The Impact of Different Pediatric Estimated GFR Equations on Reporting of Renal Outcomes

Sickle cell nephropathy is associated with significant morbidity and mortality. It is well established that systematic bias exists with estimated GFR (eGFR) equations but the time and cost associated with measured GFR (mGFR) prohibits its use in annual monitoring. The National Kidney Foundation has endorsed three pediatric eGFR equations using either cystatin (CyC) alone, serum creatinine (SCr) alone, or a combination of CyC and SCr. It has been suggested that mGFR should be performed when differences >10mL/min/1.73m² exist between the CyC and SCr based eGFR equations. This study tested the hypothesis that the majority of patients would have a difference in eGFR <10mL/min/1.73m². Furthermore, we tested the hypothesis that using the different eGFR equations would replicate the results demonstrated in a prior research study on early childhood hyperfiltration.

Methods: In the UAB Pediatric Sickle Cell Nephropathy cohort, we abstracted data from 84 patients during early childhood (ages 4-10 years) to calculate the mean values for each variable required to calculate eGFR using the different equations. In total, we identified 285 visits (mean: 3.4 eGFR measurements/patient from 4-10 years). At Children's of Alabama, eGFR is reported from cystatin C using the formula: log10 eGFR=1.962 + [1.123 x log10(1/CyC)]. The three additional pediatric eGFR equations adopted by the National Kidney Foundation include: (1) Schwartz Creatinine equation: eGFR=41.3 x (ht(m)/SCr), 2) Schwartz cystatin C equation: eGFR =70.69 x (CyC)-^0.931, and 3) CKiD Cystatin-Creatinine equation = 39.8 x (ht(m)/SCr)^0.456 x(1.8/CyC)^0.418 x (30/BUN)^0.079 x (1.076^male)(1.0^female) x (ht(m)/1.4)^0.179). We performed Bland-Altman analysis to determine differences between eGFR equations. We then determined the number (percent) of patients identified with a discrepancy between SCr and CyC equations >10 mL/min/1.73m². To evaluate the impact of the three different equations on clinical research, we re-analyzed our data that previously showed the temporal relationship between hyperfiltration and the development of albuminuria. We defined hyperfiltration as the highest 25% of mean eGFR values during early childhood for each formula. We then performed a time to event analysis for the development of the first episode of albuminuria for each formula based on early hyperfiltration vs no early hyperfiltration. All analyses were conducted using JMP14 (Cary, NC).

Results: The mean eGFR during early childhood based on the four equations were: UAB cystatin C eGFR: 171 mL/min/m², Schwartz cystatin C eGFR: 119 mL/min/m², Schwartz creatinine: 145 mL/min/m², and CKiD creatinine-cystatin: 129 mL/min/m². The mean difference between all GFR formulas was significantly different ranging from 9.5 mL/min/m² (Schwartz CyC vs CKiD CyC-SCr) to 53 mL/min/m² (Siemens CyC vs Schwartz CyC). (p<0.0001). Comparing the Schwartz CyC vs Schwartz SCr eGFR equations, the mean eGFR difference was 26.5 mL/min/m². Only 16/84 (19%) participants had a eGFR difference <10 mL/min/1.73m² between serum CyC alone vs SCr alone equations. Replicating our time to albuminuria analysis, hyperfiltration using the Schwartz CyC (p=0.04) and Schwartz SCr (p=0.02) equations remained statistically significant for development of albuminuria early in life while the eGFR equation using the combination of CyC and SCr (p=0.053) was suggestive but not significantly associated with the development of albuminuria. While both the CyC and Scr equations were statistically associated with earlier development of albuminuria, only 12 (55%) of the 22 patients identified with hyperfiltration using the Schwartz CyC equation also were identified as having hyperfiltration using the Schwartz SCr equation.

Conclusion

This data highlights the large differences in the mean eGFR between the four different pediatric eGFR equations. Over 80% of patients during early childhood were identified with a difference of > 10 mL/min/1.73m² in cystatin and creatinine based eGFR equations suggestive of the need to perform an mGFR test. Furthermore, the different eGFR equations can impact the categorization of patients and significance of research findings. Future research to improve sickle cell kidney disease must acknowledge the systematic bias in eGFR equations and, as a research community, work together to report eGFR outcomes using a similar methodology.