Characterization of rapid weight gain phenotype in children with narcolepsy

Abstract Objectives To characterize the rapid weight gain (RWG) phenotype associated with the onset of childhood narcolepsy and to determine whether it could constitute a marker of severity of the disease. Methods RWG was defined using the BMI z‐score slope reported to one year (>0.67 SD) from symptom onset to disease diagnosis. We compared the clinical, metabolic, and sleep characteristics between patients with or without RWG at diagnosis. Pharmacological management, anthropometric, and clinical progression were also evaluated during the follow‐up. Results A total of 84 de novo narcoleptic pediatric patients were included; their median age at diagnosis was 12.0 years; 59.5% boys, 90.5% cataplexy, and 98.7% HLA‐DQB1*06:02, 57% had RWG profile. RWG patients were younger at diagnosis than non‐RWG patients, despite a shorter diagnostic delay. They had a higher BMI z‐score and a higher prevalence of obesity at diagnosis, but not at symptom onset, and higher adapted Epworth Sleepiness Scale and Insomnia Severity Index scores than non‐RWG patients. No differences on nocturnal polysomnography and multiple sleep latency tests were found between groups at disease diagnosis. After a median follow‐up of 5 years, RWG patients still had a higher BMI z‐score and a higher prevalence of obesity despite benefiting from the same therapeutic management and displaying improvement in sleepiness and school difficulties. Conclusions Narcoleptic RWG patients were younger, sleepier, and the prevalence of obesity was higher at diagnosis despite a shorter diagnostic delay than that of non‐RWG patients. These patients had also a higher risk of developing a long‐term obesity, despite a positive progression of their narcoleptic symptoms. RGW could then represent a maker of a more severe phenotype of childhood narcolepsy, which should inspire a prompt and more offensive management to prevent obesity and its complications.


| INTRODUC TI ON
Narcolepsy is a rare neurological disorder characterized by excessive daytime sleepiness (EDS) and abnormal rapid eye movement (REM) sleep manifestations including cataplexy, hypnagogic hallucinations, and sleep paralysis. 1 For more than half of narcoleptic patients, symptoms appear before the age of 18 years. 2 The human leukocyte antigen (HLA) DQB1*0602 genotype is closely associated with narcolepsy. 3 It is generally believed that a hypocretin neuronal loss in the lateral hypothalamus 4 constitutes the direct cause of narcolepsy with cataplexy (NC). As the human leukocyte antigen (HLA) DQB1*0602 genotype is closely associated with narcolepsy, 3 this disappearance of hypocretin cells is generally imputed to autoimmune mechanisms. 5,6 NC with cerebrospinal fluid (CSF) hypocretin-1 levels ≤110 ng/ml is also called narcolepsy type 1 or hypocretin-deficient (NT1). Hypocretins are not only involved in sleep-wake control but have also been involved in feeding behavior and energy balance regulation. 7 More than 50% of narcoleptic children are obese/overweight 8,9 (higher prevalence than among adult patients), 10 and more than 80% of them experience a rapid weight gain (RWG) soon after symptom onset, 11 which has drawn increasing clinical attention. 12 The clinical picture and mechanisms involved in RWG remain poorly illustrated, even though the rapid body mass index (BMI) increment in NT1 children has been attributed to a decreased energy expenditure at symptom onset. 13 Notably, the characteristics of the patients who developed RGW have never been addressed. Although Ponziani et al. have analyzed the anthropometric characteristics of 30 NT1 children over a 2-year period before NT1 onset and have found a weight gain starting close to NT1 onset, 12 only the comparison between narcoleptic children with or without obesity has been made. It is noteworthy that some narcoleptic patients have developed obesity without RWG, and that some patients are not obese at diagnosis despite having experienced RWG. The present study sought to determine whether narcoleptic children with/without RWG after symptom onset had distinct clinical and polysomnographic characteristics in order to assess whether this RGW criteria could constitute a marker of the severity of the disease. Firstly, we compared the clinical, anthropometric, endocrinological, metabolic, and sleep characteristics at diagnosis between narcoleptic patients with and without RWG. Secondly, we evaluated over a follow-up their pharmacological management, clinical and anthropometric changes, and academic achievements.

