Abstract
Objective: To examine the specificity of low CSF hypocretin-1 levels in narcolepsy and explore the potential role of hypocretins in other neurologic disorders.
Methods: A method to measure hypocretin-1 in 100 μL of crude CSF sample was established and validated. CSF hypocretin-1 was measured in 42 narcolepsy patients (ages 16–70 years), 48 healthy controls (ages 22–77 years,) and 235 patients with various other neurologic conditions (ages 0–85 years).
Results: As previously reported, CSF hypocretin-1 levels were undetectably low (<100 pg/mL) in 37 of 42 narcolepsy subjects. Hypocretin-1 levels were detectable in all controls (224–653 pg/mL) and all neurologic patients (117–720 pg/mL), with the exception of three patients with Guillain–Barré syndrome (GBS). Hypocretin-1 was within the control range in most neurologic patients tested, including patients with AD, PD, and MS. Low but detectable levels (100–194 pg/mL) were found in a subset of patients with acute lymphocytic leukemia, intracranial tumors, craniocerebral trauma, CNS infections, and GBS.
Conclusions: Undetectable CSF hypocretin-1 levels are highly specific to narcolepsy and rare cases of GBS. Measuring hypocretin-1 levels in the CSF of patients suspected of narcolepsy is a useful diagnostic procedure. Low hypocretin levels are also observed in a large range of neurologic conditions, most strikingly in subjects with head trauma. These alterations may reflect focal lesions in the hypothalamus, destruction of the blood brain barrier, or transient or chronic hypofunction of the hypothalamus. Future research in this area is needed to establish functional significance.
Narcolepsy is a chronic, disabling sleep disorder affecting 1 in 2,000 individuals.1-3⇓⇓ The disease is marked by excessive daytime sleepiness, cataplexy, sleep paralysis, and hypnagogic hallucinations.2-4⇓⇓ Narcolepsy is thought to be caused by a complex interplay of environmental and genetic factors, and is tightly associated with human leukocyte antigen (HLA)-DQBI*0602 (85–95% of narcoleptics).1,3⇓ The disease is currently treated symptomatically with amphetamine-like stimulants for excessive daytime sleepiness and antidepressants for cataplexy.3,5,6⇓⇓
The quality of life of patients with narcolepsy is lower than in patients with depression or epilepsy.7 Symptoms usually appear during adolescence, when schoolwork and social interaction become increasingly important; however, diagnosis is often not made until years after disease onset (up to 10 years).8 This is due, in part, to the difficulty of diagnosis. Current diagnosis is based primarily on clinical picture, with the assistance of sleep recordings and HLA typing.2,3,5⇓⇓ A Multiple Sleep Latency Test is often used to document excessive daytime sleepiness and the occurrence of sleep onset REM periods.9 However, this test is time-consuming and expensive, and false negatives and false positives are common in clinical practice.9,10⇓ HLA typing is only supportive of a diagnosis, since the specificity of DQBI*0602 positivity is as low as 62%.11 Therefore, establishment of a highly sensitive and specific diagnostic test for narcolepsy based on the pathophysiological basis of the disease would be very useful in confirming diagnosis.
Hypocretins/orexins are recently discovered neuropeptides of hypothalamic origin.12,13⇓ A connection between hypocretin deficiency and narcolepsy was first made when discoveries in animal models demonstrated that either absence of hypocretin ligands or mutations in hypocretin receptor 2, one of two hypocretin receptors, could cause narcolepsy.14-17⇓⇓⇓ While hypocretin gene mutations in human narcolepsy are, in general, absent, most sporadic cases are now known to be associated with undetectably low hypocretin-1 levels in the CSF18 and a dramatic decrease of hypocretin peptides and transcript in the brain.19,20⇓
An extended CSF measurement study in 38 randomly selected narcolepsy-cataplexy patients recently demonstrated that abnormal (low or high) CSF hypocretin-1 levels are predictive for narcolepsy (abnormal levels in 89.5% of the overall patient population and 94.7% of HLA positive cases).21 Furthermore, undetectable levels were observed in narcoleptic patients as young as 16 years old, and as early as 3 years after disease onset,21 highlighting the potential utility of CSF hypocretin-1 measurement as a diagnostic tool for narcolepsy. However, although sensitivity of low hypocretin-1 levels for narcolepsy-cataplexy is established, specificity still needs to be determined.
