The Japanese clinical practice guideline for acute kidney injury 2016

In the past, the pathology associated with sudden renal impairment was characterized as acute renal failure (ARF). However, in the 2000s, the joint efforts of specialists in fields including nephrology, intensive care medicine, and cardiovascular medicine led to the introduction of a novel concept called acute kidney injury (AKI). Although both ARF and AKI designate clinical conditions that present with sudden renal impairment and renal tissue damage, their respective clinical backgrounds leading to onset are thought to differ. In cases of ARF, high invasiveness is assumed to lead to sudden renal impairment in patients with relatively few comorbidities. In addition, as ARF is thought to be essentially a reversible disease, there was little awareness of its poor outcomes; greater attention was paid to the differentiation of the causes and to the countermeasures against the complications associated with renal failure than to the need for early detection. However, as medical care progressed, patients such as high-risk elderly subjects who were not deemed to be candidates for invasive therapy came to be treated in intensive care units (ICUs). Eventually, there grew to be widespread awareness of the increase in cases of sudden kidney injury comorbid with sepsis and multiple organ failure, and of the incredibly poor outcomes in these cases. This led kidney injury as a subset of multiple organ failure to be re-considered as AKI in intensive care medicine. Thus, AKI was proposed as a novel disease concept to emphasize early diagnosis and early intervention for the improvement of prognoses.

Meanwhile, the RIFLE [1], AKIN [2], and KDIGO [3] diagnostic criteria were introduced in an effort to establish unified international diagnostic criteria. The present guideline recommends the use of the KDIGO diagnostic criteria (see CQ2-1). However, these criteria are based solely on the serum creatinine (sCr) and urine output; they do not take into account the cause or site of the kidney injury, or location and mode of onset of AKI. Thus, as AKI refers to a syndrome with a broad spectrum of diseases and a variety of pathophysiologies, it calls for constant differentiation of the causes and elimination of the reversible factors. The KDIGO Clinical Practice Guideline for AKI [3] also recommends searching for and assessing the cause of the syndrome whenever possible, particularly in regard to reversible causes (recommendations 2.1.3 and 2.3.1).

Open image in new window

We identified 11 observational studies that compared the KDIGO with the AKIN and RIFLE criteria and that assessed death as an outcome. However, they did not assess the initiation of dialysis. In these 11 observational studies, the comparisons of the AKI diagnosis based on the KDIGO criteria versus those based on the RIFLE and AKIN criteria showed that the KDIGO criteria is more precise than, or as precise as, the RIFLE and AKIN criteria in reflecting the-in-hospital mortality.

In the past, acute renal failure (ARF) was diagnosed and classified according to several different criteria. In response to the growing call for unified international diagnostic criteria, the acute dialysis quality initiative (ADQI) published the RIFLE (risk, injury, failure, loss, end-stage kidney disease) criteria in 2004 (Table 1) [1, 4]. The RIFLE criteria distinguished three degrees of severity (risk, injury, and failure), with the latter defined as an increase in the serum creatinine (sCr), a decline in the glomerular filtration rate (GFR), and a reduction in the urine output; and two types of clinical outcomes (Loss and End-Stage Kidney Disease). In 2004, members of the International Society of Nephrology, the American Society of Nephrology, the National Kidney Foundation (in the United States), and the European Society of Intensive Care Medicine founded the acute kidney injury network (AKIN); as a replacement for the term ARF, the AKIN advocated the concept of acute kidney injury (AKI), which encompasses earlier stages of kidney injury. On the other hand, after the RIFLE criteria were published, a mere 0.3 mg/dl increase in sCr was reported to affect the survival prognosis and the clinical course of AKI [5, 6].

In 2007, the AKIN proposed the AKIN criteria, which were a revision of the RIFLE criteria (Table 2) [2]. The AKIN criteria included milder increases in sCr (0.3 mg/dl) and added the time course of the sCr increase (within 48 h) to the diagnostic criteria. By contrast, a reduced GFR was removed from the RIFLE criteria. In addition, while both the AKIN criteria and the RIFLE criteria included the urine output, the AKIN criteria specified that when making a diagnosis based on the urine output alone, urinary tract obstructions and easily reversible causes of a reduced urine output were to be excluded, and an adjustment was to be made for the body fluid volume. In addition, the RIFLE criteria’s loss and end-stage kidney disease were judged to be outcomes of AKI and were removed from the AKIN criteria’s staging system. Furthermore, patients who had started renal replacement therapy (RRT) became classified as stage 3 regardless of their sCr and urine output prior to the RRT initiation.

In 2012, the Kidney Disease: Improving Global Outcomes (KDIGO) group assembled all the available evidence into their own clinical practice guideline for AKI and proposed the KDIGO criteria, which integrate the RIFLE and AKIN criteria (Table 3) [3]. The KDIGO criteria diverge from the AKIN criteria in that the time course for a 1.5-fold increase in sCr from baseline was changed from within 48 h to within 7 days. Thus, as the KDIGO criteria encompass more gradual increases in sCr, they have made the number of patients diagnosed with AKI likely to increase.

As described above, three sets of diagnostic criteria for AKI have been proposed: the RIFLE, AKIN, and KDIGO criteria. The utility of the KDIGO criteria, the most recent of the three sets, has been compared with that of the two older sets. In a prospective, multicenter observational study of 3107 intensive care unit (ICU) patients, Luo et al. reported the percentages of patients diagnosed with AKI according to the RIFLE, AKIN, and KDIGO criteria using both the sCr and urine output, and compared their in-hospital mortality rates [7]. The percentages of patients diagnosed with AKI according to the RIFLE, AKIN, and KDIGO criteria were 46.9, 38.4, and 51.0%, respectively; thus, the number of patients diagnosed with AKI was significantly higher when using the KDIGO criteria than when using either the AKIN or RIFLE criteria. The patients diagnosed with AKI based on the KDIGO criteria had poorer survival outcomes than those diagnosed using the AKIN criteria, although there was no significant difference in the survival outcomes of patients diagnosed using the RIFLE criteria. In a retrospective multicenter study of 1005 adult patients hospitalized for acute heart failure, Li et al. compared the percentages of patients diagnosed with AKI within 7 days of hospitalization using the KDIGO, AKIN, and RIFLE criteria, as well as the patients’ in-hospital mortality rates [8]. Using only the sCr criterion, the percentages of patients diagnosed with AKI according to the KDIGO, AKIN, and RIFLE criteria were 38.9, 34.7, and 32.1%, respectively. A total of 110 patients (10.9%) were diagnosed with AKI with the KDIGO criteria but not with the RIFLE or AKIN criteria. A total of 18.4% of the patients who died in the hospital were diagnosed with AKI according to the KDIGO criteria only; this group was at a high risk of in-hospital death. In a study of 1050 patients hospitalized for acute myocardial infarction, Rodrigues et al. compared the percentages of patients diagnosed with AKI according to the RIFLE and KDIGO criteria using the sCr criterion only, as well as their mortality rates [9]. A total of 14.8% of patients were diagnosed with AKI with the RIFLE criteria, versus 36.6% with the KDIGO criteria. In comparison with patients without AKI, the 30-day and 1-year mortality hazard ratios for patients diagnosed with AKI according to the KDIGO criteria but not to the RIFLE criteria were 2.55 and 2.28, respectively.

In other studies of the AKI diagnostic criteria in hospitalized patients [10, 11], ICU patients [12, 13, 14], acute decompensated heart failure [15], patients after cardiac surgery [16], and sepsis [17], the KDIGO criteria were reported to be equal or superior to the RIFLE and AKIN criteria in their ability to predict the survival outcomes. Based on the above, the KDIGO criteria are considered to be more useful in their survival outcome prediction ability than the RIFLE or AKIN criteria for the diagnosis of AKI.

PubMed was searched for relevant studies published between January 1990 and July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Several methods have been suggested to estimate the baseline renal function. However, compared to the use of the known baseline function, all of these methods have been reported to yield a certain rate of false positives or false negatives in their AKI diagnoses and mortality predictions.

The diagnosis of acute kidney injury (AKI) requires the baseline renal function; however, in actual clinical practice, the patient’s history of examination and his/her baseline renal function are often unknown. These cases require an estimation of the baseline renal function, and many methods have been proposed (Table 4).

To compare the estimated baseline renal function with the known baseline renal function, we identified seven observational studies that used the AKI diagnosis as an outcome [18, 19, 20, 21, 22, 23, 24], and two observational studies that used the all-cause mortality as an outcome [20, 22]. While two of the seven studies included all hospitalized patients [19, 22], two of them limited the subjects to intensive care unit (ICU) patients [23, 24], two used only patients undergoing cardiac surgery [20, 21], and one study only included patients with cirrhosis [18]. All these studies assumed the lower limit of the normal renal function to be an estimated glomerular filtration rate (eGFR) of 75 ml/min/1.73 m² (as suggested by the KDIGO clinical practice guideline [3]) and examined a way to back-calculate the serum creatinine (sCr) based on the MDRD equation. In the six studies whose subjects included all hospitalized patients, the ICU patients only, and the cardiac surgery patients only, an assumed baseline renal function of eGFR 75 ml/min/1.73 m² yielded false positive AKI diagnoses. Four of these studies [19, 21, 23, 24] stated that false positives were especially frequent in patients with a known eGFR < 60 ml/min/1.73 m². On the contrary, in the study whose subjects included cirrhosis patients only, an assumed baseline renal function of eGFR 75 ml/min/1.73 m² yielded false negative AKI diagnoses. In a study that was not taken into account due to its unsuitable outcome, Zavada et al. indicated that the estimated sCr was higher than the known sCr in young people [25]. The two observational studies that used the all-cause mortality as an outcome reported that the mortality rates were reduced by sCr estimation methods that frequently yielded false positive AKI diagnoses, while the mortality rates were increased by estimation methods that frequently yielded false negative diagnoses.

In conclusion, there is currently no specific baseline sCr estimation method on par with a measured baseline sCr. The easy method that involves the calculation of the sCr based on an eGFR of 75 ml/min/1.73 m² is tolerable; however, this method often overestimates the sCr in young people and cirrhosis patients, and underestimates it in chronic kidney disease (CKD) patients. Therefore, we suggest that whenever possible, the baseline renal function should be determined using multiple methods, while also confirming whether the CKD and other comorbidities are present based on methods such as image searches to check for renal atrophy.

PubMed was searched for relevant studies published up to July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

We identified seven observational studies that used death as an outcome. In the studies of ICU patients, the inclusion of the urine output as a criterion significantly improved the survival outcome predictions. In one of these studies, the renal outcome prediction was also improved; however, a study of patients after cardiac surgery indicated a potential for overdiagnosis. Because no study of outpatients or general ward patients have been conducted, it is unclear whether the above results can be generalized.

During the 10 years since the concept of AKI was introduced, three sets of diagnostic criteria/classifications have been proposed: the RIFLE, AKIN, and KDIGO. All these criteria sets enable the diagnosis and staging of AKI based on changes in the serum creatinine (sCr) or the urine output [1, 2, 3]. In many previous clinical studies, AKI was diagnosed and staged according to the sCr alone, and a slight increase in the sCr was reported to affect the survival outcomes [3, 6]. However, few clinical studies have used the urine output as a criterion for the diagnosis and staging of AKI. Therefore, we examined whether the urine output reflects the survival outcomes of AKI as accurately as the sCr, and whether the inclusion of the urine output in the determination of the AKI stage reflects the survival outcomes more accurately than determination based on the sCr alone.

To compare the sCr and the urine output, we adopted seven observational studies that used death as an outcome [26, 27, 28, 29, 30, 31, 32]. All these studies were conducted in intensive care units (ICUs); none of them involved outpatients or patients in general wards. Regarding the AKI diagnostic criteria, three studies used the RIFLE criteria [26, 27, 30], two used the AKIN criteria [31, 32], and two used the KDIGO criteria [28, 29]. In six of these studies, the inclusion of the urine output with the sCr in the AKI diagnosis significantly improved the survival outcome predictions [27, 28, 29, 30, 31, 32]; furthermore, in one of these six studies, the renal outcome prediction ability was also improved [27].

In an analysis of 155,624 patients hospitalized in ICUs on an emergency admission, Harris et al. reported that the urine output was a more powerful predictor of the survival outcomes than the sCr [26]. In a study of 32,045 adult ICU patients classified according to the KDIGO sCr and urine output criteria, Kellum et al. demonstrated that patients who fulfilled both the sCr and urine output criteria were at the highest risk of death and the initiation of permanent renal replacement therapy (RRT), while isolated oliguria was associated with a long-term risk of death even when the sCr criterion was not fulfilled [27]. Similarly, in an analysis of 390 septic shock patients, Leedahl et al. reported that persistent oliguria was a risk factor for death by day 28 [28]. In a study of 260 ICU patients, Wlodzimirow et al. compared the combined use of the RIFLE’s sCr and urine output criteria (RIFLEsCr + urine output) with the use of the sCr criterion (RIFLEsCr) alone; they reported that the RIFLEsCr was associated with a delayed AKI diagnosis and higher mortality [29]. Furthermore, Han et al. [30] and Macedo et al. [31] also reported that the addition of the urine output criterion enabled a more accurate AKI diagnosis than the use of the sCr criterion alone. Although some studies have featured different urine output criterion values, overall, the assessment of the urine output has been shown to improve the accuracy of the AKI diagnosis. However, in a comparison of the sCr criterion alone with the urine output criterion alone for the diagnosis of AKI in patients after cardiac surgery, Lagny et al. indicated that the use of the urine output criterion alone could lead to overdiagnosis [32]. In a recent multicenter prospective study that assessed the association between the hourly urine output and mortality, Vaara et al. reported that patients who fulfilled both the sCr and urine output criteria had the highest rate of RRT initiation and the highest 90-day mortality, while isolated oliguria was associated with poor outcomes; these results affirm the importance of measuring the hourly urine output and the need to combine the urine output criterion with the sCr criterion [33]. Moreover, in using the urine output criterion in the diagnosis and staging of AKI, there is a concern that the use of diuretics may change the urine output, causing underestimation of the AKI severity. However, in their analysis of the effects of diuretics on the AKI diagnosis, Han et al. reported that the inclusion of the urine output criterion alongside the sCr criterion played an additional role in the diagnosis and staging of AKI regardless of whether diuretics were used [30].

To summarize the above studies, the inclusion of the urine output along with the sCr to determine the AKI stage improves the sensitivity of the AKI diagnosis, and yields more accurate reflections of the survival and renal outcomes than AKI staging based on the sCr alone. Therefore, we suggest that AKI staging should involve the urine output whenever possible.

