Gestational diabetes mellitus (GDM), defined as glucose intolerance that is first detected during pregnancy, is a major public health burden with prevalence ranging from 1% to 14%. Moreover, it is a well-established risk factor for adverse maternal and infant health outcomes, including preeclampsia, cesarean delivery, fetal macrosomia, and fetal death, as well as long-term risk of obesity, metabolic syndrome, and diabetes mellitus in both mother and offspring. Numerous studies have identified risk factors for GDM, such as advance maternal age, previous GDM, member of some ethnic groups, and obesity.1–4

Obstructive sleep apnea (OSA) is a common disorder characterized by the presence of repetitive episodes of total or partial airflow interruption in the upper airway during sleep.5 Prevalence of OSA during pregnancy increases due to some physiological changes, such as weight gain, hormonal changes, or modifications in the upper airway.6–8

Previous studies suggest a possible association between OSA and adverse pregnancy and fetal outcomes, such as GDM, preeclampsia, preterm birth, and neonatal low weight.9–11 Some risk factors for OSA, including obesity and increasing age, are shared by GDM. In addition, there is growing evidence that OSA is an independent risk factor for type 2 diabetes, and it has an adverse impact on glycemic control.12

There are several previous studies that evaluated the association between GDM and OSA, but most of them relied on self-related OSA symptoms,13 were retrospective,14,15 or did not measure electroencephalogram signals (EES).9,16 Moreover, reported findings have been inconsistent. Poor sleep quality as well as disturbed sleep at some time points are extremely common during pregnancy,17 but very few studies with polysomnography (PSG), the gold standard for the diagnosis of OSA during pregnancy,7 have been published, and they showed contradictory results.18–21 Furthermore, most of them were limited because they included small samples with high-risk pregnant women, mostly with obesity. Methodological differences, in terms of sleep assessment procedures, and study population may have contributed to the inconsistent findings of the relationship between OSA and GDM. Therefore more studies with attended PSG are clearly needed to clarify the influence of important confounding factors such as obesity, age, ethnicity, and comorbidities.

To address these limitations, we designed a case control study to compare the prevalence of OSA detected with PSG between patients with GDM and a control group of glucose-tolerant pregnant women. Moreover, we aimed to assess the potential association between OSA and GDM determinants. Lastly, we explored the relationships between sleep characteristics and carbohydrate metabolism.

The main findings of the present study are: (1) OSA prevalence during the 3T was not significantly different in patients with GDM than in glucose-tolerant pregnant women. (2) We did not find significant associations between OSA and GDM. (3) HOMA-IR showed an inverse relationship with TST, and a direct correlation with maximum duration of apnea-hypopneas, while there was a direct relationship between T90 and HOMA-IR, and QUICKI.

Most previous studies that have evaluated the relationship between GDM and OSA were based on OSA symptoms,13 were retrospective,14,15 or used type 2–3 sleep devices.16 Moreover, there are some meta-analyses evaluating this association, but with conflicting findings.29–32 Taking into account all these studies, and its inconsistent conclusions, it is still unclear the existence of an association between OSA and GDM. The largest prospective study of OSA in pregnancy until now successfully performed home type 3 sleep tests to 2474 women at late second or early 3T. The prevalence of pre-pregnancy obesity was 23.3%, whereas OSA prevalence was 8.4% in mid-pregnancy. GDM was detected in 96 women, which was independently associated with OSA after adjustment for age, BMI, chronic hypertension, and weight gain (OR: 2.79, CI: 1.63–4.77).9 However, this study was designed and powered to determine if OSA was a risk factor for preeclampsia. Additionally, as a secondary objective, the authors aimed to examine the association with GDM. Furthermore, some women were classified as GDM using medical record abstraction when glucose tolerance test were not available. Nonetheless, the main limitation acknowledged by the authors was that OSA was diagnosed with a home sleep-recording device without EEG variables, that is not validated in pregnancy, which underestimate the burden and severity of the disease, and it does not allow to measure important variables like sleep architecture, arousals, or REM sleep apneic episodes.

