#2188 Erythropoietin resistance index (ERI) independently predicts mortality in hemodialysis patients. So, what's beneath the surface?

Abstract Background and Aims The management of anemia is a cornerstone of therapy in hemodialysis patients (HDps), but in most cases high doses of erythropoiesis-stimulating agents (ESAs) are needed for obtaining good Hemoglobin levels, due to a persistent low-grade inflammation and alterations in martial balance (MB) and mineral metabolism (MM). Recent data would identify the erythropoietin resistance index (ERI) as an indicator of all-cause and cardiovascular (CV) mortality risk in HD-ps. Despite this, its use and validation in clinical practice are lacking, as is the evaluation of its potential relationship with other relevant biochemical markers. Aims of the present study were to evaluate: (1) the power of ERI to independently predict mortality in a cohort of HD-ps; (2) the influencing factors of ERI at baseline. Method An observational-retrospective study was performed on 100 prevalent HD-ps, excluding those with acute or chronic infections (3 months), neoplasia, or immunosuppressive therapy. Also, patients transferred to other Centers or that underwent kidney transplant were excluded from the analyses. Patients were divided into two groups based on the median ERI level (9.26 IU*dl/kg/g). ERI was calculated as follows: ESA (IU/week)/weight (kg, post-dialysis)/hemoglobin level (g/dL). At baseline, biochemical characteristics and some specific inflammatory markers were assessed as the average of three measurements taken over three consecutive months. All-cause mortality was considered as primary outcome along a 3-year follow-up (FU), while CV mortality was a secondary outcome. Results HD-ps, predominantly males, had advanced mean age, good Hb values and fair biochemical control (Table 1). As comorbidities, 46% were diabetic and 83% hypertense. After 3 years of FU, 58% of HD-ps died. High ERI levels strongly predicted all-cause mortality (P = 0.008), with a significant difference between groups after every year (Fig. 1). Interestingly, these significances were lost when considering hemoglobin or weight alone. Notably, no significant evidence was found between ERI and CV mortality (P = 0.096). We noticed that, at baseline, ERI showed significant correlations with some elements related to inflammation, MB and BM, such as c-terminal FGF23 (Table 1). At logistic regression, ERI remained mainly influenced by markers related to MB (Table 2a), including Activin A, a protein previously suggested to have pro-inflammatory effects and to promote anemia in patients with chronic kidney disease (CKD). Despite these relevant associations at baseline, on the long run ERI remained highly significant in predicting all-cause mortality at multivariate analyses, together with age, C-reactive protein and hepcidin (Table 2b). Conclusion Our data showed a strong value for ERI in predicting all-cause mortality in a 3-year FU. On the opposite, this was not confirmed for CV mortality, probably because of the low numerosity of the cohort. Furthermore, ERI was found to be related to several factors, including cFGF23 and Activin A, both known to be implicated in maintaining an inflammatory state and anemia in CKD patients. Also, multivariate Cox regression suggested that many factors related to inflammation (i.e. age, hepcidin and C-reactive protein) are relevant for survival, even without affecting the high significance of ERI. In fact, our results suggest the possible presence of some pathological mechanisms that act silently on the development of ESAs resistance, with a strong negative impact on patients’ survival. For its simplicity, ERI should be included in clinical practice, to carefully look at patients’ health and ameliorate their management. Certainly, increased numbers and a prospective design could better clarify the explored relationships in the future.