Effetti dell'esercizio acuto o degli interventi di esercizio a breve termine sui marcatori metabolici durante la privazione del sonno indotta sperimentalmente nell'uomo: una revisione sistematica della letteratura

1. Introduction

Sleep is a non-negotiable biological state required for the maintenance of human life. Nonetheless, the prevalence of insufficient sleep is high in modern society, such that one-third or more of adults in the USA, Canada, UK, and Singapore sleep less than the 7 h per night that is recommended by public health authorities. Insufficient sleep is associated with an increased risk of developing cardiovascular diseases, type 2 diabetes (T2D), , obesity, , subclinical atherosclerosis, and other metabolic-related issues. Studies on acute (up to 2 nights of 4-h sleep restriction) and short-term (up to 7 nights of 5-h sleep restriction) experimentally-induced sleep loss have revealed an impairment in glucose tolerance and insulin resistance, , , , , , which may contribute to the development of T2D in adults. Given those constraints, adhering to an otherwise healthy lifestyle may be a positive approach among individuals with habitual short sleep durations.

Considering the challenges of ensuring adequate sleep, alternative strategies to mitigate these risks are critical. Physical activity is a viable alternative strategy for enhancing glucose tolerance and insulin sensitivity, and thereby reducing the risk of developing T2D, , an effect that is also observed in the presence of insufficient sleep. One prospective cohort study of 502,612 participants utilizing data from the UK Biobank revealed that individuals who engaged in physical activity at or above the levels recommended by the World Health Organization (WHO) guidelines (600 metabolic equivalent task (MET) min/week) were able to mitigate most of the harmful associations between poor sleep and mortality. Additionally, a beneficial impact of moderate-to-vigorous and strenuous structured exercise on the incidence of T2D is established, while high-intensity interval training (HIIT) has been shown to decrease glucose and insulin concentrations compared to control groups. Notably, in the context of sleep loss, 3 sessions of high-intensity interval exercise (HIIE) may counteract the adverse effects of experimentally-induced sleep loss on glucose tolerance. These findings suggest that exercise may be a critical intervention for mitigating the detrimental effects of sleep loss on metabolic health.

Therefore, the purpose of this review was to systematically evaluate the existing literature on whether acute and short-term exercise interventions can mitigate negative effects of experimentally-induced sleep loss on metabolic markers in blood and skeletal muscle in humans.

2. Methods

This systematic literature review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines ( ) and was registered on PROSPERO (Protocol No. CRD42024469566). The established protocol plan was followed without significant deviations.

### 2.1. Search strategy

A systematic electronic search of the literature was conducted in 6 databases, including PubMed/Medline, Scopus, Web of Science, Embase, Cochrane, and SPORTDiscus, from journal inception until November 2023 and updated on June 15, 2024. The search strategy was developed by VSF and BE a priori and was used for all databases: (“short sleep duration” OR “sleep deprivation” OR “sleep restriction” OR “sleep loss” OR “sleep insufficiency” OR “insufficient sleep” OR “sleep deficiency” OR “lack of sleep” OR “sleep absence"” OR “sleep deficit”) AND (“exercise” OR “physical performance” OR “endurance performance” OR “strength performance” OR “physical training” OR “physical effort” OR “physical activity”). The filters for “studies in English”, “humans”, and “peer reviewed articles” were activated.

### 2.2. Eligibility criteria

The Population, Intervention, Comparator, Outcomes, and Study design (PICOS) framework was adopted for eligibility criteria ( ). Original articles were included in this review if (a) it was conducted in adults (aged 18 years or over); (b) acute exercise or short-term exercise intervention (up to 6 exercise sessions) and experimentally-induced sleep loss (up to 7 nights of sleep loss) were presented in detail; (c) at least 1 outcome of interest must be measured, such as fasting or post-prandial (after a standardized meal) lipid and metabolomic profiles, glucose and insulin concentrations, insulin sensitivity, or molecular markers in the form of metabolic or mitochondrial gene expression or protein content in skeletal muscle; and (d) studies were published in the English language. Experimentally-induced sleep loss included studies of an entire night of sleep deprivation (“sleep-deprived”) and studies where sleep was restricted by a specified number of hours (“sleep-restricted”). In the latter case, models of both early and late sleep restriction were included. Early sleep restriction (i.e., delayed sleep onset) describes interventions limiting sleep during the earlier hours of the sleep period (typically the first half, e.g., 11:00 p.m. to 3:00 a.m.), whereas late sleep restriction (i.e., earlier than usual waking) describes interventions limiting sleep during the later hours of the sleep period (typically the second half, e.g., 3:00 a.m. to 7:00 a.m.).

