr measured at least daily and more frequently after an exposure, and critically ill patients should undergo urine output monitoring. Th is will necessitate urinary bladder catheterization in many cases, and the risks of infection should also be considered in the monitoring plan. Many opportunities for prevention and earlier recognition of AKI at emergency admissions may be missed. For example, a recent clinical practice assessment of emer gency admissions in the UK highlights missed oppor tu nities for prevention and earlier recognition of AKI [19]. B1: We recommend that patients be stratifi ed for risk of AKI according to their susceptibilities and exposures (Grade 1B). B2: Manage patients according to their susceptibilities and exposures to reduce the risk of AKI (see relevant guideline sections) (not graded). B3: Test patients at increased risk for AKI with measurements of SCr and urine output to detect AKI (not graded). Individualize frequency and duration of monitoring based on patient risk and clinical course (not graded).

C. Evaluation and general management
AKI is one of a number of conditions that aff ect kidney structure and function. Because the manifestations and clinical consequences of AKI can be quite similar (even indistinguishable) regardless of whether the etiology is predominantly within the kidney or predominantly from outside stresses on the kidney, the syndrome of AKI encompasses both direct injury to the kidney as well as acute impairment of function. Since treatments of AKI are dependent to a large degree on the underlying etiology, this guideline focuses on specifi c diagnostic   Figure 1). However, since general therapeutic and monitoring recommendations can be made regarding all forms of AKI, our approach will be to begin with general measures (Figure 2).
Th e clinical evaluation of AKI includes a careful history and thorough physical examination. Drug history should include over-the-counter formulations and herbal remedies or recreational drugs. Th e social history should include exposure to tropical diseases, and physical examination should include evaluation of fl uid status, signs for acute and chronic heart failure, and infection. Measurement of cardiac function and intra-abdominal pressure should be considered in the appropriate clinical context. Laboratory parameters -including SCr, blood urea nitrogen, and electrolytes, complete blood count and diff erential -should be obtained. Urine analysis and microscopic examination as well as urinary chemistries may be helpful in determining the underlying cause of AKI. Imaging tests, especially ultrasound, are important components of the evaluation for patients with AKI. Finally, a number of biomarkers of functional change and cellular damage are under evaluation for early diagnosis of AKI, risk assessment for AKI, and prognosis of AKI. Although an evidence-based analysis of the role of biomarkers was beyond the scope of this guideline, recent work suggests in particular that the prognostic utility of newer urinary biomarkers -including neutrophil gelatinase-associated lipocalin, kidney injury molecule-1, and IL-18 -added to urine microscopic examination is signifi cantly higher over clinical assessment alone [20].
Because the stage of AKI has clearly been shown to correlate with short-term [6,7,11,13] and even longerterm outcomes [21], it is advisable to tailor management to AKI stage. Figure 2 lists a set of actions that should be considered for patients with AKI. Note that for patients at increased risk, these actions begin even before AKI is diagnosed. Note also that management and diagnostic steps are both included in this fi gure. Th is is because response to therapy is an important part of the diagnostic approach. Th ere are few specifi c tests to establish the etiology of AKI. However, a patient's response to treatment (for example, discontinuation of a possible nephrotoxic agent) provides important information as to the diagnosis.

Fluids and vasopressors
Despite the recognition of volume depletion as an important risk factor for AKI, there have been no randomized controlled trials (RCTs) that directly evaluated the role of fl uids versus placebo in the prevention of AKI, except in the fi eld of contrast-induced AKI (see Lameire et al. immediately following this review). While fl uid resuscitation is widely believed to be protective, large multicenter studies have also shown that a positive fl uid balance is associated with increased 60-day mortality [22][23][24]. Results of the Saline vs. Albumin Fluid Evaluation study -a RCT comparing 4% human albumin in 0.9% saline with isotonic saline in ICU patients -indicated that albumin is safe, albeit no more eff ective than isotonic saline for fl uid resuscitation [25]. Th e study demonstrated no diff erence in need for and duration of RRT [25]. Very few patients in the trial received large volume fl uid resuscitation (>5 l) and thus the results may not be applicable to all patients.
