Chapter 42: Renal Replacement Therapies

As kidney function declines, the products of cellular metabolism accumulate, leading to retention of both the nitrogenous products of metabolism, typically assessed by the measurement of urea and creatinine but also potassium accumulation, and the development of a metabolic acidosis. Other potential toxins also accumulate,¹ leading to the term azotemia rather than uremia. This patient has AKI, and as such the time course of the illness is short, and only a limited amount of toxins will have accumulated. However, patients with AKI may have to be started on renal replacement therapy (RRT) because of hyperkalemia refractory to standard medical therapy or increasing pulmonary edema compromising oxygenation, and brain hypoxia (Table 42-1).

Historic data suggest that “early” initiation of RRT in AKI is associated with improved survival,^{2, 3, 4, 5, 6} but the evidence base is not sufficiently robust to allow a specific recommendation and the decision to initiate RRT should remain a clinical one. Whereas the decision to initiate RRT is straightforward in those patients with refractory hyperkalemia, severe metabolic acidosis and volume overload, and/or overt azotemic symptoms, in the absence of these overt manifestations, there is debate as to the optimal time to initiate renal support (see Table 42-1). Early introduction of RRT as soon as a patient enters AKIN (Acute Kidney Injury Network) stage 3 (see Chapter 41) may be of benefit, so that the patient is not exposed to the potential deleterious effects of metabolic abnormalities and/or volume overload. However, early initiation of RRT will result in some patients suffering the adverse consequences of treatment, such as venous thrombosis and bacteremia secondary to vascular access catheters, hemorrhage from anticoagulants, and other treatment-related complications, in particular hypotension, which may exacerbate cerebral ischemic injury.

Initial reports, some dating back 50 years, suggested a clinical benefit of early initiation of RRT.^{2, 3, 4, 5, 6, 7} These and other studies formed the basis for standard clinical practice that dialytic support should be instituted when the serum urea reached 80 mg/dL.^{2, 3, 4, 5, 6, 7} However, not all studies have shown a survival advantage with early initiation.⁸

Although the current consensus from retrospective and observational studies suggests that “early” initiation of RRT in AKI may be associated with improved patient survival, this has not been proven.⁸ In patients with acute traumatic brain injury (TBI), RRT not only adds complexity to patient management but also may cause additional complications in terms of changes in ICP and mean arterial blood pressure.⁹ Thus, in cases of AKI developing in patients with TBI, early initiation of RRT should only be undertaken in cases of refractory hyperkalemia or pulmonary edema; as in the majority of cases, AKI is secondary to a prerenal insult, which may be volume responsive and recover, thus avoiding the necessity of RRT. However, additional toxic insults, such as aminoglycoside antibiotics and nonsteroidal anti-inflammatory drugs (NSAIDs), should either be avoided or minimized, and every effort should be made to reduce the risk of radiocontrast-induced AKI,¹⁰ including minimizing the number of radiocontrast studies.¹¹

If kidney function continues to deteriorate, then azotemic toxins accumulate over days, and these will be retained, leading to increased intracellular osmolality. If plasma osmolality is rapidly decreased during dialysis, then the osmolality gradient between plasma and the brain may lead to a movement of water into the brain, and so potentially increase brain edema.

Although there are no randomized clinical trials, it is probably advisable to initiate RRT before the serum urea nitrogen exceeds 40 to 50 mg/dL in cases of AKI, to limit the amount of cerebral edema caused by the fall in serum urea during RRT.¹²

Patients with chronic kidney disease established on regular dialysis are susceptible to brain injury. These patients are at increased risk of subdural hemorrhage due to the prescription of coumarin anticoagulants, intracerebral hemorrhage secondary to hypertension, and cerebral abscesses secondary to central venous dialysis access catheter-associated infections. These patients have chronically high levels of urea and other azotemic toxins, and as such, delaying dialysis treatments will lead to even higher chemistries, and thereby increase the risk of RRT-induced cerebral edema. So for this group of patients, RRT should be considered once the patient has been stabilized.¹³

Choice of Renal Replacement Therapy

The key strategy of any RRT is to remove small water-soluble solutes. Urea clearance from the plasma water is faster during hemodialysis treatments than urea movement from cells into the plasma water. As such, this sets up a urea gradient. In addition, water transport across cell membranes through aquaporin channels is some 20 times faster than the corresponding transcellular passage of urea through urea transporters.¹⁴ This can result in an osmotic gradient developing between the plasma and the brain (Figure 42-1). Renal replacement therapy, particularly intermittent hemodialysis (HD), can lead to cerebral edema not only in acute experimental animal models of HD¹⁵ but also in the outpatient setting of dialyzing healthy patients with end-stage kidney failure.¹⁶ This swelling of the brain associated with HD is termed the dialysis disequilibrium syndrome.

