Pathophysiology of hypnic headache

Abstract

Background

Hypnic headache (HH) is a rare primary headache disorder that is characterized by strictly sleep related headache attacks.

Purpose

The underlying pathophysiology of HH is mainly enigmatic but some clinical characteristics such as circadian rhythmicity and caffeine responsiveness may point toward possible underlying mechanisms.

Method

Current studies that deal with the pathophysiology of HH are summarized. Data on cerebral imaging, sleep, electrophysiology studies, effectiveness of drugs, and symptomatic headache types are discussed to illuminate underlying pathophysiologic mechanisms.

Conclusion

HH can be clearly differentiated by its clinical presentation as well as imaging and electrophysiological study results from other primary headaches such as migraine or cluster headache. The underlying pathophysiology is still enigmatic but a hypothalamic involvement seems to be likely.

Introduction

Hypnic headache (HH) is primary headache disorder characterized by recurrent strictly sleep related headache attacks causing waking of patients. The headache usually occurs at the same time at night. Therefore, HH headache is sometimes referred to as alarm-clock-headache. According to the new ICHD-3 beta criteria, HH attacks have to occur on at least 10 days a month and last for 15 minutes up to 4 hours after waking (Table 1) (1). Many patients report some motor activity during the headache attacks such as reading or eating. Restlessness, a characteristic feature in cluster headache, is not observed in HH. To date, about 250 cases have been published in the literature (2). More women than men are affected by HH. In contrast with most other primary headache disorders, this headache entity usually starts later in life, with most patients over the age of 60 years. However, some younger patients and even children have been reported (3–5). Most patients report a moderate pain intensity, the pain can be localized bilateral but also unilateral. Vegetative symptoms such as nausea, phono- and photophobia are common (about 30%). About a third of HH patients have an additional history of migraine; however, most had their last migraine attacks years before the HH started. The underlying pathophysiology of HH is still rather enigmatic. This review summarizes the current knowledge regarding pathophysiologic mechanism in this headache disorder.

Table 1.

Diagnostic criteria of HH according to ICHD-3 beta.

A	Recurrent headache attacks fulfilling criteria B–E
B	Developing only during sleep, and causing wakening
C	Occurring in ≥10 days per month for >3 months
D	Lasting ≥15 minutes and for up to 4 hours after waking
E	No cranial autonomic symptoms or restlessness
F	Not better accounted for by another ICHD-3 diagnosis

Influences of the hypothalamus

The hypothalamus is considered to be a major integration center, regulating the nervous and endocrine system. It has mutual influence on pain control and sleep regulation because of its strong connections to the periaqueductal gray, locus coeruleus, and the median raphe nuclei (6,7). The trigeminohypothalamic tract connecting the posterior hypothalamus and trigeminal nucleus caudalis might be the anatomic correlate of hypothalamic influence on trigeminal pain processing (8). As of the circadian rhythmicity of HH with strict sleep related headache attacks always at the same time of night, a hypothalamic involvement is obtrusive. One study addressed the question of structural cerebral changes in HH patients that can be detected using Voxel-based morphometry (9). Brains of 14 patients with HH and 14 age- and gender-matched healthy controls were investigated. Similar to other chronic pain disorders (e.g. phantom limb pain (10,11), chronic back pain (12,13), fibromyalgia (14,15), neuropathic pain (16,17), chronic posttraumatic headache (18), migraine (19–21), and tension type headache (22)), a decrease of gray matter (GM) could be observed in areas of cortical pain processing (e.g. cingulate cortex, frontal lobe, inferior temporal gyrus). Interestingly, a significant GM decrease was also detected within the posterior hypothalamic area. In cluster headache (CH) a VBM study showed a gray matter increase in the posterior hypothalamus suggesting a role for this anatomic structure in sleep-related headache disorders (23). However, further imaging studies could not confirm the GM alterations in CH (24). The pathophysiologic relevance of the observed hypothalamic GM decrease in HH is still unclear. It might be cause but also consequence of the ongoing HH. As mainly elderly patients are affected, a neurodegenerative pathomechanism might be suspected. In HH only a left-sided GM decrease was detected. The relevance and meaning of this lateralization is still unclear. Functional MRI in other headache disorders during headache attacks showed also lateralization (25–29). However, there was no clear connection between headache side and activation side of the hypothalamus. The observed lateralization in HH may simply be based on a technical issue. A lower significance level would probably reveal bilateral changes. Another hypothesis suggested that the posterior hypothalamus might not be organized somatotopically, which would be an explanation for the observed lateralization (9).

What role does trigeminal facilitation and habituation play in hypnic headache?

