The 2nd Edition of the International Classification of Sleep Disorders (ICSD-2), in the category of hypersomnia of central origin, subdivides narcolepsy into two groups: narcolepsy with cataplexy (NA-CA) and narcolepsy without cataplexy (NA w/o CA). NA w/o CA is characterized by excessive daytime sleepiness and abnormal manifestations of rapid eye movement (REM) sleep in common with NA-CA, but no cataplexy. Patients with NA w/o CA have frequent sleep-onset REM periods, as do those with NA-CA, as revealed by performance of the multiple sleep latency test. A population-based study suggested that the prevalence of NA w/o CA is 36% of the prevalence of narcolepsy as a whole, corresponding to a point prevalence of 0.02%.1 NA-CA is tightly associated with HLA-DQB1*06:02 and orexin (hypocretin) deficiency. Almost all patients with NA-CA in many populations consistently carry DQB1*06:02, while approximately 12% of Japanese, 25% of Caucasian and 38% of African American healthy individuals are DQB1*06:02-positive.24 Low levels of orexin A in cerebrospinal fluid (CSF) (<110 pg/ml) are commonly observed in patients with NA-CA.5,6 Regarding NA w/o CA, positivity of HLA-DQB1*06:02 (30–50%) is also higher than that in the general population,710 but less than that in NA-CA. However, only approximately 20% of patients with NA w/o CA have low levels of CSF orexin A,6,10 indicating that the etiology of the majority of NA w/o CA is still unknown.

There have been a number of studies of HLA in NA-CA; results indicated that HLA-DQB1 alleles other than DQB1*06:02 modulate susceptibility or resistance to NA-CA. DQB1*06:01 and DQB1*05:01 in the Korean and Japanese populations and DQB1*06:03 in European populations are protective against NA-CA,2,7,1114 whereas individuals with DQB1*03:01 and DQB1*03:02 are at an increased risk.2,7,11,1317 In the present study, to test for associations of HLA-DQB1 alleles in NA w/o CA, we performed an association study for HLA-DQB1 in 160 Japanese patients with NA w/o CA and 1,418 control subjects.

Idiopathic hypersomnia (IHS) is a sleep disorder of presumed central nervous system origin that is associated with excessive daytime sleepiness consisting of prolonged non-REM sleep episodes. Daytime naps of IHS patients tend to be longer and less refreshing than those of NA-CA patients. IHS is a rare disease, representing 8:10 to 1:10 patients with NA-CA. This suggests that the prevalence of IHS approximates 0.005%.18 The ICSD-2 describes two clinical forms of IHS by the difference in nocturnal sleep time: IHS with long sleep time (IHS-LST) and IHS without long sleep time (IHS w/o LST). The nocturnal sleep time of IHS-LST is prolonged to at least 10 h, while that of IHS w/o LST is either normal or slightly prolonged (less than 10 h). CSF orexin A levels in IHS are normal.6 The cause and pathogenesis of IHS remain largely unknown. NA w/o CA and IHS w/o LST have several common characteristics except for REM-related symptoms. Distinguishing NA w/o CA and IHS w/o LST is impossible without the multiple sleep latency test to identify sleep-onset REM periods. According to the ICSD-2, the diagnosis is based on the number of sleep-onset REM periods, two or more in the former and less than two in the latter. In the present study, we tested whether HLA-DQB1 alleles have an influence on susceptibility to IHS w/o LST and IHS-LST.

A total of 346 Japanese patients and 1,418 Japanese healthy controls were included in this study. NA w/o CA, IHS w/o LST and IHS-LST were diagnosed according to the ICSD-2 criteria. The patient groups consisted of NA w/o CA (n=160), IHS w/o LST (n=118) and IHS-LST (n=68). We utilized HLA data of healthy individuals, who have been previously studied for disease association analyses.17,19,20 In addition, to assess genetic similarities between the above hypersomnia disorders and NA-CA, HLA-DQB1 data from 664 patients with NA-CA were utilized.17 All of the patients and controls were mainland Japanese and gave written informed consent. This study was approved by the local institutional review boards at participating institutions. Typing for the HLA-DQB1 locus was performed by a Luminex Multi-Analyte Profiling system (xMAP) with WAKFlow HLA typing kits (Wakunaga Pharmaceutical, Wakunaga, Hiroshima, Japan). Comparisons of frequencies were performed using the Chi-square test or Fisher’s Exact test as appropriate. To account for multiple testing, the significance level was adjusted by the number of HLA alleles with allele frequencies no less than 0.5% in controls (12 for HLA-DQB1 alleles). The significance level was set to be P<4.2×10−3 (0.05/12). If any of the four cells was zero, the Woolf-Haldane correction was applied (adding 0.5 to all cells). An association analysis controlling for the effects of DQB1*06:02 was needed to test the other HLA-DQB1 alleles. Specifically, the analysis was performed using counts of the other HLA-DQB1 alleles remaining after excluding allele counts of DQB1*06:02 from both cases and controls, which is the relative predispositional effects method.21 Briefly, frequencies and ORs were calculated for the alleles carried by the non-DQB1*06:02 chromosomes. To determine whether there was a different allelic distribution between two groups, the overall frequency distribution of HLA-DQB1 alleles in one group was compared with the distribution in another group by using the global Chi-square test with 12 degrees-of-freedom.

