Dialysis in patients with TBI is made more complicated by their propensity to develop pulmonary oedema, and by their sensitivity to fluctuations of blood pressure, all of which can exacerbate secondary brain injury. The normal pre-packaged dialysate fluids have the wrong electrolytes for a TBI patient, and the normal dialysis prescription (designed to maximise the efficiency of the circuit) can lead to unacceptable episodes of ICP elevation. Apart from that, circuit anticoagulation and vas cath position can play a role in promoting new complications. Having said that, it is usually impossible to completely avoid dialysis in these people. Both chronic renal failure and acute renal failure are associated with a worse outcome in TBI. Moore et al (2010) found that among patients with TBI, the mortality for those who also developed renal failure went up from 18% to 42%.
This was the topic of Question 21 from the first paper of 2018, the first time this has come up in the CICM Part II exam. The college asked for the sort of issues that would affect renal replacement therapy in a young patient with severe traumatic brain injury and raised intracranial pressure. The model answer was surprisingly pragmatic, but somewhat unstructured. Fortunately, there is some well-structured literature to support this answer. Andrew Davenport published an excellent review article in 2007. He also published another excellent review article in 2008 to describe a practical approach to the issues involved. Both reviews are updates of an article he published in 2001. A slightly more recent experiment by Yeh et al (2016) focuses on the prevention of ICP fluctuations during RRT by modifying the IHD protocol. Recommendations from these studies can be summarised as follows:
Recommendations for Renal replacement Therapy in Patients with traumatic Brain Injury Domain Recommendations Rationale Access Avoid internal jugular lines Promote venous drainage from the brain Modality Prefer CRRT Produces a more gradual solute clearance;
less likely to produce cerebral oedema
Low efficiency IHD/SLED Frequency Daily, if not continuous Daily treatments decrease the fluctuations of urea Blood flow Start low, increase slowly Minimise haemodynamic effects Dialysate flow Start low, increase slowly Minimise solute clearance Dose Under-dialyse (by half) Minimise solute clearance per unit time Solute clearance Pre-dilution haemofiltration Minimise urea clearance: decrease the resulting urea gradient between brain parenchyma and blood, minimising cerebral oedema Filtration Low volume fluid removal Minimise dialysis-associated hypotension to prevent cerebral hypoperfusion Anticoagulation Regional, or none Prevent cerebral haemorrhage extension due to anticoagulation. Minimal anticoagulation is recommended for 2 weeks following TBI. Dialysate Add sodium Minimise the hyponatremia which develops due to exposure to hyponatremic dialysate (to keep sodium around 145-150 mmol/L) Add urea Minimise urea clearance Minimise bicarbonate Prevent intracellular acidosis (may be hypthetical) Fluid warmer Temperature matching Maintain therapeutic hypothermia if this is being used for ICP control
Before deciding how to dialyse them, most reasonable people will first deal with the question, "do I need to?". Realistically, what is the imperative to dialyse these people? You might get away without it. For instance, Masud et al (2015) reported on a group of 11 patients all of whom had coexisting acute kidney injury and severe TBI, of whom none required dialysis. So, what might compel you to use RRT?
They may have end-stage renal failure. There is a growing population of patients in the community who are chronically dialysis dependent. This group is just as likely to crash motorcycles, fall over and get into fights as the rest of the population. Ergo, one should expect the population of TBI patients to contain a proportionally similar group of ESRF patients who will need to undergo routine maintenance haemodialysis while recovering from their traumatic brain injury.
Metabolic acidosis increases cerebral blood flow. Britton et al (1979) demonstrated this by infusing isocapneic fluid into the cerebral arteries; acidic fluids resulted in increased regional blood flow irrespective of CO2 content. Increased cerebral blood flow might sound like a positive thing, but in fact because blood contributes to the intravascular volume this may produce an increase in the ICP. Taha et al (2015) reviewed the outcomes of TBI patients with abnormal blood biochemistry and found that metabolic acidosis was independently associated with poor outcomes. Similarly, Rahimi et al (2014) found that paediatric TBI patients had increased risk of mortality if metabolic acidosis was present. In summary, acidosis = bad. This may be the dominant reason for starting dialysis.
Uraemia impairs platelet function. This should intuitively produce the immediate desire to remove uraemic toxins from the body of the trauma patient. Though not specific to TBI, data from acute intracerebral haemorrhage patients (Kim et al, 2013) suggests that one of the most common complications in these people is the recurrence of the haemorrhage.
Uraemia increases blood-brain barrier permeability. Liu et al (2008) injured the kidneys of several mice and then observed that Evans Blue dye tends to extravasate into their brains more effectively, demonstrating this concept. Outside of the context of brain injury, uraemia is a fairly well recognised cause of impaired cognitive function. One does not need this in traumatic brain injury, where neurological progress informs prognosis.
