Cluster headache (CH) is a primary headache belonging to the trigeminal autonomic cephalalgias (TAC) group. The earliest descriptions date back to 1641, when the Dutch physician Nicolaes Tulp, famous for Rembrandt's painting, “The Anatomy Lesson,” described a recurrent intense unilateral headache no longer than 2 h. In the Medical Observations (1), the autonomic features were characterized by Wilfred Harris (1869–1960), a London neurologist, in his classic monograph Neuritis and Neuralgia in 1926 (2); this was the first recognition of CH as a separate entity from migraine and trigeminal neuralgia (2). In 1936, Harris named these headaches migrainous neuralgia or ciliary (migrainous) neuralgia (3), where he reported the one-sidedness of the attacks, the severity, associated autonomic features, and the frequency of attacks. His description was the first recorded report of CH in the English medical literature. The same clinical features are detailed in the International Classification of Headache Disorders-3 (ICHD-3) (4).

CH has the legendary reputation of being the most severe primary headache and one of the most excruciating pain conditions a human being can experience (with terrible intensity and generating the most remarkable restlessness). It is colloquially known as the “suicide headache” as many patients may contemplate suicide during attacks. The lack of education of emergency room physicians and various specialists who are not specifically trained and experienced in the management of headache disorders leads to underdiagnosed and often inadequate treatment of CH. The simplicity of its clinical picture contrasts with the fact that the diagnosis is made with an average delay of 5 years from the first occurrence of attacks, and the correct therapeutic management is provided only to a minority of these patients (5, 6). As a natural consequence of the severity of CH, the disability it causes, and the described deficiencies in health care, patients with CH show a high proportion of sickness absence. Beyond the severe impact on the quality of life of affected CH patients, the financial benefits resulting from sick leave or statutory sick pay and the higher number of disability pensions cause a significant cost to society (5, 6).

TACs are primary headache disorders characterized by pain localized in the first division of the trigeminal nerve in parallel with ipsilateral cranial autonomic features. CH is the most common and best-studied TAC. It is characterized by severe pain attacks, strictly unilateral, of orbital, supraorbital, temporal, or any combination of these locations, the average duration ranging from 15 to 180 min and occurring from once every two days to eight times a day when in the active phase. Attacks accompany ipsilateral cranial autonomic symptoms such as lacrimation, ptosis, ocular flushing, rhinorrhea/nasal congestion, miosis, and restlessness. Most patients suffer from episodic CH with attacks occurring in episodes (i.e., clusters), usually lasting weeks to months, separated by attack-free periods, which can last between three months to several years. About 10%–20% of patients have a chronic variant without significant attack-free periods (less than three months/year). Attacks usually follow a circadian pattern, commonly occurring around the same time of the day. As attacks naturally also happen at night, they significantly impact sleep quality, leading to several consequences and adding to the disease burden (4).

The clinical picture of CH and its diagnostic criteria are defined in the current version of the International Classification of Headache Disorders (ICHD-3) published by the International Headache Society (IHS) (4).

CH affects 0.1% of the population (7) and seems more frequent in males, but there is contradictory data on this aspect. Previously, it was thought that CH mainly affects men. Still, the difference in the prevalence between sexes is decreasing, not because the prevalence is changing, but because more women are being correctly diagnosed. A typical feature of CH is the circannual variation in its incidence (more frequent in the spring and the autumn) (8). Fifty-five percent of patients with CH have suicidal ideation, although it is rare for them to commit it (4), and depression occurs almost three times more often than in controls (8). Delayed diagnosis in young people is also common (9).

A Danish study documented the personal and occupational limitations of 400 patients with CH, estimating that 94% of them had restrictions during pain attacks. Patients rated their health as poor/very poor in the episodic form in 9% compared to 1% of controls. In the chronic presentation, the odds of rating health as good/very good were ten times lower (OR: 10 < 10, 95% CI: 5.29–18.79). The odds of receiving a disability pension were five times higher in the chronic compared to the episodic form (OR: 5.0, 95% CI: 2.3–10.9, p < 0.001). The individuals who were in presenteeism from their employment despite having a CH attack are estimated to have a 65% reduction in their productivity (46).

