To the Editors:
In a recent paper by Steiner et al. it was suggested that the molar ratio of bilirubin to albumin is an important factor guiding bilirubin removal during albumin dialysis.1 This conclusion is mainly based on in vitro observations and is, in our opinion, too much of a simplification of the in vivo situation.
The only mode of movement of albumin-bound toxins across an albumin-impermeable membrane is diffusion of their free component. The magnitude of this diffusive transport is determined by the dialyzer mass transfer area coefficient (MTAC) for that solute, and the free concentration gradient between the blood and dialysate compartment.2 Factors affecting the MTAC include the membrane material and thickness and the pore size. Likely a result of the albumin coating, the MARS-flux membrane (Teraklin AG, Rostock, Germany) has an approximately four-fold higher MTAC compared with a high-flux polysulfone membrane, at least regarding unconjugated bilirubin.3 The concentration of the free component not only depends on the molar ratio of the toxin to albumin but also on its affinity to the carrier molecule (binding constant) and the interaction with other protein-bound substances.4 The free concentration gradient can be increased by increasing the dialysate flow2 or by adding albumin as a carrier to the dialysate, as is performed during albumin dialysis.1
Figure 6 from the paper by Steiner et al. suggests a transmembrane transport of unconjugated bilirubin that is almost equal to the transmembrane transport of bromosulphtalein.1 This should be considered a somewhat unexpected finding, taking into account the striking difference in affinity of both compounds for the albumin molecule.
As part of a recent study in which we evaluated the detoxifying kinetics of the MARS membrane,5 we followed the time course of the concentration of total bilirubin (tB), bile acids (BAs), and albumin (Alb) in both the blood and dialysate compartment. Ten patients (8 men, 57.1 ± 4.6 years) with acute or acute-on-chronic liver failure were included in this subanalysis. The closed-loop dialysate circuit was primed with 600 mL of a 200 g/L human serum albumin. The duration of the MARS treatment session ranged from 4 to 12 hours (median, 6 hours). The ultrafiltration rate was negligible in all sessions. Results are summarized in Table 1. At all time points the Alb concentration in the dialysate compartment was much lower than the concentration of the priming solution. This suggests that most of the Alb within the secondary circuit was not circulating but retained within the adsorbers. This retention proceeds at a slow pace during the dialysis session.
|Start||1 hour||4 hour||End|
|Alb (g/L)||29.26 ± 1.32||—||—||—|
|tB (mg/dL)||26.78 ± 4.31||23.01 ± 3.79||19.97 ± 2.96||16.70 ± 2.36|
|Molar ratio tB to Alb||1.06 ± 0.17||0.91 ± 0.15||0.79 ± 0.12||0.65 ± 0.09|
|BA (mmol/L)||0.144 ± 0.020||0.100 ± 0.017||0.080 ± 0.016||0.066 ± 0.013|
|Molar ratio BA to Alb||0.335 ± 0.053||0.232 ± 0.041||0.186 ± 0.043||0.150 ± 0.035|
|Alb (g/L)||—||92.05 ± 5.29||84.43 ± 4.83||80.81 ± 7.96|
|tB (mg/dL)||—||3.55 ± 0.88||3.49 ± 0.75||3.23 ± 0.71|
|Molar ratio tB to Alb||—||0.045 ± 0.012||0.043 ± 0.009||0.047 ± 0.010|
|BA (mmol/L)||—||0.081 ± 0.020||0.096 ± 0.023||0.093 ± 0.26|
|Molar ratio BA to Alb||—||0.056 ± 0.012||0.066 ± 0.013||0.070 ± 0.014|
At all time points, the molar ratio of tB to Alb was approximately 10-fold lower in the dialysate circuit compared with the blood compartment. This gradient is much more pronounced than observed in the in vitro experiments by Steiner et al..1, 3 Despite the preservation of this important gradient for the length of the albumin dialysis session (Fig. 1), the clearance of tB declines over time and becomes negligible and even negative after 6 hours.5 Notwithstanding the much smaller molar ratio gradient (Fig. 2), at all time points the blood clearance of BA was substantially higher compared with total bilirubin. Furthermore, the clearance of BA remained stable for the duration of the session.5
The existence of subfractions of bilirubin with different binding properties likely explains these apparently controversial findings. It has been demonstrated that in the in vivo situation, up to 60% to 80% of bilirubin binds to albumin by tight, presumably covalent, links. Covalently bound bilirubin becomes more important in patients with prolonged cholestasis.6 Whereas MARS succeeds in clearing free bilirubin and bilirubin bound to albumin by noncovalent forces (Van der Waals forces), it likely fails to eliminate covalently bound bilirubin.
In conclusion, our data illustrate that the molar ratio of bilirubin to albumin is not the only factor guiding transmembrane transport of bilirubin during albumin dialysis. The relation of the different subfractions to the overall bilirubin level as well as other (unknown) factors should be accounted for. A better knowledge of these regulating factors may help identify patients who will benefit most from albumin dialysis.