In this study, we measured hypocretins in the CSF of 235 patients with a wide range of neurologic disorders, 42 patients with narcolepsy, and 48 healthy controls to determine the specificity of low hypocretin-1 levels for narcolepsy.
Methods.
Subjects.
Forty-two patients with narcolepsy and 48 healthy control subjects were randomly recruited from the Stanford University Center for Narcolepsy, the Department of Neurology at Leiden University Medical Center (The Netherlands), and the Department of Neurology, 1st Medical Faculty, Charles University (Czech Republic). All narcolepsy subjects had a confirmed diagnosis of narcolepsy-cataplexy (reviewed by E.M.).21 All control subjects were healthy and had no clinical sleep abnormalities. Hypocretin-1 levels for 36 narcolepsy patients and 15 controls included in this study were previously reported.21
CSF from 235 subjects with various neurologic disorders was additionally provided by The Department of Neurology, Leiden University Medical Center; Kumamoto University, School of Medicine (Japan); Showa University, School of Medicine (Japan); The Department of Neurology, Charles University; The Department of Neuroscience, Neurology, University Hospital, Uppsala (Sweden); The Palo Alto VA Hospital; and The Parkinson’s Institute, Sunnyvale, CA. CSF from all neurologic patients was previously collected by lumbar puncture between the years of 1985 and 2001 for diagnostic and research purposes, and was kept at −80 °C prior to measurement. This study was approved by the local ethical committee.
Patients were grouped into clinical categories based on final diagnosis as determined from individual clinical records (table). Patients with incomplete records or unclear/missing final diagnoses were not included in this study.
CSF hypocretin-1 levels in various neurologic conditions
CSF collection.
To control for possible differences in hypocretin-1 level due to time, lumbar punctures in all patients with neurologic disease were carried out between the hours of 7 am and 5 pm, with the exception of four patients tapped between 6 pm and 7 pm, four patients tapped between 10 pm and 11 pm, and one patient tapped at 2 am. Controls were tapped between the hours of 7 am and 5 pm with the exception of three subjects tapped at 11 pm and five subjects tapped at 2 am. All patients with narcolepsy were tapped between the hours of 7 am and 5 pm.
Hypocretin-1 measurement in crude CSF.
Hypocretin-1, but not hypocretin-2, was measured in this study because hypocretin-2 is not detectable in CSF volumes less than 10 mL using currently available radioimmunoassays (RIA).21 Hypocretin-1 values were measured blind to diagnosis. In contrast to previous studies in which hypocretins were measured in extracted CSF,18,21⇓ hypocretins in this study were measured in crude (unextracted) CSF using commercially available RIA kits (Orexin A RIA kits; Phoenix Pharmaceuticals, Belmont, CA). Measuring hypocretin-1 in crude CSF simplifies the assay procedure by eliminating the extraction step. Furthermore, it significantly lowers the amount of CSF necessary for measurement (200 μL for duplicate crude measurements, vs 1 mL for the measurement of extracted samples).
Since direct application of unextracted CSF may interfere with the assay, we first evaluated the accuracy of measurements in crude CSF by comparing crude vs extracted hypocretin-1 values in subjects from two selected sample populations. Samples (1 mL) were extracted using a reverse phase column as previously described.16,21⇓ After extraction, samples were dried and resuspended in 300 μL of RIA buffer. Hypocretins were then measured in either 100 μL of crude CSF or 100 μL of extracted CSF. In addition, we assessed whether a hypocretin-1 concentration gradient exists by measuring hypocretin-1 in crude CSF in six sequential fractions of 2 mL collected from four people. Finally, we assessed handling conditions for crude samples by measuring hypocretin-1 in samples that had been repeatedly frozen and thawed (up to six times), as well as in samples that had been left at room temperature for 0, 24, 48, or 96 hours.
The detection limit was 100 pg/mL and intra-assay variability was less than 5%. To adjust for inter-assay variation, we included a reference CSF sample in every assay and adjusted final values accordingly. All comparative measurements (i.e., concentration gradient, crude vs extracted CSF) were done in a single assay. All samples were measured in duplicate.
Definition of low hypocretin-1 levels.
Control data were log transformed to achieve homogeneity of variance, and the threshold for abnormally low hypocretin-1 level was set at the mean minus 2 SD of the log-transformed control data (<194 pg/mL).
Statistics.
