REmoval of Free Light Chains. A COmpaRison of Three Different dialyzERs

Background. Immunoglobulin light chains are classified as middle molecule uremic toxins together with β2M and parathyroid hormon. They have a mean molecular weight of 25,000 daltons for monomers (k free light chains) and approximately 50,000 daltons for dimers (lambda free light chains).

Raised serum levels of polyclonal free light chains can impair neutrophil function in terms of: 1) inhibited chemotactic movement; 2) reduced activation of glucose uptake; 3) inhibited apoptosis.

Immunoglobulin light chains are able to interact with B lymphocyte membranes leading to the activation of transmembrane signaling . The ensuing impairment of neutrophil function can contribute to the chronic inflammation state of uremic patients and to the increased risk of bacterial infection. Desjardins et al. demonstrated an association between free light chain levels and vascular calcification progression in chronic kidney disease patients. Nonetheless, serum free light chains in patients with chronic kidney disease are associated with the risk of end stage renal disease and death.

It seems that high flux PMMA membranes significantly reduce the light chain level, presumably due to adsorption. This behavior has led to the use of PMMA in the removal of monoclonal free light chains during multiple myeloma.

Preliminary reports describe a significant reduction in free light chains also during hemodiafiltration with reinfusion of endogenous ultrafiltrate (on-line HFR). A recent paper by Borrelli et al. reported an improvement in chronic inflammation after chronic online HFR in dialysis patients.

The aim of this crossover observational study was to compare free light chain removal by three different hemodialysis filters in a cohort of chronic dialysis patients not affected by multiple myeloma.

Methods. One hundred and sixty-three patients on chronic hemodialysis at the Nephrology, Dialysis and Transplantation Unit of S.Orsola University Hospital in Bologna were considered for the study.

The inclusion criteria were free light chain values >100 mg/L for k chains and >50 mg/L for lambda chains. These levels were chosen arbitrarily because no cut-off levels are available in the literature for the assessment of lowered serum free light chain values during hemodialysis in patients with end-stage renal disease. Other inclusion criteria were: age >18 years, absence or <200 ml/die residual diuresis, fistula or central venous catheter with blood flow >250 ml/min. Exclusion criteria were: intradialytic hypotension during bicarbonate dialysis, multiple myeloma requiring the double PMMA filter application, poly-allergy, HIV positivity. The filters used during the study were: i) HFR17 (Bellco, Mirandola, Italy), a double chamber filter used for online HFR. The first part of the filter consisted in a polyphenylene high flux hemofilter with an ultrafiltration coefficient (Kuf) of 28 ml/h/mmHg, a surface area of 0.7 m2 and a membrane cut-off value of 35,000 daltons. The endogenous ultrafiltrate rate is obtained automatically by means of the transmembrane pressure levels in the hemofilter. These are calculated from two pressure sensors: the first is on the arterial bubble chamber and the second before the roller pump of the ultrafiltrate. The ultrafiltrate is driven from this hemofilter to a 40 g neutral styrenic resin that allows an adsorbing area of 28,000 m2. After adsorption, the ultrafiltrate is added to the whole blood that, in turn, passes into the second HFR17 filter, a polyphenylene low flux filter (Kuf 13 ml/h/mmHg, surface area 1.7 m2) where the weight loss and diffusive depuration take place. ii) PMMA (Toray Filtryzer BK-F, Tokyo, Japan) with a surface area of 2.1 m2, a membrane cut-off value of 20,000 daltons and an ultrafiltration coefficient (Kuf) of 26 ml/h/mmHg. iii) A conventional polysulfone membrane (Fresenius F7HPS, Bad Homburg, Germany) with a cut-off of 11,500 daltons, a Kuf of 16 ml/h/mmHg and a surface area of 1.7 m2 was used as a control dialyzer.