PubMed was searched for relevant studies published between January 1990 and August 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

We identified seven papers that assessed the risk of development of AKI in cardiac surgery. All of them were observational studies. Certain observational studies have stated that TAVR and TAVI, which have become more common with the recent aging of society, do not match the same risk of AKI observed in cardiac surgery.

PubMed was searched for relevant studies published between March 2011 and December 2015, and papers related to the present CQ were identified from the search results. The literature published before March 2011 was referenced from the KDIGO clinical practice guideline for AKI.

Open image in new window

Among 10 observational studies on the development of AKI following liver transplantation, 5 studies demonstrated a significant association between the development of AKI and the intraoperative blood transfusion volume. Two studies excluded chronic kidney disease (CKD), while two others found CKD to be a significant risk factor for AKI development. Two studies demonstrated that the MELD score and intraoperative hypotension or the use of vasopressors were associated with the development of AKI. Only three studies about lung transplantation and AKI development were found, and these studies did not demonstrate a consistent trend.

PubMed was searched for relevant studies published between March 2011 and December 2015, and papers related to the present CQ were identified from the search results. The literature published before March 2011 was referenced from the KDIGO clinical practice guideline for AKI.

Open image in new window

Among the 11 identified observational studies that included the development of AKI as an outcome, 5 were multicenter studies involving more than 1000 subjects. In multivariate analyses, the following risk factors were found to be significantly associated with AKI: comorbid CKD (4 studies), aging (4 studies), diabetes (3 studies), and cardiac dysfunction (3 studies). Other factors found to be associated with AKI were the diuretic resistance, hypotension (defined as a systolic blood pressure < 90 mmHg), and elevated urinary neutrophil gelatinase-associated lipocalin (NGAL) (2 studies each).

PubMed was searched for relevant studies published between March 2011 and December 2015, and papers related to the present CQ were identified from the search results. The literature published before March 2011 was referenced from the KDIGO Clinical Practice Guideline for AKI.

Open image in new window

In 6 observational studies that examined the risk of AKI development in sepsis, pre-existing renal dysfunction, aging, and the use of renin–angiotensin–aldosterone system inhibitors were found to be associated with AKI development in sepsis.

PubMed was searched for relevant studies published between March 2011 and December 2015, and papers related to the present CQ were identified from the search results. The literature published before March 2011 was referenced from the KDIGO clinical practice guideline for AKI.

Open image in new window

In a meta-analysis of eight observational studies, the mortality was significantly higher in hospital-acquired AKI than in community-acquired AKI (odds ratio 2.79, 95% CI 2.18–3.56). In studies that used the RIFLE or KDIGO criteria, community-acquired AKI featured a high rate of stage 3 AKI, while hospital-acquired AKI featured a high rate of stage 1 AKI.

Acute kidney injury (AKI) is primarily treated with conservative therapies, such as the optimization of fluid volume or blood pressure and the avoidance of nephrotoxins; in addition, identification of the cause of the kidney injury is recommended. Therefore, it is crucial to recognize the risk factors for AKI and take steps to prevent it in order to improve its outcomes [3].

Acute kidney injury encompasses a broad spectrum of diseases and can occur in the hospital or in the community. However, although community-acquired AKI occurs frequently in low- and median-income countries which account for roughly 85% of the world population [3], 80–90% of studies have examined hospital-acquired AKI in high-income countries [34]; few studies have compared hospital-acquired and community-acquired AKI. Hospital-acquired AKI is frequently caused by ischemia, nephrotoxins, and sepsis [87], while community-acquired AKI has been found to frequently derived from preventable causes such as dehydration, infection, and childbirth [88]. To determine the state of community-acquired AKI in low- and median-income countries, the multinational The 0 by 25 initiative conducted the global snapshot study in 2015 [89].

In order to develop the present guideline, PubMed was used to identify papers that compared hospital-acquired and community-acquired AKI. Eight observational studies were identified [90, 91, 92, 93, 94, 95, 96, 97]; among them, two defined AKI based on the RIFLE criteria [94, 95], two used the KDIGO criteria [96, 97], and the other four were published before the RIFLE and KDIGO criteria were proposed [90, 91, 92, 93]. Four studies were conducted in high-income countries [91, 94, 95, 96], while the other four were conducted in low- and median-income countries [90, 92, 93, 97]. In all of these studies, community-acquired AKI was associated with a lower mortality (Fig. 1) and shorter hospitalization duration. Moreover, the percentages of patients at each AKI stage (i.e. degree of severity) in the studies that used the AKIN criteria indicated that in all four studies, community-acquired AKI was more severe (i.e., with low percentages of stages 1 and 2 AKI and high percentages of stage 3 AKI), while hospital-acquired AKI showed higher percentages of mild cases (Fig. 2).

Open image in new window

Open image in new window

Thus, the above-cited studies demonstrate that hospital-acquired AKI and community-acquired AKI have different clinical pictures, as shown in the Table 8. The relationship between the severity and mortality may differ between hospital-acquired AKI and community-acquired AKI; therefore, we suggest that they be discriminated from one another.

However, all the studies used were conducted outside Japan. Further investigation comparing hospital-acquired AKI and community-acquired AKI in Japan is necessary.

PubMed was searched for relevant studies published up to December 2015, and papers that compared hospital-acquired and community-acquired AKI were identified from the search results.

Open image in new window

In a meta-analysis based on nine observational studies, compared to non-septic AKI, septic AKI resulted in a higher in-hospital mortality (odds ratio 2.48, 95% CI 1.76–3.49) and a higher ICU mortality (odds ratio 1.60, 95% CI 1.52–1.69). Although studies that assessed the in-hospital mortality featured publication bias, no such bias was observed related to ICU mortality.

In a report of a large-scale prospective observational study conducted at 54 centers in 23 countries [34], the cause of acute kidney injury (AKI) in the intensive care unit (ICU) was septic shock in 47.5% of cases and cardiogenic shock in 26.9% of cases. In a large-scale multinational multicenter prospective observational study published in 2015 [98], AKI occurred in 57.3% of ICU patients; the cause of AKI was sepsis in 40.7% of patients and cardiogenic shock in 13.2% of patients. In Japanese epidemiology, the Diagnosis Procedure Combination (DPC) database has been used to examine AKI patients who underwent continuous renal replacement therapy (CRRT) [99]. Among these patients, the most common causes of AKI were cardiovascular disease and other medical diseases, which accounted for approximately half of patients, followed by sepsis and cardiovascular surgery; compared to all other causes, mortality was low only for cardiovascular surgery.

In developing the present guideline, PubMed was used to identify papers which compared septic and non-septic AKI. Nine observational studies were identified [80, 100, 101, 102, 103, 104, 105, 106, 107]; seven of these studies were prospective, while two were retrospective. One of these studies was a retrospective study by Bagshaw et al., which utilized the Australian and New Zealand Intensive Care Society (ANZICS) database [80]; the study featured 14,039 septic AKI patients and 29,356 non-septic AKI patients, a prominently large number of patients compared to other studies. Seven studies compared in-hospital mortality [80, 100, 101, 102, 103, 104, 105], while five studies compared ICU mortality [80, 102, 103, 106, 107].

As for AKI diagnostic criteria, six studies used the RIFLE criteria [80, 101, 102, 103, 104, 105]; the remaining three studies [100, 106, 107] were published before the RIFLE criteria were proposed. Percentages of patients by RIFLE criteria severity were listed in four studies [80, 101, 102, 104]; Risk was the most common level of severity in one study [102], while Injury was the most common in two studies [80, 101], and Failure was the most common in one study [104]. Causes of sepsis were demonstrated in two studies [101, 103]; in these studies, sepsis was caused by intrathoracic infections (such as pneumonia) and intra-abdominal infections in approximately 30 and 25% of cases, respectively, thus accounting for more than half of all cases. The severities of patients’ illnesses were assessed with the APACHE II, SAPS, and SAPS II severity scores in eight studies [80, 100, 101, 102, 103, 105, 106, 107]; in all of these studies, septic AKI was more severe than non-septic AKI.

Figure 3 shows the results of meta-analysis of seven studies which compared in-hospital mortality. In-hospital mortality and ICU mortality may both be higher for septic AKI than for non-septic AKI; therefore, we suggest that the two forms of AKI be discriminated from each other. Septic AKI should be handled in specific ways, such as admission of patients to the ICU depending on severity, consideration of hemodynamic monitoring, and maintaining fluid volume and renal perfusion pressure.

Open image in new window

Searches were conducted on PubMed for literature published up to November 2015. Papers which compared septic and non-septic AKI were identified from the search results.

Open image in new window

In a meta-analysis of 10 observational studies, the in-hospital mortality from renal AKI was higher than that from pre-renal AKI (odds ratio 3.63, 95% CI 1.68–7.83). A significant publication bias was present.

Acute kidney injury (AKI) is classified as either pre-renal, renal (intrinsic), or post-renal. Pre-renal AKI is considered as azotemia resulting from a decreased renal perfusion pressure; conceptually, it is a form of renal impairment with no renal tissue damage, in which the renal function can recover rapidly with early treatment. There are two conceivable approaches to the differentiation of pre-renal AKI from renal AKI. The first approach is to comprehensively assess whether the AKI is pre-renal or renal based on an assessment of the cause of the AKI, hemodynamics, and urinalyses with measuring factors such as the body weight change, vital signs, urine osmality, fractional excretion of sodium (FENa), fractional excretion of urea nitrogen (FEUN), and urinary sediment. The second approach is to determine whether the renal function recovers immediately after fluid resuscitation. If the renal function recovers within 2–3 days after appropriate fluid resuscitation, the AKI is considered to be volume-responsive, which allows for clinical classification as pre-renal AKI. If the renal function does not recover despite fluid resuscitation, the AKI is considered to be volume-unresponsive, which corresponds to renal AKI. However, when a continued or prolonged reduced renal perfusion pressure results in renal parenchymal injury, or when the reduced renal perfusion pressure is accompanied by a low cardiac output, sepsis, or liver failure, the renal function does not necessarily recover with fluid resuscitation alone [108]. Therefore, even if the AKI is initially assessed as pre-renal, a second test should be performed within 3 days. However, even if the AKI is assessed as pre-renal, a mild elevation in the urinary biomarkers can sometimes be suggestive of a renal tissue injury [109]. As AKI is known to be involved in injuries to multiple organs, including the heart and lungs, even pre-renal AKI may affect the survival prognosis.

Many studies have reported that the in-hospital mortality is lower in volume-responsive AKI—in which the renal function recovers within 3 days of intervention—than in volume-unresponsive AKI. In a recent AKI cohort study of 283 patients in intensive care units (ICUs) at multiple hospitals, the in-hospital mortality rates for non-AKI, volume-responsive AKI, and renal AKI were 23.8, 29.6, and 38.9%, respectively; thus, renal AKI showed the worst outcomes [110]. However, in a study that evaluated AKI based on its underlying causes at diagnosis, the in-hospital mortality was 27.3% in pre-renal AKI versus 19.3% in intrinsic AKI; although the difference was not significant, pre-renal AKI tended to have worse outcomes [111].

To develop the present guideline, PubMed was used to identify papers that compared renal and pre-renal AKI in order to assess the difference in the survival outcomes. Ten cohort studies were identified; among them, three differentiated between pre-renal and renal AKI based on the underlying causes of AKI and the urine findings at diagnosis [111, 112, 113], while seven differentiated between pre-renal and renal AKI based on the volume responsiveness [109, 110, 114, 115, 116, 117, 118]. In our meta-analysis, the in-hospital mortality was significantly higher in renal AKI than in pre-renal AKI (Fig. 4).

Open image in new window

Based on the above, pre-renal AKI and renal (intrinsic) AKI have different survival outcomes; therefore, we suggest that they should be distinguished from one another.

PubMed was searched for relevant studies published up to August 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Multiple systematic reviews/meta-analyses have found the urinary NGAL and L-FABP to serve as useful markers for the early diagnosis of AKI. However, future clinical trials that compare AKI interventions based on the conventional diagnostic method using the serum creatinine levels with those based on diagnoses made with urinary biomarkers are necessary to examine whether novel urinary biomarkers are truly useful for the diagnosis of AKI.

Only one systematic review/meta-analysis has assessed the utility of the urinary cystatin C; therefore, firm conclusions as to its utility for the early diagnosis of AKI cannot be made.

The pathological condition previously recognized as acute renal failure (ARF) is now broadly understood to pose a risk of death at an earlier or milder stage than failure. This has prompted a paradigm shift from ARF to acute kidney injury (AKI). However, with the present method of diagnosis, which is based on the identification of an increased level of serum creatinine (sCr) and a reduced urine output, interventions are often mistimed; therefore, there is an urgent need for the clinical application of more sensitive biomarkers. The early diagnosis of AKI enables earlier consultation with a nephrologist, appropriate management of the renal hemodynamics, and the avoidance of exposure to nephrotoxins. Therefore, we examined whether urinary biomarkers should be used for the early diagnosis of AKI based on a relatively large number of studies on AKI in adult patients having received cardiovascular surgery and those in intensive care units (ICUs).

Neutrophil gelatinase-associated lipocalin (NGAL) is a low molecular weight protein (molecular weight: approximately 25,000) that belongs to the lipocalin protein family and is secreted by activated neutrophils. In addition to inducing kidney development and possessing renoprotective and antibacterial effects, NGAL is also expressed in the distal nephron in kidney injury. Multiple systematic reviews/meta-analyses have found the urinary NGAL to be useful for the early diagnosis of AKI [119, 120, 121, 122, 123, 124]. Among the studies cited in these systematic reviews/meta-analyses, 16 (for a total of 2194 patients) were related to the present CQ (Table 9) [125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140]. The subjects consisted of patients who had undergone cardiovascular surgery (14 studies, 1531 patients in all) [125, 126, 128, 129, 130, 131, 132, 134, 135, 136, 137, 138, 139, 140] and ICU patients (two studies, 663 patients in all) [127, 133]. The majority of these studies defined AKI according to the RIFLE or AKIN criteria (i.e. an increase in sCr) or to criteria conforming to the RIFLE or AKIN ones. A total of 549 patients (25%) were diagnosed with AKI. In an assessment of the early diagnostic capacity of the urinary NGAL over the 6-h period immediately after surgery or ICU admission, the area under the receiver operating characteristic curve (AUC) was 0.50–0.98 (0.77 with an unweighted mean). In 75% (12/16) of the studies, the AUC was ≥ 0.70, thus showing moderate-or-better diagnostic accuracy; therefore, the urinary NGAL was found to be useful for the early diagnosis of AKI. However, the clinical studies related to the present CQ raised several issues about the clinical application of the urinary NGAL, including the following: some studies did not use officially approved measurement methods; multiple measurement methods were used, and they were not standardized; there was no set cutoff value; urinary tract infections and urologic diseases increase the urinary NGAL levels [141]; and there are very few relevant clinical studies of Japanese subjects.

