Evaluation of clinical and genetic factors in obstructive sleep apnoea| ACTA Otorhinolaryngologica Italica

Abstract

Purpose. To evaluate the correlation between several presumed candidate genes for obstructive sleep apnoea (OSA) and clinical OSA phenotypes and propose a predictive comprehensive model for diagnosis of OSA.
Methods. This case-control study compared polysomnographic patterns, clinical data, morbidities, dental factors and genetic data for polymorphisms in PER3, BDNF, NRXN3, APOE, HCRTR2, MC4R between confirmed OSA cases and ethnically matched clinically unaffected controls. A logistic regression model was developed to predict OSA using the combined data.
Results. The cohort consisted of 161 OSA cases and 81 controls. Mean age of cases was 53.5 ± 14.0 years, mostly males (57%) and mean body mass index (BMI) of 27.5 ± 4.3 kg/ m2. None of the genotyped markers showed a statistically significant association with OSA after adjusting for age and BMI. A predictive algorithm included the variables gender, age, snoring, hypertension, mouth breathing and number of T alleles of PER3 (rs228729) presenting 76.5% specificity and 71.6% sensitivity.
Conclusions. No genetic variant tested showed a statistically significant association with OSA phenotype. Logistic regression analysis resulted in a predictive model for diagnosing OSA that, if validated by larger prospective studies, could be applied clinically to allow risk stratification for OSA.

Introduction

Obstructive sleep apnoea (OSA) is characterised by recurrent events of upper airway obstruction during sleep associated with clinical signs and symptoms such as excessive sleepiness, snoring, choking, and breathing interruptions during sleep ¹. OSA has a high prevalence ² and is related to increased morbidity and mortality from all causes ³.

Although considerable progress has been made in understanding the pathophysiology of OSA ^4,5, the precise mechanisms that lead to upper airway obstruction are not fully understood ⁶. Despite the heterogeneity and complexity of OSA, diagnosis, assessment of severity and management of OSA remain linked to a single indicator, the apnoea hypopnea index (AHI) ⁶.

Several factors that increase the risk of OSA are well established, such as obesity, body fat distribution, craniofacial morphology, and neural control of the upper airway muscles ^5,7. These risk factors are, in part, genetically determined. Identification of genes associated with risk factors for OSA will undoubtedly help to understand the molecular mechanisms underlying the disorder and aid in the management of OSA ^8-11.

The discovery of genotypic and clinical markers can help in the early detection of the disease, as well as in the prevention and personalised treatment of this pathology. Thus, the aim of this study was to evaluate the correlation between several presumed candidate genes, phenotypes and clinical data that may contribute to the development of OSA and to propose a comprehensive predictive model for its diagnosis.

Materials and methods

Participant identification and evaluation

A case-control study was carried out in a medical-dental clinic in Belo Horizonte, Brazil, and recruitment period was August 2018 to January 2020. Study participants were divided into two groups: patients diagnosed with OSA (case group) and non-OSA participants (control group). Each participant underwent complete OSA-focused clinical evaluation and polysomnography. The following inclusion criteria were used for defining OSA cases: diagnosis of OSA according to the American Academy of Sleep Medicine criteria ¹² and absence of craniofacial dysmorphism, genetic syndromes with OSA as part of the spectrum of manifestations, drug and alcohol abuse, psychiatric disorders, and dementia; age 18 to 85 years and body mass index (BMI) ≤ 35 kg/m². The control group consisted of individuals referred due to clinical suspicion of OSA, but who did not meet diagnostic criteria (AHI < 5 per hour). For stratification of cases and controls the following variables were used: socioeconomic status, sex, age, height, weight, and BMI (≤ 35 kg/m²). BMI was limited to ≤ 35 kg/m² and age to < 85 years to avoid increased AHI levels. In addition, dental data such as the presence of self-reported bruxism, pain in masticatory muscles, noise upon movement of the temporomandibular joint, tongue size, floor of mouth (i.e., sublingual space), presence of open or crossed bite were also analysed.

