Removal of Middle Molecules and Dialytic Albumin Loss: A Cross-over Study of Medium Cutoff and High-Flux Membranes with Hemodialysis and Hemodiafiltration

Abstract

Key Points

• HDF and MCO have shown greater clearance of middle-size uremic solutes in comparison with HF dialyzers; MCO has never been studied in HDF.

• MCO in HDF does not increase the clearance of B2M and results in a higher loss of albumin.

Background

Middle molecule removal and albumin loss have been studied in medium cutoff (MCO) membranes on hemodialysis (HD). It is unknown whether hemodiafiltration (HDF) with MCO membranes provides additional benefit. We aimed to compare the removal of small solutes and β2-microglobulin (B2M), albumin, and total proteins between MCO and high-flux (HFX) membranes with both HD and HDF, respectively.

Methods

The cross-over study comprised 4 weeks, one each with postdilutional HDF using HFX (HFX-HDF), MCO (MCO-HDF), HD with HFX (HFX-HD), and MCO (MCO-HD). MCO and HFX differ with respect to several characteristics, including membrane composition, pore size distribution, and surface area (HFX, 2.5 m²; MCO, 1.7 m²). There were two study treatments per week, one after the long interdialytic interval and another midweek. Reduction ratios of vitamin B12, B2M, phosphate, uric acid, and urea corrected for hemoconcentration were computed. Dialysis albumin and total protein loss during the treatment were quantified from dialysate samples.

Results

Twelve anuric patients were studied (six female patients; 44±19 years; dialysis vintage 35.2±28 months). The blood flow was 369±23 ml/min, dialysate flow was 495±61 ml/min, and ultrafiltration volume was 2.8±0.74 L. No significant differences were found regarding the removal of B2M, vitamin B12, and water-soluble solutes between dialytic modalities and dialyzers. Albumin and total protein loss were significantly higher in MCO groups than HFX groups when compared with the same modality. HDF groups had significantly higher albumin and total protein loss than HD groups when compared with the same dialyzer. MCO-HDF showed the highest protein loss among all groups.

Conclusions

MCO-HD is not superior to HFX-HD and HFX-HDF for both middle molecule and water-soluble solute removal. Protein loss was more pronounced with MCO when compared with HFX on both HD and HDF modalities. MCO-HDF has no additional benefits regarding better removal of B2M but resulted in greater protein loss than MCO-HD.

Introduction

Since the 2000s, on-line hemodiafiltration (HDF), a dialytic modality that combines diffusive and convective solute removal in a single treatment session, has gained wide-spread acceptance in many countries.¹ Compared with standard high-flux hemodialysis (HFX-HD), HDF provides an improved clearance of uremic retention solutes with middle-to-high molecular weight.² Several high-quality studies, including randomized controlled trials, have demonstrated its safety and improved cardiovascular and mortality outcomes.^3–8

Depending on local circumstances, the implementation of HDF can be challenging because it requires ultrapure water, specific staff training, and comes with marginally higher treatment costs compared with HFX-HD. To avoid some of these challenges while providing still better clearance of middle-to-high molecular weight solutes, a new type of dialyzer membrane—called medium cutoff (MCO) membrane—was developed. Dialyzers with MCO membranes are used with conventional HD systems and aim to clear higher molecular weight solutes as effective as HDF. Compared with high-flux dialyzers, MCO membranes showed better clearance of uremic solutes with molecular weights between 15 and 40 kDa.^9,10 However, MCO membranes also increase dialytic albumin loss.^9,10 Notably, some studies comparing conventional HD in combination with MCO dialyzers (MCO-HD) with HFX-HDF showed no difference in middle molecule removal.^11,12 To the best of our knowledge, the combined use of an MCO dialyzer with HDF (MCO-HDF) has not been studied yet presumably because of the potential for an excessive dialytic loss of beneficial substances, most importantly albumin. Therefore, it is unknown whether MCO-HDF offers an enhanced removal of β2 microglobulin (B2M) without increasing the risk of clearing beneficial molecules. We aimed to compare the removal of uremic solutes and waste of nutritional and essential molecules between dialyzers with different membrane surface area, composition, and dialysis modality, namely HFX-HD, HFX-HDF, MCO-HD, and MCO-HDF.