The L-type fatty acid-binding protein (L-FABP) is a low molecular weight protein (molecular weight: approximately 14,000) localized in the cytoplasm of human renal proximal tubular cells. By binding to free fatty acids and transporting them to mitochondria and peroxisomes, the L-FABP promotes beta-oxidation, contributes to energy production, and helps to maintain homeostasis. When the proximal tubule is subjected to ischemia or oxidative stress, the expression of the L-FABP is enhanced and its urinary excretion increases. The urinary L-FABP has been demonstrated to be useful for the early diagnosis of AKI [119, 120, 142] by multiple systematic reviews/meta-analyses. Among the studies cited in these systematic reviews/meta-analyses, seven (for a total of 2416 patients) were related to the present CQ (Table 10) [138, 143, 144, 145, 146, 147, 148]. The subjects consisted of patients who had undergone cardiovascular surgery (three studies, 271 patients in all) [138, 143, 148] and ICU patients (four studies, 2,145 patients in all) [144, 145, 146, 147]. These studies generally defined AKI according to the RIFLE or AKIN criteria (i.e. an increase in sCr). A total of 298 patients (12%) were diagnosed with AKI. In an assessment of the early diagnostic capacity of the urinary L-FABP over the 12-h period immediately after surgery or ICU admission, the AUC was 0.70–0.95 (0.81 with an unweighted mean). In all seven studies, the AUC was ≥ 0.70, thus showing moderate-or-better diagnostic accuracy; therefore, the urinary L-FABP was found to be useful for the early diagnosis of AKI. However, the timing of the urinary L-FABP measurement must be chosen carefully according to the different AKI etiologies. Measuring reagents for the L-FABP are available in Japan; these are standardized and covered by public health insurance.

Cystatin C is a low molecular weight protein (molecular weight: approximately 13,000) produced by nucleated cells all over the body that inhibits the cell injury caused by cysteine proteases. After being secreted outside cells, cystatin C is filtered by the glomerulus; 99% of it is then absorbed by the proximal tubule and catabolized. Therefore, tubular injuries are affected by the reabsorption of cystatin C; consequently, the cystatin C concentration in urine has been examined as a biomarker of AKI. There has been one systematic review/meta-analysis of the early diagnostic capacity of the urinary cystatin C [149]; the latter cited six studies [118, 125, 132, 134, 150, 151]. In a pool analysis of four studies that used urinary creatinine-adjusted data [118, 132, 150, 151], the sensitivity for the early diagnostic capacity of urinary cystatin C was 0.52, the specificity was 0.70, and the AUC was 0.64 (95% CI 0.62–0.66); therefore, the diagnostic accuracy was low, indicating that the urinary cystatin C is of limited utility for the early diagnosis of AKI. The measurement of cystatin C in urine samples is not covered by public health insurance in Japan. Although the urinary albumin has been reported to be a potential biomarker of early AKI [152], the number of relevant studies is limited; thus, the utility of the urinary albumin as a biomarker remains unknown.

As described above, the urinary NGAL and L-FABP have been identified as useful biomarkers for the early diagnosis of AKI. However, future clinical trials that compare AKI interventions based on the conventional diagnostic method using the serum creatinine levels with those based on diagnoses made with urinary biomarkers are necessary to examine whether novel urinary biomarkers are truly useful for the diagnosis of AKI.

PubMed was searched for relevant studies published up to August 2015, and papers related to the present CQ were identified from the search results.

Search query: (((“acute kidney injury”[MeSH Terms] OR “acute kidney injury”[tw] OR “acute renal failure”[tw]) AND (“biological markers”[MeSH Terms] OR “biological markers”[All Fields] OR “biomarker”[All Fields])) AND (“diagnosis”[Subheading] OR “diagnosis”[All Fields] OR “diagnosis”[MeSH Terms])) AND (Meta-Analysis[PT] OR systematic[SB]).

Open image in new window

Multiple systematic reviews/meta-analyses about the use of the urinary NGAL to predict the AKI severity and survival outcomes have suggested that the urinary NGAL is potentially useful, albeit limitedly, in predicting the severity in relation to death and renal replacement therapy initiation. The number of studies on the urinary L-FABP and cystatin C is limited; therefore, their utilities in predicting the AKI severity in relation to death and renal replacement therapy initiation are unclear.

Acute kidney injury (AKI) has been shown to be involved not only in short-term kidney injury, but also in the subsequent renal outcomes and mortality. Therefore, the prediction of these outcomes is of clinical importance. In addition to the serum creatinine (sCr), the cystatin C, and the estimated glomerular filtration rate (eGFR), all of which reflect the renal function, studies have examined the urinary NGAL, NAG, L-FABP, and cystatin C as biomarkers of kidney injury [119, 120, 121, 123, 149, 153]. The results of several systematic reviews and meta-analyses of these studies have been reported in recent years.

Regarding the utility of the urinary NGAL for the prediction of the renal outcomes and mortality, multiple systematic reviews/meta-analyses have indicated that the urinary NGAL could help to predict the AKI severity in relation to death and the initiation of renal replacement therapy (RRT). In a meta-analysis of nine studies (a total of 1948 patients), the odds ratios for RRT requirement and in-hospital mortality based on increased urinary and serum NGAL levels were 12.9 and 8.8, respectively [124]. Meanwhile, in an analysis of 13 studies (a total of 1079 patients) on the recovery of the renal function after kidney transplantation, the NGAL was shown to be a useful predictor of AKI, with an area under the curve (AUC) of receiver operating characteristic (ROC) is 0.87 [122]. However, these meta-analyses examined both urine and blood specimens; therefore, the results should be interpreted cautiously. Another cautionary point is that reports of NGAL measurements have used multiple measurement kits with different upper and lower limits.

In a meta-analysis of studies that used the urinary L-FABP to predict the need for RRT (three studies, 436 patients in all) and the in-hospital mortality (three studies, 561 patients in all), while no significant difference was observed in the need for RRT, the in-hospital mortality odds ratio was 13.7 (p = 0.008) [142]. Although measurement of the urinary L-FABP is covered by public health insurance in Japan, in principle, it can only be calculated once every 3 months. In addition, the urinary L-FABP has been reported to be elevated in patients with diabetes and chronic kidney diseases.

Regarding examinations of the cystatin C, a meta-analysis of seven studies (a total of 2941 patients) showed that high levels of cystatin C are a risk factor for death (odds ratio 2.3) [154]. However, both blood and urine samples were examined in this analysis. Although measurement of the serum cystatin C is covered by public health insurance in Japan, that of the urinary cystatin C is not.

Other AKI markers that have been used in the past include the NAG, which increases by release from the tubular epithelium brush border into urine, and the β2microglobulin (β2MG) and α1microglobulin (α1MG), which will be increased by impaired tubular epithelial cell reabsorption. However, these markers are fraught with problems, as the samples are unstable (e.g. they are subject to changes in the urinary pH) and they are easily affected by the serum concentration of β2MG and α1MG. In addition, the levels of these markers can be increased by tubular disorders caused by proteinuria associated with glomerular injuries.

In measuring and assessing these urinary biomarkers, one should be cautious about the timing of the sample collection. In surgeries that use cardiopulmonary bypasses, the levels of the urinary NGAL and L-FABP increase 2–6 h after the surgery before declining gradually. Other biomarkers’ levels also peak within a relatively short period and then decline. Therefore, if the timing of the AKI development is unknown, it is necessary to consider whether the testing was performed at the appropriate time for measurement.

PubMed was searched for relevant studies published up to August 2015, and papers related to the present CQ were identified from the search results.

Search query: (((“acute kidney injury”[MeSH Terms] OR “acute kidney injury”[tw] OR “acute renal failure”[tw]) AND (“biological markers”[MeSH Terms] OR “biological markers”[All Fields] OR “biomarker”[All Fields])) AND (“diagnosis”[Subheading] OR “diagnosis”[All Fields] OR “diagnosis”[MeSH Terms])) AND (Meta-Analysis[PT] OR systematic[SB]).

Open image in new window

Observational studies have reported that the urinary NGAL is mildly elevated in pre-renal AKI and highly elevated in renal AKI; therefore, the urinary NGAL can be useful in the differentiation of pre-renal from renal AKI. However, the measurement points and cutoff values have not yet been determined. Therefore, we recommend incorporating other laboratory findings and physical findings to differentiate pre-renal from renal AKI. The utility of other urinary biomarkers in this regard is unknown.

The conventional indicators for the differentiation of pre-renal acute kidney injury (AKI) and renal AKI include the urine osmolality, the fractional excretion of sodium (FENa), the fractional excretion of urea nitrogen (FEUN), and urine sediment findings; however, none of these tests can be considered sufficiently sensitive or specific. There have been no systematic reviews or meta-analyses of studies on the use of urinary biomarkers for the differentiation of pre-renal from renal AKI; only a small number of observational studies are available. In multiple studies in which patients diagnosed with AKI were divided into patients with pre-renal or renal AKI, the degree of elevation of the urinary NGAL was found to be potentially useful for the differentiation of these two types of AKI.

In Nickolas et al.’s examination of the urinary NGAL in 635 patients hospitalized after emergency room visits, the mean urinary NGAL level was significantly higher in renal AKI patients (n = 30; 416 ± 387 µg/gCr) than in pre-renal AKI patients (n = 88; 30.1 ± 92.0 µg/gCr) [114]. In a report of 145 hospitalized patients by Singer et al., the median urinary NGAL level was significantly higher in renal AKI patients [n = 75; 255.6 µg/L (98.5-872.9 µg/L)] than in pre-renal AKI patients [n = 32; 31.3 µg/L (15.9–75.5 µg/L)] [117]. Moreover, a urinary NGAL cutoff level of 104 µg/L yielded a high sensitivity (0.75), high specificity (0.88), and high positive likelihood ratio (5.97) for the diagnosis of renal AKI. In a report by Seibert et al., the urinary NGAL levels within 3 days of hospital admission were significantly higher in renal AKI patients (458.1 ± 695.3 ng/ml) than in pre-renal AKI patients (64.8 ± 62.1 ng/ml). In addition, a urinary NGAL cutoff level of 52 ng/ml yielded high sensitivity (0.75), high specificity (0.72), and a high area under the receiver operating characteristic curve (AUC 0.89) for the diagnosis of renal AKI [155].

Although one study found that the urinary NGAL levels of pre-renal AKI patients were not elevated [156], other studies have reported that the urinary biomarker levels were mildly but significantly higher in pre-renal AKI patients than in patients without AKI. In a study by Doi et al. in which 129 out of 337 patients who were admitted to the intensive care unit (ICU) were diagnosed with AKI and in which transient AKI (pre-renal AKI) was defined as the recovery of the serum creatinine (sCr) to within 0.3 mg/dl above baseline within 48 h, 51 patients were diagnosed with transient AKI [109]. Upon ICU admission, transient AKI patients’ levels of urinary NGAL, urinary L-FABP, NAG, and urinary albumin were mildly but significantly higher than those of non-AKI patients. Nejat et al. compared the urinary biomarkers upon ICU admission of 285 non-AKI patients, 61 pre-renal AKI patients (with pre-renal AKI defined as a FENa < 1.0% and recovery of the sCr levels within 48 h), and 114 renal AKI patients [157]. The median urinary NGAL levels of non-AKI patients and of pre-renal AKI patients were 7.7 µg/mmolCr (3.3–35 µg/mmolCr) and 14 µg/mmolCr (6.5–56 µg/mmolCr), respectively; thus, pre-renal AKI patients showed a tendency to have a mildly higher urinary NGAL level (p = 0.052). In addition, the median urinary NGAL level of renal AKI patients was 44 µg/mmolCr (16–345 µg/mmolCr), which was significantly higher than that of pre-renal AKI patients. The median urinary cystatin C levels of non-AKI patients and pre-renal AKI patients were 0.026 mg/mmolCr (0.010–0.12 mg/mmolCr) and 0.054 mg/mmolCr (0.017–0.53 mg/mmolCr), respectively; thus, the urinary cystatin C was significantly higher in pre-renal AKI patients than in non-AKI patients. The median urinary cystatin C level of renal AKI patients was 0.21 mg/mmolCr (0.05–1.9 mg/mmolCr), which was significantly higher than that of pre-renal AKI patients.

As described above, the urinary NGAL is mildly elevated in pre-renal AKI patients and highly elevated in renal AKI patients, which suggests that the urinary NGAL is potentially useful for the differentiation of pre-renal from renal AKI. However, the measurement points, cutoff values, and the need for urine creatinine correction have not yet been determined; these issues must be considered in the future. Therefore, pre-renal and renal AKI cannot be differentiated based on the urinary NGAL alone; hence, we recommend a comprehensive assessment that also incorporates other laboratory and physical findings.

PubMed was searched for relevant studies published up to November 2015, and papers related to the present CQ were abstracted from the search results.

Search query: (“acute kidney injury”[MeSH Terms] OR acute kidney failure[tw] OR acute renal failure[tw] OR acute kidney injury[tw] OR acute kidney injuries[tw] OR acute kidney injury[tw] OR acute kidney injury[tw] OR acute renal injuries[tw] OR acute renal injury[tw] OR acute kidney insufficiencies[tw] OR acute kidney insufficiency[tw] OR acute renal insufficiencies[tw] OR acute renal insufficiency[tw] OR acute tubular necrosis[tw] OR ARI[tw] OR AKI[tw] OR ARF[tw] OR AKF[tw] OR ATN[tw] AND NGAL[tw] OR neutrophil gelatinase-associated lipocalin[tw] OR l-FABP[tw] OR liver-type fatty acid-binding protein[tw] OR NAG[tw] OR N-acetyl-β-d-glucosaminidase[tw]) AND (“pre-renal”[tw] OR “pre-renal azotaemia”[tw] OR prerenal[tw]) NOT (child[tw] OR children[tw] OR infant[tw] OR pediatrics[tw]).

Open image in new window

Since atrial natriuretic peptide (ANP) preparation “carperitide” is covered by health insurance for treatment of congestive heart failure in Japan, we only investigated randomized controlled trials (RCTs) of ANP in which carperitide was used. A 2009 Cochrane review [158] suggested that low-dose ANP may reduce the frequency of renal replacement therapy (RRT) in the setting of AKI prevention. However, the 2012 KDIGO Clinical Practice Guideline for AKI [3] and a 2013 Cochrane review [159] carefully assessed individual pieces of evidence, and lead to a revised conclusion that there is insufficient evidence to declare that low-dose ANP is effective for treatment or prevention of AKI. Regarding 2009 and 2011 reports on cardiovascular surgery which focused on AKI prevention [160, 161], doubts were raised concerning issues such as randomization and blinding methods. In a 2011 paper, administration of low-dose ANP resulted in a significantly reduced rate of RRT after 1 year [161]. Since 2011, there have been two new RCTs related to AKI prevention; however, due to few numbers of patients, we judged these RCTs lack sufficient statistical power. Currently, there is no strong evidence indicating that low-dose ANP is ineffective for prevention or treatment of AKI, but rather the evidence that does indicate its effectiveness is of insufficient quality.

