Evaluation of equation as a classifier for kidney function in domestic cats: a proof-of-concept study.

Abstract

In human medicine, estimated glomerular filtration rate (eGFR) is routinely used to assess renal function. In cats, reliable eGFR formulas are not yet available. To test the hypothesis that a human eGFR equation with four metabolites can effectively distinguish cats with different renal functions, two cohorts of cats were retrospectively enrolled. Cohort 1 included 34 colony cats and cohort 2 had 29 client-owned cats. The metabolite markers were quantified in cat serum and renal function was estimated using the adapted equation (eRF_fel). Receiver Operating Characteristic Area Under the Curve for eRF_fel as a binary classifier were 0.9714 and 0.9958, and 0.9731 and 0.9643 as a ternary classifier for cohort 1 and cohort 2, respectively. In cohort 1, eRF_fel was 1.30 ± 0.04 (Mean ± SEM, mL/min/kg), 0.78 ± 0.06, and 0.36 ± 0.05 for healthy control, IRIS Chronic Kidney Disease (CKD) stages 1 and 2 (CKD1/2), and stages 3 and 4 (CKD3/4), respectively, and 1.30 ± 0.06, 0.72 ± 0.05, and 0.47 ± 0.04 for healthy control, CKD stage 2, and CKD stage 3 in cohort 2, respectively (both P < 0.0001). Although the original equation was trained on human data, eRF_fel was associated positively with urine specific gravity, but negatively with serum symmetric dimethylarginine and creatinine in cats (all P < 0.0001). The preliminary results demonstrated the effectiveness of eRF_fel as a classifier for feline renal function.