Hydroxyethylstarch (HES) is a widely used, relatively inexpensive alternative to human albumin for correcting hypovolemia. A recent Cochrane review concluded that there is no evidence that resuscitation with colloids, instead of crystalloids, reduces the risk of death in patients with trauma, burns, or following surgery [26]. In addition to some negative eff ects on coagulation, particularly with older forms of HES, development of renal dysfunction has been a concern associated with the use of mainly hypertonic HES. A recent meta-analysis described 11 randomized trials with a total of 1,220 patients: seven trials evaluating hyperoncotic albumin and four trials evaluating hyperoncotic starch [27]. Hyperoncotic albumin decreased the odds of AKI by 76% while hyperoncotic starch increased those odds by 92% (odds ratio (OR) = 1.92; 95% confi dence interval (CI) = 1.31 to 2.81; P = 0.0008). Parallel eff ects on mortality were observed. Th e renal eff ects of hyperoncotic colloid solutions appeared to be colloid specifi c, with albumin displaying renoprotection and hyper oncotic starch showing nephrotoxicity. A 7,000-patient study comparing 6% HES 130/0.4 in saline with saline alone was scheduled to begin in Australia and New Zealand in 2010. Th is study will provide further high-quality data to help guide clinical practice [28].
Th e use of isotonic saline as the standard of care for intravascular volume expansion to prevent or treat AKI is thus based upon the lack of clear evidence that colloids are superior for this purpose, along with some evidence that specifi c colloids may cause AKI, in addition to their higher costs. It is acknowledged that colloids may be chosen in some patients to aid in reaching resuscitation goals, or to avoid excessive fl uid administration in patients requiring large volume resuscitation, or in specifi c patient subsets (for example, a cirrhotic patient with spontaneous peritonitis, or in burns). Similarly, although hypotonic or hypertonic crystalloids may be used in specifi c clinical scenarios, the choice of crystalloid with altered tonicity is generally dictated by goals other than intravascular volume expansion (for example, hypernatremia or hyponatremia). In addition, isotonic saline solution contains 154 mmol/l chloride and when administration in large volumes will result in relative or absolute hyperchloremia (for a review, see Kaplan and Kellum [29]). Buff ered salt solutions approximate physiological chloride concentrations and cause less acid-base disturbances and other side eff ects associated with hyperchloremia. Whether the use of buff ered solutions results in better clinical outcomes, however, is uncertain. Once the intravascular volume has been optimized, it is not known which vasopressor agent is most eff ective for manage ment of shock in general, or for the kidney specifi cally. A large RCT comparing dopamine with norepinephrine as the initial vasopressor in patients with shock showed no signifi cant diff erences between groups with regard to renal function or mortality. However, there were more arrhythmic events among the patients treated with dopamine [30]. Vasopressin is gaining popularity in the treatment of shock refractory to norepinephrine [31]. Compared with norepinephrine, vasopressin increases blood pressure and enhances diuresis, but has not been proven to enhance survival or to reduce the need for RRT [32]. Although there is some suggestion that vasopressin may reduce progression to renal failure and mortality in patients with septic shock [33], the Work Group concluded that current clinical data are insuffi cient to recommend one vasoactive agent over another in preventing AKI, but emphasized that vasoactive agents should not be withheld from patients with vasomotor shock over concern for kidney perfusion. Indeed, appro priate use of vasoactive agents can improve kidney perfusion in volume-resuscitated patients with vaso motor shock.