Methods

Study Design

This was a prospective, cross-over study which included patients with kidney failure receiving HDF treatment at the National Institute of Cardiology Ignacio Chavez HDF clinic. MCO dialyzer (Theranova 400) was compared with standard high-flux hemodialyzer (FX CorDiax 120) used as reference both in HD and HDF modes. Inclusion criteria were older than 18 years, thrice-weekly dialysis with HD or HDF, absent residual kidney function, and hemoglobin levels above 7 g/dl. Most participants exercised on a stationary bicycle throughout dialysis as part of the routine treatment program in the facility.¹³

Study duration was 4 weeks and comprised two study treatments per week, one after the long interdialytic interval and another midweek. The study week sequence was as follows (Figure 1): week 1, HD combined with high-flux dialyzer (HFX-HD); week 2, HD with a MCO dialyzer (MCO-HD); week 3, HDF with a high-flux dialyzer (HFX-HDF); and week 4, HDF with a MCO dialyzer (MCO-HDF). A 5008S HD machine (Fresenius Medical Care, Waltham, MA) was used for all treatments. The high-flux dialyzer (FX CorDiax 120, Fresenius Medical Care, Waltham, MA) had an effective surface area of 2.5 m². The MCO dialyzer (Theranova 400 MCO dialyzer, Baxter Healthcare International, Deerfield, IL) had an effective surface area of 1.7 m². Treatment time was 240 minutes. Single-pool Kt/V was reported by the dialysis machine on the basis of on-line ionic clearance measurements. Hemoglobin levels and relative blood volume were measured in real time using Crit-Line (Fresenius Medical Care; Waltham, MA). Blood samples were collected in K2-EDTA tubes pretreatment and at minutes 10 and 230 from arterial (predialyzer) blood line and analyzed for B2M, vitamin B12, phosphate, urea, and uric acid. Dialysate outlet samples were collected at 10 minutes (actual collection time 10±1 minutes) and 230 minutes (actual collection time 228±6 minutes), respectively; samples were collected at 230 minutes and not 240 minutes because the 5008S HD machine does not provide dialysate at that time point. All dialysate samples were analyzed for albumin and total protein. Samples were stored at −80°C on site and then shipped on dry ice in <48 hours to the laboratory for further analysis.

Figure 1.

Reduction ratios (in percentage of initial concentration) during the treatments in the four study phases. Study phases: HFX-HD, high-flux hemodialysis; HFX-HDF, high-flux hemodiafiltration; MCO-HD, hemodialysis using a medium cutoff dialyzer membrane; MCO-HDF, hemodiafiltration using a medium cutoff dialyzer membrane.

The research was approved by the Research and Ethics Committee of the Instituto Nacional de Cardiologia Ignacio Chavez (number 19-103). Written informed consent was obtained from all participants. The study was conducted in accordance with the Declaration of Helsinki. The protocol was registered at ClinicalTrials.gov (NCT03938285).

Laboratory Measurements

Measurements were conducted at the Renal Research Institute, New York and Spectra laboratories, New Jersey. Urea and phosphate levels were quantified using spectrophotometry (Pentra C400 biochemistry analyzer; Horiba, Irvine, CA). Plates were read at 450 nm (Infinite M Plex plate reader; Tecan, Zurich, Switzerland). B2M was measured using Milliplex MAP magnet-bead immunoassay (Millipore Sigma, Burlington, MA). Fluorescence was then measured using a MAGPIX analyzer (Luminex, Austin, TX). Vitamin B12 was measured by using the Siemens ADVIA Centaur XP Immunoassay system (Siemens Medical Solutions USA, Inc., Malvern PA, product REF09544818).The concentrations of B2M at 230 minutes in plasma were corrected for hemoconcentration as follows¹⁴:

$${\mathrm{C}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}-\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}=\frac{{\mathrm{C}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}}{\left(1+\frac{{\mathrm{B}\mathrm{W}}_{\mathrm{P}\mathrm{r}\mathrm{e}}-{\mathrm{B}\mathrm{W}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}}{0.2\times {\mathrm{B}\mathrm{W}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}}\right)}$$

