Comparative Analysis of Hemodiafiltration Methods: Efficacy and Safety in Maintenance Hemodialysis

Introduction

The removal of uremic toxins in patients with end-stage renal disease is most effectively achieved through hemodiafiltration (HDF), which combines diffusion and convection with the use of replacement fluids to enhance solute clearance [1,2]. Currently, 2 standard approaches – pre-dilution HDF (pre-HDF) and post-dilution HDF (post-HDF) – are widely used, each with distinct operational advantages and limitations [3–5]. Post-HDF allows for maximal clearance of small and middle molecules, due to higher filtration fractions, but is often associated with increased hemoconcentration and elevated transmembrane pressure (TMP), which can compromise safety. In contrast, pre-HDF introduces substitution fluid before the dialyzer, effectively reducing TMP and hemoconcentration, but at the cost of lower solute clearance efficiency, due to dilutional effects. To address these limitations, mixed-dilution HDF (mix-HDF) has been developed as a novel approach that combines pre- and post-dilution strategies to optimize convective transport and enhance clearance [3,6–10].

Mix-HDF is a hybrid HDF technique that combines the infusion of substitution fluid before and after the dialyzer, in a controlled ratio. This approach aims to harness the benefits of pre-dilution (lower TMP, reduced hemoconcentration) and post-dilution (higher solute clearance) while minimizing their individual drawbacks. Pedrini and Zerbi (2007) systematically demonstrated that mix-HDF enables efficient middle molecule removal by better modulating convective transport without exceeding safe TMP levels or compromising membrane permeability [8]. Modern systems use TMP feedback to automatically adjust pre- and post-dilutional volumes in real time, ensuring stable filtration conditions and efficient solute removal [11]. Early implementations of mix-HDF required complex configurations, including Y-connectors, additional pumps, and manual adjustment of infusion ratios, which limited its widespread clinical use [8,9]. However, the advent of automated systems, such as the Biolight D800Plus, has simplified mix-HDF administration, enabling dynamic adjustment of substitution flow rates based on TMP regulation, and reducing operator burden [12–14]. Despite these advancements, comparative clinical data on mix-HDF vs conventional pre- and post-HDF are still limited. Previous studies have not consistently evaluated solute clearance across small (eg, urea), small-middle (eg, β2-microglobulin [β2-MG]), and large-middle molecules (eg, α1-microglobulin [α1-MG]) using automated mix-HDF devices. Additionally, the impact of different pre/post dilution ratios on TMP, convective volume completion rate (CVCR), and dialysate clearance efficiency remains poorly defined.

Therefore, this prospective single-center study aimed to evaluate 27 stable adult patients undergoing maintenance dialysis for more than 6 months to compare outcomes following mix-HDF with pre-HDF and post-HDF. We analyzed the efficacy of each modality in terms of urea reduction ratio (URR), serum reduction ratios (RR) of β2-MG and α1-MG, total dialysate removal amounts, and key dialysis parameters including TMP, blood flow, and convective volume indicators.

Material and Methods

ETHICAL APPROVAL:

This prospective clinical study was conducted at the Civil Aviation General Hospital in Beijing, China. The study was approved by the Research Ethics Committee of Civil Aviation General Hospital, Beijing, 100123, China (approval No.: 2022-L-K-03, date: January 15, 2022) and was performed in accordance with the Declaration of Helsinki. All participants provided written informed consent before enrollment.

STUDY DESIGN AND PARTICIPANTS:

This single-center, prospective observational study was conducted at the Civil Aviation General Hospital in Beijing, China, between November 2022 and October 2023. Eligible participants were patients aged 20 to 75 years who had been undergoing regular maintenance hemodialysis for more than 6 months, were in stable physical condition without recent hospitalization, had open and stable dialysis vascular access, and had no prior adverse reactions to HDF therapy. Exclusion criteria included a life expectancy of less than 12 months, active or recent malignancy, pregnancy, contraindications to anticoagulant therapy, and systolic blood pressure below 90 mmHg.

