Cerebellum
The cerebellum does not start movements — it checks them, comparing what the cortex ordered against what the body actually did and correcting the difference thousands of times a second; break it and power and sensation are untouched, so what you see is clumsiness on the same side as the lesion, and what kills is that all of this sits in a rigid box in front of the brainstem.
What it normally does
The cerebellum does not initiate movement and carries no command to any muscle. It receives a copy of every motor plan from the cortex (relayed through the pontine nuclei) and, at the same moment, a running report of what the body actually did — from muscle spindles, tendon organs, the vestibular apparatus and the eyes. It compares the two and sends a correction back through the deep nuclei to the thalamus and motor cortex [the Purkinje cells are the sole output of the cerebellar cortex and they inhibit the deep nuclei, which are themselves the output of the cerebellum as a whole].
Explains why cerebellar damage never causes weakness and never causes numbness. Power is full and sensation is intact, but the movement overshoots, corrects, and overshoots again — clumsy, not weak (ataxia, dysmetria, intention tremor, dysdiadochokinesia, and speech that comes out in uneven bursts).
The wiring crosses twice. A cerebellar hemisphere talks to the opposite motor cortex — the output crosses in the superior cerebellar peduncle — and that cortex drives the opposite side of the body, crossing again in the medulla. Two crossings cancel out.
Explains the single most useful localising rule on this page: cerebellar signs sit on the SAME side as the lesion. A right cerebellar stroke makes the right hand miss and the patient veer and fall to the right — the exact opposite of a right cortical stroke.
The cerebellum is divided by function down its length. The midline strip (vermis) and the small lobe tucked under it (flocculonodular lobe) run the trunk, stance, walking and eye movements, wired mainly to the vestibular nuclei. The two big lateral hemispheres run the limbs — especially fast, fine, distal movement — wired to the cerebral cortex.
Explains why a lesion in the middle gives a patient who cannot stand or walk but whose finger-nose test is normal, while a lesion out to the side gives one clumsy limb on the same side in a patient who sits comfortably.
It sits in the posterior fossa, a small rigid box below the tentorium, with the brainstem directly in front of it and the fourth ventricle squeezed between the two. Every drop of CSF made above has to pass from the third ventricle down the narrow cerebral aqueduct into that fourth ventricle and out through its foramina to reach the subarachnoid space, so the fourth ventricle is the last and most easily compressed point on the whole route. Its entire blood supply is the posterior circulation: the vertebral and basilar arteries, through PICA, AICA and SCA, the same vessels that feed the brainstem.
Explains why a cerebellar bleed or a swelling cerebellar infarct kills — it blocks the fourth ventricle, causing acute hydrocephalus, and crushes the brainstem — and why cerebellar signs rarely arrive alone in a stroke: the brainstem next door is fed by the same artery, which is how you get ataxia alongside a Horner syndrome, crossed sensory loss and swallowing failure.
What goes wrong
- Cerebellar infarct — posterior circulation stroke← from “The wiring crosses twice. A cerebellar hemisph…”
A clot in a vertebral artery, the basilar, or one of their branches (PICA, AICA, SCA) kills part of a hemisphere. The source is usually atrial fibrillation or atherosclerosis, but in a younger patient after neck trauma, a chiropractic manipulation or even a heavy cough it can be a vertebral artery dissection. Because the cerebellar output crosses twice, the comparator now fails for the limbs on the same side as the dead tissue.
Ipsilateral limb ataxia plus sudden vertigo plus cannot walk equals posterior circulation stroke until MRI says otherwise. Normal head impulse in acute continuous vertigo is bad news, not reassurance. Aboriginal and Torres Strait Islander Australians are hospitalised for stroke at close to twice the rate of other Australians and often a decade younger, so a young patient is no reason to relax.
You would find: Sudden vertigo, vomiting, headache and an inability to stand, with past-pointing, dysmetria and clumsy rapid alternating movements (dysdiadochokinesia) on the side of the lesion — and full power, normal sensation. This is mistaken for vestibular neuritis or gastroenteritis constantly. In acute continuous vertigo the head impulse test decides it: a NORMAL head impulse (no corrective saccade), nystagmus that changes direction with gaze, or vertical misalignment of the eyes (skew) means central. Someone who cannot sit or walk unaided is central until proven otherwise. Plain CT is done first and is good at showing blood, but it is nearly blind to posterior fossa infarction — a normal CT does not exclude this, and diffusion-weighted MRI is the test.
