Hemodiafiltration (HDF) integrates diffusive and convective solute removal, enhancing the clearance of middle-to-large molecular weight toxins, which are associated with cardiovascular morbidity. Compared to conventional high-flux hemodialysis (HD), online post-dilution HDF is increasingly recognized as beneficial, particularly at higher convection volumes (≥23 L/session). With conventional HD, ultrafiltration volume usually exceeds the desired weight loss but is automatically compensated by backfiltration. So, in fact this is a form of “low dose” HDF with volumes up to a few liters per session. This volume is unmeasurable, uncontrollable, unpredictable, and cannot be dosed.

Over the past few years, some new data on HDF has become available. This short comment discusses recent evidence on clinical and patient-reported outcomes, feasibility, economic considerations, personalized treatment, and practical implications, which hopefully is helpful for nephrologists in Korea. It will not address the use of HDF in the pediatric setting, in which evidence suggests that it is superior to standard HD as well.

Economic analysis of the recent CONVINCE estimates that HDF is potentially cost-effective. The primary cost driver is increased survival rather than per-session costs. Cost-effectiveness remains inherently multifactorial and context-specific, influenced by local resources, clinical practices, and healthcare systems [7]. Indeed, some investment may be necessary because dialysis machines need to be able to deliver HDF, and some training of the staff is mandatory to fully understand the variables determining the achievement of convection volumes, as will be discussed below.

However, one may think that there are practical adoption challenges, such as increased water use and the requirement for ultrapure water production infrastructure. In a recent paper, it was shown by using modelling supported by real-world data that, although it may be counterintuitive, the use of HDF may be associated with lower instead of higher water and energy use than standard HD [8]. This is the first paper addressing this important topic. The observations need to be confirmed by others.

Independent of what regulatory bodies have defined as quality standards for dialysate, it is important to realize that in fact both high-flux HD and online HDF should be considered needing similar water quality, i.e., HDF level. The reasoning for that is simple. Applying different quality standards for a treatment modality characterized by the infusion directly into the patients of “a few liters” (passive back filtration as in standard high-flux HD) versus 23 L or more (direct infusion into the extracorporeal circuit, as in online HDF) simply does not make sense.

Achieving high-volume HDF (≥23 L/session in post-dilution setting) is feasible in routine clinical practice. In the CONVINCE study mean convection volume was around 25 L per session and remained stable during follow-up. Real-world registries consistently demonstrate that high convection volumes can be delivered in the majority of sessions.

Because of the clear relation between dose (i.e., convection volume) and outcome, nephrologists should understand the practical aspects of prescribing online HDF. The primary determinant of clinical efficacy in online HDF is the convection volume delivered per session. A commonly referenced convection volume target based on the recent trials is at least 23 L per session. This represents a minimum effective threshold rather than a strict cutoff. Indeed, recent evidence strongly supports a dose-response relationship, emphasizing that achieving even higher convection volumes can confer additional survival benefits. Accordingly, it is beneficial to pursue a “the higher, the better” approach within safe and practical limits [2]. This also means that achieving lower volumes is not wrong or detrimental, but simply less effective.

Several modifiable treatment parameters can influence the achieved convection volume, providing clinicians with flexibility to tailor dialysis prescriptions according to individual patient characteristics, vascular access conditions, and personal preferences.

Blood flow rate (QB) directly impacts convection volume in online HDF. For optimal convective efficacy, QB should ideally be 350 mL/min and higher. This is particularly facilitated by a vascular access such as an arteriovenous fistula or graft. Additionally, clinicians should employ suitable dialysis needle sizes (15- to 16-gauge) to ensure optimal flow dynamics. While smaller (17-gauge) needles or central venous catheters do not exclude the use of online HDF, they may limit the attainable blood flow and thus complicate achieving desired convection targets.

The filtration fraction (FF), defined as the ratio of the ultrafiltration rate to blood flow, represents the fraction of plasma water removed during online HDF. Adjusting FF offers another means of enhancing convection volume. Typically, FF is set at 25%, but increasing FF from 25% to 30% at a blood flow rate of 400 mL/min yields an additional 1.2 L of convection volume per hour. An FF of 30% or higher is often possible in many patients, and therefore worth considering.

