Comparable Efficacy of Medium Cut-Off Dialysis Plus HA130 Hemoadsorption and Postfilter Hemodiafiltration in Removing Protein-Bound Uremic Toxins

Abstract

Rationale & Objective

Protein-bound uremic toxins, including indoxyl sulfate and p-cresyl sulfate, are poorly removed by conventional dialysis and contribute to cardiovascular, inflammatory, and neurological complications in end-stage kidney disease. Although postfilter hemodiafiltration enhances middle molecule clearance, its availability is limited. We evaluated whether combining medium cut-off dialysis with HA130 hemoadsorption achieves protein-bound uremic toxin removal comparable with that of optimized postfilter hemodiafiltration.

Study Design

Prospective, single-center, parallel-group, single-session comparative study.

Setting & Participants

Twenty anuric adult patients undergoing maintenance hemodialysis treated at a tertiary dialysis center were allocated to medium cut-off dialysis with hemoadsorption (n = 10) or postfilter hemodiafiltration (n = 10).

Exposure

Expanded hemodialysis using a medium cut-off membrane combined with HA130 hemoadsorption versus postfilter hemodiafiltration.

Outcomes

Corrected reduction ratios of protein-bound uremic toxins (indoxyl sulfate, p-cresyl sulfate, carboxymethyllysine, and protein carbonyls) and middle molecules (β₂-microglobulin, free light chains, prolactin, parathyroid hormone, soluble receptor for advanced glycation end-products).

Analytical Approach

Predialysis and postdialysis plasma concentrations were measured, and corrected reduction ratios were calculated after adjusting for hemoconcentration. Between-group comparisons were performed using nonparametric tests.

Results

Corrected reduction ratios for key protein-bound uremic toxins were similar between medium cut-off dialysis with hemoadsorption and hemodiafiltration. Indoxyl sulfate corrected reduction ratios were 28.0% (IQR, 21.6-41.7) with medium cut-off dialysis with hemoadsorption and 26.0% (IQR, 22.5-37.9) with hemodiafiltration, while p-cresyl sulfate corrected reduction ratios were 31.8% (IQR, 23.9-37.4) versus 36.5% (IQR, 23.9-39.7), respectively (all P > 0.05). Carboxymethyllysine and protein carbonyl removal did not differ between modalities. In contrast, hemodiafiltration achieved higher corrected reduction ratios for conventional middle molecules, including β₂-microglobulin and free light chains. Reduction of soluble receptor for advanced glycation end-products was modest and comparable across treatments.

Limitations

Small sample size, single-session design, and reliance on plasma reduction ratios without direct dialysate mass measurements.

Conclusions

In this exploratory study, medium cut-off dialysis with hemoadsorption achieved protein-bound uremic toxin removal similar to that of optimized postfilter hemodiafiltration, despite lower clearance of conventional middle molecules. Hybrid strategies integrating diffusion, convection, and adsorption may expand protein-bound uremic toxin removal options, particularly where hemodiafiltration is not feasible.

Plain-Language Summary

Patients with kidney failure requiring dialysis accumulate uremic toxins that are inadequately removed by conventional dialysis techniques. Among these, protein-bound uremic toxins are particularly difficult to eliminate and have been implicated in cardiovascular, vascular, and neurological complications. Although hemodiafiltration enhances the clearance of several uremic solutes, its availability remains limited in many dialysis centers. In this study, we investigated whether the combination of a medium cut-off dialysis membrane with a hemoadsorption cartridge could represent a feasible alternative strategy. We observed that this hybrid approach achieved comparable removal of protein-bound uremic toxins to hemodiafiltration, although it was less effective for certain middle-molecular-weight solutes. These findings suggest that hybrid dialysis strategies may help expand treatment options when hemodiafiltration is not available.

