Sleep duration, insomnia, and Parkinson’s disease

Sleep is a natural, recurring physiological state that includes diminished conscious awareness and low motor activity. It is a time of restoration and relative anabolism. In the brain, sleep enhances the consolidation of memories acquired during wakefulness, modulates the strength of synaptic connections, and promotes the removal of toxic metabolic products through the glymphatic system.

Sleep patterns change with age, and sleep is markedly disrupted in neurodegenerative disorders such as Parkinson’s disease and Alzheimer’s disease.^1,2 The effects may be bi-directional. It is not clear whether disordered sleep contributes to the progression and pathogenesis of these diseases or whether disordered sleep is an early consequence of pathological changes in the brainstem and cerebral cortex. Both scenarios, or neither, may be correct.

Parkinson’s disease is second only to Alzheimer’s disease as the most common neurodegenerative disorder. It is defined clinically by cardinal motor symptoms of tremor, rigidity, and bradykinesia, and men are affected more often than women. At a molecular level, it is characterized by the aggregation and accumulation of a misfolded protein called a-synuclein, which is microscopically visible as Lewy neurites and Lewy bodies in the substantia nigra, certain brainstem nuclei, and the cerebral cortex. Brainstem pathology, which precedes motor symptoms of Parkinson’s disease by a decade,³ may be particularly relevant to changes in sleep patterns. Sleep disruption in Parkinson’s disease takes the form of parasomnias and impaired transitions from sleep to waking. Transitional disturbances include increased sleep latency (i.e., trouble falling asleep) and frequent awakenings after sleep onset (disrupted sleep). These result in a reduction in total sleep time and sleep fragmentation.

In this issue, Beydoun et al.⁴ examine the complex relation between sleep and Parkinson risk in over 130,000 postmenopausal women enrolled in Women’s Health Initiative studies. Sleep measures for these secondary analyses were obtained at baseline and during follow-up. Women were on average 63 years of age at baseline, and 2.2% developed incident Parkinson’s disease during a mean follow-up of 16 years. The diagnosis was based largely on self-report (“Has a doctor every told you that you have Parkinson’s disease?”), with a small proportion of diagnoses derived from death certificate records or inferred from the use of anti-parkinsonian medications. Their findings provide insights regarding two key, self-reported exposures: sleep duration and insomnia.

Sleep duration was based on a “typical night” during the preceding 4 weeks, analyzed as ≤6 hours (short sleepers), 7 or 8 hours (standard sleepers), or ≥9 hours (long sleepers).⁴ Somewhat over a third of women (36%) were self-reported short sleepers. In fully adjusted models, short sleepers at baseline were no more likely to develop Parkinson’s disease than standard sleepers (hazard ratio 1.01, 95% confidence interval 0.93 to 1.09).

Only 4% of women were long sleepers, but these women were more likely to be diagnosed with Parkinson’s disease (hazard ratio 1.30, 1.11 to 1.53).⁴ The explanation is not clear, but increased sleep duration has been linked to other adverse health outcomes, including increased mortality, heart disease, and Alzheimer’s disease.^5,6

The second key measure was perceived insomnia. An insomnia score was derived from symptoms during the preceding four weeks, reported on a 0–4 Likert scale. For trouble falling asleep, waking up “several times” at night, waking up earlier than planned, and trouble falling back asleep after awakening too early, Likert ratings ranged from “no, not in the past 4 weeks” to “yes, 5 or more times a week.” A fifth symptom referred to the quality of a typical night’s sleep, from “very sound or restful” to “very restless.”

Twenty-six percent of women reported insomnia at baseline, defined by scores >9. The rating scale does not distinguish between primary insomnia and insomnia due to depression or some other condition. Women with insomnia faced a modestly higher hazard of developing Parkinson’s disease than women without these symptoms.⁴ The hazard ratio in the fully adjusted model was 1.13, with a 95% confidence interval of 1.04 to 1.23.

A final finding is worth noting. Sleep exposures were obtained at baseline and at least one later time point.⁴ Baseline analyses and time-varying exposures showed similar associations with Parkinson risk, suggesting that observed associations were already present at baseline, some years before women were aware that they had Parkinson’s disease.

What are the public health implications?

First, shorter duration of sleep — or at least a perceived shorter sleep duration — may be unrelated to Parkinson risk. This could be reassuring news for the large number of women who typically sleep six hours or less at night.

Second, insomnia symptoms may be related to Parkinson risk, but the magnitude of association appears to be small. Two of the five insomnia questions implicate total sleep time (difficulty falling asleep and early awakening). However, as already noted, short sleep duration on its own was not associated with Parkinson’s disease risk, and in previous analyses insomnia scores did not correlate with sleep duration determined by actigraphy.⁷ The other three insomnia questions implicate sleep fragmentation (waking up several times, difficulty falling back asleep after an early awakening, and restlessness). Sleep disruption may be key, perhaps by reducing time spent in deep sleep.

Results of course pertain only to postmenopausal women, the population under study. However, a large, population-based Taiwanese cohort of adult men and women found that a medical diagnosis of a sleep disorder (sleep apnea was excluded in these analyses) was associated with an 18% increased risk of Parkinson’s disease (95% confidence interval 1.11 to 1.26), with higher hazard linked to the specific diagnosis of chronic insomnia.⁸

If the relation is causal, sleep disruption might promote the development and progression of a-synucleinopathy. Putative mechanisms include impaired immune function, oxidative stress, inflammation, and impeded glymphatic clearance of toxic substances from the brain during deep sleep.² This scenario raises the prospect that a lifetime of better sleep hygiene might reduce Parkinson’s disease risk and, to the extent to which there are shared mechanisms in common, perhaps the risk of other neurodegenerative disorders such as Alzheimer’s disease.

However, once insomnia symptoms have appeared, it may be too late to reduce the threat of Parkinson’s disease to a meaningful extent. First, the risk associated with insomnia is small — only about 13% for women included in this study — suggesting that any potential benefit might be small as well. Second, the association was already present at baseline, suggesting that an intervention would need to be implemented many years before the emergence of cardinal motor symptoms.

Even if the relation between sleep and Parkinson’s disease is not causal, insomnia or an objective measure of sleep fragmentation could represent an early marker of Parkinson pathology and help predict those at increased risk of clinical Parkinson’s disease. However, insomnia is very common in the population (over a fourth of women in this study) in relation to Parkinson’s disease (only 2% of women developed Parkinson’s disease during the follow-up period). As such, the positive predictive value of insomnia during midlife or old age would be low. This inference contrasts with rapid eye movement sleep behavior disorder, a parasomnia that is highly predictive for the development of Parkinson’s disease⁹ but was not analyzed in this study.⁴ There might be some value, though, in considering insomnia as one component of an aggregated risk score.

In summary, findings of Beydoun et al.³ reinforce the relevance of sleep fragmentation to Parkinson’s disease, with implications concerning disease pathogenesis and early, premotor manifestations. The results, however, may or may not help guide approaches to the prevention, treatment, or early identification of Parkinson’s disease.