Serial polysomnography in hypnic headache

Abstract

Background: Hypnic headache (HH) is a rare primary headache disorder characterized by strictly sleep-related headache attacks. Most patients are over the age of 50 and usually awake at the same time at night with dull bilateral head pain. The pathophysiology of this headache disorder is still enigmatic but association with rapid eye movement (REM) sleep and sleep-disordered breathing (SDB) has been suggested.

Methods: Six patients with HH according to the current International Classification of Headache Disorders (ICHD-II) criteria (code 4.5) were investigated. Serial polysomnography (PSG) was performed in each patient for four consecutive nights.

Results: A total of 22 HH attacks were recorded from all patients during PSG. Six of the monitored headache attacks arose from REM sleep; 16 attacks, however, arose from different non-REM (NREM) sleep stages. Five patients showed an increased apnoea/hypopnoea index (>5), indicating obstructive sleep apnoea (OSA) on some but not the majority of nights. Headache onset and occurrence of SDB were not temporally connected.

Conclusions: This prospective study shows that the onset of HH was not associated with sleep stage. These results contradict the current belief that REM sleep and SDB play a crucial role in the pathophysiology of HH.

Introduction

Hypnic headache (HH) is a rare primary headache disorder characterized by exclusively sleep-related headache attacks, which clearly distinguishes this headache entity from other primary headache disorders. Most of the patients are over the age of 50 and awake with dull headache, usually at the same time at night. Therefore, this headache entity is also referred to as “alarm clock headache”(1). HH pathophysiology remains obscure, but rapid eye movement (REM) sleep and sleep-disordered breathing (SDB) were suggested to be headache-triggering factors.

The current literature reports a total of 36 polysomnographically (PSG) recorded HH attacks so far, most of them single case and single night studies without longer observation periods. In these studies, the majority of monitored attacks arose from REM sleep (2–4) suggesting that HH is mainly driven by the REM sleep. This hypothesis was supported by the clinical observation that many patients describe vivid dreams before awakening with HH. Possible REM influence was explained by absent dorsal raphe and locus coeruleus nuclei activity during this sleep state leading to a reduction of antinociceptive effects (5). However, recent reports suggest that headache attacks might also arise from NREM sleep stages (2,6). Obstructive respiratory events leading to decreased nocturnal oxygen saturation were monitored in several PSG studies (2–4,6,7) supporting the assumption that SDB plays a crucial role in the pathophysiology of HH.

Here we report serial PSG findings in six HH patients to determine the temporal association of headache and REM sleep and SDB over four consecutive nights.

Patients and methods

Patients

Six patients suffering from HH according to the ICHD-II criteria (code 4.5) (1) were investigated. No patient did take a HH related medication while participating in this study. One patient took l-thyroxin because of hypothyroidism; two patients had a rheumatoid arthritis and occasionally used non-steroidal anti-inflammatory drugs.

PSG sleep study

Each patient underwent complete overnight PSG monitoring for four consecutive nights using the Embla-System N 7000, Medcare (REMBrandt Manager 7.5) to evaluate the relationship of nocturnal headache attacks to distinct sleep stages, as well as to any occurrence of SDB apnoea. PSG was performed between 22.00 and 07.00 hours, including two-channel electroencephalography, electrooculography, chin electromyography (EMG), electrocardiography, thoracic and abdominal respiratory efforts measured by impedance plethysmography, body position monitored by a position sensor, oxygen saturation measured by pulsoxymetry (ResMedModel 305A, San Diego, CA, USA), surface EMG of tibialis muscles and oronasal airflow recorded by nasal cannula. During PSG, patients were monitored by infrared video surveillance. Patients were instructed to sleep as “normally” as possible. Sleep stages were determined visually in accordance to the standard criteria (8). Arousals were scored according to the criteria of the American Sleep Disorder Association (9).

Apnoeas and hypopnoea were defined according to the conventional criteria (American Sleep Disorder Association). Apnoea was defined as cessation of airflow or reduction of the thermistor signal <10% of the normal flow for ≥10 seconds, hypopnoea as an observable reduction of at least 10 seconds’ duration followed either by arousal or a desaturation of 4% or more. Events were classified as obstructive or central according to the respiratory effort channels.

