Introduction

Narcolepsy is a disabling sleep disorder clinically characterized by sleep manifestations such as excessive daytime sleepiness and abnormal rapid eye movements (REMs), including cataplexy, sleep paralysis, and hypnagogic hallucination [1–3]. Narcolepsy is caused by the selective lack of orexin signaling due to loss of the hypothalamic neurons normally producing this neuropeptide [4, 5]. The striking precision of the attack on orexin neurons is hypothesized to be due to autoimmune/inflammatory processes [6]. In this line, several recent observations of narcolepsy related to H1N1 virus infection and vaccination supported the hypothesis of an immunologic trigger of this sleep disorder in genetically predisposed individuals [7]. To elucidate its pathogenesis, several cerebrospinal fluid (CSF) biomarkers have been investigated, such as beta-amyloid, tau proteins, and neuron-specific enolase (NSE) [8–12]. However, conflicting results emerged from the literature regarding the assessment of CSF biomarkers in narcoleptic patients, probably due to confounding factors such as small and heterogeneous samples of patients [8–12]. In fact, whereas reduced CSF beta-amyloid and normal tau protein levels have been described in narcoleptic drug-naïve patients [8, 9, 12], on the contrary, increased CSF levels of beta-amyloid, tau proteins, and NSE were documented in patients with long disease duration receiving central stimulant medication [10].

To determine whether CSF biomarkers involved in inflammatory and neurodegenerative processes (beta-amyloid and tau proteins) change in narcolepsy over time, we evaluated these biomarkers in a population of narcoleptic drug-naïve patients ranging from early to late phases of the disease.

Methods

Participants and study design

We consecutively recruited drug-naїve patients admitted to the Sleep Center of the University of Rome “Tor Vergata” from September 2011 to December 2014. All patients underwent a standard screening including physical and neurological examination, laboratory tests, HLA-DQB1*06:02 haplotype determination, full laboratory polysomnography followed by the multiple sleep latency test, brain MRI, and lumbar puncture (LP) for analysis of CSF biomarkers, including orexin, beta-amyloid1–42 (Aβ42), total tau (t-tau), and phosphorylated tau (p-tau) proteins. All of them were diagnosed as narcolepsy type 1 (N1) or narcolepsy type 2 (N2) according to the International Classification of Sleep Disorders, 3rd edition (2014) [1].

We also enrolled controls of similar age and sex with narcolepsy patients. All of them were patients at the same neurology clinic, which underwent neurological examination, brain MRI, and LP for diagnostic purposes, who were considered healthy at the end of the diagnostic investigations which resulted negative.

Narcoleptic eligible patients and controls were required to meet the following inclusion criteria: no concomitant neurological or psychiatric diseases, no use of drugs active on the CNS, and no use of caffeine, tobacco, and/or alcohol at the time of the investigations. Exclusion criteria for patients and controls were the following: neoplastic or thyroid illness, inflammatory/infectious conditions, anti-inflammatory or corticosteroid therapies, and abnormal cell count (>4 cells/μl) at the CSF sample analysis. Additional exclusion criterion for controls was the presence of sleep complaints investigated by means of a structured interview performed by certified sleep medicine experts (AR, FI, and FP).

Narcoleptic patients were divided into two groups based on the estimated disease duration. Disease duration was defined as the elapsed time from the onset of the first symptom (i.e., sleepiness or cataplexy, including “cataplectic facies”) to clinical observation. Based on the median length of the disease duration in the whole group, a cutoff to define two subgroups of equal size was chosen. Therefore, we considered two subgroups: patients with a short disease duration (SdN) and patients with a long disease duration (LdN).

The study protocol was approved by the internal review board of the local ethical committee. Written informed consent was obtained from all participating in the study.

CSF collection and analysis

All the CSF samples were obtained by LP performed in decubitus position between 8:00 and 9:00 AM using an atraumatic needle. Blood specimens were also obtained at the same time of LP procedure. CSF samples were collected in polypropylene tubes using standard sterile techniques [13]. The first 4-ml CSF sample was used for biochemistry routine analysis including total cell count. The second 4-ml CSF sample was centrifuged to eliminate cells and cellular debris and immediately frozen at −80 °C until the analysis to assess orexin, t-tau, p-tau, and Aβ42 amount.

The CSF Aβ1–42, t-tau, and p-tau levels were determined according to previously published standard procedures, using commercially available sandwich enzyme-linked immunosorbent assays (ELISAs) (Innotest β-Amyloid 1–42, Innotest h-T-tau, Innotest Phospho-T-tau 181; Innogenetics, Ghent, Belgium) [13].

The CSF orexin levels were detected according to previously published standard procedures with commercially available ELISA kit (Orexin A/Hypocretin-1 EIA Kit; Phoenix Pharmaceuticals, Burlingame, CA, USA), based on the principle of competitive enzyme immunoassay [9, 14]. The unknown sample concentrations were calculated on the corresponding standard sigmoid curve equation (peptide standards ranging from 0 to 1000 pg/ml); pathological cutoff was set at ≤50 pg/ml [9, 14]. The intra-assay and inter-assay coefficients of variation were 2.5 and 3.9 %, respectively [14]. The minimum detectable concentration found was 26.2 pg/ml [14]. All the samples were analyzed each time on the same ELISA plate, in duplicate and in random order. The experiments were replicated three times to confirm the values and to demonstrate the reliability of the assay. Researchers (MN and SB) completely blinded to the study participants’ clinical status performed the analysis.

