MO802: Genetics Plays a Limited Role in ESA-Hyporesponsiveness and Haemoglobin Outcomes in End-Stage Renal Disease Patients on Haemodialysis

Abstract BACKGROUND AND AIMS Patients with end-stage renal disease on haemodialysis (HD) frequently have resistance or hyporesponsiveness (hypoR) to erythropoiesis-stimulating agents (ESAs) and are also often iron deficient. Despite evidence of increased comorbidities impacting outcomes such as treatment response and achieving targeted haemoglobin (Hb) and iron levels, there is substantial residual variability in these outcomes that remains unexplained. The ability to identify higher-risk sub-populations may provide additional opportunities to intervene. Genetic loci associated with Hb concentration have been identified in genome-wide association studies. Genetic risk scores (GRS) that provide a means to aggregate the health-related risk of multiple genetic alleles into a single number for risk assessment have been validated using these variants. We evaluated if GRS associated with Hb concentration, in addition to clinical factors, improves prediction of ESA hypoR, Hb and iron outcomes better than either one used alone. METHOD A collaborative, non-interventional study between GlaxoSmithKline and DaVita enrolled 3192 African American (AA) and European American (EA) HD patients. Patient demographics, laboratory and vital measurements, concomitant medications, and comorbidity records were drawn from Electronic Health Records, with patient blood samples obtained for genetic analysis. We created GRS for transethnic and euro-centric populations for Hb concentrations from the literature, and a GRS specific to EGLN1/2/3 genes. ESA hypoR status, iron replete status, and time weighted Hb concentration were evaluated as outcomes. ESA hypoR for epoetin alfa was defined by an erythropoietin resistance index ≥ 2.0 U/kg/wk/g/L [intravenous (IV)] or ≥ 1.33 U/kg/wk/g/L [subcutaneous (SC)], or dose ≥ 450 U/kg/wk (IV) or ≥ 300 U/kg/wk (SC). Iron replete was defined as a ferritin ≥ 100 ng/mL and transferrin saturation ≥ 20%. Heritability was estimated using genome complex trait analysis (GCTA). All analyses were performed individually for each population (AA and EA). Significant P-value threshold was 0.0167 (0.05/3 main effects). RESULTS None of the genetic risk scores tested had any clinically meaningful association with outcomes. Though the GRS specific to EGLN1/2/3 was significantly associated with Hb concentration (P < .01) in the AA population, the Beta squared effect size (ES) was very low (ES = 0.005). GCTA analyses were inconclusive with moderate heritability (h2 >8%) and high standard error (>30%), suggesting high variability in heritability estimates for outcomes. Ignoring GRS, several baseline measurements were significantly associated with outcomes. Baseline Hb was consistently and significantly associated with ESA hypoR and Hb concentration in both AA and EA populations, while significantly associated with iron replete status in only the EA population (P < .0001). Other variables such as age and baseline albumin were significantly associated with ESA hypoR and iron replete status, but only in either the EA or AA population (Table). Significant differences were also observed in baseline characteristics between the EA and AA patients. AA patients were 8 years younger and on HD for 2 years longer (P < .0001) at enrolment compared with EA patients. Baseline ferritin, creatinine, albumin and hypertensive status were also significantly different at baseline (P < .01). CONCLUSION There was limited utility for genetics in predicting ESA hypoR, haemoglobin and iron outcomes in HD patients in AA and EA populations. GCTA analyses were inconclusive, suggesting high variability in heritability estimates. Significant differences in baseline variables between the AA and EA populations that were also significantly associated with ESA hypoR, haemoglobin and iron outcomes suggested that population-level baseline differences must be considered when identifying at-risk sub-populations.