Genetic analysis

Genomic DNA was extracted from peripheral blood ¹³ or saliva ¹⁴. 20 ng of DNA of each genotyped individual was used for TaqMan SNP genotyping assays for PER3 gene variants (rs228697, rs228727, rs228729 and rs10462020), BDNF (rs6265), NRXN3 (rs10146997), APOE (rs7412 and rs429358), HCRTR2 (rs2653349), MC4R (rs17782313) according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA). Genotyping was performed by real-time polymerase chain reaction using allelic discrimination using a Stratagene Mx3005 instrument (MxPro QPCR System, 2007 Software, La Jolla, CA). At least 10% of the samples were genotyped twice for quality control.

Data analysis

Data were analysed using the statistical program R (version 3.6.3). In univariate analysis, logistic regression models with a logarithmic link function were adjusted using the participant’s diagnosis as a dependent variable (case-control) and as an independent variable all clinical-epidemiological variables using the generalised linear model function. For genetic variables, the analysis was carried out based on five genetic models: co-dominance, dominance, recessive, over dominant and log-additive with the SNPassoc package.

Predictive OSA model using logistic regression

A logistic regression model was developed to predict OSA. The variables that reached a level of significance less than 0.20 in the univariate analysis were considered suitable for the final model. However, only variables with p < 0.05 remained in the multivariate logistic regression model.

Results

Overall, 161 OSA cases and 81 controls were included. The mean age of the participants was 51.5 ± 14.3 years with a male predominance (57.0%) and a mean BMI of 26.6 ± 4.3 kg/m². The mean age of cases was 53.5 ± 14.1 years and mean BMI was 27.5 ± 4.4 kg/m²; mean age of the controls was 47.4 ± 13.9 years and the mean BMI was 24.8 ± 3.8 kg/m². The polysomnographic variables of the sample participants showed a mean of AHI 17.05 ev/h, AHI in REM 21.05 ev/h, minimum saturation 82.01%, desaturation index 16.13/h. Sociodemographic, clinical, and dental variables are presented in Table I. Dental data were not available in the medical records of 16% patients.

Polysomnographic data showed a significant difference between cases and controls for AHI, AHI REM, AHI NREM, AHI dorsal, AHI non-dorsal, mean time of apnoea/hypopnoea, score 3 of arousal threshold, minimum saturation, and desaturation index (Tab. II).

Patients with OSA reported significantly more persistent nocturnal snoring than controls, with rates of 34.3% and 17.4% (p < 0.001), respectively. In addition, there was a significant difference between cases and controls for the presence of hypertension (23.1% vs 6.2%, p = 0.015), lung disease self-reported (asthma, bronchitis, chronic obstructive pulmonary disease) (3.3% vs 0.4%, p = 0.038) and presence of mouth breathing (28.5% vs 4.9%, p < 0.001) in patients with OSA compared with controls. Regarding dental data, there was a significant difference between cases and controls for enlarged tongue (54.5% vs 16.5%, p = 0.000), high palate (54.5% vs 18.2%, p = 0.003) and Angle class II malocclusion (27.3% vs 6.2%, p = 0.007) in patients with OSA (Tab. III).

Univariate analysis results revealed a significant difference for gender, age, weight and BMI between cases and controls (Tab. III).

Analysis of genetic factors

BDNF (rs6265) and PER3 (rs228729) were the only variants that showed a significant association with OSA (p = 0.013; p < 0.001, respectively). However, after adjusting for age and BMI, BDNF (rs6265) and PER3 (rs228729) no longer exhibited a significant association with OSA phenotype (p = 0.062; p = 0.066, respectively) (Tab. IV). None of the other genotypes had an association between the tested genotype and OSA phenotype (Tab. IV).