Data and statistical analysis

The Kolmogorov-Smirnov test was used to check for normal distribution of data. Student’s t test was used in order to compare demographic and CSF data between the two groups (narcolepsy versus controls; SdN versus LdN). The one-way ANOVA was used to compare CSF results between SdN, LdN, and controls. The post hoc analysis was performed using Tukey’s HSD test.

Within the narcolepsy group, correlations between levels of CSF biomarkers (orexin, Aβ42, t-tau, p-tau) and clinical data were separately performed by utilizing the Pearson correlation test. An additional multiple regression analysis between demographic, clinical, and CSF data was also performed. P value <0.05 was considered to be statistically significant.

Results

Demographic and clinical data

Twenty-six consecutive drug-naїve narcolepsy patients were enrolled from September 2011 to December 2014. Based on the median length of the disease duration of the whole group, a cutoff of 5 years was chosen to define two subgroups of equal size: SdN patients reporting a disease duration ≤5 years and LdN patients reporting a disease duration >5 years. Therefore, 13 narcolepsy patients were included in the SdN group whereas the remaining 13 narcolepsy patients constituted the LdN group. The control group consisted of 17 healthy subjects of similar age and sex with narcolepsy patients. Considering narcolepsy diagnosis, 12 patients were included in the N1 group (7 SdN and 5 LdN) whereas 14 patients were counted in the N2 group (5 SdN and 7 LdN).

Demographic and clinical features of patients and controls are summarized in Tables 1, 2, and 3.

Table 1 Demographic, clinic, MSLT, and CSF data of narcoleptic patients and controls
Full size table
Table 2 Demographic, clinic, MSLT, and CSF data of narcoleptic patient subgroups
Full size table
Table 3 Demographic, clinic, MSLT, and CSF data of N1 and N2 patients
Full size table

Between-group analysis

By comparing CSF Aβ42 levels between the whole narcolepsy population and the control group, we found significantly lower CSF Aβ42 levels in narcolepsy patients than controls (see Table 1). On the other hand, we did not find differences in CSF t-tau and p-tau levels between patients and controls (see Table 1). As expected, narcolepsy patients showed lower CSF orexin levels with respect to controls (see Table 1).

We next compared CSF biomarkers between SdN, LdN, and controls. Statistical analysis documented reduced CSF Aβ42 levels in the SdN group compared to both LdN and control groups (see Table 2 and Fig. 1). Furthermore, LdN narcoleptic patients showed lower CSF Aβ42 levels with respect to controls (see Fig. 1).

Fig. 1
figure 1

Cerebrospinal fluid (CSF) beta-amyloid1–42 (Aβ 42 ) levels in narcoleptic patients with short disease duration (SdN) and long disease duration (LdN) and controls (expressed in pg/ml)

Full size image

Considering t-tau and p-tau levels, the statistical analysis between the three groups showed an increase of p-tau levels in LdN patients compared to both SdN and controls (see Table 2 and Fig. 2).

Fig. 2
figure 2

Cerebrospinal fluid (CSF) phosphorylated tau (p-tau) levels in narcoleptic patients with short disease duration (SdN) and long disease duration (LdN) and controls (expressed in pg/ml)

Full size image

Comparing N1 and N2 patient groups, we did not find differences in CSF t-tau, p-tau, and Aβ42 levels. Consistently, N1 patients showed reduced CSF orexin levels compared to N2 patients (Table 3).

Correlations in patients and control groups

In the whole narcoleptic group, the Pearson correlation test proved a significantly positive correlation between CSF Aβ42 levels and disease duration (R = 0.54, P = 0.005; Fig. 3). On the other hand, we did not find correlations between CSF and clinical data. Moreover, we found no correlations between all the parameters analyzed in control groups.

Fig. 3
figure 3

Correlation between cerebrospinal fluid (CSF) beta-amyloid1–42 (Aβ 42 ) levels (pg/ml) and disease duration (years) in narcoleptic patients

Full size image

We also performed in the narcolepsy group an additional multivariate regression analysis among demographic, clinical, and CSF data revealing the significant interplay between CSF Aβ42 levels and disease duration (β = 0.39 and P < 0.05).