The patients enrolled were randomized into two groups of treatments lasting six weeks each. Group A: 1st and 2nd weeks with bicarbonate dialysis and filter PMMA, 3rd and 4th weeks with filter HFR17, 5th and 6th weeks with bicarbonate dialysis filter F7. Group B: 1st and 2nd weeks with filter HFR17, 3rd and 4th weeks with bicarbonate dialysis and filter PMMA, 5th and 6th weeks with bicarbonate dialysis filter F7. The dialysis sessions carried out during weeks 1, 3 and 5 were considered washout sessions between weeks 2, 4 and 6 when the assessment of lambda and k light chains, β2M, C reactive protein and albumin was scheduled. Week number 1 was the washout period between the usual dialytic treatment of the patients enrolled and the beginning of the study. The checking dialysis session was the midweek session and the blood was drawn on starting dialysis (time 0), at two hours (time 120') and at dialysis end (time 240'). All the dialysis sessions lasted four hours. Mean blood flow was 310±30 ml/min, the mean ultrafiltration rate was 700±200 ml/h, during HFR the endogenous ultrafiltrate rate was 2.3±0.4 ml/h. Dialysate flow was 500 ml/min. Low molecular weight heparin enoxaparin (Clexane®, Sanofi, Milan, Italy) was used for anticoagulation of the extracorporeal circuit. Doses of 2000 IU (patients <50 kg of body weight), 4000 IU (patients between 50 to 90 kg of body weight) or 6000 IU (patients >90 kg of body weight) were administered in a single bolus on starting dialysis. Fresenius 5008 and Bellco Flexya dialysis machines were used. The concentrations of k and lambda light chains and β2M were measured by nephelometry (kit Freelite k/lambda, The Binding Site Group Ltd., Birmingham, United Kingdom; IIMAGE/IMMAGE 800 Beckman Coulter instrument, Brea California USA, Beckman Coulter β2M kit). Normal values: k light chains 3.3 -19.9 mg/L, lambda light chains 5.7-26.3 mg/L, β2M 0.7 - 2 mg/L. Molecular weights: k light chains: 22,500 daltons; lambda light chains: 45,000 daltons; β2M: 11,818 daltons. The reference range for patients with normal kidney function was considered between 0.26 and 1.65 according to Bourguignon et al.. CRP concentration was measured by turbidimetry (CRPLX, Tina-quant C-Reactive-Protein; Roche/Hitachi 902 analyzer). CRP normal value < 0.8 mg/dl, molecular weight 120,000 daltons. Albumin was assessed using the common laboratory method. The reduction rate per session (RRs) was calculated as follows:

RRs = (Cpre- Cpost-corr)/ Cpre x 100 (1) where Cpre is the predialysis solute level, and Cpost-corr is the post-dialysis solute concentration. The values measured during dialysis were corrected for hemoconcentration due to the patient's weight loss assuming a unicompartmental behavior of light chains described by the following formula: Cpost-corr = Cpost/ {1+ [DeltaBW/(0.2xBWpost)]} (2) where Cpost-corr is the post-dialysis solute concentration and Cpost-corr is the concentration of light chains corrected for the hemoconcentration, Delta body weight (DeltaBW) is the weight subtracted during dialysis, and BW (body weight) is the body weight at the end of dialysis.

Statistical analyses. Statistical analysis was performed using the "Statistical package for the social sciences (SPSS)" statistical software. The data are presented as mean ± standard error of the mean. The Shapiro-Wilk test, suitable for small populations, showed a non-normal distribution of the variables, for which the non-parametric Friedman test was used. The p value <0.05 was considered normal. The Wilcoxon Signed Rank Test was subsequently used to see which pairwise comparisons resulted statistically different. In addition, as a multiple comparison was done, we applied the Bonferroni correction, thereby setting a p value = 0.05 / 3 = 0.0167. For each of the sampling times 0.120 and 240 minutes, we compared the results and the RRs obtained with the different filters (HFR17, PMMA, F7).