Share this article on:

Simplified Measurement of Middle Molecular Toxin Removal During Continuous Therapies

Lebau, Olaf; Kleinekofort, Wolfgang

doi: 10.1097/MAT.0b013e31829ed9d8
Kidney Support/Dialysis/Vascular Access

The removal of middle molecules during continuous extracorporeal detoxification therapies is a scientific field of increasing attention. However, it is difficult to find suitable and inexpensive markers and methods regarding the in vitro clearance of molecules in a weight range between 10 and 20 kDa. We present an easy and reliable cytochrome C–based photometric method to compare the in vitro clearance of different hemofilters during hemodialysis therapy. Cytochrome C is an inexpensive globular protein that shows a relative absorption maximum at λ = 410 nm, allowing photometric concentration assessment in the dialysate outflow line. Various hemofilters were evaluated to assess the removal of cytochrome C during simulated continuous venovenous hemodialysis. Although reported sieving coefficients for globular proteins and surface area of all hemofilters were similar, clearance differences of up to 55% were observable at blood and dialysate flows of 150 ml/min and 4,000 ml/h, respectively. We found that even in a strict hemodialysis setting with moderate blood flow, a remarkable diffusive cytochrome C clearance can be achieved. Hereby, diffusion is eventually supported by transmembrane flow and backflow in the proximal and distal sections of the hollow fiber of a hemofilter.

From the Department of Physical Technology, RheinMain University, Ruesselsheim, Germany.

Submitted for consideration November 2012; accepted for publication in revised form June 2013.

Disclosures: In 2011, Dr. Kleinekofort received honoraria from Fresenius Medical Care AG. The authors have no other conflicts of interest to report.

Reprint Requests: Wolfgang Kleinekofort, Department of Physical Technology, RheinMain University, Am Brueckweg 26, 65428 Ruesselsheim, Germany. Email:

Continuous venovenous hemodialysis (CVVHD) and continuous venovenous hemofiltration (CVVH) are widely accepted as standard therapies for patients with symptoms of fluid overload, intoxication, and kidney failure. The range of molecular weights (MWs) that are removed during CVVHD and CVVH includes smaller molecules <500 Da and middle molecules between 500 and 20,000 Da. Toxins with a MW >20,000 Da are generally classified as “larger solutes.”1 Although the benefits of middle molecular toxin removal are widely accepted, there is no easy way to assess the clearance of larger molecules in a given treatment setting. Clearance of these solutes by CVVH is dependent on the ultrafiltration rate and the sieving coefficient of the hemofilter for a given substance, whereas clearance during CVVHD mainly depends on blood flow QB and dialysate flow rate QD.2 With increasing solute size, diffusivity decreases and the solute transport across the membrane decreases according to Fick’s first law. However, Messer et al.3 could show that, depending on the treatment parameter settings, higher clearance of middle molecules can be achieved with CVVHD than with CVVH. Thus, the common assumption underlying continuous extracorporeal therapies that a convective therapy always shows better clearances regarding the middle MW toxins than a diffusive therapy remains unclear.

When pursuing a strategy to improve the removal of middle molecules, hemofilters with a better dialyzing performance remain a further option in addition to adjusting the treatment parameters and switching the therapy mode.4 However, a direct comparison of the technical filter data such as effective surface area, ultrafiltration coefficient, or cutoff point will not always help to find the best filter for the removal of middle molecules. Even for a strict CVVHD setting, a convective flow through the membrane of the dialyzer can be postulated—known as internal filtration and driven by the differences among dialyzer inlet pressure (DIP), transmembrane pressure, and venous pressure (PV). The intraluminal pressure drop along the hollow fibers of the hemofilter leads to hidden fluid shifts from the blood compartment to the dialysate side and vice versa and enhances convective solute transport.5,6 It follows that besides membrane thickness, pore size, and surface area of the hemofilter, the flow resistance and pressure gradient along the hollow fiber at a given blood flow are important parameters.7 Thus, the flow resistance of the hemofilter mainly depends on the following: 1) total number of hollow fibers, 2) inner diameter of the hollow fibers, and 3) fiber length. It is somehow surprising that these data cannot be found routinely within the technical data sheets of the manufacturers. As a consequence, direct benchmarking of different hemofilters regarding the clearance of middle molecules cannot be done on the basis of limited technical data and thus has to be performed with an in vitro test setting. We present a simplified method for the assessment of middle molecular toxin removal during continuous therapies.