Although RRT is primarily designed to remove small water-soluble toxins that accumulate in kidney failure, it is also used to correct the metabolic acidosis associated with kidney failure, and therefore dialysates either contain bicarbonate or lactate in supraphysiologic concentrations. The rapid correction or increase in plasma bicarbonate, rather than correcting intracellular acidosis, may paradoxically worsen intracellular acidosis,¹⁷ as bicarbonate being charged does not readily cross lipid-rich cell membranes, unlike carbon dioxide (Figure 42-2). Intracellular acidosis increases cell swelling and will exacerbate any underlying cerebral edema. These changes in plasma bicarbonate compared to cerebrospinal fluid bicarbonate typically lead to changes in the respiratory center in the medulla, with alteration in breathing patterns in healthy outpatient dialysis subjects.¹⁸ Thus, too rapid a correction of plasma bicarbonate in nonventilated brain-injured patients may potentially adversely affect cerebral oxygenation.

In addition, the connection of a patient to an extracorporeal circuit can lead to hypotension, which can potentially compromise cerebral perfusion. The passage of blood across the dialyzer, particularly those with a very negative surface charge (zeta potential), can lead to the generation of potent vasodilators, bradykinin, nitric oxide, and the anaphylotoxins C3a and C5a,¹⁹ and it is now recognized that both cardiac and cerebral perfusion can fall within the first few minutes of initiating standard outpatient hemodialysis.^{20, 21} This effect on cardiac perfusion may be exacerbated in cases of subarachnoid hemorrhage or severe cerebral edema, owing to the associated neurogenic cardiac stunning. Bradykinin and complement activation by the extracorporeal circuit can be reduced by rinsing the circuit with isotonic bicarbonate rather than saline, and some centers advocate priming the dialyzer with albumin to coat the membrane to reduce membrane reactions. However, priming the circuit with blood, particularly aged blood, may exacerbate any reactions owing to the acidic nature of stored blood.²² Similarly, angiotensin-converting enzyme inhibitors prevent bradykinin degradation, and as such should be avoided.

Several different modes of RRT are potentially available.

Peritoneal dialysis

Peritoneal dialysis (PD) is a form of continuous renal replacement therapy (CRRT), and as such the fall in serum urea and osmolality during treatment are slower than that compared to intermittent HD,²³ but even so the dialysis disequilibrium syndrome has still been reported during PD.²⁴ Peritoneal dialysate fluids are hyponatremic relative to blood, with a sodium concentration of 132 mEq/L, but are hypertonic owing to a high glucose content, varying from 13.6 to 38.6 g/dL, and so may cause patients to become hyponatremic. Large intraperitoneal volume exchanges of hypertonic glucose solutions can adversely impact cerebral perfusion and cerebral perfusion pressure by reducing right atrial filling and cardiac output.²⁵ To minimize these potential changes, peritoneal dialysis prescriptions should use the lowest glucose concentrations possible and avoid major swings in intraperitoneal volumes. This can best be achieved by using peritoneal dialysis cycler machines with a tidal exchange.

Intermittent hemodialysis/hybrid therapies

Brain edema occurs during routine outpatient thrice-weekly HD.¹⁶ Thus, intermittent HD increases brain swelling in the patient with acute neurologic injury.²⁶ ICP increases not only because of changes in osmotic gradients and a rapid increase in arterial pH but also because of abrupt falls in mean arterial pressure, and thus cerebral perfusion pressure.²⁷ Intermittent RRT should therefore only be considered as a treatment option in cardiovascularly stable patients without changes of increased ICP or a midline shift on cerebral imaging (See Figure 42-1). Small solute removal is more efficient by diffusion down a concentration gradient rather than convection, where there is removal of plasma water and the solutes contained within it. Thus, urea removal is faster with hemodialysis than hemofiltration (Figure 42-3). Hence, hemofiltration or hemodiafiltration are preferable treatments, and the rate of urea and other small solute clearances can be further slowed by infusing the replacement fluids prefilter/-dialyzer rather than postfilter/-dialyzer (Figure 42-4).

