Decreased CSF Histamine in Narcolepsy With and Without Low CSF Hypocretin-1 in Comparison to Healthy Controls

Abstract

Study Objective:

To examine whether cerebrospinal fluid (CSF) histamine contents are altered in human narcolepsy and whether these alterations are specific to hypocretin deficiency, as defined by low CSF hypocretin-1.

Methods:

Patients meeting the ICSD-2 criteria for narcolepsy with and without cataplexy and who had CSF hypocretin-1 results available were selected from the Stanford Narcolepsy Database on the basis of CSF availability and adequate age and sex matching across 3 groups: narcolepsy with low CSF hypocretin-1 (n = 34, 100% with cataplexy), narcolepsy without low CSF hypocretin-1 (n = 24, 75% with cataplexy), and normal controls (n = 23). Low CSF hypocretin-1 was defined as CSF ≤ 110 pg/mL (1/3 of mean control values). Six of 34 patients with low CSF hypocretin-1, six of 24 subjects with normal CSF hypocretin-1, and all controls were unmedicated at the time of CSF collection. CSF histamine was measured in all samples using a fluorometric HPLC system.

Results:

Mean CSF histamine levels were: 133.2 ± 20.1 pg/mL in narcoleptic subjects with low CSF hypocretin-1, 233.3 ± 46.5 pg/mL in patients with normal CSF hypocretin-1 (204.9 ± 89.7 pg/mL if only patients without cataplexy are included), and 300.5 ± 49.7 pg/mL in controls, reaching statistically significant differences between the 3 groups.

Conclusion:

CSF histamine levels are reduced in human narcolepsy. The reduction of CSF histamine levels was more evident in the cases with low CSF hypocretin-1, and levels were intermediate in other narcolepsy cases. As histamine is a wake-promoting amine known to decrease during sleep, decreased histamine could either passively reflect or partially mediate daytime sleepiness in these pathologies.

Citation:

Nishino S; Sakurai E; Nevsimalova S; Yoshida Y; Watanabe T; Yanai K; Mignot E. Decreased CSF histamine in narcolepsy with and without low CSF hypocretin-1 in comparison to healthy controls. SLEEP 2009;32(2):175-180.

NARCOLEPSY IS A CHRONIC SLEEP DISORDER CHARACTERIZED BY EXCESSIVE DAYTIME SLEEPINESS (EDS), CATAPLEXY, AND OTHER MANIFESTATIONS OF dissociated rapid eye movement (REM) sleep, such as sleep paralysis and hypnagogic hallucinations.^1,2 A common cause of narcolepsy is defective hypocretin/orexin transmission, either as a result of mutations in the hypocretin receptor 2 [hcrtr-2] (in dogs and mice^3,4) or a loss in ligand production (in humans, canines, and mice).^5–9 In human patients, hypocretin deficiency is clinically evaluated by measuring cerebrospinal fluid (CSF) hypocretin-1.^10–12 Low CSF hypocretin-1 levels, typically found in 90% to 95% of narcolepsy-cataplexy patients and in 10% to 25% of patients without cataplexy, are now considered diagnostic for narcolepsy in the international classification of sleep disorders [ICSD-2].¹³ Studies in human and animals have also shown that some hypocretin deficient narcoleptic patients are obese and have altered energy homeostasis (i.e., reduced metabolism).^8,12,14,15 In contrast, the cause of narcolepsy with normal CSF hypocretin-1 is unknown. Narcolepsy in these patients is generally less severe; patients frequently do not have cataplexy and have less metabolic abnormalities.^12,16 Narcolepsy with normal CSF hypocretin-1 may involve minor defects in hypocretin transmission or/and other causes.