Despite remarkable technological progress in renal replacement therapy, patients receiving maintenance hemodialysis (HD) continue to face high morbidity and mortality, with annual death rates commonly exceeding 15%-20%, largely because of cardiovascular events and infections.¹ Cardiovascular disease alone accounts for nearly half of all deaths in this population, and life expectancy remains dramatically shortened compared with the general population. These outcomes reflect, in part, the incomplete correction of the uremic milieu, in which the accumulation of inflammatory and oxidative toxins, particularly medium and large middle molecules and protein-bound uremic toxins (PBUTs), plays a central pathophysiological role.1, 2, 3

PBUTs are small hydrophobic solutes that bind tightly to albumin, leaving <10% in a free, dialyzable form that varies with ionic strength, pH, and binding competitors.⁴ They arise from endogenous metabolism, oxidative and glyco-oxidative reactions, and gut microbial activity.² Beyond simple retention, PBUTs promote oxidative stress, inflammation, endothelial dysfunction, insulin resistance, and fibrosis. Even modest increases in their plasma levels activate pro-oxidant and proinflammatory pathways, such as reduced nicotinamide adenine dinucleotide phosphate oxidase signaling, cytokine release, and endothelial senescence, contributing to the persistent cardiovascular and systemic risk in dialysis patients despite adequate small-solute clearance.^2,5

Hemodiafiltration (HDF) is currently considered the most advanced technique for solute removal.³ Several studies have demonstrated that achieving convection volumes >23 L/1.73 m² body surface area per session improves clinical outcomes, a threshold supported by trials such as ESHOL and CONVINCE.^6,7 However, even under these conditions, the clearance of PBUTs, such as indoxyl sulfate and p-cresyl sulfate, remains inadequate, as demonstrated by clinical outcomes.^7,8 The difficulty in mobilizing the albumin-bound fraction is highlighted by the study of Escudero-Saiz et al,⁹ who demonstrated that the use of ibuprofen, a competitive displacer of albumin-bound toxins, significantly increased reduction ratios for indoxyl sulfate and p-cresyl sulfate during postfilter HDF, underscoring the limited contribution of convection alone to PBUT elimination.

Although postfilter HDF achieves high clearance of middle molecules, its implementation is not universal. The need for on-line replacement fluid, dedicated equipment, and high-quality water limits its availability in many dialysis centers, particularly in settings with constrained infrastructure or resources.¹⁰ In this context, medium cut-off and super-flux membranes have emerged as potential alternatives to enhance middle molecule clearance through optimized membrane pore size and internal filtration in a technique called expanded hemodialysis (HDx).^11,12 Although such membranes achieve superior removal of larger uremic toxins compared with conventional high-flux membranes, their efficacy for PBUT clearance remains limited. Kim et al¹³ reported no differences in indoxyl sulfate or p-cresyl sulfate removal between HDx, postfilter HDF, and high-flux HD. Our group similarly found that medium cut-off membranes did not improve carboxymethyllysine clearance, indicating that increased membrane permeability alone cannot overcome protein binding. These findings reinforce the need for complementary strategies such as hemoadsorption.^13,14

In this context, hemoadsorption has emerged as a promising adjunct to dialysis techniques to overcome the physicochemical limitations that restrict protein-bound solute clearance. Modern sorbent cartridges use neutral mesoporous styrene–divinylbenzene beads with a broad pore-size distribution (5-60 kDa), allowing the removal of middle and protein-bound molecules through hydrophobic, ionic, and van der Waals interactions.^15,16 Clinical evidence shows that adding hemoadsorption to HD improves the removal of multiple uremic toxins, including β₂-microglobulin, cytokines, and advanced glycation end-products, and enhances biocompatibility and inflammatory profiles in patients with kidney failure.17, 18, 19, 20

These findings suggest that hemoadsorption can complement membrane-based clearance by targeting solutes that remain largely protein-bound. Given the wider availability of HDx, combining hemoadsorption with this modality (HAHDx) may offer a practical way to broaden toxin removal. Accordingly, this study evaluated whether HAHDx enhances middle molecule and PBUT clearance compared with postfilter HDF.

Methods

This was a prospective, single-center, parallel-group, single-session comparative study evaluating 2 extracorporeal solute removal strategies. The study was conducted at the Nephrology Service, Las Higueras Hospital (Talcahuano, Chile) between August and September 2025. The objective was to evaluate the efficacy of combining hemoadsorption with HDx compared with postfilter HDF in the removal of selected PBUTs and middle molecules.

Patient Selection

A total of 20 anuric adult patients undergoing maintenance HD were included. Inclusion criteria were age ≥18 years, receiving maintenance HD for ≥3 months, a permanent vascular access (arteriovenous fistula or tunneled catheter) capable of achieving a blood flow rate ≥350 mL/min, and no residual kidney function. Exclusion criteria included active malignancy, known hypersensitivity to dialysis membranes, life expectancy <6 months, pregnancy, or a diagnosis of inflammatory bowel disease.