If intermittent hemodialysis is the only renal replacement modality available, then the prescription should be modified to minimize cardiovascular instability, using a high sodium and calcium dialysate concentration, cooling the dialysate to 35°C, and minimizing changes in effective blood volume.²⁸ In addition, the rate of change in serum osmolality should be minimized by utilizing slower blood pump and dialysate flows, with smaller surface area dialyzer membranes, and lower dialysate bicarbonate concentrations,²⁹ thus moving from a standard intermittent hemodialysis prescription to one of slow extended dialysis, or a hybrid therapy. Whereas the standard practice for outpatient hemodialysis is thrice weekly, patients with acute brain injury should be treated daily, to minimize the peaks and troughs in blood urea, thus reducing the time-averaged urea concentration. Hypotension during intermittent hemodialysis is typically due to a high ultrafiltration rate leading to plasma volume depletion by being faster than compensatory refilling from extracellular fluid.

There are no randomized prospective trials that have investigated the optimum predialysis urea to minimize changes in ICP during dialysis; however, clinical practice suggests that a predialysis blood urea nitrogen less than 30 to 35 mg/dL reduces the risk of ICP increasing during treatment.²⁹

Continuous dialysis/hemofiltration therapies

In patients with increased ICP, continuous forms of arteriovenous hemofiltration and/or dialysis have been shown to cause fewer changes in ICP and central perfusion pressure (CPP) than intermittent therapies^{30, 31} because of a combination of slower changes in plasma osmolality and greater cardiovascular stability (See Figure 42-2).¹² However, with the introduction of pumped venous hemofiltration, dialysis and hemodiafiltration, and in particular high-volume exchange, greater changes in osmolality have become possible (Figure 42-5). Thus, when performing CRRT in critically ill patients with AKI, hemofiltration is preferable to dialysis, as this leads to a slower rate of change in serum urea and other small solutes and also greater cardiovascular stability owing to additional cooling, particularly with predilutional fluid replacement³² (See Figure 42-4). Sodium balance during hemofiltration tends to be positive, as the amount of sodium in the ultrafiltrate is typically less than that in the plasma. The ratio of ultrafiltrate to plasma is termed the sieving coefficient, and for sodium the sieving coefficient is less than 1.0. Even so, a replacement fluid with a sodium concentration > 140 mmol/L should initially be used to prevent hyponatremia.³³ At the start of treatment, in critically ill patients with increased ICP, small volume exchanges should initially be used, and only when the patient has been shown to be stable should the exchange volume be increased.

Renal replacement therapy can be used to clear several drugs and/or toxins in cases of coma due to drug self-poisoning, toxicity, and some metabolic encephalopathies (eg, methylmalonic acidemia), particularly if the toxin is water-soluble and has a limited volume of distribution. Intermittent hemodialysis will more quickly remove such drugs and/or toxins compared to CRRT, provided the patient has adequate cardiovascular stability, but then may be followed by a rebound in plasma concentrations, especially if the toxin has a large volume of distribution. In such cases CRRT following a hemodialysis session may be helpful in preventing rebound. Slow extended hemodialysis or hybrid therapy would also be a therapeutic option.

