Validation of the CKiD formulae to estimate GFR in children post renal transplant

Background

The Chronic Kidney Disease in Children (CKiD) study reported new formulae to estimate glomerular filtration rate (eGFR). The study reported here aimed to assess the accuracy of these formulae in estimating levels and changes in GFR in pediatric renal transplant recipients and generate a new formula in our cohort.

Methods

Two hundred and fifty-two studies of plasma disappearance of ¹²⁵I-iothalamate (C_IO) were used to measure GFR in 155 renal transplant recipients. The CKiD bedside formula (CKiD-BS) was compared with C_IO. A mixed logistic regression model was fit to evaluate the performance of estimating change in posttransplant C_IO using CKiD-BS. We used mixed-effects linear regression to fit a multiplicative model of C_IO. The CKiD cystatin-C-based formula (CKiD-Cys) was also used for comparison in 32 additional transplant recipients. Comparisons were made using Bland–Altman plots.

Results

CKiD-BS underestimates C_IO by 20 % for GFR >25 ml/min per 1.73 m². Percentage change in CKiD-BS performed reasonably well in estimating 15 % change of C_IO beginning 6 months posttransplant [area under the curve (AUC)?=?0.791)] The multiplicative constant in the CKiD-BS was recalibrated [R-Bedside?=?0.461?×?ht(cm)/S_Cr).]A GFR model [GFR-M)?=?10.73?×?[(ht(cm)]0.51/(S_Cr)0.90?×?(BUN)0.23] has higher specificity but similar sensitivity for C_IO compared with R-Bedside. CKiD-Cys overestimates C_IO by 10 ml/min per 1.73 m² across a broad range of GFR.

Conclusions

In our cohort, the CKiD-BS underestimates C_IO; however, changes in CKiD-BS can be used to estimate changes in C_IO. CKiD-Cys overestimates C_IO and is not accurate in estimating C_IO.