Testing New Biomarkers for Acute Kidney Injury: Association, Prediction, and Intervention

In Spring 2004, the Board of Advisors and the Council of the American Society of Nephrology conducted retreats to steer research priorities in an era of limited resources. These retreats included leaders in the fields of nephrology, critical care medicine, and cardiology, along with representation from the National Institutes of Health and the US Food and Drug Administration. In the area of acute kidney injury (AKI), the group criticized nephrologists' widespread dependence upon serum creatinine, a poor marker of true AKI, and argued that this reliance was stifling therapeutic progress. The group ultimately recommended that the highest research priority be the standardization and/or discovery of new biomarkers of AKI.1 The AKI research community responded to this recommendation, and published more than 50 articles on over 20 unique biomarkers of AKI in the last 7 years.

In this issue of the American Journal of Kidney Diseases, Haase et al publish a systematic review and meta-analysis of the most popular AKI biomarker to date, neutrophil gelatinase–associated lipocalin (NGAL). The authors identified 13 studies using urine NGAL and 9 studies using serum NGAL.2 Global enthusiasm for new biomarker development is evident from the nearly equal representation of studies from the United States and abroad. NGAL was also tested across a broad range of clinical settings, in both adults and children. In their review, Haase et al conclude that urine and serum NGAL are useful as early markers of AKI, and that NGAL levels are also associated with clinical outcomes, such as the requirement of dialysis or mortality.

Important limitations of existing knowledge also emerged from this analysis. First, each of the studies was conducted in a single center. Second, the setting of transient elevations in serum creatinine after surgery was most frequently studied, with a relative deficiency of more complex forms of AKI, or AKI among patients with multiple comorbid conditions such as diabetes or older age. Third, the number of patients studied was relatively small; the number of patients with AKI ranged from 3 to 99 across studies, and the number who died or needed short-term dialysis was even lower (Table 1). Fourth, the reference standard used to define AKI was usually a 50% increase in serum creatinine. While uniformly defined across the studies, this still has all of the limitations of a surrogate outcome. Fifth, a majority of studies used a “home-grown” enzyme-linked immunosorbent assay (ELISA) to measure NGAL. The reference range in the non-AKI control population varied widely, from <10 ng/mL to <550 ng/mL, across these studies. Sixth, there is the potential for publication bias with negative results of novel biomarker studies, as investigators may not write up the results or change the focus of their investigation. Negative studies also are difficult to publish as abstracts in national meetings or as articles in scientific journals.

Table 1. Summary for the 3 End Points Examined in the Systematic Review

			No. of Patients Reaching End Point		Percentage of Patients Reaching End Point
	End Point	No. of Studies	Total	Range in individual studies	All studies	Range in individual studies
	Acute kidney injury	19	487	3-99	19.2	4.3-59.1
	Short-term dialysis	9	84	4-22	4.3	1.9-15.4
	Death	6	88	3-52	5.4	1.3-17.3

Source: Haase et al.2

The review by Haase et al allows us to reflect on where we are as a community in AKI biomarker testing and guides our next research priorities. First, the validation of biomarker results in large multicenter studies is essential. Through their collaboration, Haase et al show the importance of networks of clinical investigators working together to advance the field. Second, it is vital that large studies demonstrate the association between biomarkers and “harder” outcomes, such as need for short-term dialysis, cardiovascular events, and death. Third, we should understand the clinical outcomes of individuals who do not meet the AKI definition by serum creatinine level, but who test positive for a putative AKI biomarker. Poor outcomes in this subset of patients will be strong evidence of biomarkers showing appropriate sensitivity for the detection of AKI. Fourth, we need to study more patients with chronic kidney disease, a group of patients that frequently develops AKI and in whom new biomarkers such as NGAL may be less effective.3 Fifth, we need to move beyond home-grown ELISAs in order to standardize clinical platforms and reference laboratories so investigators are using the same methodology. To enhance such standardization, the renal community may wish to adopt solutions used by other groups of biomarker investigators, such as the National Cancer Institute–sponsored Early Diagnosis Research Network (EDRN) (www.edrn.nci.nih.gov). Sixth, we need to look beyond association studies. The 19 studies summarized by Haase et al successfully demonstrate that NGAL can separate patients according to likelihood of developing AKI and also possibly by need for short-term dialysis. We now need to know the additional contribution NGAL and other biomarkers make to clinical risk prediction, ie, whether the new marker adds significant incremental prognostic information to a model that includes the established risk markers. Finally, we need to use these new biomarkers to improve patient outcomes. Proving that new biomarkers are reliable surrogate outcomes (ie, showing that decrements in biomarker values correlate with important reductions in definitive outcomes) in large, multicenter, randomized clinical trials would advance the field. In addition, these biomarkers can be used as inclusion criteria for randomized trials to show an identifiable group of patients who benefit from early treatment of AKI.

By following this trajectory of testing, we as a community can move beyond the conduction of association studies to studies that prove the utility of new biomarkers in prediction and ultimately in intervention.