TREATMENT PROCEDURES:

A total of 27 patients were enrolled and treated sequentially using different HDF modalities. Each patient received dialysis 3 times per week, with each session lasting a minimum of 4 h. The treatments were performed using a D800Plus dialysis machine (Biolight, Guangzhou, China) and a single high-flux dialyzer (FX1000HDF, Fresenius Medical Care, Bad Homburg, Germany) with a membrane surface area of 2.2 m².

The 5 treatment protocols were as follows: (1) T1: pre-HDF, substitution volume: 40 L/session, dialysate flow: 500 mL/min; (2) T2: post-HDF, substitution volume: 20 L/session, dialysate flow: 500 mL/min; (3) T3: mix-HDF (pre: post=20: 10), substitution volume: 30 L/session, dialysate flow: 500 mL/min; (4) T4: mix-HDF (pre: post=15: 10), substitution volume: 25 L/session, dialysate flow: 500 mL/min; and (5) T5: mix-HDF (pre: post=15: 15), substitution volume: 30 L/session, dialysate flow: 500 mL/min.

Each patient completed all 5 treatment modalities in randomized sequence over a defined observation period. The study flow diagram is illustrated in Figure 1.

STUDY VARIABLES:

Baseline demographics and clinical characteristics were collected at enrollment, including sex, age, dialysis vintage, dry weight, hemoglobin, and serum albumin levels. Before and after each dialysis session, blood samples were collected to measure biochemical parameters including urea, creatinine, uric acid, potassium, sodium, calcium, phosphate, chloride, glucose, bicarbonate (via total CO₂), β2-MG, and α1-MG. Additionally, spent dialysate samples were collected throughout each dialysis session for further solute quantification.

URR was calculated using pre- and post-dialysis urea levels with the formula:

RR for β2-MG and α1-MG was calculated using post-concentration (Cpost) corrected for hemoconcentration using the Bergstrom formula:

where BW=body weight.

Total dialysate removal amount of β2-MG and α1-MG calculation with the following formula;

CVCR was calculated using actual convective volumes and target convective volumes with the formula:

OUTCOMES:

The primary outcomes assessed were differences in URR, RR of α1-MG/β2-MG serum levels, and total dialysate removal amounts of β2-MG and α1-MG among the 5 dialysis patterns. URR represented small molecule removal, β2-MG represented small-to-medium molecules, and α1-MG represented large-to-medium molecules. Secondary outcomes included blood flow, TMP, actual convective volumes, and CVCR.

STATISTICAL ANALYSIS:

Continuous variables were reported as mean±standard deviation (SD) or median (range), as appropriate. Inter-group differences across the 5 treatment arms were evaluated using the Kruskal-Wallis non-parametric test. Pairwise comparisons were performed using the Wilcoxon signed-rank test, and P values were corrected using the Benjamini-Hochberg procedure. A 2-tailed P value <0.05 was considered statistically significant. All statistical analyses were conducted using SPSS version 22.0 (IBM Corp, Armonk, NY, USA).

Results

BASELINE PATIENT CHARACTERISTICS:

A total of 27 patients consented to and participated in the study. The baseline characteristics are summarized in Table 1. The mean age of participants was 54.96±11.43 years, with 66.6% being men. The average dialysis duration was 115.15±40.72 months. Other clinical parameters included hemoglobin (110.96±12.67 g/L), albumin (40.07±4.89 g/L), total protein (68.03±5.13 g/L), hematocrit (34.62±3.95%), and ultrafiltration of 2.844±0.71 L.

PRIMARY OUTCOME:

The analysis of primary outcomes is presented in Figure 2 and Table 2. The URR was 67.81±9.38% in T1; 71.15±6.01% in T2; 71.5±7.52% in T3; 70.38±8.08% in T4; and 68.23±11.8% in T5. No significant differences in URR were observed among the 5 groups (P>0.05, Kruskal-Wallis test; Figure 2A). The β2-MG RR was 69.23±5.21% in T1; 78.38±4.17% in T2; 76.32±5.14% in T3; 73.35±5.03% in T4; and 77.68±5.01% in T5. The β2-MG RR was significantly lower in the T1 group than in the other groups (P<0.05), with no significant differences observed among the T2, T3, and T5 groups (Figure 2B). The α1-MG RR was 8.23±6.18% in T1; 13.06±6.04% in T2; 11.62±6.62% in T3; 7.10±6.56% in T4; and 8.96±6.58% in T5. The T2 group demonstrated significantly higher α1-MG RR, compared with T1 and T4 (P<0.05), with no significant differences between T2, T3, and T5 (Figure 2C).

