Should the novel biomarkers be incorporated in future definitions of acute kidney injury?

Nephrologists widely accept that glomerular filtration rate (GFR) is the best overall index of kidney function in health and disease.

Acute kidney injury (AKI) encompasses a loose collection of syndromes characterised by a sudden (hours or 2–5 days) decrease in GFR. Although this definition focuses on the decline in GFR, molecular and cellular damage may occur at the tubular level before the decline in GFR occurs. This tubular damage may act as a predictor of the later decline in GFR. The recent molecular approach to AKI has led to the recognition that many separate subtypes of tubular damage exist, each with distinct patterning of molecular markers [1].

It is difficult to measure abrupt changes in GFR and therefore international criteria for describing the presence and severity of AKI are based on changes in serum creatinine (sCr) and/or changes in urinary output (UO), used as surrogates of kidney function (i.e. GFR). It cannot be denied that current AKI criteria [e.g. those from Kidney Disease: Improving Global Outcomes (KDIGO)] [2] have brought harmonization in the definitions of AKI and better understanding of its epidemiology and have facilitated efforts to address current variations in the delivery of AKI care. However, sCr values are influenced by many non-renal determinants and both its generation rate and the distribution volume of creatinine are not stable in critically ill patients. A delay in diagnosis and a wrong judgement of renal recovery can easily occur in patients with decreased creatinine generation, e.g. in sepsis [3] or in case of dilution due to fluid overload [4]. Also, reliance on the UO criterion when expressed in mL/kg/h in obese patients is associated with higher sensitivity for diagnosing AKI, but decreases the specificity, potentially leading to AKI over-diagnosis [5, 6].

The definition and staging of AKI as proposed by (Risk, Injury, Failure, Loss of kidney function, and End-stage kidney disease) RIFLE, AKIN (Acute Kidney Injury Network) and KDIGO has been criticized because it fails to capture important clinical differences across and within stages of AKI [7, 8]. Despite the clear recommendations of the KDIGO guidelines that in every AKI patient the cause and aetiology of AKI and possibly the underlying pathophysiology of AKI should be explored, some authorities feel that not enough attention has been paid to these important aspects. In 2006 we proposed that additional parameters such as the most important causal factors responsible for AKI and, importantly, pre-existing kidney function should be added to classification systems of AKI [9].

We believe it is against this background that the review by Koyner et al. [10] published in this journal, should be read and interpreted. The authors suggest that in a clinical context of risk of AKI, use of novel biomarkers, presumably reflecting underlying tubular damage, may improve individual patient care when integrated with traditional biomarkers such as sCr and UO. The review further discusses the concept of ‘renal angina’, the diagnostic role of the furosemide test and the role of electronic warning systems and proposes to integrate the use of these biomarkers in a ‘personalized’ approach to the patient at risk of AKI.

To date, the two most extensively studied biomarkers are neutrophil gelatinase-associated lipocalin (NGAL) and the product of tissue inhibitor of metalloproteinase-2 (TIMP-2) and insulin-like growth factor-binding protein 7 (IGFBP7). Only these biomarkers are reimbursed in selected countries and TIMP-2-IGFBP7 is only approved for use in critically ill/high-risk patients [11]. These limitations explain why Koyner et al. [10] focus mainly on AKI in intensive care unit (ICU) patients and on the role of the two above-mentioned biomarkers; this obviously creates some unbalance in the discussion.

It is beyond the scope of this editorial comment to review all the suggestions made by Koyner et al. [10] on how these novel biomarkers could be implemented in daily clinical care.

We therefore decided to focus on one particular item, i.e. the suggestion to incorporate the biomarkers in the identification of different ‘phenotypes’ of AKI and, as an extension, to incorporate them in the definition of AKI by ‘creating’ phenotypes like ‘subclinical’ AKI and acute kidney disease (AKD). We have recently published a number of reservations regarding the ‘subclinical’ AKI phenotype [12]. ‘Subclinical AKI’ refers to a situation of biomarker positivity, presumably reflecting cell stress and damage but without functional repercussion as evidenced by sCr increase or oliguria [13].

It is well known that a ‘normal’ sCr does not exclude a decline in GFR or subclinical structural damage, while a normal or even a supranormal GFR also does not exclude kidney functional changes or subclinical structural damage. When estimating renal function, particularly in AKI, the concept of renal functional reserve (RFR) should be taken into account. RFR is defined as the increase in GFR above basal fasting values, which can be activated by stress, oral protein load, amino acid, dopamine or glucagon infusion. The fact that in ‘subclinical’ AKI the sCr is not yet increased does not exclude that in many of these cases the true GFR may have declined even by 50%. In such a situation, many cases of ‘subclinical AKI’ may be an ‘artefact’, reflecting the ‘weakness’ of sCr to measure true GFR. It is thus not surprising that many patients with so-called ‘subclinical AKI’ have a prognosis that is not much different from the outcome of ‘intrinsic’ or ‘true’ AKI [14]. Alternatively, it can also merely indicate that these patients are severely ill, explaining both higher marker levels (influenced by severity of illness and comorbidities) and higher mortality rates.

The other requirement for the label ‘subclinical’ AKI is the assumption that the increased urinary excretion of a biomarker is a true reflection of tubular damage.