The pathophysiology of CH still needs to be understood. The current understanding is based on preclinical, clinical, and imaging studies in patients with the disease. However, research in CH needs to catch up to other primary disorders, such as migraine, as animal models remain scarce, and clinical research is hampered by the severity and short duration of attacks and attack bouts, which complicate the recruitment of study subjects. Finally, because of the relatively low prevalence compared to migraine, funding for CH research remains very limited and is mainly directed to other more prevalent primary headache disorders, such as migraine (6).

To better understand the changes that occur in the clinical picture of CH, nitric oxide (NO) and its prodrug nitroglycerin have been used to provoke migraine-like attacks in patients previously diagnosed with migraine and cluster-like attacks in patients which have been diagnosed with cluster headaches and are in the bout. Nitroglycerin increases CGRP levels by causing vasodilatation and hyperactivity of trigeminal nociceptive fibers, as demonstrated by Fanciullacci in his studies (47, 48). The same occurs following the administration of CGRP (49, 50). Whether the effect of NO is via CGRP or if the impact attack-triggering effects are independent of each other has yet to be clarified as the available data are inconsistent. However, NO and CGRP can most likely trigger attacks by separate mechanisms, although the two may interact with each other. Other inducers of CH attacks can include, besides CGRP, PACAP38, and VIP, which can act on the mast cell (15, 51–53).

The neuroanatomical and functional systems involved in the pathophysiology of CH can be divided into three principal components: (1) the trigeminovascular system, (2) the trigeminal-autonomic reflex (sphenopalatine ganglion stimulation), and (3) the hypothalamic system (15). The interaction of these three components is responsible for the characteristic clinical presentation of CH. We will discuss these components in detail in Figures 1–3.

The diagnosis of CH is primarily clinical, and neuroimaging studies are indicated in specific cases to rule out secondary headaches (4). Frequently, there is a significant delay in diagnosing CH, which is essential when planning optimal medical management. In this context, Byung-Su Kim et al. showed in a multicenter registry in South Korea with a 4-year follow-up period that in 36.4% of patients, the average diagnostic delay was 5.7 ± 6.7 years (77). A Danish study reported that the leading causes of diagnostic delay included prolonged attack duration (greater than 180 min), migraine-like clinical features, and predominantly nocturnal episodes (75). A survey on CH conducted in the United States revealed that 21% of patients received an adequate diagnosis at the onset of the problem. There could be an average delay of 5 years after the initial start of the CH (76).

When CH is misdiagnosed, the clinical picture is most confounded as migraine, either because the clinical picture is misinterpreted or because both headache disorders occur as comorbidity, which is then not recognized. Other common misdiagnoses include trigeminal neuralgia, sinusitis, or dental and jaw disorders, leading to unnecessary treatments and procedures and increased anxiety, depressive disorders, and even suicidal ideation (79, 80). The Erwin test is a tool that identifies patients with CH. This test consists of 3 questions: Is this the worst pain you have experienced? Does your pain last less than 4 h? Do one or more of the following symptoms or signs occur during the headache: unilateral red eye, unilateral lacrimation, unilateral rhinorrhea, or unilateral nasal congestion? In the case of an affirmative response to all three questions, a sensitivity of 85% and specificity of 89% of correct CH diagnosis is obtained (81), proving to be a valuable and easy-to-use tool for proper diagnosis.

