Another Strike Against Sleepability

Concerns about the multiple sleep latency test (MSLT) as the gold standard for sleepiness are nearly as old as the MSLT itself.^1–3 Although the MSLT proved undeniably valuable in early studies of sleep deprivation and sleep disorders,^4,5 its current clinical applications are problematic. Fundamentally, it is hampered by the fact that it measures “sleepability,”⁶ i.e., the ability to fall asleep quickly on command, rather than directly measuring the pathologic experience of being unable to stay awake when desired, sleeping too long, or struggling to awaken. More direct measures of difficulty remaining awake, such as the Maintenance of Wakefulness Test and the Epworth Sleepiness Scale, correlate only moderately with MSLT findings and individuals can show discordant results across these paradigms.^7,8 Furthermore, the MSLT has poor test-retest reliability in some hypersomnolence disorders^9,10 and in the general population,¹¹ has a very high rate of positive results (i.e., mean sleep latency [MSL] < 8 minutes) at the population level,¹² and is frequently normal in patients with clinically significant hypersomnolence.^13,14

In the current issue of JCSM, Plante and colleagues highlight another major limitation of the MSLT using the large, population-based Wisconsin Sleep Cohort.¹⁵ Analyzing data from 1,155 adults and adjusting for available confounders (demographics, medical conditions, substance use, sleep disordered breathing, and reported sleep time), the authors found an inverse relationship between depression and sleepability on the MSLT, such that the presence of depression associated with decreased odds of an MSL < 8 minutes (OR 0.76, 95% CI 0.63–0.92). At the same time, subjective measures of hypersomnolence, both elevated Epworth Sleepiness Scale scores and long reported habitual sleep times, were found with significantly increased odds in people with depression. They concluded, given this discrepancy between subjective and objective results, the MSLT is a poor objective tool to quantify sleepiness in patients with depression.

Hypersomnolence and depression have a complex interrelationship. Hypersomnolence is a core feature of depression, embedded in screening tools (e.g., potentially impacting both the S [sleep] and E [energy] in SIG-E-CAPS¹⁶ and several questions, excluded in the Plante analyses, on the Zung depression scale¹⁵). Patients with hypersomnolence disorders face increased psychosocial burdens,¹⁷ and, at least for those disorders where exclusion of depression is not part of the diagnosis, demonstrate higher than expected depression rates.¹⁸ Perhaps nowhere is this diagnostic dilemma highlighted more aptly than in the International Classification of Sleep Disorders, Third Edition, in which the diagnosis of idiopathic hypersomnia requires ruling out mood disorders that could better account for symptoms and the criteria for hypersomnolence associated with a mood disorder require ruling out idiopathic hypersomnia.¹⁹

As discussed by Plante et al., there are reasons to suspect that hypersomnolence in depressed individuals might be explained, at least in part, by misperceived fatigue or long times spent awake in bed. So it is tempting to speculate, based on the results of the Plante study, that the lack of objective MSLT findings in patients with depression might be clinically useful, because it could allow for segregating hypersomnolence associated with depression from other causes of hypersomnolence. It could be hypothesized, perhaps, that if the MSLT sleep latency is normal, then treatment for depression is indicated and if the sleep latency is abnormal, then treatment for idiopathic hypersomnia is warranted. However, both the depression and idiopathic hypersomnia literatures offer direct evidence to contradict this approach. Approximately one quarter of patients with hypersomnolence associated with a mood disorder have an MSL of less than 8 minutes.²⁰ Conversely, patients with idiopathic hypersomnia who have objective evidence of increased sleep propensity (defined as polysomnographic-measured total sleep times of greater than 600 minutes) are more than twice as likely to have a normal MSL than an abnormal one.¹⁴ Clearly, although differences can be seen across large groups, the MSLT is problematic for separating hypersomnolence with and without associated depression at the individual level.

What, then, is the solution to the hypersomnolence diagnosis problem? So far, a clearly superior alternative to the MSLT has yet to emerge, despite the clinical need. Plante et al. suggest that, because hypersomnolence in depressed patients is multi-faceted, multiple measures of hypersomnolence are likely to be needed.¹⁵ This is almost certainly true for the remaining central hypersomnolence disorders as well. To date, the assessment of excessive sleepiness has strongly emphasized the wake-to-sleep transition, through the MSLT, MWT, and the Epworth, while the experience of hypersomnolence can also include prolonged sleep times, decreased quality of wakefulness, and difficulty with the sleep-to-wake transition. Sleep duration has recently entered the diagnostic algorithm for idiopathic hyper-somnia,¹⁹ and measures of waking function, such as driving simulators and psychomotor vigilance, are receiving attention as possible clinical tools.^21,22 The final component of hypersomnolence, i.e., difficulty with the transition from sleep to wake, may offer another avenue for hypersomnolence measurement. This is particularly true for idiopathic hypersomnia, for which the subtype of “hypersomnia with sleep drunkenness” was proposed decades ago to capture those patients in whom prolonged difficulties with awakening were even more problematic than daytime sleepiness.²³ But difficulties with awakening are also a common feature of mood disorders, with three quarters of patients with unipolar depression endorsing difficulty with getting out bed in the morning²⁴ and depressed patients endorsing more difficulties with physical, cognitive, and emotional aspects of awakening than healthy controls.²⁵ As with time in bed, difficulty with awakening in patients with mood disorders may reflect factors other than increased sleep propensity, such as dread about starting the day,²⁴ but this is an area that requires more detailed subjective and objective testing before firm conclusions can be drawn. Some investigations into this realm, e.g., evoked potentials upon forced awakening, show preliminary promise.²⁶ But the lesson learned about the MSLT, from the work by Plante et al. and others, is that we cannot continue to rely on “sleepability” as our most fundamental measure of the complex and multifaceted experience of hypersomnolence.

DISCLOSURE STATEMENT

This work was supported by the National Institute of Neurological Disorders and Stroke (grant K23 NS083748) from the National Institutes of Health. Dr. Trotti reports research support to her institution from Jazz Pharmaceuticals.

CITATION

Trotti LM. Another strike against sleepability. J Clin Sleep Med 2016;12(4):467–468.