Narcolepsy is a serious chronic neurological sleep disorder affecting between 0.02–0.05% of the general Caucasian population 1. It is characterized by excessive daytime sleepiness, cataplexy, hypnagogic hallucination, sleep paralysis, and nocturnal fragmented/disorganized sleep 23. Narcolepsy causes cognitive dysfunction, low academic performance and interpersonal problems 4. Cataplexy is a pathognomonic clinical symptom required for the diagnosis of narcolepsy, according to the International Classification of Sleep Disorders 5. It is a symptom that occurs when the muscle tension, in various areas of the body, is suddenly decreased involuntarily and lasts from few seconds to several minutes. Cataplexy is generally induced by laughter, excitement anger, and other emotional changes 5. Mutations in genes of the hypocretin (orexin) neurotransmitter system cause narcoleptic symptoms in animal models 6. Although most patients with narcolepsy-cataplexy have a reduction of hypocretin concentration in the cerebrospinal fluid 7, mutations or polymorphisms in hypocretin-related genes are extremely rare 8. A potential autoimmune mechanism has been suggested, supported by the finding that in most narcolepsy patients, 80–90% of hypocretin cells in the hypothalamus were destroyed 9. The lost of hypocretin neurons 10 has also been shown in a few post-mortem cases. In addition, the transcription of preprohypocretin mRNA was significantly decreased in the brain of these patients 11. This wake-promoting neuropeptide is involved in sleep regulation, energy homeostasis, reward-seeking, learning, and memory; and it is also involved in the body temperature regulation, endocrine function, and the cardiovascular system, among other systems 12. Recent studies also indicate that hypocretin/orexin neurons can alter their intrinsic electrical activity according to fluctuations in the levels of nutrients and appetite-regulating hormones 13.

Narcolepsy transmission is polygenic, environmentally influenced and genetic factors play an important role in its expression. Family studies indicate the presence of a 20–40 times increased risk of disease expression in first-degree relatives, and monozygotic twin studies showed that concordance is partial (25–31%) 1. Early onset French and Chinese patients suffering from narcolepsy have shown a positive family history as compared with late-onset patients 1415, suggesting that the disease may be more likely due to genetic factors in this subgroup of patients.

Based on the strong HLA association, family study segregation 1617 and the finding of reduction of hypocretin-1 levels in the cerebrospinal fluid of DQB1*0602 positive patients 1819, an autoimmune mediated destruction of hypothalamic neurons secreting hypocretin has been suggested as the cause of the disease. However, serum autoantibody markers have not been detected and no immunological abnormalities have been found in patients with narcolepsy 20.

The disease is strongly associated in Caucasians and Japanese with the DRB1*1501-DQA1*0102-DQB1*0602 haplotype. In African Americans, it is associated with DQB1*0602 haplotypes bearing different DRB1 alleles (*1101 and *1503) suggesting that DQA1 and DQB1 play a primary role in susceptibility 17. DQB1*0602 rather than DRB1*1501 has been found increased in the patients, indicating that the disease susceptibility allele or locus is within, or, in the vicinity of the DQ region 21. Haplotype analysis and contiguous genomic sequencing across the region have identified no other candidate gene 22. Two-five fold increased risk in DQB1*0602 homozygous vs. heterozygote patients has been demonstrated in different ethnic groups 23. DQB1*0301/*0602 carriers are also at an increased risk, whereas DQB1*0602/*0601 and DQB1*0602/*0501 heterozygotes have a lower disease risk 232425.

Several authors reported that 85 to 95% of patients with narcolepsy carry DQB1*0602 when cataplexy is clinically typical or severe. However, only 40–60% of the patients are DQB1*0602 when mild, atypical or no cataplexy exists 2526. TNF alpha region has also been claimed to be involved in susceptibility independently of class II loci 27.

Genetic factors in other chromosomes have also been implicated 28. A gender dimorphism and a strong effect of the catechol-O-methyltransferase (COMT) genes seem to influence symptoms. COMT genotype distribution between male and female patients was associated with the response to modafinil in Caucasians, since the optimal dose of modafinil was approximately 100 mg lower in females with narcolepsy, suggesting that females are better responders to the drug 29.

The HLA genetic profile of Mexican and other Central American patients has not been published; therefore the aim of this study was to investigate the class II-DRB1/DQB1 allele distribution in a group of sporadic Mexican Mestizo patients with narcolepsy and to explore if the HLA association is gender related.