C_Post–corr is the corrected concentration; BW_Pre and BW_Post are patient's body weight before and after dialysis, respectively. C_Post is measured postdialysis plasma levels at 230 minutes.

Vitamin B12 is highly protein-bound in plasma,¹⁵ and its plasma level at 230 minutes was corrected with the method proposed by Schneditz et al.¹⁶:

$${\mathrm{C}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}-\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}}={\mathrm{C}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}\times {\mathrm{h}}_{\mathrm{p}}$$

$${\mathrm{h}}_{\mathrm{p}}=\frac{{\mathrm{H}}_1\left(100-{\mathrm{H}}_0\right)}{{\mathrm{H}}_0\left(100-{\mathrm{H}}_1\right)}$$

Where C_Post is the plasma concentration at 230 minutes; H₀ and H₁ are hematocrits at 0 minutes and 230 minutes.

Reduction ratios (RR) were calculated as follows:

$$\mathrm{R}\mathrm{R}=\frac{{\mathrm{C}}_{\mathrm{P}\mathrm{r}\mathrm{e}}-{\mathrm{C}}_{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}-\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}}}{{\mathrm{C}}_{\mathrm{P}\mathrm{r}\mathrm{e}}}\times 100$$

RR of solutes was first calculated for each patient on a treatment level. Then the average RR of middle molecule was calculated as the mean RR of two study sessions for each patient. If RR was available for one session only, that result was used. Skewed data were reported as medians (interquartile ranges; IQR) of these averages across patients for each modality. RR of water-soluble solutes was calculated with data from the session after the long interdialytic interval, instead of using the average of two sessions.

Dialysate total protein concentration was determined by using the Bradford protein assay (Thermo Fisher Cat 23200). Dialysate albumin concentration was measured using a microalbumin 2 CP cassette (REF 1300032563) on the Horiba Pentra C400 clinical chemistry analyzer. If the measured albumin concentration in dialysate samples were lower than 10 mg/L, samples were concentrated 16 times using 10K MWCO protein concentrators (Thermo Fisher Cat 88516) and measured again on Pentra C400. Considering concentration and automatic postdilution, the linear range of albumin measurement was from 0.6 to 3000 mg/L.

The cumulative albumin and total protein loss per treatment were calculated as follows:

$$\mathrm{C}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\hspace{0.17em}\mathrm{l}\mathrm{o}\mathrm{s}\mathrm{s}\hspace{0.17em}\mathrm{p}\mathrm{e}\mathrm{r}\hspace{0.17em}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}={\mathrm{M}}_{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{c}.}\times \left(\mathrm{Q}\mathrm{d}+\mathrm{U}\mathrm{F}\mathrm{R}\right)\times \mathrm{T}$$

Where M_Conc. (g/L) is the logarithmic mean of dialysate concentrations of either albumin or total protein at 10 minutes and 230 minutes, Qd is dialysate flow (L/min), UFR is ultrafiltration rate (L/min), and T is treatment time; 240 minutes was used for all treatment session for calculation. Outliers of cumulative albumin or total protein loss data (>1.5×IQR+third quartile or <first quartile −1.5×IQR) were removed from final data analysis.