While the risks and benefi ts of so-called early goaldirected therapy are unclear and three large trials are underway in the USA, Australia and the UK, there is some evidence that protocolized resuscitation may be better than standard care. A recent meta-analysis concluded that protocolized therapies (regardless of the protocol) with specifi c physiological goals can significantly reduce postoperative AKI [34]. A problem in interpreting these studies is the lack of standardized hemodynamic and tissue oxygenation targets and manage ment strategies used to verify the effi cacy of these measures over standard perioperative care. A heterogeneous collection of study populations, types of surgical procedures, monitoring methods, and treatment strategies comprise this recent meta-analysis [34]. Th e basic strategy of goal-directed therapy to prevent AKI in the perioperative period is based on protocols that avoid hypotension, optimize oxygen delivery, and include careful fl uid management, vasopressors when indicated, and inotropic agents and blood products if needed [34]. Given the limitations of the current studies and the lack of comparative eff ectiveness studies of individual protocols, we can only conclude that protocols for resuscitation in the setting of septic shock and high-risk surgery appear to be superior to no protocol. D1: In the absence of hemorrhagic shock, we suggest using isotonic crystalloids rather than colloids (albumin or starches) as initial management for expansion of intravascular volume in patients at risk for AKI or with AKI (Grade 2B). D2: We recommend the use of vasopressors in conjunction with fl uids in patients with vasomotor shock with, or at risk for, AKI (Grade 1C). D3: We suggest using protocol-based management of hemodynamic and oxygenation parameters to prevent development or worsening of AKI in high-risk patients in the perioperative setting (Grade 2C) or in patients with septic shock (Grade 2C).

Nutrition and glycemic control
Pooled analyses of early multicenter studies have failed to confi rm the early observations of benefi cial eff ects of intensive insulin therapy on renal function; the risk of hypoglycemia with this approach is signifi cant, and the survival benefi ts of intensive insulin therapy are in doubt [35,36]. Th e international Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation study found a 90-day mortality of 27.5% in the intensive insulin therapy group (target blood glucose range 81 to 108 mg/dl (4.5 to 6.0 mmol/l)) and a 90-day mortality of 24.9% in the conventional glucose control (target ≤180 mg/dl (≤10.0 mmol/l)) (OR for intensive control = 1.14; 95% CI = 1.02 to 1.28; P = 0.02) [37]. Th e treatment eff ect did not diff er signifi cantly between surgical patients and medical patients. Th ere was no signifi cant diff erence between the two treatment groups in incidence of new RRT (15.4% vs. 14.5%). Severe hypoglycemia (blood glucose level ≤40 mg/dl (≤2.2 mmol/l)) was reported in 6.8% in the intensive-control group and in 0.5% in the conventional-control group (P <0.001).
Considering the balance between potential benefi ts and harm, the Work Group suggests using insulin for preventing severe hyperglycemia in critically ill patients but in view of the danger of potentially serious hypoglycemia, we suggest that the average blood glucose should not exceed 149 mg/dl (8.3 mmol/l), but that insulin therapy should not be used to lower blood glucose to <110 mg/dl (6.1 mmol/l). Th e Work Group recognizes that these thresholds have never directly been examined in RCTs but are interpolated from the comparisons so far tested in the trials. Several expert panels have developed clinical practice guidelines for the nutritional management of patients with AKI, whether treated with or without RRT [38][39][40][41][42]. Observations in critically ill patients provide a rationale to maintain a total energy intake of at least 20 kcal/kg/ day but not more than 25 to 30 kcal/kg/day, equivalent to 100 to 130% of the resting energy expenditure. Energy provision should be composed of 3 to 5 g (maximum 7 g) per kilogram body weight carbohydrates and 0.8 to 1.0 g per kilogram body weight fat.
When continuous renal replacement therapy (CRRT) techniques are used it should be realized that they may result in additional losses of water-soluble, lowmolecular-weight substances, including nutrients [43]. Normalized protein catabolic rates of 1.4 to 1.8 g/kg/day have been reported in patients with AKI receiving CRRT [44][45][46] and about 0.2 g amino acids are lost per liter of fi ltrate, amounting to a total daily loss of 10 to 15 g amino acids. In addition, 5 to 10 g protein are lost per day, depending on the type of therapy and dialyzer membrane. Similar amounts of protein and amino acids are typically lost by peritoneal dialysis. Nutritional support should account for these losses by providing a maximum of 1.7 g amino acids/kg/day. Enteral feeding is associated with improved outcome/survival in ICU patients [47,48] and should be recommended for patients with AKI.