The wake-active amine, histamine, is another neurotransmitter of possible importance in the mediation of daytime sleepiness in hypocretin deficient narcolepsy.¹⁷ We have previously reported that brain contents of histamine are reduced in hcrtr-2-mutated and hypocretin ligand-deficient narcoleptic dogs.¹⁸ Histaminergic neurons are located exclusively in the tuberomammillary nucleus (TMN) of the posterior hypothalamus and project their axons to various brain regions involved in the regulation of the sleep-wake cycle, such as the cerebral cortex, thalamus, anterior hypothalamus, forebrain, and brainstem cholinergic and monoaminergic structures.¹⁷ Together with other wake-active amines such as noradrenaline, histaminergic neurons of the TMN fire most rapidly during the waking state, less during slow wave sleep, and cease firing during REM sleep.^19,20 In vivo animal studies have also shown increased release during the active phase of the light dark cycle of rodents compared to the inactive phase.²¹ Histaminergic neurotransmission plays a pivotal role in cortical activation through interactions with the cholinergic neurons as well as under the influence of hypothalamic wake (i.e., hypocretin) and sleep regulatory (i.e., ventrolateral preoptic GABAergic/glanine) neurons.¹⁷ Most interestingly, hypocretin neurons project densely to the TMN and stimulate these neutrons through the hcrtr-2,²² the receptor subtype that was mutated in narcoleptic canines. Furthermore, the wake-promoting effects of centrally administered hypocretin-1 are abolished in histamine H₁ receptor gene knockout mice, suggesting that the wake-promoting effects of hypocretin-1 are dependent on the availability of histaminergic neurotransmission.²³ Finally, it is well known that decreasing histaminergic transmission using H₁ receptor antagonists induce sleep,²⁴ and that mice deficient in histamine or histamine H₁ receptors have sleep and locomotor activity abnormalities suggestive of sleepiness, especially when placed in a novel environment.^25,26 All of the above suggest that histamine may be a prime candidate in maintaining wakefulness in normal subjects and could thus be impaired in disorders of centrally mediated daytime sleepiness, including narcolepsy.

Interestingly, low concentrations of histamine can be measured in human and rat CSF; increased CSF histamine levels have been reported in aseptic meningitis²⁷ and febrile children.²⁸ Whether or not the level increase is of central origin is unknown, as histamine is highly concentrated in mast cells and blood.²⁹ In rats, we recently found that CSF histamine contents were moderately higher during the active period and were increased after administration of a histamine H₃ receptor antagonist (that enhances neural histamine release and wakefulness) and after forced wakefulness,³⁰ paralleling findings using in vivo dialysis.²¹ These results demonstrate that central histaminergic activity (as measured using in vivo dialysis) is reflected in CSF histamine content, at least in the context of atraumatic taps. In this study, we therefore used previously collected CSF samples to study histaminergic transmission in patients with narcolepsy.

METHOD

Subjects

All patients were from the Stanford Sleep Clinic and Department of Neurology and the First Faculty of Medicine, Charles University; control subjects were recruited at the Stanford Sleep Clinic. The ethics committee in each institute approved the study. Patients and controls gave informed consent for the study. Fifty-two subjects with narcolepsy-cataplexy (age = 42.6 ± 2.1 [SEM] y), 6 subjects with narcolepsy-without cataplexy (29.5 ± 4.5 y), and 23 age-matched control (38.0 ± 2.5 y) subjects (22 Caucasians and one African American) were included (Table 1).

Table 1.

CSF Histamine in Narcolepsy With and Without Low CSF Hypocretin-1 in Comparison to Healthy Controls

	n	Age	M/F	BMI	HLA*	Med**	MSLT***		hypocretin-1 (pg/mL)	Histamine (pg/mL)	Leptin (ng/mL)
							SL	SOREMPs
Healthy controls (C)	23	38.0±2.5	13/10	24.4±0.9	34.8%	0	na	na	269.2±9.5	300.5±39.7	0.307±0.018
Narcolepsy with low CSF hypocretin-1(D)
narcolepsy w. cataplexy	34	42.9±2.9	13/21	28.2±1.1	100%	28	2.7±0.68	2.9±0.25	18.7±4.6	133.2±20.1	0.332±0.022
Narcolepsy with normal CSF hypocretin-1 (ND)	24	38.8±2.4	7/17	24.6±1.0	29.2%	18	6.3±0.80	2.1±0.30	270.7±11.2	226.2±40.5	0.307±0.082
narcolepsy w. cataplexy	18	41.9±2.6	5/13	24.9±1.1	22.2%	14	6.6±1.3	1.9±0.33	233.3±46.5	233.3±46.5	0.312±0.021
narcolepsy w/o. cataplexy	6	29.5±4.5	2/6	24.0±2.6	50%	4	3.9±1.9	2.6±0.34	275.7±15.0	204.9±89.7	0.250±0.042
ANOVA (C vs D vs ND)		0.8		0.015	<0.001		D vs ND:	D vs ND:	< 0.0001	0.009	0.83
					(Chi-squares)		0.001	0.03
Post hoc (LSD)				C vs D:					C vs D:	C vs D:
				0.013					< 0.0001	0.002
				D vs ND:					C vs ND:	C vs ND:
				0.017					P < 0.0001	0.048

