Sleep and Insulin Resistance in Adolescents

In the past 20 years, the prevalence of obesity, as well as impaired glucose tolerance and type 2 diabetes mellitus, has increased in children.^1–2 Over the same time period, it has been suggested that sleep duration has been diminishing—a 2006 survey of the National Sleep Foundation found that 75% of US high school seniors reported not getting enough sleep.³ Several epidemiological studies have been published that suggest a link between short sleep duration and obesity in children and adolescents.^4,5 However, to date, relatively few community-based^6,7 and clinic-based studies^8–10 have examined the relationship between short sleep and metabolic parameters in children and adolescents. Two recent community-based studies used wrist actigraphy monitoring for 5-7 days to assess habitual sleep at home. Javaheri et al. studied a cohort of adolescents and reported a lack of association between short sleep duration and markers of glucose metabolism after adjusting for measures of adiposity.⁶ In the other community-based study of children between 5 to 8 years of age, obese children had higher fasting glucose and insulin concentrations and displayed a greater variability of sleep time on weekends compared to weeknights; however, their average sleep time was comparable to that of normal weight and overweight children.⁷ This finding suggests that sleep irregularity may be a more significant predictor of insulin resistance than short sleep duration in young children. The two clinic-based studies that reported a positive association between short sleep duration and glucose metabolism dysregulation used in laboratory polysomnography to assess total sleep time.^9,10 In contrast, using a sample of patients recruited from the obesity clinic, Sung and colleagues⁸ reported a lack of association between short sleep duration measured by 1-week of wrist actigraphy and insulin resistance. This inconsistency between the clinic-based studies may be due to the different methods used to assess sleep duration—wrist actigraphy vs. in-laboratory polysomnography, the latter not necessarily reflective of habitual sleep habits—as well as different patient characteristics.

In this issue of SLEEP, Matthews and colleagues¹¹ report the results of a cross-sectional study conducted in 245 healthy teenagers (14-19 years old) attending an urban public high school in the United States. They obtained fasting blood samples for glucose and insulin to calculate Homeostatic Model Assessment-Insulin Resistance (HOMA-IR), a measure of insulin resistance in the fasting state that has been validated in children.¹² Sleep duration was derived from both self-reports and 7 consecutive days of wrist actigraphy monitoring. The authors observed an inverse linear relationship between sleep duration (assessed by either method) and increased insulin resistance (HOMA-IR), which persisted after adjusting for a variety of possible confounders. The association between sleep duration and HOMA-IR was stronger in males compared to females. HOMA-IR was not correlated with actigraphic indices of sleep fragmentation in either sex, but in males, fasting glucose concentrations were significantly associated with sleep fragmentation, after adjusting for body mass index (BMI), race, and socioeconomic status. Importantly, controlling for adiposity did not eliminate the relationship between short sleep duration and insulin resistance, contrary to what was observed in the Cleveland Children Sleep and Health Study.⁶

The study by Matthews et al.¹¹ has several strengths. First, it focuses on adolescents, a population that undergoes substantial changes in sleep and has received relatively little attention so far in studies of sleep and metabolism, despite the increase in rates of obesity and type 2 diabetes mellitus in the past two decades. Of note, 48% of the cohort was overweight or obese (BMI ≥ 25 kg/m²). Secondly, it involves a fairly large gender-balanced community-based sample, with a good representation of African-American participants. Interestingly, no significant interactions by race were observed, suggesting that the impact of sleep loss on metabolism is at least as significant in young African Americans, a group at disproportionately high risk of diabetes, as it is in whites. Finally, sleep was evaluated with an objective method that has been validated against polysomnography and allows characterization of sleep habits in the field.¹³

Notwithstanding the strengths of the study by Matthews and colleagues,¹¹ there are also several noteworthy limitations. The authors have identified some of them in their discussion, such as the cross-sectional design of the study that limits the assessment of causality, or the lack of objective evaluation of sleep-disordered breathing, which is likely to have been prevalent in this population and could represent a further confounder. It is important to note that without using validated questionnaires, parental report of symptoms related to sleep-disordered breathing is inaccurate during preschool years and is likely to be in accurate as well during adolescence.^14–16 Another limitation is the reliance on fasting blood samples, which give a limited picture of glucose metabolism. HOMA-IR, although widely validated, is a measure of insulin-resistance, but no information about glucose tolerance or beta-cell function can be derived from it. It is also noteworthy that most controlled laboratory studies of partial sleep restriction in adults have reported alterations in glucose metabolism during dynamic conditions such as oral glucose tolerance testing, frequently sampled intravenous glucose tolerance testing, and hyperinsulinemic euglycemic clamp. However, these studies did not find any change in parameters of glucose metabolism in the fasting state.¹⁷ The fact that fasting alterations in glucose metabolism were detected by Matthews et al.¹¹ may suggest a strong effect in this younger population, which could be particularly susceptible to the effects of sleep loss. However, they did not specify if the wrist actigraphy monitoring started on the same day of the week for all participants. Given that sleep duration was about 1.5 h longer on weekends, if the blood sample was taken on a Monday as opposed to a Friday, results may have been influenced by the relative sleep recovery taking place on weekends. Finally, Matthews et al. do not indicate if the students took daytime naps, which could also have somewhat confounded the results. One poll suggested that 38% of high school students reported having had at least 2 naps in the 2 previous weeks, and the naps lasted on average 1.3 hours. In the same poll, 55% school children sleeping less than 8 hours reported taking at least 2 naps in the previous 2 weeks.¹⁸

Matthews et al.¹¹ distinguished between weekday and weekend sleep; the relationship between sleep duration and insulin resistance was only present for week nights. The difference in sleep duration between work/school nights and weekend nights, which has been referred to as social jetlag and could indicate circadian disruption, has also been linked to obesity and could represent another factor contributing to the development of insulin resistance.¹⁹ Indeed, social jet lag has been found to be most pronounced among adolescents compared to other ages.

Besides the main finding of an impact of sleep duration on glucose homeostasis, Matthews et al.¹¹ confirm yet again that American adolescents are chronically sleep deprived; the average sleep duration reported is strikingly short (6.4 h/night), and approximately 75% of the cohort slept less than 6.5 h per night on week nights. These sleep durations are significantly shorter than what was reported in the last National Sleep Foundation poll, which assessed sleep duration by self-report,¹⁸ and considerably lower than the recommended amount (about 9 hours across all years of adolescence).²⁰ Adolescents are naturally delayed in terms of circadian phase^21,22 and the organization of the school day in the United States, with early start times for adolescents, is at odds with this circadian chronotype. However, other factors (e.g., cultural, psychosocial) may play a role, as average sleep duration on weekends, although longer, was on average 7.5 h.

In conclusion, the results of Matthews et al.¹¹ suggest that the relationship between short sleep duration and alterations in glucose homeostasis, which has been extensively demonstrated in adults in both epidemiologic and laboratory-based studies, is also present in adolescents.²³ Given the potentially negative impact these alterations could have in the future metabolic and cardiovascular health of adolescents, it is clear that there is a need for controlled clinical trials to assess the impact of sleep extension and improved sleep hygiene on glucose metabolism in this vulnerable age group. Moreover, longitudinal studies are needed to better establish the direction of causality and to dissect the mechanisms underlying the relationships between short or poor quality sleep and impairment of glucose metabolism.

CITATION

Morselli LL; Knutson KL; Mokhlesi B. Sleep and insulin resistance in adolescents. SLEEP 2012;35(10):1313–1314.

DISCLOSURE STATEMENT

Dr. Mokhlesi has received research support from Philips/Respironics. The other authors have indicated no financial conflicts of interest.