Ongoing Clinical Trials in AKI

Sample Size and Endpoints for AKI Prevention Trials

What then is the appropriate sample size with adequate power for a prevention trial in AKI? In the recent Translational Research Investigating Biomarker Endpoints in AKI Consortium biomarker study, the incidence of severe AKI (doubling of serum creatinine or RRT requirement) postcardiac surgery was 5% (22). Thus, with this incidence rate, α=0.05, power of 0.9, and 30% effectiveness of an intervention (i.e., 30% relative reduction in the development of severe AKI from 5% to 3.5%), 3799 patients per arm would be needed. Even when a trial is enriched with high-risk patients, the incidence of AKI after either contrast administration or cardiac surgery is still generally low; for example, the risk of AKI requiring RRT after cardiac surgery in patients with a creatinine of 2–4 mg/dl is 10%–20% (23). In this population, with 20% incidence and 30% effectiveness, 822 patients per arm would be required to successfully show a reduction in the need for RRT. Thus, it is notable that only 3 of 28 contrast studies and only 3 of 30 cardiac surgery trials in adults are enrolling 800 or more patients in total. Although a number of these trials are pilot studies, which are useful for assessing feasibility of protocol design and execution of future trials, many of the ongoing trials are underpowered, and their results will have to be interpreted with caution. An additional caution regarding the interpretation of pilot and feasibility studies is that estimates of effect sizes from small pilot studies for use in sample size/power calculations for full-scale trials may be unreliable (24). In regard to sample size, the Prevention of Serious Adverse Events Following Angiography (PRESERVE) Trial is notable in that it plans to enroll 8680 patients with preangiography estimated GFR <60 ml/min per 1.73 m² and diabetes mellitus or preangiography estimated GFR <45 ml/min per 1.73 m² with or without diabetes mellitus, thus enriching for high-risk patients. The PRESERVE trial will address the use of NAC versus placebo and sodium bicarbonate versus normal saline for the prevention of contrast-induced AKI in the setting of coronary and noncoronary angiography. Because the use of NAC and the appropriate volume expansion strategy for the prevention of contrast-induced AKI remain uncertain, the results of this definitive study will be anxiously awaited.

Another key consideration for the design of prevention trials (and clinical trials in general) is the proposed effect size of the intervention studied. We believe that a relative effect of 10%–30% is reasonable and that an effect size greater than 40% is unrealistic given the complex pathophysiology of AKI (25). For reference, cholesterol-lowering therapy for primary prevention resulted in an approximately 30% relative risk reduction of coronary events (26). A similar effect size was observed for the reduction in all-cause mortality in patients treated for long-term sustained hypertension (27); finally, low tidal volume ventilation resulted in a 22% relative risk reduction in mortality versus high tidal volume ventilation in patients with acute lung injury/adult respiratory distress syndrome (ALI/ARDS) (28). Certainly, however, estimates of effect size for AKI endpoints should be based on preclinical and early phase clinical trial data as well as the biologic mechanism of action of the therapy of interest. In regard to the current studies for the prevention of AKI, it is noteworthy that the largest trial for the prevention of contrast-induced AKI has a target enrollment of 8680 participants, whereas the remaining trials have target enrollments of 601–1000. The two largest trials for the prevention of cardiac-associated AKI are targeting 2070 and 1610 participants, respectively, whereas the remaining trials have enrollment targets of 16–820. Assuming a generous 20% incidence of AKI in either of these settings, a study of 1000 participants (500 per arm) has sufficient power (90%) to detect a difference with an α=0.05 if the intervention results in a 38% decrease in the incidence of AKI (i.e., to an incidence of 12%); a study of 500 participants (250 per arm) would require the intervention to reduce the incidence of AKI by more than 50% for sufficient power. Thus, it is discouraging to report that the majority of the ongoing trials for the prevention of AKI seem to be underpowered to show a biologically plausible effect for the intervention.