Differential diagnosis can be challenging. Nevertheless, it should be guided by anamnesis and initial physical examination, followed by laboratory studies to determine if there is any other etiology of the paroxysmal neurological events, e.g., serum electrolyte disturbances or hepatic or renal insufficiency, lumbar punctures in case of suspected neuro infection. Neuroimaging studies such as brain tomography or magnetic resonance imaging (MRI) are indicated in patients with an atypical clinical picture to rule out other secondary etiology because of specific vascular alterations that can be found in 37.7% (e.g., internal carotid artery dissections, arteriovenous malformations of any cerebral lobe, intracranial aneurysms [vertebral artery, inferior cerebellar artery, posterior communicating artery, multiple, internal carotid], artery dissection, intracavernous internal carotid artery thrombosis, dural arteriovenous fistulas, and cerebral venous thromboses), tumors in 32.5% (prolactinomas, epidermoid cysts, meningiomas, parietal glioma, besides others.) and inflammatory entities in 27.3% (sinusitis of any sinus, sphenoid sinus mucocele, inflammatory orbital pseudotumor, posterior scleritis, idiopathic orbital myosis, hypothalamic-pituitary granuloma, and idiopathic intracranial hypertension). These lesions typically occur ipsilateral to the symptoms and can, therefore, mimic CH (81, 82).

For these reasons, it is recommended to perform cranial MRI in patients with suspected CH (82). The American College of Radiology recommends using contrast medium on MRI (83). In patients who do not have symptomatology suggestive of pituitary adenoma, MRI with sella turcica focus is not necessary, as a retrospective study demonstrated that the prevalence of pituitary adenomas in patients with CH is like that reported in the general population (38). Since arterial dissection can simulate a CH, the European Headache Federation has recommended using supra-aortic truncal angio-resonance or carotid and vertebral artery ultrasound in selected patients (83, 84).

In summary, an MRI of the brain with contrast is recommended in every patient with CH, an MRI of the sella turcica in suspected pituitary tumors and in patients with suspected cerebrovascular lesions, an angio-MRI or MRI of aortic trunks (83).

Atypical features of CH that should alert the clinician are (1) pain attacks being exclusively ocular or retro-ocular, (2) abnormal findings on neurological examination, (3) other headache attacks between those typical of CH, (4) atypical duration of CH, (5) migraine-like symptoms, (6) Horner's syndrome, and (7) an unexpected frequency of atypical attacks, being the most relevant, in particular cranial nerve disorders, within which ophthalmological signs and symptoms are the most frequent. In all cases of primary CH, the findings of laboratory or neuroimaging examinations are expected to be normal or unrelated to the etiology of CH (84–86).

Table 2 and Figure 1 show the diagnostic approach to CH concerning other common primary headaches.

Resistance to standard treatments for CH should increase the suspicion of a secondary origin. However, acute treatments with triptans, ergotamine, analgesics, or caffeine can improve attacks by up to 46.9%, so a response to therapy does not justify conducting further investigations (87).

Misdiagnosis of CH has a dramatic impact on the patient's quality of life. A survey in the United States revealed that even if the diagnosis is established correctly, 25% of patients have lost their jobs due to the disease, and 8% are unemployed (76). In addition, as events tend to be more predominant at night, they significantly affect sleep quality, adding to the disease burden. Using the Headache Impact Test-6 (HIT-6) in patients with CH, several authors observed that up to 74% were classified as having a severe impact, with 78% reporting daily restrictions and up to 96% a need for lifestyle changes (88), demonstrating the detriment of the quality of life of these patients.

There are effective and safe pharmacological and non-pharmacological treatments with the heterogeneity of clinical trial designs for patients with CH divided into acute, transitional (short preventive treatment or bridging), and preventive long-term interventions. Acute attacks are treated using triptans, oxygen, and—in the case of episodic cluster headache—non-invasive transcutaneous vagal nerve stimulation. Prednisone is the most studied in the bridging phase. Moreover, verapamil and monoclonal antibodies are considered the first option, followed by multiple pharmacological and non-pharmacological options for preventive treatment (89).

During acute attacks of CH, the use of triptans is widely recommended. Sumatriptan 6 mg subcutaneously is one of the most effective acute treatments (90, 91). Alternative options with similar efficacy are sumatriptan 20 mg (intranasal) and zolmitriptan 5/10 mg (intranasal). With an overall response rate of triptans in CH of 80%, triptans remain one of the most effective acute treatment options in CH (92, 93). However, due to their vasoconstrictive properties, they are contraindicated in patients with comorbid cardiovascular pathologies (94). In addition, its use is limited to 2 doses in 24 h, which is problematic for patients who experience more than two attacks in 24 h. Unfortunately, Mexico has no access to intranasal or subcutaneous formulations of triptans, which commonly leads to the use of an oral triptan with an analgesic such as naproxen (or other NSAID) to abort the acute pain (95).