Patients were allocated to 2 treatment groups, ensuring similar distributions by age, sex, and dialysis vintage. Baseline clinical data, including etiology of kidney failure, comorbid conditions, dialysis duration, and session parameters were collected and recorded in dedicated electronic medical records.

Interventions

Each participant underwent a single study session after a 1-week washout period with standard high-flux HD, ensuring similar baseline conditions.

• •Group 1 (HAHDx): Treatment was performed using an HA130 cartridge (Jafron) in series with a Theranova 400 dialyzer (Vantive Health). The HA130 cartridge was inserted in prefilter position and primed and connected according to the manufacturer’s instructions. The filter featured a medium cut-off membrane made of polyarylethersulfone/polyvinylpyrrolidone blend, steam-sterilized, with a surface area of 1.7 m², dializer ultrafiltration coefficient 48 mL/h/mm Hg, and molecular weight cut-off ∼56 kDa.
• •Group 2 (HDF): Standard postfilter online HDF was carried out using an FX CorAL 800 high-flux dialyzer (Fresenius Medical Care), equipped with the Helixone hydro membrane featuring a surface area of 2.2 m², a dializer ultrafiltration coefficient of 84 mL/h/mm Hg, a β₂-microglobulin sieving coefficient of 1.0, and an albumin sieving coefficient of <0.001. HDF sessions were performed on a 5008S CorDiax system, achieving a median convection volume of ∼26 L over 4 hours.

In both groups, blood flow was maintained at ≥450 mL/min, and dialysate flow was set approximately 20% higher than blood flow. Ultrafiltration goals were individualized for each patient based on their clinical status. Unfractionated heparin was used for anticoagulation during all procedures.

Blood Sampling and Laboratory Analysis

Venous blood samples were collected immediately before and after each dialysis session using ethylenediaminetetraacetic acid tubes. Samples were centrifuged at 1,000g for 15 minutes at 2-8°C within 30 minutes of collection, and the resulting plasma was stored at −20°C until analysis.

Plasma concentrations of PBUTs and related biomarkers were quantified using commercially available enzyme-linked immunosorbent assays (ELISAs), according to the manufacturers’ instructions. Indoxyl sulfate (213 Da) was measured using a competitive ELISA kit (Antibodies.com; catalog no. A86937). p-Cresyl sulfate (108 Da) was quantified using a competitive ELISA (MyBioSource; catalog no. MBS7273551). Carboxymethyllysine (204 Da) was measured using a sandwich ELISA kit (MyBioSource; catalog no. MBS700744). Protein carbonyl content was assessed using a colorimetric ELISA (Abcam; catalog no. ab238536). Soluble receptor for advanced glycation end products (sRAGE) (55,000 Da) was measured using a Quantikine ELISA kit (R&D Systems; catalog no. DRG00) at the Biomedical Research Labs, Catholic University of Maule, Talca, Chile.

In addition, the study included the middle molecule markers β₂-microglobulin (11,800 Da), prolactin (23,000), and free light chains kappa (22,500 Da) and lambda (45,000 Da) (FLC-κ, FLC-λ), measured by immunonephelometry using the BN ProSpec System (Siemens Healthcare Diagnostics), along with the biochemical parameters ferritin, intact parathyroid hormone, calcium, phosphorus, bicarbonate, and BUN, analyzed at the Las Higueras Hospital central laboratory using standard procedures.

Calculation of Reduction Ratios

The reduction ratio was calculated using the formula:

Reduction ratio (%) = 100 (C₀ − C_end) / C₀

where C₀ is the baseline concentration, and C_end is the posttreatment concentration.

To account for hemoconcentration, the corrected reduction ratio was calculated as follows:

Corrected Reduction Ratio (%) = 100 (C₀ − cC_end) / C₀

where cC_end = C_end / (1 + [$\Delta$ BW / 0.2 (BW_post)]) and $\Delta$ BW refers to the change in body weight during the treatment, used to account for variations because of ultrafiltration.

To minimize recirculation, post-dialysis samples were obtained after reducing blood flow by 50 mL/min and briefly stopping dialysate flow and ultrafiltration for 15 seconds before collection.

