Indomethacin‐responsive headaches—A narrative review

Indomethacin is a nonsteroidal anti‐inflammatory drug whose mechanism of action in certain types of headache disorders remains unknown. The so‐called indomethacin‐responsive headache disorders consist of a group of conditions with a very different presentation that have a particularly good response to indomethacin. The response is so distinct as to be used in the definition of two: hemicrania continua and paroxysmal hemicrania.

| 701 HEADACHE continua (HC), and a recognized, yet not obligate response in others, such as cough headache. 1 The reason(s) why this drug, unique among all the nonsteroidal anti-inflammatory medications, is particularly effective is unknown. Indomethacin has also been used for other diverse indications, including the treatment of hypercalcemia associated with certain neoplastic diseases, dysmenorrhea, premature labor, and patent ductus arteriosus closure. 2 Currently, indomethacin is FDA approved only for osteoarthritis, rheumatoid and gouty arthritis, ankylosing spondylitis, or acute painful shoulder.

ME THODS
We used the PubMed and Cochrane databases in September 2020 to perform a narrative literature review and searched for the following keywords: "Indomethacin-responsive headaches," "Indomethacin-sensitive headaches," "Indomethacin pharmacology," "Indomethacin AND intracranial pressure," "Indomethacin AND prostaglandins," and "Nitric oxide AND Indomethacin." Articles addressing pharmacology, clinical trials, and observational studies were included. The search included publications in English and Spanish. The reference lists of relevant and recent articles focusing on indomethacin-responsive headaches and potential alternatives were also reviewed and added if considered appropriate.

PHARMACOKINE TIC S
Indomethacin was, along with aspirin, one of the compounds used by John Vane to discover nonsteroidal anti-inflammatory drugs' (NSAIDs) essential activity as a prostaglandin inhibitor in 1971, 3 which subsequently led to him winning the Nobel Prize in Physiology or Medicine in 1982, 4 shared with Bergström and Samuelsson for their work with prostaglandins. 5 Indomethacin (international nonproprietary name standardized by changing British spelling-th to t) is an NSAID. The chemical name of indomethacin is 1-(p-chlorobenzoyl)-5-methoxy-2-methylindole-3-acetic acid. It is relatively insoluble in water at neutral pH. 6 It belongs to the group of acetic acids, as well as sulindac and tolmetin, in contrast to other usual NSAIDs, such as naproxen or ibuprofen, which are propionic acids. 7 There are several formulations available: capsules, extendedrelease capsules, suppositories, and injectables. Dose presentations range from 25 to 100 mg and vary by country. Treatment usually starts with a minimum dose of 25 mg three times per day and, if there is no response, should be increased to a daily dose of at least 150-275 mg to evaluate its efficacy properly. 8,9 See Table 1

Absorption
Indomethacin is readily and quickly absorbed after oral administration, with peak plasma concentrations achieved after 1 or 2 h, although this rate of absorption may vary depending on particular situations, such as food ingestion. The fastest rate of absorption has been seen in fasting subjects. 10,11 The bioavailability of the capsules and suppositories is 100% and 80%, respectively, relative to the equivalent intravenous dose, 6 and there is no difference regarding effectiveness and side effects experienced between these routes of administration. 12  The pharmacokinetics of indomethacin follows a linear and dosedependent pattern. 11

Distribution
The high lipid solubility permits indomethacin to cross the bloodbrain barrier easily, 15 possibly by penetrating the meninges by simple diffusion. The drug can be detected in the CSF half an hour after administration, according to a study on 52 patients with chronic lumbar pain due to radicular compression that received a single 50 mg intramuscular injection. 13 The tight binding of indomethacin to plasma proteins, especially albumin, may play a role in the low distribution of the drug to the CSF, given the lower amount of free indomethacin available in plasma. 16 A close relationship between the levels of indomethacin in the CSF and free plasma levels has been suggested. 13 The slight elevation of CSF levels of indomethacin compared with plasma levels might be due to the binding of indomethacin in the CSF. 13 Indomethacin crosses the placenta easily, 10,17 and a low amount of the drug is excreted in human breast milk. 10