In children with AKI, physiological macronutrient requirements are age dependent, refl ecting the developmental dynamics of growth and metabolism. Although these recommendations are limited to observational studies, it is generally agreed that critically ill children, like adults, should receive 100 to 130% of the basal energy expenditure, which can be estimated with acceptable precision and accuracy by the Caldwell-Kennedy equation [49]: Resting energy expenditure (kcal/kg/day) = 22 + 31.05 × weight (kg) + 1.16 × age (years) In a recent survey of the nutritional management of 195 children with AKI on CRRT, the maximal calorie prescription in the course of treatment averaged 53, 31, and 21 kcal/kg/day, and that for protein intake 2.4, 1.9, and 1.3 g/kg/day in children aged <1 year, 1 to 13 years, and >13 years, respectively [50]. Although not validated by outcome studies, these fi gures provide an orientation for the macronutrient supply typically achieved in and tolerated by children with AKI receiving CRRT.

Diuretics
On the basis of various mechanistic studies and support from preclinical data [51][52][53][54], loop diuretics (especially furosemide) have long been prescribed in the acute-care setting [55][56][57], and a number of RCTs have tested whether furosemide is benefi cial for prevention or treatment of AKI. Specifi cally, prophylactic furosemide was found to be ineff ective or harmful when used to prevent AKI after cardiac surgery [52,53], and to increase the risk of AKI when given to prevent contrast-induced AKI [54]. Epidemiologic data suggest that the use of loop diuretics may increase mortality in patients with critical illness and AKI [58], along with confl icting data that suggest no harm in AKI [59]. Finally, furosemide therapy was also ineff ective and possibly harmful when used to treat AKI [51,60]. A systematic review and meta-analysis by Ho and Power also included six studies that used furosemide to treat AKI, with doses ranging from 600 to 3,400 mg/day [61]. No signifi cant reduction was found for in-hospital mortality or for RRT requirement. Furosemide may be useful in achieving fl uid balance to facilitate mechanical ventilation according to the lung-protective ventilation strategy in hemodynamically stable patients with acute lung injury. However, a benefi cial role for loop diuretics in facilitating discontinuation of RRT in AKI is not evident from clinical studies [62,63]. Th e often retrospective and/or underpowered studies using prophylactic mannitol did not meet the criteria of the Work Group to be included in formulation of recommendations. Mannitol is often added to the priming fl uid of the cardiopulmonary bypass system to reduce the incidence of renal dysfunction, but the results of these studies are not very convincing [64]. Two small randomized trials -one in patients with pre-existing normal renal function [65], the second in patients with established renal dysfunction [66] -did not fi nd diff erences for any measured variable of renal function. More convincing are the results obtained with the preventive administration of mannitol, just before clamp release, during renal transplantation [67,68]. Th e sparse controlled data available have shown that 250 ml of 20% mannitol given immediately before vessel clamp removal reduces the incidence of post-transplant AKI, as indicated by a lower requirement of post-transplant dialysis. However, 3 months after transplantation, no diff erence is found in kidney function compared with patients who did not receive mannitol [69]. Finally, it has been suggested that mannitol is benefi cial in rhabdomyolysis by stimulating osmotic diuresis and by lowering the intracompartmental pressure in the aff ected crushed limbs [70][71][72]; again, these studies were either not random ized or were underpowered. A separate guideline on crush injury associated with disasters, mainly earthquake victims, has now been published by the International Society of Nephrology Renal Disaster Relief Task Force [73]. D9: We recommend not using diuretics to prevent AKI (Grade 1B). D10: We suggest not using diuretics to treat AKI, except in the management of volume overload (Grade 2C).

