Explaining Haemodiafiltration (HDF)
There are three principle ways by which any solute (= any substance dissolved in water) can be removed from solution:
The major mechanisms of dialysis use all three, but to varying extents, depending on the method.
Diffusive removal depends primarily on the creation of a concentration gradient between two solutions that are separated by a partially (or semi-) permeable membrane across which solutes can pass. Solutes will pass from the more concentrated to the less concentrated side of the membrane. This works best for the removal of small molecules (like urea, or potassium). The rate of diffusion will depend on:
- The design of the membrane
- How thick it is
- How porous it is
- How well it “sieves”…think of the differences between what could pass through fly wire when compared to chicken wire!
Since diffusion diminishes as the size of the solute molecules increases, diffusion begins to fail as target toxins get larger in size.
Convective removal is based on the rate water flow through a membrane and is driven by the pressure difference (known as the hydrostatic pressure) between the two sides of the membrane. A higher pressure in the blood compartment…or a lower pressure on the dialysate side…will force water to squeeze across the membrane from the higher pressure to the lower pressure side. The greater the unopposed pressure, the greater will be the water flow. Just think of one of those porous garden hoses!
In this situation, solutes are pulled across the membrane with water (this is known as solute drag), with the driving force now being a pressure gradient, not a concentration gradient. Depending on the porosity of the membrane (= governed by the size of the pores), larger molecules can be more effectively removed by convection than by diffusion. In convection, both solutes and water are more efficiently removed.
Adsorption principally affects plasma proteins and any solutes that might be bound to them. Plasma proteins (especially those of low molecular weight) stick to the membrane surface and are, to a greater or lesser degree, removed by membrane binding. High flux membranes adsorb protein-bound solutes better than low flux membranes. Until recently, adsorption has been an essentially uncontrollable factor and has thus been largely ignored. But, as membranes can now be nano-engineered, interest (and success) in the manipulation of protein-bound solutes is rapidly becoming a future growth area in membrane biophysics.
While all dialysis techniques mix and match these three processes, in practical terms, early haemodialysis was a mainly diffusive process using low-flux (low permeability) membranes. As time passed, and as membranes trended towards high flux (leakier), and our ability to ensure water quality and safety improved, dialysis slowly crept up the diffusion vs. convection slope towards a greater reliance on convective forces (see diagram below). While lots of things will affect the efficiency of the convective “transport” of a molecule across a membrane, the most important include:
- Water flux (rate and volume)
- Membrane pore size (big or little holes, and their respective ratios)
- The pressure difference (hydrostatic pressure) applied to and across the membrane
- Viscosity of the fluid within the membrane pores
- The size, shape and electrical charge of each molecule.
While haemodialysis is primarily diffusive, haemodiafiltration is increasingly convective in nature. And, the greater the convective force, the greater will be the generated volume of the pressure-driven ultrafiltrate.
So…running that thought again…the greater the convective force, the greater will be the rate of water loss across the membrane (the ultrafiltration rate). The extension of this thought is that if convection is uncontrolled, a great many litres of body water would be lost across the membrane. This would result in rapid circulatory collapse and death unless the dialytic process adopts one of two choices:
- Using conventional haemodialysis methods, the ultrafiltration volume is limited to only that which is must be removed to achieve target weight. This has long been the mainstay of dialysis-as-we-know-it.
- Using convective force, maximize ultrafiltration and solute drag but, at the same time, replace the convective-driven ultrafiltration volume (minus any desired ultrafiltration loss needed to attain target weight) with a physiologically optimum, sterile and pure ‘replacement fluid’. This replacement volume needs to be directly infused back into the plasma volume.
Thus, to avoid the need for fluid replacement, conventional dialysis has needed to limit the rate of ultrafiltration and larger solute removal, and even high flux dialysis has been unable to utilize the full benefits of increasing convection.