Excretion
The excretion of indomethacin is shared; about 60% corresponds to renal excretion as a glucuronidated form, and the remaining 40% has a rapid biliary secretion. 6,10 This is associated with biliary recycling. 14 Three metabolites that are pharmacologically inactive can be recovered from urine, and the proportion of unchanged recovered indomethacin increases with alkalinization. 11 The drug has a plasma clearance of 1-1.5 ml/kg/min. 2

PHARMACODYNAMIC S
Adverse events can be present in more than 30% of patients, and the majority of them are within the gastrointestinal spectrum. Side effects such as dyspepsia, nausea, vomiting, abdominal pain, or diarrhea were frequently reported, 8,18 and some patients may mitigate these effectively with gastric protection agents, such as ranitidine, 18  Hypotensive drugs including angiotensin converting enzyme inhibitors, diuretics, or probenecid may also interact with indomethacin, 16 and careful attention must be paid when administered with other drugs with renal clearance, such as acyclovir or aminoglycosides, and particularly given the patient population, lithium, 22 to avoid nephrotoxicity. Digoxin, methotrexate, and lithium levels should be monitored regularly, as these may increase. 21

MECHANIS M OF AC TION
The unique effect of indomethacin on certain headache disorders is unknown and may not be the same for each disorder. Aside from the well-known inhibition of cyclooxygenase, in other activities, such as reduction of cerebral blood flow (CBF), the inhibition of nitric oxide (NO) pathways or attenuation of oxidative stress 24 may be important.
This inhibition, which could be easily correlated with its antiinflammatory and antipyretic action, is less likely to explain all of its analgesic activity. 3

REG UL ATION OF INTR ACR ANIAL PRE SSURE AND CEREB R AL B LOOD FLOW
The regulation of intracranial pressure may play an important part in the action of indomethacin on traumatic brain injury, as well as cough, exertional, or sexual headaches, when a peak in intracranial pressure is proposed as the underlying pathophysiology. 33 In experimental models of intracranial hypertension, indomethacin caused a decrease of intracranial pressure, and at higher doses, also CBF, venous pH, and electrocortical activity, along with an increased difference in arteriovenous oxygen. 34 In human studies, indomethacin was able to reduce the CSF opening pressure in patients with idiopathic and posttraumatic intracranial hypertension. 35 Furthermore, it causes a reduction in CBF after intravenous and rectal administration 36 and a decreased mean flow velocity in the middle cerebral artery when measured with transcranial Doppler, 37 without ischemic changes. 38-4 0 Indomethacin-induced CBF decrease was restored with hypoxia and hypercapnia. 41,42 Acetazolamide, a carbonic anhydrase inhibitor, causes extracellular acidosis and an increase in CBF by vasodilation. Compared with diclofenac, another COX inhibitor, indomethacin alone was able to reverse the increment in CBF caused by acetazolamide in rats. 43 This reduction in vessel diameter may, therefore, be mediated by other mechanisms that are not related to the inhibition of prostaglandins.

PROS TAG L ANDINS
The role of prostaglandins has been studied in different nociceptive pathways. The intramuscular, subcutaneous, and intravenous injections of prostaglandins, used in the 1970s to induce abortion, were reported to be extremely painful. 44,45 Other studies have proved nociceptive activity in humans 46 and in animals, 47 and nociceptive behavior has been inhibited with antagonists of prostaglandin receptors, in preclinical studies. 48 The function of prostaglandins in headache is not clear. The levels of prostaglandins were elevated and reached a peak at 2 h after the initiation of a spontaneous migraine attack in migraineurs. 49 Prostaglandins have been shown to be capable of triggering an attack. 50,51 However, following the administration of one antagonist of prostaglandin receptors, neither the headache intensity nor the superficial temporal artery or middle cerebral artery diameter changed. 52 Curiously, salivary levels of prostaglandins are modulated by placebo and nocebo effects, 53 which may be related, to a certain extent, to the benefit seen in some headaches responsive to other cyclooxygenase inhibitors like ibuprofen, where a placebo effect has been seen in up to 50% of patients. 54 Prostaglandins are not the only substances with vasoactive activity released in the trigeminal endings that may be involved in the nociceptive activity of indomethacin. Levels of molecules including nitrates or calcitonin gene-related peptide (CGRP) were found elevated during the first hour of a spontaneous migraine attack, 49,55,56 and a reduction in the levels of CGRP and vasoactive intestinal peptide was seen after treatment with indomethacin in a patient with chronic paroxysmal hemicrania (CPH). 57