Central facilitation or sensitization of trigeminal processing was described in various headache disorders such as chronic migraine (30), chronic tension type headache (31), and medication overuse headache (32,33), as well as during acute migraine attacks (34,35). A lack of habituation is a characteristic feature of primary headache disorders and especially migraine (36), but also other chronic pain conditions such as chronic back pain (37) or fibromyalgia (38). It was even suggested that a habituation deficit might be an endophenotypic marker for presymptomatic migraine in unaffected relatives of migraineurs (39). Interestingly, a habituation deficit can also be observed in other chronic conditions without pain such as tinnitus (40), schizophrenia (41), as well as Parkinson’s disease (42). In this context, a ‘unifying thesis’ was proposed, which states that neuronal excitability changes only reflect chronicity itself and not distinct diseases (43). In line with this hypothesis, one would assume that a chronic headache disorder such as HH should display a distinct overexcitability and habituation deficit. Only one electrophysiological study investigated central facilitation of trigeminal pain processing and habituation pattern in 15 patients with HH compared with healthy controls using the nociceptive blink reflex (nBR) and trigeminal pain related evoked potentials (PREP) (44). This study did not show any central facilitation and/or habituation deficit in these patients. It could be concluded that chronicity itself might not be the only predictor for electrophysiologically detectable cerebral over-excitability. These electrophysiological results underscored that HH is a primary headache disorder that can be clearly distinguished from other primary headache disorders not only based on clinical presentation, but also electrophysiologically measurable pattern on cerebral facilitation and habituation.

Is hypnic headache a REM-sleep disorder?

The occurrence of strictly sleep related headache attacks is a pathognomonic feature of HH. Early on it was suggested that HH attacks might be strictly occurring REM (rapid eye movement)-sleep associated. Initial case reports appeared to confirm this hypothesis suggesting HH to be a REM-sleep disorder (45–51). Newer data rebut this assumption, showing that the majority of HH attacks arise from NREM (non-REM)-sleep stages, mainly sleep stage 2 (for review (2)). Closer analysis did not reveal a REM or NREM sleep subtype of HH as both headache attacks could be detected in the same patient within the same night (52). However, the pathophysiologic correlate of HH might rather lie within microstructural changes of sleep. Capuano et al. analyzed the cyclic alternating pattern (CAP) in a single HH patient over 12 consecutive nights (53). CAP reflects sleep disturbance and drug influence as well as subjective quality of sleep. After successful treatment with amitriptyline the CAP rate increased, suggesting that nocturnal hypoarousal might be involved in the pathophysiology of HH. Similar patterns can also be observed in migraine(54–56).

Why do hypnic headache patients drink coffee?

The good therapeutic response to caffeine is a characteristic clinical feature in HH, and may provide insight into the pathophysiology of the disease itself. Drinking a cup of coffee is the firstline treatment recommendation for prophylactic as well as acute treatment (2,57). Typically, patients figure out this treatment option themselves, before they are diagnosed properly. Many of those report a good therapeutic effect of caffeine containing analgesics but not pure analgesic. There are different mechanisms of caffeine in pain regulation. Preclinical data and animal experiments showed antinociceptive effects at doses between 35 and 100 mg kg⁻¹ (for review (58)). These antinociceptive properties seem to result from the interaction of caffeine with adenosine-receptors (A_2A and A_2B) mainly on peripheral sites. One might hypothesize that HH patients have a different adenosine receptor profile, but study data in this regard are missing as yet. This should be addressed urgently. Another potential effect of caffeine in HH is its influence on sleep. Through cerebral A_2A receptors, caffeine influences the key signaling pathway of sleep induction (59). The caffeine induced wakefulness is mainly based on antagonism at this receptor. However, HH patients do not report pronounced sleeping difficulties after drinking coffee before going to bed or during headache attacks. Whether this is a clinical feature unique to HH patients remains unknown.

Do hypnic headache patients suffer from obstructive sleep apnea syndrome (OSAS)?

Many polysomnographic data in HH suggest a high prevalence of obstructive sleep apnea syndrome (OSAS) in HH patients. However, the observed oxygen desaturations in almost all reported cases are not temporally correlated to the headache attacks and headache and oxygen desaturation occurs quite randomly (46,47,52,60–62). The observed high prevalence is rather biased by the higher age of affected patients but not the disease itself. Age-matched controls show similar prevalence rates of OSAS in up to 80% of non-HH patients (63). OSAS should be ruled out only in HH patients who suffer additionally from other symptoms such as excessive daytime sleepiness, unrefreshing sleep, fatigue, and unintentional sleep episodes during wakefulness. Only in two patients could HH headache attacks be treated efficiently by improving severe OSAS (50,64). Therefore, OSAS seems not to be the driving force in the underlying pathophysiology of HH.

Is hypnic headache an indomethacin-sensitive headache?