In Figure 3 and Table 3, the analysis of the total removal amounts of β2-MG and α1-MG in dialysate are shown. The total β2-MG removal was 193.31±50.07 mg in T1; 293.4±70.05 mg in T2; 233.72±60.59 mg in T3; 231.48±57.09 mg in T4; and 234.91±70.65 mg in T5. The removal of β2-MG was higher in T2 to T5 than in T1 (P<0.05), with no significant differences among T2, T3, T4, and T5 (Figure 3A). β2-MG extraction followed the trend post-HDF ≈ mix-HDF > pre-HDF. The total α1-MG removal was 20.57±7.26 mg in T1; 70.46±28.48 mg in T2; 39.46±18.77 mg in T3; 35.55±18.62 mg in T4; and 54.87±33.49 mg in T5. The removal of α1-MG in T2, T3, T4, and T5 was higher than that in T1 (P<0.05). The α1-MG removal in T2 was better than that in T3, T4, and T5 (P<0.05), while T5 demonstrated a higher removal than T3 and T4. RRs between T3 and T4 showed no significant difference. The α1-MG extraction trend was post-HDF > mix-HDF (15: 15) >mix-HDF (20: 10)≈mix-HDF (15: 10) >pre-HDF (Figure 3B).

SECONDARY OUTCOMES:

Secondary outcomes focused on dialysis parameters across the various models, as shown in Figure 4 and Table 4. The blood flow rates were 262.58±19.73 mL/min in T1; 265.19±19.49 mL/min in T2; 264.07±17.82 mL/min in T3; 259.26±19.79 mL/min in T4; and 261.48±15.62 mL/min in T5, with no statistical differences observed among the groups (P>0.05; Figure 4A). The CVCR was 98.58±3.8% in T1; 92.3±8.94% in T2; 97.43±6.52% in T3; 97.1±8.0% in T4; and 93.64±13.9% in T5. The T1 group had a higher CVCR than did T2 and T5 (P<0.05). No differences were observed among T3, T4, and T5, but T3 showed a higher CVCR than T2 (P<0.05). CVCR in T2 and T5 did not differ significantly (Figure 4B). The CVCR trend was pre-HDF≈mix-HDF (20: 10)≈mix-HDF (15: 10) >mix-HDF (15: 15)≈post-HDF. The mean TMP was 123.44±27.08 mmHg in T1; 215.98±40.7 mmHg in T2; 199.87±34.31 mmHg in T3; 191.81±30.69 mmHg in T4; and 233.45±45.95 mmHg in T5. The TMP in T5 was significantly higher than in the other groups (P<0.05). T2 exhibited higher TMP than T1, T3, and T4 (P<0.05), with no significant differences between T3 and T4 (Figure 4C). TMP gradually increased during dialysis, without rapid spikes, and starting levels varied among groups (Figure 4D).

Discussion

Traditionally, 2 approaches to HDF – post-HDF and pre-HDF – have been used, each with inherent limitations, due to their principles [5,11,14]. Post-HDF maximizes the removal of uremic substances; however, it is often associated with increased technical challenges, including elevated TMP and blood concentration issues. Conversely, pre-HDF mitigates these technical complications but reduces clearance efficiency due to dilutional effects. Mix-HDF has emerged as a novel modality that combines the advantages of pre-HDF and post-HDF, while addressing their limitations. By injecting substitute fluid at specific ratios at both the inlet and outlet of the dialyzer, mix-HDF improves water retention and enhances membrane solute permeability. Recent advances in automated dialysis systems, such as the Biolight D800Plus, have facilitated the safe implementation of mix-HDF, without the need for manual adjustment or auxiliary connections [8,9,12]. However, comparative clinical studies assessing its performance remain limited [13,14].