NEUROPEP TIDE S
Colocalization of nitric oxide synthase (NOS) and CGRP, which has a crucial role in the trigeminovascular system, has been shown in the trigeminal ganglion. 58 NO may lead to the release of CGRP as shown on a rat model experiment, 59 and in cluster headache patients, nitroglycerin, an NO donor, was able to induce headache attacks with elevated plasma CGRP. 60 The serum levels of kininogen, an anti-inflammatory mediator and precursor of bradykinin, one of the substances known to induce the release of NO from endothelial cells, 61,62 along with histamine, acetylcholine, or N-methyldaspartate (NMDA), 63 were found to be reduced in patients with CPH, 64 which could be translated as an increased production of NO.

NITRIC OXIDE
NO is one of the fundamental molecules involved in the regulation of cerebral metabolic activity and blood flow. 65 The release of NO is responsible for the formation of cyclic guanosine monophosphate (cGMP) by the enzyme guanylate cyclase, from Mg 2+ and guanosine triphosphate. 66 cGMP acts as a second messenger for different biological processes such as the relaxation of smooth muscle or the inhibition of platelets. 66 Increase in cGMP levels has been seen in some brain areas, 63 and an increase in cGMP concentration in the brainstem was seen after cortical spreading depression (CSD) was evoked. 67 The synthesis of NO in the CNS is dependent on the presence of free intracellular concentrations of Ca 2+ , 63,68 and increased concentrations of intracellular Ca 2+ may be one of the stimuli needed for the synthesis of NO and the activation of guanylate cyclase in the brain. 61 The local release of NO has also been associated with the activation of NMDA receptors, which are usually linked to ion channels with high Ca 2+ permeability and sensitivity to Mg 2+ , 63,69 and the prolonged activation of these receptors may lead to neuronal cytotoxicity. 61 NO and prostaglandins may share common pathways, where they act synergistically, 70 and the synthesis of prostanoids may play a role in the duration and maintenance of the headache. 49 79 The spinal expression of nNOS and eNOS was increased following the intradermal administration of capsaicin in a rat model and was appropriately inhibited by specific inhibitors. 80 Likewise, NO may be implicated in neuronal sensitization in patients with migraine, which would be translated clinically as increased cutaneous allodynia. 81 In a preclinical model of trigeminovascular nociception, indomethacin proved to inhibit dural vasodilation elicited by using electrical stimulation or NO donors. However, it was unable to inhibit CGRP-induced dural vasodilation. 82 Histamine, another substance with vasoactive properties and known to cause headache when infused in humans, 83 was also a plausible candidate to be the target of indomethacin acting through NO. However, the infusion of mepyramine, an antagonist of the histamine H1 receptor, was unable to prevent headache induced by nitroglycerine. 84 Indomethacin, as well as sumatriptan, partially blocked the dilation of meningeal blood vessels caused by NO, which was not reproducible with either the histamine receptor antagonists or flunarizine. 85 Notwithstanding, the efficacy of indomethacin may not even be related to its vasoactive properties, as has been the case with other NSAIDs, 86,87 as it is not plausible that vasodilation acts as an exclusive cause of pain in primary headache disorders. 88 The direct implication of indomethacin in the CNS is supported by a recent study using an established animal model of trigeminovascular nociception. This study compared the response of postsynaptic second-order neurons in the trigeminocervical complex (TCC) to indomethacin and naproxen at therapeutic doses, when activated by local micro-iontophoresis of L-glutamate or administration of a NO donor. While naproxen was able to inhibit dural-evoked firing and the firing evoked by l-glutamate, indomethacin was the only drug capable of inhibiting, along with the two previous triggering mechanisms, the firing induced by an NO donor. 89 The nitrergic hypothesis promoted the investigation of the inhibition of the synthesis of NO, which has been tested with general and specific NOS inhibitors. The synthesis of NO was first inhibited by Palmer and colleagues by using l-NG-monomethyl arginine (l-NMMA). 62 Since then, the arginine-based inhibitors of NOS have been tried in headache.
The inhibition of the synthesis of NO in preclinical models, using the unselective NOS inhibitor Ng-nitrol-arginine methyl ester (l-NAME), blocked the associated hyperemia during CSD, but not the neuronal firing, provoked after needle stick injury. 65 When using l-NAME or another nonselective inhibitor, NG-nitrol-arginine (l-NNA) in CSD, pial arteriolar dilation was significantly reduced. [90][91][92] When indomethacin was administered, arterioles were dilated. When administered together with l-NNA, indomethacin abolished the effect of l-NNA on CSD-induced dilation. l-NNA inhibited topical acetylcholine-induced arteriolar dilation, and this effect was not altered by indomethacin. 92