| Patients
In this retrospective study, data regarding de novo pediatric narcoleptic patients were collected from the pediatric sleep unit of the

| Diagnostic procedure
Patients filled up a sleep log during the 15 days preceding the sleep laboratory evaluation. Each patient underwent nocturnal polysomnography (PSG) followed by the multiple sleep latency test (MSLT) with 4 or 5 nap opportunities; this assessment was stopped after 20 min if no sleep occurred, and after 15-min asleep if sleep occurred. 14 The polysomnography included eight electroencephalogram (EEG) electrodes, two electrooculograms (EOG), and surface electromyograms (EMG); the latter were located on the mentalis muscle, the left and right anterior tibialis muscles, and the oral thermistor, thoracic, and abdominal belts; electrocardiogram and nasal pressure were recorded using cannulae. 14 Sleep stages, arousals, and respiratory events were scored visually according to standard pediatric criteria by an experienced sleep specialist (PF). 14

| Criteria for idiopathic narcolepsy diagnosis
The criteria for the diagnosis of idiopathic narcolepsy 15 were: (1) complaints of excessive daytime sleepiness for at least 3 months,  16 HLA-DQB1*0602 genotyping was determined for all patients, and CSF hypocretin levels were measured for some patients. Brain magnetic resonance imaging was carried out to exclude secondary narcolepsy.

| Questionnaires
During the hospitalization for the diagnosis of narcolepsy, daytime sleepiness was evaluated using the adapted Epworth Sleepiness Scale (AESS), 17 for which a score >10 is considered pathological.
The severity of cataplexy (1=moderate weakness, for example, head drop or jaw opening; 2=can maintain posture with external support; 3=loss of posture and falls to the ground) 18 19 were evaluated.
Depressive feelings were evaluated using the Children's Depression Inventory (CDI), 20 for which a score ≥16 is considered pathological.
Attention-deficit/hyperactivity disorder symptoms were scored by parents using the revised version of the Conners Parents Rating Scale, 21 for which a score ≥65 or ≥75 corresponds to moderate or severe symptoms, respectively. Insomnia was evaluated using the Insomnia Severity Index (ISI), 22 for which a score ≥10 is considered pathological. The severity of the disease was assessed using the Narcolepsy Severity Scale (NSS), 23 a self-administered 15-item scale evaluating the severity, frequency, and impact of five narcolepsy symptoms (EDS, cataplexy, hallucinations, sleep paralysis, and disrupted nighttime), the item regarding driving was ignored, and higher scores indicating the more severe symptoms.

| Anthropometric measurements
Height and weight at diagnosis were measured in upright position during the hospitalization and those at symptom onset were collected from the carnet de santé (record of all anthropometric, vaccinal, and health-related data about a child from birth by a pediatrician). Anthropometric data at a time-point close to symptom onset were considered for analysis when there were no available data at symptom onset. Symptom onset was defined as the occurrence of sleepiness because it appears before other symptoms, such as cataplexy in the population studied herein. The time of PSG evaluation corresponded to the diagnostic time. The BMI was computed at symptom onset and disease diagnosis. Overweight and obesity were defined as BMI was > +1 standard deviation (SD) and > +2 SD for sex and age in children >5 years old, respectively, and as BMI was > +2 SD and > +3 SD for sex and age in children ≤5 years old, respectively, in accordance with the criteria recommended by the World Health Organization (WHO). 24 BMI z-score (i.e., weight adjusted for height, sex, and age) was also computed at symptom onset and disease diagnosis. 25 No underweight child was included in our study.
The RWG phenotype was defined by a BMI z-score slope reported to one year (difference between the BMI z-score at symptom onset and at disease diagnosis divided by the diagnostic delay expressed in years) >0.67 SD; the non-RGW phenotype was defined by a BMI z-score slope reported to one year ≤0.67 SD. A similar criterion has been used in the evaluation of RGW in infancy and early childhood. 26,27 A gain >0.67 SD may be clinically interpreted as upward centile crossing through at least one of the centile bands in childhood growth charts (e.g., 2nd, 10th, 25th, 50th, 75th, 90th, and 98th centile lines). 26,28
Precocious puberty was determined by breast-tanner stage 2 in girls and testicular volume >4 ml in boys (genital tanner stage 2) and confirmed by the endocrinologist with plasma luteinizing hormone (LH), follicle stimulating hormone (FSH) levels, bone age, and pelvic ultrasound in girls if necessary.

