This subject matter occasionally rises from the depths of the examiners' subconscious to menace the advanced trainees. Though it has never appeared in the Part II exam in any meaningful way, anecdotal recollections of survivors seem to suggest that it occasionally occurs in viva stations and confronting bedside tutorials. Fortunately, the prevalence of this in the actual arena of clinical practice has decreased so much with modern methods of dialysis that authors have been referring to it as a "vanishing complication".
At a basic level this is a condition which results from the cerebral oedema that occurs when an ineffective osmole is enthusiastically extracted from the extracellular fluid by dialysis. In this, it resembles the cerebral oedema that can develop with the rapid overcorrection of glucose in a patient with a hyperosmolar hyperglycaemic state. Its one of those rare disease states that, when first recognised, was immediately and fully understood by the authors of the seminal article, with little else added in the subsequent years.
That seminal paper was Kennedy et al (1962), and is really just a letter to the editor of the Lancet. This group of clinicians noted that among renal failure patients with uraemia, a group failed to improve following dialysis - instead developing worsening confusion even while their biochemistry looked better. They made the connection between this phenomenon and the urea concentration, taking a hint from contemporary neurosurgical practices (where at the time it was routine to give doses of urea for cerebral oedema, in the same way as in the modern era one might give mannitol). Coming to the conclusion that some sort of a reverse version of this effect must have been operating in their patients, they followed this up with a series of experiments, demonstrating the lag between blood urea and CSF urea concentrations. In this manner, Kennedy et al were able to demonstrate that a concentration gradient is established during and after dialysis.
This was developed further by Pappius et al (1967) whose experimental animals ended up with a 12 mmol/L urea gradient across their blood-brain barrier while undergoing outpatient-style haemodialysis. This gradient develops because, though urea is an ineffective osmole, it's still effective enough over short timeframes. For the blood-brain barrier, the reflection coefficient (σ) of urea is about 0.44–0.59, where a σ value of 0.0 would be expected for a perfectly ineffective osmole and 1.0 for a perfectly effective one (for example, the σ for mannitol is 0.9). Without going into too much detail, this abstract ratio is a measure that describes the difference between the measured osmotic pressure across a membrane, and the pressure that would be expected from the van 't Hoff equation (which assumes the membrane is perfectly impermeable). The upshot of this measured value being not 1.0 is that urea can exert some osmotic pressure across the blood-brain barrier, which would give rise to a movement of water, and result in oedema. Moreover, during a long period of extracellular uraemia, idiogenic osmoles may also be generated by the cells of the central nervous system as a compensatory move.
So: from the above, two questions arise:
During a routine outpatient haemodialysis session, the dialytic and convective clearance of solutes results in an extracellular osmolality change of something like 20-30 mOsm/kg (Hagstam et al, 1969). Of these solutes, the vast majority must surely be urea. The other main solutes which contribute to osmolality are sodium (which remains stable because the dialysate contains a physiologically normal amount of sodium) and glucose (which would be expected to remain stable because the intensivist and the pancreas are both rather invested in maintaining a survivable blood glucose concentration in the patient). Ergo, a urea drop of 20-30 mmol/L over 3-4 hours is probably going to be well tolerated by otherwise well haemodialysis outpatients, which would correspond to a maximum drop of around 10 mmol/L/hr.
Even with this clinically irrelevant and unnoticed osmolality change, the brain parenchyma does tend to take on a lot of water, as demonstrated through pre and post dialysis CTs by La Greca et al (1982) and through MRI by Chen et al (2007). Some of these patients were symptomatic (headaches and nausea) even with a modest fall in osmolality - around 23 mOsm/kg in the Chen paper, which reported on ESRD patients having their first dialysis session. Therefore, dialysis disequilibrium syndrome could still occur in vulnerable brains even with a supposedly "safe" rate of urea drop. Mistry (2019) quoted reports of proper coma-and-seizures DDS in patients with a urea drop ranging from 65% to a mere 28%.
All that nausea and vomiting reported in the studies mentioned above suggests that some kind of change in intracranial pressure might have occurred. As none of the dialysis disequilibrium syndrome patients ever get sick enough for long enough to warrant an intracranial pressure monitor, our evidence about this is collected opportunistically in studies of patients who have an ICP monitor for some serious intracranial reason and have some dialysis because they also incidentally have renal failure. Thus we have data on ICP changes with dialysis in patients with intracranial haemorrhage as in Krane, 1989, or traumatic brain injury as in Lund et al (2018). In the latter paper, the ICP increased by an average of about 5-10 mmHg, and was predictably highest in patients who were receiving rapid haemodialysis (occurring 75 minutes into the treatment).
So, does that mean we can give them hypertonic saline and mannitol?
In short, yes. And there is probably more data and experience for mannitol here, probably because this condition has existed for a very long time, as has mannitol, whereas hypertonic saline is a relatively recent development. Giving mannitol to dialysis patients dates back to Hagstam et al (1969) and Rodrigo et al (1977). There was definitely an effect on the clinical features of disequilibrium syndrome, particularly in the latter study. Mannitol is still recommended for the management of acute severe dialysis disequilibrium syndrome by the sort of nepherologists who write narrative reviews for Seminars in Dialysis (Patel et al, 2008).
Unsurprisingly, these are very similar to other clinical features of cerebral oedema and raised intracranial pressure:
Patel et al list these with references and percentage values, but most reasonable people would agree that knowing the exact percentages would not have any meaningful benefit. "Mild, transient, and self-limited" is how most authors describe the natural course of this disease in most patients.
From everything above, the following preventative steps logically follow: