The Use of Kidney Disease Improving Global Outcomes AKI Definitions in Retrospective AKI Research: CON

Introduction

AKI is defined by abrupt changes in serum creatinine (SCr) and reduction in urine volume. The definition of AKI has evolved over the years, with the most recent Kidney Disease Improving Global Outcomes (KDIGO) definition using lower SCr thresholds than prior versions. The increased sensitivity in the definition was primarily supported by the prognostic evidence from large epidemiologic studies demonstrating an increase in short- and long-term mortality and resource utilization with small increases in SCr. However, this definition has limitations when applied to retrospective AKI research and clinical care.

A Simple Numerical Definition Discourages Deep Investigation

The current definition of AKI is objective and comprehensible, but lacks clinical nuance and discourages medical reasoning and investigations. In decades past, the definition incorporated some aspect of pathophysiology, even if elementary—AKI was, at a minimum, broken down into prerenal, intrinsic, and postrenal etiologies—but this has since been neglected. Research studies on AKI now commonly focus on KDIGO thresholds rather than investigating the pathophysiology and causes of AKI (Figure 1). This limits research on other important aspects of AKI; for example, long-term outcomes after AKI may vary by etiology of AKI, but focusing on numbers alone makes deeper investigation challenging. SCr and urine output (UOP) may be better suited as screening tests for AKI, but they have little phenotypic potential for meaningful interpretations from retrospective studies.

Figure 1.

SCr is not specific for AKI because of different etiologies, which have variable effects associated with management and short- and long-term outcomes. Dotted lines represent causes of increased SCr that tend to be reversible with treatment or removal of the offending agent. ADHF, acute decompensated heart failure; AIN, acute interstitial nephritis; ATI, acute tubular injury; HRS, hepatorenal syndrome; RAAS, renin-angiotensin-aldosterone system; SCr, serum creatinine; SGLT2i, sodium-glucose cotransporter-2 inhibitors; TMA, thrombotic microangiopathy.

Increases in Creatinine Can Be Expected and Even Beneficial

Many causes of increased SCr are not associated with impaired renal function or kidney parenchymal injury despite being defined by KDIGO as AKI. Table 1 outlines various causes of increased SCr, but here we focus on a few essential categories because of the relevance of these conditions in clinical practice.

Table 1.

Some causes of increased serum creatinine that are not indicative of kidney failure

Cause of Increased SCr	Example(s)	Pathophysiology	Effect on Retrospective Research
Concentration/dilution	Diuresis	Reduction of blood volume; intrarenal vascular hemodynamics	Not captured; adds heterogeneity; may make results and associations with outcomes invalid
Creatine/protein ingestion	Cooked meat Creatinine supplementation Protein intake	Increased reactants in creatinine formation	Data in diet usually not available; misclassification of AKI events
Inhibition of secretion	H2 histamine blockers Pyrimethamine Salicylates Trimethoprim	Competitive inhibition of creatinine secretion in renal tubules	May be available if medications reliably documented, but likely under-reported; contributes to misguided interpretation of SCr changes
RAAS inhibition	ACEIs ARBs ARNIs	Preferential dilation of renal efferent arteriole	Overestimates AKI in many settings; prognostic estimates are likely underestimates as long-term effects are beneficial
Tubuloglomerular feedback	SGLT2i	Decreased sodium reabsorption and tubuloglomerular feedback	Overestimates AKI in many settings; prevents adequate study of medication effects due to discontinuation

ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin-receptor blockers; ARNIs, angiotensin receptor/neprilysin inhibitors; DM, diabetes mellitus; RAAS, renin-angiotensin-aldosterone system; SCr, serum creatinine; SGLT2i, sodium-glucose cotransporter-2 inhibitor.

The first example of changes in SCr as a manifestation of a disease process is in heart failure, best observed in patients with acute decompensated heart failure (ADHF). Patients with ADHF often have altered SCr from their baseline on admission. SCr higher than baseline may represent true AKI due to hemodynamic effects of ADHF itself, mainly secondary to venous congestion, and diuresis tends to improve AKI. However, at times, SCr on presentation may be lower than baseline because of increased plasma volume or hemodilution;¹ therefore, increases in SCr with decongestion can sometimes be attributed to appropriate correction of blood volume with removal of excess fluid. Furthermore, baseline SCr values can be challenging to determine for these same reasons. In addition, findings from multiple studies have suggested that increased SCr in ADHF are not associated with the same adverse outcomes as true AKI. For example, a post hoc analysis of the Diuretic Optimization Strategies Evaluation trial, which examined diuretic strategies in ADHF admissions, showed that increased SCr was associated with improved outcomes.² The fact that SCr changes in ADHF have clear physiologic underpinnings and can even be beneficial suggests that SCr increases of ≥0.3 mg/dl should not always be considered AKI; this clinical state has been termed “permissive hypercreatinemia,” with recommendations to continue diuretics.³ Continuing diuresis to achieve optimal decongestion should be prioritized over avoiding SCr increases.

