Race-Based Creatinine Egfr Equations Outperform Nonrace-Based and Cystatin C-Based Equations in a Population of Adults with Sickle Cell Disease

Background:

Sickle cell disease (SCD) progressively alters kidney function, increasing morbidity and mortality. In the setting of recurrent and chronic tubular injury unique to SCD, established equations to determine glomerular filtration rate (GFR) have limitations. This study compares estimated GFR (eGFR) derived from serum cystatin C (Cys)- and creatinine (Cr)- based equations to iothalamate-measured GFR (mGFR) to determine the more accurate noninvasive marker of kidney function in adults with SCD. We hypothesized that Cys-based equations would better predict mGFR than Cr-based equations. We explored additional markers of kidney function, including endothelin-1, a proinflammatory vasoconstrictor; serum β2-microglobulin (β2M), which is elevated with glomerular dysfunction; and renal blood flow.

Methods:

Adults with SCD who were free of vaso-occlusive disease were recruited from the outpatient setting. We calculated eGFR using the 2009 Cr Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), 2012 CKD-EPI Cys, 2021 CKD-EPI Cr-Cys, 2021 CKD-EPI Cr, and an SCD-specific equation (eGFR Jamaica); 2021 equations exclude race as a coeffciient. We measured renal blood flow via MRI and serum endothelin-1 and β2M by clinical laboratory testing. Patients performed a 24-hour urine collection for albumin excretion. The iothalamate plasma decay method was used to determine mGFR. Pairwise comparisons of the equations were made using the Wilcoxon signed rank test to evaluate the absolute value of the discrepancies between the measurement and iothalamate mGFR. Bland Altman plots were used to visually assess agreement and bias. We used Spearman correlations to assess the relationship between novel markers and mGFR or albuminuria.

Results:

Sixty-seven patients with SCD (Male: 39, Female 28; mean age 36.9 + 10.9 years; HbSS: 64, HbSβ⁰-thalassemia: 3) completed laboratory evaluations. Most were on stable hydroxyurea doses (n=51, 76%), and a subset were on kidney protective medications (n=15, 22%). Median mGFR was 154mL/min/1.73 m² (Q1:123, Q3:231). Of the estimating equations, the 2009 Cr-based equation had the lowest median absolute values of discrepancy: 25.6mL/min/1.73m², while the 2012 Cys had the highest discrepancy: 33.8 mL/min/1.73 m² (p < .01 for all comparisons). The Jamaican eGFR was most similar to the 2012 cys equation (p=0.099). The data show strong general GFR underestimation from the Jamaican equation. In patients with albuminuria (>30mg/24hours, n=32), the 2009 Cr equation had the lowest median absolute value of discrepancy: 22.95 mL/min/1.73 m², and the 2012 Cys equation had the highest: 42.55 mL/min/1.73 m². The 2009 Cr equation showed the best agreement with mGFR; the 2012 Cys equation showed a consistent tendency to under-estimate results and had the poorest agreement. The 2021 Cr-based equation showed lower discrepancy and higher agreement than the 2021 Cr-Cys equation across GFR 0-150 mL/min/1.73 m².

There was a moderate inverse correlation between β2M and mGFR (r=-0.45, p=0.0002). Similarly, β2M correlated with albuminuria (r=0.47, p=0.0004). Average kidney blood flow was weakly correlated with iothalamate mGFR (r=0.34, p=0.009) but not with albuminuria (r=-0.07, p=0.67). Endothelin-1 levels did not correlate with mGFR or albuminuria.

Conclusion:

Of the estimating equations, the 2009 Cr-based eGFR had the highest agreement, and lowest absolute values of discrepancy with mGFR, even when controlling for albuminuria. Both Cys-based equations produced higher discrepancies with a bias to under-predict mGFR. These data suggest Cys-based eGFR is inferior to Cr-based eGFR in adults with SCD and normal kidney function/ hyperfiltration. The heightened inflammatory state in SCD may contribute to higher Cys levels and, therefore, depressed eGFR, increasing the discrepancy. These data demonstrate Cys’s accuracy at predicting mGFR in SCD is limited. The 2009 Cr equation, which includes a race coefficient, also outperformed the 2021 Cr equation, which excludes race, in terms of discrepancy. In addition, our data confirm β2M‘s correlation with mGFR and albuminuria. Further, renal blood flow may play a small role in GFR changes. A larger data set in SCD patients with lower mGFR between 30-90 mL/min/1.73m² and longitudinal evaluations may better define the most accurate equations to assess kidney function at different stages of CKD in patients with SCD.