In summary, there are several reasons as to why renal dysfunction might eventually require renal replacement therapy in TBI, even though the college answer to Question 21 from the first paper of 2018 cautions the candidates to have a "higher threshold for commencement in the context of raised ICP".
The main reason to consider this question is the effect of RRT on fluid shifts in the brain parenchyma, i.e on cerebral oedema. Specifically, large movements of solutes are the thing that needs to be avoided. Urea exchanges slowly with the blood compartment: the movement of water in and out of the brain tissue is approximately 20 times faster (Trinh-Trang et al, 2005). Thus, when the circuit removes a vast amount of urea rapidly, the difference between serum urea and brain tissue urea equilibrates comparatively slowly. The difference gives rise to an osmotic gradient which attracts water into the brain. This post-dialysis cerebral oedema is generally known as "dialysis dysequilibrium syndrome" in patients who have not had a head injury.
As the dialysis circuit removes fluid and urea, the extracellular compartment becomes short on both substances. Doubly so when it comes to fluid, as one one hand it is disappearing into the effluent, and on the other being taken up by swelling cells due to the osmotic shifts described above. In short the fluid is disappearing out of the intravascular space. The effect of this is decreased MAP and decreased cardiac output. This has predicable effects on cerebral perfusion. At this stage nobody has demonstrated any increased brain tissue lactate or increased cerebral oxygen extraction ratio associated with this, but the fact that the MAP and CPP are affected is well-documented.
Davenport's 2008 article contains an excellent graph (reproduced here with no permission whatsoever) which illustrates this effect in what seems like real recording taken from a TBI patient undergoing intermittent haemodialysis. Unfortunately Davenport's article does not describe the experimental setup and there is no reference associated with the graph, but it is no less effective as an illustration of what usually happens. Approximately one or two hours into the dialysis session (when the rate of urea removal is at its peak) the ICP starts going up. From the graph it would appears that ICP peaks of nearly 60 mmHg are to be expected, which seems a little extreme.
The ICP changes during RRT are clearly the consequence of excess urea removal, and so the choice of modality therefore depends on picking something which might achieve a less efficient urea clearance rate. In essence, the trick is to underdialyse them; and there are obviously several possible solutions to this.
Use CRRT. A continuous low efficiency circuit will produce a more controlled and gradual rate of fluid and solute removal. One may run the circuit all day long and still ultimately achieve the same goal of adequate toxin clearance as IHD may try to accomplish over a 4-hour run.
Use SLED. Yeh et al (2016) did essentially this. They compared two groups of patients with TBI, where one group received conventional 200ml/min IHD over 4 hrs removing 5% of their body weight (3.5L net negative) every second day, and the other 100ml/min flow rates only aiming to remove 2.5% body weight (1.75L), but daily. The total "dose" of dialysis over a week would have been roughly the same, The low efficiency group had better ICP control during dialysis - they peak at 24 mmHg on average, whereas the high-efficiency group had ICPs around 36 mmHg on average. The ICP in both groups peaked around the second hour of dialysis, to give you an idea of when to expect this complication.
Use pre-dilution haemofiltration because it is less efficient at clearing urea; the college mentions this in their answer to Question 21 from the first paper of 2018. It seems weird to choose a dialysis strategy because of one of its known deficiencies, but in this context it has positive effect.
If using intermittent dialysis, do it daily. Davenport (2007) recommends this because the urea will not have risen very far between dialysis sessions. With a lower serum (and therefore brain) urea, the gradient will never be so high as to generate much oedema.
Add sodium. Normal dialysate and replacement fluid will usually have a normal sodium concentration, but in TBI you ted to aim for higher levels. Vinsonneau et al (2006) were able to dialyse patients with an additional 10mmol/L of sodium in the dialysate without any scary haemodynamic effects.
Add urea. To add urea to the dialysate results in a decreased diffusion gradient for urea, and therefore a decreased rate of urea removal. This is a card you can play if you want to dialyse the patient reasonably vigorously to remove some other dialysable toxin (eg. lithium) but the patient also has TBI and you do not want to risk cerebral oedema. Adding urea to the dialysate should put you in a position where the dose of dialysis can be safely increased without major effects on the ICP.
Reduce bicarbonate. There is a widespread perception that exchanging a large amount of bicarbonate into the patient will give rise to an intracellular acidosis, as CO2 migrates across cell membranes and dissolves there, affecting intracellular pH. It is not clear whether this is a real in vivo effect or an artifact of buffering cells in a culture. Furthermore, the argument that hypothetical intracellular acidosis will do damage in traumatic brain injury is countered by the argument that the real damage from uncontrolled extracellular acidosis will probably be a bigger problem.
People who write recommendations about bleeding risk in chronic renal failure patients (eg. Janssen et al, 1996) suggest that dialysis modalities using systemic anticoagulation (including circuit heparin) should be avoided in the two weeks following major head injury or neurosurgery. Regional anticoagulation techniques such as citrate are favoured.