| Personal history
Personal history of children regarding birth weight, narcolepsy history, current cataplexy symptoms, hypnagogic hallucinations, sleep paralysis, night eating, academic situation (school difficulties, grade repetition, and absenteeism), and H1N1 vaccination (yes/no answer for all these characteristics) were also collected by the pediatric sleep specialist. The socioeconomical level (SEL) was evaluated based on the parents' occupational levels.

| Follow-up
Patients were treated following the French recommendations for the management of narcoleptic patients (modafinil, methylphenidate, and sodium oxybate) associated or not with venlafaxine for cataplexy. 34,35 BMI, BMI z-score, AESS, and academic situation were collected at the last follow-up visit. All the patients were treated and followed by the same pediatrician sleep specialist (P.F). Only children for whom the duration of follow-up was ≥1 year were included in the analyses.
Overall, our primary objective was to compare at the diagnosis time, the narcoleptic symptoms and comorbidities (anthropometry, metabolic, and endocrinal characteristics) between RWG and non-RWG groups; the second objective was to study at diagnosis the polysomnographic characteristic of these two groups; finally, we evaluated if the pharmacological management and the follow-up outcomes (clinical and anthropometric changes and academic achievements) were different between these patients.

| Statistical analyses
Statistical analyses were conducted using the R software (version 3.6.3, Vienna, Austria).
Continuous variables were expressed as median (range).
Dichotomous and polytomous variables were expressed as count (percentage). Comparisons between groups (RWG vs non-RWG) were performed using the unpaired Wilcoxon tests for continuous A bivariate analysis was conducted and identified factors that were assessed in a multivariable linear model at disease diagnosis, in order to evaluate their association with the BMI z-score slope reported to one year adjusted for sex (these factors were BMI z-score at symptom onset, age at diagnosis, and Epworth Sleepiness Scale score at diagnosis). A square root transformation (sqrt) was used to correct the skewed distribution of the slope. For a better readability, figures report BMI z-score slope per year without square root transformation. Statistical significance value was set to a p-value <0.05.

| Demographic and general characteristics
A total of 84 narcoleptic pediatric patients were included in the study; there were 48 patients (57%) in the RWG group and 36 patients (43%) in the non-RWG group ( Table 1; Figure 1). The median (range) BMI z-score slope was 1.8 (0.7-6.2) for RWG patients and 0.2 (−0.6-0.6) for non-RGW patients, i.e., significantly higher in the RWG group (p < 0.001; Figure 2). There was a tendency for more boys in the RWG than no-RWG group, but no significant difference was reached. The median age at diagnosis was significantly lower for RWG patients (10.4 vs 12.9 years, p = 0.018). Additionally, the median diagnostic delay was shorter in the RWG group (1.1 vs 2.1 years, p < 0.001). CSF hypocretin levels were evaluated for 35 patients and were <110 pg/ml for all of them. There was no significant difference in terms of CSF hypocretin concentrations, HLA-DQB1*06:02 positivity, H1N1 vaccination prior to symptom onset, birth weight, and parents' socioeconomic levels between RWG and non-RWG groups ( Table 1).

| Narcoleptic symptoms and comorbidities
No difference regarding the severity of cataplexy (type 1, 2, 3) or frequency of different types of cataplexy was observed between the two groups (Supplementary Table 1). There was no difference between groups regarding hypnagogic hallucinations and sleep paralysis. Similarly, the frequency of night eating was not significantly different between groups. Compared to non-RWG patients, narcoleptic patients with RWG had a significantly higher median AESS score (17.0 vs 15.5, p = 0.013) and had higher % of patients with AESS score greater than 10 (100.0% vs 86.1%, p = 0.012) Additionally, narcoleptic patients with RWG had a higher ISI score (14.5 vs 11.0, p = 0.021) and tended to have a higher Conners score (12.5 vs 9.0, p = 0.056). The depression and severity of the disease scores were similar between patients with and without RWG ( Table 2). There was no significant difference in terms of academic problems (school difficulties, grade repetition, and absenteeism) between both groups (Supplementary Table 2).