NO synthesis blockers and headache
The first attempt to inhibit nonselectively NOS enzymes with

Indomethacin-responsive headaches
The group of primary headaches known as "indomethacinresponsive" has, paradoxically, only a moderate level of evidence, as indomethacin has not been tested in randomized, controlled trials (  123 and the consecution of this reflex explains the relationship between the pain and the characteristic ipsilateral CAS. 124 Indomethacin was able to inhibit neuronal firing in the TCC evoked by the stimulation of the SSN. 125 The hypothalamus has been hypothesized as a key structure in the generation of headache attacks. 126 This is partly based on its implication in circadian rhythms and the typical periodicity of the most common TAC, cluster headache. 127 Indeed, one of the hypotheses points to increased production of NO by the suprachiasmatic nucleus of the hypothalamus. 128,129 Specifically, the firing of the A11 hypothalamic nucleus was attenuated when an NO donor was infused, and this attenuation was blocked in animals pretreated with indomethacin. 130 Functional neuroimaging studies have shown activation of the ipsilateral hypothalamus in PH 131 and HC, 131 and an activation of the posterior hypothalamic and dorsal rostral pontine areas was seen in imaging studies using positron emission tomography. 132

Paroxysmal hemicrania and hemicrania continua
The first case of PH was described by Sjaastad and Dale in the 1970s, and the extraordinary response to indomethacin has been noted since then. 64,133 The first case of HC was described by Sjaastad and Spierings in 1984 134 and was probably missed as an entity a few years earlier. 135 The clinical presentation of these may overlap and can be difficult to distinguish from other primary headaches. 136 An in-principle issue that needs to be borne in mind when considering these syndromes is how to treat partial indomethacin responders or patients who cannot be exposed for various reasons. We would argue that retaining an absolute requirement for an indomethacin effect is essential to make progress in research in this area to understand indomethacin's effect.
This approach does not materially alter clinical practice.
PH is defined by ICHD-3 as severe and strictly lateralized attacks, occurring at least 5 times a day, and lasting for 2-30 min, associated with either restlessness or CAS and with an absolute response to indomethacin. In the chronic form, attacks occur for more than 1 year or with remission periods lasting less than 3 months. 1 PH has a prevalence of 0.5 per 1000 individuals in the general population.
In contrast with cluster headache, which has a male predominance, PH has been documented more frequently in women between 30 and 40 years of age. 2 Indomethacin can abolish spontaneous attacks of CPH and also those triggered by the NO donor glyceryl trinitrate (GTN), supporting the nitrergic hypothesis. 137 Headache response to indomethacin has been described from 12.5 to 300 mg daily. 138 In a series of 74 patients with CPH, around 75% had a consistent response to indomethacin.
The majority of patients had an average effective daily dose of 75 mg.
Five patients did not respond to maximal doses of 150 mg. 8 Two patients had an initial response to indomethacin, and this effect was lost.
With cessation of the treatment, 9/13 patients could discontinue indomethacin without headache recurrence. 104 However, in a study on PH that involved 31 patients, two thirds experienced side effects. 105 HC is defined as a lateralized headache that is persistent for more than 3 months and is also associated with ipsilateral CAS or restlessness, sometimes migrainous features, and responds absolutely to therapeutic doses of indomethacin. In the remitting subtype, the pain can be interrupted by remission periods that last for at least 24 h, whereas in the unremitting subtype the pain is continuous, without remissions of more than 24 h, for more than 1 year. 1 Because of the differential diagnosis with chronic migraine, the prevalence of HC is probably underestimated. In line with this, a high incidence of HC was demonstrated when reviewing consecutive patients attending an orofacial pain clinic. 139 In a retrospective review of 192 cases using indomethacin 150 mg or more for 2 weeks or 225 mg for 1 week, 43 had an absolute response to indomethacin. In another retrospective analysis of 62 patients with diverse clinical presentations, daily doses of 25-300 mg were used. The response to indomethacin may be delayed up to 4 weeks for some patients 103 and an incomplete relief of the symptoms is frequent, although a differential diagnosis with lateralized migraine in such cases remains possible. 140 Trials with either oral or parenteral doses of indomethacin, the so-called "Indotest," are encouraged in all patients presenting with lateralized headache. 102,141,142 The interval between indomethacin discontinuation and pain reappearance ranges from 4 to 28 h. 142 Long-term treatment with indomethacin is safe and generally well tolerated in most patients, with no major adverse events. With time the dose can be reduced for better longer term tolerability. 18 Regarding the headache-associated features, nausea, dizziness, and photophobia or photophobia, typically associated with migraine, have been documented in a small proportion of patients. 108,109,144,145 Vertigo, ataxia, and syncope have been reported as potential red flags that point to a possible secondary cause. 109 The differential diagnosis with secondary or symptomatic cough headache is important. The most common reported cause of secondary cough headache is Chiari malformation type 1. 145