Atriuretic peptide (ANP) is a circulating hormone that was discovered in Japan. Along with brain natriuretic peptide (BNP) and C-type natriuretic peptide (CNP), they make up the natriuretic peptide family [162, 163, 164]. In healthy conditions, ANP is produced from the atria; however, in heart failure, the production and secretion of ANP from both the atria and ventricles are enhanced [164, 165]. ANP possesses multiple independent modes of actions, including vasodilation, inhibition of sodium reabsorption, inhibition of water reabsorption, elevation of glomerular filtration rate via afferent arteriole dilation and efferent arteriole constriction, reduction of renin activity, angiotensin II concentration, aldosterone concentration in the blood, and sympathetic nerve inhibition [166]. Combined together, continuous infusion of ANP or BNP on laboratory animals and humans exerts a powerful natriuretic effect [167]. Therefore, in the prevention or treatment of AKI, ANP is expected to elicit renoprotective effect through diuresis and increase of glomerular filtration rate. Many clinical studies have been carried out in this context. However, administration of high-dose ANP reduces systemic blood pressure, thereby potentially cancelling out the above-mentioned renoprotective effect. Therefore, it is crucial to identify the optimal dose of ANP for the achievement of renoprotective effect. Based on the 2012 KDIGO Clinical Practice Guideline for AKI, the present guideline defined low-dose ANP as ≤ 50 ng/kg/min and high-dose ANP as ≥ 100 ng/kg/min.

As to assessment of therapeutic effect of ANP after development of AKI, there are two large-scale randomized controlled trials (RCTs) in which more than 200 participants were assigned into two arms. In both RCTs, high-dose ANP (200 ng/kg/min, 24 h) failed to reduce the incidence of RRT [168, 169]. In a later small-scale RCT using low-dose ANP (50 ng/kg/min, mean 127 h), the ANP group exhibited a significant reduction in the frequency of RRT as compared with the placebo group [170]. There have been no subsequent RCTs investigating the therapeutic effects of low-dose ANP for AKI. Therefore, the present guideline could not offer a definitive recommendation.

On the hand, concerning prevention of AKI by ANP, 13 RCTs were found (excluding AKI from contrast-induced nephropathy); all of these were Japanese clinical trials which used low-dose ANP. In most of them, the serum creatinine (sCr) values became significantly lower in the ANP group than in the control group. However, based on a strict application of the AKI diagnostic criteria from the 2012 KDIGO Clinical Practice Guideline for AKI [3], we found no studies in which the ANP group demonstrated a significant reduction in the incidence of AKI.

In a 2009 Cochrane review by Nigwekar et al. [158], low-dose ANP was reported to potentially reduce the need for RRT during prevention of AKI in major surgery, particularly cardiovascular surgery. However, the administration of high-dose ANP for AKI treatment was shown to increase the frequency of adverse events such as hypotension and arrhythmia. The 2012 KDIGO Clinical Practice Guideline for AKI and a 2013 Cochrane review by Zacharias et al. [159] judged that the numbers of patients, the details of the randomization and blinding, and the rigor of the endpoint definitions were insufficient in previous studies. Consequently, the effectiveness of low-dose ANP for the prevention of AKI was deemed inconclusive.

Although RCTs conducted in Japan have suggested that ANP is useful for the prevention of AKI, the quality of the research methods used was debatable. Therefore, we conclude that evidence for the effectiveness of low-dose ANP both for the prevention and treatment of AKI is insufficient, making a definitive recommendation impossible. The ANP preparations used in RCTs in Japan and in the West are carperitide (product name: Hamp^®) and anaritide, respectively. Although carperitide has been available in Japan for the treatment of congestive heart failure since 1995, its use for the prevention or treatment of AKI is not covered by health insurance. Urodilatin (product name: Ularitide^®) is ANP-related hormone with four-amino-acid residues added to the N-terminus of ANP; it is produced in the distal nephron [171]. Clinical trials for ularitide have been conducted outside Japan [172], and we did not mention it in the present guideline.

Searches were conducted for relevant studies published between January 2008 and August 2015. The literature published prior to 2008 was referenced from a 2009 Cochrane review by Nigwekar et al. [158]. All RCTs related to contrast-induced nephropathy were excluded.

Open image in new window

Previous guidelines and systematic reviews do not recommend the use of loop diuretics for the prevention or treatment of AKI. There have been no new RCTs to contradict the results of previous clinical trials on loop diuretics for AKI.

Loop diuretics inhibit the sodium reabsorption and exert a diuretic effect by inhibiting the Na-K-2Cl cotransporter in the thick ascending limb of the loop of Henle. Due to their theoretical effectiveness against acute kidney injury (AKI), clinical trials involving loop diuretics have long been performed. For example, by ensuring a diuretic effect, loop diuretics can prevent the tubular obstruction induced by cell shedding; in addition, they increase the medullary oxygenation and the renal medullary blood flow.

Three randomized controlled trials (RCTs) have compared the use of loop diuretics to that of a placebo or to standard therapy for the prevention of AKI [173, 174, 175]. In a meta-analysis by Ho et al., loop diuretics failed to yield a significant improvement in the in-hospital mortality or in the percentage of patients who required renal replacement therapy (RRT) [176]. Moreover, although different RCTs have defined AKI differently, none has yet demonstrated a statistically significant reduction in the incidence of AKI in a loop diuretics group. In fact, in an RCT by Lassnigg et al., the loop diuretics group showed an increased incidence of renal dysfunction (14.6 vs 0%, p < 0.01) [174]. Based on the above, the present guideline does not recommend the use of loop diuretics for AKI prevention.

Seven RCTs have also compared the use of loop diuretics to that of a placebo or to standard therapy for the treatment of existing AKI [177, 178, 179, 180, 181, 182, 183]. In the above-mentioned meta-analysis, the loop diuretics group did not demonstrate a significant improvement in the in-hospital mortality or the percentage of patients who required RRT [176]. Although different RCTs have used different definitions of recovery from renal dysfunction, no RCT to date has demonstrated a significant increase in the percentage of patients who recovered from renal dysfunction in the loop diuretics group. Among the above-mentioned seven RCTs, two limited their subjects to AKI patients who underwent RRT; in both of these, the loop diuretics group did not demonstrate a significant reduction in the duration of the RRT or in the early recovery from renal dysfunction [182, 183]. In addition, one meta-analysis showed that high-dose furosemide, which is often used to treat AKI, significantly increased symptoms such as tinnitus and temporary deafness (as compared with their incidence in control groups) [184]. Based on the above, the present guideline does not recommend the use of loop diuretics for AKI treatment.

On the other hand, in AKI with a reduced urine output, loop diuretics may help to correct the fluid overload and to improve any electrolyte imbalance (such as hyperkalemia). However, there are currently no RCTs in which loop diuretics were administered specifically to treat AKI with these sorts of clinical manifestations—hence the above suggestion. The guidelines published by the KDIGO and NICE (National Institute for Health and Clinical Excellence) do not exclude the use of loop diuretics for the correction of fluid overload.

As for diuretics other than loop diuretics, RCTs have also examined mannitol for the prevention of AKI. In a meta-analysis by Yang et al., mannitol did not demonstrate evident effectiveness for the prevention of AKI [185]. In subsequent RCTs, the mannitol groups also failed to demonstrate significant improvement in their RRT initiation rates or in-hospital mortality [186, 187].

PubMed was searched for relevant studies published between January 2012 and April 2015, and papers related to the present CQ were identified from the search results. The literature published before January 2012 was referenced from the KDIGO Clinical Practice Guideline for AKI.

Open image in new window

The KDIGO Clinical Practice Guideline for AKI suggests not to use low-dose dopamine to prevent or treat AKI. Since the publication of the KDIGO guideline, the efficacy of low-dose dopamine for the prevention of AKI has been examined in five RCTs. None of them have found low-dose dopamine to be effective.

Dopamine has been widely used to treat severely ill patients especially before 2000. The administration of dopamine, particularly low-dose (1–3 µg/kg/min), to healthy individuals, is considered to bring increases in renal vasodilation, in natriuresis, and in the glomerular filtration rate (GFR); therefore, low-dose dopamine had been anticipated to have a renoprotective effect. However, many of the clinical studies on dopamine have been found to be of poor quality due to a variety of issues, including small numbers of patients, unsuitable randomization, insufficient statistical powers, and unsuitable outcomes related to the clinical utility. Furthermore, due to the negative results in several randomized controlled trials (RCTs) that applied appropriate statistical powers and sample sizes, the use of dopamine is less commonly recommended today [188]. In addition, the renal vasodilation effect observed in healthy individuals has been found not to occur in acute kidney injury (AKI) patients [189].

However, there is only limited evidence of harm caused by the use of low-dose dopamine to prevent or treat AKI. Although a 2005 meta-analysis by Friedrich et al. [190] did not find low-dose dopamine to significantly increase the incidence of adverse effects, there is many literature related to the adverse effects of dopamine. The potential adverse effects of dopamine include tachycardia, myocardial ischemia, a reduced intestinal blood flow, hypopituitarism, and the inhibition of the T-cell function.

Friedrich et al. conducted a meta-analysis of studies in which low-dose dopamine was used to treat or prevent AKI [190]. Their analysis of 61 randomized and semi-randomized trials determined that low-dose dopamine did not prolong survival, reduce the rate of dialysis initiation, or improve the renal function; in addition, the urine output was found to be only improved on the day the dopamine treatment was initiated. Based on the absence of positive studies about the use of dopamine to prevent and treat AKI, and in consideration of the information about the previously-described adverse effects of dopamine, the 2012 KDIGO Clinical Practice Guideline for AKI recommends that low-dose dopamine should not be used to prevent or treat AKI (1A).

For the present CQ, we searched the literature to retrieve new evidence that has emerged since the publication of the KDIGO guideline. The literature review and assessment of the abstracts revealed five trials that potentially contained new evidence not included in the existing meta-analyses [191, 192, 193, 194, 195]. The subjects were heart failure patients in three trials, laparoscopic surgery patients in one trial, and severe obstructive jaundice patients in one trial. In the three trials involving heart failure patients, dopamine failed to improve the outcomes. The trials involving laparoscopic surgery patients and severe obstructive jaundice patients did not examine any clinically useful outcomes.

Based on the above-described KDIGO Guideline recommendation and on the fact that no subsequent trials have demonstrated low-dose dopamine to be effective in the prevention or treatment of AKI, we offer the same quality of evidence and strength of recommendation as the KDIGO guideline.

PubMed was searched for relevant studies published between December 24, 2009 and December 22, 2014, using “dopamine”, “AKI”, and “RCT” as search terms.

Search query: ((“dopamine”[MeSH Terms] OR “dopamine”[All Fields]) AND ((“kidney”[MeSH Terms] OR “kidney”[All Fields]) OR renal[All Fields]) AND low[All Fields]) AND (Randomized Controlled Trial[ptyp] AND “2009/12/24”[PDAT]:“2014/12/22”[PDAT] AND “humans”[MeSH Terms]).

Open image in new window

Since the publication of the KDIGO Guideline, there have been no RCTs regarding nutritional support with subjects limited to AKI patients. The KDIGO Guideline recommends a calorie intake of 20–30 kcal/kg/day for AKI patients of any stage. The desired amounts of protein administration are 0.8-1.0 g/kg/day in hypermetabolic AKI patients who do not require dialysis, and 1.7 g/kg/day in hypermetabolic patients undergoing CRRT; when possible, nutrition through the enteral route is preferred.

Acute kidney injury (AKI) is the form of organ failure to which severely ill patients are most susceptible. Thus, the metabolism is greatly affected by the primary disease, the malnutrition severity, the presence of comorbid organ failure, and the performance of renal replacement therapy (RRT). Therefore, the target levels of the calorie intake and the necessary protein should ideally be tailored to the individual pathology; however, the specific efficacy of nutritional support for AKI has not been demonstrated [196, 197].

Enteral nutrition is considered to be more effective than intravenous nutrition for intestinal mucosa maintenance, bacterial translocation, and the prevention of organ dysfunction. In meta-analyses of studies involving critically ill patients (including AKI patients), the initiation of enteral nutrition within 24 h after intensive care unit (ICU) admission was shown to significantly reduce the mortality and the incidence of infectious complications, and to shorten the hospital stay lengths; however, negative results have also been reported [198, 199, 200, 201]. In order to provide sufficient calorie, amino acids, and protein, a combination of intravenous and enteral nutrition is sometimes considered. A group of patients who only received vitamins and trace elements through enteral nutrition for the first 7 days following ICU admission and started intravenous nutrition on day 8 (late-initiation group) demonstrated a significant increase in early discharge (alive) from the ICU/hospital, as well as reductions in the incidence of infections, the number of patients on mechanical ventilation for > 2 days, the duration of RRT, and the health care costs [202].

The target calorie provision levels can be determined with a simple body mass conversion equation (25 kcal/kg/day), a calorie consumption prediction equation (Harris–Benedict equation), or the measurement of the energy consumption with an indirect calorimeter. In the first 7 days of sepsis treatment of patients who are not yet critically ill and are not malnourished, energy replenishment by enteral nutrition is recommended; however, supplemental intravenous nutrition to reach the target energy level is not recommended, as it can affect the prognosis adversely [203]. The preferable method to reach the target energy level is to start with a small amount of energy and to increase it gradually based on factors such as the presence of aspiration and/or regurgitation of gastric contents and diarrhea. In obese patients, it must be noted that the use of the actual body weight in the prediction formula will cause an overestimation of the target energy provision level. Based on reports that intensive insulin therapy is not useful for mortality reduction, the initiation of insulin control at a blood glucose level ≥ 180 mg/dl and the setting of a target blood glucose level of 144–180 mg/dl can be considered valid in severe AKI [3, 204, 205]. At the same time, AKI is reported to occur frequently in acute myocardial infarction patients with a blood glucose level ≥ 200 mg/dl at hospital admission [206, 207].