Finally, prevention trials must examine an endpoint that is clinically important, particularly because the number needed to treat to prevent one episode of AKI will be high. Although small increases in serum creatinine are associated with adverse outcomes, exposing a large percentage of patients to a medication or intervention that may have risk may not outweigh the potential benefit of preventing AKI. Thus, we suggest that clinical trials examine the endpoints of severe AKI (e.g., doubling of serum creatinine), requirement of RRT, impaired quality of life measures, development of CKD, and/or mortality, recognizing that adequately powered studies that examine these endpoints will need to be very large; therefore, examination of a composite of these endpoints is reasonable. Interventions that improve these outcomes are the most likely to be used in clinical practice or gain approval by the Food and Drug Administration (FDA). Indeed, the FDA has historically strongly preferred patient-centered composite endpoints for clinical trials (for example, in the case of CKD progression, doubling of serum creatinine or progression to end stage kidney disease). It is important to emphasize that there may be a difference in what the FDA might approve and what a nephrologist would use if clinically available. For example, if NAC was definitively proven to reduce the incidence of contrast-induced AKI as defined by a 0.3-mg/dl increase in serum creatinine, it is likely that most, if not all, nephrologists would use it for prophylaxis, because it is relatively inexpensive and safe. Thus, studies examining AKI with this definition are of interest to nephrologists and are of value. The FDA, however, has expressed reluctance to approve agents based on this definition of AKI, and thus, more stringent criteria likely would need to be examined if an intervention were to seek FDA approval. However, as additional data emerge supporting a potential link between AKI and progression to CKD and suggesting that CKD progression may occur with mild AKI, ongoing discussions with the FDA will need to occur regarding appropriate endpoints for AKI clinical trials.

Regarding studies for the prevention of contrast-induced or cardiac surgery-associated AKI, we note that, despite a now-accepted definition of AKI proposed by the AKI Network that rests on an increase in serum creatinine of at least 0.3 mg/dl or 50% over baseline or a decrease in urine output (if measured) to <0.5 ml/kg per hour for >6 hours (29), few studies use these criteria as the definition of AKI. Furthermore, the nomenclature used to refer to AKI after contrast administration also varies widely (e.g., a search of acute kidney injury on clinicaltrials.gov identifies only a minority of contrast-induced AKI prevention trials). Thus, we recommend a common nomenclature and definition for AKI after contrast administration, and we suggest that the phrase contrast-induced AKI be used and that the definition of AKI is the same as the AKI Network criteria.

Paucity of AKI Studies in Children

Only eight trials were identified in children, suggesting that this population is understudied. This finding is unfortunate, because studies of children have provided important insights into AKI, especially because they do not possess most of the chronic diseases that confound AKI studies in adults. For instance, most of the AKI biomarker validation work (30,31) and the studies showing the association of fluid accumulation with increased mortality were first performed in children (32–34). Children are a particularly important and useful cohort for the study of AKI postcardiac surgery, because the incidence rate of AKI is much higher (30%–50%) (30,35) and AKI-associated mortality and morbidity rates in critically ill children are similar to adults when the serum creatinine doubles (35). Thus, fewer patients would be required for an adequately powered study. Although protection from harm is a paramount concern, exclusion of children from potentially beneficial therapeutic trials could also be viewed as a form of harm, because pediatric subspecialists must often extrapolate data from adults and treat children with off-label medications or devices. Separate pediatric trials are difficult to develop for many reasons, including the lack of specific funding, need for a separate research infrastructure, and inadequate sample size for acceptable power; therefore, we recommend that children >13 years of age be considered for inclusion in AKI prevention and treatment trials.

