Estimation of Glomerular Filtration Rate (GFR) Using Serum Cystatin C and Creatinine Has Poor Precision and Accuracy in Adults with HbSS Sickle Cell Disease

Introduction: Renal damage is a progressive complication of sickle cell disease (SCD) that begins with hyerfiltration in childhood followed by decline in glomerular filtration rate (GFR) and chronic renal failure in up to 12% of HbSS adults. Albuminuria is a marker of glomerular damage but does not predict GFR. Serum creatinine (SCr) is typically low in SCD despite glomerular damage, in part because tubular dysfunction leads to increased creatinine secretion. Cystatin C (CysC) is a low molecular weight protein, eliminated exclusively by glomerular filtration, and correlates with GFR in children and adults with chronic kidney disease (CKD). Previously, estimated GFR (eGFR) has been compared to measured GFR (mGFR) using equations that incorporate CysC and SCr in young children with SCD; however no studies have examined the precision or accuracy of GFR formulas in adults and older children with SCD.

Methods: Adults and children ≥10 years with HbSS and albuminuria (albumin/creatinine ratio ≥30 mg/g) who were on hydroxyurea therapy for ≥6 months were enrolled on a trial of losartan therapy to determine effects on albuminuria and renal hemodynamics. GFR was measured by urinary clearance of intravenous iohexol at baseline and 2 time points subsequent to starting losartan. Serum creatinine and CysC were measured using standard clinical laboratory techniques. For subjects age ≥18 years, eGFR was calculated using 3 CysC-based equations that have high accuracy in adults with CKD:

(1) eGFR = 76.7 x [CysC(mg/L)]^-1.19;

(2) eGFR = 127.7 x [CysC(mg/L)]^-1.17 x age^-0.13 x 0.91^(female) x 1.06^{(African American)};

(3) eGFR = 177.6 x [SCr(mg/dL)]^-0.65 x [CysC(mg/L)]^-0.57 x age^-0.20 x 0.80^(female) x 1.11^{(African American)}.

In subjects <18 years, eGFR was calculated by:

(1) modified Schwartz formula, eGFR = height in cm x 0.41/SCr(mg/dL), and

(2) CKiD Schwartz formula: eGFR = 39.1 x [height (m)/SCr]^0.516 x [1.8/CysC (mg/L)]^0.294 x [30/BUN (mg/dL)]^0.169 x 1.099^(male) x [height(m)/1.4]^0.188 .  

The relation of serum CysC with mGFR was determined by Pearson’s correlation. Bias was assessed as the median difference between eGFR and mGFR, and precision as the interquartile range (IQR) for the difference. Accuracy was assessed as the percent of estimates within 10% (P₁₀), 30% (P₃₀) and 50% (P₅₀) of mGFR.

Results: In 15 subjects (4 female, 1 <18 years), there were 33 iohexol GFR measurements (15 prior to losartan, 18 on losartan). Eight (53%) had macroalbuminuria (ACR >300 mg/g).  Mean SCr was 0.58 mg/dL, mean CysC was 0.74 mg/L (normal 0.5 – 1.0 mg/L), and mean mGFR was 106.9 ml/min/1.73 m² (range 68 - 147). Four (27%) subjects had stage 2 CKD (GFR <90 ml/min/1.73 m²) measured at 7 time points prior to losartan. CysC correlated inversely with mGFR (r= -0.58, p=0.0003) (Figure) and was higher in those with stage 2 CKD vs. stage 1 (mean 0.95 vs. 0.68 mg/L, p=0.0013). CysC was above laboratory-defined normal range in 2/26 (7%) with stage 1 CKD and 3/7 (43%) with stage 2 CKD (p=0.052). For subjects ≥18 years (31 measurements), bias, precision, and accuracy of all 3 equations is shown in the Table. Bias tended to be less for measurements taken while on losartan; however this was not a significant difference for any eGFR equation. For the 1 subject <18 years (2 measurements), the modified Schwartz formula had less bias than the CKiD formula (12.4 vs. 83.7 ml/min/1.73 m²).

	Bias – ml/min/1.73 m2 (95% CI)	Precision – IQR ml/min/1.73 m2 (95% CI)	P10 - %	P30 - %	P50 - %
Equation 1	21.9 (-7.8, 29.8)	50.1 (6.3, 66.6)	12.9	61.3	96.7
Equation 2	33.5 (-4.7, 47.6)	58.0 (12.4, 81.5)	16.1	51.6	74.2
Equation 3	64.3 (51.7, 80.0)	42.2 (8.4, 76.9)	9.7	16.1	38.7

Conclusions: In adults with HbSS, the equation that best estimated GFR incorporated serum CysC alone without age, sex, race, or SCr; however accuracy and precision of all equations was unacceptably low and tended to overestimate GFR. Equation 3 which incorporated SCr had the greatest positive bias. Cystatin C correlates with GFR in SCD, and is elevated more frequently in stage 2 CKD, thus it remains an important marker of renal impairment in this population. As direct GFR measurement is not practical in the general clinical setting, further studies are needed to develop eGFR equations with improved accuracy for the SCD population.