Vasodilator therapy: dopamine, fenoldopam, and natriuretic peptides
Th ree systematic reviews have reached identical conclusions that dopamine does not provide any benefi t for prevention or early treatment of AKI [74][75][76]. Th ere is also limited evidence that the use of dopamine to prevent or treat AKI causes harm. Dopamine can trigger tachyarrhythmias and myocardial ischemia, decrease intestinal blood fl ow, cause hypopituitarism, and suppress T-cell function [77]. Fenoldopam mesylate is a pure dopamine type-1 receptor agonist that has similar hemodynamic renal eff ects as low-dose dopamine, without systemic αadrenergic or β-adrenergic stimulation [78]. A metaanalysis found that fenoldopam reduces the need for RRT and in-hospital death in cardiovascular surgery patients [79]. However, the pooled studies included both pro phylactic and early therapeutic studies, as well as propensityadjusted case-matched studies (rather than purely random ized trials). A 1,000-patient RCT of fenoldopam to prevent the need for RRT after cardiac surgery is currently underway (ClinicalTrials.gov: NCT00621790); meanwhile, this remains an unproven indication for fenoldopam therapy. Our analysis of existing data from suitable prophylactic studies of adequate size and study design that reported AKI incidence in patients randomized to fenoldopam versus placebo revealed a pooled relative risk (RR) of 0.96 (95% CI = 0.76 to 1.2; P = not signifi cant). Only one study reported mortality (8-day) in sepsis patients randomized to fenoldo pam (35%, n = 150) versus placebo (44%, n = 150), with a RR of 0.79 (95% CI = 0.59 to 1.05; P = 0.1) [80]. As therapy for AKI, only one study reported (21-day) mortality in critically ill patients with early AKI random ized to fenoldopam (11/80, 13.8%) versus placebo (n = 19/75, 25.3%; P = 0.068) [81]. Another study reported the change in renal function in AKI patients randomized to fenoldopam (n = 50) versus dopamine (n = 50), defi ned by the absolute SCr change between the beginning and end of the study drug infusion and the maximum decrease from study entry, which were signifi cantly larger in the fenoldopam group: -0.53 ± 0.47 vs. dopamine: -0.34 ± 0.38 md/dl, P = 0.027 [82]. Overall, therefore, no data from adequately powered multicenter trials with clini cally signifi cant end-points and adequate safety are available to recommend fenoldopam to either prevent or treat AKI. Th e guideline recommendation against using fenoldopam places a high value on avoiding potential hypotension and harm associated with the use of this vasodilator in high-risk perioperative and ICU patients, and a low value on potential benefi t, which is currently only suggested by relatively low-quality single-center trials.
Nigwekar and colleagues recently conducted a systematic review and meta-analysis of ANP for manage ment of AKI [83]. Th ey found 19 relevant studies, among which 11 studies were for prevention and eight were for treatment of AKI. Pooled analysis of the eight treatment studies, involving 1,043 participants, did not show signifi cant diff erence for either RRT requirement or mortality between the ANP and control groups. However, low-dose ANP preparations were associated with significant reduction in RRT requirement (OR = 0.34; 95% CI = 0.12 to 0.96; P = 0.04). Th e incidence of hypotension was not diff erent between the ANP and control groups for low-dose studies, whereas it was signifi cantly higher in the ANP group in the high-dose ANP studies (OR = 4.13; 95% CI = 1.38 to 12.41; P < 0.01). Finally, a pooled analysis of studies that examined oliguric AKI did not show any signifi cant benefi t from ANP for RRT requirement or mortality. Only two of the treatment studies included in Nigwekar and colleagues' analysis [84,85] were of adequate size and quality to meet the criteria for our systematic review, which found no signifi cant inconsistencies in the fi ndings of both trials that (combined) included 720 subjects (351 treated with ANP). Th erefore, although subset analyses separating low-dose from highdose ANP trials suggest potential benefi ts, the pre ponderance of the literature suggests no benefi t of ANP therapy for AKI. Th e Work Group therefore suggests this agent not to be used to prevent or treat AKI.