Statistical Analysis

All quantitative variables are expressed as median and interquartile range (IQR) and categorical variables as counts and percentages. Data distribution was assessed using the Shapiro–Wilk test. Comparisons of corrected reduction ratios and biochemical parameters between groups were performed using nonparametric tests: the Mann–Whitney U test for continuous variables and the χ² test for categorical data. P ≤ 0.05 was considered statistically significant.

All analyses and graphical representations were performed in SPSS 25.0 and Python (version 3.12) using the pandas, scipy.stats, numpy, and matplotlib libraries.

Ethical Considerations

This study was conducted in compliance with the principles of the Declaration of Helsinki and the Good Clinical Practice guidelines. The Metropolitan Oriente Health Service Ethics Committee (approval code SSMOriente120825-3) approved this study on August 12, 2025. Written informed consent was obtained from all participants before inclusion in this study in accordance with the Declaration of Helsinki.

Results

Baseline Characteristics

A total of 20 patients receiving maintenance HD were included, divided equally between the HAHDx and postfilter HDF groups. The median age was 58.5 years (IQR, 40.5-64.0), with no significant difference between groups (59.5 [IQR, 40.0-63.8] vs 55.0 [IQR, 41.3-67.0]; P = 0.880). The median single-pool Kt/V was 1.9 (IQR, 1.6-2.2), significantly higher in the HDF group (2.1; IQR, 2.1-2.2) compared with the HAHDx group (1.7; IQR, 1.6-1.8; P = 0.038). Median blood flow was 526 mL/min (IQR, 491.8-540.5), and session duration averaged 242.5 minutes (IQR, 242.0-244.3), with no between-group differences (P = 0.570 and P = 0.124, respectively).

The median convection volume achieved in the HDF group was 26.4 L (IQR, 22.1-27.1). The median dialysis vintage was 1.9 (IQR, 0.7-2.9) years in HAHDx and 3.0 (IQR, 2.6-3.3) years in HDF (P = 0.226).

Baseline biochemical parameters were comparable between groups, with no significant differences in albumin, ferritin, hematocrit, bicarbonate, calcium, or phosphorus levels (P > 0.05 for all) (Table 1). Likewise, the distribution of comorbid conditions, including hypertension, diabetes, coronary disease, and heart failure, as well as vascular access type and etiology of kidney disease were similar between the 2 groups.

Table 1.

Baseline Characteristics and Laboratory Parameters

Variable	Total	HAHDx	HDF	P
Age, y	58.5 [40.5-64.0]	59.5 [40.0-63.8]	55.0 [41.3-67.0]	0.880
Women	35% (7)	20% (2)	50% (5)	0.211
Hypertension	85% (17)	100% (10)	70% (7)	0.210
Diabetes	25% (5)	30% (3)	20% (2)	1.000
Heart failure	15% (3)	0% (0)	30% (3)	0.210
Cardiovascular diseases	35% (7)	30% (3)	40% (4)	1.000
BUN (pre), mg/dL	62.4 [48.4-67.3]	64.0 [59.3-66.2]	57.9 [44.4-63.3]	0.344
BUN (post), mg/dL	10.6 [5.9-12.6]	11.5 [10.5-14.8]	7.6 [5.6-10.6]	0.104
Ferritin, ng/mL	651.0 [530.7-836.7]	563.5 [528.2-738.7]	815.0 [579.0-1053.2]	0.121
Albumin, g/dL	3.8 [3.7-4.1]	3.8 [3.7-3.9]	3.9 [3.8-4.1]	0.305
Bicarbonate, mmol/L	19.0 [18.0-21.0]	20.0 [18.2-20.7]	19.0 [18.2-20.5]	0.619
Calcium, mg/dL	8.7 [8.4-8.9]	8.7 [8.6-9.0]	8.7 [8.4-8.9]	0.520
Phosphorus, mg/dL	3.9 [3.3-5.8]	4.3 [3.4-7.3]	3.9 [3.3-5.5]	0.678
OCM	1.9 [1.7-2.3]	1.9 [1.8-2.0]	2.2 [1.7-2.5]	0.427
spKt/Vurea	1.9 [1.6-2.2]	1.7 [1.6-1.8]	2.1 [2.1-2.2]	0.038
QB, mL/min	526.0 [491.8-540.5]	510.0 [489.3-537.8]	533.0 [515.3-541.5]	0.570
Time, min	242.5 [242.0-244.3]	243.5 [242-245]	242.0 [241-243.8]	0.124

Note: Values are median [interquartile range] or % (n).