Protein restriction for the prevention or delay of RRT initiation is not recommended; however, it can be considered when an advanced electrolyte imbalance is present. In non-hypermetabolic AKI patients who do not require RRT, a protein provision of 0.8–1.0 g/kg/day is recommended. A protein provision of < 1 g/kg/day can induce negative nitrogen balance in patients undergoing RRT, due to factors such as the loss of approximately 10–15 g/day of amino acids, especially in continuous renal replacement therapy (CRRT). Therefore, in hypermetabolic patients undergoing CRRT, the KDIGO guideline recommends the administration of 1.7 g/kg/day of protein to account for the amount of protein loss, while a protein intake of 2.5 g/kg/day is reportedly needed to achieve a positive nitrogen balance [208, 209]. However, the excess provision of amino acids is indicated to potentially cause azotemia and prolonged RRT [210]. During CRRT, commercially available dialysates and replacement fluids in Japan can cause hypokalemia and hypophosphatemia, and the latter is reported to delay weaning from mechanical ventilation; therefore, appropriate supplementation of potassium and/or phosphate through intravenous or enteral nutrition can be beneficial [196, 211, 212, 213]. Outside of Japan, a CRRT dialysate containing 4.0 mEq/L potassium and 3.7 mg/dl phosphorus has been developed [214]. However, a switch from CRRT to intermittent renal replacement therapy (IRRT) can easily cause an electrolyte imbalance, thereby calling for a reexamination of the content of the intravenous nutrition or enteral nutrition, including the total fluid volume; in particular, the risk of hyperkalemia must be kept in mind.

There is no clear evidence to recommend nutritional support for mild AKI without fluid overload, dehydration, or an electrolyte imbalance. The International Nutrition Survey conducted a recent international cross-sectional study on nutritional intervention, in which nine Japanese facilities participated. In this study, the calorie sufficiency rate, protein sufficiency rate, nutrition provision rate, and in nearly all other parameters in Japanese ICU patients were found to be below the global mean; in addition, the initiation of enteral nutrition was demonstrated to be late. Further research is needed.

PubMed was searched for relevant studies published between January 2012 and April 2016, and papers related to the present CQ were abstracted from the search results. The literature published before January 2012 was referenced from the KDIGO Clinical Practice Guideline for Acute Kidney Injury.

Open image in new window

Out of nine relevant RCTs, three were performed at a single center (two involving patients having undergone cardiac surgery, one involving ICU patients); in all three RCTs, the early initiation of blood purification was associated with a reduced mortality. However, in a meta-analysis that included multicenter RCTs, the efficacy of the early initiation of blood purification was not supported.

There has been a consensus that emergent blood purification should be initiated for serious, life-threatening disease states. The KDIGO Clinical Practice Guideline for Acute Kidney Injury (AKI) also states that renal replacement therapy (RRT) should be initiated immediately in the event of potentially fatal changes in the body fluid, electrolyte, and acid-base balance (Not Graded). The indications for emergent RRT in clinical settings are listed in the Table 11.

In observational studies from the 1960s that examined the timing of blood purification initiation, hemodialysis was demonstrated to improve the survival rates not when initiated after AKI had progressed to the point at which symptoms of uremia were present, but when initiated before that point. Since 2000, multiple observational studies have examined the initiation of blood purification when the blood urea nitrogen (BUN) is at a level even lower than that set in the aforementioned observational study. Bagshaw et al. conducted a meta-analysis of 15 clinical studies [including two randomized controlled trials (RCTs), four prospective observational studies, and nine retrospective observational studies] [215]. Although the early initiation of blood purification was found to be associated with favorable outcomes, the timing of the initiation across the different studies examined was significantly heterogeneous; therefore, the early initiation of blood purification could not be definitively recommended.

The effect of the early initiation of blood purification on death has been examined in nine RCTs to date—including some that were not covered in Bagshaw et al.’s meta-analysis. Bouman et al. randomly assigned intensive care unit (ICU) patients presenting with oliguria to early high-volume, early low-volume, and late low-volume hemofiltration, with the survival at day 28 and the recovery of the renal function as primary endpoints; however, the three groups did not demonstrate any differences in their survival at day 28 or their recovery of the renal function [216]. An Indian RCT involving patients who developed AKI in the hospital divided the subjects into two groups: a group in which dialysis was initiated early, when the BUN was ≥ 70 mg/dl or the serum creatinine (sCr) was ≥ 7 mg/dl (n = 102), and a control group in which dialysis was initiated upon fluid overload, hyperkalemia, or any other indication for emergent dialysis (ultimately, BUN: 100.9 ± 32.6 mg/dl, sCr: 10.41 ± 3.3 mg/dl; n = 106). The comparisons of these two groups revealed no significant differences in the mortality or the recovery of the renal function [217]. In a Canadian open-label pilot trial [218] reported in 2015, patients with volume-replete AKI were randomly assigned to an early treatment group (n = 48) or a standard treatment group (n = 52); the mortality and the recovery of the renal function did not show any significant differences between the two groups. In two single-center RCTs involving post-cardiac surgery patients [219, 220], the early initiation of blood purification was associated with a reduced mortality. In a multicenter RCT (the HEROICS study) [221], patients experiencing shock requiring catecholamine support following cardiac surgery were randomly assigned to one of two groups: an early hemofiltration group (80 ml/kg/hr for 48 h) or a standard therapy group that included continuous hemodiafiltration (CHDF) if necessary (n = 112 for both groups). The two groups did not demonstrate a significant difference in their mortality or recovery of the renal function.

Two additional RCTs involving AKI patients in the ICU were reported in May 2016. In a French multicenter RCT (the AKIKI trial) [222], critically ill patients with severe AKI (stage 3) who required mechanical ventilation or catecholamine infusion were randomly assigned to one of two groups: an early initiation group (n = 311), in which RRT was initiated immediately after randomization, or a delayed initiation group (n = 308), in which RRT was not initiated until a criterion such as hyperkalemia, metabolic acidosis, pulmonary edema, a high BUN level (> 112 mg/dl), or oliguria (> 72 h) was met. The investigation of the utility of early RRT did not reveal a significant difference in the 60-day mortality of the two groups. In a German single-center RCT (the ELAIN trial) [223], 231 critically ill patients with stage 2 AKI and a plasma NGAL > 150 ng/ml were randomly assigned to an early group (initiation of RRT immediately after randomization) or a delayed group (initiation of RRT upon progression of AKI to stage 3 or the presence of any absolute indications); the assessment of the 90-day mortality found the latter to be significantly reduced in the early group. However, in the AKIKI trial, the RRT initiation in the early group was late, upon validation of stage 3 AKI; this timing corresponded to the delayed group in the ELAIN trial. Moreover, in the ELAIN trial, RRT was initiated as continuous renal replacement therapy (CRRT) and was performed for at least one week in all patients; however, in the AKIKI trial, CRRT was performed as the sole method of RRT in only 30% of patients. Although both of these RCTs examined the early initiation of RRT, it must be noted that they differed in their timing of initiation and their treatment modalities.

Among the nine RCTs reported to date, those that recorded the 28- or 30-day mortality were subjected to a meta-analysis; this meta-analysis did not support the efficacy of early initiation (Fig. 5). Moreover, a meta-analysis that included the AKIKI [222] and ELAIN [223] trials was published [224]. In this analysis of six RCTs, including the AKIKI and ELAIN trials as well as the three previously-mentioned RCTs involving non-post-cardiac surgery patients [216, 217, 218], the early initiation of blood purification did not show evident efficacy in terms of either mortality (relative risk 0.93, 95% CI 0.68–1.26) or recovery of the renal function (relative risk 0.88, 95% CI 0.48–1.62).

Open image in new window

As of August 2016, the multicenter STARRT RCT [225] is ongoing. In France, a multicenter RCT assessing the utility of early initiation of RRT for septic AKI (corresponding to KDIGO stage 3) is ongoing (the IDEAL-ICU study [226]). Both of these RCTs are larger in scale than previous investigations; therefore, the results obtained may lead to new directions about early initiation.

PubMed was searched for relevant studies published up to August 2015, and papers related to the present CQ were identified from the search results. Important manuscripts published after the search period ([222, 223, 224]) were also incorporated into our recommendation.

Open image in new window

There have been no RCTs relevant to the discontinuation of blood purification. In three observational studies, the urine output and SOFA score were reported to be predictors of the possibility to wean patients from blood purification.

The blood purification for acute kidney injury (AKI) is stopped when the renal function sufficiently recovers. However, very few studies have examined the criteria for discontinuation of the blood purification for AKI. Our literature search found three observational studies that identified predictors of the possibility to wean patients from blood purification to date. Wu et al. retrospectively examined AKI patients (n = 304) who required renal replacement therapy (RRT) in the intensive care unit (ICU) after surgery [227]. Of the 94 patients in whom RRT was discontinued, 30 patients needed to resume it within 30 days; the duration of the RRT [odds ratio (OR) 1.06, 95% confidence interval (CI) 1.02–1.10], the SOFA score at cessation (OR 1.44, 95% CI 1.13–1.83), oliguria (< 100 ml over 8 h) (OR 4.17, 95% CI 1.07–16.13), and an age over 65 years (OR 6.35, 95% CI 1.61–24.99) were identified as predictors of discontinuation failure. Kawarazaki et al. retrospectively examined AKI patients (n = 343) in Japanese ICUs who required continuous renal replacement therapy (CRRT) [228]. The comparison of an early recovery group (those who could discontinue CRRT within 48 h of initiation; n = 52) and a control group that excepted patients who died early (n = 239) revealed that the urine output upon CRRT initiation (ml/h) (OR 1.02, 95% CI 1.01–1.03), the SOFA score upon CRRT initiation (OR 0.87, 95% CI 0.78–0.96), and the time from ICU admission to CRRT initiation (in days) (OR 0.65, 95% CI 0.43–0.87) were significantly associated with early weaning from CRRT. It is of note that the urine output and SOFA score data were collected upon CRRT initiation and some patients who ceased the CRRT within 48 h may not have required blood purification; therefore, these results should be interpreted with caution.

The urine output is perhaps the most clinically emphasized predictor; one useful reference is a sub-analysis of the BEST study (n = 1,006), in which Uchino et al. examined AKI patients in ICUs in 23 countries [229]. Those who did not require RRT for at least 7 days after the initial discontinuation were defined as the success group (n = 313), while those who had to resume RRT within 7 days after the initial discontinuation were defined as the repeat-RRT group (n = 216). Comparisons of the two groups revealed that the urine output was the most useful predictor of RRT weaning; the cutoff values for the use and the non-use of diuretics were 2330 ml/day (approximately 100 ml/h) and 436 ml/day (approximately 20 ml/h), respectively.

In the previously-cited sub-analysis of the BEST study, the serum creatinine (sCr) was also reported to be a significant predictor of weaning (OR 0.996, 95% CI 0.994–0.998). Creatinine is produced from creatine in the muscle tissue and is released into the blood. Both blood purification and the kidneys of the patients remove creatinine. The balance of creatinine between muscle production and elimination by blood purification and kidneys defines the sCr. Therefore, if the sCr level remains constant for at least 2–3 days, the production and elimination can be considered equal. The phenomenon by which the amount of sCr remains constant for several days before suddenly decreasing sharply without changing blood purification doses—called “spontaneous fall”—indicates that the renal function has recovered. In the VA/NIH ATN study [230] that examined the association between the dialysis doses and the AKI outcomes, the recovery of the renal function was defined as a urine output > 30 ml/h in 6 h of collection or a spontaneous fall in the sCr. The following protocol was adopted: if the creatinine clearance in the 6-h urine collection was > 20 ml/min, the CRRT was discontinued; if the creatinine clearance was < 12 ml/min, the CRRT was continued; if the creatinine clearance was between 12 and 20 ml/min, the decision to continue or discontinue the CRRT was left to the clinician.

However, in AKI and advanced chronic kidney disease (CKD), creatinine not only undergoes glomerular filtration, but is also excreted into urine due to re-secretion from the renal tubule; consequently, the creatinine clearance is greater than the actual glomerular filtration rate (GFR). In addition, if the sCr continues to decrease in 6-h urine collection, the sCr value selected for use in the GFR calculation may result in an overestimation or underestimation of the GFR. At the AKI recovery stage, in which the renal function fluctuates dynamically, the sCr and creatinine clearance are markedly unreliable; however, due to the absence of other appropriate endpoints, the sCr may be an acceptable basis upon which to determine whether to discontinue the blood purification, with the above-described background considered.

PubMed was searched for relevant studies published up to August 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Increasing the dose of blood purification for AKI to higher than the level recommended as the international standard (20–25 ml/kg/h) has not been reported to improve the AKI outcomes. No RCTs have compared the blood purification dose covered by health insurance in Japan (10–15 ml/kg/h) to that recommended internationally; only two observational studies have evaluated this issue, and neither observed a significant difference in mortality. Thus, there is no definitive evidence to support the need to change the dose used in Japan to that recommended as the international standard.

The appropriate dose of renal replacement therapy (RRT) for acute kidney injury (AKI) has been investigated so far. Several studies have reported that increased doses do not lead to improved outcomes [216, 230, 231, 232]; in addition, there is insufficient evidence to determine an optimal dose.

The dose of continuous renal replacement therapy (CRRT) for AKI was firstly examined by a 2000 study by Ronco et al. [233]. A total of 425 AKI patients who required continuous hemofiltration (CHF) were randomly assigned a filtration flow rate (QF) of 20, 35, or 45 ml/kg/h. The comparisons of the three groups revealed respective survival rates of 41, 57, and 58%; the 20 ml/kg/h group demonstrated a significantly lower survival rate than the other two groups, while there was no significant difference between the 35 ml/kg/h group and the 45 ml/kg/h group. Since then, two multicenter, large-scale randomized controlled trials (RCTs) have been reported (in 2008 and 2009, respectively) [230, 232]; unlike in the trial reported by Ronco et al., these two RCTs found that increased doses of RRT for AKI did not improve the outcomes. In the ATN study [230], 1124 AKI patients who required RRT were randomly assigned to a standard therapy group or an intensive therapy group, and the two groups’ mortality and recovery of the renal function were compared. In the standard therapy group, hemodynamically stable patients underwent hemodialysis (HD) three times a week, while hemodynamically unstable patients either underwent continuous hemodiafiltration (CHDF) at a rate of 25 ml/kg/h or received sustained low-efficiency dialysis (SLED) three times a week. In the intensive therapy group, hemodynamically stable patients underwent HD six times a week, while hemodynamically unstable patients either underwent CHDF at a rate of 35 ml/kg/h or received SLED six times a week. The comparisons revealed no significant differences in the mortality or recovery of the renal function of the two groups. In the RENAL study [232], 1508 AKI patients were randomly assigned to an intensive therapy group (35 ml/kg/h CHDF) or a standard therapy group (25 ml/kg/h CHDF), and the two groups’ mortality and recovery of the renal function were compared. Likewise, the two groups in this study did not demonstrate any significant differences in their mortality or recovery of the renal function. Based on the results of two multicenter, large scale RCTs, the recent KDIGO Clinical Practice Guideline for AKI [3] recommends a CRRT dose of 20–25 ml/kg/h. However, both the ATN and RENAL studies examined AKI collectively with a wide variety of causes, including ischemia, nephrotoxic substances, and sepsis; to date, there have been very few studies of the optimal doses in function of the underlying disease. Regarding septic AKI, four RCTs [234, 235, 236, 237] have compared the CRRT outcomes from doses of 35–45 ml/kg/h and larger doses (65–100 ml/kg/h); in these RCTs, increased doses were not found to improve the outcomes. Based on the above, there is currently no definitive evidence allowing for the recommendation of optimal doses according to the underlying disease. However, in the event of acute hyperkalemia—such as in tumor lysis syndrome—the blood purification doses must be temporarily increased or otherwise tailored to the individual pathologies.