Statistical Analysis

Data are presented as mean±SD or median (IQR). Dialysate outlet concentrations and RR differences were compared between study phases using one-way ANOVA and in predefined post hoc analyses using pairwise dependent samples t tests with Bonferroni correction. Comparisons of albumin and total protein levels between each study phase were performed using the linear mixed-effects model. Differences in dialysis treatment settings, including Qd, blood flow (Qb), substitution volume, ultrafiltration volume as surrogate of weight loss, and transmembrane pressure, were compared between study phases using the pairwise t test. A two-sided P-value < 5% was considered significant. Analyses were conducted using R (version 4.1.1; R Foundation for Statistical Computing, Vienna, Austria) (Table 1).

Table 1.

Study design

Monday	Tuesday	Wednesday	Thursday	Friday		Saturday
Study visits				Routine care
Session 1		Session 2		Nonstudy session
Participant 1–6	Participant 7–12	Participant 1–6	Participant 7–12	Participant 1–6	Participant 7–12

Results

All 12 patients (6 female patients; mean age 44±19 years; dialysis vintage 35.2±28 months) completed all eight study treatments.

Table 2 presents the treatment characteristics for each study phase. Table 3 presents convective volume-related variables. Supplemental Table 1 and Figure 1 show RRs of solutes during the treatments in the four study phases. There was no significant difference in removal of solutes between groups, just a tendency of higher removal in B2M with convective treatments. Supplemental Table 2 and Figure 2 show dialysate albumin and total protein concentrations at 10 and 230 minutes as well as estimated loss per treatment on the basis of these two time points, respectively. MCO dialyzer showed higher albumin loss per treatment than HFX dialyzer when compared with the same modality (P < 0.001, estimate= −3.23 for comparing MCO-HD with HFX-HD; P < 0.001, estimate= −4.37 for comparing MCO-HDF with HFX-HDF). HDF groups showed higher albumin loss than HD groups (P < 0.001, estimate=4.05 for comparing MCO-HD with MCO-HDF and 0.01 > P >0.001, estimate=3.00 for comparison between groups used HFX-HD with HFX-HDF).

Table 2.

Summary of dialysis characteristics for each study phase

Variables	MCO-HD	HFX-HD	MCO-HDF	HFX-HDF
Qb (ml/min)	367.1±23.9	371.7±27.1	371.6±15.9	365.7±25.5
Qd (ml/min)	538.3±52.0c	547.1±48.0	449.4±18.6b,d	448.8±29.9a
TMP (mm Hg)	29.6±4.5c	33.8±11.5	175.2±25.8b,d	168.7±22.3a
RBV minimum (%)	81.5±6.0	81.9±5.9	79.9±5.2	79.3±5.5
spKt/V	1.6±0.3c	1.7±0.5	1.8±0.4d	1.9±0.4
Intradialytic weight change (kg)	2.4±0.8	2.3±0.8	2.5±0.8	2.4±0.8

Data presented as mean±SD.

All comparisons executed by using the pairwise dependent sample t test. Study phases: MCO-HD, hemodialysis using a medium cutoff dialyzer membrane; HFX-HD, high-flux hemodialysis; MCO-HDF, hemodiafiltration using a medium cutoff dialyzer membrane; HFX-HDF, high-flux hemodiafiltration. Qb, blood flow; Qd, dialysate flow; TMP, transmembranous pressure; RBV, relative blood volume; UFV, ultrafiltration cutoff

Hypothesis testing:

^aP < 0.001 comparing HFX-HDF with HFX-HD.

^bP < 0.001 comparing MCO-HDF with HFX-HD.

^cP < 0.001 comparing MCO-HD with HFX-HDF.

^dP < 0.001 comparing MCO-HDF with MCO-HD.

Table 3.

Summary of convective volume-related variables for each study phase

Variables	MCO-HD	HFX-HD	MCO-HDF	HFX-HDF
Substitution volume (L)	n/a	n/a	24.2±3.5	27.0±4.1
Convective volume (L)	2.7±0.7c	2.7±0.8	27.0±3.2b,d	29.8±4.1a
Filtration fraction (%)	3.1±0.8c	3.1±1.1	30.2±2.9b,d	33.9±3.9a
Ultrafiltration volume (ml)	2730.3±697.3	2728.8±808.1	2795.1±776.9	2765.0±718.1
Quf (ml/min)	11.4±2.9	11.4±3.4	11.6±3.2	11.5±3.0
Qi (ml/min)	n/a	n/a	100.7±14.4	112.7±17.3

Data are presented as mean±SD.