PRIMARY E XERCIS E HE ADACHE
The prevalence of primary exercise headache ranges from 1% to 30% 150,151 without a clear gender predisposition. 108,150,[152][153][154] The highest prevalence was found in adolescents. 150 Onset of head pain occurs during or within 30 min of exercise termination 1,153,155 and is typically a pulsating unilateral or bilateral frontal headache. 150,153,155 Headache duration is typically between 1 min and 1 day although rarely it can last up to 2 days. 152,153,155 Associated features, typically associated with migraine such as nausea, vomiting, photophobia, and phonophobia, can be associated with the headache of primary exercise headache and are more frequently reported in females. 150,156 Most patients undergo spontaneous resolution within 1 year of onset. 152 Different hypotheses have been proposed for the pathophysiology of headache related to exercise, and the majority involve a change in intracranial pressure. 33 First-line management of primary exercise headache is modification of exercise for prevention. In cases where this is not effective or ideal, indomethacin 25-150 mg daily is effective in the prevention or amelioration of primary exercise headache in over 80% of patients. 110,157,158 We have recently used gepants, CGRP receptor antagonists, as an unlicensed pretreatment approach and found them useful.

PRIMARY HE ADACHE A SSOCIATED WITH S E XUAL AC TIVIT Y
Despite being roughly described by Hippocrates as headache related to "immoderate venery", it was not until the early 1970s when that the first series of patients were reported. [159][160][161][162] Over time, three different types of headaches have been described, 108 although the distinction is no longer in vogue. 1 The so-called type 1 was characteristically defined as tension that occurred and progressed with sexual excitement. 159 On the other hand, type 2 was described as sudden, explosive, and would occur close to or with orgasm, being called orgasmic headache. 159 Currently, either one of the two or both types are accepted in the classification and included under the same nomenclature. 1 The headache can last with severe intensity from 1 min to 24 h. The pathophysiology of "type 1" has been related to muscle tension, particularly involving the jaw muscles.
Lance hypothesized type 2 was due to a hyperdynamic circulatory state, and this has not been refuted to date. 159 Indeed, a relationship between this and exertional headache, currently known as primary exercise headache, was suspected, and the coexistence of both was documented in many patients. 111 The coexistence has not been our experience; indeed headache associated with sexual activity in our experience almost never occurs with cough or exertional headache.
Hemodynamic changes leading to stretching of certain intracranial structures were hypothesized for both headaches. 163 The third type of headache, no longer considered sexual headache, appears after sexual intercourse and resembles that of spontaneous intracranial hypotension and is attributed to a CSF leak. 1 Limited noncontrolled data on the best treatment exist. As a preventive therapy, beta-blockers were efficacious in 15 out of 18 patients 111 and have been recommended as a first-line drug. 109 Regarding indomethacin 25-50 mg, this showed efficacy as a preventive in 9 out of 10 patients, taken 30-60 min prior to intercourse.
The same dose of indomethacin has also shown efficacy as an acute medication. Interestingly, this contrasts with a lack of efficacy of other NSAIDs pointing to a different mechanism of action. Triptans have also demonstrated utility as acute medications in some cases. 111 Again, we have recently used gepants, CGRP receptor antagonists, as an unlicensed pretreatment approach and found them useful. Our experience is that diltiazem can be useful for persistent headache associated with sexual activity.