Prediction of the presence of OSA using logistic regression model

The values obtained in the multivariate logistic regression model (Tab. V) indicated that the probability of a patient having OSA can be best predicted in the equation derived from the final model:

z = -3.705-0.855 × (X1) + 0.031 × (X2) + 1.643 × (X3) + 1.105 × (X4) + 1.093 × (X5) + 1.209 × (X6) + 0.528 × (X7)

Where: Dx = diagnosis; exp = exposure; X1 = 1, if the gender is female, otherwise 0; X2 = age of the patient in years; X3 = 1, if subjective snoring is present, otherwise 0; X4 = 1, if hypertension is present, otherwise 0; X5 = 1, if not breathing well through the nose, otherwise 0; X6 = number of T alleles of PER3 (rs228729).

The area under the ROC curve was 79.4% (CI 95% 72.2%-86.6%) suggesting that the model achieves 76.5% specificity and 71.6% sensitivity assuming a 0.763 cutoff bridge (Fig. 1).

Discussion

In this report, the contribution and association of sequence variants of several OSA candidate genes ^15-20 was tested in Brazilian OSA cases. None of these variants was significantly associated with OSA phenotype.

The choice of variants for the present study was based on previous studies. For example, Canales et al. ²³ found an association between OSA and decreased morning expression of clock genes, such as PER3. Yuksekkaya et al. ²¹ reported that the BDNF rs6265 (196G/G) genotype may be useful to assess OSA in non-obese patients. Furthermore, findings showed that BDNF, but not NGF, was significantly increased in a subpopulation of muscle fibres in patients with snoring and OSA ²⁰. The MC4R gene is largely expressed in the hypothalamus and is intimately involved in appetite regulation, autonomic and endocrine functions, and insulin resistance. OSA is one of the deleterious consequences of obesity. A case report of a child with obesity at 23 months of age showed homozygous mutations of the MC4R gene (Pro299his) through DNA sequencing of the genes involved in early-onset obesity. Overnight polysomnography was significant for severe OSA (AHI 36 events/h and oxygen saturation 50%). The authors suggested that future research may provide more information on possible associations between MC4R and OSA ¹⁷. Central abdominal fat is a strong risk factor for diabetes, cardiovascular disease and OSA. Common variants in NRXN3 are associated with waist circumference, BMI and obesity ²⁴. A case-control study investigated whether variations in the APOE-ε gene were associated with craniofacial changes, AHI and BMI in patients with OSA. The polymorphisms that define the APOE-ε1-4 allele rs429358 and rs7412 were genotyped. There was no association of the APOE-ε4 allele with facial profile among these patients with OSA. However, the authors cautioned that in relation to genetic research, this study was underpowered due to the small sample size ¹⁸. The orexin receptor type 2 (Ox2R or OX2), also known as hypocretin receptor type 2 (HcrtR2), is a protein that in humans is encoded by the HCRTR2 gene. Peever et al. ¹⁵ demonstrated that genioglossus activity was increased by intracerebroventricular injection of orexin, therefore reductions in orexin may contribute to the suppression of upper airway dilator activity, which may facilitate OSA.

A previous study evaluated the association of the BDNF variant rs6265 (G > A) with OSA ²¹. To avoid the confounding effect of obesity, the authors divided participants into four groups based on the presence/absence of OSA and/or obesity. No significant differences were observed regarding the BDNF (rs6265) polymorphism between patients with and without OSA, and there was also no significant association with the BDNF gene and OSA polymorphism in a regression analysis. In our study, unlike the study by Yüksekkaya et al. ²¹, age and BMI did affect the results, since BDNF 270C/T was significantly associated with OSA in a codominant, dominant, recessive, and log-additive model, although after adjusting for age and BMI there was no longer a significant association with OSA.

