**5. The future of citrate anticoagulation in hemodialysis**

The fundamental roadblocks to widespread implementation of regional citrate anticoagulation are fear of electrolyte or metabolic disturbances and the relative laboriousness of this mode of anticoagulation. These two domains are interconnected. What current citrate dialysis algorithms have in common is that they are empiric. There is some degree of individualization, but only on a relatively low level. As a consequence, while these algorithms may work for the average patient, or even a majority of patients, there will always be the concern that the characteristics of a particular patient situation are not captured adequately, leading to unexpected and possibly dangerous changes in electrolyte or acid-base parameters. And for this very reason, these algorithms will never help eliminate the intensive laboratory monitoring that, at least initially, is currently required for regional citrate anticoagulation.

Tailoring the citrate infusion rate to the blood flow rate alone is a crude oversimplification. Anticoagulation along the extracorporeal circuit depends on a myriad things, such as the hematocrit, the void volume fraction, the plasma water calcium concentration, the composition of the other ionic species in the multi-ionic milieu of the plasma, the ultrafiltration rate, the type, size and geometry of the dialyzer used and, consequently, its solute removal characteristics, the blood and dialysate flow rates, the concentration of the citrate infusion (high concentrations entail low infusion rates, which may cause mixing issues or discontinuous, pulsatile flow), the dialysate composition (e.g., in terms or calcium, magnesium and citrate concentration), the plasma protein concentration, the rates of citrate generation and metabolism, the systemic citrate levels, the degree of access recirculation, the patient's capacity to buffer changes in extracellular calcium concentration, and so on, to name but a few. Some of these have greater impact than others; some are easier to model than others. But if the kinetics of calcium and citrate are to be predicted (not on average, but

staff resources and, consequently, can make citrate dialysis more costly than standard heparin dialysis. The prolonged filter patency times seen with citrate anticoagulation, however, may also introduce cost savings compared to heparin dialysis in continuous dialysis therapies [20]. The administration of buffer base in the form of citrate can further lead to metabolic alkalosis [20-22]. Hypernatremia can occur secondary to the additional sodium load administered with the citrate infusion (e.g., in the form of trisodium citrate, which carries 3 moles of sodium for each mole of citrate) [5, 21]. With high citrate infusion rates and/or in patients with impaired liver function (liver failure, cirrhosis), systemic citrate accumulation may occur. Measurements of plasma citrate concentrations are not usually readily available in clinical laboratories, but citrate accumulation may be detected by looking for its effects on calcium levels: citrate accumulation traps calcium in the form of calcium-citrate complexes. The growing plasma pool of calcium-citrate complexes and the insufficient release of calcium from this pool via citrate metabolism lead to a drop in systemic iCa which is spotted in systemic iCa measurements and countered by an increase in the calcium substitution rate in order to restore systemic iCa to physiologic levels. Under such conditions, the amounts of free calcium, calcium-protein complexes and the increased amount of calcium-citrate complexes add up to an increased total calcium concentration. Therefore, citrate accumulation may be detected by an increased total calcium concentration or an increased ratio of total to ionized serum calcium concentration [23]. An increased

The fundamental roadblocks to widespread implementation of regional citrate anticoagulation are fear of electrolyte or metabolic disturbances and the relative laboriousness of this mode of anticoagulation. These two domains are interconnected. What current citrate dialysis algorithms have in common is that they are empiric. There is some degree of individualization, but only on a relatively low level. As a consequence, while these algorithms may work for the average patient, or even a majority of patients, there will always be the concern that the characteristics of a particular patient situation are not captured adequately, leading to unexpected and possibly dangerous changes in electrolyte or acid-base parameters. And for this very reason, these algorithms will never help eliminate the intensive laboratory monitoring that, at least initially, is currently required for

Tailoring the citrate infusion rate to the blood flow rate alone is a crude oversimplification. Anticoagulation along the extracorporeal circuit depends on a myriad things, such as the hematocrit, the void volume fraction, the plasma water calcium concentration, the composition of the other ionic species in the multi-ionic milieu of the plasma, the ultrafiltration rate, the type, size and geometry of the dialyzer used and, consequently, its solute removal characteristics, the blood and dialysate flow rates, the concentration of the citrate infusion (high concentrations entail low infusion rates, which may cause mixing issues or discontinuous, pulsatile flow), the dialysate composition (e.g., in terms or calcium, magnesium and citrate concentration), the plasma protein concentration, the rates of citrate generation and metabolism, the systemic citrate levels, the degree of access recirculation, the patient's capacity to buffer changes in extracellular calcium concentration, and so on, to name but a few. Some of these have greater impact than others; some are easier to model than others. But if the kinetics of calcium and citrate are to be predicted (not on average, but

anion gap may also point towards citrate accumulation [24].

regional citrate anticoagulation.