To our knowledge, there are only four studies that have evaluated the potential association between GDM and OSA employing PSG recordings, and they deserve some comments. Louis et al.19 performed home PSG in obese pregnant women (76% Afro-American/Hispanic-American), and OSA prevalence was 15.4%. GDM was detected in 17 of them, but the prevalence of GDM was similar among OSA and non-OSA women. Similar results were found in a small case-control study that included 26 women with GDM. All participants underwent unattended home PSG during 3T. GDM women had significant higher ESS compared with control subjects, but neither AHI nor OSA prevalence were significantly different among study groups.18 On the contrary, Retrakul et al.20 found a higher prevalence of OSA in a group of 15 patients with GDM compared with 15 non-diabetic controls (73 vs 27%, p=.01). This study is also limited because of its small sample size and the inclusion of mostly non-Caucasian women with obesity. In keeping with these results, a recently small case-control study performed PSG in 46 women with GDM and in 46 controls, who were matched for age, gestational age, BMI, race, and parity.21 The frequency of OSA was found to be higher in GDM cases than in controls (22 vs. 9%). OSA was found to be significantly associated with the risk of GDM (OR, 1.81). Nevertheless, the main limitation of this study was that mean pre-pregnancy BMI was 30kg/m2, maybe because they over-included subjects with overweight/obesity. Furthermore, more than 75% of them were Afro-American/Hispanic-American. Obesity, and the high percentage of the previously mentioned ethnic groups in the study samples, which are known risk factors for both GDM and OSA, might have an impact on the main findings,1,33 which makes difficult any generalization of their results. As far as we know, our study has included the largest number of pregnant women studied by PSG in the evaluation of OSA as a risk factor for GDM to date. Overall, obesity rate was 10.2%, and, as expected, we found that women with GDM had a higher BMI than controls, which represents the usual clinical finding of patients with GDM. Despite this fact, we did not find that neither AHI nor prevalence of OSA were different among the study groups.

The relationships between GDM and OSA are complex, and any interactions between contributing factors are still unclear. Obesity is one of the main determinants to GDM risk. Besides, ethnicity and age could play major roles.1,22 In study of 105 pregnant women (75% African-American, BMI=33.4), a 27% of OSA frequency has been reported, moreover, with every BMI increase of 5kg/m2, subjects were nearly two times more likely to have OSA.34 Hence, larger studies with PSG are needed to better clarify the complex interrelationships between OSA and GDM, as well as the influence of important cofounding factors, such as obesity, ethnic background, and maternal age.

Specific hormonal and physiological changes during pregnancy could lead to OSA.7,8 The mechanisms responsible for the potential association with GDM have been poorly explained. Among others, it has been proposed that intermittent hypoxia, and arousals could lead to an increase in systemic inflammation, in oxidative stress, and in sympathetic activation, which could contribute to increase cortisol secretion, promoting IR, impairing glucose uptake, and increasing gluconeogenesis.7,10,11,16 However, most evidence of these multiple mechanistic pathways has yet to be demonstrated whether they are also involved during pregnancy.

We did not find significant associations between arousal index and IR. Moreover, we did not find differences in AHI, but patients with GDM had longer mean and maximum obstructive apneas length, in addition to lower nocturnal minimum SaO2. Besides, HOMA-IR was related to maximum duration of respiratory events, and both HOMA-IR and QUICKI were directly proportional to CT90 independently of obesity. Longer sleep event duration has been previously associated with lower nocturnal SaO2 and longer CT90 in OSA.35 Similarly, other groups showed that nocturnal hypoxia markers correlated positively with fasting glucose levels,20,36 IR, and beta cell dysfunction.37 Based on these findings, nocturnal hypoxia could be a potential intermediate mechanism and a better factor in the complex relationship between glucose metabolism and OSA during pregnancy.38 However, a limited number of studies with very small samples, and some confounding variables have examined whether nocturnal hypoxia increases the risk of IR. The percentage of women with obesity was high in most studies, and it could also be possible that the increased risk may be related to obesity or other comorbidities and not to OSA per se.

