**3. Studies comparing the effect of hemodiafiltration versus hemodialysis on arterial stiffness**

Uremic toxin accumulation, particularly retention of protein-bound solutes and middleweight molecules as a result of their inadequate clearance through conservative dialytic modalities, is proposed to play a prominent role in promoting vascular atherosclerosis and pathogenesis of cardiovascular disease among dialysis patients. In support of this notion, background and clinical studies have shown that accumulation of *p*-cresol and indoxyl sulfate, two protein-bound uremic toxins, acts as a triggering factor for the expression of pro-inflam‐ matory cytokines and adhesion molecules, induces shedding of endothelial microparticles, and disrupts the nitric oxide signaling pathway [27–29]. Importantly, in recent prospective observational studies, high concentrations of both *p*-cresol and indoxyl sulfate in hemodialysis patients have been associated with increased risk of cardiovascular morbidity and mortality independently from other traditional cardiovascular risk factors [30–32].

Hemodiafiltration is a dialytic modality that uses a combination of convective transport and diffusion to enhance the removal of middle-to-high molecular weight solutes in comparison with standard hemodialysis [33, 34]. In some clinical studies, enhanced middle-molecule clearance achieved through convective dialytic modalities was shown to be associated with improvement in phosphate control, better preservation of intradialytic hemodynamic stability, as well as, with a number of beneficial actions on vasculature, such as reduction in circulating markers of vascular inflammation and oxidative stress and improvement in flow-mediated endothelium-dependent vasodilatation [11–14, 35]. These beneficial effects of hemodiafiltra‐ tion on the vasculature have generated the hypothesis that switching ESRD patients from conventional hemodialysis to high-efficiency hemodiafiltration may be a therapeutic maneuver with potential advantages in causing regression of arterial stiffness. This hypothesis was tested in a number of clinical studies summarized in **Table 1** and discussed in detail below.

**Figure 1.** Pathophysiological role of increased arterial stiffness in ESRD.

**on arterial stiffness**

48 Advances in Hemodiafiltration

The close pathophysiological association of arterial stiffness with promotion of end-organ damage is in line with a strong epidemiological association of increased arterial stiffness with worse cardiovascular outcomes. Among dialysis patients, prospective observational studies have for long-connected higher aortic PWV with increased risk of all-cause and cardiovascular mortality independently from other cardiovascular risk factors [7]. In the first study conducted in the late 1990s in a cohort of 241 hemodialysis patients prospectively followed for a mean period of 6 years, Blacher et al. [6] showed that the fully adjusted odds ratio (OR) for aortic PWV > 12.0 versus PWV < 9.4 m/s was 5.4 [95% confidence intervals (CIs): 2.4–11.9] for allcause mortality and 5.9 (95% CI: 2.3–15.5) for cardiovascular mortality [6]. The strong prog‐ nostic association between aortic PWV and cardiovascular outcomes was confirmed in several subsequent cohorts of hemodialysis patients [15, 23]. Similarly to patients receiving hemo‐ dialysis, more recent observational studies have demonstrated the strong and independent prognostic significance of arterial stiffness in the whole spectrum of CKD, showing that aortic PWV is an independent predictor of mortality in patients receiving peritoneal dialysis [24] in renal transplant recipients [25] and in patients with CKD not yet on dialysis [26]. Most importantly, regression of arterial stiffness in response to BP-lowering was shown to be associated with improvement in survival [9], providing evidence that arterial stiffness is not simply a risk predictor, but a true cardiovascular risk factor in the dialysis population.

**3. Studies comparing the effect of hemodiafiltration versus hemodialysis**

Uremic toxin accumulation, particularly retention of protein-bound solutes and middleweight molecules as a result of their inadequate clearance through conservative dialytic



**Table 1.** Prospective studies comparing the effect of hemodiafiltration versus standard hemodialysis on arterial stiffness.