Studies were excluded from this review if (a) they were editorials, theses, posters or conference abstracts or presentations, or opinion pieces; (b) the sleep loss was not clearly specified, such as early or late sleep restriction, sleep fragmentation, or total sleep deprivation; (c) the population of interest were military or cadets in field training, which introduces a wide range of confounders, such as caloric restriction, emotional stress, and immeasurable exercise features (volume, intensity, and rating of perceived effort); (d) they were conducted in the presence of a diagnosis of chronic disease (e.g., diabetes, metabolic syndrome, or cardiovascular disease), psychiatric disorder (depression or anxiety), sleep disorder (e.g., insomnia, sleep apnea, or narcolepsy), or eating disorder; (e) any drug was administered during the experimental design, including melatonin, benzodiazepines, anti-depressants, and sedating anti-histamines; (f) they were an animal or cell culture study; or (g) they were an abstract without full-text.

### 2.3. Study selection and data extraction

The literature search was completed by the first author (VF) and confirmed by the last author (BE). Data were exported from the aforementioned 6 databases and uploaded to the Covidence Systematic Review Management Software (Veritas Health Innovation, Melbourne, Australia). Covidence was used to eliminate duplicates. Title and abstract were screened independently by 2 authors (VSF and LM). Then the full texts were assessed against the pre-determined inclusion and exclusion criteria. Disagreements were resolved by consensus between the authors and a third author (ADOH or BE). Reference lists of relevant articles were then screened in order to explore the potential for additional articles not included in the original search strategy. After compiling the final list of relevant articles, data extraction was conducted by 2 researchers (VSF and BE) working independently. The following data were extracted to an electronic spreadsheet: authors and study design, participant characteristics, interventions, outcomes of interest, and differences between conditions (detailed in and , and ).

### 2.4. Risk of bias assessment

The methodological quality of the included studies was critically-appraised using 2 distinct but closely-related tools as appropriate to the respective experimental design: “Cochrane Risk of Bias Tool for Randomized Trials 2” and “Cochrane Risk of Bias Tool for Randomized Crossover Trials 2” (RoB 2.0). , These tools contain 6 domains, including: (a) bias arising from the randomization process; (b) bias due to the intended interventions; (c) bias due to missing outcome data; (d) bias in measurement of the outcome data; (e) bias in selection of the reported result; with a further domain for crossover trials; and (f) bias arising from the carryover effects. Domain-level judgments about risk of bias were classified as “low risk of bias”, “some concerns”, and “high risk of bias”. The overall judgment was calculated using the lowest score in the domain (e.g., in a case where a study was classified “low risk of bias” for 4 domains and “some concerns” for 1, the overall judgement was “some concerns”). None of the included studies employing a crossover design reported carryover effects; therefore, a low risk of bias was assigned to the sixth domain in those studies. Studies employing a parallel group design were marked as not applicable (N/A) for the sixth domain. Two authors (VSF and LM) assessed the risk of bias independently and any discrepancies were resolved by a third author (BE or ADOH).

3. Results

### 3.1. Literature search and quality assessment

Of the total 7989 articles retrieved, 4026 were excluded as title duplicates. The remaining 3963 articles were screened by title, and 3578 articles were excluded as being an irrelevant topic or not pertinent to the research question. A further 327 records were excluded from the remaining 385 articles after screening of the abstract contents, resulting in 58 potentially eligible studies, after which 1 additional study was identified through a supplementary search, resulting in the screening of 59 full-text articles. Subsequently, 12 papers were included in this review. The flow chart for the literature search and selection of studies is presented in . All studies were classified as “some concerns”. Individual scores by domain are presented in .

