Direct comparison of ultrafiltration to pharmacological decongestion in heart failure: a per-protocol analysis of CARRESS-HF.

Abstract

Mechanical ultrafiltration (UF) involves the removal of an iso-osmotic filtrate from the blood. Its benefit in acute decompensated heart failure, however, remains inconclusive. We sought to better understand the direct effects of UF in comparison to an aggressive, urine output-guided pharmacological protocol for decongestion on fluid loss, renal function, and neurohormonal activation. A per-protocol analysis of the Cardiorenal Rescue Study in Acute Decompensated Heart Failure (CARRESS-HF) trial (n = 188) was performed. Participants were included if randomized to UF and had UF output collected, or if randomized to the pharmacological arm and had urine but not UF output collected. Using these definitions, there were 163 participants at 24 h, 156 at 48 h, 129 at 72 h, and 106 at 96 h. UF was associated with higher cumulative fluid loss (P = 0.003), net fluid loss (P = 0.001), and relative reduction in weight (P = 0.02). UF was also associated with higher serum creatinine and blood urea nitrogen by 72 h (P-interaction <0.05 for both), lower serum sodium by 48 h (P-interaction <0.01) and increased plasma renin activity by 96 h (P = 0.04). The pharmacological arm was associated with higher serum bicarbonate after 24 h (P-interaction <0.002). There were no differences in 60-day outcomes between the UF and pharmacological arms. Ultrafiltration vs. pharmacological therapy was associated with more fluid removal but also rise in serum creatinine and neurohormonal activation. Additionally, loop diuretic use vs. UF was associated with an increase in serum bicarbonate despite less decongestion, data which question the commonly held conception of a 'contraction alkalosis'.