^*Percent of HLA DQB1*0602 positive subjects;

^**Number of medicated subjects at the time of the CSF sampling;

^***Mean sleep latency (SL) and sleep onset REM periods (SOREMPs) During MSLT

Diagnosis was made according to the 2nd edition of the International Classification of Sleep Disorders (ICSD-2).¹³ Thirty-four narcolepsy-cataplexy subjects (30 Caucasians, 2 Asians, 1 African American, and 1 mixed ethnicity) were shown to have low CSF hypocretin levels (18.7 ± 4.6 pg/mL, measured in extracted CSF) (Group D: hypocretin-deficient narcolepsy), while hypocretin-1 levels in 18 narcolepsy-cataplexy subjects (270.7 ± 11.2 pg/mL) (17 Caucasians and 1 mixed ethnicity) and 6 narcolepsy-without cataplexy subjects (233.3 ± 46.5 pg/mL) (4 Caucasians, 1 African American, and 1 mixed ethnicity) were in the range of normal controls. These subjects were pooled and studied as a single group of non hypocretin-deficient narcolepsy subjects (Group ND). All control subjects (Group C), 6 of 34 subjects in Group D, and 6 of 18 subjects in Group ND had never been medicated. Medication prescribed in the other subjects included amphetamine and amphetamine-like stimulants (n = 17), modafinil (n = 18), sodium oxybate (n = 6), tricyclic antidepressants (n = 4), and fluoxetine (n = 4). Human leukocyte antigen (HLA) DQB1*0602 status and mean body mass index (BMI) are reported in Table 1.

CSF Collections

Lumbar punctures were all performed between 09:00 and 17:00. CSF was collected in three 2–3 mL fractions (2–3 mL in each tube) and frozen (−80°C) within 30 min. Some of these aliquots have been previously used for the measurement of hypocretin-1 and leptin.^11,12,16 All CSF samples were stored in the same freezer for 1 to 3 years. The CSF samples were thawed, 40 μl of CSF was aliquotted, and aliquots were acidified with 2 μl of 60% perchloric acid and kept frozen. The histamine measures were carried out within a month after aliquoting the samples. In separate pilot experiments, we observed that the number of fractions of CSF collection (3 fractions up to 12 mL) did not affect histamine content (n = 6, df = 2, F = 0.20, P = 0.82, repeated measures ANOVA), while repetitive freeze-thaw cycles significantly reduced histamine levels. Indeed, thaw-freezing twice before acidifying the samples reduced histamine levels by 44.4% (286 pg/mL vs. 159 pg/mL, n = 17, df = 16, T = 3.8, P < 0.001 by paired t-test), while a much smaller effect was observed in samples thaw-frozen only once (248 pg/mL vs. 231 pg/mL by 6.9%, n = 18, df = 17, T = 0.41, P = 0.69, by paired t-test). In animals, we also found that contamination of CSF by more than 20% (v/v) of blood (that induced a considerable red coloring) significantly increased histamine levels.³⁰ For these experiments, we therefore only selected CSF samples that were blood-free and were thaw-frozen no more than once in each group before acidifying the samples (Group C, D, and ND). Blood contamination was checked visually, and CSF samples with any red or yellow coloring (about 2% of all CSF samples in our database) were excluded from the study.