### 3.2. Study and participant characteristics

The eligible articles included in this review were published between 1984 and 2024 (mean ± standard deviation = 2015 ± 12; median = 2020), including 2 non-randomized , and 7 quasi-randomized/randomized crossover trials , , , , , , and 3 parallel randomized controlled trials (RCTs). , , The studies were conducted in Australia ( n = 5), , , , , the USA ( n = 2), , and Brazil, Tunisia, Canada, the UK, and Republic of Korea ( n = 1 for each country). Across the 12 studies, a total of 177 participants ( n = 14.8 ± 7.3; range: 5–32; median = 12) were included in this review. Ten of the studies included males only ( n = 154), , , , , , , , , , 1 study included a mixed cohort (7 males and 6 females), and 1 study included only females ( n = 10); the overall proportion of females is 9.0%. The participants’ training background ranged from sedentary/inactive to athletes, while 3 studies did not clearly report participants’ training status. , ,

The interventions varied in terms of type of sleep loss and whether a single session of exercise or short-term exercise interventions were employed ( ). Prior to the intervention period, a wide array of questionnaires was utilized, including the Epworth Sleepiness Scale, Pittsburgh Sleep Quality index, Horne-Osteberg morningness-eveningness, Berlin Questionnaire, UNIFESP Sleep Questionnaire, Mini-Sleep questionnaire, and Sleep questionnaire/diary ( ). In terms of sleep intervention, 6 studies were conducted in a laboratory setting, , , , , , whereas 4 studies were conducted in free-living conditions. , , , One study did not specify whether participants were in a laboratory or free-living condition, and another was unclear on whether all conditions were conducted in a laboratory setting or if it was only the sleep-restricted conditions ( ). One study did not employ any tool to evaluate sleep, while the remaining studies used sleep diary ( n = 7), , , , , , , wrist actigraphy ( n = 9), , , , , , , , , and/or polysomnography ( n = 2) , ( ). Regarding sleep interventions, total sleep deprivation was applied in 3 studies, , , while the remaining studies implemented partial sleep restriction in a single night ( n = 3) , , or a short-term period ( n = 6), , , , , , in which 4 h early sleep restriction was the most common ( n = 6) ( ). , , , , , Regarding exercise intervention, 4 studies employed a single exercise session, , , , while the remaining employed 2–6 exercise sessions spread over a 2-week period ( ). , , , , , , , The most common activities in the included studies were cycling ( n = 6) , , , , , and walking/running ( n = 4), , , , followed by a multimodal exercise ( n = 1) and resistance exercise ( n = 1). Regarding study outcomes, blood metabolomic ( n = 1), fasting ( n = 8) , , , , , , , and post-prandial glucose ( n = 3) , , and insulin ( n = 5) , , , , concentration, insulin sensitivity ( n = 4), , , , free fatty acids ( n = 2), , skeletal muscle transcriptome ( n = 2), , and gene expression and molecular markers ( n = 1) were reported ( ).

### 3.3. Blood markers

Eight studies examined fasting glucose , , , , , , , and 3 studies examined post-prandial glucose concentration. , , Regarding fasting measures, 5 studies revealed a higher fasting glucose concentration in sleep-restricted , , or sleep-deprived , participants compared to control, while 3 studies demonstrated no difference between conditions. , , In this subset of studies, glucose concentration was restored to normal values after exercise intervention. , , , In terms of post-prandial glucose concentrations, one of the studies demonstrated a higher glucose concentration in sleep-restricted participants when compared to control, while another reported no difference between conditions. Glucose concentration was increased immediately after exercise intervention in sleep-restricted participants in comparison to control ( ).

