Estimating Glomerular Filtration Rate in Patients With Acute Kidney Injury

Yosu Luque; Eric Rondeau


Nephrol Dial Transplant. 2020;35(11):1834-1836. 

The 2012 Kidney Disease: Improving Global Outcomes consensus definition of acute kidney injury (AKI) has been very useful to classify patients and to compare the different studies of AKI.[1] However, this definition is based on urine output and serum creatinine levels that are suboptimal markers of renal function. It is known that serum creatinine depends not only on renal function, but also on muscular body mass, protein intake, hydration and medication. The other parameter used to define AKI stages is urine flow, which lacks sensitivity and specificity.[2]

During steady-state chronic kidney diseases (CKDs), it has been shown that estimated glomerular filtration rate (eGFR) by different creatinine-based formulae can be used in current practice to control the evolution of CKD and to predict the time for terminal renal failure and dialysis.[3] During AKI, it is acknowledged that the patient is not stable and that there may be a marked variation in blood pressure and hydration from one hour to another. It appears thus that serum creatinine may vary independent of changes in GFR. In addition, there is a delay between the acute decrease in GFR and the progressive accumulation of creatinine, and so any evaluation of GFR based on serum creatinine in the acute phase is erroneous. That is why the classic Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) formulae that are used in CKD cannot be used in AKI, as they overestimate GFR in this setting.[4]

In the present issue of NDT, Pelletier et al. used a prospective study performed in three tertiary care centres in Canada to examine the performance of two different methods of non-steady GFR estimation [the Jelliffe equation and kinetic eGFR (KeGFR)] when compared with a direct measurement of GFR by a radioisotopic method using the plasma clearance of technetium-99m diethylenetriaminepentaacetic acid (99mTc-DTPA). The Jelliffe equation takes into account the serum creatinine, as well as the sex, age and body surface area of the patient, and was developed for patients with unstable renal function without the need for urine sampling.[5] First described in 2013, the KeGFR equation allows for acute variations of serum creatinine between two time points, as well as the initial creatinine content, the distribution volume and the creatinine production rate.[6] Recent studies have shown that the KeGFR equation has better predictive performances for severe AKI and renal replacement therapy initiation than the MDRD equation and that it could also be predictive for AKI recovery.[7,8] It has also been suggested that it could be helpful in drug dosage adaptation.[9] However, this is the first study comparing these formulae to direct GFR measurement in AKI.

The authors used the KeGFR and Jelliffe equationd to estimate GFR in 119 adult AKI patients, mostly from the intensive care unit (ICU) (63%). They included a majority of moderate AKI patients (71% of Stage 1 AKI), and oliguric patients were excluded. GFR radioisotopic measurement was performed 24 h after inclusion in the majority of cases with a single 99mTc-DTPA intravenous bolus.

This is a highly important study, as it is the first to evaluate these equations compared with a validated GFR measurement in the AKI setting. Previously they have only been compared with steady-state formulae, such as the MDRD, which are not applicable in AKI. The results show that the Jelliffe and KeGFR equations both underestimate GFR but show a good correlation with measured GFR.

The authors chose as a reference a validated radioisotopic GFR measurement with 99mTc-DTPA. This technique, which is currently widely used, correlates well with the gold standard method of inulin clearance, which is now performed less frequently due to anaphylactic issues, costs and feasibility. The authors' chosen method uses the slope of decrease of plasma 99mTc-DTPA, assuming that there is no other way of elimination than the kidney and that all the elimination is related to GFR with no tubular secretion. However, any significant fluid infusion or fluid losses during the time period of measurement (4 h) can either increase or decrease the slope, and thus the apparent GFR. Other fluorometric methods have been developed, based on the same assumption that the plasma clearance reflects renal clearance, which itself reflects GFR. It has to be said, however, that it is not routine clinical practice and it is generally performed only in the context of clinical research.[4]

It is not possible to assume that the results are applicable to all stages of AKI, as oliguric AKI was excluded, with the majority of patients having Stage 1 AKI at inclusion.

