COMMENTVolume 404, Issue 10464P1703-1705November 02, 2024Download Full Issue

Haemodiafiltration improves survival in patients receiving dialysis

Bruno Ranchin, Rukshana Shroff

Lancet 2024; 404: 1703-1705

Chronic kidney disease is a global public health problem, affecting about 10% of the world's population,¹ with increasing numbers requiring dialysis. Haemodialysis is the most commonly used dialysis modality, accounting for 89% of long-term dialysis worldwide.²Haemodialysis saves lives, but offers a 5-year survival of only 50% even in high-income countries,^3,4 and carries a high burden of cardiovascular morbidity and results in inestead of carries.² Increasing the dialysis dose, and thereby improving the urea clearance, has been shown not to improve survival.⁵ This led to a focus on middle molecular weight uraemic toxins that cannot be eliminated by the simple diffusion offered by haemodialysis, and resulted in the development of haemodiafiltration as a technique for maintenance dialysis.⁶

Haemodiafiltration is a blood purification technique combining both diffusive and convective solute removal by ultrafiltration of at least 20% of the blood volume processed through a high-flux dialyser.⁷ To achieve this high ultrafiltration rate and maintain fluid balance, haemodiafiltration requires an infusion of sterile replacement fluid into the patient's blood; this is called the convection volume. Higher convection volumes (greater than 23 L per session; determined by statistical stratification from randomised trials) closely correlate with an increased clearance of β2-microglobulin, a prototype middle molecule, and translate to improved survival.⁶

To date five randomised controlled trials (RCTs) have been performed to compare all-cause and cardiovascular mortality between haemodiafiltration and haemodialysis in adults receiving maintenance dialysis for at least 1 year. Of these, the CONVINCE⁸ and ESHOL⁹ trials that consistently achieved high convection volumes, showed a significant reduction in all-cause mortality with haemodiafiltration compared with high-flux haemodialysis. Notably, there was marked heterogeneity across these RCTs with the use of low-flux or high-flux haemodialysis and variable low-dose or high-dose haemodiafiltration techniques, vascular access types, and the actual delivered convection volume. In addition, substantial methodological shortcomings, including informative selection bias with the exclusion of patients with insufficient blood flow rate from the haemodiafiltration group after being randomly assigned, have confounded results, and current clinical guidelines have not reached a consensus on the treatment benefit of haemodiafiltration. Also, none of the RCTs were powered for subgroup analyses, nor did they systematically adjust for potential confounders.¹⁰

In this issue of The Lancet, Robin W M Vernooij and colleagues¹¹ present a meta-analysis of pooled individual patient data from the five RCTs (conducted in Canada, France, Germany, Hungary, the Netherlands, Norway, Portugal, Romania, Spain, Türkiye, and the UK), assessing subgroup effects with higher power and better adjustment for potential confounders. The analysis involves 4153 patients (mean age 63·5 years [SD 14·3], 1549 [37%] were female, 1295 [32%] had diabetes, and 1601 [40%] had a history of cardiovascular disease) randomly assigned to receive haemodialysis or haemodiafiltration in a thrice weekly schedule with an average of 4 h per session, with additional follow-up data on survival collected for 541 (91%) of 597 patients who were censored alive at the point of discontinuing randomised treatment. The analysis showed that all-cause mortality was 16% lower in the haemodiafiltration than in the haemodialysis group (hazard ratio 0·84, 95% CI 0·74–0·95) after a median follow-up of 30 months (IQR 24–36). Also, the study reported a 22% reduction in cardiovascular mortality in the haemodiafiltration group compared with the haemodialysis group (hazard ratio 0·78, 95% CI 0·64–0·96), largely driven by a 33% reduction in cardiac mortality. Although there is a trend towards differences in all-cause mortality—paradoxically older patients appear to have a survival advantage—the wide overlap in CIs negates between-group differences. Similarly, for cardiovascular mortality, women, older patients, and those without diabetes appear to have a non-statistically significant trend towards improved survival. When multivariable-adjusted analyses were restricted to patients who achieved a convection volume greater than 23 L per session, the differences with the haemodialysis group increased. While improved cardiac or cardiovascular outcomes (variously adjudicated in different RCTs) have been shown with haemodiafiltration in several observational studies and RCTs, notably in CONVINCE, the largest haemodiafiltration trial, there was no change in cardiovascular mortality, but instead a reduction in infection-related deaths in the haemodiafiltration group; these hypothesis-generating data require further exploration.

The Article by Vernooij and colleagues is the largest and most rigorous analysis available to date. Yet, individual patient data analysis can only be as good as the individual trial data input into it. Most importantly, the RCTs do not represent the real-world situation seen in dialysis clinics; the RCTs have a younger population,^3,4 with a lower prevalence of diabetes^3,12 and cardiovascular disease,^2,3,12 and a higher prevalence of arteriovenous fistula use,³ who are therefore more likely to achieve higher blood flows and the desired convection volume that are associated with superior outcomes. While an absolute convection volume greater than 23 L per session is associated with the greatest chance of survival, standardising convection volumes to body surface area might reveal that small (ie, both in terms of height and weight) patients achieve superior outcomes even at somewhat lower convection volumes. Furthermore, some outcome measures could show an improvement at lower convection volumes depending on the kinetics of different uraemic toxin clearance. Also, as convection volume depends on the blood flow rate (for an equivalent treatment time), these data would be strengthened by exploring an association between the blood volume treated and mortality in the haemodialysis groups of these RCTs. Finally, the combination of comorbidities that is often seen in patients on dialysis requires further study. Further studies are also needed to confirm the better preservation of health-related quality of life on haemodiafiltration reported in a sub-analysis of the CONVINCE trial.¹³

To conclude, this study harnesses the power of cumulative evidence from RCTs, reinforcing the superior survival outcomes seen with haemodiafiltration compared with conventional haemodialysis that have been reported in some RCTs.^8,9 Haemodiafiltration therapy provides hope for the growing dialysis populations worldwide who have among the highest mortality and the lowest quality of life reported in patients with chronic diseases.