In about 70% of patients with HH, indomethacin at a dose between 25 and 150 mg/d has shown good therapeutic response in prophylactic treatment (for review (65)). Indomethacin is a non-steroidal anti-inflammatory drug (NSAID) and has potent inhibitory effects on synthesis of prostaglandins. Some primary headache entities, that is paroxysmal hemicrania and hemicrania continua, are characterized by their marked good response to indomethacin, which is required for diagnosis by the ICDH criteria (1). Some other headache disorders also show good therapeutic response to indomethacin including primary stabbing headache, primary cough headache, and primary exertional headache (66). Indomethacin has some distinct properties that differentiate it from other NSAIDs, which might explain why indomethacin-responsive headaches exist (for review (65)). Indomethacin crosses the blood brain barrier better than naproxen and ibuprofen (66), suggesting a potential central site of action in headache treatment (67). In animal experiments, Summ et al. detected indomethacin specific inhibition of NO (nitric oxide) induced dural vasodilation. In contrast, naproxen and ibuprofen did not reduce NO dependent dural vasodilation. NO is known to have a crucial role in headache pathophysiology. NO donors can be given to induce migraine (68,69) as well as cluster headache attacks (70). These NO-induced migraine-like and cluster-like headaches occur with delay of minutes or hours. In contrast, in paroxysmal hemicrania NO-induced headache attacks occur immediately (71); however, data on HH in this regard are missing as yet. This observation supports the assumption that direct and indirect (e.g. via second messenger) mechanisms may be involved in the effects of NO in headache (65), and particular subtypes of these mechanisms might differentiate several primary headache disorders, that is the indomethacin-sensitive headaches such as HH and other headache entities from migraine and cluster headache.

Symptomatic hypnic headache: hints to pathophysiology?

Several symptomatic HH cases have been described in the literature. In particular, brain tumors (i.e. hemangioblastoma of the cerebellum (72), nonfunctioning pituitary macroadenoma (73), growth hormone secreting pituitary tumor (74), posterior fossa meningioma (75), and brain stem lesions (76)) have been observed mimicking HH. Therefore, cerebral imaging should be performed in all patients presenting with symptoms of HH to rule out a symptomatic subtype. Additionally, nocturnal arterial hypertension is associated with HH-like headache attacks and can be treated efficiently with antihypertensive agents in some patients (77,78). However, in most patients with HH who suffer from high blood pressure, this is well-controlled and treatment did not lead to attack termination. Therefore, nocturnal arterial hypertension cannot be assumed to be the underlying pathophysiologic correlate of HH.

More interestingly, transient HH symptoms were observed after withdrawal of long-term lithium treatment (79), which is also effective in treatment of this disorder (2). Lithium is known to influence the human circadian clock by increasing the expression of transcription factor BMAL1, resulting in a disruption of the natural cycle (80). It was proposed that this mechanism might help to restore healthy circadian rhythms in affected patients. Additionally, it was hypothesized that lithium may interact with NO and N-methyl-D aspartate (NMDA) receptor signaling in the cerebral nociceptive processing system (81). However, the exact underlying physiological mechanisms of lithium efficacy in HH patients are still unknown.

Is hypnic headache a subtype of migraine?

Thirty-six percent of patients suffering from HH report an ongoing migraine or a migraine in the previous medical history (for review (2)). In many patients migraine stopped before HH attacks started. It was hypothesized that HH might be a subtype of migraine. However, the different clinical presentation with its strict sleep related headache attacks, as well as electrophysiological and imaging data supports the assumption that HH is a distinct primary headache disorder, as mentioned by the IHS criteria. The high prevalence of migraine in HH patients remains unclear for the time being, but a common pathophysiologic pathway can be expected.

What future studies are needed regarding the pathophysiology of hypnic headache?

Understanding the pathophysiology of HH will be essential to improving its treatment. Further polysomnographic studies should be conducted to investigate the role of microstructural changes in sleep pattern in these patients as strict sleep dependency is the pathognomonic feature in HH. Longitudinal imaging studies would help to illuminate further a possible neurodegenerative nature of this disease. The role of hormones known to be involved in the sleep cycle, such as melatonin and oxytoxin, is still unknown and would be an interesting target for further HH research.

Clinical implications

HH is a unique primary headache disorder, with clinical presentation as well as underlying pathophysiology that clearly distinguishes it from other primary headache disorders.

As for the circadian rhythmicity and strict sleep association of headache attacks a hypothalamic involvement seems logical, and is supported by imaging data.

Therapeutic response to indomethacin, as well as caffeine, is a characteristic feature of HH.

HH is not a REM sleep disorder, as headache attacks can also arise from NREM sleep stages in the same patient even in the same night.

Trigeminal facilitation and changes in habituation pattern do not seem to be pathophysiologic correlates of HH.