Abstract: SA-PO424

Performance of a Pure Metabolite Panel Estimate of GFR (accuGFR)

Session Information

Category: Chronic Kidney Disease (Non-Dialysis)

  • 302 CKD: Estimating Equations, Incidence, Prevalence, Special Populations

Authors

  • Coresh, Josef, Johns Hopkins Medical Institutions, Baltimore, Maryland, United States
  • Levey, Andrew S., Tufts Medical Center, Boston, Massachusetts, United States
  • Inker, Lesley, Tufts Medical Center, Boston, Massachusetts, United States
  • Chen, Jingsha, Johns Hopkins Medical Institutions, Baltimore, Maryland, United States
  • Eiriksdottir, Gudny, Icelandic Heart Association, Kopavogur, Iceland
  • Gudnason, Vilmundur, Icelandic Heart Association, Kopavogur, Iceland
  • Torres, Vicente E., Mayo Clinic, Rochester, Minnesota, United States
  • Ford, Lisa, Metabolon, Inc. , RTP, North Carolina, United States
  • Oyaski, Maria, Metabolon, Inc. , RTP, North Carolina, United States
  • Perichon, Regis, Metabolon, Inc. , RTP, North Carolina, United States
Background

We showed that panels of metabolites can provide an accurate estimate of glomerular filtration rate (GFR) without creatinine and demographic characteristics. We present a Laboratory Developed Test (LDT) to estimate GFR (accuGFR) and estimate its performance in subgroups where eGFRcr is less accurate. Study Population: GFR measurements (mGFR) on 3,236 individuals in the AASK, MDRD, AGES and the CRISP studies were divided randomly into a development sample (50%) and validation samples (25% complete, 25% unused backup).

Methods

Laboratory Methods: Stored (-70°C) serum specimens were subjected to targeted UPLC-mass spectrometry assays for 4 metabolites and serum creatinine (CV <5%). Data Analysis: accuGFR estimated using linear regression of log mGFR on log metabolites in the development sample. Performance measured using large errors (1-P30 and 1-P20, the percentage of estimates deviating from the mGFR by >30% and >20%). We compared the accuGFR to eGFR by CKD-EPI equations.

Results

mGFR spanned a wide range (mean 55, SD 26 ml/min/1.73m2). accuGFR based on 4 metabolites without demographics had substantially better accuracy than eGFRcr and eGFRcys but similar performance to eGFRcr-cys in the accuGFR development and validation samples. In subgroups where there is concern that eGFRcr may be inaccurate, accuGFR often outperformed even eGFRcr-cys, particularly for 1-P20.

Conclusion

accuGFR based on four metabolites without serum creatinine or demographics nearly halved the rate of large errors compared to eGFRcr and appears to be robust across a range of relevant subgroups. The utility of the accuGFR test based on a single blood draw should be tested in diverse clinical and research populations.