The common dose of CRRT for AKI in Japan (10–15 ml/kg/h) is generally smaller than the recommended dose outside of Japan (20–25 ml/kg/h). This seems to be because Japanese health insurance only covers a dialysis dose of approximately 15 L/day. No RCTs have compared the standard Japanese dialysis dose of 10–15 ml/kg/h with the recommended international dose of 20–25 ml/kg/h; however, two retrospective observational studies [238, 239] have concluded that the standard Japanese dose did not lead to worse outcomes. As mentioned above, although Ronco et al. reported an optimal blood purification dose of 35 ml/kg/h, this was controverted by two subsequent large-scale RCTs. However, there is currently insufficient evidence to determine that 20–25 ml/kg/h is an optimal dose, and the standard Japanese dose must be examined as well. Moreover, while it is unknown whether the reduction of doses to below the standard Japanese dose of 10–15 ml/kg/h would worsen the outcomes, we would like to address that there is no evidence to recommend the reduction of dialysis doses.

The appropriate doses of HD for AKI have been examined in three RCTs [216, 230, 240]. In those, the different groups demonstrated no significant differences in their mortality or recovery of the renal function. For intermittent renal replacement therapy (IRRT) or extended RRT, the KDIGO Clinical Practice Guideline for AKI recommends a weekly standardized dialysis dose (Kt/V) of 3.9. However, there is insufficient evidence to establish this as the optimal hemofiltration dose for AKI. An RCT that compared HD and predilution on-line HF rather than HD doses was reported in 2012 [241]. The mean volume of infusate in predilution on-line HF was 81 L; the HF and HD groups did not demonstrate significant differences in their mortality or recovery of the renal function. In a prospective study by Schiffl et al. that compared a daily HD group with an alternate-day HD group [240], the daily HD group demonstrated a significantly lower mortality and a significantly earlier recovery of the renal function. However, several issues have been raised with this study: the HD doses were extremely low, and the randomization was inadequate. The Hannover Dialysis Outcomes Study randomly assigned 156 AKI patients to a standard dialysis group and an intensified dialysis group, and compared the mortality and the recovery of the renal function in the two groups [242]. The standard dialysis was dosed to maintain a blood urea nitrogen (BUN) level of 120–150 mg/dl, while the intensified dialysis was dosed to achieve a target BUN level of < 90 mg/dl. However, the two groups did not demonstrate significant differences in their mortality or recovery of the renal function.

PubMed was searched for relevant studies published up to July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Several RCTs and meta-analyses have compared CRRT with IRRT, but none have demonstrated differences in mortality. It must be noted that some of these RCTs excluded hemodynamically unstable patients, and that none limited their subjects to hemodynamically unstable patients. In meta-analyses of sustained low-efficiency dialysis (SLED), which combines the advantages of both CRRT and IRRT, comparisons with CRRT revealed no significant difference in the mortality.

The optimal modality of blood purification for acute kidney injury (AKI) has been investigated. The primary continuous and intermittent modalities used in Japan are continuous hemodiafiltration (CHDF) and hemodialysis (HD), respectively [243]. The choice of therapy is generally based on a consideration of various factors, including the patient’s hemodynamics and anticoagulation, the facility’s equipment, and the staff’s experience and manpower. The advantages and disadvantages of these modalities are summarized in the Table 12. The greatest advantage of continuous renal replacement therapy (CRRT), which is mostly conducted with CHDF in Japan, is that its effects on the hemodynamics are minimized as it removes the body fluids and solutes gradually. CRRT is also associated with a reduced risk of cerebral edema [244]. However, continuous blood purification not only restrains the patient over a long period of time, but also places a great burden on the medical care staff. In addition, the continuous administration of anticoagulants increases the risk of hemorrhage. Intermittent renal replacement therapy (IRRT), which is mostly conducted with HD in Japan, removes the body fluids and solutes quickly, thereby easily affecting the hemodynamics and increasing the risk of cerebral edema. However, in addition to being completed faster, IRRT places a lesser burden on the staff and poses a lower risk of hemorrhage than CRRT. As CRRT and IRRT possess evidently different characteristics, direct comparisons of their utility have been found to be worthless [245].

For the present CQ, we abstracted a total of 15 randomized controlled trials (RCTs) [246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260] and eight meta-analyses [261, 262, 263, 264, 265, 266, 267, 268] that compared the utility of the two modalities of renal replacement therapy (RRT) for AKI. A primary meta-analysis was reported as part of a Cochrane joint project in 2008 [267]. An analysis of 15 RCTs involving a total of 1550 AKI patients who required RRT revealed no significant differences between CRRT and IRRT in the in-hospital mortality, the ICU mortality, or the discontinuation of RRT in surviving patients. Several other meta-analyses have demonstrated similar results [262, 263, 264, 265]. However, in a meta-analysis of 13 studies (including three RCTs) by Kellum et al., while CRRT and IRRT yielded no significant difference in mortality, adjustments for the severity of the illness and the study quality revealed that the risk of death was significantly reduced in CRRT [relative risk (RR) 0.72, 95% confidence interval (CI) 0.60–0.87] [261]. In Schneider et al.’s meta-analysis of 23 studies (including 7 RCTs) related to the rates of dialysis dependence, although the risk of dialysis dependence was significantly higher in patients undergoing IRRT (RR 1.73, 95% CI 1.35–2.20), there was no significant difference when the analysis was limited to RCTs [266]. The KDIGO Guideline and the Surviving Sepsis Campaign Guideline (SSCG) 2012 [269] considered the same matter as the present CQ and gave similar recommendations based on the results described above.

It must be noted that two RCTs [247, 256] excluded hemodynamically unstable patients. Theoretically and empirically, CRRT has been considered useful, and has therefore been used, for the treatment of hemodynamically unstable patients. No RCT has ever compared CRRT and IRRT in hemodynamically unstable patients. Based on the above, our expert opinion is that CRRT is preferable for hemodynamically unstable patients. However, there is no standard opinion on the degree of instability at which CRRT should be chosen. In addition, many hemodynamically unstable patients are of course critically ill and may have comorbid coagulation disorders, such as disseminated intravascular coagulation (DIC). HD may be more useful in patients with an advanced bleeding tendency, as it can be performed with only a short course of anticoagulant administration. In addition, the selection of a modality must take into account the equipment at the facility, the staff’s experience and resources, and a variety of other factors unrelated to the patient. The modality should be decided by a physician with sufficient knowledge and experience of blood purification (i.e. an intensive care specialist, nephrologist, or dialysis physician) according to the patient’s disease condition.

Sustained low-efficiency dialysis (SLED), or extended daily dialysis (EDD), uses the lower blood flow and the lower dialysate flow, and is performed more frequently than standard HD; it incorporates the advantages of both CRRT and IRRT, and is now performed widely. Although no RCTs have compared SLED and intermittent hemodialysis (IHD), the effects of SLED on the hemodynamics are reported to be the same as those of CRRT [270]. Recently, Zhang et al. conducted a meta-analysis of 17 studies (including 7 RCTs) that compared EDD and CRRT for AKI [268]. In the RCTs, no significant difference was observed in the mortality of the different groups; however, in the 10 observational studies, the risk of death was lower with EDD (RR 0.86, 95% CI 0.74–1.00). No significant difference in the recovery of the renal function was evidenced by the RCTs or the observational studies. By establishing appropriate implementation conditions, not only CRRT but also EDD may be performed in hemodynamically unstable patients.

PubMed was searched for relevant studies published up to July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

There have only been two RCTs involving the use of nafamostat mesilate as an anticoagulant during blood purification for AKI (nafamostat mesilate versus no anticoagulant). No significant difference was observed in the survival outcomes. Two observational studies that compared the use of nafamostat mesilate and heparin also found no significant difference in the survival outcomes.

Blood coagulates when it comes into contact with anything including artificial materials other than vascular endothelial cells. Therefore, renal replacement therapy (RRT)—which involves extracorporeal circulation—usually requires the use of anticoagulants. Patients with severe acute kidney injury (AKI) requiring RRT frequently present with a hemorrhagic complication. Thus, the use of anticoagulants that pose the lowest possible risk of bleeding is required. Japanese health insurance currently covers four types of anticoagulants for RRT: unfractionated heparin, low molecular weight heparin (LMWH), nafamostat mesilate (NM), and argatroban. Citrate is widely used as an anticoagulant outside of Japan. Although it can be used in Japan as well, this is not frequent, since its use as an anticoagulant for RRT is considered off-label. In the BEST kidney study [34], which included an examination of the anticoagulants used in continuous renal replacement therapy (CRRT) for AKI, the most commonly used anticoagulant was unfractionated heparin (42.9% of patients), followed by no anticoagulant (33.1%), citrate (9.9%), NM (6.1%), and LWMH (4.4%).

When possible, CRRT without anticoagulation is the safest option for AKI patients, as it does not increase the risk of bleeding; however, the lifespans of the filter and circuit may be shortened, and CRRT cannot be performed without anticoagulation in all patients. In Japan, NM is widely used as an anticoagulant in CRRT for AKI since it has a short half-life and is associated with a relatively lower risk of bleeding than other anticoagulants. However, it has not been approved only in the limited countries. In addition, NM causes adverse effects, including agranulocytosis, hyperkalemia, and anaphylactoid reactions [271, 272, 273, 274]. In Japan, unfractionated heparin is the most commonly used anticoagulant for hemodialysis (HD) in chronic dialysis patients. However, due to concerns over the risk of bleeding, it is seldom used as an anticoagulant in CRRT for AKI in Japan. Similarly, although LMWH is also associated with a lower risk of bleeding than unfractionated heparin, the test that indicates anticoagulant action (the anti-Xa assay) is not common. Therefore, LMWH is infrequently used as an anticoagulant in CRRT for AKI [34].

Five studies have examined the use of NM as an anticoagulant in CRRT for AKI [34, 275, 276, 277, 278]. Only two of these studies were randomized controlled trials (RCTs) [275, 276], while one was a prospective observational study [34] and two were retrospective studies [277, 278]. Both of the RCTs compared NM with no coagulation, and neither observed a significant difference in the survival outcomes of the groups. In one RCT, the lifetime of the hemofilter was found to be significantly longer with NM than without coagulation. The two groups demonstrated no significant difference in their hemorrhagic complications. A prospective observational study the BEST kidney study and a retrospective study by Hwang et al. [278] also observed no significant differences in the survival outcomes. In a retrospective observational study in which continuous hemodiafiltration (CHDF) was performed without anticoagulation in patients at a high risk of bleeding, Baek et al. reported that only NM was used when the hemofilter lifespan was less than 12 h [277]. The in-hospital mortality was significantly lower in the NM group (anticoagulation-free group versus NM group: 64.6 vs 41.9%, p = 0.003), while there was no significant difference between the groups in the transfusion volume (anticoagulation-free group versus NM group: 0.7 units/day versus 0.7 units/day). In addition to the five studies cited above, another study in which NM was compared with heparin was recently published in Japan [279]. Despite being a retrospective observational study, its analysis featured propensity score-matched cohorts. Although the mortality was not examined, hemorrhagic complications were significantly less frequent in the NM group, while there was no significant difference in the filter lifespan.

Citrate is commonly used as an anticoagulant in CRRT for AKI outside of Japan. Although the safety and efficacy of citrate have been assessed, its use as an anticoagulant in CRRT for AKI is considered off-label in Japan. Although no RCTs have compared citrate with NM, 10 RCTs have compared citrate and unfractionated heparin [280, 281, 282, 283, 284, 285, 286, 287, 288, 289]; in the six RCTs that examined the survival outcomes [281, 284, 285, 286, 288, 289], no significant difference was observed. Eight studies found the filter lifespan to be significantly longer with citrate [280, 281, 284, 285, 286, 287, 288, 289], while two observed no significant difference [282, 283]. The frequency of hemorrhagic complications with citrate and unfractionated heparin was found to be either equal, or significantly lower with citrate. Five more RCTs examined the use of LMWH as an anticoagulant in CRRT [290, 291, 292, 293, 294]; three of them [290, 291, 293] compared LMWH with unfractionated heparin, while two [292, 294] compared LMWH with citrate. Only one of the five RCTs [292] examined the clinical outcomes; in this RCT, the mortality was significantly lower in the citrate group. With regard to bleeding, only one RCT found a significant difference [293]; however, the results related to the filter lifespan varied across studies.

PubMed was searched for relevant studies published up to December 2015, and papers related to the present CQ were identified from the search results. The study by Makino et al. [279], which was published after the search period, was found through a hand search.

Open image in new window

The majority of blood purification filters currently used in Japan are biocompatible high-flux membranes. However, no studies have found the differences in these membranes to affect the AKI outcomes or the recovery of the renal function. For AKI—and particularly septic AKI—blood purification is performed in Japan to improve hypercytokinemia by using the principle of adsorption; however, there is no high-level evidence for the effect of this blood purification method on the outcomes.

The blood purification membrane materials currently used in Japan include cellulose triacetate (CTA), polymethyl methacrylate (PMMA), ethylene vinyl alcohol (EVAL), polysulfone (PS), polyethersulfone (PES), polyarylethersulfone (PAES), and polyester polymer alloy (PEPA). These materials will activate complement system less than cuprophan and other regenerated celluloses that have been used since the 1960s, and are considered to be highly biocompatible membranes. Many of these membranes currently in use have been developed as high-flux (HF) membranes, with the goal to remove the β2 microglobulin (β2MG) and other small-molecule proteins.

The therapeutic effects of each membrane material have been compared in a few small-scale randomized clinical trials (RCTs) that primarily featured comparisons of regenerated cellulose with synthetic polymeric membranes. These RCTs have also featured comparisons of HF membranes with the low-flux (LF) membranes that were developed prior to the existence of HF membranes. With regard to the present CQ, we found seven relevant RCTs that compared different types of membranes [295, 296, 297, 298, 299, 300, 301].