Nesiritide (b-type natriuretic peptide) is the latest natriuretic peptide introduced for clinical use, and is approved by the US Food and Drug Administration only for the therapy of acute, decompensated congestive heart failure. Meta-analysis of outcome data from these and some other nesiritide congestive heart failure trials has generated some controversy [86][87][88]. Sackner-Bernstein and colleagues analyzed mortality data from 12 randomized trials; three trials provided 30-day mortality data, and found a trend towards an increased risk of death in nesiritide-treated subjects [86]. In another meta-analysis of fi ve randomized trials that included 1,269 subjects [87], the same investigators found that there was a relationship between nesiritide use and worsening renal function, defi ned as SCr increase >0.5 mg/dl (>44.2 μmol/ l). Nesiritide doses ≤0.03 μg/kg/minute and even at doses ≤0.015 μg/kg/minute signifi cantly increased the risk of renal dysfunction compared with non-inotrope-based controls or compared with all control groups (including inotropes). Th ere was no diff erence in dialysis rates between the groups. Another retrospective study determined independent risk factors for 60-day mortality by multivariate analysis in a cohort of 682 older heart-failure patients treated with nesiritide versus those who were not [89]. When patients were stratifi ed according to nesiritide usage, AKI emerged as an independent risk factor for mortality only among patients who received the drug. Strikingly, among these heart-failure patients who developed AKI, nesiritide usage emerged as the only independent predictor of mortality. A 7,000-patient multicenter RCT in acute decompensated heart failure has recently assessed the clinical eff ective ness of nesiritide therapy for acute decompensated heart failure (the Acute Study of Clinical Eff ectiveness of Nesiritide in Decom pensated Heart Failure; Clinicaltrials.gov: NCT00475852).
A prospective, randomized clinical trial (the Nesiritide Study) found no benefi t of nesiritide for 21-day dialysis and/or death in patients undergoing high-risk cardiovascular surgery [90]. However, this study did demonstrate that the prophylactic use of nesiritide was associated with reduced incidence of AKI in the immediate postoperative period (nesiritide 6.6% vs. placebo 28.5%; P = 0.004). Recently, Lingegowda and colleagues investigated whether the observed renal benefi ts of nesiritide had any long-term impact on cumulative patient survival and renal outcomes [91]. Data on all 94 patients from the Nesiritide Study were obtained with a mean follow-up period of 20.8 ± 10.4 months. No diff erences in cumulative survival between the groups were noted, but patients with in-hospital incidence of AKI had a higher rate of mortality than those with no AKI (41.4% vs. 10.7%; P = 0.002). Th e possible renoprotection provided by nesiritide in the immediate postoperative period was not associated with improved long-term survival in patients undergoing high-risk cardiovascular surgery.
Although evidence from a variety of small studies suggests the potential for therapy with natriuretic peptides to be useful for the prevention or treatment of AKI in a variety of settings, there are no defi nitive trials to support the use of atrial natriuretic peptide (ANP), btype natriuretic peptide, or nesiritide for these purposes. Th e Work Group therefore suggests these agents should not be used for prevention or treatment of AKI. D11: We recommend not using low-dose dopamine to prevent or treat AKI (Grade 1A). D12: We suggest not using fenoldopam to prevent or treat AKI (Grade 2C). D13: We suggest not using ANP to prevent (Grade 2C) or treat (Grade 2B) AKI.