PRIMARY S TAB B ING HE ADACHE
In comparison with the other primary indomethacin-sensitive headache disorders described herein, the prevalence of primary stabbing headache is dramatically greater in approximately one third of the general population, 177 although this finding has not been reproduced in clinic-based studies. 116,[178][179][180] There is a female predominance 180 where the female-to-male ratio in population-based studies is 1.49:1. 181 The mean age of onset, in adult populations, ranges from 28 to 53 years of age with eldest reported age of onset at 83. 114,177 The mean range of onset in pediatric cases range between 4.5 and 12 years of age with cases reported in children as young as 1.5 years of age. [182][183][184][185] In our clinical experience, primary stabbing headache can be exceptionally disabling in children in a manner not often seen in adults.
In approximately 70% of cases, the pain is experienced at extratrigeminal locations and can be either unilateral or bilateral. 114,178 Attacks typically range from 0 to 3 s without discernable regularity 114,177 although attacks lasting up to 120 s have been reported but are rare. 177 These are usually not associated with CAS, and typical migraine features of photo sensitivity, phonosensitivity, and nausea are infrequent. 114,115 Approximately two thirds of patients with primary stabbing headache will have a partial or complete response to indomethacin with doses between 50 and 75 mg. 114,115 In a smaller cohort study, up to 70% have a partial response, and 20% had complete remission with 120 mg of indomethacin daily. 116 Indomethacin 50 mg three times daily in a small study of five patients resulted in a significant improvement. 158

HYPNIC HE ADACHE
In 1988, Raskin described a headache distinct to cluster that also occurred with clockwork regularity and was responsive to lithium. 173 Hypnic headache is a rare primary headache disorder with a prevalence between 0.07% and 0.1% of the population reviewed for headache. 174,175 Presenting patients typically present in their fifth decade of life with an average age of 58. Hypnic headache is most frequently described in the adult population with an age range between 18 and 85. 113 However, five pediatric cases have been described. [176][177][178] It is more common in females with a ratio of 2:1. 113,179 Attacks most commonly occur between the hours of 2 and 4 a.m. and last for an average of 90 min. 113 The pain experienced is typically dull and in over half of patients bilateral. 113,179 Migrainous symptoms such as nausea, photosensitivity, and phonosensitivity are reported in less than a third of patients. 113,179 CAS such as lacrimation, ptosis, or rhinorrhea are reported in less than 10% of patients. 113,179 The pathophysiology of hypnic headache may be related to a degenerative dysfunction of the hypothalamic-pineal axis. This system involves the connections between the midbrain periaqueductal gray matter, the noradrenergic locus coeruleus, and serotonergic raphe nuclei with the suprachiasmatic hypothalamic nuclei, which may influence the release of melatonin. 179 In fact, melatonin is a molecule with a structure analogous to that of indomethacin, 180 and a comparable clinical efficacy has been reported. 113,181 Interestingly, melatonin was higher when comparing serum levels of nine hypnic headache patients and controls at different times, especially during the headache attack, although the differences did not reach statistical significance. 182 Moreover, structural changes associated with this axis have been described. Using voxel-based morphometry in magnetic resonance images of patients with hypnic headache, a decrease in gray matter was observed in the periaqueductal area, as well as other cortical areas related to nociceptive processing. 183 Hypnic headache does not seem to be related to the sleep stage, according to polysomnography studies, 184 and it may be related to interrupted rapid eye movements (REM) sleep or hypoarousals due to a higher cyclic alternating pattern. 185 Indomethacin has been proposed as the third-line prophylactic treatment after caffeine and lithium. 186 A number of case reports and case series have described partial to complete relief or prevention of attacks with the use of oral indomethacin, 187,[192][193][194][195][196][197][198] and doses between 25 and 150 mg orally before bed have been reported to be effective. In one case series, indomethacin has been proposed to be more efficacious in patients who only experience unilateral pain. 198 Paradoxically, some patients who report resolution of nocturnal headache with indomethacin experience diurnal attacks, which resolve with cessation of indomethacin.