Disease groups of particular interest were compared with the control group to determine whether changes in hypocretin-1 levels occur in various neurologic conditions (see figure). Significance of the difference (p < 0.05) was determined on the log-transformed hypocretin-1 values using one-way analysis of variance with post-hoc Fisher analysis.
Figure. Hypocretin-1 levels in controls and major neurologic disorder groups.
The number of subjects in each group is listed in parentheses. Thick bars represent mean values and thin bars represent median values. Control subjects that were tapped at night (6 pm to 2 am) are indicated with diamonds. Statistics were applied on the log-transformed values for each disease group (one-way analysis of variance and Fisher’s post hoc test). The acute lymphocytic leukemia and acute nonlymphocytic leukemia (ALL/ANLL) group (p < 0.05) and tumor group (p = 0.002) were significantly different, as was the head trauma group (p = 0.0002), infection group (p = 0.014), and Guillain–Barre syndrome (GBS) group (p = 0.0006). Narcolepsy was highly significant (p < 0.0001). The dotted line represents the cut-off point for abnormally low hypocretin-1 levels. The ALL/ANLL group includes 9 patients with other non-CNS malignant disorders, and the GBS group includes 2 patients with Fisher syndrome.
Results.
Hypocretin-1 measurement in crude CSF.
We found a significant correlation between hypocretin-1 levels measured in crude or extracted CSF in two sample populations selected for their wide variation in hypocretin levels (population 1: [crude range: 221–428 pg/mL]; y = 0.86 [extracted value] + 69.0; r = 0.63, n = 32; population 2: [crude range: 149–864 pg/mL]; y = 0.96 [extracted value] + 65.6; r = 0.82, n = 18). No CSF concentration gradient for hypocretin-1 was found (up to six fractions; 12 mL total; Friedman test; p = 0.70). Furthermore, hypocretin-1 levels in crude samples were unaffected by repeated freezing and thawing up to six times (p = 0.83) or by exposure to room temperature up to 96 hours (p = 0.85).
Hypocretin-1 in healthy controls.
Hypocretin-1 levels in the crude CSF of all 48 healthy controls were well above the detection limit, with an average (±SD) of 344 ± 107 pg/mL (range: 224–653 pg/mL) (see table and figure). Hypocretin-1 values in extracted CSF from 32 of these healthy controls were 285 ± 37.2 pg/mL (range: 176–385 pg/mL) compared with 372 ± 115 pg/mL (range 231–653 pg/mL) in crude CSF. Thus, interindividual variation was larger in crude samples than in extracted samples. There was no difference in CSF hypocretin-1 levels in the control group with respect to age, gender, time of lumbar puncture (night or day) or season of lumbar puncture.
Hypocretin-1 in narcolepsy patients.
Hypocretin-1 was undetectably low (<100 pg/mL) in 37 of 42 patients with narcolepsy (see table and figure). Four of the five patients with detectable hypocretin-1 had levels within the healthy control range (274 pg/mL, 297 pg/mL, 624 pg/mL, and 649 pg/mL); the fifth patient had a level of 169 pg/mL. The five patients with detectable levels ranged in age from 35 to 61, and duration of illness prior to the lumbar puncture ranged from 11 to 44 years, suggesting that detectable CSF hypocretin-1 level in these subjects does not correlate with age or the course of the disease.
Hypocretin-1 in patients with various neurologic conditions.
Patients were classified into 15 major categories based on clinical description and final diagnosis. Median CSF hypocretin-1 values and number of subjects that showed low levels (<194 pg/mL) for each category are listed in the table (additional material related to this article can be found on the Neurology Web site. Go to www. neurology.org and scroll down the Table of Contents for the December 26 issue to find the title link for this article).
Hypocretin-1 was within the control range in the majority of neurologic patients (excluding patients with narcolepsy) tested (85.6%). All patients with neurodegenerative diseases, such as PD and AD, had normal levels (see figure). Further, all patients with MS were within the control range.
Several disease categories, however, contained one or more patients with low hypocretin levels. We selected 12 major disease categories or subcategories of interest and determined significance of lower levels (see figure). Several patients with acute lymphocytic leukemia (ALL) and acute non-lymphocytic leukemia (ANLL), intracranial tumors, craniocerebral trauma, CNS infections and Guillain–Barré syndrome (GBS) had low levels, and these disease groups had statistically lower hypocretin levels when compared to controls. Two patients with vascular disease and two patients with epileptic seizures also had low levels, but these disease categories as a whole were not statistically different from controls.