HLA-DQB1 allele frequencies of 160 patients with NA w/o CA and 1,418 control subjects were determined, and are shown in Table 1. DQB1*06:02 was significantly associated with NA w/o CA (allele frequency: 16.6 vs. 7.8% in controls, P=1.1×10−7, OR=2.36). Regarding carrier frequencies for HLA-DQB1, 31.3% of the patients carried DQB1*06:02, vs. 14.7% of controls (P=7.6×10−8, OR=2.64) (Supplementary Table 1). We tested whether there were other independent HLA-DQB1 associations aside from DQB1*06:02 using the relative predispositional effect method.21 DQB1*06:04 showed a marginal association (P=5.9×10−3, OR=1.80) (Table 2). However, the P value did not reach the threshold corrected for multiple comparisons. NA w/o CA was subdivided into DQB1*06:02-positive and -negative groups for further analyses, because previous studies suggested that these two disease groups have different etiology, clinical characteristics and electroencephalographic findings.10,2224 The analyses were performed by controlling for the effects of DQB1*06:02 (Table 2). No significant association with HLA-DQB1 alleles was found in DQB1*06:02-positive NA w/o CA. DQB1*06:04 was nominally associated with DQB1*06:02-negative NA w/o CA patients (P=5.6×10−3, OR=1.88).

Table 1 HLA-DQB1 allele frequencies of patients with NA w/o CA
Table 2 Frequencies of HLA-DQB1 alleles after removal of DQB1*06:02 effects

Next, we tested whether HLA-DQB1 allele distributions (except for DQB1*06:02) in DQB1*06:02-positive and -negative NA w/o CA patients showed similarities with those of NA-CA patients (Table 2 and Supplementary Figure 1). The allele distribution of DQB1*06:02-negative NA w/o CA was significantly different from that of NA-CA (P=5.8×10−7). On the other hand, no significant difference was observed in the distribution between DQB1*06:02-positive NA w/o CA and NA-CA (P=0.97). When we focused on HLA-DQB1 alleles with frequencies no less than 5% in DQB1*06:02-positive NA w/o CA, all the ORs of the alleles were found in the same direction as those of NA-CA (Table 2). These results suggest that a common etiological pathway might exist for DQB1*06:02-positive NA w/o CA and NA-CA, whereas DQB1*06:02-negative NA w/o CA has a different etiological pathway from NA-CA. Orexin A in the CSF of typical NA-CA patients is known to be reduced or undetectable.5,6 A minority of NA w/o CA patients have low concentrations of CSF orexin A levels, although a majority of NA w/o CA patients have normal concentrations;6,10 thus, an etiologic heterogeneity in NA w/o CA is suggested. In addition, most of the NA w/o CA patients with low CSF orexin A are DQB1*06:02-positive, as in NA-CA.6,10 In contrast, DQB1*06:02 positivity of patients with normal CSF orexin A is not higher than that seen in the general population.10 Therefore, the heterogeneity in NA w/o CA might be divided by DQB1*06:02 positivity, and DQB1*06:02-positive NA w/o CA might have the same etiology as NA-CA from the view point of orexin deficiency. In addition, clinical characteristics and electroencephalographic findings have been reported to be different between DQB1*06:02-positive and -negative NA w/o CA groups.23,24 These findings correspond well to our result.

Patients with IHS w/o LST and IHS-LST were also typed for HLA-DQB1 in the present study. HLA-DQB1 allele and carrier frequencies are shown in Table 3 and Supplementary Table 2, respectively. DQB1*06:02, known to be associated with NA-CA and NA w/o CA, was not associated with IHS w/o LST or IHS-LST. Although DQB1*05:02 (P=6.3×10−3) and DQB1*03:01 (P=0.04) showed nominally significant associations with IHS-LST, there were no HLA-DQB1 alleles that reached the threshold after correction for multiple comparisons. The similarity of HLA-DQB1 allele distribution between NA-CA and IHS w/o LST or IHS-LST was tested after controlling for the effects of DQB1*06:02, and significant differences were found: for IHS w/o LST: P=2.1×10−4 and for IHS-LST: P=2.2×10−8. Taken together, these results indicate that IHS w/o LST and IHS-LST are caused by different etiological genetic factors than those that give rise to NA-CA.

Table 3 HLA-DQB1 allele frequencies of patients with IHS w/o LST and IHS-LST

The International Classification of Sleep Disorders was recently revised for the 3rd Edition (ICSD-3). When we had recruited patient samples, the 2nd edition (ICSD-2) was used. Main differences between the ICSD-2 and ICSD-3 regarding NA w/o CA, IHS w/o LST and IHS-LST are as follows. The terminology has been changed from ‘narcolepsy without cataplexy (NA w/o CA)’ to ‘narcolepsy type 2’. The concept of NA w/o CA and narcolepsy type 2 is almost the same. IHS w/o LST and IHS-LST were unified to ‘idiopathic hypersomnia (IHS)’ regardless of the extension of the sleep time. Therefore, we combined our IHS w/o LST and IHS-LST data to analyze the HLA-DQB1 data as IHS (Supplementary Tables 3 and 4). As a result, no significant association was observed between each HLA-DQB1 allele and IHS. Distribution of HLA-DQB1 alleles in IHS was also significantly different from that of NA-CA, even after the effect of DQB1-06:02 was controlled (P=8.6×10−9).

To conclude, the association of DQB1*06:02 in NA w/o CA was confirmed. Our results also suggested an immunological pathogenesis of DQB1*06:02-positive NA w/o CA, which is similar to that of NA-CA. DQB1*06:02-negative NA w/o CA, IHS w/o LST and IHS-LST may have a different etiology, which is not well understood.