The metabolites that define a bad night’s sleep

Slow reactions, sleepy eyes, or a tired voice are just some of the obvious signs of a sleepless night. But to truly understand the biology of a bad night’s sleep, researchers need markers that are a little more precise. Quantitative biomarkers for impaired sleep will help researchers not only develop systems to prevent traffic accidents but will also improve diagnosis and treatment of chronic sleep disorders such as insomnia or obstructive sleep apnea.

Various tools are currently available for this purpose, but they face limitations: They are costly, time-consuming, and some of them are difficult to carry out outside of specialized settings. For instance, polysomnography, the gold standard diagnostic test for sleep-related breathing disorders, most of the time can only take place within a sleep laboratory with a specialized technician present. To aid in diagnosis, clinicians may also use electroencephalography (EEG) to measure brain activity, ocular tests to monitor eye movements, or a microphone to record snoring behavior. However, limited accessibility to these diagnostic tools contributes to underdiagnosis of some disorders — such as sleep apnea, which is often overlooked in patients with other medical conditions (1,2).

During the wake period, you use energy. When you’re sleeping, you stop eating. You stop moving. So, a lot of it has to do with cycles in energy, and energy really is about metabolism at the end of the day.
- Aalim Weljie, University of Pennsylvania

Now, scientists are discovering that they can identify the effects of a bad night’s sleep by analyzing the metabolites that course through the bloodstream or emerge from breath or urine. “A lot of what happens [in sleep] is about energetics,” said Aalim Weljie, a biochemist who studies sleep at the University of Pennsylvania. “During the wake period, you use energy. When you’re sleeping, you stop eating. You stop moving. So, a lot of it has to do with cycles in energy, and energy really is about metabolism at the end of the day,” he said.

Even just one or a few nights of poor sleep alter metabolite composition. Mice placed in cages with rotating metal bars that disrupted their sleep for five days had significant changes in the abundance of many metabolites (3). In humans, 24 hours of wakefulness increased levels of 27 of 171 plasma metabolites quantified (4). Multiple studies have also concluded that people under sleep deprivation show significant differences in their metabolic profile, namely increased levels of lipids and amino acids, compared to healthy control individuals (5).

Transcriptomics and proteomics have contributed to the search for potential sleep loss signatures. Yet, “what is really interesting about metabolomics is that it’s the end product of whatever ends up being present in the blood,” said Marie Gombert-Labedens, a biological rhythms specialist at SRI International. Weljie concurred, “Because metabolites sit at the end of this chain, they are a more acute reflection of what’s going on [in the body].”

While the search for sleep biomarkers is in its early stages, recent advances in metabolomics are paving the way for more accessible and accurate markers to assess short-term exhaustion and to improve the diagnosis of sleep disorders.

Metabolic signs of a sleepless night

Even if a single night without rest might not cause the severe health consequences linked to long-term sleep deprivation, insufficient sleep for one or two days can lead to fatal accidents on the road or at work due to decreased alertness and reduced cognitive functions.

Clare Anderson, a sleep and circadian rhythm scientist at the University of Birmingham, has spent more than a decade developing and assessing tools to detect insufficient sleep and its associated performance decline, such as ocular and EEG-based measures. But these methods are costly and time-consuming, she said. Moreover, the field has recently pushed towards finding instead a biological signature that reveals how long somebody’s been awake, similar to how the hormone melatonin indicates the body’s internal time, she explained.

A man in yellow T-shirt, blue pants, and a cap sleeps sitting on the floor leaning against a wall, next to a pile of boxes. — A sleepless night can affect work performance and even lead to serious accidents. Using metabolomics, scientists identify key metabolites that reflect this state, aiming to prevent fatal outcomes.
Credit: iStock.com/AlbertoChagas

To achieve this, her team collected blood samples from 23 young healthy adults undergoing various sleep deprivation experiments. They analyzed how 929 metabolites changed as sleep duration did, both at the group and individual levels. Then, “we wanted to go from 929 down to as few as possible,” Anderson said, but without losing accuracy. So, they applied a series of steps to reduce this number, including a machine learning algorithm that estimated the relative importance of each compound for predicting sleep loss in those individuals. By the end, they had a metabolomic biomarker composed of five metabolites: a phenylsulfate, a gut bacteria-derived compound, a monosaccharide, and two phospholipids (6).

