Cluster Headache: Symptoms, Attack Treatment and Prevention

The diagnosis rests on the characteristic clinical picture described in the ICHD-3 classification. The pain is one-sided, very severe or extremely severe, boring, stabbing or burning in character, and located in the orbital, supraorbital or temporal region. A single attack lasts 15–180 minutes untreated, and in the active period can recur from once every other day up to eight times a day. At least one of two features is required: autonomic symptoms on the same side and/or restlessness or motor agitation.

The name comes from the grouping of attacks into series. In the episodic form (most patients) the cluster period lasts from 7 days to a year, and between series there are remissions lasting at least 3 months — sometimes many months or years. In the chronic form attacks occur for more than a year without remission, or with breaks shorter than 3 months. Patients can move from one form to the other, which matters for the choice and length of preventive treatment.

Many people notice a seasonality of series (more often in spring and autumn) and the constancy of the attack hours. These observations agree with the involvement of the hypothalamus — the structure that controls the circadian rhythm — in the pathophysiology of the disease. It was originally assumed that the disease mainly affects men (a ratio of about 4.3:1), but newer papers point to a smaller difference (1.3–2.6:1), which is attributed to earlier misdiagnosis in women — one more argument for the pattern of pain, not sex, deciding the diagnosis.

The first layer is the trigeminal-autonomic reflex. Activation of the ophthalmic branch of the trigeminal nerve (the source of the pain around the eye) triggers a connection with the parasympathetic system through the superior salivatory nucleus and the sphenopalatine ganglion, which explains the tearing, conjunctival congestion and blocked nose on the same side. Constriction of the pupil and drooping of the eyelid relate to involvement of sympathetic fibres. This is why the pain is sometimes mistaken for eye, sinus or dental disease.

The second layer is the hypothalamus, which appears to be the 'generator' of the series. Functional brain studies (including PET and functional MRI) have described activation of the posterior hypothalamus during an attack, and the circadian and seasonal rhythms and changes in hormone secretion, including melatonin, point to a disturbance of the biological clock. It is worth stressing that this is mainly mechanistic knowledge: it explains why attacks are punctual and why fixed sleep times are sometimes important, but it does not translate directly into proven treatment methods. Triggers during a cluster include alcohol (even a small amount) and, in some people, nitroglycerin, high temperatures or changes in sleep rhythm. Outside a cluster alcohol usually does not provoke attacks.

The relationship between cluster headache and sleep runs both ways: attacks often wake people at night, and some observations suggest more frequent coexistence of sleep apnoea in patients; the evidence here is mainly observational and it is not known whether treating apnoea changes the course of the pain. We write about sleep apnoea in separate articles, and the causes of night-time attacks are worth discussing with a neurologist rather than guessing on your own.

Two methods have a strong recommendation for an attack in the European treatment guideline (EAN 2023): inhaling 100% oxygen at a flow of at least 12 L/min for 15 minutes, and sumatriptan 6 mg subcutaneously. The choice between them depends on availability, contraindications and the number of attacks in 24 hours. Let us stress: oral drugs (including triptans in tablets) are not suitable, because the attack peaks within a few minutes and usually passes before a tablet starts to work.

What follows in practice? Oxygen is a safe method and can be used repeatedly in 24 hours, which makes it attractive when attacks are frequent; however, it needs a cylinder or concentrator and a suitable mask (a high flow and a mask with a reservoir bag are recommended). Injected sumatriptan acts fastest but is contraindicated, among other things, in coronary disease, previous stroke and uncontrolled hypertension, and the number of doses per day is limited by the leaflet and the doctor. This text is no substitute for a prescription: access to home oxygen and the dosing schedule are discussed with a neurologist. Conclusions about availability and reimbursement in Poland are not the subject of this article.

Because attacks can occur several times a day, treating the attack alone is not enough — in the active period prevention is used. The 2023 guideline of the European Academy of Neurology recommends verapamil at a dose of at least 240 mg a day as the first-choice drug (the maximum dose depends on efficacy and tolerability). Because of its effect on conduction in the heart, treatment requires an ECG before starting and when increasing the dose; doses in cluster headache are sometimes higher than in hypertension, so medical monitoring is essential.

Corticosteroids play the role of a 'bridge': they act quickly, so they are given briefly at the start of a series until the more slowly building verapamil takes effect. They are not suitable for long-term use because of adverse effects. Greater occipital nerve block (an injection of local anaesthetic and steroid) is another transitional method. Lithium, topiramate and galcanezumab are considered when verapamil is ineffective or poorly tolerated; each has its own limitations, and the choice belongs to the neurologist.

In migraine, antibodies against CGRP (galcanezumab, fremanezumab, erenumab) have become an important preventive drug, so it was natural to test them in cluster headache. The results, however, are more nuanced than in migraine. Galcanezumab has the position of an alternative in the EAN guideline only for the episodic form, and a recent network meta-analysis showed that the pooled picture is modest.

Conclusion: anti-CGRP antibodies are not 'the solution' for cluster headache and should not be presented as such. In some patients with the episodic form they may help, but the pooled data do not show a large, certain effect. This is an example of how a drug effective in one disease (migraine) does not necessarily carry its success over to a related one. In addition, the results of individual trials in narrow subgroups matter a great deal, not general marketing slogans.

Cluster headache is rare, so primary-care doctors, dentists and ophthalmologists see it only occasionally. The typical route of a patient to the diagnosis runs through diagnoses of sinusitis, toothache, temporomandibular joint problems, glaucoma or migraine — sometimes with an unnecessary tooth extraction. A clue that it is not the sinuses is the attack-like course (minutes to hours, not days), motor restlessness and repeatability at a fixed time; we write about the sinuses in a separate article on chronic sinusitis.

A first attack of this kind always requires neurological assessment. The 2024 review mentioned above notes that magnetic resonance imaging of the brain is mandatory to exclude secondary causes (tumours in the pituitary region, vascular malformations, carotid artery dissection and others). We write about the examination itself in the entry on magnetic resonance imaging.

Cluster headache is one of those diseases in which the right diagnosis changes everything, because effective treatments exist but are sometimes unused for years owing to diagnostic mix-ups. The strong evidence concerns aborting the attack (oxygen, subcutaneous sumatriptan), while prevention rests mainly on verapamil and short-term steroid 'bridges'. Newer, expensive therapies such as anti-CGRP antibodies so far give less than their position in migraine would suggest. The best thing a patient can do is to keep an attack diary and see a neurologist quickly, ideally one who deals with headaches.