Reappraisal of Hemodiafiltration for Managing Uremic Complications

In this issue of CJASN, Kang et al. report on a study comparing the effects of high-flux hemodialysis (HD) with online hemodiafiltration on neuropathy (¹). Although two prior small studies found an improvement of nerve function in patients who switched to hemodiafiltration, this randomized controlled trial (RCT) could not confirm these findings. Actually, this is just one more investigation that disproves the beneficial effect of hemodiafiltration on a clinical outcome described in nonrandomized studies. Whereas dialysis fluid was not ultrapure in the HD arm of several older studies, in this study, both HD and hemodiafiltration were performed using ultrapure dialysis fluid and a high convection volume in hemodiafiltration. So, why are the results of well-performed studies on intermediate end points with hemodiafiltration negative, while a recent meta-analysis of four RCTs showed that hemodiafiltration is associated with a lower mortality risk (²)? How can hemodiafiltration affect mortality, without beneficial effects on individual organs? Let us go back to the early days of dialysis.

Of the first 15 patients with acute kidney injury dialyzed by Kolff during World War II, only one survived. In the following decades, major developments in patient care and dialysis technology led to a massive increase in the number of maintenance dialysis patients worldwide. Since clinical outcome was extremely poor in the early days, Henderson et al. performed the first hemofiltration in 1975 in an attempt to remove toxic middle molecular weight substances, which were held to be responsible for the uremic syndrome (³). In subsequent years, this concept was refined by combining the diffusive capacity of HD with the convective features of hemofiltration, which was called hemodiafiltration. Membranes were developed with high sieving coefficients to allow the passage of solutes up to the size of albumin. Online production of ultrapure dialysis fluid and sterile, nonpyrogenic substitution fluid enabled treatment with high convection volumes. Only after these developments could hemodiafiltration grow toward its current share of 10% of dialysis treatments worldwide. Yet, the evidence that modern online postdilution hemodiafiltration improves clinical outcome is scanty. In the following paragraphs we will shortly touch upon the effects of hemodiafiltration on selected biomarkers, organ (dys)function, quality of life (QoL), and survival.

While prior studies found less oxidative stress and inflammation with hemodiafiltration, the differences with HD disappeared when ultrapure dialysis fluid was used in both modalities. Although a large RCT showed differences in C-reactive protein levels despite ultrapure dialysis fluid in both arms (⁴), the clinical impact of this finding remains unclear because the change was only 1 mg/L over 3 years. As for middle molecular weight compounds, hemodiafiltration can remove an array of substances, including β₂-microglobulin (molecular weight, 11.8 kiloDalton) and fibroblast growth factor 23 (molecular weight, 32 kiloDalton). Yet, although a clear relation was observed between these substances and survival, additional lowering by hemodiafiltration did not result in superior clinical outcomes. Considering anemia control, conflicting results are reported. While observational and crossover studies suggested improvements in erythropoiesis-stimulating agent resistance and iron use, three RCTs were inconclusive, varying from no effect to a reduction in erythropoiesis-stimulating agent requirements.

As outlined below, the beneficial effect of hemodiafiltration on survival is mainly due to a reduced risk of fatal cardiovascular events. It should be mentioned, however, that cardiovascular disease in patients with kidney failure is determined not only by uremic alterations of the heart and blood vessels, but also by an inappropriate vascular response to hormonal and neurologic stimuli. In addition, fluid overload and hypertension play an important role. So far, however, it has not been convincingly shown that hemodiafiltration improves any factors such as left ventricular mass and vascular stiffness, compared with HD. Considering the deterioration of residual kidney function, only a few studies have addressed the issue. Neither a large observational study nor a recent RCT showed any longitudinal differences between HD and hemodiafiltration in this respect. As for neurologic complications, which can be categorized into peripheral and central, both neuropathy of the limbs and impaired cognition are reported in >70% of the patients. Yet, randomized comparisons between HD and hemodiafiltration have not been performed so far.