All comparisons were executed by using the pairwise dependent sample t test. Study phases: MCO-HD, hemodialysis using a medium cutoff dialyzer membrane; HFX-HD, high-flux hemodialysis; MCO-HDF, hemodiafiltration using a medium cutoff dialyzer membrane; HFX-HDF, high-flux hemodiafiltration. Quf, ultrafiltration rate; Qi, infusion flow rate.

Hypothesis testing:

^aP < 0.001 comparing HFX-HDF with HFX-HD.

^bP < 0.001 comparing MCO-HDF with HFX-HD.

^cP < 0.001 comparing MCO-HD with HFX-HDF.

^dP < 0.001 comparing MCO-HDF with MCO-HD.

Figure 2.

Cumulative albumin and total protein loss during treatments for four study phases. Study phases: HFX-HD, high-flux hemodialysis; HFX-HDF, high-flux hemodiafiltration; MCO-HD, hemodialysis using a medium cutoff dialyzer membrane; MCO-HDF, hemodiafiltration using a medium cutoff dialyzer membrane.

Total protein results showed trends similar to those of albumin results. MCO dialyzer showed higher loss of total protein than HFX dialyzer when compared with the same modality (P < 0.001, estimates= −4.34 for comparing MCO-HD with HFX-HD; 0.01 > P > 0.001, estimates= −6.15 for comparing MCO-HDF with HFX-HDF). HDF groups showed higher total protein loss than HD groups (P < 0.001, estimates=6.13 for comparing MCO-HD with MCO-HDF and 0.01 > P > 0.001, estimates=4.43 for HFX-HD with HFX-HDF).

The albumin to total protein ratio (in percentage) across all samples detected in dialysate was 51.8% for MCO-HD, 9.5% for HFX-HD, 57.1% for MCO-HDF, and 33.5% for HFX-HDF. The albumin to total protein ratio in dialysate was higher for the MCO dialyzer than for the HFX dialyzer. (P < 0.001, estimates= −41.00 for MCO-HD compared with HFX-HD; P < 0.001, estimates= −16.64 for MCO-HDF compared with HFX-HDF). HDF compared with HD contributed to a higher percentage loss of albumin in HFX groups (P < 0.001, estimates=27.42, HFX-HD to HFX-HDF); however, no difference was found when comparing HD with HDF in MCO groups (P > 0.1, estimates=2.92).

Discussion

The main finding of our clinical cross-over study is that the use of an MCO dialyzer with HD (MCO-HD) was not associated with improved removal of B2M and water-soluble uremic solutes when compared with a high-flux dialyzer. Albumin and total protein loss were higher in MCO-HD than in HFX-HD. The use of MCO in HDF did not provide additional benefits for better B2M removal and was associated with increased protein loss compared with the other three treatment modalities. Regardless of dialytic modality, high-flux dialyzers did not differ in loss of vitamin B12 and small water-soluble uremic solutes. It is important to note that the MCO membrane was neither designed nor approved for use with HDF and that the MCO-HDF arm represents off-label use of the MCO membrane. We included this arm in this mechanistic study because it is not unusual that off-label use of devices or medicinal products provides novel insights.

