The results indicated that the removal performance varied depending on the substitution volume in pre- and post-OHDF, as in previous reports [6,7,8, 15]. In particular, the tendency was remarkable in middle molecules, which are predominantly removed by filtration rather than diffusion.

An interesting finding of this study was that the removal performance of small-molecular-size substances, such as urea, UA, CRE, and iP, was not inferior in pre-OHDF to that in post-OHDF. Theoretically, small-molecular-size substances are predominantly removed by diffusion; therefore, their removal performance should be superior in post-OHDF. The concentration of such substances is diluted in pre-OHDF, and the concentration difference between the blood side and dialysate is less in pre- than in post-OHDF. However, in this study, the reduction ratios of small-molecular-size substances were almost the same regardless of the dilution mode. This might reflect the excellent removal performance of ABH-PA through the diffusion of small-molecular-size substances using an improved three-dimensional hollow fiber structure.

Meanwhile, post-OHDF was superior to pre-OHDF in terms of removal performance of α1-MG, which is predominantly removed by filtration. The removal performance of the middle molecules increases as the substitution volume increases in the same dilution mode, and comparing the dilution modes, post-OHDF is superior to pre-OHDF. This result resonates with that of previous studies [20].

For β2-MG, unlike α1-MG, no significant difference was observed in the removal performance regardless of the dilution mode or substitution volume, as with small-molecular-size substances. Therefore, β2-MG, with a molecular weight of 11,800, was thought to be removed mainly by diffusion rather than by filtration. This is consistent with the findings of previous reports that compared the removal performance of pre- and post-OHDF [5, 20].

As β2-MG is removed mainly by diffusion, it would be insufficient to attribute the improvement in mortality resulting from OHDF to β2-MG removal only. Instead, it is necessary to also focus on the removal of substances that have a higher molecular weight than β2-MG and are mainly removed by filtration. Among these substances, clinical manifestations related to dialysis are alleviated when the reduction ratio of α1-MG exceeds 35% [5]. When attempting to increase the reduction ratio of α1-MG, albumin will inevitably leak out simultaneously. This is because the Stokes radii of free α1-MG and albumin are similar, at 28.6 Å [21] and 35.5 Å, respectively, even though their molecular weights differ, at approximately 33,000 and 66,000, respectively. α1-MG has been reported to form a complex with IgA, prothrombin, and albumin within the body [22]. However, the Stokes radii of α1-MG combined with IgA, prothrombin, and albumin remain unknown. New types of hemodiafilters that can selectively remove α1-MG with low albumin leakage are under development.

In the removal performance of the selectivity of the α1-MG removal for albumin leakage, some differences were observed between pre- and post-OHDF. SRIA, an index of the selective removal capacity of α1-MG, was slightly better in pre- than in post-OHDF. However, SRIA tends to be high when the amount of albumin leakage is low, even if the removal amount of α1-MG is minimal. In this study, the maximum average removal amount of albumin for each dilution mode was 2.46 g in pre-OHDF with 48 L substitution volume and 3.51 g in post-OHDF with 10 L substitution volume, which were too low to assess SRIA. Therefore, the variation in substitution volume in each group was not sufficient for detailed discussion, and further studies are required to elucidate the selective removal capacity of α1-MG with different dilution modes.

In this study, the levels of the inflammatory markers such as hs-CRP and IL-6, both before and after treatment, demonstrated no significant difference between pre- and post-OHDF. The inflammatory response was measured during post-OHDF at a substitution volume of 10 L, and QB was 280 mL/min; thus, the ratio of filtration flow rate (Qf) to QB was approximately 15%. This value is lower than the standard ratio in EU countries. In a study regarding EU countries, more than half of post-OHDF patients were treated with a substitution volume of ≥ 20 L, and the average QB of the patients was 342 mL/min, in which case the ratio of Qf to QB was ≥ 24% [23]. Some differences in inflammatory response may be observed under conditions of higher blood concentrations. Further studies with a high Qf to QB ratio are required to confirm this assumption.

There were no device-related AEs, device malfunctions, or elevations in TMP in either the pre- or post-OHDF modes. Therefore, both pre- and post-OHDF can be regarded as sufficiently safe for treatment under the conditions evaluated in this study.

This study has some limitations. First, no female patients were involved in the study, which means that the current findings are not representative of what occurs in both males and females. Second, the findings were associated with OHDF employing only ABH-22PA. Therefore, if other instruments are employed, the results may vary. Third, there were no data on high volumes of substitution fluid as the maximum amount of substitution fluid was based on the average in Japan. Further studies employing a larger population, the use of other instruments, and higher volumes of substitution fluid are required.

This study aimed to confirm removal performance when the same hemodiafilters were used in the same patients; thus, mortality and other clinical benefits were not evaluated. Further studies are required to evaluate the relationship between substance removal, especially of middle molecules, and mortality.