AKI trial endpoints: surrogate markers of AKI-associated complications

In considering optimal AKI trial endpoints, the purpose of the trial impacts endpoint options. AKI prevention and attenuation trials endpoints should ideally account for the incidence, severity, and duration of AKI allowing for gradation across the spectrum of AKI. Long-term outcomes (e.g. Major Adverse Kidney Events —MAKE—death, initiation of dialysis, and persistent kidney dysfunction) are not ideally suited as their incidence in non-enriched populations in prevention trials will be extremely low (< 5–10%). Thus, using this outcome would necessitate trials of several thousand patients to demonstrate efficacy [129]. Treatment trials should utilize patient-centered long-term outcomes (e.g. MAKE, hospital and dialysis free days) although depending on the cohort, clinical setting, and intervention, one component of MAKE may dominate the overall outcomes.

Importantly, when considering AKI as an outcome for a trial, short-term and transient changes in creatinine - and urine output-based endpoints are practical and will undoubtedly increase event rates, facilitating smaller and faster trials. However, it remains unclear how small changes in serum creatinine reliably predict long-term outcomes. While some epidemiology studies demonstrate links between Stage 1 AKI and adverse outcomes, the data is inconsistent [130–132]. However trials performed in critically ill patients, are different from those looking at community-acquired or ward-based AKI, and ICU investigations are better suited to look at more severe and persistent AKI and MAKE.

Furthermore, MAKE, especially mortality, is influenced by factors unrelated to AKI (e.g. post-operative sepsis or stroke), and mortality is a competing risk for long-term outcomes such as CKD and dialysis-dependence/ESKD [133]. Strategies for investigating CKD progression as an outcome after AKI include: ensuring adequate proportions of patients in each stage of AKI severity, matching patients based on age, baseline kidney function, and comorbidities, and frequent laboratory follow-up to analyze trajectory of kidney function [134]. Importantly, establishing baseline (pre-enrolment) kidney function can be problematic, depending on the availability of prior serum creatinine data, and this can impact AKI incidence, AKI severity, and the incidence of recovery. In patients without known CKD, when a reliable premorbid creatinine (or cystatin C) value is unavailable, the 2012 KDIGO guideline recommends estimating baseline creatinine either by back-calculation using the Modification of Diet in Renal Disease equation (assuming a baseline eGFR of 75 mL/min/1.73 m²) or using the lowest in-hospital creatinine value as a surrogate, provided unrecognized CKD has been excluded [15]. However, back calculation can both over and under estimate the incidence of AKI depending on the cohort. It is more likely to underestimate the incidence in younger healthy cohorts and over estimates in sicker and older cohorts [135, 136]. To overcome these issues, albuminuria may be used as a surrogate marker of CKD progression. The prospective, longitudinal, observational ASSESS-AKI study found that higher urine albumin-creatinine ratio 3 months after hospitalization with AKI was associated with increased risk of CKD progression (HR 1.53, 95% CI 1.45–1.62) [137].

Albuminuria could serve as a surrogate to predict patients at risk of CKD progression after AKI as well as a therapeutic target, such as with renin-angiotensin-aldosterone inhibitors, sodium-glucose cotransporter 2 inhibitors (SGLT2i) or glucagon-like peptide-1 receptor agonist (GLP-1) [137–142]. Other structural and functional biomarkers, including plasma soluble tumor necrosis factor receptor 1 and 2 (sTNFR1, sTNFR2), urine neutrophil gelatinase-associated lipocalin (uNGAL), urinary tissue inhibitor of metalloproteinase 2 and insulin-like growth factor–binding protein 7 (TIMP-2*IGFBP7), have been shown to predict CKD progression or recovery from AKI and may have a role as a surrogate outcome in AKI trials, however this requires future investigation and validation [143–149]. This is separate from the use of damage and stress biomarkers as an enrichment strategy to increase AKI incidence and severity [150–154].

