Biomarkers for Evaluation of Clinical Outcomes of Hemodiafiltration

Hemodiafiltration (HDF) can efficiently remove uremic toxins, from low-molecular-weight to large-molecular-weight substances, by using a combination of diffusion and convection [1,2,3]. More specifically, the benefits of HDF can be best utilized and good therapeutic efficacy achieved in the treatment of insomnia, pruritus, irritability, restless legs syndrome (RLS), anemia, dialysis amyloidosis (DA), etc., when removal performance of low-molecular-weight protein (LMWP) is improved [4,5,6,7].

However, if the conditions under which HDF is performed are unsuitable, the removal performance of HDF is not as high as expected, and it is difficult to differentiate from hemodialysis (HD) with a high-performance dialyzer.

β₂-Microglobulin (β₂-MG) is the substance that causes DA, and it is widely accepted that its predialysis value is useful for evaluating the quality of dialysis therapy (quality of water purification, dialysis membrane, dialysis time, dialysis frequency, presence or absence of inflammation) [8,9,10,11]. In addition, β₂-MG is also an important biomarker for evaluating the dialysis efficiency of HD and HDF [12].

However, β₂-MG has a molecular weight of 11.8 kDa, which is toward the lower end of the molecular weight range of LMWPs (MW: 10-55 kDa), and since it can be efficiently removed by diffusion with existing high-performance dialyzers, a higher molecular weight substanceshould be used to evaluate removal efficiency in evaluations of the performance of HDF, in which diffusion and convection are performed simultaneously.

In this paper, we state based on our own cases what kind of marker should be used and how conditions for HDF should be set up in order to make the best use of the distinctive features and benefits of HDF.

On-line HDF became formally covered by the Japanese National Health Insurance system (JNHAS) in 2010. However, four conditions were stipulated for performing it: (1) use of approved dedicated HDF equipment, (2) use of an approved hemodiafilter, (3) guarantee of dialysis water purification by each institution, and (4) DA and dialysis difficulty as indications.

We changed the dialysis conditions at our own institution to conform to the conditions stipulated for coverage by JNHAS in 17 cases that had become stable on HDF with a super high-flux dialyzer, and the result was that by one month after changing the dialysis conditions, the symptoms in 13 of the 17 cases had become worse. In 6 of these 13 cases, the dialysis method had been changed to HD with a super high-flux dialyzer, and in the other 7 cases it had been changed to HDF with a hemodiafilter that had been approved.

We then changed the dialysis conditions again in the 13 cases in which the symptoms had become worse. More specifically, we attempted to improve removal performance of dialysis by performing HDF with the super high-flux dialyzer that had been used before or performed HD with a higher performance super high-flux dialyzer, etc. The result was that one month later ADL had improved in 11 of the 13 cases, and their QOL had returned to its previous level. An improvement in QOL was also noted in the other 2 cases.

α₁-Microglobulin (α₁-MG) removal rates in the 13 cases whose symptoms changed from ‘stable' to ‘worse' and then to ‘improved' changed significantly almost exactly in tandem with the changes in symptoms from 35.5 ± 7.7 to 23.2 ± 4.9% and then to 32.3 ± 5.6% (n = 13, all p < 0.01). However, the corresponding changes in Kt/V from 1.59 to 1.65 and then to 1.56 and in the β₂-MG removal rate from 78.2 to 75.2% and then to 77.3% only fluctuated within narrow ranges. These changes show that the symptom worsening occurred because the α₁-MG removal rate had decreased, i.e. they occurred because of insufficient removal of uremic toxins having a molecular weight of 30 kDa or more, and the improvement in symptoms occurred because the removal of uremic toxins having a molecular weight of 30 kDa or more had increased. They also showed that it is impossible to infer associations between changes in symptoms and LMWPs by assessing only Kt/V and β₂-MG removal rates.

The symptoms that occurred were either the development or aggravation of bone and/or joint pain (10/13 cases), reduced activity (7/13 cases), pruritus (4/13 cases), RLS (3/13 cases), irritability (2/13 cases), etc. Dialysis vintage in these 13 cases was 21.0 ± 10.1 years, and dialysis vintage in the group that developed or experienced aggravation of bone and/or joint pain was 24.3 ± 9.0 years (group without these changes: 8.4 ± 4.1 years). When the dialysis vintages and the symptoms that developed were reviewed together, these episodes showed that continuing HDF at an α₁-MG removal rate of 35% is advisable to prevent the onset of DA or mitigate these symptoms.

RLS symptoms have been classified into four stages by International Restless Legs Syndrome score (IRLS score): 1-10, mild; 11-20, moderate; 21-30, severe; 31-40, very severe [13]. Even a review of the IRLS score and α₁-MG removal rate together in the 7 cases in which RLS developed as a complication revealed that while the symptoms were alleviated up to an α₁-MG removal rate of 35%, the patients' RLS was not cured, and that an α₁-MG removal rate of 38% or more was necessary in order to cure their RLS (fig. 1). This shows that it is impossible to hope for a cure of RLS even by increasing removal of uremic toxins from a molecular weight of 10-20 kDa, and that aggressively removing uremic toxins having molecular weights of 30 kDa or more leads to a cure of RLS [14].

These assessments showed that α₁-MG should be used to correctly evaluate the distinctive features and benefits of HDF.

α₁-MG has a molecular weight of 33 kDa and a Stokes radius of 28.4 Å. Its predialysis values are not correlated with dialysis vintage, and no associations between predialysis values and dialysis complications have been found in any studies conducted thus far (fig. 2). The predialysis α₁-MG values of dialysis patients in our institution were 117.0 ± 19.9 mg/l (n = 126), and the mean value of healthy subjects was 12.5 ± 2.3 mg/l (n = 31).

