Comparison of race-based and non-race-based glomerular filtration rate equations for the assessment of renal functional risk before nephrectomy

Abstract

Objective:

To compare the performance of race-based and race-neutral estimated glomerular filtration rate (eGFR) calculators in patients undergoing kidney surgery.

Methods:

Analysis of institutional kidneys surgeries from 2006–2021 was conducted. Demographics, serum creatinine (SCr), protein dipstick, and creatinine clearance (CrCl) were assessed within one week prior to surgery. SCr was used to calculate eGFR using 3 models: MDRD, CKD-EPI 2009, and CKD-EPI 2021. Patients were classified based on KDIGO criteria for CKD staging and prognosis, with urine CrCl treated as benchmark for analysis. ROC analysis evaluated accuracy of eGFR calculators’ binary discrimination of eGFR less than 60 ml/min. CKD stage agreement between eGFR and urine creatinine clearance was assessed.

Results:

A total of 554 kidney surgeries in 336 patients had necessary laboratory data for analysis. The cohort was 62% male, with a median age of 47. Within this cohort, 8.1% (n=45) were Black, and 80% (n=441) were White. GFR from 24-hour urine creatinine clearance normalized by BSA did not vary significantly from eGFR by serum creatinine based calculators. The proportion of patients with eGFR <60 significantly differed when using MDRD, CKD-EPI 2009, and CKD-EPI 2021 equations when compared to creatinine clearance (p < 0.001). Still, they performed equivalently in the staging of CKD, as well as in predicting GFR of less than 60, and classifying CKD prognosis of “moderately increased or higher”.

Conclusion:

A race-neutral eGFR calculator can perform equivalently to established eGFR calculators, with the added benefit of mitigating biases that account for racial disparities in nephrectomy decision making.

Introduction

Preoperative estimation of glomerular filtration rate (GFR) prior to partial or radical nephrectomy for renal cell carcinoma has several clinical implications. The current American Urologic Association (AUA) guidelines emphasize the accurate staging of preoperative kidney disease for assessment of renal functional reserve and appropriateness of surgical intervention.¹ Additionally, for systemic treatment, GFR can be a factor for clinical trial eligibility, choice of therapy and dosage.² If renal function is expected to be severely compromised by surgical extirpation, routine renal function assessments may offer insight into candidacy for short or longterm dialysis or renal transplantation. Since GFR is a determinant of several clinical parameters, there is a need for a standardized, accurate method of estimating renal function prior to nephrectomy for renal carcinoma.

Creatinine is the most utilized and cost-effective endogenous filtration marker.³ Assessment of GFR by creatinine clearance via 24-hour urine collection offers a direct, informative measure of renal function that is independent of variability in serum creatinine levels.^4,5 However, in practice, it is not routinely assessed in the preoperative setting.⁶ GFR is more routinely estimated by serum creatinine with equations that model differences in creatinine production, secretion, and reabsorption by modifiers such as age, sex, and self-reported race.^7–10 Among these parameters, race has not been established to accurately account for non-GFR-related serum creatinine parameters, and therefore may introduce unintentional bias into GFR estimation and subsequent surgical algorithms.¹¹

To date, several race-based equations have been developed to calculate estimated GFR (eGFR) using serum creatinine. Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration 2009 (CKD-EPI 2009) equations are among the most common.^{9, 10} CKD-EPI 2009 has been demonstrated to have the highest accuracy in estimating GFR and was recommended in the 2012 guidelines for Kidney Disease: Improving Global Outcomes (KDIGO) clinical practice guidelines.¹² Additionally, it is the currently accepted standard by the AUA for preoperative assessment of renal carcinoma.^{1, 10, 13} Both the MDRD and CKD-EPI 2009 eGFR equations include a race coefficient to modify serum creatinine concentration.^{9, 10} National Kidney Foundation (NKF) and American Society of Nephrology (ASN) recently developed CKD-EPI 2021 as part of an initiative to reduce the bias of self-identified race in clinical decision-making by eschewing race in their algorithms.⁸

The aim of this study is to compare the accuracy of three eGFR calculators: namely, the new, non-race-based CKD-EPI 2021 to the race-based CKD-EPI 2009 and MDRD calculators in a cohort of patients undergoing partial or radical nephrectomy. A reference standard of preoperative urine creatinine clearance and proteinuria will serve as important measures for validation. eGFR equations have been compared in various medical settings, including in patients with heart failure and myocardial infarction, but no studies have validated the use of CKD-EPI 2021 or compared the use of various GFR calculators to accurately assess renal function prior to nephrectomy.¹⁴

Methods

We retrospectively queried our institutional registry of patients undergoing kidney cancer surgery from 2006–2021 at the National Cancer Institute. Demographic data, serum creatinine, urinalysis, and creatinine clearance were assessed within one week before surgery.