CSF Histamine, Hypocretin-1, and Leptin Measures

Histamine levels were measured in acidified CSF (40 μL with 2 μL of 60% perchloric acid) using a fluorometric HPLC system developed by Yamatodani et al.³¹ Drs. Yanai and Sakurai carried out all histamine assays. The HPLC system includes a cation exchange column (TSK-gel SP-2SW; 6 mm i.d. × 150 mm) to separate histamine and uses a mobile phase of 0.125 M KH₂PO₄ at 0.6 mL/min. Elutants are then mixed with 0.1% o-phthaldialdehyde (OPA) and 2M NaOH in a reaction coil made of polytetrafluoroethylene tubing (0.5 mm i.d. × 5m) at 45°C and pH 12.0, and finally, 10% of sulfuric acid is added to reduce the pH to 3.0. OPA, NaOH, and sulfuric acid solutions are pumped out at 0.2–0.25 mL/min. Fluorescence is monitored using a fluorescent detector (L-7480, Hitachi, Japan) using excitation and emission wavelengths of 360 and 450 nm, respectively. The detection limit was 10 pg/mL of histamine.

Histamine contents were calculated using histamine standard solutions (Sigma, St. Louis, MO). In follow-up assays, values of 2 studies (this study and a companion study by Kanbayashi et al³² were adjusted according to the values of reference CSF samples measured in 2 separate HPLC assays. All measurements were done blindly; each sample was measured twice, and mean values are reported. Inter-assay variation was < 5%. CSF hypocretin-1 and leptin values were measured using radioimmunoassays in samples extracted by Sep-pac 18 (for hypocretin-1) and non-extracted (for leptin) CSF samples; values for these measures were reported in previous studies.^11,12,16 The use of extracted CSF hypocretin-1 measurements versus direct assays explains the slightly lower concentration observed in the healthy subjects than usually reported.^11,12

Data Analysis

Comparisons were performed between D, ND, and C, unless otherwise specified. All values are expressed as mean ± SEM. For multiple group comparisons, one-way analysis of variance (ANOVA) was first applied to determine overall significance, followed by the Fisher least significant difference (LSD) multiple comparison tests as appropriate. For 2-group comparisons, unpaired t-test was used. Pearson correlation coefficient was used to determine possible relationships between CSF histamine, hypocretin-1, leptin levels, BMI, and multiple sleep latency test (MSLT) values (i.e., mean sleep latency and number of sleep onset REM periods [SOREMPs]). When needed, data were transformed (i.e., square root or log transformation) to obtain normal distribution and equality of homogeneity of variance before any parametric comparison was performed. Systat^© 11 (Systat Sofware Inc. Richmond, CA) was used for the statistical data analysis. P < 0.05 (2-tailed) was considered statistically significant.

Results

We found significantly lower CSF histamine levels in hypocretin-deficient narcoleptic subjects (133.2 ± 20.1 pg/mL) than controls (300.5 ± 49.7 pg/mL) (Mean square error [MSE] = 4.001, df = 78, pairwise mean difference [PMD] = −1.709. P < 0.002, ANOVA with Fisher's Protected Least Significant Difference [PLSD]) (Table 1, Figure 1). Furthermore, decreased histamine levels were partially dependant on hypocretin status. Indeed, histamine levels in narcolepsy with normal CSF hypocretin-1 whether with or without cataplexy, were intermediate (233.3 ± 46.5 pg/mL and 204.9 ± 89.7 pg/mL, respectively); this was defined as the ND group, and its values were significantly lower than in controls (MSE = 4.001, df = 78, PMD = −1.172. df = 2,78, P = 0.048). CSF histamine levels in non-medicated subjects (54.5 ± 46 pg/mL) of group D were lower than that in medicated subjects in Group D (150.0 ± 21.4 pg/mL), although the difference did not reach a significant level (df = 32, t = 3.529, P = 0.069, Student t-test). There was no significant difference in CSF histamine levels between non-medicated (221.6 ± 82.7 pg/mL) and medicated (228.3 ± 47.7 pg/mL) subjects in Group ND (df = 22, t = 0.005, P = 0.94, Student t-test).

Figure 1.

CSF histamine levels in controls (Group C), narcolepsy with low CSF hypocretin-1 (Group D), and narcolepsy with normal CSF hypocretin-1 (Group ND). All 34 subjects for narcolepsy with low CSF hypocretin-1 exhibited cataplexy, while 6 of 24 of narcolepsy with normal hypocretin-1 did not have cataplexy. The patients with medication (at the time of CSF sampling) are indicated with closed circles and the non-medicated patients are indicated with open circles. HLA DQB1*0602 negative subjects are indicated with circles with horizontal bars.