In 2017, a new HD modality concept was built around the use of MCO membranes and forced internal filtration exchange by modified fibers engineering design. MCO membranes have demonstrated a higher removal of 25–45 kDa molecules and were proposed as a more efficient alternative to HDF.¹⁷ Laboratory and clinical trials of MCO dialyzers showed an increased clearance of middle molecules when compared with HD using a high-flux dialyzer.^9,10 Several studies reported decreased blood levels of middle molecules, such as λ and ĸ light chain, over the study course,^10,18–21 while others reported nonsignificant effects of MCO dialyzers on blood levels of cytokines (IL-6, IL-1β, IL-17, IL-10, and TNF) and other middle molecules, such as myoglobulin, FGF-23, and complement factor.^10,19–22 However, other studies showed no superiority of MCO in comparison with HDF regarding removal of B2M, cystatin C, myoglobin, prolactin, and α1-glycoprotein.¹²

Kirsch et al. reported 39 patients randomized to three prototype MCO dialyzers combined with HD (MCO-HD) or HDF with a high-flux dialyzer (HFX-HDF). This study showed a higher clearance of λ-free immunoglobulin light chains with MCO.²³ In other studies that evaluated the performance of MCO compared with HDF with high substitution volume and a filtration fraction >20%, MCO did not demonstrate superior clearance of urea, creatinine, B2M, myoglobin, prolactin, α1-microglobulin, and α1-acid glycoprotein.^11,24 There are reports in the literature suggesting that MCO-HD provides better removal of middle molecules than HFX-HD and sometimes equals the removal of middle molecules with HFX-HDF. Our study could not corroborate these findings regarding B2M.

Differences in the removal of small water-soluble molecules have not been reported in the literature. In our study, there was no difference between groups regarding clearance of low–molecular weight solutes. We did observe a tendency for greater removal of B2M with convective modalities. This result is expected because convection is the major driving force of middle molecule removal.^1,25 A Qb equal or greater to 350 ml/min and a substitution volume of 21 L (the prescription used in our study) can provide higher RR of the middle molecules (B2M, myoglobin, prolactin, a1 microglobulin, and a1 acid glycoprotein) than HDF treatment with lower Qb and substitution volume using an FX CorDiax 80 dialyzer.²⁶ Our study demonstrates that B2M removal is similar to an MCO membrane (surface area 1.7 m²) when compared with a HFX dialyzer with 2.5-m² membrane surface area, smaller pore size, and significantly lower albumin loss. We speculate that the similar clearance of small solutes and B2M by MCO-HD and HFX-HD despite a different surface area may be achieved through a back-transport phenomenon with MCO dialyzer design. Specifically designed studies are warranted to address this point. In other words, the results in our study demonstrate that a similar performance of Theranova 400 (surface area 1.7 m²) on B2M removal can be achieved when compared with a HFX dialyzer with a larger membrane surface area (2.5 m²) without a larger pore size and with a significantly higher albumin loss. These findings should be further addressed.

A large portion of vitamin B12 is protein-bound,¹⁵ which may explain a much lower RR (range from 11.6% to 23.6%) when comparing with the extraction ratio (approximately 60% in both Theranova 400 and FX CorDiax 120, at Qb 400 ml/min) tested in vitro. While comparing HDF with HD with the same dialyzers, we only observe a minor increase (P = 0.01 for comparing MCO-HDF with HFX-HD and P = 0.03 for comparing HFX-HDF with HFX-HD) in B2M removal in the HDF groups.

Albumin loss was higher with HDF groups, with a significantly higher loss in the off-label use of MCO-HDF, which was studied for mechanistic purposes and is not meant to inform clinical practice. This finding is novel and has not been previously reported. The loss of albumin on HFX-HDF in our study is within the range reported in the literature. For instance, the study by Combarnous et al. reported an albumin loss in effluent that ranged between 1.09 and 6.82 g/treatment while undergoing predilutional on-line HDF using a HFX hemodiafilter DIAPES HF800 with a membrane surface area of 1.8 m²,²⁷ and Shinzato et al. found a loss of 18.9±3.5 g of albumin/treatment in HDF compared with 3.4±0.7 g/treatment with HD.²⁸ Albumin loss has been reported to be higher while using the postdilutional mode when compared with HFX-HD (range: 0.08–7 g/treatment)^29,30 and increases with higher convective volumes.³¹