Abbreviations: BUN, blood urea nitrogen; HAHDx, expanded hemodialysis with hemoadsorption; HDF, hemodiafiltration; OCM, online clearance monitor; Q_B, blood flow rate; spKt/V, single-pool Kt/V.

Pre- and post-dialysis plasma concentrations of the investigated middle molecules and PBUTs are provided in Table S1.

Protein-Bound Uremic Toxins

The evaluation of protein-bound solute removal showed comparable efficacy between HAHDx and postfilter HDF (Table 2; Fig 1). The median corrected reduction ratios for the analyzed solutes were similar across both groups: protein carbonyls reached 22.8% (IQR, 20.3-32.3) with HAHDx and 24.9% (IQR, 24.2-32.7) with HDF, while carboxymethyllysine achieved 53.9% (IQR, 41.9-62.1) versus 56.2% (IQR, 32.2-59.0). For p-cresyl sulfate and indoxyl sulfate, corrected reduction ratios values were 31.8% (IQR, 23.9-37.4) versus 36.5% (IQR, 23.9-39.7) and 27.9% (IQR, 21.6-41.7) versus 25.9% (IQR, 22.5-37.9) for HAHDx and postfilter HDF, respectively (all P > 0.05).

Table 2.

Corrected Reduction Ratio of Middle Molecules and Protein-Bound Uremic Toxins

Biomarker	HAHDx	HDF	P
Middle molecules
FLC-λ	69.4 [67.6-74.2]	80.8 [79.3-82.5]	<0.001
FLC-κ	69.2 [66.7-71.7]	80.2 [77.7-82.8]	<0.001
Prolactin	61.3 [54.1-63.9]	72.2 [67.6-77.8]	0.009
β2-microglobulin	83.0 [81.8-84.4]	88.3 [86.5-88.5]	0.003
PTH	74.4 [49.3-76.3]	78.6 [69.8-86.9]	0.327
sRAGE	21.0 [9.4-29.9]	20.6 [12.8-34.4]	0.571
PBUTs
Protein carbonyls	22.7 [20.3-32.3]	24.9 [24.2-32.7]	0.427
Carboxymethyllysine	53.9 [41.9-62.1]	56.2 [32.2-58.9]	0.571
p-Cresyl sulfate	31.8 [23.9-37.3]	36.5 [23.9-39.7]	0.623
Indoxyl sulfate	27.9 [21.6-41.7]	25.9 [22.5-37.9]	0.910

Note: Comparison between HAHDx and HDF groups for conventional biomarkers and PBUTs. Values are median [interquartile range].

Abbreviations: FLC λ, lambda free light chain; FLC κ, kappa free light chain; HAHDx, expanded hemodialysis with hemoadsorption; HDF, hemodiafiltration; PBUT, protein-bound uremic toxin; PTH, parathyroid hormone; sRAGE, soluble receptor for advanced glycation end-products.

Figure 1.

Removal of protein-bound uremic toxins (PBUTs) by dialysis modality. Corrected reduction ratios (%) for carboxymethyllysine (CML), soluble receptor for advanced glycation end-products, protein carbonyls, indoxyl sulfate, and p-cresyl sulfate are shown for the expanded hemodialysis with hemoadsorption (HAHDx) and hemodiafiltration (HDF) groups. No significant differences were observed between modalities for any PBUT, indicating comparable efficacy in removing protein-bound solutes.

No statistically significant differences were observed between modalities, although a slight trend toward higher indoxyl sulfate removal was noted with HAHDx.

Middle Molecule Reduction Ratio

Reduction ratios of middle molecules, including β₂-microglobulin, prolactin, FLC-κ, and FLC-λ, were consistently higher with postfilter HDF compared with HAHDx (Table 2; Fig 2). The median corrected reduction ratio for FLC-λ was 80.8% (IQR, 79.3-82.5) in HDF versus 69.4% (IQR, 67.6-74.2) in HAHDx, while for FLC-κ, values were 80.3% (IQR, 77.7-82.8) and 69.2% (IQR, 66.7-71.7), respectively (P < 0.001 for both).