MIG R AINE
Migraine is not considered as an "indomethacin-responsive" headache. 8 In comparison with other NSAIDs-aspirin, diclofenac, ibuprofen, and naproxen-there is a paucity of evidence to support the use of indomethacin for acute treatment of migraine. 199 More recently, indomethacin in combination with prochlorperazine and caffeine has been compared with sumatriptan in two trials. The first study compared a combination of indomethacin 25 mg, prochlorperazine 4 mg, and caffeine 75 mg rectally to rectal sumatriptan 25 mg. 200 The combination was superior to sumatriptan in aborting attacks. 200 There was no difference in pain reduction between the combination and sumatriptan. 200 In the second trial, indomethacin 25 mg, prochlorperazine 2 mg, and caffeine 75 mg orally was compared against oral 50 mg sumatriptan. 201 In this study, there was no difference between the combination and sumatriptan in reducing pain intensity or successful abortion of a migraine attack 2 h after treatment. 201

S ECONDARY HE ADACHE S
Within the category of secondary headaches, we find several diagnoses whose pathophysiology is related to an increase of intracranial pressure, such as those attributed to craniotomy, idiopathic intracranial hypertension, or traumatic brain injury. 1 Severe intracranial hypertension following a headtrauma is one of the main predictors of poor outcome. 40 Indomethacin causes a lasting decrease in intracranial pressure, as well as in CBF when administered intravenously to patients with secondary intracranial hypertension. The effects of indomethacin 30 mg bolus followed by 30 mg/h for 7 h were compared with those of hyperventilation in patients with severe traumatic brain injury, showing a decrease in intracranial pressure (ICP) similar to that of a decrease in PaCO 2 of 6.6 mmHg, 39,202 although not all the patients responded to both mechanisms, suggesting an independent action.
Higher doses (50 mg bolus) reduced intracranial hypertension. 203 This also improved cerebral perfusion pressure in patients with severe traumatic brain injury, but this change was not significant with a prolonged infusion. The perioperative infusion of indomethacin 50 mg also decreased intracranial pressure in patients intervened for supratentorial tumors. 38

New approaches to indomethacin-sensitive headache disorders
Although the possible link is ill-understood, recent evidence suggests indomethacin-responsive headaches may also respond to noninvasive vagal nerve stimulation (nVNS). 204 The nVNS was approved for the prevention of cluster headache following two randomized controlled trials. 208,209 Similar studies are lacking on indomethacin-responsive headaches. However, a clinical audit in 15 patients with HC or PH showed efficacy with at least a reduction in headache intensity in 11/15. nVNS was similarly useful for both conditions, with efficacy observed in seven of nine patients with HC and four of six patients with PH. 210 Based on this potential link, a patient with indomethacin-responsive cough headache who had a contraindication for indomethacin was effectively treated with nVNS. 211

CON CLUS IONS
The only headache disorders that include responsiveness to indomethacin as part of the diagnostic criteria are HC and PH. 1 Yet, other headache types described herein have been effectively treated with indomethacin. The mechanism behind this distinctive response, as previously discussed, has not been fully elucidated and could be linked to a direct effect on the CNS.
Indomethacin, albeit generally well tolerated, has important side effects, particularly gastrointestinal intolerance, and CNS side effects, and remarkably can induce headache of a generalized type in patients who also have a personal or family history of migraine. A better understanding of the mechanism of action of indomethacin will lead to more specific and better tolerated treatments for rare yet highly disabling disorders.