Preoperative 24-hour urine creatine clearance (CrCl) was used as a benchmark. Urine creatinine clearance was normalized to body surface area (BSA) from patient height and weight¹². Estimated GFR was calculated based on serum creatinine (sCr) according to the following equations:

• CKD-EPI equation 2009: eGFR = 141 × min(sCr/κ, 1)α × max(sCr/κ,1)−1.209 × 0.993age × 1.018 (if female) × 1.159 (if Black)

  • κ is 0.7 for females and 0.9 for males,
  • α is −0.329 for females and −0.411 for males
• CKD-EPI equation 2021: eGFR = 142 × min (standardized sCr/K, 1)α × max(standardized sCr/K, 1)-1.200 × 0.9938Age × 1.012 (if female)

  • κ = 0.7 (females) or 0.9 (males)
  • α = −0.241 (female) or −0.302 (male)

• MDRD equation: eGFR = 186 × sCr^(−1.154) × Age^(−0.203) × (0.742 if female) × (1.210 if Black)

Patients were assigned to CKD stages based on each of the eGFR estimates with the use of KDIGO guidelines, with stage 1, 2, 3a, 3b, 4, and 5 defined by eGFR ≥90, 60 to 89, 45 to 59, 30 to 44, 15 to 29, and <15 ml/min/1.73 m², respectively¹². Patients were assigned KDIGO Prognosis of CKD risk into “low risk”, “moderately increased risk”, “high risk” and “very high risk” categories based on degree of albuminuria on protein dipstick and creatinine clearance by 24-hour urine collection. Albuminuria was also categorized by KDIGO guidelines, where none or trace proteinuria, 1+, >2+ (including 3+) designated in categories A1, A2, A3, respectively¹².

Statistical analysis

Categorical variables were expressed as count and percentages, and continuous variables were expressed median and interquartile range. Continuous variables were compared using Wilcoxon rank sum, while categorical variables were compared using Pearson’s Chi Square.

Receiver operating characteristic (ROC) curve analysis was used to assess the accuracy of each eGFR calculators’ binary discrimination of estimated eGFR less than 60 ml/min/1.73 m² (Figure 2A) and moderate to very high risk by KDIGO (Figure 2B) designated by urine clearance and albuminuria as the benchmark. AUC is reported with standard error.

Figure 2.

Receiver operator characteristic (ROC) curves for (A) GFR < 60 and (B) KDIGO Prognosis Moderate Risk of Higher

Confusion matrices were used to illustrate each calculator’s accuracy of CKD staging compared to urine creatinine clearance. Agreement was assessed using Cohen’s kappa without weighting with standard error.

Two-sided p values < 0.05 were considered statistically significant. All data transformation and statistical analyses were performed using R version 4.1.1 and SPSS v.1.0.0.1089.

Results

Study Population

Over the period of 2006–2021, 1,159 surgeries were performed on 765 patients. 554 kidney surgeries conducted on 336 patients had all necessary laboratory data and represented the final study population for analysis. The cohort was 62% male, with a mean age of 47 at the time of surgery. Within this cohort, 8.1% (n=45) was Black, and 80% (n=441) was White (Table 1).