Six studies assessed insulin concentration in the fasting state or in response to an oral glucose tolerance test (OGTT), , , , , , of which, 4 studies assessed insulin resistance through the homeostatic model assessment of insulin resistance (HOMA-IR) in fasting samples, , , , and 3 studies assessed insulin sensitivity through the Matsuda Index based on the results of the OGTT. , , Three of the aforementioned studies reported a higher insulin concentration in sleep-restricted or deprived , participants, while 2 studies revealed no difference between conditions. , Furthermore, 3 studies reported no differences between sleep-restricted and exercised participants compared to control, , , while 2 studies reported a reduced insulin concentration in sleep-restricted or deprived following exercise in comparison to a sleep-loss condition ( ). In terms of insulin resistance and sensitivity, 1 study presented an increase in insulin resistance (HOMA-IR) and reduction in insulin sensitivity (Matsuda index), while 3 studies showed no difference between control, sleep-restricted/deprived, and sleep-restricted/deprived plus exercise for HOMA-IR , , and Matsuda index , ( ). Noteworthy is that HOMA-IR values were elevated at baseline in one of the studies, likely because the participants were described as obese (sleep restriction (SR) = 2.7 ± 0.4 vs. sleep restriction plus exercise (SREX) = 2.8 ± 0.5). A positive mean difference of +11.1% was observed in the sleep-restricted participants, whereas a negative mean difference of –28.6% was noted in the exercised participants under sleep restriction as compared to baseline.

In 2 studies, free fatty acids (FFAs) concentration was increased in sleep-restricted or deprived participants in respect to their controls. One of the above studies reported a higher concentration of FFA in sleep-restricted and exercised participants in comparison to baseline data, while another study reported no differences among control, exercised, and sleep-restricted and exercised participants. Lastly, investigation of the circulating metabolome revealed that 4 h of sleep opportunity reduced the concentration of 5 metabolites, while resulting in increases in 2 metabolites. Furthermore, a single session of continuous exercise significantly affected 18 metabolites in the same cohort ( ).

### 3.4. Molecular markers in skeletal muscle

Two studies employed a multi-contrast analysis to examine the skeletal muscle transcriptome, utilizing reactome gene ontology to compare transcriptomic alterations across different groups ( ). , Here, a 5-night period with only 4 h of sleep opportunity resulted in a decreased enrichment of genes associated with mitochondrial function, including pathways related to oxidative phosphorylation. Similarly, cellular pathways were downregulated in response to 3 nights and 9 nights of 5-h sleep opportunity, including oxidative metabolism, respiratory electron transport, complex 1 biogenesis, and citric acid cycle. Conversely, 3 sessions of HIIE or resistance exercise led to an increase in several pathways associated with mitochondrial function or oxidative metabolism.

One study found that mitochondrial respiratory function and biogenesis were negatively impacted by sleep restriction, while no significant differences were observed in mitochondrial activity and content across conditions such as normal sleep (NS), SR, and HIIE SR + HIIE. However, HIIE mitigated the adverse effects of sleep loss on mitochondrial respiratory function and biogenesis. Additionally, there were no significant differences within (pre vs . post) or between groups (NS, SR, and SR + HIIE) for peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α), dynamin-related protein 1 (DRP1), mitofusin 2 (MFN2), tumor protein p53, basic helix-loop-helix ARNT-like protein (BMAL), and solute carrier family 2 member 4/glucose transporter 4 (GLUT4) content ( ).

4. Discussion

This review describes a systematic search and review of studies investigating whether acute and short-term exercise interventions can mitigate negative effects of experimentally-induced sleep loss on metabolic markers in blood and skeletal muscle in humans. While the number of studies in this domain is small ( n = 12) at present, the key finding is that exercise is effective as a therapeutic intervention to mitigate the negative effects of sleep loss on metabolic markers of glucose metabolism and insulin sensitivity, at least in the short-term intervention studies upon which we focused our search.

Experimentally-induced sleep loss has somewhat consistent negative impacts on glucose homeostasis, although the degree of sleep loss and duration imposed are influential factors. Most studies reported a higher fasting glucose concentration in participants subjected to a single night of total sleep deprivation , or 5 consecutive nights of 4-h sleep restriction. , On the other hand, the remaining studies reported no differences between conditions, which consisted of a single night of 4-h sleep restriction or 5 nights of 6-h sleep restriction. These results suggest a period of “moderate” sleep restriction (e.g., 5 nights of 6-h sleep restriction) does not cause the same magnitude of changes as total sleep deprivation or consecutive nights of partial sleep restriction (e.g., up to 7 nights of 5-h time in bed), which are both well-established to negatively impact glucose concentration. , , , Further research is required to fully describe the effects of diverse durations of sleep loss (per night and number of days) on glucose homeostasis in adults.