Barriers to Successful Drug Trials in the Treatment of AKI

Of the 23 trials addressing general management of AKI, 22 trials are being conducted in the ICU, with 21 trials related to RRT. There is only one drug trial, and the drug being tested is furosemide. Given the history of failed drug trials in the treatment of AKI, the lack of drug trials is not surprising. Extensively reviewed elsewhere (36,37), reasons for failure of drug trials include the heterogeneity of AKI and underpowered studies. In particular, many studies failed to properly apply the preclinical bench work to the design and conduct of the clinical trial; for example, hypotension was a known complication of atrial natriuretic peptide (ANP) in preclinical studies, prompting intrarenal administration or coadministration with dopamine (38); however, clinical trials used systemic ANP. Predictably, hypotension occurred, thus obscuring any potential benefit of ANP to accelerate renal recovery. In addition, many drugs have been administered late in the course of AKI; notably, human IGF-1 reduced renal injury in rodent models of ischemic AKI when administered up to 4 hours postreperfusion (39). However, with the serum creatinine definition of AKI used in the human trial, patients were likely entered between 24 hours and 6 days after AKI onset (40).

Thus, a key element of future drug development studies will be to establish the therapeutic window for each agent in preclinical studies. In animal models, most agents are studied prophylactically; however, agents that improve kidney function when administered before the onset of AKI are typically less effective or ineffective when administered even just 1 hour after the initiation of AKI. Agents with just a 1-hour therapeutic window are unlikely to be clinically useful, and it is critical for drugs to be tested at longer time points after the onset of AKI. In this regard, the Early Intervention in Acute Renal Failure trial showed the difficulty of early randomization, even when cognizant of the therapeutic window; in this trial, urinary biomarkers were used to randomize patients with sepsis to treatment with erythropoietin or placebo, which occurred at a median of 13 hours after the renal insult (41). As the investigators acknowledge, this amount of time is well outside of the proposed therapeutic window of 6 hours for erythropoietin (42). Although point of care biomarker testing with a short turnaround time might assist with rapid randomization in future clinical trials, successful drug intervention for the treatment of AKI might ultimately rest on finding a drug that is effective ≥12 hours after the initiation of AKI. Thus, agents that accelerate recovery of AKI or decrease systemic complications of AKI need to be developed and studied in preclinical models.

An additional barrier to successful clinical trial execution includes certain difficulties with trial enrollment. Because nephrologists are typically consultants, they must obtain approval from the treating physician (e.g., intensivists and surgeons) for study enrollment. Also affecting trial enrollment is that many critically ill patients with AKI cannot consent for themselves because of their underlying illness, and therefore, surrogate consent is required. Thus, it is essential that collaborations outside of the nephrology community with intensivists and other specialists be developed in the design of future clinical trials to assist with both enrollment and obtaining consent.

Finally, single-center trials suffer from numerous drawbacks, including limited sample size and restriction to prespecified clinical settings. Given the lack of evidence-based consensus practices in AKI, single-center trials may heavily reflect local medical practice variations, which may lead to study results that are either invalid or lack external generalizability. Furthermore, as a community, we need to carefully consider the validity of positive results obtained from underpowered, single-centered trials, because false-positive results from underpowered studies can inappropriately influence medical practice. In terms of the publication of such trials, the treatment effect should be reported, and the biologic plausibility of the treatment effect needs to be fully considered. Although financial concerns are often cited as a limiting factor in the conduct of multicenter trials, one can argue that the cost of small uninformative trials may not be justified. Thus, because of these multiple concerns, we discourage the pursuit of small single-center, underpowered trials unless they have clearly been designed as exploratory or feasibility studies as a prelude to large trials.