Back to Top | Article Outline

Materials and Methods

Total aqueous in vitro clearances for cytochrome C as tracer molecule during simulated dialysis with four different hemofilters were measured. In contrast to partly linear dextran or poly(ethylene oxide)n tracers which are sometimes used for the assessment of middle molecular clearances, cytochrome C (MW 12.4 kDa) is a monomeric globular, highly water-soluble protein. It consists of a 104 amino acid chain with a single heme group, which is covalently attached to Cys14 and Cys17. The heme group shows a relative absorption maximum at λ = 410 nm, allowing photometric concentration assessment during extracorporeal therapy, and solubility in water is approximately 100 g/L.8 We took this tracer molecule as it is cheap, has a high molar extinction coefficient, nearly round shape, and minimal binding to blood tubing and hemofilters without negative surface charge. It has a similar MW like β2-microglobulin and myoglobin and thus could serve as an inexpensive surrogate marker for middle molecules between 10 and 20 kDa. However, a direct comparison of membrane permeabilities with cytochrome C, myoglobin, and β2-microglobulin has not been made. Although this is a major methodological deficiency of our study, a relative comparison of the middle molecular clearance of different hemofilters at identical treatment parameters is still possible.

All measurements were performed on a Fresenius Medical Care multiFiltrate device (Bad Homburg, Germany) in CVVHD mode at a constant blood flow of 150 ml/min. Net ultrafiltration was set to zero to avoid additional convective transport effects overlaying diffusion and internal filtration within the hemofilter. To calculate the magnitude of internal filtration, DIP and PV were recorded for each hemofilter. Dialysate flow QD was varied between 1,000 and 4,000 ml/h in steps of 1,000 ml/h. The temperature of all fluids was adjusted to (294 ± 0.8) K. Table 1 summarizes the hemofilters used for the described experiment.

Fifteen liters of dialysate (HF11, Fresenius Medical Care) and 3.72 g cytochrome C from equine heart (Cat-Nr. 250600; Calbiochem®; Merck KGaA, Darmstadt, Germany) were mixed and stirred continuously in a canister to achieve a start concentration of c0 = 2 × 10−5 mol/L. The absorbance of the stock solution was checked with a Varian Cary 50 photometer (Varian Medical Systems Inc., Palo Alto, CA), and the canister was connected to the arterial and venous blood line of the dialysis machine. After removing air from the extracorporeal system, the blood and dialysate side of the hemofilter of interest were flushed in recirculation mode for a minimum of 10 min with the protein solution. This procedure was chosen to account for possible protein absorption at the surface of the hemofilter membranes and thus to allow protein saturation of the membrane material.

During recirculation, the absorbance was measured continuously to account for potential absorption effects of the inner surfaces of the extracorporeal circuit and hemofilter. The preparation phase was followed by a data acquisition period starting with a cytochrome C–free HF11 dialysate bag connected to the dialysate inflow line and an empty bag connected to the venous blood line (single pass mode). Absorbance at λ = 410 nm as a function of time was measured continuously every 10 s with a flow cuvette integrated in the dialysate outflow line. After reaching a stable plateau that took approximately 8–10 min following dialysate connection or QD changes, the absorbance at steady state conditions was assessed over a period of 5 min (i.e., 30 data points). Separated measurements were performed at dialysate flows of QD = 1,000; 2,000; 3,000; and 4,000 ml/h, and mass balance calculation was used as plausibility check for the measurements. Clearance calculation was made if the deviation between pre-filter protein concentration times blood flow (Cin×QB) and the sum of post-filter concentration times blood flow (Cout×QB) and dialysate outflow concentration times dialysate flow (CDia×QD) was <5%.

A total of three clearance procedures were recorded with each single hemofilter, and three different filters of one type were investigated separately. This results in nine independent assessments for each clearance value on a database of 270 single absorbance measurements. Results are presented as mean clearance K ± standard deviation. All fits and calculations were performed using Microsoft Excel. During all measurements, the net ultrafiltration was set to zero, and mass balance on the dialysate was controlled gravimetrically by the multiFiltrate device. A scheme of the experimental setup is shown in Figure 1.