Other pharmacologic therapies
AKI occurs in 60% of neonates suff ering from perinatal asphyxia [92], and experimental studies have indicated an important role for adenosine-mediated vasoconstriction in neonatal kidneys exposed to normocapnic hypoxemia [93]. A potential renoprotective eff ect of theophylline in perinatal asphyxia has been assessed in three randomized, placebo-controlled clinical trials [94][95][96], including a total of 171 term neonates. Th eophylline was uniformly administered in the fi rst hour of life as a single intravenous bolus at a dose of 5 mg/kg [94,96] or 8 mg/kg [95]. All three studies observed signifi cantly higher glomerular fi ltration rate, higher urine output with more negative fl uid balance, and lower urinary β 2 -microglobulin excretion with theophylline as compared with placebo during the fi rst 3 to 5 days of life. In each study, theophylline treatment was associated with a signifi cantly reduced risk of severe renal dysfunction (17 to 25% vs. 55 to 60% in the placebo group; RR = 0.3 to 0.41). Th e benefi cial eff ect was selective for kidney function, whereas the incidence of extra-renal complications was unaltered. Patient survival was not aff ected by treatment. In line with these studies in mature neonates, a similar improvement of glomerular fi ltration rate and urine output was observed during the fi rst 2 days of life by administration of 1 mg/kg theophylline versus placebo in 50 very preterm neonates with respiratory distress syndrome [97]. Follow-up of renal function throughout the fi rst year of life by Bhat and colleagues found equally normal glomerular and tubular function in both groups from 6 weeks of age onward [95]. Hence, while theophylline clearly improves renal function in the fi rst week of life in postasphyctic neonates, the overall benefi t from this intervention in neonatal intensive care is less evident in view of the complete long-term recovery of renal function in the placebo-treated controls and the absence of an eff ect on patient survival. By contrast, adenosine antagonism does not appear benefi cial in cardiorenal syndrome. Th ree pivotal phase III trials in a total of 2,500 patients were recently completed, aiming to corroborate the renoprotective eff ects of rolofylline in patients with cardiorenal syndrome, and to establish drug safety. Th e fi nal results of the PROTECT trial have recently been published [98]. Rolofylline, as compared with placebo, did not provide a benefi t with respect to the three primary end-points: survival, heart-failure status, and changes in renal function. Persistent renal impairment developed in 15.0% of patients in the rolofylline group and in 13.7% of patients in the placebo group (P = 0.44). By 60 days, death or readmission for cardiovascular or renal causes had occurred in similar proportions of both groups of patients. Adverse-event rates were similar overall; however, only patients in the rolofylline group had seizures, a known potential adverse eff ect of A 1 -receptor antagonists. Rolofylline therefore does not appear to be eff ective for treatment of cardiorenal AKI.
Based on an analysis of the three RCTs with insulin-like growth factor-1 that are currently available [99][100][101] and which were overall negative or at least equivocal, the Work Group recommends against its use in patients with AKI. D14: We recommend not using recombinant human insulin-like growth factor-1 to prevent or treat AKI (Grade 1B). D15: We suggest that a single dose of theophylline may be given in neonates with severe perinatal asphyxia, who are at high risk of AKI (Grade 2B).

Avoiding nephrotoxins
Aminoglycosides exhibit a number of favorable pharmaco kinetic and pharmacodynamic advantages, but a major dose-limiting toxicity of the aminoglycosides remains the risk of drug-induced AKI [102]. Th e risk of AKI attri butable to aminoglycosides is suffi ciently high (up to 25% in some series, depending upon the defi nition of AKI used and the population studied) [103][104][105][106][107][108][109] that they should no longer be used for standard empirical or directed treatment, unless no other suitable alternatives exist. When still required, the potential effi cacy of singledose daily regimens (or other extended dosing treatment programs) of aminoglycosides versus multiple-daily dosing strategies has been extensively studied in numerous controlled and uncontrolled clinical studies over many years [110][111][112][113][114][115][116][117][118][119][120][121], and the subject has been the focus of a number of formal meta-analyses [122][123][124][125][126][127]. Th e cumulative results of this evidence-based review and numerous meta-analyses indicate that once-daily dosing strategies generally result in less AKI when compared with multiple-dose dosing strategies, although the benefi t accrued by the single-daily dose strategy is modest and inconsistent across a number of these studies.