β₂-MG is known to be the substance that causes DA, and many studies have been conducted regarding its etiology and β₂-MG. β₂-MG has a molecular weight of 11.8 kDa and a Stokes radius of 16 Å. The β₂-MG predialysis values of dialysis patients in our institution were 26.5 ± 5.3 mg/l (n = 126), and the mean value of healthy subjects was 1.4 ± 0.3 mg/l (n = 31).

The changes in α₁-MG and β₂-MG values between August 2011 and August 2012 in all of the cases as a whole (n = 126) were: α₁-MG, 119.8 → 117.0 mg/l, and β₂-MG, 27.4 → 26.5 mg/l, and thus there was hardly any change in either of them. In an assessment of the HDF group (n = 11) alone, the changes were: α₁-MG, 119.1 → 120.3 mg/l, and β₂-MG, 24.0 → 22.9 mg/l, and a tendency for the β₂-MG values to decrease was observed, but the difference was not significant.

Many studies have been conducted in this field since the Gejyo study in which β₂-MG was identified as the precursor protein of DA, but no correlations have been reported between predialysis β₂-MG values and the onset of DA or the severity of the symptoms [8,9,15,16,17]. Nor have there been any reports of studies that investigated the relationship between α₁-MG and dialysis complications. Based on our study as well, at present predialysis α₁-MG values have not been found to indicate the severity of the pathology or the appropriateness of the dialysis conditions, etc., but further study is needed.

Uremic toxins whose molecular weights are greater than that of β₂-MG may be involved in the development of clinical manifestations such as insomnia, pruritus, irritability, RLS, anemia, and osteoarticular pain.For that reason, it has been said that it will be necessary to remove a wider range of LMWPs than β₂-MG and to remove protein-bound toxins in order to treat these symptoms, and the same appeared to be true in the present study of our own cases.

We wish to emphasize that the reason for removing α₁-MG is not because α₁-MG itself is toxic, but because α₁-MG has a molecular weight of 33 kDa and should be used as a biomarker for evaluation of clinical outcomes of HDF. We also concluded that an increase in therapeutic efficacy, an increase in patient QOL, and a higher survival rate can be expected when the distinctive features and advantages of HDF are best utilized by setting the HDF performance conditions with an α₁-MG rate of 35% as the goal.

When HDF is performed under these conditions, a β₂-MG removal rate of 80% can be easily achieved. On the other hand, even if the replacement fluid volume is increased, a β₂-MG removal rate of 80% and α₁-MG removal rate of 15% are likely to be achieved by HDF with a small-pore size dialyzer or hemodiafilter.

We used two high-flux dialyzer models (FX, Fresenius Medical Care; FDX, Nikkiso Co., Ltd.) and one super-high dialyzer model (FDY, Nikkiso) having a dialysis area of 1.8 m² and assessed the relationship between β₂-MG and α₁-MG removal rates and the amount of albumin loss when HD, predilution on-line HDF, and postdilution on-line HDF were performed (fig. 3). Although there was a large amount of albumin loss when we performed HDF with each of the dialyzers, β₂-MG removal rate was almost constant. This means that β₂-MG is mainly removed by diffusion.

There is a twofold difference between the molecular weight of albumin (MW: 66 kDa, Stokes radius: 35.5 Å) and α₁-MG, but there is a mere 20% difference in their Stokes radius. Because of the difference in their Stokes radius, it is impossible to differentially remove α₁-MG and albumin with the dialysis membranes currently in use, and a correlation exists between the amount of α₁-MG removed and the amount of albumin loss [18]. There is a strong likelihood that a certain degree of albumin loss has a positive impact on the body from the standpoint of removing uremic toxins that fall into the category of protein-bound solutes and of removing albumin that has lost its antioxidant capacity [19,20,21].

We performed 50-liter predilution online HDF (for 4 h) at Q_B 250 ml/min (Q_D total 500 ml/min) with TDF-20H (Toray Medical Co., Ltd), ABH-21P (Asahi Kasei Medical Co., Ltd), and MFX-21U (Nipro Co.), and -assessed removal performance (fig. 4). These three products can be described as high-performance versions of hemodiafilters that are currently on the market in Japan.

Under these assessment conditions, all three models yielded a good β₂-MG removal rate of 80% [amount removed (mg): TDF, 186; ABH, 169; MFX, 190], but there were large differences in their α₁-MG removal rates: TDF, 18.2% (amount removed: 93 mg), ABH, 25.9% (amount removed: 126 mg), and MFX, 37.5% (amount removed: 179 mg), respectively. Thus, differences between the performance of the hemodiafilters became increasingly marked as molecular weight increased from the 11.8 kDa for β₂-MG, to 23 kDa for prolactin, and to 33 kDa for α₁-MG.If the removal performance of the hemodiafilters had been evaluated on the basis of β₂-MG alone, the results for all three hemodiafilter models would have been the same.In other words, to make the best use of the distinctive features and benefits of HDF, it is important to use α₁-MG as the biomarker to evaluate the removal performance of HDF, and to choose a hemodiafilter that provides an α₁-MG removal rate of 35%. An improvement in clinical manifestations can be expected by doing so, and it increases patients' QOL and, in turn, their survival rate.

The removal performance of HDF should be assessed by using α₁-MG as a biomarker. Further, the distinctive features and benefits of HDF are best utilized by selecting a suitable hemodiafilter and setting an α₁-MG removal rate of 35% as the target value.

I have no relationship with the industry or financial associations that might pose a conflict of interest in connection with the submitted article.