Session duration is another important determinant of convection volume. Clinical experience shows that extending treatment duration by 30 minutes generally increases the convection volume by approximately 2.5 L, while an additional 60 minutes may add approximately 6 L, depending on other parameters such as QB and FF. Although extending dialysis sessions can significantly enhance convective volume and clinical benefit, feasibility is sometimes limited by patient preference.

Online HDF requires the use of high-flux dialyzers, available in various sizes and membrane configurations. Selection of larger dialyzers (e.g., dialyzer surface area ≥2.0 m²) is preferred to maximize convection at relatively lower transmembrane pressures, thus reducing hemoconcentration and clotting risks. To what extent other dialyzer characteristics affect the efficacy of an HDF treatment session, is poorly defined. Real-world experience from millions of online HDF treatments indicates that high convection volumes (>23.9 L) are achievable in most patients when clinicians carefully optimize these parameters. A successful prescription typically requires integration of all above mentioned variables.

In summary, prescribing high-volume online HDF effectively involves a targeted and individualized approach, leveraging adjustments in blood flow, FF, session length, and dialyzer characteristics. Through careful attention to these practical principles, nephrologists can reliably deliver the clinically beneficial convection volumes based on recent evidence. The above considerations are based on the results in European patients while on post-dilution HDF [9–12]. An Asian population is generally characterized by a smaller body size. We have proposed a model on how to adjust for body size [10]. When pre-dilution is prescribed, then the rule of thumb is to aim at a convection volume twice that of post-dilution.

When aiming for an upgrade of current standard high-flux HD, one may consider medium cutoff dialyzers and HDF [13]. These dialyzers are built upon high-flux dialyzer technology, featuring larger pores to enhance middle-molecule clearance (up to 45 kDa) while minimizing albumin loss. In addition, the medium cutoff fibers have a narrower intraluminal diameter or prolonged length, facilitating convective clearance through enhanced internal-filtration backfiltration. This convective clearance is uncontrollable, unmeasurable, and unpredictable in clinical practice, while being considered a key contributor to solute clearance. In high-flux HD, convective volumes are estimated to be several liters per treatment. A medium cutoff dialyzer delivers approximately 7.1 L to 12.7 L convective volumes during 4-hour experimental sessions, depending on filter size and blood flow rates. The medium cutoff dialyzers offer a combination of “higher” high-flux HD and “very low dose” HDF. So far, clinical studies offer limited argumentation for a switch from high-flux to medium cutoff dialyzers. Further, it is completely unknown how prescription variables can be used to dose the treatment [13].

As indicated above, the relation between dose (i.e., convection volume) and outcome is established in trials in the European population and confirmed in large real-world data sets, also including the Asian population. It would be good to learn more about this relation in an Asian population, which is typically characterized by a smaller body size. It is very well possible that the dose and response curve, as we have presented recently [2], is slightly shifted to the left in an Asian population. Thus, that similar benefit can be obtained when achieving a bit lower convection volumes. Also, more detailed information on how to achieve the convection volume target, when taking above mentioned prescription variables into consideration, could be helpful. These and other questions could be addressed in Korea when implementing HDF.

As I have learned recently, presently around 25% of Korean dialysis patients are treated with HDF. The level of evidence that is available today justifies wider implementation of online HDF [14]. The choice of high-dose HDF is supported by solid evidence of superiority on clinical endpoints based on both trial and real-world data. Further, trial data show beneficial effects on relevant domains of QoL. Prescription variables of HDF are well defined, easy to handle, and clearly related to the outcome. It has been argued that further trials are not necessary anymore, because it is highly unlikely that the overall conclusion of superiority of HDF on all-cause mortality as compared to standard HD will change as a result of new studies [15,16]. So in fact, HDF offers a change in the basic principles of dialysis therapy, which for the first time in decades has turned out to be clinically meaningful for our patients (Table 1).