Further complicating SCr changes in ADHF are medications used in treatment of heart failure, such as renin-angiotensin-aldosterone system inhibitors (RAASis) and sodium-glucose cotransporter-2 inhibitors (SGLT2is), which are often held, restarted, or adjusted in some way during ADHF admissions. However, these medications have implications beyond ADHF because they are used to manage many other conditions, including diabetes, hypertension, and CKD. RAASis cause predictable increases in SCr because they preferentially vasodilate the renal efferent arteriole; however, it is debatable whether this should be considered AKI because the underlying physiology that causes decreased GFR is due to decreased glomerular hemodynamic pressure. In addition, studies have shown that the initial rise in SCr after starting RAASis predicts slower long-term loss of kidney function,⁴ although others have found some evidence of adverse outcomes associated with larger changes in SCr after RAASi initiation.⁵

SGLT2is cause increased SCr and decreased GFR through tubuloglomerular feedback mechanisms leading to afferent arteriole constriction, but similarly have various cardiovascular and renal benefits, and have even been shown to lower the risk of AKI in some populations.⁶ Diuresis, RAASi, and SGLT2i are thus three causes of SCr increases that should not always be interpreted as AKI and are crucial to recognize to provide optimal, life-prolonging care for patients with and without kidney disease. Nuances of medical decision making in heart failure—as-needed diuretic administration and dosing, continuation, or discontinuation of RAASis and SGLT2is—are not adequately captured in retrospective studies. This makes it challenging to interpret results, as summarized in Table 1. Notably, much of the data that informed the current definition of AKI were from before the advent of SGLT2is and when there was no evidence for restarting RAASis in the hospital—another tangible limitation of this definition.

Small Changes May Not Be Clinically Relevant

The current SCr-based definition of AKI, defined by an abrupt rise of just 0.3 mg/dl or 50% increase from baseline, is largely based on prognostic studies suggesting that changes this small may be associated with increased mortality and other negative outcomes.⁷ It is not surprising that small changes in SCr may be associated with worse outcomes—whether with SCr, BP, erythrocyte sedimentation rate, or other clinical biomarkers, disruptions to homeostasis are typically not favored by the organism. However, this definition is oversimplified, and clinical trials have questioned whether these small changes are clinically actionable and clinically relevant. For instance, one trial showed that an electronic alert system that informed providers about stage 1 AKI did not improve clinical outcomes among hospitalized patients.⁸ Furthermore, a 2023 multicenter cohort study showed that stage 1 or 2 AKI does not contribute to CKD progression.⁹ These findings suggest that the sensitive definition of AKI has limited benefit in clinical practice and management of patients. In clinical research, this could lead to misclassification, especially in clinical trials, and, therefore, an underestimation of the true morbidity and mortality of AKI compared with if more stringent criteria were used.

The Role of UOP

UOP is included in KDIGO definitions for AKI, and using UOP criteria instead of solely SCr to identify AKI has been shown to improve detection of AKI and the association of AKI with mortality.¹⁰ However, this is less commonly used in retrospective studies than SCr, largely because of a number of logistic barriers—for example, it is often inaccurately and unreliably recorded, especially outside of intensive care units and in patients without indwelling catheters, and it is influenced by factors such as diuretic administration.

Other Tools in the Toolkit

If the current definition for AKI lacks nuance, has poor specificity for AKI, and is limited by logistic challenges, how can we identify true AKI? Two underutilized tools at our disposal include biomarkers and urine sediment. Many blood and urinary biomarkers have been identified that can detect AKI earlier, and with better sensitivity, than SCr; some also offer insight into mechanisms, site, and etiology of AKI that SCr cannot provide.¹¹ These can serve as damage biomarkers, complementing the functional biomarkers of SCr and UOP. Urine sediment is also useful as an adjunct to our current repertoire, mainly because of its ability to elucidate etiology of AKI, such as with granular casts in acute tubular injury. The exclusion of biomarkers and urine sediment in KDIGO definitions of AKI discourages their routine collection, making the data unavailable for analysis in retrospective studies and limiting our understanding of AKI in existing research.

The definition of AKI has evolved to a point that it is now routinely recognized but perhaps more poorly understood. The current KDIGO definition is sensitive for AKI, but perhaps too sensitive—stage 1 AKI, in particular, has limited clinical relevance by itself while stages 2 and 3 may be of greater utility. Furthermore, the definition adds noise and heterogeneity in research studies, and it discourages both investigation into the cause of AKI and acceptance of SCr increases when they are expected and even favorable. In many of these cases, AKI is probably an inappropriate term, despite meeting KDIGO criteria, because there is no parenchymal damage. Retrospective AKI research is limited by using these simple numerical definitions for AKI without appreciation of nuance. Other tools such as serum and urine biomarkers and urine sediment should be considered in future definitions of AKI to enhance our understanding of kidney disease and evidence from research studies moving forward.

Disclosures

C.R. Parikh reports the following: Consultancy: Genfit Biopharmaceutical Company; and Research Funding: National Heart, Lung and Blood Institute (NHLBI) and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). C.L. Tamargo reports the following: Ownership Interest: Amazon, Amedisys, AstraZeneca, Ayro, General Motors, Microsoft, Uber, and Vulcan.

Funding

C.R. Parikh: NIDDK (U01DK114866, U01DK106962, U01DK129984, and R01DK093770).

Author Contributions

Conceptualization: Chirag R. Parikh, Christina L. Tamargo.

Supervision: Chirag R. Parikh.

Visualization: Chirag R. Parikh, Christina L. Tamargo.

Writing – original draft: Christina L. Tamargo.

Writing – review & editing: Chirag R. Parikh, Christina L. Tamargo.