Surprisingly, we found that three patients with GBS had levels that fell beneath the detectable range, making GBS the only disease category besides narcolepsy to contain patients with undetectable hypocretin-1 levels.
Hypocretin-1 in children with neurologic disease.
Neurologic conditions typically observed in children are listed separately (see table). Although hypocretin-1 levels in the majority of children (as young as 1 day old) were within the normal adult range, four children (9 months to 3 years old) with various neurologic conditions had low levels.
Clinical description of patients with low hypocretin-1 levels.
To further evaluate the significance of the low levels found in various neurologic disease groups, all individual cases with low hypocretin-1 level are listed with accompanying clinical and demographic data in the Appendix to the table (additional material related to this article can be found on the Neurology Web site. Go to www. neurology.org and scroll down the Table of Contents for the December 26 issue to find the title link for this article).
Three patients with GBS had undetectable CSF hypocretin-1 levels. These patients all had severe GBS, with respiratory failure and tetraplegia. Additionally, three more patients with GBS (including one case of Fisher syndrome) had low levels. Two of these patients received multiple CSF taps: a 19-year-old woman (156 pg/mL) and a 69-year-old man (<100 pg/mL). Although the patient with undetectable levels still had undetectable levels 34 days later, the patient with low levels showed recovery of hypocretin-1, with levels increasing from 156 pg/mL to 274 pg/mL after 24 days, and reaching a level of 317 pg/mL after 44 days.
A majority of patients with craniocerebral trauma showed decreased CSF hypocretin-1 levels (CSF collected 5 to 11 days after trauma in four patients, and 7 months after trauma in one patient). All patients experienced moderate disturbance of consciousness. The low levels observed in one patient tapped 7 months after trauma (a 44-year-old man) suggest the possibility of long-lasting alterations of the hypocretin system.
Patients with CNS infectious diseases associated with low hypocretin-1 levels also had disturbance of consciousness or seizures. Some cases of generalized or partial recurrent seizures and status epilepticus had low levels as well.
In one patient with a germ cell tumor (136 pg/mL) that extended to the hypothalamus, hypothalamic dysfunction, including obesity and diabetes insipidus, was observed. However, low levels in other patients with tumors were not linked to obvious hypothalamic dysfunction, nor to specific localization of the tumors.
All patients with ALL and ANLL associated with low CSF hypocretin-1 had no other CSF abnormalities and no leptomeningeal localization, and low levels did not appear to be due to chemotherapy, as three patients were receiving no treatment at the time of CSF collection. However, one patient with a T-cell lymphoma did have leptomeningeal localization.
Since only CSF samples collected by lumbar puncture are included in this study, only a few patients with vascular disease are included. A single patient with subarachnoid hemorrhage and one of five patients with infarction (recurrent cases) had low hypocretin-1 levels.
Four infants, including one with fever, one with developmental abnormalities, one with generalized epileptic seizures, and one with Reye syndrome, had decreased hypocretin-1 levels, but none fell below the detection limit.
Discussion.
We report CSF hypocretin-1 levels in a large population of patients with neurologic disease to determine specificity of undetectable CSF hypocretin-1 to narcolepsy. Using a simplified RIA procedure that can be applied to 100 μL of crude CSF, we confirmed previous results demonstrating undetectable hypocretin-1 in the majority of patients with sporadic narcolepsy.18,21⇓ Most other samples were detectable, with limited variation based on age (1 day to 85 years)22, suggesting early maturation of hypocretin transmission and stability with age. A limitation of the crude CSF assay is the larger variability of the reading, probably the result of the smaller sample volume applied (100 μL vs 300 μL). We also obtained overall higher mean levels of hypocretin-1 in the crude assay, making it difficult to compare exact values using crude vs extracted samples. Of note, high CSF hypocretin-1 levels in a few patients with narcolepsy were previously reported as abnormal, and thus potentially predictive of the disease.21 In this extended study, while some narcolepsy patients had elevated levels, several control subjects also had high levels, making the significance of elevated levels in narcolepsy patients uncertain.
Our results also indicate that CSF hypocretin-1 levels are not altered in most chronic neurologic conditions. CSF hypocretin-1 levels in patients with AD and PD, two conditions with established sleep abnormalities,23,24⇓ were normal. Dysfunction of other neurochemical systems, for example dopaminergic and cholinergic systems in PD and AD, may be more directly involved in sleep abnormalities in these subjects.23,24⇓ Levels in patients with MS were also within the normal range, an important result considering the fact that the disease is associated with the same HLA haplotypes as narcolepsy (e.g., DR2, DQB1*0602)25 and involves autoimmmune inflammation of the CNS, a mechanism that has been proposed in narcolepsy.