| Anthropometric characteristics
The median BMI z-score was significantly higher among narcoleptic patients with RWG than that of non-RWG patients at disease diagnosis (3.5 vs 2.0, p = 0.001); however, no significant difference was was found between groups ( Table 3).

| Metabolic and endocrinal characteristics
There was no difference regarding the total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, fasting glucose, basal insulin, HOMA, leptin, ghrelin, TSH, T3, and T4 levels between groups. No difference regarding the blood pressure was noted between the two groups. A total of 5/29 (17.2%) RWG narcoleptic children displayed a MetS, whereas none did in the non-RWG group (p = 0.142). No difference in terms of precocious puberty or hypothyroidism was noted between groups (Supplementary Table 3).

| Sleep-related characteristics
No significant difference was noticed between the two groups in terms of nocturnal respiratory parameters and sleep characteristics such as total sleep duration, sleep quality, and alteration of sleep architecture ( Table 4). Regarding MSLT, it was evaluated with 5 nap opportunities for 13 patients and with 4 for the remaining 71 patients. There were 3 out of 84 patients who had a mean sleep latency ≥8 min (8, 8.18, 9.4), and all of them presented at least two SOREMPs. Two patients had only one SOREMPs while their mean sleep latency was 4 min (<8 min). There was no significant difference regarding mean sleep latency, REM latencies, and SOREMPs between groups ( Table 4).

| Follow-up
No follow-up after diagnosis occurred for 4 patients, and 3 patients were excluded as their follow-up lasted less than 1 year. Most narcoleptic children were followed-up until they were 18-year-old by the same pediatrician sleep specialist (P.F.) they consulted at diagnosis ( Figure 1). The age at the last follow-up visit was similar between groups. The follow-up duration was longer for RWG patients than that for non-RWG patients due to the earlier disease diagnosis among patients with RWG. At last follow-up, narcoleptic children with RWG still had a higher BMI z-score (3.4 vs 1.9, p = 0.011; Figure 3) and were more frequently obese (50.0% vs 21.2%, p = 0.033) compared TA B L E 1 Demographic and general characteristics among narcoleptic patients with and without rapid weight gain (RWG)

| Multivariable regression
The adjusted analysis, explaining 28% of the BMI z-score slope per year variability, found significant independent associations between the BMI z-score slope per year and the age at disease diagnosis (p < 0.001, Figure 4) and the Epworth Sleepiness Scale at diagnosis (p = 0.003, Figure 5). The higher the increase of BMI z-score slope, the lower the age at diagnosis and the worse the Epworth Sleepiness Score at diagnosis. There was no other significant effect.

| DISCUSS ION
Compared to non-RWG patients, despite a shorter diagnosis delay, RWG narcoleptic patients were younger, sleepier, and their prevalence of obesity was higher at disease diagnosis but not at symptom onset, which is indicative of a more accelerated pathological process. At last follow-up, both groups displayed improvement in sleepiness and school difficulties; however, RWG patients still had a higher BMI z-score and a higher prevalence of obesity compared to that of non-RWG patients, despite in the RWG group, a decrease F I G U R E 1 Flowchart in pediatric RWG and non-RWG narcolepsy. BMI, body mass index; RWG, rapid weight gain F I G U R E 2 BMI z-score slope (reported to one year) in pediatric narcoleptic patients with and without rapid weight gain. Each line corresponds to the BMI z-score slope (reported to one year) of a patient, which is the difference between the BMI z-score at symptom onset and at disease diagnosis divided by the delay in years between both. The red segments represent RWG patients (slope >0.67), the grey ones represent non-RWG patients (slope ≤0.67). The red bold line and the grey bold line correspond to the median BMI z-score slope (reported to one year) of RWG patients (slope =1.8) and non-RWG patients (slope = 0.2), respectively. The difference of the slope was represented using asterisk (*p < 0.05, **p < 0.01,***p < 0.001). BMI, body mass index; RWG, rapid weight gain in the proportion of obese patients compared to disease diagnosis ( Figure 6).
Studies investigating the prevalence and the pathophysiology of obesity in narcoleptic adults remain controversial. 36,37 An early age at disease diagnosis seems to increase the susceptibility of developing obesity. 8,12 Due to the fast growth, metabolism, and weight changes occurring during childhood and adolescence, the loss of hypocretin might yield exaggerated weight gain, perhaps due to relatively decreased activities and metabolic consumption. 13 Additionally, an impairment in histamine neurotransmission, from another brain wake-promoting system located in close proximity to hypocretin neurons, has been reported in the CSF of NT1 pediatric patients, 38 but not in those of adults. 39 Histaminergic neurons could also be involved in the rapid weight gain process as a mild obesity has been observed in knockout mice lacking either histamine or hypocretins. 40  Abbreviation: AESS, adapted Epworth Sleepiness Scale; CDI, Children's Depression Inventory; ISI, Insomnia Severity Index; NSS, Narcolepsy Severity Scale; RWG, rapid weight gain. The significance level was set at 5%.    Abbreviation: AHI, apnea-hypopnea index; REM, rapid eye movement; RWG, rapid weight gain; SOREMPs, sleep onset rapid eye movement periods; TST, total sleep time. The significance level was set at 5%.