The lack of agreement that the equations demonstrate with the measured GFR could be explained in part by the fact that >50% of patients had a decreasing serum creatinine when the GFR measurement was performed (sometimes 24 h after inclusion). As renal function during AKI is rapidly changing, the radioisotopic method and creatinine-based equations might not have evaluated the same time point, thus creating these discrepancies. For the moment, the considerable lack of agreement that these equations have with measured GFR precludes the use of drug dosage adaptation.

In patients who became anuric during a septic shock, there is no question that GFR is considerably reduced, if not completely suppressed. Any accurate evaluation of GFR in these cases seems futile. In patients who still produce urine during AKI, it may be useful to evaluate GFR to adapt drug dosage and prevent a superimposed nephrotoxic injury. In addition, it has been proposed that subclinical AKI, without a significant change in serum creatinine, could be detected with an accurate direct measurement of GFR, and that these subclinical AKIs may have an impact on patient prognosis, in both the short and long term.[10]

New biomarkers are available but, except for some functional markers such as cystatin C or proenkephalin,[11,12] they mostly detect tissue injury rather than altered function, performing better in severe AKI. Today, their role in clinical management at the bedside is limited. Direct measurement of GFR could be useful to classify the less severe forms of AKI, with no or limited increase in serum creatinine. As previously stated, the initial phase of AKI is critical because it is the most difficult to recognize, with serum creatinine showing little or no variation during the first hours after injury, and because it is most likely the point when therapeutic intervention would be successful. Non-invasive biomarkers, such as neutrophil gelatinase-associated lipocalin and kidney injury molecule 1 or the combined measurements of tissue inhibitor of metalloproteinase 2, a metalloproteinase inhibitor and insulin growth factor-binding protein 7, a binding protein for insulin growth factor 1, to be measured in the blood or urine were developed to detect early kidney injury.[13] These molecules are induced in the kidney at the early phase of AKI and reflect changes in the metabolic profiles of tubular epithelial cells. However, they do not reflect functional alterations, that is, a decrease in GFR. Thus, at the early phase of a potential renal injury, such as septic or haemorrhagic shock, or after administration of a nephrotoxic drug, it may be important to precisely measure GFR if any intervention is planned. In the future, if some compounds have to be tested, it may be after identification of patients with developing AKI.

The other period of time when measurement of GFR may be used during AKI is when recovery is expected. Again, direct measurement of GFR is more accurate than creatinine-based estimations, especially in the patient with AKI in the ICU. Catabolism of muscle mass, low protein intake and inflammation, which are frequently encountered in these patients, significantly change the muscle mass and affect creatinine production and serum creatinine level. Although we do not yet have an efficient therapeutic intervention that would shorten the period of recovery, we very much hope that in the future such drugs will be available and appropriately used in patients at the very beginning of AKI recovery.

In conclusion, the authors show that it is highly challenging to estimate GFR in patients with AKI, as the available equations underestimate the actual GFR and correspond unsatisfactorily with measured GFR. The discrepancy between eGFR and measured GFR may be related to several factors, either intrinsic to the formula, which thus needs to be improved and validated over multiple populations and degrees of AKI, or because each is creatinine-based and serum creatinine depends not only on GFR, but also on muscle mass and metabolism, fluid balance and nutritional support. Previous studies have shown that GFR estimating equations could be predictive of severe renal failure, but Pelletier et al. have demonstrated that they cannot precisely estimate GFR in order to, for example, adapt drug dosage. In the future, we hope that a combination of tissue injury and functional biomarkers, or GFR estimation formulae, could better define a global phenotype of AKI, including its severity, prognosis and recovery (Figure 1).

Figure 1.

AKI phenotyping.

Today, the direct measurement of GFR in AKI is performed in clinical research, not in routine practice, but in the future, if technically feasible, and if a therapeutic intervention becomes available, it may be required at the early infra-clinic phase.