Presentation Summary

Written by Jasna Trbojevic-Stankovic
Reviewed by Esteban Porrini

Accurate measurement of glomerular filtration rate as an indicator of renal function is of crucial importance because clinical decision-making at many occasions depend on the results obtained. In routine clinical practice, renal function is commonly evaluated with formulas that provide estimations of glomerular filtration rate (eGFR). These equations are mathematical models relying on the levels of endogenous markers, such as serum creatinine or cystatin-C, and variables such as weight, height, and gender. The reliability of such formulas is still ambiguous, putting clinicians in front of a significant dilemma whether rely on them or not, and to which extent. The error of eGFR calculated by any of the currently available equation is common and wide. The average deviation from the actual level of renal function is 30%. This may lead to misstating of chronic kidney disease in nearly one third of the cases, and failure to detect a substantial number of cases with hyperfiltration (1).

What do we know today about the error of formulas that estimate GFR?
Inulin clearance is considered the gold standard for obtaining mGFR, however it is expensive and difficult to keep in solution. In the last years, there has been a noticeable trend toward iohexol plasma clearance, particularly in Europe. The iohexol technique is sensitive, inexpensive, and practical, since it does not necessarily require 24h urine collection. Thus, a recent study analyzed the level of agreement between eGFR determined by creatinine and cystatin C–based formulas, and GFR measured using plasma clearance of iohexol (Table 1).

Table 1. Analysis of agreement between GFR estimated with creatinine and cystatin C–based formulas and measured by plasma clearance of iohexol in a representative group of 14 subject (1, 2)

Cases 1 and 2 had measured GFR of 17 mL/min, and almost all formulas underestimated (Case 1) or overestimated the actual GFR (Case 2). Cases 5 and 6 had similar mGFR values of 48 mL/min and 49 mL/min respectively, whereas formulas either overestimated, as in Case 1, or underestimated, as in Case 2, eGFR. Thus, the eGFR error was random and unpredictable, suggesting that current estimation formulas should be used and interpreted with caution (1).
According to Porrini et al. the average error of the currently available formulas either based on creatinine or cystatin-c to estimate GFR is 30% or more, and, even more disturbing, it shows no consistency whatsoever (3). Such large aberration way overpass the acceptable level of 10% and cause incorrect staging of chronic renal disease in as many as 30 to 60% of patients (Figure 1).

Figure 1. The average error of currently available formulas to estimate GFR (2, 3)

Evidence of the errors of estimated GFR in patients with diabetes
The utility of eGFR in type 2 diabetes is still unclear. Several studies evaluated the agreement between eGFR and mGFR in this group of patients yielding contradictory results (1). Luis-Lima et al. analyzed a large cohort with different underlying renal diseases, normo-, micro- and macroalbuminuria, and mGFR ranging from 8.5 to 180.6 mL/min. The authors also estimated GFR with 61 different formulas based on creatinine and/or cystatin-C levels. The error of eGFR calculated with any equation was common and random, averaged 40% of real renal function, and was larger in patients with mGFR below 60 mL/min. In addition, nearly 30% of the individuals were misclassified in chronic renal disease stages, and 25% of those with hyperfiltration were not diagnosed (1). Thus, the currently available formula to estimate GFR may be considered unreliable in type 2 diabetes.

Besides the actual GFR value, its decline also cannot be accurately estimated in type 2 diabetes (4). In a large study involving 600 patients with type 2 diabetes, Gaspari et al. evaluated the agreement between GFR measured by different methods at baseline, month 6, and long-term follow-up. The results confirmed that both the simplified MDRD equation and the CKD-EPI formula underestimate mGFR in diabetic subjects, and that this limitation must be extended to other 13 commonly used formulas implemented over the past fifty years (4). The most probable cause for these discrepancies is the fact that creatinine and cystatin-C are not accurate enough markers of renal function.

What can we do to improve GFR assessment in everyday practice?
Whenever the clinical situation requires meticulous evaluation of renal function mGFR should be obtained. As an example of good practice, at the University of La Laguna on Tenerife the Laboratory of Renal Function was founded to validate the plasma clearance of iohexol for clinical purposes ( The method used to evaluate plasma clearance of iohexol is dry blood spot (DBS) sampling, which proved to be cheap, safe, and simple, not requiring any special equipment, such as freezers, tubes, or centrifuge (Figure 3). The test costs 50€ and gives an accurate assessment of renal function.

Figure 2. Determining plasma clearance of iohexol using DBS sampling (2, 3)


1. Luis-Lima S, Higueras Linares T, Henríquez-Gómez L, et al. The Error of Estimated GFR in Type 2 Diabetes Mellitus. J Clin Med. 2019;8(10):1543. doi: 10.3390/jcm8101543.

2. Porrini E. Formulas that estimate GFR in diabetes: can we trust them? Presented at the 57th European Renal Association – European Dialysis Transplantation Association congress (fully virtual), June 6, 2020. Available at the Virtual Meeting.

3. Porrini E, Ruggenenti P, Luis-Lima S, et al. Estimated GFR: time for a critical appraisal. Nat Rev Nephrol. 2019;15(3):177-190. doi: 10.1038/s41581-018-0080-9. PMID: 30518813

4. Gaspari F, Ruggenenti P, Porrini E; GFR Study Investigators. The GFR and GFR decline cannot be accurately estimated in type 2 diabetics. Kidney Int. 2013;84(1):164-73. doi: 10.1038/ki.2013.47. PMID: 23447062.

NEP – E Summary Articles