Figure 2.

Reduction ratios of middle molecules. Box plots represent the corrected reduction ratios (RRc, %) of middle molecules, comparing the HAHDx and postfilter HDF groups. Free light chains λ and κ, prolactin, and β₂-microglobulin showed significantly higher RRc values with HDF (P < 0.001, P < 0.001, P = 0.009, and P = 0.003, respectively), whereas PTH removal was comparable between modalities (P = 0.327). Abbreviations: FLC, free light chain; HAHDx, expanded hemodialysis with hemoadsorption; HDF, hemodiafiltration; PTH, parathyroid hormone; sRAGE, soluble receptor for advanced glycation end-products.

Similarly, β₂-microglobulin and prolactin showed significantly greater reduction ratios with HDF (88% vs 83% and 72% vs 61%, respectively; P = 0.003 and P = 0.009). In contrast, parathyroid hormon showed similar median reductions across modalities (79% vs 74%; P = 0.327), with greater variability in the HAHDx group. Notably, sRAGE, which also falls within the middle molecule range, showed comparable median reductions between the 2 groups (21.0% [IQR, 9.4-29.9] in HAHDx vs 20.6% [IQR, 12.8-34.4] in HDF; P = 0.571).

Overall, these findings demonstrate that HDF outperformed the HAHDx configuration in removing classical middle molecules. At the same time, both treatments maintained excellent circuit performance and safety, with no episodes of clotting or technical complications.

Discussion

In this study, the combination of a medium cut-off membrane with hemoadsorption achieved a high removal efficiency across a broad spectrum of uremic solutes, demonstrating overall performance comparable to postfilter HDF under optimized convective conditions. Although HDF showed superior clearance of conventional middle molecules, including β₂-microglobulin, prolactin, and free light chains, both modalities achieved substantial reduction ratios, markedly higher than those typically reported for standard high-flux or medium cut-off membranes used alone.^{13,14,18,21,22} On the other hand, the level of efficiency in the HDF group was extremely high and uncommon in routine dialysis. In settings where such high convection volumes are not feasible, the gap between HDF and HAHDx would likely narrow, as HAHDx provides stable efficiency independent of convection volume.

In previous comparative studies, Maduell et al^21,22 reported β₂-microglobulin reduction ratios ranging from 75%-80% and free light chain (λ and κ) removal between 40% and 55% using medium cut-off membranes, whereas postfilter HDF achieved values of 82%-86% and 45%-50%, respectively. Similarly, Kim et al¹³ compared high-flux HD, postfilter HDF, and HDx and found that β₂-microglobulin removal was slightly higher with HDF (79.5%) than with HDx (75.3%). In contrast, FLC-λ removal favored HDx (≈51.5% vs 43.5%).

Taken together, these data show that both HDF and HDx achieve efficient middle molecule clearance. However, PBUT removal remains suboptimal with these modalities, highlighting a persistent therapeutic limitation. This gap defines the next frontier in uremic toxicity research and underscores the need for strategies specifically designed to enhance PBUT elimination and address their systemic effects.²³ In our study, the addition of hemoadsorption to HDx yielded reduction ratios similar to or higher than those previously reported for HDx or HDF alone, suggesting a possible additive contribution of adsorption. These findings support the concept that hemoadsorption may complement diffusive–convective transport to extend the spectrum of solute removal without compromising circuit performance or safety.

In our study, the reduction ratios of p-cresyl sulfate, indoxyl sulfate, and protein carbonyls were comparable between the HAHDx and HDF groups, with no statistically significant differences. These findings indicate that the level of PBUT removal achieved with postfilter HDF can also be obtained using the simpler HAHDx configuration. Thus, even under optimized convective conditions, PBUT clearance remains modest.^22,24,25 In a longitudinal cohort study of postfilter HDF using a polymethyl methacrylate membrane, Fabbrini et al²⁵ reported a mean indoxyl sulfate reduction ratio of 33% with HDF with a polymethyl methacrylate filter, compared with 30% with conventional HDF and only 12% with HD, observing a significant decline in indoxyl sulfate levels after 6 months only in the HDF with a polymethyl methacrylate filter group. In contrast, p-cresyl sulfate removal showed no significant differences among modalities, highlighting the heterogeneous behavior of PBUTs depending on their protein-binding strength and distribution kinetics.