Five of these RCTs primarily compared the effects of the differences in the membranes’ biocompatibility, along with the effects of the differences between LF and HF membranes. Schiff et al. [295] examined the recycled cellulose Cuprophan^® (LF membrane) and the synthetic polymer polyacrylonitrile (PAN; HF membrane) in 2 groups of 26 postoperative acute kidney injury (AKI) patients each, and compared the results of the two groups. In later trials, Jörres et al. [296] compared 76 AKI patients in whom Cuprophan^® (LF membrane) was used with 84 patients in whom PMMA (LF membrane) was used. Meanwhile, Gastaldello et al. [297] and Albright et al. [298] compared cellulose acetate (LF membrane) with PS (HF membrane)—both of which are improved versions of Cuprophan^®—in AKI patients. However, none of the above three trials observed differences in the outcomes or in the recovery of the renal function. In 2008, Cochrane reported a meta-analysis of 1100 patients from the five RCTs cited above and from five others (10 RCTs in all); the comparisons of the biocompatible membranes (synthetic polymeric membranes, n = 575) with the bioincompatible membranes (regenerated cellulose membranes, n = 525) revealed no significant differences in the mortality (relative risk 0.93, 95% confidence interval 0.81–1.07) or in the recovery of the renal function (n = 1,038, relative risk 1.09, 95% confidence interval 0.90–1.31) [302]. Although these results are not directly relevant to comparisons of the synthetic polymeric membranes mainly used today, it should be acknowledged that no significant differences have been found between synthetic polymeric membranes and so-called bioincompatible regenerated cellulose membranes.

Jones et al. [299] compared the survival rates for the use of the two synthetic polymeric membranes PAN (n = 97) and PS (n = 100) (both HF membranes) in the continuous hemodialysis (CHD) of ventilated patients with AKI. No significant difference was observed (PAN: 29%, PS: 27%). In the 2000s, another trial compared HF and LF membranes made of the same material. Ponkivar et al. [300] compared HF membranes (n = 34) and LF membranes (n = 38) both made of PS in AKI patients; the two groups’ results were similar. Based on the above, none of the blood purification membranes currently used in Japan can produce better AKI treatment outcomes.

The core of the pathophysiology of septic AKI is assumed hypercytokinemia. The improvement of hypercytokinemia may be useful for improvement of the performance status and of AKI. This has led to attempts at blood purification designed to remove all types of cytokines. Haase et al. [301] conducted a crossover RCT in which 10 sepsis patients with AKI classified by the RIFLE criteria as Failure underwent hemodialysis (HF) using a standard HF membrane (with in vivo molecular wright cutoff values of 15–20 kD) and a membrane with a larger pore size (50–60 kD). Comparisons of the cytokine removal efficiency revealed that the use of a membrane with a larger pore size significantly reduced the blood concentrations of the cytokines IL-6, -8, and -10 at 4 h of HD. In Japan, PMMA membranes [303] and AN69ST membranes [304], which are based on the principle of adsorption and are considered to highly efficient at removing cytokines, have been used in attempts at blood purification. Since 2014, AN69ST membranes have been covered by health insurance in Japan for patients with severe sepsis and septic shock. However, there is no high-level evidence of the clinical efficacy of membranes with large pore sizes or of adsorption membranes. Therefore, in regard to blood purification for the treatment of sepsis, the KDIGO Guideline states, “Until further evidence becomes available, the use of RRT to treat sepsis should be considered experimental”. The collection of further evidence is anticipated.

PubMed was searched for relevant studies published up to July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

At present, there have been no RCTs to examine the long-term outcomes of AKI (≥ 12 months following onset). However, there have been systematic reviews and meta-analyses of observational studies related to the survival outcomes, cerebrovascular and cardiovascular, and renal outcomes; the most reliable and most recent one is a study by Sawhney et al. Based on the search query used in that systematic review, we further searched the literature for studies related to the survival outcomes, the cardiovascular and cerebral disease outcomes, and the renal outcomes with an extended search period. We then conducted a meta-analysis of the search results, along with a consideration of any new studies related to the survival outcomes and renal outcomes, and of observational studies related to the cerebral and cardiovascular disease outcomes. In the results of our meta-analysis, the long-term outcomes of AKI patients were consistently poor. Moreover, although there have been no meta-analyses of the long-term QOL, some observational studies report that the onset of AKI is associated with a reduced long-term QOL.

The term “acute renal failure” (ARF) was used for the first time in writing by Heberden et al. in 1802 [305]. Although ARF was once considered to be reversible and therefore to have a favorable outcome, Hishida et al. reported that ARF patients have extremely poor survival outcomes, and cited multiple organ failure as a crucial underlying factor [306]. There were already multiple definitions of ARF [4]. In order to avoid confusion over the definition of ARF and to define acute syndromes related to the renal function more broadly, the term “acute kidney injury” (AKI) was suggested globally.

As the concept of AKI spread worldwide, many clinical studies on AKI have been performed. These studies have shown that the survival outcomes of AKI is poor [307, 308, 309, 310], that its long-term outcomes is also poor [311], and that the stage of AKI in the intensive care unit (ICU) is correlated with the mortality [312]; as a result, perspectives on the outcomes of AKI have been changing. In 2015, Sawhney et al. reported the results of a systematic review that assembled the results of individual studies [313]; the survival outcomes and renal outcomes 1 year after the AKI onset were both shown to be poor. However, other clinically important outcomes such as the cerebral outcomes, the cardiovascular outcomes, and the quality of life (QOL) have not been examined.

Based on the search query used by Sawhney et al. in their systematic review [313], we developed a search query to encompass four types of outcomes: survival outcomes, cerebral and cardiovascular outcomes, renal outcomes, and the QOL. PubMed was searched for relevant studies published between January 1, 2005 and April 30, 2015. In regard to the long-term survival outcomes and the long-term renal outcomes, we abstracted results from beyond the subject period used in existing systematic reviews. With regard to the titles and abstracts, we conducted a preliminary review and selected potentially relevant manuscripts; we then conducted a secondary review of these manuscripts (full-text assessments) to identify our final target manuscripts.

Open image in new window

Age ≥ 3 months: Two single-center retrospective observational studies have assessed the KDIGO diagnostic criteria in sufficiently large cohorts; these studies consistently demonstrated that the KDIGO diagnostic criteria were useful for the prediction of the mortality and of other outcomes.

Age < 3 months: Two review papers have examined the diagnosis of AKI in neonates and have commented on the results obtained in a total of 11 observational studies. The modified neonatal KDIGO criteria was suggested with the data of associations with the AKI onset, mortality, and neurological outcomes.

The early diagnosis and treatment of acute kidney injury (AKI) is crucial to the improvement of the outcomes not only in adults, but also in children. Several diagnostic criteria have been suggested for children. These have included the Pediatric RIFLE (pRIFLE) (Table 13), AKIN (Table 2), and KDIGO (Table 3) criteria. It is known that the normal serum creatinine (sCr) values change with the age (Table 14) [319]. As urine collection is difficult in children, the pRIFLE classification uses a Schwartz formula-based [320] calculation of the estimated glomerular filtration rate (eGFR) [321]. Due to differences between the Japanese and Western body constitutions and renal functions, the assessment of the GFR based on the Schwartz formula is considered unsuitable for Japanese children [322]; therefore, another equation has been proposed to estimate the GFR in Japanese children [323]. The AKIN and KDIGO classifications of AKI are based on the sCr and on the duration of oliguria/anuria (i.e. a urine output < 0.5 mLlkg/h). Hereafter, we will describe studies that have compared these multiple diagnostic criteria for pediatric AKI.

Sutherland et al. compared the pRIFLE, AKIN, and KDIGO diagnostic criteria in 14,795 children aged under 18 who were hospitalized for AKI [324]. The AKIN and KDIGO classifications, which both use the sCr criteria, were almost completely in agreement; however, as the eGFR-based pRIFLE classification has a higher incidence for stage 1 than the AKIN or KDIGO classifications; a larger number of patients were diagnosed with mild AKI. In all three classifications, the mortality was higher in patients with AKI than in those without AKI; particularly in the intensive care unit (ICU), the increasing severity of AKI (according to all three classifications) was associated with increased mortality. Selewski et al. used the KDIGO classification to examine the AKI outcomes in a cohort of 2415 patients in pediatric ICUs [325]. In comparison with patients who did not develop AKI, pediatric AKI patients demonstrated a significantly increased length of mechanical ventilation, a longer ICU stay, a longer duration of hospitalization, and a higher mortality rate. In addition, the length of the ICU stay was proportional to the worsening of the KDIGO AKI stages. These two single-center retrospective observational studies involved sufficient numbers of patients to demonstrate that the KDIGO classification is useful for the diagnosis of pediatric AKI. Moreover, as the KDIGO classification does not involve an estimation of the GFR but instead allows to stage AKI based on the sCr, it can be considered superior to the pRIFLE classification. Therefore, we suggest the use of the KDIGO diagnostic criteria for pediatric AKI patients aged ≥ 3 months. However, it must be noted that the use of the AKI diagnostic criteria has not yet been specifically evaluated in Japanese children (Table 15).

Neonates and children aged < 3 months possess a unique background that includes immaturity and perinatal factors; therefore, children under 3 months must be considered separately from those aged ≥ 3 months. Although there have been investigations of the AKI diagnosis, treatment, and outcomes in neonates, there used to be no definitive diagnostic criteria for neonatal AKI [326, 327]. As the use of adult diagnostic tools such as the RIFLE, AKIN, and KDIGO criteria spread, their use for the diagnosis of AKI in neonates came to be researched too. Although the pRIFLE classification [320, 321] was proposed for pediatric use, it requires calculation of the eGFR and is therefore unsuitable for neonates, in whom the eGFR cannot be calculated. In 2014, Jetton et al. and Askenazi et al. introduced the neonatal modified KDIGO criteria (Table 14), which are based on the KDIGO diagnostic criteria [326, 327]. Similarly to the adult and pediatric KDIGO criteria, the neonatal modified KDIGO criteria define stages 1, 2, and 3 AKI primarily according to sCr values 1.5–1.9, 2.0-2.9, and ≥ 3 times higher than baseline, respectively. Although the baseline sCr is the minimum value prior to AKI diagnosis, it is only established at age ≥ 3 months in Japan [319], and not in children under 3 months. The level of sCr in neonates immediately after birth is extremely close to the level of maternal sCr (generally ≤ 1 mg/dl) [328]. It peaks at day 0–3, and declines to a minimum value (0.2–0.5 mg/dl) over the following 1 week to 20 months [328, 329, 330]. (Note that prematurity (in terms of gestational age and birth weight) is reported to affect the postnatal sCr levels and the speed at which they decline [328, 330]). Going forward, it is necessary to collect data on Japanese neonates to establish their baseline sCr levels. It should be taken into consideration that the current absence of established baseline levels requires multiple measurements.

Koralkar et al. used the neonatal modified KDIGO criteria to examine AKI and mortality in 229 very low-birth-weight infants both at 36 weeks of gestational age and with a birth weight of 500–1500 kg [331]; the very low-birth-weight infants diagnosed with AKI had a significantly higher mortality than those not diagnosed with AKI. In an examination of 455 very low-birth-weight infants using the neonatal modified KDIGO criteria, Carmody et al. found AKI to be associated with mortality and prolonged hospitalization [332]. In addition, a gestational age < 28 weeks was strongly associated with the onset of AKI; furthermore, all infants with a gestational age < 24 weeks were diagnosed with AKI, which indicates an association between prematurity and AKI. Rhone et al. used the neonatal modified KDIGO criteria to examine the association between the AKI onset and nephrotoxic medications (acyclovir, amphotericin B, gentamicin, ibuprofen, indomethacin, iohexol, tobramycin, and vancomycin) in 107 very low-birth-weight infants; consequently, these drugs were shown to be associated with the onset of AKI [333]. In an examination of 96 neonates with moderate to severe asphyxia who underwent therapeutic hypothermia, Sarkar et al. demonstrated that abnormal brain MRIs at 7–10 days of age were significantly more frequent in infants diagnosed with AKI according to the neonatal modified KDIGO criteria [334]. As detailed above, many recent studies have employed the neonatal modified KDIGO criteria for the diagnosis of neonatal AKI.

PubMed was searched for relevant studies published between January 1, 1980 and August 1, 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Many studies have indicated that biomarkers such as NGAL, cystatin C, L-FABP, IL-18, and KIM-1 may be useful for the early diagnosis of AKI and to predict the survival outcomes in children. However, interventions based on these indicators have not been reported to improve the renal or survival outcomes of AKI; therefore, their utility is limited.

Neutrophil gelatinase-associated lipocalin (NGAL) is a secretory protein that has a molecular weight of 25,000 Da and is secreted from activated neutrophils and tubular epithelial cells; the levels of NGAL in the blood and urine are known to be elevated in the hyperacute phase (i.e. the initial 2–4 h) of kidney injury. In an examination of 71 children undergoing a cardiopulmonary bypass (CPB) [335], the children who developed acute kidney injury (AKI) showed significantly elevated levels of serum and urinary NGAL 2 h after the CPB, with areas under the receiver operating characteristic curve (AUC) of 0.998 and 0.906, respectively; this study was the first to indicate the utility of biomarkers for the early diagnosis of AKI. An examination of 311 children undergoing cardiac surgery for congenital heart disease registered at three institutions [336] also indicated that the urinary NGAL is useful for the early diagnosis of AKI, despite a relatively low AUC of 0.71. The urinary NGAL was also reported to be useful for the early diagnosis of AKI in a heterogeneous pediatric intensive care unit (PICU) patient cohort which had undergone mechanical ventilation and bladder catheterization [337]. Likewise, in a systematic review/meta-analysis of 19 studies [124], a subgroup analysis of six studies featuring populations of pediatric patients only demonstrated the utility of NGAL for the early diagnosis of AKI. With regard to the survival outcomes, two studies have reported that NGAL is significantly associated with mortality [338, 339].

Cystatin C is a low-weight molecular protein (molecular weight: approximately 13,000 Da) produced by nucleated cells all over the body. It is unaffected by environmental changes inside or outside the cells, and is produced and secreted constantly; therefore, its concentration in the serum is constant. In addition, cystatin C is unaffected by factors such as inflammation, aging, the gender, the muscle mass, or exercise. The serum cystatin C passes freely through the glomerular basement membrane and is filtered by the glomerulus. As more than 99% of the serum cystatin C is reabsorbed by the proximal tubule and catabolized, healthy individuals excrete only a minimal amount of it in their urine. The serum cystatin C has been indicated to be useful for early, accurate diagnoses of AKI. The cystatin C concentrations in the serum and urine are known to increase 12–24 h after the onset of kidney injury. In an examination of 374 children undergoing CPB [340], AKI patients demonstrated significantly elevated serum cystatin C levels 12 and 24 h after the onset of AKI, with AUCs of 0.81 and 0.84, respectively; thus, the serum cystatin C was shown to be a useful biomarker for the early diagnosis of AKI. It was also reported to be useful for the early diagnosis of AKI in a study of 288 children undergoing cardiac surgery [341]. While measurement of the serum cystatin C is covered by insurance in Japan, that of the urinary cystatin C is not.