When they used the biomarker to predict sleep deprivation based on blood samples from a different cohort of healthy sleep-deprived adults, it showed a 94.7 percent accuracy. According to the authors, this is the highest level of accuracy achieved with a metabolomic biomarker for sleep loss so far.

Anderson foresees using this metabolomic biomarker for forensic purposes, among other applications. “If somebody has been involved in a serious crash or accident and you want to detect whether sleep deprivation was a significant factor, it would be really beneficial to have a kind of marker that does that … just like you were trying to detect the presence of drugs or trying to detect the presence of alcohol,” she said. In an occupational setting, it could also serve as an objective marker to show if someone is sleep deprived and not able to perform a critical task, she added. However, for this to become a reality, she and her team plan to test their biomarker in bigger and more diverse populations.

Improving diagnoses of sleep disorders

Unlike the effects of short-term sleep loss, chronic sleep disorders develop over the course of years, and data from temporal deprivation experiments don’t accurately represent them. Studying their metabolic signatures is also more challenging, said Arjun Sengupta, a bioanalytical chemist who works in Weljie’s laboratory at the University of Pennsylvania. One reason is that disorders such as insomnia or obstructive sleep apnea can look very different from person to person. Patients may experience diverse symptoms or respond differently to treatment.

Various research teams have attempted to characterize chronic sleep pathologies using metabolomics. Notably, Weljie, Sengupta, and their colleagues discovered that people with insomnia had substantial differences in their blood metabolic profile compared to good sleepers (7). Those with insomnia showed desynchronized central energy metabolism with lower lactate levels throughout the entire day and higher glucose levels at night, suggesting that the two groups use and process nutrients differently. Their branched-chain amino acid metabolism also showed nighttime alterations. These amino acids — leucine, isoleucine, and valine — are vital for protein synthesis, and disruptions in their breakdown associate with type 2 diabetes and obesity (8). Due to the small and relatively homogeneous sample size of just 15 participants with insomnia and no other comorbidities, researchers will need to further validate this metabolic signature before using it as a diagnostic tool.

Arjun Sengupta wears purple gloves and holds an instrument. He is standing next to a nuclear magnetic resonance machine. — Using diverse spectroscopic techniques, University of Pennsylvania’s bioanalytical chemist Arjun Sengupta delves into sleep’s metabolic signatures.
Credit: Arjun Sengupta

The metabolic profile of people with obstructive sleep apnea, on the other hand, is quite different, said Weljie. “In sleep apnea, you have hypoxic events which cause the lack of oxygen, and that lack of oxygen is going to affect many metabolic processes in the cell,” he explained.

Urinary samples of people with obstructive sleep apnea — diagnosed via polysomnography — had increased levels of fatty acids, total cholesterol, and triglycerides, suggesting an increased risk for cardiovascular diseases (9). They also showed alterations in branched amino acid levels, a metabolic signature that may link to the reported association of the disease with type 2 diabetes (9,10). The analysis of the exhaled breath of a large cohort of individuals with clinical suspicion of the disorder also led to the identification of a panel of 33 metabolites that could serve as biomarkers in a potential fast and noninvasive diagnostic tool (11).

Furthermore, Weljie noted that within the population experiencing obstructive sleep apnea, there are various subtypes and differences in how the disorder affects people cognitively and in terms of sleep quality. “We have no idea, based on current measures, what the actual consequences to the individual’s performance and cognitive function [are],” he said. Obtaining that information through a blood sample would be very powerful, he added. His team is currently addressing that challenge.

A better diagnostic tool for insomnia and sleep apnea will benefit typically underdiagnosed populations such as women, in whom sleep disorders are often overlooked. For instance, “menopause is an important moment of sleep disturbance onset,” said Gombert-Labedens, co-occurring with many metabolic changes that would be worth further study. Metabolomics could help identify unknown mechanisms linking sleep and hormonal cycles in women, leading to more accurate diagnostics, she added.

There are challenges ahead for the development of sleep disorder biomarkers using this approach. A key one is the replication of laboratory studies in less controlled environments, said Anderson. “Metabolites change as a function of the environment changing — sleep deprivation being one of them, but also movement being another, and diet being another,” she said. Thus, metabolic biomarkers should function in everyday contexts.

Overall, she said, “Sleep biomarkers have a huge future. I think they’re really important, both in occupational health and safety but also within sleep health.”