In trials of patients with AKI who require dialysis, standardized definitions of recovery and dialysis dependence are lacking; as such, several definitions have characterized renal recovery or dialysis dependence including continued dialysis at hospital discharge, 28 days, and/or 90 days [155–157]. Dialysis dependence at 90 days is consistent with the definition of ESKD, and while this is certainly a patient-centered outcome, assessment before 90-days is equally relevant. Among these trials, definitions of recovery from dialysis dependence also vary, such as being alive without dialysis for 14 consecutive days as used in the LIBERATE-D trial, or lack of need for dialysis with a minimal creatinine clearance of 20 ml/min used in a post hoc analysis of the Acute Renal Failure Trial Network (ATN) study [155, 156]. Days alive and dialysis-free is a novel alternative endpoint that accounts for a patient-centered outcome as well as mortality associated with in-hospital AKI [155, 158–162]. However, this is inherently a composite outcome that treats death and dialysis dependence as equally important. Hierarchical composite endpoints and calculations of win ratios have been explored to account for the relative importance of different outcomes and provide more nuanced interpretations of results of AKI trials where death and dialysis-dependence are competing outcomes [133, 163]. Even among patients who no longer require dialysis after AKI, the risk of recurrent AKI and development of ESKD remains high, and future studies should investigate how to optimize care after incident AKI [164].

Systemic complications from AKI are often included in secondary, post hoc, or safety outcomes of AKI trials. Examples of systemic complications include prolonged mechanical ventilation, cardiac arrhythmias, exposure to hypotension / hemodynamic instability, delirium, infections, and bleeding. Table 3 describes outcomes from AKI trials related to other organ systems. Since AKI is associated with these complications and these influence patient outcome, these complications can also be used as a composite endpoint in treatment trials.

Finally, given the high morbidity associated with hospital-acquired AKI, it is important for AKI trials to increase and expand the measurement of patient- and caregiver- centered outcomes for those who survive their critical illness. Length of stay and readmission rates, while often regarded as hospital quality metrics, also represent time patients spend in the hospital rather than at home and can be a useful outcome in AKI prevention and post-discharge AKI care trials [150, 165–167]. However, this endpoint should only be used in a context-specific manner (e.g. in the context of sepsis-associated AKI, but not for trials investigating cardiac surgery associated AKI). This endpoint should also differentiate between time at a non-hospital facility (e.g. skilled nursing facility or long-term acute care facility) and time at home.

Financial burden and impact on employment are equally life-changing patient-centered outcomes, but are seldom explored in AKI studies [158]. Health-related quality of life and physical function have frequently been shown to decrease after hospitalization with AKI [168–170]. Trials surrounding dialysis initiation for AKI, such as STARRT-AKI and AKIKI2 trials, have thus included measures of quality of life and activities of daily living as secondary outcomes [159–161]. To optimize post-ICU physical function, some studies have explored the feasibility and effects of early mobilization while on continuous kidney replacement therapy (CKRT) [171–173]. Future AKI trials should incorporate outcomes that pertain to patients’ and caregivers’ perceptions and experiences during and after hospitalization with AKI.

Outcome | Interaction with AKI | Examples from prior trials
Mechanical ventilation/Ventilator-Free Days | Fluid overload can lead to or exacerbate pulmonary edema, hypoxemia, and decreased lung compliance | Delayed initiation or conservative dialysis strategies do not appear to increase days on mechanical ventilation, with the caveat that protocols allowed for dialysis per clinician judgement [155, 161, 162, 174]
Hypophosphatemia from prolonged CKRT can contribute to neuromuscular weakness | Hypophosphatemia during CKRT was associated with lower chance of successful extubation [175, 176] and fewer ventilator-free days [177]
Cardiac arrhythmias | Electrolyte abnormalities related to AKI can precipitate cardiac arrhythmias | Delayed initiation or conservative dialysis strategies do not appear to increase incidence of cardiac arrhythmias [155, 161]
Hemodynamics | Volume balance and dialysis can impact the need for vasoactive medications | Intermittent KRT modalities were associated with hypotension and/or cardiac arrhythmias during KRT [178]
Accelerated or delayed dialysis initiation strategies were not associated with vasopressor-free days [159, 161]
Delirium/Coma | Accumulation of uremic toxins or neurotoxic medications/metabolites can contribute to delirium/coma | A more-delayed dialysis initiation strategy was associated with a lower chance of awakening among comatose patients with AKI [78]
Infections | Placement of dialysis catheters could increase risk for bloodstream infections | An accelerated dialysis initiation strategy was not associated with catheter-related bloodstream infections [159]
Clearance by dialysis can affect antimicrobial drug levels | Patients on CKRT experience high rates of multi-drug resistant infections [179]
Postoperative bleeding | Accumulation of uremic toxins and fluid overload could contribute to bleeding after surgery | Postoperative AKI has been associated with increased risk of bleeding complications [180, 181]