Relationship between poor sleep and daytime cognitive performance in young adults with autism

Abstract

Poor sleep is a common feature in autism even though patients themselves do not necessarily complain. The impact of poor sleep on daytime cognitive functioning in autism is not well-known and we therefore investigated whether sleep in autism correlates with daytime cognitive performance. A battery of non-verbal tasks was administered, in the morning after a second night of sleep in the laboratory, to 17 young adults with autism and normal intelligence, and 14 typically developed individuals matched for age and IQ; none of the participants complained about sleep problems. Two dimensions of attention (sustained and selective) and 4 types of memory (working, declarative, sensory-motor and cognitive procedural) were tested. Individuals with autism showed clear signs of poor sleep. Their performance differed from the controls in response speed but not in accuracy. Signs of poor sleep in the autism group were significantly correlated with either normal performance (selective attention and declarative memory) or performance inferior to that of the controls (sensory-motor and cognitive procedural memories). Both groups presented a significant negative correlation between slow-wave sleep (SWS) and learning a sensory-motor procedural memory task. Only control participants showed a positive association between SWS duration and number of figures recalled on the declarative memory task. Correlation patterns differed between groups when sleep spindles were considered: they were negatively associated with number of trials needed to learn the sensory-motor procedural memory task in autism and with reaction time and number of errors on selective attention in the controls. Correlation between rapid eye movements (REMs) in REM sleep and cognitive procedural memory was not significant. We conclude that some signs reflecting the presence of poor sleep in adults with high-functioning autism correlate with various aspects of motor output on non-verbal performance tasks. The question is raised whether poor sleep in non-complaining persons with autism should be treated.

Introduction

Sleep is a natural, spontaneous phenomenon that involves the activation of neurobiological systems shared with the waking state (Brown, Basheer, McKenna, Strecker, & McCarley, 2012). Many studies show a relationship between sleep and daytime cognitive performance in healthy individuals as well as in various medical conditions (Chee and Chuah, 2008, Diekelmann and Born, 2010, Maquet et al., 2003, Wang et al., 2011). Unfortunately, this type of protocol has never been adapted to research on individuals with autism spectrum disorders (ASD). As a result, very little is known about sleep and EEG in autism, and, a fortiori, about possible relationships between these measures and the clinical and cognitive picture of autism.

Neurobiological correlates of autism include atypical micro- and macroscopic brain structure and functional reallocation of brain activity (Belmonte et al., 2004). Sleep disorders are reported and documented in adults with ASD (Godbout et al., 2000, Limoges et al., 2005, Tani et al., 2003). Limoges et al. (2005) investigated both subjective and objective accounts of sleep in normal IQ adults with ASD and normal IQ adults without ASD while controlling for confounding factors, such as psychiatric comorbidity and age differences between the clinical and comparison groups. They found 3 main categories of sleep disturbances. First, symptoms of poor sleep, i.e., trouble initiating and maintaining sleep as quantified by increased sleep latency, increased light (stage 1) sleep, and increased awakenings after sleep onset. Second, disorders of EEG synchronization, i.e., low slow-wave sleep (SWS) percentage and low densities of phasic activities during stage 2 sleep (i.e., decreased EEG sleep spindles). Third, hypoactivation of rapid eye movements (REMs) during REM sleep (i.e., decreased REMs).

Several studies have related sleep quality to attention and memory in neurotypical populations. Cognitive deficits due to sleep disorders, such as insomnia, have been described (Fortier-Brochu et al., 2012, Göder et al., 2007, Nissen et al., 2011). The literature shows that long-term declarative memory and working memory deficits as well as alertness and sustained attention problems associated with poor quality of sleep can be attributed to daytime sleepiness (Hauri, 1997, Mendelson et al., 1984, Vignola et al., 2000). Experimental protocols, to evaluate the effects of complete, partial or selective sleep deprivation on subsequent cognitive performance, have suggested specific associations between sleep architecture and cognitive functions, including attention and memory (Forest and Godbout, 2005, Maquet et al., 2003, Smith et al., 2004). In particular, early nocturnal sleep (rich in non-REM sleep) has been linked with better performance in verbal and non-verbal declarative long-term memory tasks (Plihal and Born, 1997, Plihal and Born, 1999). Many other studies have positively associated non-REM sleep with declarative long-term memory (Fowler et al., 1973, Mazzoni et al., 1999) and also with sensory-motor procedural memory (Smith and MacNeill, 1994, Smith et al., 2004). In contrast, late nocturnal sleep (rich in REM sleep) has been related to verbal and non-verbal procedural memory tasks (Plihal and Born, 1997, Plihal and Born, 1999). REM sleep phasic activity has also been linked with cognitive procedural memory in other studies (Fogel et al., 2007a, Smith and Smith, 2003). Attention processes seem to be more sensitive to a general loss of sleep quality and quantity (such as increased sleep latency or decreased sleep efficiency) rather than to selective sleep deprivation (Forest & Godbout, 2005), even though a clear correlation between selective attention and stage 2 sleep spindles has been reported (Forest et al., 2007). Given the documented relationship between nocturnal sleep and daytime cognitive performance in humans, markers of poor sleep and cognitive performance documented in ASD are therefore expected to correlate.

There is no straightforward pattern of intact vs. impaired components of attention in high-functioning ASD. Sustained attention has been preserved in most studies (Burack, 1997, Goldstein et al., 2001, Minshew et al., 1997, Pascualvaca et al., 1998). Selective attention appears to be superior to that of non-autistic individuals in visual search tasks (Plaisted, O’Riordan, & Baron-Cohen, 1998). However, impaired flexibility of attention (disengaging and re-orienting) has been observed (Burack, 1994, Burack, 1997, Goldstein et al., 2001, Rinehart et al., 2001, Townsend et al., 1996). Regarding the pattern of memory performance in high-functioning ASD, intact verbal working memory contrasts with growing evidence of abnormal visuo-spatial working memory (Belleville et al., 2006, Geurts et al., 2004, Minshew et al., 1999). Declarative episodic memory processes (encoding, storage and retrieval) were spared in most studies (Ben Shalom, 2003, Minshew and Goldstein, 2001), while categorization during memory tasks (Bowler, Matthews, & Gardiner, 1997) and balance between the superficial and deep effects of cueing favor “superficial” aspects of memorized material (Mottron et al., 2001, Toichi and Kamio, 2002). Representation system and semantic memory are also spared (Ben Shalom, 2003). Procedural memory has received very little attention. Mostofsky, Goldberg, Landa, and Denckla (2000) published the only study on motor procedural memory in children with high-functioning ASD, reporting slower motor procedural learning in a serial reaction time (RT) task. However, impairments on tasks, such as Tower of London and Tower of Hanoï (Geurts et al., 2004, Ozonoff et al., 1991), first taken to reflect executive functions, might also reflect cognitive procedural memory when considering the learning effect of problems (Beauchamp et al., 2003, Ouellet et al., 2004).

The present study aimed at verifying whether markers of poor sleep, documented in ASD, correlate with non-verbal cognitive performance. To achieve this goal, we first analyzed the cognitive performance of participants with and without autism. Then, correlations between cognitive variables known to be associated with sleep performance were tested. To avoid performance variability due to test modality, only non-verbal tasks were used. To our knowledge, this is the first study to simultaneously investigate sleep architecture and cognitive performance in ASD individuals.