- Malignant cerebellar swelling — space-occupying infarct and hypertensive cerebellar haemorrhage← from “It sits in the posterior fossa, a small rigid …”
Oedema after a large cerebellar infarct peaks two to four days later; a hypertensive bleed into the region of the dentate nucleus and deep cerebellar white matter does the same thing in minutes. The posterior fossa has no room to give. The fourth ventricle is squashed, so cerebrospinal fluid cannot get through it and acute obstructive hydrocephalus develops; the brainstem is pushed forward against the clivus; the tonsils cone down through the foramen magnum, and the swollen cerebellum can also push upward through the tentorial hiatus.
Drowsy on day three after a cerebellar stroke means swelling: rescan, call neurosurgery. Sudden occipital headache and vomiting in someone anticoagulated with a cerebellar bleed on CT means reverse the anticoagulation now.
You would find: A stroke that looked survivable, then worsening headache and vomiting and a conscious level that slides on day two to four. Look for new sixth nerve palsy, small reactive pupils, and bradycardia with rising blood pressure. This is the one cerebellar emergency with a surgical answer — suboccipital decompressive craniectomy, with an external ventricular drain for the hydrocephalus — and patients who get it in time can walk out.
- Alcohol-related cerebellar degeneration, and Wernicke encephalopathy← from “The cerebellum is divided by function down its…”
Chronic heavy drinking plus a poor diet strips thiamine (vitamin B1), the cofactor the mitochondria need to burn glucose; alcohol also blocks its absorption, its storage in the liver and its conversion to the active phosphate. Direct ethanol toxicity to Purkinje cells contributes as well, which is why abstinence helps but does not always reverse the gait. The neurons with the highest metabolic traffic die first, and in the cerebellum those are the Purkinje cells of the anterior part of the midline strip — the vermis. The vermis runs trunk and legs; the hemispheres are largely spared.
Gait ataxia with normal arms in a drinker is anterior vermis. Alcohol is the commonest cause of chronic cerebellar degeneration in Australia; bread-making flour has been thiamine-fortified here since 1991, which cut Wernicke-Korsakoff cases, but it protects nobody whose calories come mostly from drink. Alcohol-related harm and hospitalisation fall disproportionately on Aboriginal and Torres Strait Islander communities, even though a higher proportion abstain completely than in the rest of the population — so ask about the pattern of drinking, never assume it from the person.
You would find: A wide-based, stumbling, unsteady gait developing over months to years, with an abnormal heel-shin test, but a normal finger-nose test, normal speech and normal arms — legs and trunk far worse than arms. Add acute confusion, unsteadiness and eye signs (nystagmus, a sixth nerve palsy or gaze palsy) and it is Wernicke encephalopathy, a medical emergency; most cases do not show all three parts of the triad, so one part is enough to treat.
- Drug-induced cerebellar toxicity← from “The cerebellum does not initiate movement and …”
The comparator circuit depends on neurons firing fast and precisely, so drugs that damp high-frequency firing hit it before they hit anything else. Phenytoin and carbamazepine block sodium channels in a use-dependent way, so the fastest-firing cells are silenced first. Lithium is not a sodium channel blocker — it reaches the same cells by a different route, accumulating when clearance falls and injuring Purkinje cells, but the clinical picture it produces is the same. Chronic high phenytoin concentrations go further and kill Purkinje cells permanently.
New nystagmus and ataxia in a patient on phenytoin, carbamazepine or lithium is a drug concentration, not a stroke call. Most acute toxicity reverses, but not all of it: years of high phenytoin concentrations leave permanent cerebellar atrophy, and a severe lithium toxicity can leave permanent ataxia and nystagmus long after the level is normal again (the syndrome of irreversible lithium-effectuated neurotoxicity). That is the argument for catching it early, not for reassuring the patient.
You would find: The order of appearance is the giveaway: nystagmus first, then limb and gait ataxia, then slurred speech and drowsiness — coming on over days after a dose change, a new interacting drug, dehydration or an intercurrent illness. Lithium: ataxia with a coarse tremor, vomiting and diarrhoea, usually after starting an NSAID, ACE inhibitor or thiazide, or after a gastro illness.
- Midline lesions — vermis and flocculonodular lobe← from “The cerebellum is divided by function down its…”
A lesion in the middle cuts the trunk, stance and eye-movement channel while leaving the limb channel out to the sides intact. In a child the classic cause is a tumour arising in the vermis and filling the fourth ventricle (medulloblastoma), which also blocks CSF drainage. In adults it is alcohol, a demyelinating plaque, or a bleed into the vermis.
Truncal ataxia with normal limb testing equals midline. Cerebellar patients are unsteady with their eyes open; if closing the eyes is what makes it dramatically worse, the problem is in the dorsal columns or peripheral nerves, not the cerebellum.