A key explanation is the very high protein binding of these solutes (>95%), leaving only a minimal free fraction available for diffusive, convective, or adsorptive transport. Above this threshold, extracorporeal removal becomes intrinsically limited, as shown by Kawasaki et al,²⁶ who demonstrated that clearance depends mainly on the dynamic release of the unbound fraction. In vitro studies similarly showed minimal removal of highly bound solutes, indicating that the bottleneck is free solute availability rather than sorbent capacity.^24,26

Overall, these data suggest that improving indoxyl sulfate and p-cresyl sulfate removal will likely require combining extracorporeal techniques with albumin-binding displacers, rather than relying on greater convection or adsorption alone. Recent mechanistic reviews reinforce that effective clearance of highly bound solutes depends on disrupting protein–toxin interactions.^9,27 Nevertheless, in the meantime, the addition of hemoadsorption represents an important opportunity to optimize clearance, especially in patients who cannot afford high blood flows or extremely high fluid exchange volumes. Emerging electrostatic sorbents such as pHA130 also show promise for enhancing PBUT removal, underscoring the potential of next-generation materials to further complement these strategies.^28,29

Regarding carboxymethyllysine, our results demonstrate that coupling hemoadsorption with a medium cut-off membrane achieved removal efficiencies similar to those obtained with postfilter HDF. This contrasts with prior evidence showing no improvement of HDx over high-flux HD when adsorption is not included. In our previous pilot study, single-session HDx versus high-flux HD produced comparable corrected reduction ratios (≈36.9% vs 35.6%), confirming that increased membrane permeability alone does not overcome the protein-binding barrier for advanced glycation end-products.¹⁴ By contrast, adding hemoadsorption to high-flux HD significantly increased the carboxymethyllysine-corrected reduction ratio from 39% to 65%.¹⁸ In the present study, the carboxymethyllysine-corrected reduction ratio reached 53.9% (IQR, 41.9-62.1) with HAHDx and 56.2% (IQR, 32.2-59.0) with HDF, showing equivalent efficacy between both strategies. Accordingly, the present study was not intended to isolate the independent effect of hemoadsorption, but rather to evaluate the performance of composite extracorporeal strategies.

Importantly, sRAGE, a soluble decoy receptor with vascular protective and anti-inflammatory properties, was minimally affected and remained comparable between both modalities, consistent with our previous findings using high-flux HD plus hemoadsorption.¹⁸ This preservation of sRAGE supports the safety of incorporating hemoadsorption into medium cut-off-based dialysis strategies.

This study has several limitations that should be acknowledged. First, the small sample size and single-center design may limit the generalizability of the findings and increase the risk of type II error when comparing groups. Accordingly, the limited sample size precludes any formal assessment of equivalence or noninferiority; therefore, the present findings should be interpreted as exploratory rather than confirmatory. Second, only a single treatment session per modality was analyzed, which restricts the evaluation of long-term or cumulative effects on solute kinetics and systemic toxin burden. Third, although reduction ratios provide an accurate intrasession estimate of removal efficiency, they do not account for interdialytic toxin rebound or tissue redistribution, processes that may attenuate net detoxification over time. Furthermore, the study did not include measurements of the free fraction or plasma protein-binding dynamics, which would have provided additional insight into removal mechanisms. Future multicenter crossover trials with longitudinal follow-up, including biomarker and clinical endpoints, are warranted to confirm these preliminary findings and determine their clinical relevance.

In conclusion, these findings demonstrate that hybrid extracorporeal strategies integrating diffusion, convection, and adsorption can broaden the spectrum of uremic toxin removal without compromising safety. By increasing the clearance of selected advanced glycation end-products such as carboxymethyllysine while preserving protective mediators like sRAGE, the combination of medium cut-off membranes with hemoadsorption represents a physiologically balanced approach that performs similarly to HDF.

Future investigations should confirm these observations in larger, crossover, and longitudinal trials, and evaluate complementary strategies—such as albumin-binding displacers, gut-targeted interventions, or next-generation sorbents with enhanced specificity for PBUTs (electrostatic resins)—to further improve the removal of tightly bound solutes. Importantly, this technique remains applicable when HDF is unavailable or suboptimally delivered, in which it may provide comparable removal efficiency.

Supplementary Materials

Supplementary Material (PDF)

Table S1