Clinical trials in Acute Kidney Injury: is the end in sight?

Clinical trials in patients with acute kidney injury (AKI) have been stymied by a lack of consensus on suitable renal-specific endpoints. In a recent analysis, Grams et al. suggest that a sustained reduction in estimated glomerular filtration rate after hospital discharge might be a suitable intermediate endpoint for AKI clinical trials.

Acute kidney injury (AKI) is a common condition in acutely ill patients for which there are currently no effective treatment options other than supportive care. Clinical trials for AKI have been hampered by challenges inherent to their design — including a lack of appropriate renal-specific study endpoints. An analysis performed by Grams and colleagues examined the utility of changes in estimated glomerular filtration rate (eGFR) over 30–360 days as a surrogate end-point for end-stage renal disease (ESRD) in 161,185 patients who underwent major surgery at a Veterans Administration hospital with the aim of identifying potential end-points for trials of AKI prevention and/or treatment.(1)

A total of 19,025 patients in the study cohort developed AKI according to the KDIGO creatinine-based criteria.(2) Creatinine measurements made during routine clinical care were used to estimate glomerular filtration rate (GFR) between 30 and 360 days. Data from prior studies have indicated that a 30% decline in eGFR could be used as a surrogate endpoint for progression of chronic kidney disease (CKD) to ESRD in trials of small sample size, rather than using the traditional endpoint of a doubling of serum creatinine that is more akin to a 50% reduction in eGFR (3). In comparison, little data has been provided thus far that support a threshold of eGFR that would be a meaningful endpoint in the context of AKI. The analyses by Gram et al. focused on the association of 30–40% declines in eGFR with adverse outcomes; patients had a mean age of 64 years, ~96% patients were Caucasian males, and the average pre-hospitalization eGFR was 80 ml/min/1.73m². For this population, a 30% decline in eGFR translates to a 0.3 mg/dl increase in creatinine from a baseline of 1.0mg/dL.

Not surprisingly, those with higher stage AKI were more likely to exhibit a marked decline in eGFR, and those who experienced AKI together with a decline in eGFR had a greater risk of ESRD compared to those who experienced AKI without a decline in eGFR or those who did not have AKI. A greater reduction in eGFR was associated with greater risk of ESRD, regardless of baseline AKI stage or status. Among those who did not exhibit AKI, a small subset of subjects experienced a 30–40% decline in eGFR and were also at increased risk of developing ESRD, although the risk was lower in these individuals compared to those who concomitantly experienced AKI. This increase in ESRD risk was noted as early as 30 days after hospitalization (2.6-fold increased risk) and persisted up to 1 year (7.3-fold increased risk). There was no association between the severity of AKI and risk of ESRD after stratifying by the change in follow-up eGFR. In a mediation analysis, a decline in eGFR explained the majority of the increased risk of ESRD after AKI.

The risk of death was also found to be greater with larger eGFR decline, regardless of baseline AKI stage or status, but the magnitude of the risk of death associated with eGFR decline was lower compared with the risk of ESRD. For example, at 30 days, a 30% decline in eGFR in those who had experienced AKI was associated with a hazard ratio (HR) for ESRD of 5.60 (95% CI 4.06–7.71) and a HR for death of 1.55 (95% CI 1.42–1.69). Finally, the sensitivity and specificity of changes in eGFR decline to predict ESRD were examined. A notable decline in eGFR (40%) was less sensitive and more specific than a decline of 30% in eGFR decline, For example, among those with AKI, an eGFR decline of 40% at 30 days had 27% sensitivity and 92% specificity for ESRD comparared to 41% sensitivity and 85% specificity for an eGFR decline of 30%. Furthermore, the overall sensitivity and positive predictive value of eGFR decline was low. Amongst those with AKI, a 30% decline in eGFR at 180 days had 51% sensitivity and only a 9% positive predictive value for ESRD.

This study was designed following similar analyses that examined CKD progression to ESRD and is the first large study to systematically examine the association between eGFR decline and risk of ESRD after AKI. From their data, the researchers propose that a 30–40% decline in eGFR measured 30–180 days after onset of AKI could be a useful intermediate endpoint for AKI clinical trials. Major strengths of this analysis include the rigorous systematic approach to examining changes in eGFR decline, and the availability of follow-up on all individuals, as this analysis was conducted using data from the Veterans Affairs healthcare system.

Some important caveats to this study are worth noting when assessing these data. First, additional studies are needed to confirm the magnitude of the effect in other populations. Specifically, this study included a predominantly middle-aged, Caucasian, male population; only 12% had CKD because of the original cohort design. As an extreme contrast, one could imagine that if the study population had a baseline eGFR of 20–30 ml/min/1.73m², a 30% decline in eGFR might have very different performance characteristics (in terms of sensitivity, specificity, and positive predictive value) with regards to ESRD prediction. This caveat is relevant to clinical trial design, as many studies to date have selected patients with CKD at increased risk for AKI as part of the target study population.(4–6) Second, because the creatinine measurements were obtained as part of routine clinical care rather than having been systematically collected, those with more frequent measurements are likely somewhat different than those with missing or fewer measurements with respect to overall health and comorbidities.

We consider that additional studies of more diverse, large populations are now needed to confirm and extend these findings. Two studies that may shed additional light on the relationship between eGFR decline and ESRD are the NIDDK-sponsored ASSESS–AKI study (7) and the VA and NIDDK-sponsored PRESERVE study (8). ASSESS–AKI is a parallel, matched cohort study of subjects with and without AKI during an index hospitalization, and the included subjects encompass a range of baseline eGFR values.(7) Both pre-hospitalization baseline eGFR and post-hospital discharge eGFR are being systematically and longitudinally captured in this study. As suggested by Grams et al, an important approach is to analyze the relationship of eGFR decline with ESRD in the context of a clinical trial so that the impact of treatment on both the intermediate outcome of eGFR decline and longer term outcomes can be tested. PRESERVE is a randomized controlled trial of bicarbonate and N-acetylcysteine to prevent contrast nephropathy. The primary endpoint of the study is a 90-day composite of death, dialysis, and persistent decline in renal function (defined as ≥50% rise in creatinine). This definition was chosen based on the association of changes of this magnitude at 90 days with adverse clinical outcomes at 1 year using data from the population of interest.(8) If changes in eGFR of 30–40% are associated with subsequent CKD progression in PRESERVE, this study would serve as even stronger evidence that such changes could be considered as clinical trial endpoints.

These latest data from Grams et al highlight some of the challenges inherent to studies of AKI therapeutics — perhaps most importantly, the difference in magnitude of the competing risk of death, which is >50-fold greater than the risk of ESRD in this population. During the follow-up period, 43,668 deaths occurred, compared to 787 cases of ESRD, and it would now be of interest to understand the causes of death in this population and whether these deaths could be attributed to complications of AKI. Although death is arguably the most important patient-centered outcome, an AKI therapeutic that does not decrease the risk of death might have limited power to demonstrate benefit even in a large clinical trial because of the magnitude of the competing risk. A number of other challenges to trial design exist and have been discussed in the context of critical care, including identification of study subjects sufficiently early that they may benefit from therapy, the use of therapies that target only a single pathophysiologic process, and the relative heterogeneity of study populations (9, 10) Nonetheless, the current study represents an important step towards developing appropriate renal-based study endpoints for AKI therapeutics trials that may be particularly useful in peri-operative AKI or contrast-induced AKI, where mortality is substantial but lower than in intensive care unit or sepsis-based populations. Future studies should attempt to confirm and extend these findings in other populations, including higher risk populations with a higher severity of illness or a greater burden of CKD, as well as populations that more closely reflect the overall US population.