The results of this study are considered preliminary, due to the small sample size. However, it is important to mention that we are not "Hypothesis generating" but "hypothesis testing". As demonstrated by many statistician experts in the HLA field: "If an association is detected in the first case, it can be tested and confirmed in the latter without having to correct in multiple comparisons" 33. Nevertheless, future studies in Hispanic admixed populations are needed to confirm the presence of other HLA allele associations. It is worth to mention, that DRB1/DQB1 association has been tested and repeatedly demonstrated by many authors in different ethnic groups 23242534; therefore, the associations shown here, are real. In this study, we report on the genetic profile in a sample of narcoleptic Mestizo patients. DRB1*1501 (OR = 8.2; pc < 0.0001) and DQB1*0602 (OR = 8.4; pc < 0.0001) were the strongest associated alleles found in narcoleptic Mexicans, similarly to African, White Americans and different Oriental groups 23242534353637. Fifteen of the 27 DRB1 typed patients, were positive for DRB1*1501 (allele frequency = 27.8 vs. 4.4% in the controls) and 21 of the 32 DQB1 typed patients were positive for DQB1*0602 (allele frequency = 32.8% vs. 5.4% in the controls). In five DQB1 typed patients, DRB1 was not typed because of insufficient DNA, thus we were not able to assemble the DRB1-DQB1 combinations in them. Interestingly, the DRB1*1501-DQB1*0602 haplotype was present in most of the DQB1*0602 positive patients but other DR2 allele combinations were also found. Indeed, two narcoleptic patients had DRB1*1502-DQB1*0602 and two had DRB1*1503-DQB1*0602. A DRB1*1503 but not DRB1*1502 association with DQB1*0602 has been reported in patients from Martinique 38 and in African Americans 23. These results show, beyond doubt, that DQB1*0602, rather than DRB1*1501 is the major narcolepsy susceptibility allele in Mestizos. Interestingly, none of the DR52 associated-DQB1*0602 haplotypes were found in Mexican patients, but their frequency is low in the general healthy population. As an example, in 160 Mexican Mestizo healthy individuals typed in our laboratory, the haplotype frequency was 0.31% for each of the following combinations: DRB1*1101-DQB1*0602, DRB1*1201-DQB1*0602 and DRB1*1301-DQB1*0602 (unpublished data). Some of these haplotypes have been reported in narcoleptic patients in other populations, most notably in African Americans 23.

No increase in DQB1*0602 homozygosity was found, as previously reported in White Americans and in African American patients, in whom a two to four fold higher risk has been described, compared to heterozygotes 23. The same has been shown in Japanese patients 24. This fact cannot be explained only based on the different allele frequencies of DQB1*0602 across ethnic groups, since the frequency shown in Japanese (AF = 6.4%) 24 was similar to the one found in the control group of the present study (AF = 5.5%). The lack of homozygote patients may be due to the reduced number of cases. Thus, again, a larger number is needed to confirm these results.

As in other populations, other DQB1 alleles, beside DQB1*0602, influence narcolepsy susceptibility 2439. The analysis of DQB1 distribution in DQB1*0602 negative patients showed that DQB1*0301 (OR = 2.7, p = 0.03) was significantly increased in this subgroup of patients. DQB1*0301 had also the second strongest susceptibility effect, after DQB1*0602 in Africans, Japanese and White Americans 24. In Mexican patients, DQB1*0301 occurred in the context of several HLA haplotypes that included DRB1*1101, *1303, *1304, *0806 and *1602; however the number of patients in this group was insufficient to perform additional comparisons and to explain the possible independent contribution of the mentioned DRB1 alleles to susceptibility. None of these patients had the DRB1*04-DQB1*0301, found associated in White Americans 24. Genotype distribution in patients and controls showed that DQB1*0602/DQB1*0301 conferred the highest risk for susceptibility (OR = 11.4) compared to DQB1*0602/X (non *0301) (OR = 9.4). The former combination was also described as the one with the highest risk for the development of narcolepsy, across three different ethnic groups 24.

DRB1*0407, which is the most frequent allele in Mexican population, seemed to be linked to protection in the present study. This allele is in strong linkage disequilibrium with DQB1*0302 in Mexicans and the haplotype frequency of DRB1*0407-DQB1*0302 is 13.1% 31. In Koreans DRB1*0406-DQB1*0302 was found protective in patients with narcolepsy 39. None of these studies confirmed a possible effect of DRB1*04 in susceptibility as previously shown in Whites and Japanese 24. It may be claimed that the DQB1 locus may also be involved in protection, since in Korean as well as in Mexican patients, DQB1*0302 was decreased, although combined with *0406 in Koreans and with *0407 in Mexicans. DQB1*0601 has also been associated with protection in Koreans 2539 and Japanese 40, but the latter was not related with protection in Mexicans, perhaps due to its low frequency (AF = 0.5%) 31. Similar findings have been published recently in Koreans, where again, DRB1*0406 was found negatively associated and DQB1*0301 was described as a susceptibility allele. The authors claim based on their own work, and previous work, that a remarkable consistency of the HLA association pattern across multiple ethnic groups and cultures exists 41. Thus, even if our sample size is small and the results of protection and secondary association may be regarded as preliminary, our data are consistent with those published 21232425263941.

