A swollen brain is a finding, not a diagnosis. Three different processes produce
it, they look identical on the monitor, and they take different treatments — so
the useful first move is not to reach for a drug but to name which one you are
looking at.
The three
Vasogenic. The barrier is physically disrupted, or its tight junctions are
functionally altered, and fluid moves out of cerebral vessels by bulk flow down
a transmural pressure gradient. It is the pattern around brain metastases,
abscesses, trauma and haemorrhage, and it lives predominantly in white
matter — most commonly the deep white matter of the hemispheres. This is the
kind that responds to corticosteroids, and the response is dramatic enough to
have shaped everyone’s intuitions about steroids and brain swelling.
Cytotoxic. The barrier is intact and the water is inside the cells. Energy
failure disables the sodium–potassium ATPase, and the swelling is intracellular.
The obvious causes are toxicity, ischaemia and hypoxia — and the less obvious
ones are the reason this column matters to you: hypo-osmolar states,
including dilutional hyponatraemia, acute sodium depletion and SIADH, and
osmotic disequilibrium syndromes such as haemodialysis and diabetic
ketoacidosis. Give enough hypotonic fluid and you are on that list.
Two features distinguish it. It involves both grey and white matter, in
contrast to vasogenic. And in the early stages fluid simply moves from the
extracellular to the intracellular compartment with no net change in brain
volume, which is why cytotoxic oedema tends to produce less mass effect than
its severity suggests — until the extracellular compartment equilibrates with
the intravascular one.
Interstitial. The barrier is intact and the cells are working. Cerebrospinal
fluid is being forced into periventricular white matter because something is
impeding its circulation or absorption. This is the oedema of hydrocephalus, and
it is the one no drug reaches: the treatment is to relieve the obstruction, which
is the subject of day 12.
Real brains do not respect the taxonomy. A contused hemisphere has a leaking
barrier and failing pumps in the same field, and a patient with a glioma can be
hyponatraemic as well. The point of separating them is that the treatments
separate cleanly even when the pathology does not.
One imaging note, because it is the only thing that reliably separates the first
two. The appearance of oedema is essentially similar whatever produced it — dark
on CT, dark on T1, bright on T2 — with one exception: diffusion is reduced in
cytotoxic oedema and increased in vasogenic oedema. Location and DWI are the
discriminators; everything else looks the same.
What the steroid actually does
Corticosteroids are very effective at decreasing vasogenic brain oedema. In
patients with peritumoral oedema they typically reduce intracranial pressure and
improve neurologic symptoms within 48 hours.
Dexamethasone is the agent of choice — minimal mineralocorticoid effect, high
potency, long half-life. It is typically started before surgery with a bolus of
10 mg followed by 4 mg every 6 to 8 hours, continued through the craniotomy,
and tapered rapidly afterwards. Glucose should be monitored in everyone on it,
and patients on higher doses watched for psychiatric disturbance, which is
routinely missed because nobody is looking.
Two things about that regimen are worth knowing. Peritumoral oedema can persist
or even worsen after the mass is removed, which is why the drug is continued
through surgery rather than stopped at induction. And the reflex to increase the
dose when improvement is slow has been tested: in a randomized comparison of 4, 8
and 16 mg a day in metastatic brain tumours, 4 mg produced the same improvement
in Karnofsky score as 16 mg after a week, with toxic effects significantly more
frequent at the higher dose.
What the steroid does not do
Corticosteroids are not only ineffective in cytotoxic oedema — they are
relatively contraindicated in cerebral ischaemia.
The definitive test of the generalization was MRC CRASH. 10,008 adults with
clinically significant head injury were randomized to a 48-hour infusion of
methylprednisolone or placebo. Death within two weeks occurred in 21.1% of
the corticosteroid group and 17.9% of the placebo group — relative risk
1.18 (95% CI 1.09–1.27), P = 0.0001. The trial was stopped early.
That result is worth carrying in its full form, because the short version — “no
steroids in head injury” — is a rule, and rules get forgotten. What happened is
that a drug with a real and visible effect on one kind of brain swelling was
extended to another kind on mechanistic grounds, and when someone finally
measured it, the extension was killing people.
The one you cause
Within the cytotoxic column sits the only mechanism on this page that your own
decisions produce, and it is worth pulling out.
The cerebral capillary excludes even sodium and chloride, so water movement
across it is governed by total osmolality, not by oncotic pressure. Let the
plasma osmolality fall and water crosses into a brain whose barrier is entirely
intact — and ends up, as the taxonomy says, inside the cells. Lactated Ringer’s
at about 273 mOsm/L against a plasma value near 295 is the everyday version;
free water and dextrose solutions are the emphatic one.
The treatment is not a drug. It is to stop producing the gradient, and then to
correct it slowly enough that the correction is not its own injury. That
physiology — reflection coefficients, osmolality against oncotic pressure, and
why hypertonic solutions work at all — is the next day. What the correction rate
has to respect is day 15.
A note on what “reduces oedema” is worth
There is a related result worth knowing while you are here. In the traumatic
brain injury cohort of the SAFE study, resuscitation with albumin rather than
saline was associated with higher mortality at 24 months: 33.2% versus 20.4%,
relative risk 1.63 (95% CI 1.17–2.26), and in severe injury 41.8% versus 22.2%.
Albumin is not a steroid and the mechanism is different. What it shares with
CRASH is the shape: an intervention chosen for a sensible physiological reason,
applied to an injured brain, and found on measurement to be doing harm. Two large
trials, two plausible drugs, the same lesson about the distance between mechanism
and outcome.