#### **3.1. Acute effects of hemodiafiltration on arterial stiffness**

In a nested case-control design, Georgianos et al. [36] compared the acute changes in aortic PWV from pre- to postdialysis in 24 ESRD patients receiving hemodiafiltration and in 24 ageand sex-matched controls receiving low-flux hemodialysis. Aortic PWV was not significantly changed from pre- to postdialysis during both the first and second weekly dialysis sessions. These modest acute changes in aortic PWV from pre- to postdialysis were no different between the hemodiafiltration and low-flux hemodialysis groups in both dialysis sessions studied [36]. With regards to wave reflections, augmentation index and related parameters were signifi‐ cantly reduced from pre-to postdialysis in both dialysis sessions and patient groups. Similarly to aortic PWV, intradialytic reduction in wave reflection indices was no different between patients treated with hemodiafiltration and standard low-flux hemodialysis [36]. These findings suggest an acute intradialytic improvement in wave reflections from the periphery but not in aortic stiffness, an effect that was independent of the mode of dialysis. However, these comparable acute alterations in large-artery cushioning function do not necessarily prespecify the pattern of long-term progression of the arterial stiffness in patients treated with different dialytic modalities.

#### **3.2. Long-term effects of hemodiafiltration on arterial stiffness**

**Author Year n Patient**

50 Advances in Hemodiafiltration

Mostovaya et al. [39]

Charitaki et al. [38]

Georgianos et al. [36]

stiffness.

**characteristics**

participating in the CONTRAST

2014 189 ESRD patients

trial

2014 289 ESRD patients

HD

2014 48 ESRD patients

HD

HD, hemodialysis; RCT, randomized controlled trial.

**3.1. Acute effects of hemodiafiltration on arterial stiffness**

on maintenance

on maintenance

**Design Intervention Follow-up**

line HDF versus low-flux HD

RCT Post-dilution on-

Observational 69 patients on

Observational HDF versus lowflux HD

**Table 1.** Prospective studies comparing the effect of hemodiafiltration versus standard hemodialysis on arterial

In a nested case-control design, Georgianos et al. [36] compared the acute changes in aortic PWV from pre- to postdialysis in 24 ESRD patients receiving hemodiafiltration and in 24 age-

low-flux HD versus 78 patients switched from low-flux HD to HDF versus 142 patients on HDF

**(months)** 

**Change in arterial stiffness over time**

36 Aortic PWV remained unchanged over time with both dialytic modalities (annual rate of PWV change: HDF group: −0.01, 95% CI: −0.41 to 0.40 m/s/year; HD group: −0.04, 95% CI: −0.31 to 0.23 m/s/ year; p value HDF versus HD: 0.89)

6 Aortic PWV increased over time in the lowflux HD group (9.5 ± 1.9 versus 10.2 ± 2.2 m/s, p < 0.01) as well as in the HD to HDF group (9.4 ± 1.9 versus 10.1 ± 2.2 m/s, p < 0.01), but remained constant in the HDF group (9.9 ± 2.1 versus 10.1 ± 2.2

m/s)

Aortic PWV remained unchanged from preto postdialysis either with HDF or with lowflux HD (HDF: 9.3 ± 0.5 versus 9.4 ± 0.5 m/s, p = 0.686 low-flux HD: 9.1 ± 0.4 versus 8.9 ± 0.4 m/s, p = 0.396)

