eGFR, MDRD equation

Clinical background

Serum creatinine is the commonest marker of renal function in clinical practice, but its concentration is influenced by more than the glomerular filtration rate (GFR) alone: muscle mass, sex, age and ethnicity all alter the relationship between serum creatinine and true filtration. Without an estimating equation, the interpretation of a single creatinine value is uncertain, particularly in older patients and those with low muscle mass, in whom kidney damage may be masked by a normal creatinine.

The MDRD equation (Modification of Diet in Renal Disease) was developed to replace Cockcroft–Gault as the dominant estimation method and to provide a standardised estimate of GFR expressed in mL/min/1.73 m², normalised to body surface area. During the first decade of the 2000s it became the equation most laboratories reported alongside the serum creatinine [2]. Today CKD-EPI has largely replaced MDRD for new assessments, since it is more accurate in the higher GFR range, but MDRD lives on in older records, in the reporting of some laboratories and in clinical studies where continuity with earlier data is required.

Calculating eGFR with the MDRD equation

The four-variable MDRD equation, in its re-expressed, IDMS-standardised form, is calculated as:

$$\text{eGFR} = 175 \times S_{cr}^{-1{.}154} \times \text{age}^{-0{.}203} \times 0{.}742^{[\text{female}]} \times 1{.}212^{[\text{Black}]}$$

where $S_{cr}$ is the serum creatinine in mg/dL, age is in years, and the exponents $[\text{female}]$ and $[\text{Black}]$ take the value 1 when the variable applies and 0 otherwise. The result is given in mL/min/1.73 m². The coefficient 175 replaces the original 186 of the older, non-standardised version, and requires the laboratory to calibrate its creatinine measurement against IDMS reference methodology.

The derivation cohort consisted of 1,628 patients with chronic kidney disease recruited into the MDRD study, an American multicentre study from the 1980s [1]. Of these, 1,070 patients formed the training set and 558 the validation set. The patients had uniformly impaired renal function (mean GFR approximately 40 mL/min/1.73 m²) and the equation was modelled against iothalamate clearance as the gold standard. In the validation set the model explained 90.3% of the variance in log GFR [1]. This is the central limitation of the derivation: the equation was constructed in patients with already established kidney disease, not in healthy individuals or those with normal or near-normal filtration.

Interpretation in practice

The MDRD equation reports eGFR in mL/min/1.73 m² and can be used to stage chronic kidney disease according to the KDIGO classification:

A value that the laboratory reports as "≥60" means that the MDRD equation is not considered reliable enough to distinguish stages in that range, and that early kidney disease cannot be excluded on the eGFR alone. Albuminuria or a cystatin C-based estimate is then needed for further assessment.

Validation and performance

The MDRD equation was originally validated in the African American Study of Kidney Disease and Hypertension (AASK), a cohort with hypertensive nephrosclerosis, in which performance was acceptable in the reduced GFR range [2]. In populations with normal or near-normal renal function, however, the equation has shown systematic bias. A systematic review and meta-analysis of 48 studies with a total of 26,875 patients in populations comparable to primary care found that both MDRD and CKD-EPI underestimated measured GFR, but that CKD-EPI was less biased and had a higher proportion of estimates within 30% of measured GFR (P30) [3]. The difference between the equations grew the higher the true GFR was, confirming that the weakness of MDRD lies precisely in the range in which many primary care patients fall [3].

The NKF-ASN Task Force noted in its 2021 interim report that MDRD was still the most widely used equation in American laboratories, even though KDIGO had recommended CKD-EPI as first choice as early as 2012 [2]. The task force further noted that the race coefficient in MDRD (1.212) was somewhat larger than the corresponding one in CKD-EPI 2009 (1.16), and that the race coefficients in both equations rested on observations from American studies in which African Americans had a higher serum creatinine at the same measured GFR as White patients [2]. The biological mechanism behind this difference has never been fully explained.

Limitations

The MDRD equation does not apply in acute kidney injury, since during rapid changes the serum creatinine is not in steady state with the GFR. Nor is it validated in children, in pregnancy or in individuals of extreme body size (severe undernutrition, marked obesity or amputation), in whom the relationship between muscle mass and creatinine production differs from that in the derivation cohort.

The most important systematic weakness is underestimation of GFR above approximately 60 mL/min/1.73 m². This is why many laboratories round the report to "≥60" and do not give an exact value in that range. For patients in whom a decision on dose adjustment of toxic drugs or on the administration of contrast media depends on an accurate estimate in the borderline range, MDRD should not be used on its own. A cystatin C-based estimate or measured GFR is then preferable.

The race coefficient is no longer recommended. In 2021, and definitively in its final report later that year, the NKF and ASN stated that the race coefficient should be removed from creatinine-based eGFR equations [2]. The reasons are partly scientific (race as a biological variable is hard to define and the mechanism unexplained) and partly ethical (the coefficient may delay diagnosis and referral to nephrology or transplant assessment for Black patients by estimating their GFR as higher). The calculator allows the race coefficient to be set, but modern practice is to select Exclude.

MDRD in contemporary laboratory practice

For several years, laboratories have reported eGFR based on CKD-EPI, not MDRD, for new samples. The MDRD equation may still appear in older record entries and in study protocols. Routine eGFR reporting increasingly omits any race coefficient, consistent with the international move towards race-neutral equations. When a patient has an earlier MDRD value in the record and a new CKD-EPI value, it should be borne in mind that the change of method may in itself cause a small shift, particularly in the higher GFR range, and this should not automatically be interpreted as a real change in renal function.

References

1. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med. 1999;130(6):461–70. PMID: 10075613
2. Delgado C, Baweja M, Burrows NR et al. Reassessing the Inclusion of Race in Diagnosing Kidney Diseases: An Interim Report From the NKF-ASN Task Force. Am J Kidney Dis. 2021;78(1):103–115. PMID: 33845065
3. McFadden EC, Hirst JA, Verbakel JY et al. Systematic Review and Metaanalysis Comparing the Bias and Accuracy of the Modification of Diet in Renal Disease and Chronic Kidney Disease Epidemiology Collaboration Equations in Community-Based Populations. Clin Chem. 2018;64(3):475–485. PMID: 29046330