In view of the high variability of the pharmacokinetic characteristics of aminoglycosides, therapeutic drug monitor ing in combination with or independent from single-dose daily treatment regimens is recommended. In single-dose or extended-dose treatment strategies, the peak drug level should be at least 10-fold greater than the minimum inhibitory concentration of the infecting microorganism. Th e trough level should be undetectable by 18 to 24 hours to limit accumulation of amino glycosides in renal tubular cells and to minimize the risk of AKI.
Aminoglycoside aerosol delivery systems are now in use to provide high intrapulmonary antibiotic levels with minimal systemic and kidney concentrations of the antibiotic. However, signifi cant nephrotoxicity with the use of inhaled tobramycin has been described in at least two cases [128,129].
Th e safety and effi cacy of lipid formulations of amphotericin B have been studied in numerous experimental and clinical trials with conventional amphotericin B as the comparator [130][131][132][133][134][135][136][137][138][139][140][141][142]. A detailed analysis of these various trials, and a number of meta-analyses that have analyzed this clinical question, concluded that the lipid formulations are less nephrotoxic than amphotericin B deoxycholate [133,135]. When feasible, we recommend that lipid formulations supplant the use of conventional amphotericin B to reduce the risk of nephrotoxicity. Alternatively, when feasible, it may be best to avoid polyene antifungal agents entirely and use alternative agents, such as the azoles and echinocandins [143][144][145][146][147]. D16: We suggest not using aminoglycosides for the treatment of infections unless no suitable, less nephrotoxic, therapeutic alternatives are available (Grade 2A). D17: We suggest that, in patients with normal kidney function in steady state, aminoglycosides are administered as a single dose daily rather than multiple-dose daily treatment regimens (Grade 2B). D18: We recommend monitoring aminoglycoside drug levels when treatment with multiple daily dosing is used for more than 24 hours (Grade 1A). D19: We suggest monitoring aminoglycoside drug levels when treatment with single-daily dosing is used for more than 48 hours (Grade 2C). D20: We suggest using topical or local applications of aminoglycosides (for example, respiratory aerosols, instilled antibiotic beads), rather than intravenous application, when feasible and suitable (Grade 2B). D21: We suggest using lipid formulations of amphotericin B rather than conventional formulations of amphotericin B (Grade 2A). D22: In the treatment of systemic mycoses or parasitic infections, we recommend using azole antifungal agents and/or the echinocandins rather than conventional amphotericin B, if equal therapeutic effi cacy can be assumed (Grade 1A).

Surgical patients
A comprehensive meta-analysis examining off -pump versus conventional coronary artery bypass surgery found that the off -pump technique was associated with a statistically signifi cant 40% lower odds of postoperative AKI and a nonsignifi cant 33% lower odds for dialysis requirement [148]. Within the selected trials, off -pump coronary artery bypass graft surgery was not associated with a signifi cant decrease in mortality. It is apparent from this meta-analysis that the trials were clinically heterogeneous, particularly with regard to their defi nitions of kidney outcomes, and mostly were of poor to fair quality (based on the Jadad score). Th e very low event rates (often zero or one patient) make the estimates suspect and highly imprecise. Th ere is also a question of publication bias. Th ere are several large trials in progress that are likely to generate more defi nitive data. Th e Work Group concluded that there was not enough evidence at present to recommend off -pump coronary artery bypass for reducing AKI or the need for RRT. A meta-analysis did not fi nd evidence that N-acetylcysteine used perioperatively could alter mortality or renal outcomes after major cardiovascular or abdominal cancer surgery when radiocontrast agents are not used [149]. Only a single study has compared N-acetyl cysteine with placebo in critically ill patients with hypotension and was also negative [150]. D23: We suggest that off -pump coronary artery bypass graft surgery not be selected solely for the purpose of reducing perioperative AKI or need for RRT (Grade 2C). D24: We suggest not using N-acetylcysteine to prevent AKI in critically ill patients with hypotension (Grade 2D). D25: We recommend not using oral or intravenous N-acetylcysteine for prevention of postsurgical AKI (Grade 1A).