A Call for Best-Practice Studies

Although successful drug development to treat AKI is an important pursuit, improving AKI-associated morbidity and mortality does not rest solely with the development of drug-based interventions. In this regard, useful lessons to guide the nature of clinical trials in AKI can be learned from the progress made in the treatment and management of acute myocardial infarction (AMI) versus ALI/ARDS. Specifically, drug treatments, including aspirin, glycoprotein IIb/IIIa inhibitors, β-adrenergic antagonists, angiotensin-converting enzyme inhibitors, and thrombolytics, have revolutionized the management of acute coronary syndromes and dramatically decreased mortality rates. In contrast, no drug has been shown to improve the mortality of ALI/ARDS. Remarkably, however, mortality in ALI/ARDS has decreased from 40% to 20% in the past 15 years (28,43) because of the study of best practice, including ventilator management (low versus high tidal volume), protocolized fluid management, and infection control measures such as prevention of ventilator-associated pneumonia, that are now standard of care.

Although AMI is a relatively homogeneous disease with what might be considered a single cause (i.e., acute arterial occlusion), ALI/ARDS is a heterogeneous disease with many different etiologies, although the use of a simple definition of ALI/ARDS, without regard to etiology, has been critical for identifying and studying large numbers of patients in clinical trials. Thus, AKI is more similar to ALI/ARDS than AMI, because AKI is also a heterogeneous disease with multiple etiologies and a simple definition that ignores etiology. By extension, therefore, we suggest that large clinical trials devoted to clarifying best practice regarding RRT and other management aspects of AKI will have the highest likelihood of having a significant impact on AKI-related outcomes at present. We suspect that a large portion of the morbidity and mortality of patients with AKI rests on its late recognition and complications and the delivery of RRT. For example, medications are likely over- and underdosed in patients with AKI. Overdosing of medications probably occurs at the initiation of AKI before the detection/recognition of AKI based on changes in serum creatinine; conversely, medications are likely underdosed after the onset of RRT because of increased clearance from high-flux dialyzers. Accumulation of vasoactive drugs may lead to systemic consequences such as hypotension and prolongation of AKI, whereas underdosing of antibiotics may prolong an episode of sepsis; either of these complications may subsequently lead to increased mortality. Thus, the study of medication dosing in patients with AKI may be a worthy pursuit to reduce complications in patients with AKI. One ongoing trial will examine the role of a pharmacist-based monitoring and intervention regarding drug dosing and use of nephrotoxic medications; indeed, reduction of medication errors and avoidance of nephrotoxic drugs may be a key intervention to reduce AKI duration and morbidity, which was shown in a study using computer-based alerts regarding serum creatinine in patients receiving nephrotoxic medications that led to earlier medication changes and lower relative risk of serious kidney injury (44).

Another important area of research regarding best practice involves prevention of complications related to RRT such as avoidance of intradialytic hypotension, which occurs in >20% of treatments (45). Finally, improved methods of anticoagulation during RRT to reduce dialyzer clotting are also needed. On a practical level, frequent dialyzer clotting exerts a significant impact on the delivery, cost, and labor required to perform continuous RRT. A final advantage of a large clinical trial that examines best practice is that drug treatments for AKI may be examined within the trial, similar to the design of the ARDS Network Trials.

Many clinical trials are currently examining aspects of best practice regarding RRT, including studies of anticoagulation, dialysis catheter management to avoid bacteremia, fluid removal strategies, hemodynamic monitoring to avoid intradialytic hypotension, and pharmacokinetic studies to optimize drug dosing. Ethanol lock for dialysis catheters is being tested in one of the largest trials (n=1300). The results of the anticoagulation studies (n=190–250), hemodynamic studies (n=40–600), and drug dosing studies (n=8–60) may be too small to show an impact on AKI recovery rate or mortality; however, these initial studies will hopefully be informative for the design of larger trials.

Another important management issue in patients with AKI is fluid management. Numerous observational studies have shown that fluid overload is associated with worse mortality in patients with AKI; however, it remains to be shown whether a conservative fluid management approach or early initiation of RRT for the purpose of fluid removal will impact outcomes. In this context, an ongoing management trial in children will use neutrophil gelatinase-associated lipocalin as a biomarker of AKI to guide fluid management.