Single dialysis session

**Overall effect**

Neutral

Better

Neutral

Studies evaluating the long-term effects of hemodiafiltration relative to hemodialysis on arterial structure and function provided contradictory results. Beerenhout et al. [35] random‐ ized 40 ESRD patients treated with conventional low-flux hemodialysis to switch to highefficiency pre-dilution on-line hemodiafiltration or to continue on the same dialytic modality. After 12 months of follow-up, aortic PWV increased similarly in both low-flux hemodialysis (12 ± 3 versus 13 ± 5 m/s) and on-line hemodiafiltration groups (12 ± 3 versus 13 ± 5 m/s). Notably, change in 48-h ambulatory systolic and diastolic BP and in LV mass index over time were also no difference between the two dialytic modalities [35]. Furthermore, on-line hemodiafiltration was not superior to low-flux hemodialysis in inhibiting the formation of advanced glycation end-products and reducing the circulating levels of asymmetric dimethy‐ larginine (ADMA) and markers of oxidative stress or total anti-oxidant capacity. In a subse‐ quent study, 42 ESRD patients were randomly assigned to switch from high-flux conventional hemodialysis to high-efficiency post-dilution on-line hemodiafiltration or to remain on highflux hemodialysis for a mean follow-up period of 4 months [37]. Arterial stiffness assessed with the use of the distensibility co-efficient of the common carotid artery was improved in the on-line hemodiafiltration group, but not in the high-flux conventional hemodialysis group (between-group difference: −6.7 kPa−1 × 10−3, 95% CI: −9.9 to −3.5 kPa−1 × 10−3, p = 0.048) [37]. Improvement in carotid artery distensibility was accompanied by a significant improvement in Kt/V urea, predialysis levels of β2-microglobulin, circulating levels of ADMA and tumor necrosis factor (TNF)-a, and brachial artery flow-mediated endothelium-dependent vasodila‐ tation. In a multiple regression analysis model, hemodiafiltration-induced improvement in conduit artery endothelial dysfunction and stiffness was associated with the changes in Kt/V urea and predialysis levels of β2-microglobulin, suggesting that enhanced clearance of middleto-high molecular weight solutes is one factor potentially contributing to the beneficial effect of hemodiafiltration on large-artery stiffness.

A beneficial effect of hemodiafiltration on arterial stiffness is supported by another observa‐ tional study, in which aortic PWV measurements were performed 6 months apart in three different groups of ESRD patients [38]. The first group consisted of 69 ESRD patients receiving conventional low-flux hemodialysis, the second group consisted of 78 ESRD patients who were switched from low-flux hemodialysis to on-line hemodiafiltration, and the third group included 142 ESRD patients receiving long-term renal replacement therapy with on-line hemodiafiltration. Over the 6-month observational period, a significant increase in aortic PWV was noted in those patients treated with hemodialysis (9.5 ± 1.9 versus 10.2 ± 2.2 m/s, p < 0.01) as well as in those switched from hemodialysis to hemodiafiltration (9.4 ± 1.9 versus 10.1 ± 2.2 m/s, p < 0.01); in contrast, aortic PWV remained unchanged in the group of hemodiafiltration (9.9 ± 2.1 versus 10.1 ± 2.2 m/s) [38]. The most important finding of this study was that aortic PWV remained constant during follow-up only in those patients receiving long-term treatment with hemodiafiltration, whereas aortic PWV increased in patients who were switched from hemodialysis to hemodiafiltration. This observation could be interpreted in two different ways: either the 6-month-long therapy with hemodiafiltration might be inadequate in order to modify the arterial wall structure and stiffness, or aortic PWV increased in those patients switched to hemodiafiltration due to a carry-on effect of previous long-term therapy with conventional hemodialysis.

The above beneficial impact of hemodiafiltration in causing regression of arterial stiffness was not confirmed in a recent subanalysis of 189 prevalent dialysis patients participating in the CONvective TRAnsport STudy (CONTRAST) trial [39]. In this study, ESRD patients receiving conventional low-flux hemodialysis were randomly assigned in a 1:1 ratio for treatment with on-line hemodiafiltration or continuation of low-flux hemodialysis for a mean follow-up period of 36 months. Median aortic PWV at baseline was 9.8 m/s (interquartile range: 7.5–12.0 m/s). Aortic PWV was not significantly changed over time, and the annual rate of PWV change had no difference between the on-line hemodiafiltration and hemodialysis groups (hemodia‐ filtration group: −0.01 m/s/year, 95% CIs: −0.41 to 0.40 m/s/year; hemodialysis group: −0.04 m/ s/year, 95% CI: −0.31 to 0.23 m/s/year; p value for the between-group comparison: 0.89) [39]. The absence of difference between the two dialytic modalities in the rate of PWV change was consistent across subgroups of age, sex, residual renal function, dialysis vintage, diabetes, and history of pre-existing cardiovascular disease. Of note, the annual rate of PWV change had once again no difference between the two dialytic modalities regardless of the convection volume used for on-line hemodiafiltration (convection volume <18.9 L/session: 0.37 m/s/year; 95% CIs: −0.25 to 0.98 m/s/year, p = 0.23; convection volume >18.9 L/session: −0.01 m/s/year; 95% CIs: −0.59 to 0.57, p = 0.99) [39].