**5. The future of citrate anticoagulation in hemodialysis** 

for a particular patient) with any degree of reliability, then these factors must be taken into account. Needless to say, the interactions between all these factors cannot possibly be assessed (let alone integrated over an entire treatment and beyond) based on intuition or clinical experience. Computer-aided calcium and citrate kinetic modeling is the only way to simulate in detail the processes during regional citrate anticoagulation. We have recently published a comprehensive, yet versatile, mathematical model for citrate dialysis [25]. A refinement of this model (comprised of our original model combined with a statistical correction component), recently presented as a talk at the XLVII ERA-EDTA conference in Munich, Germany, showed excellent prediction quality [26]. When applied to 120 patients on pure dialysate-side citrate dialysis (dialysate containing 2.4 mEq/L citrate and 2.25 mEq/L calcium), the model overestimated end-dialysis ionized calcium levels by only 0.026 mmol/L on average. While current clinical citrate dialysis algorithms are only applicable to a rather narrow setting for which they have been developed, computer-aided calcium and citrate kinetic modeling affords much greater flexibility and could possibly even be adapted on-the-fly to different conditions.

As was mentioned above, the calcium substitution in regional citrate anticoagulation is currently dosed empirically and adjusted so as to keep systemic iCa within the physiologic range. However, it must be born in mind that this approach pays no heed to the question of calcium mass balance. This is, of course, not done deliberately but simply from necessity, because clinicians have no way of assessing intradialytic calcium mass balance reliably, let alone under such complex conditions as occur in regional citrate anticoagulation. The difference between calcium substitution and calcium loss across the dialyzer membrane determines the intradialytic calcium mass balance, and from this perspective, the calcium substitution should be chosen so as to effect the desired mass balance. The challenges with determining what calcium mass balance is required for a given patient is a related but separate issue and shall not be discussed here. With higher citrate infusion rates, and accompanying citrate accumulation and calcium "trapped" systemically in the form of calcium-citrate complexes, calcium mass balances can easily become positive. In practice, this point is often dismissed and calcium substitution rates justified with reference to the need to maintain serum ionized calcium within the normal range. What becomes clear, however, when simulating citrate dialysis is that many roads lead to Rome, and, within limits, different calcium mass balances can be achieved without compromising the extracorporeal anticoagulation by modifying parameters such as dialysate calcium and citrate concentrations and blood and dialysate flow rates. Dialysis dose issues certainly have to be considered, and the combination of calcium and citrate kinetic modeling with urea kinetic modeling would be a particularly powerful tool. Conversely, the same calcium mass balance can be achieved in different ways, potentially allowing for individualization of the citrate dialysis prescription according to particular patient characteristics, such as impaired liver function or reduced calcium buffering capacity. In view of the ever-increasing awareness of the potential importance of calcium mass balance for long-term outcomes in hemodialysis patients, calcium and citrate kinetic modeling offers a unique opportunity for actively incorporating this parameter into the dialysis prescription. This may turn out to be crucial for translating the compelling short-term benefits associated with regional citrate anticoagulation into long-term improvements in cardiovascular outcomes and ultimately survival. Currently, this mode of anticoagulation is thoroughly ignoring this aspect and is lagging behind the trend towards neutral calcium mass balance seen in standard heparin hemodialysis. Similar to calcium mass balance considerations, dialysis-related sodium

Citrate Anticoagulation in Hemodialysis 225

[10] Pedersen KO: Protein-bound calcium in human serum. Quantitative examination of

[11] Toffaletti J, Gitelman HJ, Savory J: Separation and quantitation of serum constituents associated with calcium by gel filtration. Clin Chem 1976;22(12):1968-1972. [12] Bohler J, Schollmeyer P, Dressel B, Dobos G, Horl WH: Reduction of granulocyte