An ideal biomarker should be measurable from non-invasive sources and the test should be easy to perform with rapid turnover and high reliability. The biomarker should be organ specific, its levels should correlate with the severity of damage and it should be able to pick up AKI in a stage when functional damage, at least estimated by using sCr, is not yet detectable.

The results of recent molecular studies, already mentioned above, make the clinical interpretation of the excretion patterns of novel biomarkers quite confusing. However, these results suggest that the molecular and cellular responses that accompany different stimuli, which in the past were considered part of the same disease spectrum, do not indicate that these diseases are a continuum or reflect a single common final pathway [1].

Different biomarkers have different windows of opportunitiy, which makes it difficult to define optimal timing of their analysis after renal injury, especially when the exact moment of the renal insult is unknown [15]. Based on the requirements that define a reliable biomarker, it is far from evident that the presently available biomarkers are ready for clinical implementation and the fact that these markers can be influenced by chronic comorbidities and acute severity of illness, independent of AKI, and thus be false positive, should not be disregarded [16].

For example, NGAL not only reflects tubular damage, but is also released by systemic inflammation. In sepsis, there is a strong correlation between serum and urinary NGAL (uNGAL) and uNGAL levels, indicating that increased levels of uNGAL can be the result of overspill from the systemic circulation, blurring the discriminative value of NGAL as a biomarker for AKI [17]. On the other hand, the source of the increased urinary excretion of the cell cycle biomarkers in experimental AKI is unclear since their genes and their protein products may be expressed ubiquitously within as well as outside the kidney [18]. These experimental results suggested that than rather increased glomerular damage, decreased tubular reabsorption and urinary TIMP2 × IGFBP7 leakage seem to be the most likely mechanisms contributing to the elevated urinary levels.

In addition, uncertainty about several crucial points cannot be neglected, such as whether or not to normalize biomarkers for urinary flow rate [19], what the normal age and gender-related values in the no-AKI population are, how to standardize assays and how these biomarkers perform in chronic kidney disease (CKD), which is a prominent risk factor for AKI and vice versa.

Three studies used biomarkers to enrich the study population in clinical investigations. In the study by Wald et al. [20], NGAL levels were not different between patients treated with RRT and those who recovered spontaneously in the standard group. Meersch et al. [21] used the cell cycle inhibitors to identify high-risk patients and compare the use of standard care versus application of the KDIGO care bundle in the prevention of AKI after cardiac surgery. Although there was a decrease in AKI incidence and biomarker levels in the intervention group, there was no difference in mortality or need for RRT. Göcze et al. [22] studied the effect of the same intervention in patients who underwent major abdominal surgery and had a positive screening for cell cycle inhibitors. Although the authors also found a benefit in the intervention group, one is left to wonder whether this result indicates the clinical usefulness of the markers or the beneficial effect of applying a care bundle in patients at risk for AKI [23]. Overall, one could assume that all ICU patients and/or patients undergoing major surgery are at risk to develop AKI and applying the KDIGO care bundle should be the standard of care in all patients instead of only those with a positive biomarker result.

In our opinion ‘overoptimistic’ belief in biomarkers also partly explains the creation of the concept of AKD. AKD was recognized as a potential antecedent to CKD. The conceptual overlap among AKI, AKD and CKD was first illustrated in the 2011 AKI guideline [2]. AKD was originally ‘designed’ to name the no man’s land between sustained AKI (which is arbitrarily set at 7 days maximum) and CKD (which can only be diagnosed at least 90 days after AKI). Until recently, the magnitude of the problem and its consequences were not known. Emerging evidence has highlighted that AKD is common, nearly three times more prevalent than AK,I and AKI is associated with increased risks of death and development or progression of CKD [24].

However, AKD also names the scenario where patients after AKI are no longer fulfilling the KDIGO AKI criteria but do respond to the below-threshold criteria and/or have increased damage marker levels without an increase in sCr or a decrease in UO [25]. Also, patients who never experienced AKI, at least based on the actual criteria, can, according to the proposed definition, be diagnosed as having AKD.

As a consequence, the concept of AKD can create confusion and will inevitably classify the majority of patients as being diseased. It refers to a non-homogeneous group of patients including patients with e.g. 15 days of prolonged AKI according to the KDIGO criteria and patients who never experienced AKI according to those criteria but display 15 days of biomarker increase. We believe that in the near future an in-depth reflection and discussion on the definition of AKD and subclinical AKI is necessary before these concepts can be proposed to the clinical community.

In conclusion, although the biomarker research has contributed to increased attention to and reflection on the problem of AKI, addition of the new biomarkers into the present classification method and ‘creating’ new subtypes of AKI is in our view presently not yet justified based on the majority of current evidence. For the moment, this subtyping only complicates the diagnostic landscape without adding much benefit.

The acceptance or rejection of the use of new biomarkers in the field of AKI is an illustration of what Thomas Kelly, in his famous article ‘Dissent, Dogmatism and Belief Polarization’, published in the Journal of Philosophy [26]. He described the phenomenon called ‘belief polarization’, by which exposure to the same evidence, far from bringing those who have different opinions closer, usually makes the disagreement between them more pronounced, as we are more demanding with anything that contradicts our belief and more permissive with what favours our own point of view.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflict of interest related to this article. This work has not been published previously in whole or part.

(See related article by Koyner et al. The impact of biomarkers of acute kidney injury on individual patient care. Nephrol Dial Transplant 2020; 35: 1295–1305)

REFERENCES