As measurements were performed without convection (no ultrafiltration), calculation of all clearances K were based on the following simplified equation9:

where K is the clearance in ml/min, QD is the dialysate flow in ml/min, C(Dialysate) and C(Blood) represent the concentration of cytochrome C in the dialysate outflow line and the blood inflow line, respectively. According to the Lambert-Beer law, absorbance A is the product of extinction coefficient ε, solute concentration c, and cuvette length l. As extinction coefficient and cuvette length are constant during measurements, concentration in Equation 1 can be replaced by absorbance, and the clearance K can be calculated easily from photometric data:

The diffusive transport of cytochrome C can be assessed on the basis of Fick’s first law:

The term j represents solute flux in mol/s, D is the translational diffusion coefficient in m2/s, dc/dx is the concentration gradient across the membrane in mol/m4, and A is the surface area of the hemofilter in m2.10 The translational diffusion coefficient D of cytochrome C as globular monomeric protein was calculated with the boundary integral method of solving the steady state Stokes equation.11

Back to Top | Article Outline


All dialyzers were optically clean from cytochrome C at the end of the sessions, demonstrating only minimal protein absorption within the hollow fibers. This observation was supported by the measurement of stable absorbance values at the arterial blood side during initial recirculation. At a blood flow of 150 ml/min, DIP was 143, 142, 151, and 152 mbar for EMiC2, AV1000S, HCO1100, and SepteX, and PV was 121, 121, 121, and 122 mbar. It follows that the pressure difference along the hollow fiber as driving force for hidden convective fluid shifts amounts to 22, 21, 30, and 30 mbar, respectively. Mean dialysate pressures were between 132 and 137 mbar. Due to the very low flow resistance within the dialysate flow pathway, a pressure gradient between dialysate inlet and dialysate outlet in a dialysate flow range between 1,000 and 4,000 ml/h was not observable. Assuming an ultrafiltration coefficient of k(UF) = 100 ml/(mbar·h) and a dialysate pressure as arithmetic average between DIP and PV, the internal filtration amounts to 18, 17.5, 25, and 25 ml/min for EMiC2, AV1000S, HCO1100, and SepteX.12,13 Thus, internal filtration can be estimated to contribute between 12% and 17% to the measured middle molecular clearance at a blood flow of 150 ml/min.

Figure 2 and Table 2 summarize the clearance K of cytochrome C as a function of dialysate flow for the different hemofilter types during CVVHD. The curves represent quadratic fits to the data points with a second-order polynome of type y = a + bx + cx2. The maximum theoretically achievable clearance for cytochrome C K(max) would be reached if the concentrations at the arterial blood side and the dialysate waste side were identical. Calculated K(max) values according to Equation 2 are 16.7 ml/min (QD = 1,000 ml/h), 33.3 ml/min (QD = 2,000 ml/h), 50.0 ml/min (QD = 3,000 ml/h), and 66.7 ml/min (QD = 4,000 ml/h), respectively. All hemofilters allow for the passage of cytochrome C through the membrane, and cytochrome C clearance strongly depends on dialysate flow.

A remarkable difference in practically achievable clearance is observable between the hemofilters. Over the whole range, EMiC2 shows the best performance regarding cytochrome C removal, followed by HCO1100 and AV1000S. Under identical conditions, clearance of the SepteX hemofilter is <50% of the K value measured with EMiC2 (Table 2). Although sieving coefficients for myoglobin of EMiC2, HCO1100, and SepteX are similar (Table 2), divergences in cytochrome C clearance are evident. The results are not fully consistent with the data sheets of some hemofilters. For example, CVVHD clearance K (QD = 1,000 ml/h, ultrafiltration of 0 ml/h) for myoglobin in plasma for the SepteX filter should be K = 13 ml/min at QB = 80 ml/min and K = 18 ml/min at QB = 200 ml/min.

Back to Top | Article Outline


We present an easy and reliable cytochrome C–based photometric check to compare the in vitro clearance of different hemofilter types during simulated hemodialysis therapy. Even in a strict hemodialysis setting with moderate blood flow, a remarkable diffusive clearance can be achieved, partly supported by unobservable transmembrane flow and backflow in the proximal and distal section of the hollow fiber of a hemofilter (internal filtration). This result is in good agreement with the observation that adding a diffusive component to therapy also increases effectivity of purely convective therapies.14,15 The finding that the overall effectivity of continuous hemodialysis and continuous hemofiltration is comparable for the removal of both small and middle molecules supports our hypothesis.16 The assumption that only convective therapies are optimal for a high clearance of middle molecules may not always be true. Diffusive elimination of middle molecules can serve as an explanation for the observed middle molecular clearance: low dialysate flows that are typical for CVVHD settings offer enough time for the dialysate to equilibrate with the blood side, and the observed high clearance values seem to be the result of a diffusive protein transport rather than a convective protein mobilization. Our assumption is congruent with the finding that high-flux membranes are also high diffusion membranes.6,17