TA B L E 3 Anthropometric characteristics of narcoleptic patients with and without rapid weight gain (RWG)
F I G U R E 3 BMI z-score progression over time in pediatric narcoleptic children with and without rapid weight gain The red line represents the BMI z-score progression in RWG patients, three red points represent the median BMI z-score at symptom onset, disease diagnosis, and last follow-up of RWG patients, respectively; the grey line represents the BMI z-score progression in non-RWG patients, three grey points represent the median BMI z-score at symptom onset, disease diagnosis, and last follow-up of non-RWG patients, respectively. The difference between RWG patients and non-RWG patients was presented using asterisk (*p < 0.05, **p < 0.01, ***p < 0.001) at symptom onset, disease diagnosis, and follow-up, respectively. The difference between symptom onset, disease diagnosis, and follow-up was presented using hashtag (# p < 0.05, # # p < 0.01, # # # p < 0.001) within RWG patients and within non-RWG patients, respectively. BMI, body mass index; RWG, rapid weight gain TA B L E 5 Follow-up of narcoleptic patients with and without rapid weight gain (RWG) +0.67 SD in weight-for-age, height-for-age, and weight-for-height zscores. 53 Using weight-for-age variation greater than +1 SD z-score has also been reported to assess rapid growth, 54 and did not change significantly the results presented here, except than the RWG children were also younger at disease onset. Moreover, to understand the mechanisms involved in the RWG phenotype, some data such as genetic, antenatal, and postnatal environmental risk factors, 28,53 but also serum and CSF proinflammatory cytokine levels and autoreactive CD8 + /CD4 + T-cells at symptom onset were missing. 45,46,55 In conclusion, despite a similar BMI z-score at symptom onset and a shorter diagnostic delay, narcoleptic pediatric patients with RWG were younger, sleepier, and more obese in proportion than non-RWG patients. Given that the autoimmune etiology of narcolepsy plays a critical role in its pathophysiology, RGW in narcoleptic children could reflect severer autoimmune-inflammatory attacks and could inspire the design of specific treatments such as immunological therapy in these patients. 56,57 RWG could represent a more severe phenotype of the disease with a long-term persistence of obesity at follow-up. As the persistence of obesity could represent a risk of developing metabolic and cardiorespiratory complications and poor quality of life, we recommend to assess the growth curve of the narcoleptic children to determine if they display a RWG profile, and to ensure a prompt and more offensive management to prevent obesity and its complications.

ACK N OWLED G M ENTS
The authors thank all the patients for their collaboration and Hélène Boyer for help in English revision of this manuscript. Min Zhang is supported by a PhD fellowship from China Scholarship Council.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available upon request to the authors.

F I G U R E 6
Characterization of rapid weight gain (RWG) and non-RWG phenotype of childhood narcolepsy. Compared to non-RWG patients, RWG narcoleptic children had shorter diagnosis delay, were younger, more obese, sleepier with more dyssomnia at diagnosis, but were also more obese at follow-up. RGW could represent a maker of a more severe phenotype of childhood narcolepsy. BMI, body mass index; RWG, rapid weight gain