Background: Hemofiltration in the form of continuous venovenous hemofiltration (CVVH) is increasingly used to treat acute renal failure. Compared to hemodialysis, hemofiltration provides high clearances for large solutes but its effect on protein-bound solutes has been largely ignored.
Methods: Standard clinical systems were used to remove test solutes from a reservoir containing artificial plasma. Clearances of the protein-bound solutes phenol red (C(PR)) and indican (C(IN)) were compared to clearances of urea (C(UREA)) during hemofiltration and hemodiafiltration. A mathematical model was developed to predict clearances from values for plasma flow Q(p), dialysate flow Q(d), ultrafiltration rate Q(f), filter size and the extent of solute binding to albumin.
Results: When hemofiltration was performed with Q(p) 150 mL/min and Q(f) 17 mL/min, clearance values were C(PR) 1.0 +/- 0.1 mL/min; C(IN) 3.7 +/- 0.5 mL/min; and C(UREA) 14 +/- 1 mL/min. The clearance of the protein-bound solutes was approximately equal to the solute-free fraction multiplied by the ultrafiltration rate corrected for the effect of predilution. Addition of Q(d) 42 mL/min to provide HDF while Q(p) remained 150 mL/min resulted in proportional increases in the clearance of protein-bound solutes and urea. In contrast, the clearance of protein-bound solutes relative to urea increased when hemodiafiltration was performed using a larger filter and increasing Q(d) to 300 mL/min while Q(p) was lowered to 50 mL/min. The pattern of observed results was accurately predicted by mathematical modeling.
Conclusion: In vitro measurements and mathematical modeling indicate that CVVH provides very limited clearance of protein-bound solutes. Continuous venous hemodiafiltration (CVVHDF) increases the clearance of protein-bound solutes relative to urea only when dialysate flow rate and filter size are increased above values now commonly employed.