Different laboratory detection methodologies were used for albumin and total protein assays, but the results of the two analyses followed a similar pattern. The protein loss in the dialysate was reported to be approximately 1.2±0.3 g/treatment through a HFX dialyzer on HD,³² which is similar to our observations in the HFX-HD group. MCO membranes' pore size distribution shifts toward larger middle molecules with a cutoff slightly higher than albumin, while high-flux membranes usually have a cutoff lower than size albumin. As a result, HFX membranes are more selective in retaining albumin than MCO membranes. The albumin to total protein ratio in MCO groups, which is 50%–60%, is closer to the percentage of albumin to total protein in serum groups. Meanwhile HFX membrane, especially when used with HD, has a much lower percentage (10%) of albumin loss. This also explains why MCO membranes have a higher loss of total protein than HFX membranes. It is possible that some larger proteins with size close to albumin, such as α-1 globulin (54 kDa), may have a higher loss in MCO membranes. MCO-HDF has only a slightly higher percentage of albumins passing through comparing with MCO-HD. This observation indicates that MCO allows most of the albumin passing through largely depends on the additional opening of the pore size compared with high-flux dialyzer and relies less on the effects of convection. While we have demonstrated an increased dialytic albumin loss by MCO, it is unclear whether this translates into a beneficial or harmful treatment effect. Our study protocol did not aim to address this question, and long-term outcome studies—ideally sufficiently powered randomized controlled trials—are warranted to quantitate the risk/benefit ratio of MCO versus HFX-HD and HFX-HDF, respectively. Of note, a meta-analysis³³ found a slight reduction in serum albumin over the short term that seemed to normalize over the long term, suggesting compensatory mechanisms such as increased albumin synthesis. Absent adequate clinical trials, clinical judgment, and understanding of pathophysiology will inform treatment choices made by the dialysis practitioner caring for HD patients. Albumin loss has been extensively researched in patients treated with peritoneal dialysis, where daily protein loss ranges between 5 and 15 g/d³⁴ without clear adverse effects on patient outcomes. In some studies, this loss reflects a higher clearance of middle size and protein-bound uremic toxins, such effect does not have a negative effect on survival when liver function has the capacity of compensating albumin loss. One may hypothesize that is the effect which can be attained because of back-transport phenomenon in an MCO dialyzer design. Our study finds comparatively large total protein and albumin removal with MCO. While the albumin to total protein fraction was lower with HFX membranes, the absolute quantity of nonalbumin protein removal was higher with MCO. We speculate that the nonalbumin includes numerous large middle molecules. It is likely that some of these molecules are harmful and others beneficial. Our study was not designed to evaluate the net clinical benefit. Adequately powered clinical outcome studies (ideally randomized controlled trials) are warranted to address that question.

Our trial has some limitations: First, there is a significant difference in the dialyzer surface area between MCO and HFX dialyzers that may have influenced the results; to explore this topic further, clinical studies may not be adequate but require rather sophisticated bench tests that keep other potential influencing factors (e.g., hematocrit, protein concentration, dialyzer geometry, membrane composition, pore sizes, sterilization procedure, etc.) constant. In fact, the effect of surface area is called into question by Chapdelaine who studied HDF patients from the convective transport study and found no difference in convective volumes independently of the surface area,³⁵ and also the B2M sieving coefficient of the Theranova 400 has been reported to be 1; in comparison, the B2M sieving coefficient of all the Fx Cordiax Filters is 0.9, which should confer a superiority to Theranova 400.^12,36 Second, we did not measure middle or larger sized molecules other than B2M. Third, this was a short-term study (1 week of each dialysis method), which was not designed to explore systemic effects.