CSF histamine values did not differ by gender and did not correlate significantly with age, HLA-DQB1*0602, CSF hypocretin-1, leptin, or BMI (in subjects with normal hypocretin or across all subjects). Furthermore, CSF histamine and CSF histamine/hypocretin ratios did not correlate significantly with objective sleepiness (MSLT mean sleep latency) or number of SOREMPs in all narcoleptics, whether with and without low CSF hypocretin-1 or with and without treatment. As previously reported, mean BMI of Group D was higher than that of Group C or Group ND (28.2 ± 1.1 [D] vs. 24.4 ± 0.9 [C] and vs. 24.6 ± 1.0 [ND], df = 2,77, F = 4.429, P = 0.015 by ANOVA). Furthermore, CSF leptin significantly correlated with BMI across all subjects (r = 0.359, n = 76, P = 0.014, Pearson correlation).

Discussion

In the current study, we measured CSF histamine contents in human narcolepsy using the same CSF samples we used for the measurement of hypocretin-1 and leptin.^11,12,16 Together with a companion publication,³² we are now reporting that CSF histamine levels are reduced in human narcolepsy with and without low CSF hypocretin-1 levels.

The nature and origin of CSF histamine is not well understood. Histamine is released in the brain not only from TMN histamine neurons but also from brain mast cells.³³ Brain mast cells are enriched in the choroid plexus,³⁴ and histamine release by these cells may thus contribute to CSF histamine levels. We recently demonstrated, however, that CSF histamine contents in rats increased significantly after 6 hours of sleep deprivation and after administration of a histamine H₃ receptor antagonist that enhances terminal release of histamine and wakefulness.³⁰ Furthermore, rat CSF histamine levels were high during the active period and low during the resting period. Taken together with the fact that there is a significant delay and slow equilibrium between central CSF compartments and the lumbar sac in humans,³⁵ lumbar CSF histamine levels, like lumbar CSF hypocretin-1 levels, are likely to represent overall histamine release over the 24 h of a day, mostly occurring during the active period.

In this context, decreased CSF histamine seen in hypocretin deficient narcoleptic patients is not surprising. Indeed, hypocretin projects densely onto the TMN, where it activates histaminergic neurons via hcrtr 2,^17,22 the receptor most relevant to the mediation of the narcolepsy phenotype.^3,4 We have also previously reported that brain histamine content is significantly decreased in hcrtr 2-mutated narcoleptic Dobermans and in 3 hypocretin ligand-deficient sporadic narcoleptic dogs, although in these sporadic cases, breed-matched controls were not available.¹⁸ Whether or not this observation could be extended to humans was unknown. Also unknown was if observations in histamine brain content, which likely reflects histamine storage in neurons and brain mast cells,³³ could be applicable to histamine release, and these questions are now addressed by our finding.

More surprisingly, we also found that decreased CSF histamine was also observed in narcolepsy with normal hypocretin levels, independently of cataplexy status (Table 1). One possibility may be that some of these subjects have partial hypocretin deficiency, not reflected in CSF hypocretin-1, but already secondarily affecting CSF histamine levels. A recent population study has found increased DQB1*0602 prevalence in subjects with multiple SOREMPs.³⁶ Furthermore, in a recent analysis, we found a very slight increase (35% vs 25%) in HLA-DQB1*0602 positivity in narcoleptic subjects with normal CSF hypocretin.³⁷ In our sample, however, this possibility is rather unlikely, as low CSF histamine did not correlate with HLA status in subjects with normal CSF hypocretin-1 (Figure 1). In addition, low CSF histamine levels were also found in patients with idiopathic hypersomnia in a companion paper, suggesting that this parameter correlates with sleepiness rather than with REM sleep abnormalities.³² Levels were lowest in untreated patients with low CSF hypocretin-1, most likely to be most sleepy, followed by treated hypocretin deficient patients and were intermediate in other groups who were less sleepy. Thus, it is more likely that CSF histamine levels reflect the severity of EDS independently of hypocretin status, although we did not find such a direct correlation in this study.