[13] Gritters M, Grooteman MP, Schoorl M, Bartels PC, Scheffer PG, Teerlink T, Schalkwijk

[14] Hofbauer R, Moser D, Frass M, Oberbauer R, Kaye AD, Wagner O, Kapiotis S, Druml

[15] Hetzel GR, Schmitz M, Wissing H, Ries W, Schott G, Heering PJ, Isgro F, Kribben A,

[16] Kutsogiannis DJ, Gibney RT, Stollery D, Gao J: Regional citrate versus systemic

[17] Monchi M, Berghmans D, Ledoux D, Canivet JL, Dubois B, Damas P: Citrate vs.

[19] Oudemans-van Straaten HM, Bosman RJ, Koopmans M, van der Voort PH, Wester JP,

[20] Morgera S, Scholle C, Voss G, Haase M, Vargas-Hein O, Krausch D, Melzer C, Rosseau

[21] Mehta RL, McDonald BR, Aguilar MM, Ward DM: Regional citrate anticoagulation for

[22] Silverstein FJ, Oster JR, Perez GO, Materson BJ, Lopez RA, Al-Reshaid K: Metabolic alkalosis induced by regional citrate hemodialysis. ASAIO Trans 1989;35(1):22-25.

venovenous hemofiltration. Crit Care Med 2009;37(2):545-552.

experience. Nephron Clin Pract 2004;97(4):c131-136.

prospective randomized study. Intensive Care Med 2004;30(2):260-265. [18] Bagshaw SM, Laupland KB, Boiteau PJ, Godinez-Luna T: Is regional citrate superior to

haemodialysis. Nephrol Dial Transplant 2006;21(1):153-159.

1972;30(3):321-329.

Nephrol 1996;7(2):234-241.

Kidney Int 1999;56(4):1578-1583.

Kidney Int 2005;67(6):2361-2367.

2011;26(1):232-239.

2005;20(2):155-161.

1990;38(5):976-981.

binding and its variables by a molecular binding model and clinical chemical implications for measurement of ionized calcium. Scand J Clin Lab Invest

activation during hemodialysis with regional citrate anticoagulation: dissociation of complement activation and neutropenia from neutrophil degranulation. J Am Soc

CG, Spreeuwenberg M, Nube MJ: Citrate anticoagulation abolishes degranulation of polymorphonuclear cells and platelets and reduces oxidative stress during

W: Effect of anticoagulation on blood membrane interactions during hemodialysis.

Himmele R, Grabensee B, Rump LC: Regional citrate versus systemic heparin for anticoagulation in critically ill patients on continuous venovenous haemofiltration: a prospective randomized multicentre trial. Nephrol Dial Transplant

heparin anticoagulation for continuous renal replacement in critically ill patients.

heparin for anticoagulation in continuous venovenous hemofiltration: a

systemic heparin anticoagulation for continuous renal replacement therapy? A prospective observational study in an adult regional critical care system. J Crit Care

van der Spoel JI, Dijksman LM, Zandstra DF: Citrate anticoagulation for continuous

S, Zuckermann-Becker H, Neumayer HH: Metabolic complications during regional citrate anticoagulation in continuous venovenous hemodialysis: single-center

continuous arteriovenous hemodialysis in critically ill patients. Kidney Int

loading is another topic that has been receiving more and more attention in recent years and is another domain of solute kinetic modeling that should ultimately be integrated into citrate dialysis modeling, particularly given the additional sodium load administered with the use of regional citrate anticoagulation.

The use of dialysate-side citrate anticoagulation (i.e., the use of a citrate- and calciumcontaining dialysate without arterial citrate infusion or venous calcium substitution) has sparked interest recently for its alleged heparin-sparing potential and its safety and ease of use [27-29]. At unchanged heparin doses, using citrate-containing dialysate (instead of bicarbonate dialysate acidified with acetate) appears to improve solute removal [30].

Citrate anticoagulation holds great promises for improving the outcomes of hemodialysis patients. Ultimately, kinetic modeling will be essential for taking this therapy to the next level (i.e., a high degree of individualization and increased safety through accurate prediction of electrolyte and acid-base kinetics) and to facilitate its widespread use in routine clinical practice.