You would find: Truncal ataxia: the patient cannot sit unsupported or walk heel-to-toe and sways with feet together whether their eyes are open or closed — but their finger-nose test is normal when they lie down, and their arms look fine. Nystagmus is common. In a child, add early-morning headache and vomiting and a head that is tilted. Distinguish from sensory ataxia, where the sway is far worse the moment the eyes shut (a positive Romberg sign).
What we give, and how it works
Open a drug to see what it binds, what that does to the cell, and what you then see in the patient.
- Binds
- Not a receptor. Thiamine is phosphorylated to thiamine pyrophosphate, the obligatory cofactor for pyruvate dehydrogenase, alpha-ketoglutarate dehydrogenase, branched-chain ketoacid dehydrogenase and transketolase.
- Which does
- Restores the entry of pyruvate into the Krebs cycle and the flow through the pentose phosphate pathway. Without it, oxidative ATP production collapses in the neurons with the highest metabolic traffic, lactate accumulates locally and glutamate is not cleared.
- So you see
- The recovery runs in a fixed order, and the order matters more than the exact timing. Eye signs go first — ophthalmoplegia and nystagmus often start improving within a day. Confusion lifts more slowly, over days to weeks. Ataxia improves last, partly, and often not completely. Purkinje cells already dead do not come back, so an established wide-based gait stays. Untreated, Wernicke encephalopathy becomes permanent amnesia with confabulation (Korsakoff syndrome).
- And the same mechanism causes
- The drug itself is close to inert — the danger runs the other way. Glucose given first is pushed into oxidative metabolism, and the thiamine-dependent steps immediately downstream of glycolysis — pyruvate dehydrogenase, and transketolase in the pentose phosphate pathway — consume the last remaining thiamine, which can tip a marginal patient into acute Wernicke encephalopathy within hours. Thiamine goes in before, or at the same time as, the carbohydrate.
- Handling
- Oral absorption is saturable and is poor in heavy drinkers and in vomiting, so the first doses go intravenously or intramuscularly. This is a route decision, not a potency one.
Catches people out: There is no useful bedside test and no time to wait for one. The common error is withholding thiamine until the diagnosis is certain; by then the amnesia is permanent. Anaphylaxis to intravenous thiamine happens but is rare, and is not a reason to delay. Magnesium is a cofactor for the thiamine-dependent enzymes, so a patient who does not respond to thiamine may be hypomagnesaemic — correct that too.
- Binds
- Cyclo-oxygenase-1 in the platelet: aspirin acetylates serine 530 in the active site, irreversibly.
- Which does
- Thromboxane A2 can no longer be synthesised. The platelet has no nucleus, so it cannot make new enzyme — the block lasts that platelet's whole 7 to 10 day life, which is why the effect outlives the tablet. Clopidogrel works at a different target, irreversibly blocking the platelet P2Y12 ADP receptor, which is why the two add up rather than duplicate.
- So you see
- Less platelet aggregation on ruptured atherosclerotic plaque in the vertebrobasilar vessels, so fewer recurrent strokes. The risk of recurrence is highest in the first days, which is why two antiplatelets are used briefly then reduced to one.
- And the same mechanism causes
- The same COX-1 in the gastric mucosa makes the prostaglandins that maintain mucus, bicarbonate and mucosal blood flow. Block it there and you get dyspepsia, ulceration and gastrointestinal bleeding — the same mechanism, a different tissue. And because the platelet is inhibited irreversibly, stopping the drug does not switch bleeding off: the effect wears away only as the marrow replaces the circulating platelets over 7 to 10 days, which is why urgent surgery or major bleeding may need a platelet transfusion rather than time.
Catches people out: Do not start it until imaging has excluded a cerebellar haemorrhage — a bleed and an infarct look identical at the bedside, and this is the one thing plain CT does well. If the patient has been thrombolysed, antiplatelets are held for 24 hours and until the follow-up scan is clear. If the source is atrial fibrillation, the answer is anticoagulation, not aspirin. If it is a vertebral artery dissection in a younger patient, that is a specific decision for the stroke team.
- Binds
- Dopamine D2 receptors, with additional H1 histamine and muscarinic blockade.
- Which does
- The area postrema sits outside the blood-brain barrier and triggers vomiting when it samples the blood; D2 blockade there switches that trigger off. The antihistamine and anticholinergic actions are the ones that quieten the vestibular nuclei — a pure D2 blocker does not do this.
- So you see
- Nausea and vertigo settle and the patient looks much better. Gait ataxia and limb dysmetria are untouched — that is the finding that still separates a central lesion from an inner ear one after the drug has worked.