The calculated translational diffusion coefficient D of cytochrome C as globular monomeric protein at T = 294 K is approximately D = (1.20 ± 0.24)×10−10 m2/s, and the result is in good agreement with experimentally determined values of proteins with similar mass.18 Even for a high dialysate flow of QD = 4,000 ml/h, blood flow is 2.25 times QD. Thus, for the ease of calculation of the diffusive transmembrane protein flux j, protein concentrations at the blood side were assumed to be nearly constant. A calculation of the protein flux through a 1.8 m2 membrane with 35 μm thickness (Fresenius Medical Care hemofilters) and a cytochrome C concentration of c = 2 × 10−5 mol/L at the blood side according to Fick’s first law (Equation 3) leads to a maximum transmembrane flux of j = 1.23 × 10−7 mol/s. The above calculation is made under the assumption that the concentration gradient between blood side and dialysate side is always maximal, which only holds true for infinite blood flow. Assuming a more or less linear decrease in the concentration gradient along the dialysate pathway, the average cytochrome C concentration in the filter can be roughly estimated to 50% of the measured dialysate outflow concentration. It follows that a realistic mean transmembrane protein flux in our experimental setting is approximately j = 6.15 × 10−8 mol/s. Thus, it would take approximately 325 s for an arbitrary dialysate volume to fully equilibrate with the blood side. Hydraulic residence times of dialysate in a filter housing with a priming volume of 150 ml on the dialysate side are 540 s at a dialysate flow of 1,000 ml/h, 270 s (QD = 2,000 ml/h), 180 s (QD = 3,000 ml/h), and 135 s (D = 4,000 ml/h), respectively. Within these time limits, a nearly full diffusive equilibration between blood and dialysate side for a 1.8 m2 membrane would be expected for dialysate flows below QD = 1,660 ml/h. For a 1.1 m2 membrane with 50 μm thickness (Gambro hemofilters), a nearly full diffusive equilibration should only be achievable at dialysate flows below QD = 712 ml/h. None of the hemofilter reaches a clearance of 100% (EMiC2: 84%, HCO 1100: 74%, AV1000S: 66%, and SepteX: 35% for QD = 1,000 ml/h).

As cytochrome C diffusivity is also a function of temperature, the results have to be scaled up to an experimental setting at body temperature of 310 K to bridge the gap between laboratory and dialysis unit. According to the Stokes-Einstein equation, D increases with increasing temperature.19 Assuming a nearly stable viscosity of water between 294 and 310 K, the translational diffusion coefficient D for cytochrome C at body temperature is D = (1.26 ± 0.24) × 10−10 m2/s. Within the limits of accuracy for the diffusion coefficient of cytochrome C (±20%), the temperature has only a minor impact on D for a protein with a mass of 12.4 kDa in a temperature range between 290 and 310 K. The presented method for the assessment of middle molecular toxin removal indicates that the real achievable clearance can deviate from the values presented in the data sheets of hemofilter manufacturers. Although the presented setup allows measurement of clearance under carefully controlled conditions, the applicability of the results to the clinical situation is limited because of the use of a plasma-free solution on the blood side. All the membranes included in this study will absorb plasma proteins to a greater or lesser extent and those adsorbed proteins will affect membrane permeability properties, with considerable differences between membranes fabricated from nominally similar materials. To yield clinically relevant information, the experiments should be performed with a protein-containing solution on the blood side of the membrane, either by using plasma or by adding albumin at physiologic concentrations to the test fluid. Although a full transferability of the presented in vitro results is limited, the presented photometric measurement of aqueous cytochrome C clearance still allows a relative comparison of hemofilters regarding middle molecular toxin removal.