The apnoea/hypopnoea index (AHI) was used for quantification of sleep apnoea (number of apnoeas and hypopnoeas per hour of sleep) in accordance to the American Academy of Sleep Medicine Task Force (9). Sleep apnoea was considered mild with AHI between 5 and 15, moderate between 15 and 30 events per hour, and severe if AHI was >30. Oxygen indices were calculated by software from the oxygen saturation (SaO₂ curve with minimal SaO₂ being the lowest saturation reached during sleep with average minimal SaO₂ being the mean of all saturation values reached during all respiratory events). Clear oxygen saturation artefacts were excluded manually prior to analysis.

In case of headache the patients were instructed to inform the attending nurse as well as record the exact time of awakening and the headache features in a standardized questionnaire.

The protocol of this study was reviewed and approved by the Ethics Committee of the Faculty of Medicine, University of Duisburg-Essen, and written informed consent was obtained from all participants prior to study inclusion.

Statistical analysis

Comparison of percentage of HH attacks occurring from REM sleep stage and proportion of REM sleep of total sleep time (TST) was performed by Chi-square test. Level of significance was set to p < .05. Statistics were calculated with SPSS 16 (SPSS, Inc., Chicago, IL, USA).

Results

Individual patients’ characteristics are displayed in Table 1, summarized PSG results are shown in Table 2. Twenty-two headache attacks were recorded; six arose from REM sleep (27%), and 16 from NREM sleep stages (73%) (stage 1 = 0, stage 2 = 12, stage 3 = 2, stage 4 = 2). The proportion of observed REM sleep of TST was 21 ± 5%, which is in line with previous PSG studies in healthy elderly people (10,11). Chi-square test revealed no significant differences between the percentage of REM HH attacks and REM TST (29 % vs. 21 %; χ²= 0.987, degrees of freedom [df] = 1, p value = .321), suggesting that there is no or at least no strong association between REM sleep and HH attacks.

Table 1.

Clinical characteristics of six hypnic headache patients

Patient no.	1	2	3	4	5	6
Gender	W	W	W	W	W	M
Age at study inclusion (years)	63	44	66	63	69	55
Age at onset (years)	56	29	47	55	59	51
BMI (kg/m2)	21	23	24	26	27	21
Duration of attacks (min)	60–180	60–90	180	60–240	240	60–180
Frequency of attacks/month	14	25	18	10	28	31
Frequency of attacks/night	1–2	1–2	1	2	1	1–2
Localization	Bilateral	Bilateral	Bilateral	Right	Right	Bilateral
Pain quality	Dull	Dull	Dull	Dull	Dull	Dull
Headache intensity (NRS)	8	6	8	8	10	4
Photophobia	–	–	–	–	–	–
Phonophobia	–	–	–	–	–	–
Nausea	–	–	–	Yes	–	–
Vomiting	–	–	–	–	–	–
Trigemino autonomic symptoms	–	–	–	–	–	–

W, woman; M, man; BMI, body mass index; NRS, numeric rating scale.

Table 2.

Sleep characteristics of nights with hypnic headache

	N (patients) or mean ± SD [range]
Total sleep time (TST) [min]	370 ± 34 [330–416]
Total wake time (TWT) [min]	71 ± 30 [28–106]
Sleep latency [min]	13 ± 3 [9–18]
Sleep stage measures, % of total sleep time
Stage 1	7 ± 3 [3–13]
Stage 2	48 ± 9 [34–66]
Stage 3	14 ± 3 [7–18]
Stage 4	10 ± 8 [1–27]
REM	21 ± 5 [10–30]
AHI index	10.9 ± 6.1 [1.4–17.1]
Nadir SaO2	87.7 ± 2.6 [84.0–90.5]
Arousal index	20.2 ± 5.9 [12.2–24.1]
Sleep efficiency	75.0 ± 8.7 [64.0–87.1]

SD, standard deviation; AHI index, apnea/hypopnea index; REM, rapid eye movement; SaO₂, oxygen saturation.

No patient showed exclusive REM sleep–associated headache attacks across the four investigated nights. Serial polysomnography of one example patient is shown in Figure 1. Wake-up times due to headache and corresponding sleep stages, as well as corresponding AHI indices, are summarized in Table 3 (please also see Supplementary Table 1 for individual data). Five patients showed signs of SDB (AHI > 5), but no temporal relationship of SDB with headache occurrence could be established in any of the patients.

Figure 1.