In summary, we conclude that MCO with HD is not superior to HFX-HD for both B2M and water-soluble solute removal. Protein loss was more pronounced with MCO when compared with HFX on both HD and HDF modality; other studies need to be designed to evaluate the effect on survival and nutritional status of such protein loss. MCO-HDF did not provide additional benefits for better B2M removal and was associated with increased protein loss when comparing with the other three treatment groups. It is important to note that the MCO membrane was neither designed nor approved for use with HDF and that the MCO-HDF is off-label. We included MCO-HDF in this mechanistic study to explore dialytic technologies beyond current device claims because it is not unusual that off-label use of devices or medicinal products provides novel insights. In brief, findings of our study are in line with a recent systematic review exploring the effects of MCO dialysis with HFX-HD and HDF.³³

Disclosures

B. Canaud reports the following—employer: Scientific consultant for Fresenius Medical Care, Germany up to December 2022—retired from December 2022 and consultancy: Senior scientist consultant for Fresenius Medical Care up to December 2022. G. Ferreira Dias reports the following—consultancy: Renal Research Institute; research funding: Renal Research Institute; honoraria: Renal Research Institute; and patents or royalties: Renal Research Institute. N. Grobe reports the following—employer: Telephonics; research funding: Fresenius Medical Care; and patents or royalties: Fresenius Medical Care. P. Kotanko reports the following—ownership interest: Fresenius Medical Care; research funding: Fresenius Medical Care, KidneyX, NIH; honoraria: HSTalks; patents or royalties: multiple patents in the kidney space; and advisory or leadership role: Editorial Board of Blood Purification; Editorial Board of Kidney and Blood Pressure Research; Editorial Board of Frontiers in Nephrology. M. Madero reports the following—consultancy: I am a consultant for Astra Zeneca and Boheringer; research funding: Abbvie, Astra Zeneca, Bayer, Boehringer; honoraria: Baxter, Astra Zeneca, Boehringer, Fresenius Medical Center; advisory or leadership role: American Journal of Kidney Disease, Kidney Disease Improving Global Outcomes (KDIGO) Executive Committee, International Society of Nephrology, Abbvie Advisory Boards, Astra Zeneca, Bayer; and speakers bureau: Astra Zeneca. J. Raimann reports the following—ownership interest: owning shares of stock in Fresenius Medical Care and other interests or relationships: Member of the Board of Directors “Easy Water for Everyone” (501c3). S. Thijssen reports the following—research funding: Fresenius Medical Care and other interests or relationships: I hold performance shares (virtual shares) in Fresenius Medical Care. X. Wang reports the following—research funding: Renal Research Institute. The remaining authors have nothing to disclose.

Funding

Fundacion Gonzalo Rio Arronte Grant 93131700.

Author Contributions

Conceptualization: Armando Armenta, Salvador Lopez-Gil, Magdalena Madero.

Data curation: Armando Armenta, Nadja Grobe, Peter Kotanko, Ivan Armando Osuna-Padilla, Jochen Raimann, Xia Tao.

Funding acquisition: Magdalena Madero.

Formal analysis: Armando Armenta, Salvador Lopez-Gil, Magdalena Madero, Xia Tao.

Investigation: Armando Armenta, Joshua Chao, Gabriela Ferreira Dias, Salvador Lopez-Gil, Magdalena Madero, Xiaoling Wang.

Methodology: Armando Armenta, Bernard Canaud, Salvador Lopez-Gil, Jochen Raimann.

Project administration: Salvador Lopez-Gil, Magdalena Madero.

Resources: Peter Kotanko, Stephan Thijssen.

Supervision: Bernard Canaud, Peter Kotanko, Jochen Raimann, Stephan Thijssen.

Validation: Armando Armenta, Nadja Grobe, Peter Kotanko, Jochen Raimann, Xia Tao.

Writing – original draft: Armando Armenta, Salvador Lopez-Gil, Magdalena Madero.

Writing – review & editing: Armando Armenta, Bernard Canaud, Peter Kotanko, Salvador Lopez-Gil, Magdalena Madero, Jochen Raimann, Xia Tao.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/KN9/A367.

Supplemental Table 1. Summary of average reduction ratios (in percent of starting level) in the four 2 study phases. Data are presented as median (interquartile range).

Supplemental Table 2. Summary of average protein dialysate loss in the four study phases. Data are presented as median (interquartile range).