Back to Top | Article Outline


We found that commercially available hemofilters differ with respect to the clearance of cytochrome C, which could serve as an inexpensive surrogate marker for β2-microglobulin and myoglobin in the future. In clinical practice, dialysis efficiency regarding middle molecular toxin reduction is usually not assessed, and the hemofilter is chosen mainly by habit or availability. Our in vitro data show that the clearance of middle molecules is strongly influenced by the hemofilter choice. Commercially available hemofilters differ with respect to their diffusive middle-molecule clearance. With regard to the maximum achievable clearance of middle molecules in the 10–20 kDa range like β2-microglobulin and myoglobin, the “real” effectivity of the filters at a typical dialysate flow of 2,000 ml/h is between 76% (EMiC2) and 35% (SepteX) of the maximum theoretically achievable clearance. In summary, a hemofilter with a low performance in the middle molecular range will lower the clearance of toxins with a MW similar to cytochrome C.

Back to Top | Article Outline


1. Vanholder R, De Smet R, Glorieux G, et al.European Uremic Toxin Work Group (EUTox). Review on uremic toxins: Classification, concentration, and interindividual variability. Kidney Int. 2003;63:1934–1943
2. Matzke GR, Frye RF, Joy MS, Palevsky PM. Determinants of ceftriaxone clearance by continuous venovenous hemofiltration and hemodialysis. Pharmacotherapy. 2000;20:635–643
3. Messer J, Mulcahy B, Fissell WH. Middle-molecule clearance in CRRT: In vitro convection, diffusion and dialyzer area. ASAIO J. 2009;55:224–226
4. Uhlenbusch-Körver I, Bonnie-Schorn E, Grassmann E, Vienken J Understanding Membranes and Dialysers. 2004 Lengerich, Germany Pabst:pp. 103–172
5. Leypoldt JK, Schmidt B, Gurland HJ. Net ultrafiltration may not eliminate backfiltration during hemodialysis with highly permeable membranes. Artif Organs. 1991;15:164–170
6. Maeda K, Shinzato T. AU Importance of convection in artificial kidney treatment, Contributions to Nephrology—Effective Hemofiltration: New Methods. 1994;vol 108 Basel Karger:pp. 53–70
7. Dellanna F, Wuepper A, Baldamus CA. Internal filtration—Advantage in haemodialysis? Nephrol Dial Transplant. 1996;11(suppl 2):83–86
8. Skulachev VP. Cytochrome C in the apoptotic and antioxidant cascades. FEBS letters. 1988;423:275–280
9. Hoenich NA, Woffindin C, Ronco CJacobs C, Kjellstrand CM, Koch KM, Winchester JF. Haemodialysers and associated devices, in Replacement of Renal Function by Dialysis. 19964th ed Dordrecht Kluwer Academic Publishers:pp. 188–230
10. Cussler EL Diffusion—Mass Transfer in Fluid Systems. 1997 Cambridge/New York Cambridge University Press
11. Brune D, Kim S. Predicting protein diffusion coefficients. Proc Natl Acad Sci U S A. 1993;90:3835–3839
12. Boschetti-de-Fierro A, Voigt M, Storr M, Krause B. Extended characterization of a new class of membranes for blood purification: The high cut-off membranes. Int J Artif Organs. May 10, 2013 [Epub ahead of print]
13. Ward RA. Protein-leaking membranes for hemodialysis: A new class of membranes in search of an application? J Am Soc Nephrol. 2005;16:2421–2430
14. Saudan P, Niederberger M, De Seigneux S, et al. Adding a dialysis dose to continuous hemofiltration increases survival in patients with acute renal failure. Kidney Int. 2006;70:1312–1317
15. Hofmann CL, Fissell WH. Middle-molecule clearance at 20 and 35 ml/kg/h in continuous venovenous hemodiafiltration. Blood Purif. 2010;29:259–263
16. Ricci Z, Ronco C, Bachetoni A, et al. Solute removal during continuous renal replacement therapy in critically ill patients: Convection versus diffusion. Crit Care. 2006;10:R67
17. Naitoh A, Tatsuguchi T, Okada M, Ohmura T, Sakai K. Removal of beta-2-microglobulin by diffusion alone is feasible using highly permeable dialysis membranes. ASAIO Trans. 1988;34:630–634
18. Venable RM, Pastor RW. Frictional models for stochastic simulations of proteins. Biopolymers. 1988;27:1001–1014
19. Einstein A. Über die von der molekularkinetischen Theorie der Wärme geforderte Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen. Annalen der Physik. 1905;17:549–560

hemodialysis; clearance; middle molecules

Copyright © 2013 by the American Society for Artificial Internal Organs