In this study, we used the Biolight D800Plus dialysis machine to compare the effects of various mix-HDF regimens with pre-HDF and post-HDF. Mix-HDF demonstrated excellent small-molecule clearance, comparable to both pre-HDF and post-HDF. For small-middle molecules, represented by β2-MG (molecular weight 12 kDa), mix-HDF exhibited removal efficiencies similar to post-HDF and significantly superior to pre-HDF. These findings are consistent with those reported by Pedrini et al and Park et al, who found improved β2-MG clearance in mixed modes relative to pre-HDF [6,13]. In contrast, for large-middle molecules, such as α1-MG (molecular weight 33 kDa), post-HDF achieved the highest removal rates, followed by mix-HDF, with pre-HDF showing the least efficacy. Within mix-HDF modalities, the T5 (15: 15) configuration demonstrated superior α1-MG clearance compared to other mixed-dilution groups, likely due to a higher post-convective volume. Overall, the clearance of large-middle molecules, including α1-MG, was inferior to that of small-middle molecules such as β2-MG, regardless of the HDF modality. This limitation is primarily attributed to the dialyzer membrane’s performance, which restricts the removal of substances with molecular weights exceeding 30 kDa [15,16]. Alternative approaches, such as medium cut-off dialyzers, may offer enhanced removal of such molecules, although they were not evaluated in this study [11].

TMP, a critical safety parameter during dialysis, varied significantly across modalities. The highest TMP was observed in mix-HDF (15: 15), followed by post-HDF, and then mix-HDF (20: 10) and mix-HDF (15: 10), with the lowest TMP in pre-HDF. Higher TMP associated with post-HDF and certain mix-HDF configurations increases the risk of complications, necessitating adjustments in convective volume to ensure safety. Clinical practice often limits convective volumes in post-HDF to 50% of the plasma water volume, to avoid excessive TMP and hemoconcentration. Mix-HDF offers flexibility in balancing clearance efficiency and safety by adjusting pre- and post-dilutional flow rates. Among mix-HDF configurations, the T5 (15: 15) group demonstrated the highest post-convective volume and better removal of large-middle molecules, such as α1-MG, compared with other mix-HDF settings. However, mix-HDF (20: 10) provided the optimal balance between clearance efficiency and safety, as it maintained TMP within acceptable limits while delivering effective solute removal.

This study underscores the importance of automated systems that dynamically regulate pre- and post-dilutional flow rates in mix-HDF based on patient parameters, including hemoconcentration and TMP variations. Such systems preserve membrane permeability and ensure consistent performance. Our findings align with those of prior research, suggesting that individualized control of substitution flow enhances both efficacy and safety [6,10,13].

Limitations of this study include its single-center design, relatively small sample size (n=27), and short-term nature. Additionally, the study did not assess inflammatory markers, long-term clinical outcomes, or patient-reported quality-of-life indices, which could provide further insight into the overall benefits of mix-HDF. Despite these limitations, the prospective crossover design and comprehensive biochemical assessment offer valuable preliminary evidence supporting mix-HDF as a clinically viable modality. Further multicenter studies with larger cohorts and extended follow-up periods are warranted to validate these findings and determine the optimal substitution ratios for long-term application.

Conclusions

In conclusion, mix-HDF represents a viable and effective alternative to conventional HDF techniques. It combines the advantages of pre-dilution and post-dilution modalities while minimizing their respective limitations. Although post-HDF remains superior in removing large-middle molecules, such as α1-MG, mix-HDF – especially the T5 (15: 15) configuration – achieved comparable removal efficiency for small and small-middle molecules, such as urea and β2-MG. Moreover, mix-HDF offered improved safety by modulating TMP within acceptable limits. These findings support the potential clinical adoption of mix-HDF, particularly in settings equipped with automated feedback systems to optimize filtration dynamics and ensure treatment stability. Further large-scale studies are warranted to validate long-term efficacy and safety.