Although the majority of patients with chronic neurologic conditions had CSF hypocretin-1 levels within the control range, a subset of subjects with acute or subacute neurologic disorders had decreased levels. This subset included ALL/ANLL, intracranial tumors, cardiovascular events, craniocerebral trauma, CNS infections, and GBS. The most striking findings were in the head trauma and encephalitis subgroups, in which the majority of patients had levels below 2 SD of normal values. These findings are particularly interesting considering that these conditions are frequently associated with acutely disturbed consciousness and residual sleep disturbances. Furthermore, symptomatic cases of narcolepsy have been reported after head trauma and encephalitis,26 suggesting possible interactions between genetic susceptibility and these triggering events. In fact, lower hypocretin-1 levels have been reported in three symptomatic cases of narcolepsy, including a case of narcolepsy associated with an enlargement of the third ventricle in a heredodegenerative disease and two cases of narcolepsy occurring after surgical removal of a hypothalamic tumor.27-29⇓⇓
GBS was the only disease, with the exception of narcolepsy, that was associated with undetectable hypocretin-1 levels. The three patients with undetectable levels had severe GBS, with paralysis of all four limbs and respiratory insufficiency. Three other GBS cases had low but detectable levels. This result is notable, considering the fact that hypocretin-1 was normal in subjects with other polyneuropathies (including 10 cases of Chronic Inflammatory Demyelinating Polyneuropathy). Decreased hypocretin-1 levels in GBS are not likely to be iatrogenic or secondary to severely impaired health conditions, because in one case (a 69-year-old man), hypocretin-1 was already undetectable before respiratory failure occurred. In this case, no treatment was initiated and no increase in CSF protein was observed at the time CSF was first collected and undetectable levels were measured. Interestingly, hypocretin-1 was measured twice in two patients and levels increased after recovery in one of the patients.
The pathophysiological mechanism leading to decreased hypocretin levels in various acute or subacute neurologic disorders, such as GBS, head trauma, and encephalitis, is unknown. Most of these acute conditions are associated with alterations or destruction of the blood-brain (or spinal) barrier.30 This could lead to CSF dilution or dynamic alterations of CSF hypocretin-1 elimination. Alternatively, hypothalamic focal lesions (tumors, contusions) or global hypothalamic hypofunction secondary to a primary neurologic condition may reduce hypocretin-1 production in some cases. In addition to destruction of the blood-brain barrier, hypothalamic dysfunction may also be involved in GBS cases. In GBS, multiple autoantibodies are detected31 and a transient autoimmune process could affect hypocretin-1 neurons. In fact, although GBS is most often characterized as a peripheral disease, global hypothalamic dysregulation with syndrome of inappropriate secretion of antidiuretic hormone32 or diabetes insipidus33 has been reported in some severe cases.
Undetectable CSF hypocretin-1 levels are highly sensitive and specific for narcolepsy, indicating diagnostic utility of CSF hypocretin-1 measurement. However, more than 10% of narcolepsy-cataplexy patients have normal CSF hypocretin-1 levels, suggesting that multiple pathophysiologies may result in a narcolepsy phenotype.16 For this reason, physicians should exercise caution when interpreting a normal CSF hypocretin-1 level. Conversely, a decreased level may not always suggest a diagnosis of narcolepsy, as seen with patients with ALL/ANLL, tumors, craniocerebral trauma, infections, and GBS. More work in determining the functional significance of altered hypocretin transmission is warranted.
Acknowledgments
Supported by grants from the National Institutes of Health (NS 27710, NS23724, NS 33797, MH40041, and MH01600). S.O. was supported by the Foundation “De Drie Lichten” and Hersenstichting Nederland. S.N. was supported by grant agency, GAUK 56/99 and MSM 111100001. A.M. was supported by grants from the Swedish Society of Medicine and the Selander Foundation. M.H. was supported by the Deutsche Forschungsgemeinschaft (DFG-HU 827/2-1).
Acknowledgment
The authors thank Drs. Greer Murphy and Takashi Kanbayashi for helpful comments and support, and Will Rogers for technical assistance.
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the December 26 issue to find the title link for this article.
- Received June 14, 2001.
- Accepted September 10, 2001.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.