Another accessible and highly effective acute treatment is high-flow oxygen, which can be administered at the patient's home or in the emergency room. A non-rebreather mask is used at a flow of 100% oxygen at a rate of 7–15 L/min. Response rates range from 62% to 100% of patients with acute attacks, with a positive answer in 12–15 min on average (96–98).

Based on the high response rates, intranasal and subcutaneous triptans, and oxygen are considered first-line options for the acute treatment of CH attacks. Additionally, non-invasive transcutaneous vagal nerve stimulation is a non-pharmacological option proven effective in episodic CH, reducing the pain intensity within 15 min of stimulation with a favorable safety profile (99).

In the transitional phase, using prednisone orally at 100 mg/day for five days and gradually decreasing the doses is recommended (100). Another alternative, which is probably similarly effective but better tolerated, is the greater occipital nerve block (101, 102).

Verapamil, a calcium channel blocker, is the first-choice pharmacological treatment for preventing episodic and chronic CH using at least 240 mg/day (103). When using verapamil, it is essential to conduct an electrocardiogram before initiating treatment and after each dose increase to identify potential cardiac side effects. Even after a stable dose has been determined, electrocardiograms should be conducted regularly, as cardiac side effects can appear with a significant delay (104–106). Interestingly, in Mexico and other Latin American populations, verapamil doses of no more than 240 mg/day are used, unlike in different latitudes where the recommended doses are up to 720 mg/day of this drug, which possibly has a genetic explanation (103, 107, 108).

The monoclonal antibody galcanezumab is also an FDA-approved preventive treatment option for CH (300 mg/month for two months). The respective randomized controlled trial showed that it decreased weekly attack frequency by 8.7 compared to 5.2 in the placebo group at week three. The most frequent side effects were nasopharyngitis and pain at the application site (109, 110). Unfortunately, the study on chronic CH did not reach statistical significance in a trial with 237 participants (108). In Mexico, galcanezumab has yet to be approved for use in CH. Besides, galcanezumab does not exist at 300 mg presentation in Mexico. Minor dose regimens have proven successful in some patients, but no data from a randomized clinical trial would confirm these observations (111).

The second line includes lithium, civamide, melatonin, topiramate, sodium valproate, baclofen, gabapentin, and transcutaneous non-invasive vagal nerve stimulation, followed by botulin toxin and sphenopalatine ganglion stimulation as a third line of treatment (9, 54). However, most of these studies have methodological shortcomings, such as, for example, an open-label design or a small sample size. Unfortunately, ATI, the company that made this stimulator, went bankrupt.

For refractory CH, the use of invasive treatments such as sphenopalatine ganglion blockade with radiofrequency ablation or neuromodulation may be an option in these patients as in one case series it led to a decrease of pain in 31% at six months of follow-up after radiofrequency treatment or 75% at 24 months of follow-up after neuromodulation (112); however, 81% of patients showed maxillary paresthesia when using neuromodulator treatments or were re-operated (113).

Greater occipital nerve stimulation (ONS) has been successfully used in cases with chronic CH refractory to medical treatment. In the early Fontaine et al. open-label trial, a 77% response rate (improvement >50%) was observed after ONS, with a 68% reduction in frequency and a 49% reduction in severity of CH attacks (114). Therefore, based on the feasibility and costs associated with the different surgical procedures currently available, ONS could be considered the first therapeutic strategy for patients with refractory chronic CH, as shown by the current evidence.

A recent meta-analysis of 45 ONS studies for refractory CH showed a pooled response rate of 57%. Deep brain stimulation (DBS) was the second most studied surgical treatment, with a pooled response rate of 77% (115–45).