With regard to best practice for RRT, the Veterans Affairs/National Institutes of Health Acute Renal Failure Trial Network trial (n=1124) and Randomized Evaluation of Normal Versus Augmented Level of Replacement Therapy trial (n=1508) show that large, adequately powered studies regarding RRT may be performed. These successful trials convincingly showed that more intensive dosing of RRT does not confer a mortality benefit in patients with AKI and have informed clinical practice. Based on the information from these trials, additional studies regarding dose of dialysis are unnecessary at this time.

Although a clinical trial examining the appropriate timing for initiation of RRT is highly desired by clinicians, it is unlikely that such a trial will be conducted in the near future. Complexities of such a trial include the development of criteria to define the onset of AKI to assess early versus late RRT (e.g., serum creatinine, BUN, and emerging biomarkers), criteria to predict which patients have severe AKI and will need RRT so that RRT is not performed in patients who will rapidly recover, and agreement among nephrologists and non-nephrologists regarding what would constitute equipoise between an early versus late start. Thus, additional work to clarify these issues will be important for the design and execution of such an important trial.

In terms of prioritization of the research agenda for clinical trials, we believe that several criteria must be used to prioritize treatments for study; these criteria would include a plausible mechanism of action for an intervention, the existence of phase 2 human data to guide study design and assure patient safety, a high perceived likelihood that the intervention would benefit patients with AKI, and most importantly, the feasibility of conducting the study.

Study of Non-ICU Patients with AKI

Although most of the AKI management trials are conducted in the ICU, non-ICU patients have increased morbidity and mortality from an episode of AKI, and we recommend that larger studies of non-ICU patients with AKI be conducted. This population may be particularly well suited to management studies designed to prevent worsening of early AKI. A recent small clinical trial examined the impact of early nephrology consultation on outcomes in patients with a 0.3 mg/dl rise in serum creatinine, and it found that the peak creatinine was lower in patients who received early nephrology consultation versus those patients who received usual care (46). This study suggests that early evaluation of patients with AKI to correctly determine diagnosis (e.g., prerenal versus intrinsic AKI), optimize fluid management, and avoid nephrotoxic insults may improve patient outcomes, and this concept certainly warrants evaluation in larger studies.

Clinicaltrials.gov and the AKI Advisory Group Clinical Trial Online Resource

How then do we move forward? We believe that the development of a United States-based clinical trials network focused on AKI, similar to the National Heart, Lung, and Blood Institute-sponsored ARDS Network, would be the most important advance to move the AKI research field forward. The ARDS Network has been highly successful in designing and executing definitive negative and positive trials on the pharmacologic and supportive management of acute lung injury. In this model, areas of best practice may be examined along with specific interventions. For example, low versus high tidal volume ventilation was examined at the same time as ketoconazole for patients with acute lung injury; although the ketoconazole trial aspect was negative, low tidal volume ventilation is now standard of care for patients with ALI/ARDS. Until the development of an analogous AKI Network, we suggest that clinicaltrials.gov may be a useful, currently available tool to enhance communication among AKI investigators. In this platform, studies can be registered and updated, and investigators with similar research interests can be identified for potential collaborations. It is the intention of the AKI Advisory Group to maintain an up to date resource of prevention and management clinical trials using clinicaltrials.gov that will be available at www.ASN-online.org. Thus, we earnestly encourage investigators to update trials regarding recruitment status, results, and publications using this existing mechanism. Although we have made every effort to identify and summarize all of the current trials in this report, we note that many trials have not been recently updated and that trial and publication statuses are inaccurate. Additionally, we suggest that all trials regarding AKI use a common nomenclature, especially studies of contrast-induced AKI, and that acute kidney injury should be a key phrase in all registered trials. Ultimately, the kidney community must focus less on small trials and meta-analyses and rather, find ways to collaborate to conduct large clinical trials. We hope that the AKI Advisory Group Clinical Trial Online Resource may be an additional mechanism to promote this collaborative approach and start a national dialog about clinical trials in this area.