Discussion

In this paper, we investigated the CSF beta-amyloid and tau protein levels in a large clinically based cohort of narcoleptic drug-naïve patients. Dividing narcoleptic patients into two subsets on the basis of the median disease duration of the whole group, we documented lower CSF beta-amyloid levels in patients near to disease onset compared to patients with long disease duration. In this way, it is conceivable to highlight the impairment of beta-amyloid metabolism and clearance, manifesting with reduced CSF Aβ42 levels, in narcoleptic patients near to disease onset. On the other hand, we found in patients with a long disease duration higher CSF Aβ42 concentrations than SdN, documenting the trend in restoring the beta-amyloid metabolism to physiological conditions along with the disease course. Consistently, we also documented the significant positive correlation, associated with the significant mutual interplay in the multiple regression analysis, between Aβ42 levels and disease duration, corroborating the hypothesis that the beta-amyloid metabolism is more altered near to the disease onset and progressively increases to normal levels. However, we documented that both SdN and LdN patients showed lower CSF Aβ42 concentrations with respect to controls, confirming in an enlarged group of patients previous findings regarding the alteration of the beta-amyloid cascade in narcolepsy [9]. The cause of the beta-amyloid metabolism dysregulation in narcolepsy is still unidentified. In our previous study, we hypothesized that both inflammatory conditions present at disease onset—related to the immunopathogenesis of narcolepsy—and the presence of an “amyloidogenic” pathway of beta-amyloid catabolism—owing to alpha-secretase enzyme deficiency—may be involved [4, 9, 15–18]. In keeping with this hypothesis, the present data allow to suppose two different phases of beta-amyloid metabolism dysregulation in narcolepsy.

The first phase, probably acute or subacute, near to disease onset, is characterized by transient inflammatory processes well known in narcolepsy [6, 8, 15, 19], which may dramatically alter the beta-amyloid cascade producing the marked reduction of CSF Aβ42 levels evident in SdN patients. In fact, it has been demonstrated that the beta-amyloid metabolism could be affected by inflammatory processes [20]; moreover, the activation of the immune system may alter the beta-amyloid metabolism [21]. Indeed, narcolepsy is considered an inflammatory/autoimmune disease, since it has been associated with HLA-DQB1*06:02 allele and it has been considered to be triggered by environmental factors such as streptococcal and influenza A H1N1 infection and vaccination [7, 15, 22, 23]. Furthermore, taking into account that in narcolepsy the inflammatory responses are confined to the lateral hypothalamus, thus reducing the opportunity to find detectable levels of inflammatory markers such as cytokines diffusing to the CSF [24, 25], we could speculate that the alteration of beta-amyloid metabolism may be a detectable evidence of the immune processes present in narcoleptic brains.

The second phase, evident along with the disease course, in which inflammatory triggers are lowered and beta-amyloid metabolism could be mainly affected by the underlying deficiency of alpha-secretase enzymes, was recently documented in narcolepsy patients [16–18]. In fact, a reduced activity of alpha-secretase enzymes may be responsible for an incomplete catabolism of beta-amyloid proteins, thus causing the reduction of CSF Aβ42 concentrations [16, 26]. These effects could result in a less severe reduction of CSF Aβ42 levels evident in LdN patients.

Therefore, we hypothesize that SdN patients show lower CSF Aβ42 concentrations than LdN patients since they are co-affected by inflammatory processes and alpha-secretase enzyme deficiency. On the other hand, LdN patients present a less severe impairment of the beta-amyloid metabolism, because the inflammatory drives are less evident in the chronic phase of the disease, and then the patients suffer predominantly from the deficiency of alpha-secretase enzymes. However, we are aware that Aβ42 changes in narcolepsy could be more informative if assessed over time within the same patient.

Recently, it has been described that narcoleptic patients with a long disease duration and under stable treatment with psychostimulants showed increased t-tau, p-tau, and Aβ42 levels [10]. Authors postulated that these findings may be the expression of an altered cell metabolism, featured by changing in plasticity and regenerative capacity of the neuronal networks of the brain [10]. In agreement with this observation, we found higher CSF p-tau levels in LdN patients than in both SdN patients and controls. Taking into account that narcoleptic patients included in our study were drug naïve, these data may encourage the emerging hypothesis of an altered neuronal plasticity, involving the phosphorylated tau proteins, evident in narcolepsy along with the disease course. However, few studies showed that narcolepsy seems to not predispose to Alzheimer’s dementia (AD) [11, 27, 28], thus proving that beta-amyloid and tau protein metabolism dysregulation does not take part in prearranging to AD pathology.

We next found no differences in the levels of CSF biomarkers between N1 and N2 patients. We may hypothesize that both types of narcolepsy show similar pathological pathways of beta-amyloid and tau protein metabolism alteration, disregarding orexin signaling.

In conclusion, beta-amyloid metabolism seems to be impaired in narcolepsy not only at the onset but also along the disease course although it shows a compensatory profile over time. At the same time, also CSF biomarkers regulating the neuron dynamic and structure appear to be altered in narcolepsy, as documented by the trend in an increase of CSF p-tau levels after a long disease duration. However, the mechanism by which beta-amyloid is reduced and tau metabolism is dysregulated during narcolepsy, although largely hypothesized, remains still unclear and deserve to be better elucidated in larger groups of narcolepsy patients.

Study limitations

Few limitations of this study should be considered when interpreting the results. First, the sample size is small although similar to other single-center studies on narcolepsy. Second, the mechanism by which beta-amyloid and tau metabolism is dysregulated is still unclear and should be better evaluated in larger multicenter prospective trial in order to clarify the role of these markers in narcolepsy. Therefore, this can be considered as a pilot study to determine a sample size and optimization for a future multicenter trial.