Multiple mechanisms underpin the abnormal glucose metabolism following sleep loss. These mechanisms involve an increment of inflammatory markers (e.g., C-reactive protein (CRP), interleukin 1beta (IL-1β), IL-6, IL-17, and tumor necrosis factor-alpha (TNFα)), , , hormone imbalance (e.g., insulin, growth hormone, cortisol, testosterone, leptin, and ghrelin), , , , , increased sympathetic nervous system activity, , , elevated non-esterified fatty acids concentration, , , , , and a reduction in sensitivity of the insulin signaling pathway. , , For example, an experimental study recruited 17 health participants to evaluate cerebral metabolic rate and neuronal synaptic activity following 85 h of total sleep deprivation. The findings reveal a significant decrease in brain glucose utilization following 24 h of sleep deprivation. Relatedly, a randomized crossover trial employed 4 days of 4.5-h sleep opportunity and reported a downregulation of Ser 473 phosphorylation of Akt (pAkt) and an approximately 30% reduction in the insulin signaling pathway in adipocytes. In contrast, another study observed no downregulation in pAkt signaling in skeletal muscle following 2 nights of 50% sleep restriction, but there was a reduction of ∼19% in whole-body insulin sensitivity measured by Matsuda index. Analogously, a literature review examining the effects of sleep restriction (e.g., up to 14 nights of 5.5 h of sleep restriction) and fragmentation (e.g., up to 3 nights of suppression of slow wave sleep and/or rapid eye movement) on energy metabolism revealed a reduction in insulin sensitivity, ranging from 16% to 32% across 9 experimental studies.

The development of T2D is intricately linked to the complex interaction between inadequate sleep and metabolic dysfunction, a relationship that has been the subject of extensive research. , Physical activity and/or exercise have been widely implemented in both healthy and diseased individuals to enhance metabolic system functioning. In adults at risk for T2D, even minor adjustments, such as replacing prolonged sitting with standing or stepping, can improve glucose tolerance and insulin sensitivity, both of which are closely linked to a reduced incidence of T2D. The protective effect of physical activity on the risk of T2D is observed even among individuals with insufficient sleep. Engaging in physical activity or structured exercise training could therefore serve as a potential alternative for managing glucose homeostasis and insulin sensitivity, as well as for reducing the risk of T2D even in a condition of insufficient sleep. , Based on the studies included in our analysis, acute and short-term exercise interventions may offset the negative effects of sleep loss on fasting glucose , , and insulin concentrations , , , , as well as on insulin sensitivity. Sleep loss is likely to negatively impact peripheral insulin sensitivity, , , , and exercise may improve glucose tolerance and insulin sensitivity through insulin-dependent and -independent mechanisms. , One of these mechanisms is related to elevated FFA concentrations that are associated with increased oxidative stress and inflammation, which further exacerbate insulin resistance. Importantly, some of the beneficial effects of exercise may be due to positive effects on fatty acid uptake and oxidation. , , In the context of the present review, it is notable that 6 sessions of HIIE mitigated elevations in FFA concentrations induced by sleep loss. These findings suggest the potential efficacy of exercise interventions for mitigating the adverse metabolic effects of sleep deprivation could include effects on lipid metabolism. Yet, despite the largely positive effects noted in our findings, further research is necessary to determine whether exercise training can mitigate adverse effects over the longer term, such as with chronic sleep loss, particularly in shift workers who are at an elevated risk of developing T2D. ,