Several studies in humans have suggested that PER3 plays a key role in maintaining circadian rhythm. Although there are still few studies on the relationships between circadian clock genes and OSA, it is important to note that the sleep-wake cycle can modulate circadian clock genes and, in turn, the sleep cycle circadian rhythm can affect the occurrence and duration of apnoea ²². A study conducted in a cohort of elderly and obese veterans compared the relative gene expression of clock genes between those with and without OSA or related nocturnal hypoxaemia. The results showed that PER3 expression was significantly decreased by 35% among those with OSA compared to those without. In addition, a trend of downregulation of PER3 was observed with increasing severity of OSA ²³. The present study is the first to examine the association of PER3 (rs228729) with OSA. We showed that this variant was significantly associated with OSA in a codominant, dominant, recessive, and log-additive model, although after adjusting for age and BMI there was a significant association for dominant and heterozygosity, but there was no further significant association for OSA when analysing the log-additive.

Although the evidence points to the possibility of changes in the BDNF and PER3 genes being relevant for the development of OSA ^21-23, the type of analysis carried out in the present study did not support the notion of such an association. These results may be due to the fact that there is indeed no relationship between the specific BDNF and PER3 gene polymorphisms and OSA. Alternatively, the small sample size of the present cohort prevented any subtle association from being detected.

Using clinical and phenotypic data collected from patients with OSA, an algorithm was constructed that allows the prediction of OSA cases with reasonable specificity and sensitivity, which, if validated, may at least be applicable in South American populations. Although the PER3 (rs228729) variant, after adjusting for age and BMI, resulted in a significant association for dominant and heterozygosity, but no additional significant association for OSA when analysing the log-additive, it showed a p < 0.05 and remained in the model of multivariate logistic regression. Therefore, the results and the investigation provide a reference for investigations focusing on the identification of OSA in Brazilian and similar populations using the equation:

A limitation of this study is the fact that some polysomnographic examinations were performed in different clinics. Differences in the performance and analysis of polysomnographic examinations are known ²⁵, which may lead to the need for validation of algorithms in different contexts. The results of this study should be replicated in larger and more independent Brazilian samples, preferably involving multicentre studies to minimise the study bias.

Conclusions

In conclusion, no genetic variant tested showed a significant association with OSA phenotype. Notwithstanding, logistic regression analysis enabled risk stratification for OSA diagnosis in Brazilian cases. These results should be replicated in a larger ethnically diverse cohort of cases in a prospective manner.

Conflict of interest statement

The authors declare no conflict of interest.

Author contributions

LB-R, EF, LDM: study conception and design; MLRG, PGA, BG-F: acquisition, analysis and interpretation of data; RPS: statistical analysis; MLRG: drafting of the manuscript; LB-R, EF, LDM: critical revision of the manuscript; LB-R: study supervision.

All authors read and approved the final manuscript.

Ethical consideration

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee (Ethics Committee of the Universidade Federal de Minas Gerais #2.980.453) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Figures and tables

Figure 1.The ROC curve for the logistic regression model was based on the definition of the chance of having OSA. The area under the ROC curve was 79.40%.