Table 1:

Baseline Patient Characteristics

Characteristic	N = 554
Age at Surgery (years)	47 (36, 57)
Number of Previous Surgeries	1 (1, 2)
Race
American Indian or Alaskan	1 (0.2%)
Asian	17 (3.1%)
Black/African American	45 (8.1%)
Multiracial	10 (1.8%)
Pacific Islander	8 (1.4%)
Unknown Race	32 (5.8%)
White	441 (80%)
Gender
Female	211 (38%)
Male	343 (62%)
BMI	29 (25, 33)
Proteinuria Category
A1 (None to Trace)	485 (88%)
A2 (1+)	37 (6.7%)
A3 (2+ to 3+)	32 (5.8%)
Baseline Renal Function
eGFR CKD EPI (2009)	78 (60, 99)
eGFR CKD EPI (2021)	80 (60, 101)
eGFR MDRD	71 (56, 89)
Urine Cr. Clearance	81 (60, 103)
% of Surgeries with GFR <60 ml/min/1.73 m 2
eGFR CKD EPI (2009)	139 (25%)
eGFR CKD EPI (2021)	134 (24%)
MDRD	183 (33%)
Urine Cr. Clearance	140 (25%)

¹n (%)

²Median (IQR)

Median GFR varied by equation. The absolute proportion of preoperative GFR values less than 60 ml/min/1.73 m² varied depending on the calculator employed (Figure 1). Based on urine creatinine clearance, 108 patients (19%) had a GFR of less than 60 ml/min/1.73 m². MDRD, CKD-EPI 2021, and CKD EPI 2009 all differed significantly from urine creatinine clearance in their classification of patients with a GFR of less than 60 ml/min/1.73 m² (183 (33%),134 (24%), and 139 (25%), p<0.001) respectively.

Figure 1:

Histogram illustrating distribution of GFR values

Comparison of accuracy (ROC and kappa analysis) by eGFR equation

Receiver operator characteristic (ROC) curves and area under curve (AUC) were equivalent across all calculators, indicating moderate accuracy in their ability to discriminate renal function less than 60 ml/min and moderate to very high risk by KDIGO (Figure 2A-2B) based on urine creatinine clearance standardized by BSA as benchmark. In CKD staging, all three calculators demonstrated equivalent agreement when individually compared to urine creatinine clearance (Supplemental Figure 1). The three equations agreed on the classification of 62% of patients to an eGFR <60 ml/min defined by urine creatinine clearance.

Comparison of renal function estimates in Black and non-Black race

In a sub-analysis comparing between the “Black” and “non-Black” cohort, gender was distributed equally (p=0.3) (Table 2). Patients undergoing surgery who identified as Black were older (55 years, IQR: 40–61 vs. 47 years, IQR: 36–56) (p=0.003) and had a larger BMI and incidence of proteinuria compared to non-Black counterparts (Table 2).

Table 2:

Demographic and GFR Sub analysis of Black and Non-Black Patients

Characteristic	Black, N = 45	Not Black, N = 509	p-value
Age at Surgery (years)	55 (40, 61)	47 (36, 56)	0.003
Gender			0.3
Female	14 (31%)	197 (39%)
Male	31 (69%)	312 (61%)
BMI	34 (30, 39)	28 (25, 33)	<0.001
Proteinuria Categories			<0.001
A1	30 (67%)	455 (89%)
A2	8 (18%)	29 (5.7%)
A3	7 (16%)	25 (4.9%)
Baseline Renal Function
eGFR CKD EPI (2009)	72 (45, 92)	79 (61, 100)	0.031
eGFR CKD EPI (2021)	66 (40, 89)	81 (63, 102)	<0.001
eGFR MDRD	71 (44, 88)	71 (56, 90)	0.3
Urine Cr. Clearance	69 (51, 99)	82 (61, 104)	0.1
% of Surgeries with GFR <60 ml/min/1.73 m2
eGFR CKD EPI (2009)	17 (38%)	122 (24%)	0.041
eGFR CKD EPI (2021)	20 (44%)	114 (22%)	<0.001
MDRD	17 (38%)	166 (33%)	0.5
Urine Cr. Clearance	16 (36%)	124 (24%)	0.1

¹n (%)

²Median (IQR

Comparing the median eGFR of Black patients to the rest of the cohort, only the CKD-EPI 2009 equation (p=0.031) and CKD-EPI 2021 (p<0.001) equations differed by race (Table 2). GFRs acquired by urine creatinine clearance and the MDRD equation did not differ. Additionally, the absolute proportion of cases identified to have a baseline eGFR of less than 60 ml/min/1.73 m² differed by race when using CKD-EPI 2009 and CKD-EPI 2021 equations, but not the MDRD equation and urine creatinine clearance (Table 2).