Serial polysomnography of one example patient showing REM and NREM sleep associated HH attacks of four consecutive nights (A–D). Hypnogram of patient 2 is reported as example. A clear association with a distinct sleep stage cannot be observed. REM and NREM sleep stages precede the headache attack in different nights (A–C) as well as in one night (D). Arrows indicate the wake-up time of this patient with typical headache. On the left, sleep stages are shown. (REM, rapid eye movement sleep, 1–4; NREM sleep stages 1–4). Horizontal bar indicates the time of night. W, waking; MT, movement time; N, not in bed time.

Table 3.

Association of sleep stages and time of headache occurrence in six HH patients

Patient	Gender/age, [years]	No. of night	No. of headache attacks	Sleep stage before headache awakening	Time	AHI
1	W/63	1	1	NREM 2	00 : 30	32.7
		2	1	NREM 2	01 : 00	6.3
		3	1	REM	04 : 40	7.9
		4	2	NREM 3	22 : 20	1.3
				NREM 2	01 : 40
2	W/44	1	2	NREM 2	01 : 10	0.2
				NREM 3	04 : 50
		2	1	REM	01 : 40	0.9
		3	1	NREM 2	00 : 10	1.6
		4	2	NREM 4	02 : 50	2.7
				REM	04 : 45
3	W/66	1	1	NREM 4	01 : 30	9.9
4	W/63	2	1	NREM 2	02 : 30	27.2
		3	1	NREM 2	04 : 20	6.3
5	W/69	3	1	NREM 2	00 : 30	17.1
		1	2	NREM 2	01 : 45	7.6
6	M/55			REM	03 : 30
		2	2	NREM 2	02 : 00	8.0
				NREM 2	03 : 30
		3	1	NREM 2	01 : 45	9.7
		4	2	REM	01 : 30	6.5
				REM	03 : 00

W, woman; M, man; REM, rapid eye movement; NREM, non-REM; AHI, apnoea/hypopnoea index.

Discussion

We report 22 HH attacks in six patients each monitored by PSG during four consecutive nights. Six attacks occurred during REM sleep, and 16 during NREM sleep stages (predominantly stage 2). As there was 21% REM sleep of TST (predicting 5 ± 1/22 HH attacks occurring from REM sleep), our data do not support a REM association of HH attacks and point to a random occurence. These results contradict most of the previous studies that suggested REM sleep association of HH (3,4) but support more recent reports that HH attacks may not depend on REM sleep. Manni et al. recorded six HH attacks, four of them occurring from NREM sleep stages (2). Liang et al. monitored 12 HH attacks in 11 Taiwanese patients, 50% of them arising from NREM sleep stages (6). However, in all but one patient just one type of attack, either REM or NREM associated, was observed. With regard to this observation, it was suggested that there might be two subtypes of HH, one occurring from REM sleep and a NREM variant. Our longitudinal data clearly demonstrate that different sleep stages can precede HH within one single patient and even in one single night, and that HH attacks are in fact not dependent on REM sleep at all. The high prevalence of sleep stage 2 prior to arising HH attacks most likely results from the predominance of this sleep stage in TST (48 ± 9%) and probably just reflects statistic probability.

With regard to headache onset, one major problem has to be addressed. It is not possible to determine the exact time of headache onset because headache starts during sleep. Only the wake-up time of the patient with already-existing headache can be identified accurately. To what extent the sleep stage before waking up displays the true initiation stage of the headache itself can only be presumed. Nevertheless, the recorded wake-up times in our patients who awoke from NREM sleep with the HH attack are sometimes hours away from their last REM sleep phase. This makes an underlying connection unlikely.

Five patients (83%) showed at least some form of SDB in its main subtype, obstructive sleep apnoea (OSA) (AHI index > 5), but no temporal correlation of headache onset and decreased SaO₂ was detected in any of the patients. Additionally, none of the patients complained about clinical symptoms of OSA, like hypersomnolence, fatigue, impaired concentration or increased daytime sleepiness. Our results are in line with previous data of Liang et al., who detected SDB in 73% of HH patients, but without temporal relationship to headache occurrence (6). Temporal connection of SDB and HH attack was just established in one patient (7). Therefore, observed SDB might simply be based on the higher age of this patient group, as an AHI > 5 can be detected in about 80% of healthy persons over the age of 60 years (12).

Our data indicate that REM sleep and SDB cannot be considered to be the main triggers of HH. Previous pathophysiological concepts of HH have to be reviewed in light of these new results.