References

1. Epstein LJ, Kristo D, Strollo PJ. Adult Obstructive Sleep Apnea Task Force of the American Academy of Sleep Medicine. Clinical guideline for the evaluation, management and long-term care of obstructive sleep apnea in adults. J Clin Sleep Med. 2009; 5:263-276.
2. Trzepizur W, Blanchard M, Ganem T. Sleep apnea-specific hypoxic burden, symptom subtypes, and risk of cardiovascular events and all-cause mortality. Am J Respir Crit Care Med. 2022; 205:108-117. DOI
3. Dodds S, Williams LJ, Roguski A. Mortality and morbidity in obstructive sleep apnoea-hypopnoea syndrome: results from a 30-year prospective cohort study. ERJ Open Res. 2020; 6:00057-2020. DOI
4. Dempsey JA, Veasey SC, Morgan BJ. Pathophysiology of sleep apnea. Physiol Rev. 2010; 90:47-112. DOI
5. Zinchuk A.V, Gentry MJ, Concato J. Phenotypes in obstructive sleep apnea: a definition, examples and evolution of approaches. Sleep Med Rev. 2017; 35:113-123. DOI
6. Gottlieb DJ, Punjabi NM. Diagnosis and management of obstructive sleep apnea: a review. JAMA. 2020; 323:1389-1400. DOI
7. Chowdhuri S, Badr MS. Control of ventilation in health and disease. Chest. 2017; 151:917-929. DOI
8. Redline S, Tishler PV. The genetics of sleep apnea. Sleep Med Rev. 2000; 4:583-602. DOI
9. Casale M, Pappacena M, Rinaldi V. Obstructive sleep apnea syndrome: from phenotype to genetic basis. Current Genomics. 2009; 10:119-126. DOI
10. Mukherjee S, Saxena R, Palmer LJ. The genetics of obstructive sleep apnoea. Respirology. 2018; 23:18-27. DOI
11. Mohit Shrivastava A, Chand P. Molecular determinants of obstructive sleep apnea. Sleep Med. 2021; 80:105-112. DOI
12. American Academy of Sleep Medicine International classification of sleep disorders. American Academy of Sleep Medicine: Darien, IL; 2014.
13. Lahiri DK, Nurnberger JI. A rapid non-enzymatic method for the preparation of HMW DNA from blood for RFLP studies. Nucleic Acids Res. 1991; 19:5444. DOI
14. Aidar M, Line SR. A simple and cost-effective protocol for DNA isolation from buccal epithelial cells. Braz Dent J. 2007; 18:148-152. DOI
15. Peever JH, Lai YY, Siegel JM. Excitatory effects of hypocretin-1 (orexin-A) in the trigeminal motor nucleus are reversed by NMDA antagonism. J Neurophysiol. 2003; 89:2591-2600. DOI
16. Burioka N, Koyanagi S, Endo M. Clock gene dysfunction in patients with obstructive sleep apnoea syndrome. Eur Respir J. 2008; 32:105-112. DOI
17. Pillai S, Nandalike K, Kogelman Y. Severe obstructive sleep apnea in a child with melanocortin-4 receptor deficiency. J Clin Sleep Med. 2014; 10:99-101. DOI
18. Roedig JJ, Phillips BA, Morford LA. Comparison of BMI, AHI, and apolipoprotein E ε4 (APOEε4) alleles among sleep apnea patients with different skeletal classifications. J Clin Sleep Med. 2014; 10:397-402. DOI
19. Akbarian SA, Salehi-Abargouei A, Pourmasoumi M. Association of brain-derived neurotrophic factor gene polymorphisms with body mass index: a systematic review and meta-analysis. Adv Med Sci. 2018; 63:43-56. DOI
20. Shah F, Forsgren S, Holmlund T. Neurotrophic factor BDNF is upregulated in soft palate muscles of snorers and sleep apnea patients. Laryngoscope Investig Otolaryngol. 2018; 4:174-180. DOI
21. Yüksekkaya M, Tutar N, Büyükoğlan H. The association of brain-derived neurotrophic factor gene polymorphism with obstructive sleep apnea syndrome and obesity. Lung. 2016; 194:839-846. DOI
22. Yang MY, Lin PW, Lin HC. Alternations of circadian clock genes expression and oscillation in obstructive sleep apnea. J Clin Med. 2019; 8:1634. DOI
23. Canales MT, Holzworth M, Bozorgmehri S. Clock gene expression is altered in veterans with sleep apnea. Physiol Genomics. 2019; 51:77-82. DOI
24. Heard-Costa NL, Zillikens MC, Monda KL. NRXN3 is a novel locus for waist circumference: a genome-wide association study from the CHARGE Consortium. PLoS Genet. 2009; 5:e1000539.
25. Mjid M, Ouahchi Y, Toujani S. Variabilité inter-nuits du syndrome d’apnée-hypopnée obstructive du sommeil [Night-to-night variability of the obstructive sleep apnoea-hypopnoea syndrome]. Rev Mal Respir. 2016; 33:775-780. DOI