Discussion

Renal function is difficult to measure directly and can vary based on the biomarker or equation that is utilized. eGFR calculators are established using measurements of kidney function based on radiotracer infusion and urinary clearance to generate multivariate equations with age, gender, and racial coefficients. However, of these coefficients, variability of GFR by race may not be substantiated with evidence. These claims were determined by reference population that was used to derive the MDRD and CKD-EPI 2009 race-based calculators were only 12 and 32% Black respectively. Within these populations and also in our study, there was unbalanced comorbidities that confounded the true estimation of renal function. In light of this analysis, among the many eGFR calculators available to clinicians, CKD-EPI 2021 offers a more straightforward estimation of GFR. Additionally, CKD-EPI 2021 creates a more equitable appraisal of GFR prior to nephrectomy as well as in the hospital stay, and postoperative period.

Compared to urine creatinine clearance, all calculators equivalently classify urine creatinine clearance as less than 60 ml/min with moderate accuracy. Additionally, all three calculators could also equivalently establish CKD prognosis as “moderately high or greater” based on KDIGO predictions from urine creatinine clearance and proteinuria, though less optimally. This suggests that assessment of proteinuria can account for additional risk not estimated by GFR alone as previously indicated¹⁵. Lastly, all three calculators demonstrated equivalence in the staging of CKD. Urine creatinine clearance can still be evaluated in patients with marginal renal function to establish an accurate baseline prior to surgery

In a sub analysis of comparing patients who self-identified as “Black” to the rest of the cohort, only the GFR values estimated by CKD-EPI 2009 equation (p=0.031) and CKD-EPI 2021 (p<0.001) equations differed by race. GFR values did not differ by race when determined by MDRD or urine creatinine clearance. Black patients within this cohort were older, and this could account for this observed difference for CKD-EPI 2009 and CKD-EPI 2021 as the age is weighted higher for these calculators compared to MDRD. Additionally, Black patients in this study had an accompanying higher median BMI and incidence of proteinuria which suggests increased comorbid conditions that are not estimated by eGFR alone. Additionally, removal of the race multiplier within CKD-EPI 2021 results in an increased proportion of Black patients identified as having lower than 60 ml/min/1.73 m² GFR.

Limitations of this study include retrospective design. Therefore, it does not definitively address the optimal equation to use when predicting post-operative functional outcomes, mortality, or acute kidney injury. Future studies should evaluate performance of each calculator in assessing renal functional deficits in the acute and chronic post-operative stages. Additionally, reference standards apart from urine creatinine clearance, such as iothalamate infusion, should be considered due to inherent challenges in 24 hour urine collection and evaluation. It must be noted that a large proportion of patients included in our cohort of nephrectomies have a known or suspected predisposition to the development of hereditary kidney cancer. Hence, they represent a population at risk of progressive renal functional decline, and attributable risk of surgical CKD may be higher compared to sporadic renal carcinoma patients which can affect the generalizability of these results. Black patients only comprised 8.1% of this dataset. This is likely attributable to individual and systemic barriers in the recruitment, enrollment, and participation of minority groups in clinical trials¹⁶. A more racially heterogenous population should be studied to evaluate the race-specific effect of nonrace-based eGFR in clinical decision-making, such as nephrectomy and therapeutic trials.

The use of race adjustments can inadvertently create barriers to nephrectomy and assessment of outcomes following nephrectomy in Black individuals. For example in operative planning with patients with advanced CKD, the impact of the race coefficient on eGFR may alter a clinical judgment for partial versus radical nephrectomy candidacy. Landmark values of renal function are built into the urologic guidelines recommendations, and they are primarily based on SeCr based estimates of GFR, which further stresses the importance of standardization of eGFR calculators utilized for patients.

Conclusions:

Our study suggests that choice of eGFR calculator changes CKD risk estimation. Interestingly in the designation of patients with a GFR of less than 60 ml/min/1.73 m², all three models of GFR measurements reached consensus for only 62% of patients identified by urine creatinine clearance, leaving a large margin of reclassification based on calculator employed. Nevertheless, all three calculators could still effectively predict urine creatinine clearance less than 60 ml/min/1.73 m². This finding indicates that a non-race based GFR calculator performs equivalently to other clinically established eGFR calculators, while also addressing racial biases that account for health disparities in clinical nephrectomy decision-making.¹⁷

Supplementary Material

1

Supplemental Figure 1: Contingency table displaying staging of CKD using eGFR relative to urine creatinine clearance