- And the same mechanism causes
- The same D2 receptors sit in the nigrostriatal pathway. Block them and you get acute dystonia — oculogyric crisis, torticollis, jaw spasm — hours after a single dose, classically in young patients, reversed with an anticholinergic such as benztropine, plus drug-induced parkinsonism and akathisia with continued use. One receptor, two territories, two entirely different consequences. The antimuscarinic action that helps the vertigo is also the one that causes dry mouth, urinary retention and confusion in an older patient.
Catches people out: Give it for days, not weeks. Prolonged use delays the brain's own vestibular compensation, so the dizziness lasts longer, and its sedation can mask the falling conscious level of a swelling cerebellar infarct. Before anyone treated with it is discharged, watch them walk. Do not substitute metoclopramide here and call it the same drug: metoclopramide is a D2 antagonist with essentially no antihistamine or antimuscarinic action, so it treats the vomiting but is not a vestibular suppressant — and it still carries the full dystonia risk.
- Binds
- Voltage-gated sodium channels, bound preferentially in their inactivated state.
- Which does
- Recovery from inactivation is slowed, so a channel that has just fired stays unavailable. The block is use-dependent: the harder a neuron is firing, the more it is suppressed. That is why it stops a seizure spreading while leaving ordinary traffic alone.
- So you see
- Seizure control at usual concentrations. Above them, the neurons that normally fire fastest are silenced first, and they are in the vestibulo-ocular and cerebellar circuits.
- And the same mechanism causes
- The toxic syndrome appears in a fixed order that mirrors which circuits fire fastest: nystagmus first, then limb and gait ataxia and slurred speech, then drowsiness and coma. Early on it is dose-related and fully reversible. Years of high phenytoin concentrations kill Purkinje cells outright, leaving permanent gait ataxia and visible cerebellar atrophy on MRI. Gum overgrowth, coarse facial features and hirsutism come from a different mechanism entirely — effects on fibroblasts and connective tissue — not from the channel.
- Handling
- This part is phenytoin, not a class property. Phenytoin's metabolism saturates inside the therapeutic range (Michaelis-Menten, effectively zero-order at clinical concentrations), so a small dose increase can produce a very large jump in concentration — this is the usual reason a stable patient becomes ataxic. It is heavily albumin-bound, so in low albumin or renal failure a normal-looking total concentration can conceal a toxic free level. Rapid intravenous loading drops blood pressure and provokes arrhythmia, from the same sodium channel block in the heart plus the propylene glycol vehicle. Carbamazepine is different: its elimination is first-order, but it induces its own metabolism over the first few weeks (autoinduction) and induces CYP3A4, so its trap is drug interactions and a concentration that drifts, not a saturating one.
Catches people out: Lithium produces a similar picture by a different route, and lithium toxicity is usually precipitated by something else: dehydration, gastroenteritis, or a new NSAID, ACE inhibitor or thiazide. New nystagmus and ataxia in a patient on phenytoin, carbamazepine or lithium is a drug concentration, not a stroke call — but check the story for both. Carbamazepine has two problems of its own worth knowing: it commonly causes hyponatraemia, which produces confusion that can be mistaken for toxicity, and it carries a strongly raised risk of Stevens-Johnson syndrome in people carrying HLA-B*15:02, which is why testing is advised before starting it in patients of South-East Asian ancestry.
Three lines settle most cerebellar questions. First, the side: cerebellar signs are ipsilateral to the lesion, because the output crosses in the superior cerebellar peduncle and the corticospinal tract crosses again in the medulla — so the patient who past-points and falls to the right has a right cerebellar lesion, the reverse of a cortical stroke. Second, the level: trunk and gait bad with arms normal means midline (vermis) — alcohol, or a fourth ventricle tumour in a child; one clumsy limb with a comfortable seated patient means that hemisphere. Third, the trap: in a patient with ACUTE CONTINUOUS vertigo (the acute vestibular syndrome — not brief positional spells), a NORMAL head impulse test, direction-changing nystagmus or vertical skew means central, not the inner ear. Know what your scan can and cannot do: plain CT reliably shows acute blood, which is why it is the first test and why it is what clears a patient for aspirin, but it is nearly blind to infarction in the posterior fossa — diffusion-weighted MRI is the test for that. Two safety rules ride on top. A cerebellar stroke patient who becomes drowsy on day two to four is swelling into a rigid box — scan and call neurosurgery, do not give more antiemetic. And thiamine goes in before glucose, every time.
Now test whether it stuck
Reading this through is not the same as being able to reconstruct it. Every question in the bank is free, with a full debrief on each option.