Sleep time decreases during gestation, particularly by the 3T. Additionally, multifactorial aspects such as psychosocial, ethnic, physical, and biological factors predispose to worsen the sleep quality.17,39 There is some evidence that shorter sleep time is associated with higher glucose levels and GDM.40 We also found that objectively shorter TST was independently associated with higher HOMA-IR. Consequently, an alternative hypothesis is that sleep duration, as well as sleep quality not related to OSA, increases sympathetic activity41 and results in hypothalamic-pituitary-adrenal axis dysregulation,42 which could promote IR. Nonetheless, TST was based in only one night evaluation, and further studies are needed to explore this hypothesis.

Strengths of our study include a multicenter approach, a prospective enrolment, sufficient sample size, and the use of the same protocol in both cases and controls with an objective measure of sleep parameters with PSG performed at hospital. Additionally, laboratory researchers and study personnel were blinded to maternal status. Yet, a number of potential limitations deserve comment. First, the global refusal rate to participate was high (29%), which was though expectedly, as previous studies showed a similar and even much higher percentage of refusal rate (26–88%) both in GDM patients,18,19 and in other groups of pregnant women.7,30 The Institutional Ethic Committee did not allow the collection of information from those women who had declined participation, however we believe that the consecutive recruitment of patients, as well as the general and anthropometric characteristics of the selected women, similar to series of both GDM and healthy pregnant women from the same geographical area,43–45 suggest that the study sample could be considered as representative and makes it difficult to consider a selection bias. Second, as it is already known, we found patients with GDM had higher pregestational BMI. However, it is unlikely that this would modify our results to any important extent since obesity prevalence was relatively low (10%), and despite this, we did not find differences neither in AHI nor in OSA prevalence between both study groups. Third, since episodes of flow limitation <30% associated with arousal were not scored, we cannot ascertain whether our results would still be similar, though, it is unlikely to affect our conclusions as previous studies evaluating OSA in GDM and integrating flow limitation scoring, showed no remarkable influence, since the sleep time spent with flow limitation was similar in both groups in one of the studies,18 while the other study found that the flow limitation index was not associated with increased risk of GDM, spite of the fact the authors showed that OSA prevalence (based only in AHI) was significantly higher in GDM than in control pregnant women.21 Fourth, because we included mainly Caucasian women, our results may not be directly applicable to other ethnic groups. Fifth, further studies are required to better determine the influence of OSA, since the women with OSA of the present study mainly had very mild forms, which could limit further analysis to study the association between DGM and moderate-severe OSA. Finally, the design of the present study does not allow to determine whether OSA started previously or during early or late pregnancy.

In summary, our results did not demonstrate significant differences in OSA prevalence among patients with GDM and control pregnant women, and we did not find significant and independent associations between OSA and GDM. Nevertheless, we found that IR was negatively related with TST and positively related to respiratory apnea-hypopneas length, as well as nocturnal hypoxia, representing that OSA influence on IR during pregnancy could be more complex than its effect on AHI would implicate. The number of studies that have evaluated the interactions of OSA with GDM is still scarce, with major heterogeneity in research design, and populations included. Further, larger, multicenter, and more rigorous investigations employing PSG studies in all subjects, with attention to important confounding factors such as obesity, age and ethnic background, are clearly needed to better clarify and characterize the potential complex interrelationships between OSA and GDM,46 in the meantime, practitioners should screen and treat OSA in suspected pregnant women.

This research was partially supported by grants from SEPAR-2010-820, Ministerio de Economía y Competitividad (PI10/00495), Ministerio de Ciencia, Innovación y Universidades (PI19/00875), and Programa “Ramon Llull”, contractes per a la Intensificació de l’Activitat Investigadora a l’IdISBa 2020.

The authors declare that they have no conflict of interest.