#1329 Assessing eGFR performance against measured GFR stratifying for ethnicity and albumin: development of the AIM CKD UK study

Abstract

Abstract Background and Aims Accurate measurement of kidney function using estimated glomerular filtration rate (eGFR) is essential for appropriate risk stratification of cardiovascular events, kidney failure and prescribing medications. The AIM CKD UK study (IRAS 320215) will assess the accuracy of current and novel creatinine-based eGFR equations compared to exogenous filtration marker derived measured GFR (mGFR) for people of different ethnicities in large national cohort. We performed a preliminary analysis from a single centre in AIM CKD UK to inform future data collection and assessed: Method All adults (≥18-years) who had paired mGFR and eGFR tests within 30 days between 2009-2022 were included. Patients with mGFR ≥150 ml/min/1.73 m2, incomplete datasets and previous amputations were excluded. Age, reported ethnicity, biological sex, referral specialty, height, weight, exogenous filtration marker, creatinine and albumin concentration were recorded. eGFR was calculated from creatinine, age and sex using five equations (Fig. 1). Each eGFR equation was compared to mGFR using median bias, precision, 30% accuracy (P30). Agreement was demonstrated with Bland-Altman plots. Results were stratified for time between eGFR-mGFR (≤7 days vs 8-30 days) and albumin concentration; low (<35 g/L) vs normal / high (≥35 g/L). Assessment of equation performance in different ethnicity was also stratified for the three largest ethnicity groups (White, Black and South Asian ethnicity). Data were analysed using software R V4.2.2. Results There were 2623 patients who met the inclusion criteria. Mean age was 54.8 ± 13.1 years, with 57.8% male (N = 1515). The majority of patients were White ethnicity (58.7%; N = 1539). Median mGFR was 77 mL/min/1.73 m2 (IQR 62–91 mL/min/1.73 m2). Median serum creatinine concentration was 74 umol/L (IQR 61–90 umol/L). The CKD-EPI 2009 equation, compared to mGFR, demonstrated a median bias of +12.8 mL/min/1.73 m2, percentage bias of 16.9% and P30 of 68.6%. The Lund-Malmo revised equation performed best with median bias of 2.1 mL/min/1.73 m2 (percentage bias 2.9%) and a P30 of 82.7%. Agreement, with ethnicity stratification is summarised in Table 1. The majority of patients (N = 2439; 93.0%) had paired mGFR-eGFR tests within 7-days. There was no significant difference between the median bias of each group (P = .749). 2610 (99.5%) patients had concurrent albumin tests performed; 183 (7.0%) of these patients had low albumin concentrations. Median bias was significantly higher in this group compared to normal/high albumin for each equation (e.g. Low vs Normal/high albumin median bias for CKD-EPI equation = 20.3 vs 12.5 mL/min/1.73 m2; P <. 001). Conclusion The LM revised and European Kidney Function Consortium equations perform best in this cohort. Limitations include selection bias and retrospective data. Future data analysis for the AIM CKD study can include paired samples within 30 days as there is no significant difference between 0-7 and 8-30 days. The impact of serum albumin on mGFR-eGFR association in a larger cohort stratified for confounding variables, such as liver disease, malignancy and body mass index is required.