Which kidney failure patients would benefit most from HDF instead of HD?

Younger patients and those with fewer comorbidities benefit most from haemodiafiltration (HDF) instead of conventional haemodialysis (HD), according to a multitrial model presented at ERA 2025.

Individual-level data from 4,153 patients predicted that, compared with HD, HDF would overall extend survival by a median of 6.9 months. Moreover, the use of high-dose HDF, defined as a convection volume ≥23 L per session, would improve survival by a median of 8 months relative to HD. [ERA 2025, abstract 3321]

“First, we can see that compared with HD, there are no patients who would not benefit from HDF; in other words, all patients would have some benefit from HDF,” said Dr Robin Vernooij, a clinical epidemiologist at the Department of Nephrology and Hypertension, University Medical Center Utrecht, the Netherlands. “If we stratify it by high-volume HDF, the absolute effect is even greater and clearly [points] towards a more extreme gain in survival benefit.”

A notable finding was the prediction of which patients are expected to survive for ≥24 months with HDF vs HD. These patients were mostly younger (median age 25 years), half were male, and had a low comorbidity burden (none had diabetes, 4 percent had a history of cardiovascular [CV] disease). They had a median predialysis serum creatinine concentration of 9.7 mg/dL, a mean post-dialysis BMI of 22.7 kg/m², a mean serum albumin concentration of 4.1 g/dL, and a median C-reactive protein (CRP) level of 0.9 mg/L.

From the model-estimated patient profiles of those expected to have a survival benefit of <6 months, ≥6 and <12 months, ≥12 and <24 months, and ≥24 months, respectively, there was a “clear trend” indicating that  patients who would benefit most from HDF vs HD were younger, had a low comorbidity burden related to diabetes and CV disease, higher serum creatinine and albumin levels, and lower inflammation as indicated by CRP levels, as Vernooij described.

“It is important to note that these  characteristics are commonly available in routine clinical care,” added Vernooij. “We can use them for individualized modelling to prioritize care for those who would benefit the most from HDF.”

Survival benefit up to a couple of years

The five randomized controlled trials included in the prediction model are CONVINCE, ESHOL, the Turkish study, CONTRAST, and the French study, all of which were primarily conducted in Europe. [Lancet 2024;404:1742-1749]

“There’s a wide variation,” commented Vernooij on the predicted gain in survival. “You can appreciate that some patients only have a couple of weeks of extra survival from HDF, in contrast to other patients who would have 30 or 40 months of survival gain from HDF vs HD, which is almost 3 years.”

Individualized care model

Unlike group-level average treatment effects derived from subgroup analyses, individualised treatment effect prediction through model development can better address the heterogeneity between individuals and simulate personalised medicine to a greater extent. [Clin Kidney J 2022;15:1924-1931]

“A patient is never [defined by] one single factor,” said Vernooij. “So, we moved to a better individualized care model where we explored if certain predictors or features can better estimate the individual treatment benefits from HDF vs HD.”