Interleukin-18 (IL-18) is an inflammatory cytokine induced in the proximal tubule. In a study of 55 children undergoing CPB [342], children with AKI demonstrated a significant acute phase (4–6 h after CPB) increase in their urinary IL-18 levels. The latter peaked at 12 h and remained high at 48 h. The AUC at 12 h (i.e. at the urinary IL-18 levels’ peak) was 0.75, demonstrating the utility of the urinary IL-18 as a biomarker for the early diagnosis of AKI. In a systematic review/meta-analysis of 18 studies [343], the urinary IL-18 was also shown to be useful for the early diagnosis of AKI in a subgroup analysis of five studies featuring populations of pediatric patients only. Measurement of the urinary IL-18 is not covered by insurance in Japan.

The L-type fatty acid-binding protein (L-FABP), kidney injury molecule-1 (KIM-1), and albumin are also known to show a marked increase in urine as a result of kidney injury; these biomarkers have been studied for their potential utility in the early diagnosis of AKI. The L-FABP is a protein with a molecular weight of 14,000 Da that is expressed in the liver, the small intestine, and the proximal tubular epithelial cells; since 2011, measurement of the L-FABP as a biomarker has been covered by insurance in Japan. In a study of 40 pediatric patients having undergone CPB [344], children who developed AKI demonstrated a significant acute phase (4 h after AKI onset) increase in the urinary L-FABP. KIM-1 is a membrane-spanning glycoprotein expressed in the proximal tubular epithelial cells. In a study of 40 children undergoing CPB [345], children with AKI demonstrated a significant acute phase (12 h after CPB) increase in KIM-1. Moreover, in a prospective study of 294 children undergoing cardiac surgery, the urinary albumin/creatinine ratios 0–6 h after surgery were useful for the prediction of AKI [346].

Due to the diversity of the pathologies involved in AKI and the decline in the glomerular filtration rate (GFR), the use of a single biomarker to increase the accuracy of early diagnosis is of limited efficacy. One attempt to increase the accuracy of biomarkers for the diagnosis of AKI is to assemble a “panel” that combines multiple biomarkers and the renal angina index (RAI), an indicator of the risk of AKI onset [347, 348]. Another advantage of panels is that, as each of the biomarkers that comprise them demonstrate favorable sensitivity and specificity at different periods, these different time phases may complement one another.

The uses of biomarkers have been studied in children, though not as often as in adults. Relevant studies indicate that these biomarkers may be useful for the early diagnosis of AKI and for the prediction of the survival outcomes. However, many of these studies involved relatively homogeneous populations, e.g. children undergoing CPB; the utility of these biomarkers has not been sufficiently assessed in populations of patients with diverse pathologies. Furthermore, the interventions based on these indicators have not yet been reported to improve the renal outcomes or survival outcomes of AKI; therefore, their utility is limited.

PubMed was searched for relevant studies published between January 1980 and July 2015, and papers related to the present CQ were identified from the search results.

Open image in new window

Many observational studies have reported that pediatric AKI non-survivors exhibit fluid overload compared to survivors. Few manuscripts have discussed fluid overload in neonatal AKI; there is little evidence to support prioritization of the fluid overload assessment for determination of the indication of blood purification in neonates.

In pediatric acute kidney injury (AKI), life-threatening conditions resistant to conservative therapy, such as electrolyte disorders (hyperkalemia, etc.), fluid overload (pulmonary edema, heart failure, etc.) metabolic acidosis, and uremia symptoms (pericarditis, impaired consciousness, convulsions, etc.) are absolute indications for blood purification just like in adult AKI; in these cases, blood purification must be initiated immediately. However, for relative indications that are not considered to be immediately life-threatening, the criteria for the initiation of blood purification have not yet been defined. No randomized controlled trials have assessed the indications for blood purification and the timing of its initiation in pediatric AKI.

In 2001, Goldstein et al. conducted a single-center study [349], followed by a large-scale multi-center study of continuous renal replacement therapy (CRRT) for pediatric AKI whose results were reported in 2005 [350]. This study examined the predictors of survival and death in 116 children registered in the Prospective Pediatric Continuous Renal Replacement Therapy (ppCRRT) Registry who underwent CRRT for multiple organ failure. Even when controlling for the severity of the illness (as measured by pediatric risk of mortality [PRISM] score), the %FO at CRRT initiation was an independent predictor of survival; the %FO was significantly lower in survivors than in non-survivors (survivors: 14.2 ± 15.9 versus non-survivors: 25.4 ± 32.9, p < 0.05), while the mortality was significantly higher when the %FO was > 20% (< 20:40 vs. > 20:58%) at CRRT initiation. The same group later demonstrated that the %FO at the initiation of blood purification was correlated with mortality (< 10:29.4%, 10–20:43.1%, > 20:65.6%) [351]. Modem et al. reported that FO is a factor of poor survival outcomes [352]. Many studies of AKI in multiple organ failure [353, 354, 355], stem cell transplantation [356], and extracorporeal membrane oxygenation (ECMO) following cardiac surgery [357, 358] have also reported that a lower %FO at CRRT initiation is associated with more favorable survival outcomes. Similar results have also been reported in assessments of FO based on the body weight at hospital admission, at intensive care unit (ICU) admission, and at the initiation of blood purification [359]. Therefore, the early initiation of blood purification to prevent fluid overload may improve the survival outcomes; when determining whether blood purification is indicated in pediatric AKI, we suggest that the fluid overload assessment be taken into consideration in addition to absolute indications.

However, these results all were obtained from observational studies; there is no high-quality evidence from interventional studies. In addition, a study of blood purification in children undergoing cardiac surgery failed to find an effective timing for initiation [360], while fluid overload was reported not to be an absolute predictor of the survival outcomes [361]. Unnecessary blood purification should be avoided in cases of mild AKI, in which the renal function recovers quickly. Blood purification carries serious complications, including catheter-related infection, an increased risk of bleeding from anticoagulation, and hemodynamic fluctuations unique to children of small constitution; therefore, the indication for blood purification and the timing of initiation must be considered comprehensively.

In neonatal AKI just like in pediatric AKI, renal replacement therapy (RRT) is considered when prolonged oliguria/anuria prevents the appropriate adjustment of the body fluid, electrolytes, and blood nitrogen level. The overall mortality in neonatal AKI is reported to range between 11.3 and 48.3% [362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372]; while the mortality is reported to be 4.1–71.7% in premature neonates [331, 373, 374, 375, 376], 13.9–70.0% in asphyxiated neonates [369, 377, 378], 2.9–11.6% in neonates undergoing a cardiopulmonary bypass/cardiac surgery [379, 380, 381], 71.2% in sepsis [382], and 50–100% in neonates with AKI who undergo blood purification [363, 367, 368, 369]. Due to the different definitions of AKI in these studies and to the major differences in the standards of neonatal intensive care medicine between countries and institutions, uniform comparisons of past studies are difficult. The risk factors for death in neonatal AKI include mechanical ventilation, hypervolemia (%FO ≥ 7%), chronic heart failure, a low birth weight, hypoxia, oliguria/anuria, dialysis, and metabolic acidosis [362]. The risk of death is particularly high in neonates with oliguria [362, 363, 364, 365, 366, 371, 377, 378]. However, no studies have discussed FO in neonatal AKI; there is little evidence to support prioritization of the fluid overload assessment when determining the indication of blood purification in neonates. Low-birth-weight infants present technical problems such as vascular access; however, the indication for acute blood purification in neonates must be determined comprehensively on a case-by-case basis.

PubMed was searched for relevant studies published between January 1980 and July 2015, and papers related to the present CQ were identified from the search results. Manuscripts that supplemented the commentary were hand searched as appropriate.

Open image in new window

Observational studies of children and neonates who underwent CRRT or other modalities of blood purification have been performed. However, there has been no evidence to demonstrate the effects of different blood purification modalities on the outcomes, nor of their superiority to peritoneal dialysis.

The modalities of blood purification for acute kidney injury (AKI) include peritoneal dialysis (PD), extracorporeal intermittent renal replacement therapy (IRRT), and continuous renal replacement therapy (CRRT). In the past, PD was often the first choice; however, due to progress in the techniques of vascular access, in the types of catheters, the hemodialysis (HD) devices, and the pediatric intensive care management, extracorporeal CRRT has become more common. At present, the only studies that have compared PD and extracorporeal CRRT are observational studies [383, 384], and there is no evidence that one modality is superior to the other. However, as in adults, CRRT is considered preferable to IRRT for hemodynamically unstable patients. Many evidences that inform the blood purification modality selection for adults can be applied pediatric AKI; however, the incidence of pediatric AKI is < 1% in all hospitalized children [385], and 4.5% in children admitted to the ICU [386]. Differences in knowledge and health care resources between regions and institutions may greatly affect the selection of blood purification modalities. Further investigations are necessary to better inform the selection of suitable blood purification modalities.

In children (aside from neonates), blood purification can be performed safely with a combination of low priming volume and multipurpose blood purification devices. For vascular access, the size of the catheter is chosen to suit the patient’s constitution (Fig. 6). The standard values for the quantity of blood flow (QB), the dialysate flow rate (QD), and the filtration rate (QF) are 1–5 ml/kg/min, QB × 0.2–2.0, and 0–20% of the QB, respectively. With regard to circuit priming before the initiation of HD, a priming volume of ≥ 10% of the circulating blood volume causes hypotension at the initiation of dialysis; therefore, priming with blood products is preferred [387]. After priming with blood products, it is recommended dialyzing the priming blood in the circuit to adjust the electrolyte and acid-base balance, and by removing the potassium and citric acid in the blood products.

Open image in new window

In the past, PD used to be the method of choice for low-birth-weight infants (including neonates) due to the technical problems. However, extracorporeal blood purification can recently be performed safely. Extracorporeal blood purification has been technically possible in Japan since 2001, when blood purification devices (QB can be adjusted from 1 ml/min), filters, and other related devices became commercially available. In 2013, the Pediatric Acute Blood Purification Handbook described blood purification in neonates and the Guideline for Neonatal Extracorporeal Blood Purification was published in Japan. In addition, the Neonatal Extracorporeal Blood Purification Manual was published in 2014. Meanwhile, there have been many reports on blood purification primarily in pediatric patients (including some neonates) both in Japan and outside of Japan [350, 388, 389].

In blood purification for low birth-weight-infants (including neonates), vascular access is a specific and important issue; in addition to the standard central venous route, the umbilical arteries/veins and peripheral arteries can also be used (when the flow rate is low, the peripheral veins can sometimes also be used). There has been a Japanese case report of blood purification in an infant weighing < 500 g; however, blood purification in infants weighing < 2 kg is considered to require experienced skill. Central venous catheter size should be large. Although variations between institutions exist, the catheter sizes used for infants weighing 1, 2, and 3 kg are generally 17 G, 15 G, and 6 Fr, respectively. As in pediatric patients, the circuit is basically primed with mixed blood in order to prevent hypotension. The blood preparation is recycled and dialyzed to remove potassium and citric acid before initiating the blood purification. Prevention for hypothermia is necessary. Details are described in the above-cited guidelines and handbooks (Fig. 6).

Outside Japan, the blood purification of low-birth-weight infants (including neonates) was primarily consisted of peritoneal dialysis [390, 391, 392]. The improvements in blood purification devices have recently led to an increase in extracorporeal acute blood purification [393, 394, 395]. However, there have been no randomized controlled trials (RCTs) to examine the performance of extracorporeal blood purification in Japan or elsewhere.

The optimal blood purification modality also depends on the disease conditions. For AKI with acute brain injury, intracranial hypertension, or cerebral edema, CRRT [continuous hemodiafiltration (CHDF) or 24 h of PD] is recommended, as IRRT may cause intracranial hypertension, dialysis disequilibrium syndrome, and reduced blood pressure [396].

PubMed was searched for relevant studies published between January 1980 and July 2015, and papers related to the present CQ were identified from the search results. The literature needed for the commentary was manually extracted from PubMed as appropriate.

Open image in new window

Despite the existence of multiple case reports and case series, there is no relevant high-level evidence.

Children with severe motor and intellectual disabilities caused by factors such as chromosomal abnormalities, multiple abnormality syndromes, and neonatal asphyxia (cerebral hypoxia) are estimated to occur in approximately 0.3 out of 1000 live births. Factors such as severe asphyxia and infection frequently cause acute kidney injury (AKI) in neonates. Neonates with AKI necessitating blood purification often present comorbid serious brain injury. Children with severe motor and intellectual disabilities that correspond to grades 1–4 of Oshima’s classification have a high risk of developing severe infections, and frequently require renal replacement therapy (RRT) for AKI. However, most past reports of RRT in children with severe impairments, mostly from Japan, have involved the issues of chronic dialysis. Out of a total of 23 reports (37 patients), 20 of them (32 patients) described the initiation (or scheduled initiation) of peritoneal dialysis (PD), two (4 patients) described the initiation of chronic hemodialysis (HD), and one (1 patient) described PD and continuous hemodiafiltration (CHDF) for AKI. In another report, the patient was treated without initiating dialysis. Many of these reports have suggested that a multidisciplinary health care team should consider the patient’s case and should ultimately decide what to do after consulting the patient’s family. Meanwhile, reports from outside Japan have stated that the frequency of peritonitis in PD for children with psychomotor retardation is the same as in children without psychomotor retardation if dedicated cooperation and support are provided. There have also been reports of dialysis initiation in children with chromosomal abnormalities [397, 398, 399]. These reports have demonstrated that RRT can be performed relatively safely in children with severe motor and intellectual disabilities. However, the medical staff experience a great physical and psychological burden; therefore, the health care team also needs support.

There are no definitive criteria upon which to determine the indication for RRT in children with severe impairments; thus, it must be considered on a case-by-case basis. The health care team should decide on a therapeutic strategy after considering the patient’s present status and long-term survival prognosis amongst themselves, explaining the nature of the treatments to the patient’s family, and presenting the respective advantages and disadvantages of treatment versus no treatment. Various guidelines can be referred as well. This concept is called shared decision-making; essentially, health care professionals must share information with the patient’s family and decide on a therapeutic strategy together. The process is described below.

PubMed and Ichushi-Web (Japanese language) were searched for relevant studies published up to August 30, 2015, and papers related to the present CQ were identified from the search results. References in Japanese are not shown in this article.