As depicted in Figure 1, organ dysfunction in kidney failure does not only influence clinical outcome, but also health-related QoL (HRQoL). In this respect, peripheral neuropathy is an illustrative example, as it can be quantified objectively by measuring nerve conductivity, reflexes, and sensibility and subjectively by assessing patients’ symptoms. It is increasingly recognized that HRQoL, which includes domains of physical, mental, emotional, and social functioning, is a relevant outcome measure in kidney failure. In two large RCTs comparing HD with hemodiafiltration, HRQoL, as evaluated with the Kidney Disease Quality of Life Short Form (KDQOL-SF) questionnaire, did not differ after 2 years of treatment (⁵, ⁶). However, specific symptoms might not be captured by a QoL questionnaire. In fact, neuropathic symptomatology is only one of the >70 (sub)questions of the KDQOL-SF (“numbness in hand or feet”) or the 30 questions of the Dialysis Symptom Index (“numbness or tingling in feet”). Hence, an effect of treatment on neuropathy, although relevant, might easily be missed. In this issue, Kang et al. report on the effect of hemodiafiltration on neuropathy (as measured by a symptom score, nerve conduction, and a sensory test) in prevalent dialysis patients (¹). From a methodological point of view, the FINESSE study has notable strengths. Follow-up was 4 years, and neuropathy was assessed by gold standard methods. Both HD and hemodiafiltration were executed according to current state-of-the-art criteria, with adequate treatment times, high convection volumes, and ultrapure dialysis fluid. Outcome was assessed with linear mixed models. Since only 4% of treatments were executed in the predilution mode, and treatment adherence in the hemodiafiltration group was 82.6%, the convection volumes in patients undergoing postdilution hemodiafiltration may have been somewhat lower than the reported 24.7 L/treatment, but—hopefully—not too low to reverse neuropathic symptoms. Points of concern are the lower-than-expected baseline prevalence of moderate to severe neuropathy (27%–34% as compared with the expected 80% after 4 years in the control group) (⁷), the low number of patients who completed 4 years of follow-up (88 at 2 years, 61 at 4 years), and the imperfect randomization as manifested by a higher proportion of patients in the hemodiafiltration group with diabetes and peripheral vascular disease, which are both associated with a high prevalence of polyneuropathy. Notwithstanding these issues, a carefully executed RCT on a clinically relevant end point, as reported by Kang et al., is an important contribution to the positioning of hemodiafiltration in treating kidney failure.

Figure 1.: Complex interplay between uremic toxins, fluid overload, and organ dysfunction in patients with kidney failure. As compared with hemodialysis, hemodiafiltration lowers middle molecular weight uremic toxins more effectively and appears to be related to better survival (circles in orange).

Despite the absence of clear benefit on biochemical and intermediate clinical end points, the life expectancy of patients with kidney failure appears prolonged with high-volume hemodiafiltration. Four RCTs comparing HD with hemodiafiltration have been published. Although individual studies were inconclusive, a meta-analysis showed superior all-cause and cardiovascular survival with hemodiafiltration (²). The largest mortality reduction was obtained with the highest convection volume (>23 L/1.73 m² per session). Other causes of mortality, such as infections and malignancies, were not different.

The beneficial effect of hemodiafiltration on survival was initially suggested by studies with evidence levels B and C, and later confirmed by three level A investigations, including one RCT, one meta-analysis on individual patient data, and one Cochrane meta-analysis on aggregated data (²,⁸). In both meta-analyses, the inclusion of the Estudio de Supervivencia Hemodiafiltración On-Line (⁹) heavily contributed to the outcome, as this study provided 26%–32% of patients. Yet in this study, which is the only RCT showing a clear survival benefit of hemodiafiltration, study groups were not well balanced, since hemodiafiltration patients were younger and had a lower Charlson Comorbidity Index and less often a central venous catheter. Individuals who could not achieve the target convection volume were excluded. Since the latter criterion in particular may have prevented the inclusion of frail patients in the hemodiafiltration arm, the outcome may be influenced by selection bias.

Therefore, we are eagerly awaiting the outcome of two large ongoing RCTs comparing high-flux HD with postdilution high-volume hemodiafiltration to clarify whether hemodiafiltration is indeed associated with lower mortality (NTR7138 NTR [trialregister.nl]; https://doi.org/10.1186/ISRCTN10997319). Since HRQoL and patient-reported outcome measures are prespecified secondary objectives and both studies are large enough to avoid randomization disbalances between groups, the question of whether hemodiafiltration has a beneficial effect on quality of life will hopefully also be solved in the near future.

Disclosures

Both authors report unrestricted research grants from Niercentrum aan de Amstel, Amstelveen, The Netherlands, and from B. Braun Avitum, Germany.

Funding

None.