The importance of central histaminergic transmission in the regulation of energy expenditure and appetite has been stressed recently, with the possibility that H₁ receptor antagonists increase body weight.³⁸ A secondary goal of the study was thus to explore whether central measures of histamine transmission correlates with metabolic indices in this population. A significantly higher BMI was observed in hypocretin-deficient narcolepsy (the group with the most severe histamine changes). Furthermore, as previously reported, we found a tight positive correlation between BMI and central leptin levels.^12,16 However, histamine content did not correlate with BMI or leptin levels, suggesting histamine may not be a critical mediator of altered energy homeostasis in these patients. Additional experiments will however be needed to further confirm this finding, considering the small sample size of this study and the limitations inherent to our evaluation of central histaminergic function through CSF assessment, discussed below.

Our study had several limitations. First, CSF samples were previously collected, thus (1) CSF collection time (varied from 09:00 to 17:00) and (2) medication status were only partially controlled. Also, sleep/napping and activity status prior to CSF collection was not evaluated, and this could affect histamine levels.

In the control subjects, CSF histamine levels (287.7 ± 48.5 pg/mL, n = 16) of samples collected between 09:00 and 13:00 were not significantly different from those (329.9 ± 73.4 pg/mL, n = 7) collected between 13:00 and 17:00 (df = 21, t = 0.23, P = 0.63, Student t-test). About 60% of the subjects in each group received CSF taps between 09:00 and 11:00, and CSF histamine differences among the 3 groups were evident in these subsets ([D]: 119.8 ± 34.0 pg/mL, n = 20 (P = 0.002, Fisher LSD) and [ND]: 168.5 ± 37.9 pg/mL, n = 16 (P = 0.03, Fisher LSD) vs. [C]: 292.4 ± 40.5 pg/mL, df = 2,47, F = 5.450, P = 0.007 by ANOVA). A similar or a more significant reduction in CSF histamine levels was observed in non-medicated subjects. Furthermore, a recent animal study has shown that acute administration of amphetamine or modafinil does not alter CSF histamine,³⁰ suggesting that low CSF histamine levels in narcolepsy is not secondary to the medication.

Finally and most importantly, it is unknown whether decreased histamine in narcolepsy mediates or is simply a marker of daytime sleepiness in these patients. Central histaminergic systems, together with other wake active systems, play a critical role in cortical activation and in maintaining vigilance,^17,19,20,22 and the sedative effect of histamine antagonist suggests a possibility that a decreased histaminergic neurotransmission mediates sleepiness in at least some cases. Conversely, a number of studies have shown decreased and increased fos expression in histaminergic cells as the result of sleep and wake, respectively.³⁹ In addition, sleep was recently found to be normal in rats with their TMN histaminergic cells lesioned.⁴⁰ It is thus possible that histamine is only one of multiple redundant wake promoting systems, and that low levels passively reflect the expression of neuronal networks activated by sleep and sleepiness. Low histaminergic neuronal transmission may only contribute to the clinical picture when other circuits are also impaired, for example when multiple monoaminergic systems are inactivated, or in the context of hypocretin deficiency, where it is clearly not secondarily activated as a compensatory mechanism.¹⁸

Whether CSF histamine is a mediator or a reflection of sleepiness, further studies under more controlled conditions are needed to explore whether histamine level changes are of clinical significance. A notable example of its use would be in the differentiation of idiopathic hypersomnia and narcolepsy without cataplexy with hypersomnia secondary to depression or insufficient sleep, as therapeutic strategies are dramatically different. Indeed at least for insufficient sleep, CSF histamine would likely be increased based on animal studies.³⁰ It is also possible that histamine levels are a useful, objective marker of the severity of centrally mediated somnolence. Based on this report and the results of the companion paper by Kanbayashi,³² future clinical studies are needed to explore these possibilities, keeping in mind that the use of this marker will always be limited by the need to perform a lumbar puncture.

DISCLOSURE STATEMENT

This was not an industry supported study. Dr. Nishino has received research support from Johnson and Johnson and Jazz Pharmaceuticals. The other authors have indicated no financial conflicts of interest.