Abstract 18126: Attenuations in Serum Albumin Over Time Powerfully Predict an Amplified Risk of Death in Chronic Heart Failure

Background: Worsening heart failure is typically accompanied by diffuse inflammation, plasma volume expansion, diminished nutritional status and blunted liver biosynthetic function due to hepatic congestion. Because attenuations in serum albumin over time reliably reflects these derangements, we hypothesized that it would be predictive of death in patients (pts) with chronic heart failure (CHF) and that its prognostic power would supercede that of static single measures of albumin at baseline. Methods: We analysed the relation of serum albumin on first consultation and over time with mortality in 246 outpatients with CHF (mean [±SD] age 68±12y, LVEF 29±8%, albumin 42±5 g/L, 48% NYHA class >2, 55% ischemic). Results: On initial consultation, 12 (5%) pts had an albumin < 35g/L. Over a median (±IQR) follow-up of 13±18 months, 51 (21%) pts died, 132 (54%) pts had a fall in albumin, and 46 (19%) pts harboured attenuations > 6g/L. On Cox proportional hazards analyses, a lower albumin at baseline predicted increased mortality (HR 0.89, CI 0.84 - 0.94, χ 2 :18, P<0.0001) independently of all covariates. However, a falling albumin over time (unadjusted HR 0.89, CI 0.84 - 0.92, χ 2 53, P<0.0001) was even more predictive (Difference in ROC AUC for baseline vs Δalbumin 0.16, P<0.001) and did so independently of baseline albumin, NYHA class and changes in urea and hemoglobin (Δalbumin adjusted HR 0.84, CI 0.81 - 0.88, χ 2 54, P<0.0001). A reduction in albumin > 6 g/L optimally predicted death (ROC AUC 0.82, CI 0.76 - 0.86, P<0.0001, Fig A) and conferred a 6-fold escalated risk of mortality (HR 6.42, CI: 3.67 - 11.22, P<0.0001, Fig B). In incremental prognostic analyses, the addition of Δalbumin to the strongest 4 variable model dramatically augmented the χ 2 value (Fig C). Conclusions: Attenuations in serum albumin over time strongly correlate to an enhanced risk of death in chronic heart failure and could be utilised to further refine risk stratification.