Interestingly, the overall positive rate for DQB1*0602 in Mexican patients was 65.6%, while in Japanese, White Americans and African Americans; the rate is between 75–80% 24. To analyse if this difference was significant or not, we performed a statistical comparison between patients and controls from the present study and those from the Mignot et al. study 24. No significant difference, regarding DQB1*0602 distribution, was found between Japanese and Mexican controls but we did find a significant deviation when comparing Mexican controls with White or African Americans healthy people (p = 0.001 and p = 0.00000001, respectively). These differences are due, undoubtedly, to the lower frequency of DQB1*0602 existing in the general Mexican population 42, which is similar to the one found in Orientals, but it is lower than the frequency in Caucasian and African Americans 43. The lower frequency of the DQB1*0602 allele in Mexican patients compared with African (p = 0.0007), Caucasian (p = 0.003) and Japanese (p = 0.0009) patients, is due to our small number of cases.

Gender stratification showed a differential distribution of DRB1 and DQB1 alleles. The patients selection was unbiased since no significant difference was found when distribution of males Vs. females was compared (Z = 0.561, p = 0.29). To demonstrate if the gender selection among the controls was biased, we performed several analyses comparing the mean allele frequency in the overall control group with the frequency for HLA DQB1*0602 among male and female patients. However, the distribution of DQB1*0602 alleles among female and male controls was found in the limit of the significance (X²Y = 3.841, p = 0.05; data not shown). Therefore, it must be mentioned that even if the p value was p = 0.05, the control selection may have been biased. When HLA distribution was compared matching for gender, DRB1*1501 (OR = 15.8; pc < 0.0001) and DQB1*0602 (OR = 19.8; p < 0.0001) showed a higher risk in female patients, although the same alleles were significantly increased in males, but with less intensity. This stronger female HLA association was described for the first time. However, it is important to confirm these data with a larger sample size and in different ethnic groups. Thus, the gender association reported here should be regarded as preliminary. Even if we have a stronger HLA DRB1/DQB1 association in females, this does not imply, by any means, that females are more or less affected than males; this only means that there is a stronger HLA genetic predisposition to narcolepsy in females than in males. A better response in affected women to certain drugs 29 is not contradictory at all, with the HLA/female association.

The explanation for this gender specific association is not clear, but a sexual dimorphism and a strong effect of the COMT genotype on disease severity and response to modafinil have been shown 44. Females with narcolepsy with high COMT activity fell asleep twice as fast as those with low COMT activity during the multiple sleep latency test, while the opposite was true for men 45. A gender difference in body weight gain and leptin signaling in hypocretin/orexin deficient mouse models has been also claimed 46. Obesity was more prominent in females in both preprohypocretin knockout mice and orexin/ataxin-3 transgenic narcoleptic mice and was associated with higher serum leptin levels, suggesting a partial leptin resistance 46.

In conclusion, this is the first study reported in Mexican patients with narcolepsy-cataplexy; the profile of HLA class II allele associations was found to be complex. As in other diseases such as type I diabetes, multiple sclerosis and others, studied by our group in Mexican Mestizos, the HLA pattern in these diseases was somewhat distinct from the association typically found in Caucasians or Blacks and Orientals 4748, illustrating the importance of analysing MHC associations in populations with different ethnic backgrounds. We also demonstrated an HLA associated sexual dimorphism in this population and a protective allele effect which was also shown in some Oriental populations. Insights on the different HLA associations in different ethnic groups may prove to be an important asset in the investigation of genetic factors and the molecular mechanisms of disease expression. This knowledge may be important for the design of predictive, therapeutic and perhaps preventive approaches.

The authors declare that they have no competing interests.

CA Drafted the manuscript and supervised the laboratory work. LL Collaborated in drafting the manuscript and extracted the DNA at Stanford University. HF-A Carry out the statistical analysis and revised the drafted manuscript. MV Collaborated with the technical HLA DNA typing and revised the drafted manuscript. AM Collaborated with the technical HLA DNA typing and revised the drafted manuscript. EM Selected the clinical cases and wrote the criteria for all Centres and for the 13th International Histocompatibility Workshop and revised the drafted manuscript. RH Diagnosed and selected the cases from The Instituto Nacional de Neurologia and revised the drafted manuscript. HB Diagnosed and selected the cases from and revised the drafted manuscript. CG Coordinated all the work, revised the results and statistical analysis, revised the drafted manuscript thoroughly; revised and corrected the final version. All authors read and approved the final manuscript.