Recently isovolemic CRRT has been advocated for treating patients without AKI who have been resuscitated after cardiac arrest and not regained consciousness.³⁴ There have been initial encouraging reports of increased survival for those patients treated for 8 hours with high-volume hemofiltration.³⁵

In addition to inducing osmotic changes, the rapid diffusion during hemodialysis of supraphysiologic bicarbonate concentrations rapidly corrects plasma pH. Bicarbonate is charged, and cannot readily cross the blood-brain barrier (BBB). However, bicarbonate forms H₂CO₃ by reacting with hydrogen ions, which can dissociate into water and carbon dioxide, and carbon dioxide can readily cross the BBB. These fluxes of HCO₃^–, H₂CO₃, and carbon dioxide lead to changes in pH according to the Henderson- Hasselbalch equation, resulting in what is termed a paradoxical intracellular acidosis.³⁶ This then leads to a compensatory production of so-called intracellular osmoles and water movement into the brain, along a concentration gradient (see Figure 42-2).¹⁸ Although lactate- or acetate-based dialysates and substitution fluids result in a slower increase in plasma bicarbonate, these fluids lead to an increase in lactate and/or acetate, with greater risk of hypotension and inotrope requirement. As such lower bicarbonate concentrations, 30-32 mEq/L, are recommended rather than standard bicarbonate of 35-40 mEq/L.

Intradialytic hypotension, which is relatively common during intermittent hemodialysis,³⁷ causes a reduction in CPP, with a consequent increase in ICP, owing to compensatory cerebral vasodilatation.¹¹ Cardiovascular stability during hemodialysis is greater when higher dialysate sodium concentrations are used, up to a maximum of 10 mmol/L greater than plasma sodium.³⁸ In addition, dialysate calcium concentrations of 2.7 to 3.0 mEq/L lead to fewer episodes of intradialytic hypotension than dialysate calcium concentrations of 2.5 mEq/L or less.

Cooling the dialysate improves cardiovascular stability during hemodialysis. Increased cooling can be achieved with predilutional fluid removal and during CRRT by not warming dialysates and/or replacement fluids.

As blood passes through an extracorporeal circuit, neutrophil and monocyte activation leads to release of blebs of surface membrane, which are a potent source of tissue factor, which leads to the initiation of thrombin generation, platelet activation, and activation of the coagulation cascades. As such, the majority of patients require some form of anticoagulant during RRT to prevent clotting of the extracorporeal circuit. (One advantage of peritoneal dialysis is that no anticoagulant is required.) Anticoagulants may be either regional or systemic. During regional anticoagulation, only the extracorporeal circuit is anticoagulated, whereas with systemic anticoagulants, the patient is anticoagulated.

There is a risk of hemorrhage around ICP monitors, particularly with the more invasive intraventricular catheters and to a lesser extent with subdural devices.³⁹ In these cases, CRRT should preferably be anticoagulant free or a regional anticoagulant used.¹²

Anticoagulant-Free Extracorporeal Circuits

Clotting in the extracorporeal circuit often starts in areas of blood-air contact. Thus, for anticoagulant-free circuits, careful preparation of the circuit is required with complete flushing of all air, in particular from the dialyzer, and ensuring that all joints are airtight. Predilutional fluid replacement during hemofiltration and/or hemodiafiltration reduces the hemoconcentration along the dialyzer and dilutes both platelets and clotting factors, thus reducing the risk of circuit clotting compared to postdilutional fluid replacement (See Figure 42-4). The same effect of predilutional fluid replacement can be achieved during intermittent hemodialysis by the administration of regular isotonic saline boluses.

Unfractionated heparin is a highly negatively charged molecule and as such can adsorb to the dialyzer surface. Thus, several centers rinse the dialysis circuit with around 20,000 IU heparin in 1 L of isotonic saline for 60 minutes and then flush out the heparinized solution and dialyze patients without any additional anticoagulant. Hospal (Hospal AN69ST, Lyon, France) has altered the surface of a dialyzer in a deliberate attempt to increase heparin adsorption to allow priming with heparin but then dialysis without additional heparin.⁴⁰

Regional Anticoagulants

Citrate, nafamostat, and prostacyclin all have very short half-lives and can be used as regional anticoagulants. The risk of bleeding is markedly reduced, as these agents provide anticoagulation for the extracorporeal circuit but do not systemically anticoagulate the patient. Nafamostat maleate, a serine protease inhibitor, is extensively used in Japan. Typically a loading dose of 20 to 40 mg is followed by an infusion of 20 to 40 mg/h, titrated to achieve an activated partial thromboplastin time ratio of around 1.5 to 2.0 times normal. Prostacyclin is a potent pulmonary vasodilator and is mainly limited to European practice. There is no simple monitoring test, and most patients are empirically started on an infusion of 2.5 to 5 ng/kg per minute. As it is such a potent vasodilator, hypovolemia should be corrected before starting treatment, and some centers increase vasopressor support as prostacyclin is started. In some cases prostacyclin has been reported to reduce CPP and increase ICP.⁴¹