Herein lies the difference between haemodialysis and haemodiafiltration:
- In haemodialysis (HD), diffusion occurs down a purposefully engineered concentration gradient, and ultrafiltration volume is “controlled” by limiting the volume of water lost across the membrane. But, this also limits the additional solute drag that could be achieved if greater ultrafiltration could occur. This additional solute drag would be especially useful for removal of larger molecules (β2 microglobulin, homocysteine and small-behaves-like-big phosphate) but is design-limited in HD systems. As there is no mechanism in HD systems to replace excess UF volumes—other than by small volume resuscitative saline infusion—convection remains a minor contributor to solute removal in HD.
- In haemodiafiltration (HDF), diffusion still occurs down its purposefully engineered concentration gradient, but the ultrafiltration volumes that are lost are—at least to a large extent—uncontrolled (or at least less controlled), with large volumes of convective volume crossing the membrane. This high (or very high) volume of convective transport is driven by applying increasing trans-membrane pressure to drive water movement. The large volumes of ultrafiltrate thus created add enormously to solute drag—especially for the larger “middle molecule” solute classes. But, if not replaced (minus any calculated UF volume needed to achieve target weight), this profound fluid movement across the membrane would lead to equally profound volume depletion, rapid circulatory collapse and death. Replacement of this extra convective ultrafiltrate, milliliter for milliliter throughout the dialysis period…minus, of course, any intended ultrafiltration volume…is the key difference. And, as this replacement fluid must be given back directly to the patients’ circulation, the composition and purity of this directly injected replacement fluid is pivotal. The “re-infusate” must be absolutely sterile, free from endotoxin contamination and biochemically exact. Modern HDF systems ensure that these requirements are faithfully and reliably performed. Water quality can be assured by using a pair of in-series ultrafilters (e.g. 2 x Diasafe™ filters, or similar), with a protocol for filter replacement being strictly enforced, while the additional cost of water filtration must also be factored in to any cost/benefit analysis.
Replacement volumes of <20 litres define low volume HDF, while most high volume HDF is defined by replacement volumes of >20 litres. But, and here is an important explanatory message for NxStage users: this “replacement volume” is not the same “volume” as the 25-30 litres of dialysate that you are used to thinking about. The < or > than 20 litre HDF volume I describe here is a replacement fluid volume, and not the volume of dialysate.
The other key HDF issue—and there are arguments for and against each of the several options here—is exactly where that re-infusion process should take place:
- Pre-dialyser (the “pre-expansion” option)
- Post-dialyser (the “post-replacement” option)
- Or, in one of several ways within the dialyser itself.
There is little to be gained from exploring the technical advantages and disadvantages of each of these options here…suffice to say that most current chronic maintenance HDF systems use a within-dialyser technique.
The following schematic demonstrates the gradation as membrane transport moves from diffusion to convection (or vice versa). While low flux haemodialysis is seen to occupy the left side of the diagram, at the far right of this sliding scale lies haemofiltration, a process that, in the mid-late 1980’s, opened up a rapid expansion of ICU techniques to treat acute kidney injury. Just as haemodialysis has slid up the diffusion/convection slope, pure haemofiltration has been modified backwards down the “slope” as variants of HDF have emerged. But, whilst some combination of diffusion and convection appears to be ideal, it remains to be sorted exactly where “the sweet spot” lies…and it may prove to be different in differing clinical circumstances
HDF, in most if not all studies, has been reported to offer a smoother, less symptomatic treatment than HD. It has been shown to enhance recovery time, improve survival, and to be achievable at a line-ball cost comparison. While improved water quality requirements are paramount, this is, in truth, a good thing as it focuses attention on ensuring that optimal water filtration and preparation are achieved at all times.
In my view, facility-based HDF does trump HD…using the “trump” word in its’ intended and original sense…and, in line with many others in ANZ, Canada and Europe, we fully converted all our facility-based patients to HDF in 2014.
As regards home-based HDF, the cost/benefit effectiveness remains under evaluation, but the clinical risk/cost/benefit balance is less clear – particularly as home patients in ANZ already reap the benefits of the greater middle molecule clearances through long, slow, frequent HD, and the additional costs of ultrapure water provision at home are not insignificant.