Citrate is gaining popularity, particularly as an anticoagulant for CRRT, especially now that commercially available dialysates and replacement solutions have been developed specifically for citrate CRRT. Although citrate may occasionally accumulate—typically in cases of acute severe liver damage, cardiogenic shock, and extensive rhabdomyolysis—it is both a potent and an effective extracorporeal anticoagulant. The amount of citrate administered is adjusted according to the blood flow rate, and then as citrate chelates calcium, with loss of the calcium-citrate complex into the dialysate/ultrafiltrate effluent, calcium is then reinfused to maintain a normal systemic ionized calcium concentration.⁴²

Systemic Anticoagulants

Systemic anticoagulants include heparins and the direct thrombin inhibitors. Traditionally, unfractionated heparin has been the standard anticoagulant for extracorporeal therapies. The main risk with systemic anticoagulants is one of hemorrhage, which is greatest in the first 24 hours following surgery and then falls over the subsequent 48 hours, but thereafter there remains an increased risk of bleeding compared to no anticoagulant or regional anticoagulants.⁴³ To reduce the risk of bleeding with unfractionated heparin (UFH), lower doses are now used, although the data suggesting that lower doses reduce the risk of bleeding are limited. During CRRT many centers use a loading dose of 500 IU followed by 500 IU/h, aiming for a systemic activated thromboplastin time ratio of 1.5 to 2.0 times normal.⁴⁴ More recently low-molecular-weight heparins (LMWHs) have been introduced. These have longer half-lives and a single bolus can be used for intermittent treatments (enoxaparin 0.5 mg/kg or tinzaparin 1500 to 4500 IU depending on duration of session). LMWHs require specialized anti-Xa activity monitoring, particularly for CRRT, when infusions are required.⁴⁵

UFH is negatively charged and, particularly when contaminated with chondroitin sulfate, can increase bradykinin and C3a and C5a production as it passes through the dialyzer, risking hypotension.⁴⁶ In addition, patients may have allergic reactions to heparins, particularly if they have a porcine allergy. Occasionally patients may develop an autoimmune thrombocytopenia, termed heparin-induced thrombocytopenia (HIT), in which case patients are at increased risk of thrombosis. Management includes both withdrawal of all heparins and administration of an alternative systemic anticoagulant.

Direct thrombin inhibitors include lepirudin and argatroban and are typically used for cases of HIT. Lepirudin is an irreversible antagonist of thrombin, which has a very prolonged half-life in kidney failure, and as such is associated with a marked increased risk of bleeding. The plasma concentration of hirudin does not correlate linearly with the activated partial thromboplastin time, and thus monitoring of either plasma hirudin or the ecarin clotting time is recommended. Argatroban is a reversible thrombin antagonist and, as such, requires a bolus of 250 μg, followed by an infusion of around 2 μg/kg per minute titrated to achieve an activated partial thromboplastin ratio around 2.0.⁴⁷

When sustained surges in ICP occur during RRT, standard neurosurgical conservative management should first be instituted, checking that arterial oxygenation is adequate (Pao₂ of > 82.5 and Paco₂ of 49.5 to 55 mm Hg)⁴⁸ and the neck is not compressed.⁴⁹ Boluses of propofol and/or thiopentone can be administered if the CPP is high, whereas if the CPP is normal or low, then hypertonic saline and/or mannitol should be given. Deliberate ultrafiltration will also reduce CPP but may then result in a rebound in ICP.²⁷

Individual centers differ in the concentrations of hypertonic saline they use (ranging from 3% to 10%),⁵⁰ and these can be infused during CRRT and other renal replacement therapies,⁵¹ aiming for a serum sodium of 145 mmol/L, up to a maximum of 155 mmol/L, depending on the ICP response. If mannitol is infused during CRRT, then 100 mL of a 20% mannitol solution is effective when infused quickly over 10 to 15 minutes.¹²