Considering the adverse effects of sleep loss on skeletal muscle, a single night of sleep deprivation induces a catabolic environment, which in turn results in a subsequent increase in protein breakdown. , An experimental study including 24 healthy participants revealed that 4-h sleep opportunity is associated with a reduction in myofibrillar protein synthesis. At the molecular level, insufficient sleep has been associated with an increase in inflammatory and immune-related pathways, which can negatively influence regeneration processes in skeletal muscle. A negative enrichment of genes associated with mitochondrial function, including pathways related to oxidative phosphorylation, in response to 5 nights of 4-h sleep restriction was also observed in 20 recreationally active participants. Similarly, cellular pathways were downregulated in response to 3 nights and 9 nights of 5-h sleep opportunity, including oxidative metabolism, respiratory electron transport, complex 1 biogenesis, and citric acid cycle. As a countermeasure, 3 sessions of HIIE or resistance exercise led to an increase in several pathways associated with mitochondrial function or oxidative metabolism in sleep-restricted individuals. However, in a related analysis of the latter study, the protein content of PGC-1α, DRP1, MFN2, p53, BMAL, and GLUT4 were not different within (pre vs . post) or between groups (RS, SR, and SR + HIIE). The current literature is scarce in terms of adaptations in skeletal muscle in response to exercise training during periods of sleep loss, whether in the context of metabolic health or athletic performance. Therefore, further research is warranted to better understand the impact of prolonged periods of sleep restriction on adaptations to exercise training in health and disease. New avenues could include older adults, considering that older adults often experience reduced sleep duration and quality , , as well as age-related declines in skeletal muscle mass and function, for which exercise is an essential countermeasure.

This review has several strengths to be acknowledged. First, to the best of our knowledge, this is the first systematic literature review to evaluate the effect of acute exercise or short-term exercise interventions on metabolic markers in blood and skeletal muscle during experimentally-induced sleep loss. This topic is evidently of increasing interest given that most of the articles included in this review were published recently (e.g., 2018–2024), with the exception of 2, which were published in 1984 and 1993. Second, to ensure methodological rigor, this review was pre-registered on PROSPERO, and a comprehensive search was conducted across 6 different databases as well as in the reference lists of included papers and relevant reviews. Lastly, this review adhered to the recommended PRISMA guidelines for systematic literature reviews, with screening, data extraction, and risk of bias assessment performed independently by at least 2 researchers (VSF and LM).

On the other hand, this review can be considered to have several limitations. First, the number of studies related to each outcome and their sample sizes are limited, including lipid profile ( n = 2), blood metabolomics ( n = 1), muscle transcriptome ( n = 2), and skeletal muscle protein expression and content ( n = 1); and on the whole, there were not yet enough studies or common effects with which to proceed to meta-analysis of outcomes. Second, there is considerable variation in exercise models (e.g., walking/running and cycling), volume (ranging from minute to hour), and intensity (from maximum efforts to low-intensity exercise). Similarly, the sleep loss interventions employed in each study varied widely, ranging from total sleep deprivation (1–3 nights) to multiple nights of partial early or late sleep restriction (5 nights with 2 h or 4 h of sleep restriction). Fourth, many studies reported insufficient data or used tools/devices with low reliability and high variability to report sleep data. Relatedly, there was limited use of advanced techniques to evaluate glucose tolerance and insulin sensitivity in greater detail, such as the intravenous glucose tolerance test or the hyperinsulinemic-euglycemic clamp. These approaches, combined with tissue sampling, would provide greater insight into the mechanisms by which sleep loss negatively impacts glycemic control as well as the mechanisms by which exercise mitigates these effects. Another important methodological consideration is the timing of exercise in the temporal context of the sleep loss intervention as well as the timing of assessment of insulin sensitivity relative to the most recent session of exercise (i.e., the “last bout effect” in designs where several sessions of exercise have been performed and when sleep loss is over a number of days). In general, the considerable heterogeneity in populations, study designs, and outcome measures in this domain mean that much work remains to be done to better understand the effects of exercise as a mitigation strategy. Lastly, the literature is scarce regarding the interaction between sleep loss and exercise in female populations. Therefore, future experimental studies may wish to investigate the effectiveness of exercise intervention on metabolic parameters and molecular markers in blood and skeletal muscle in females, including across the life course given the negative impact of perimenopause on sleep duration and quality. ,

5. Conclusion

Although the majority of studies reported a favorable impact of exercise with respect to mitigating the negative effects of sleep loss on metabolic parameters, these findings were not consistent. The results were robust regarding fasting glucose and insulin concentration, but the current literature is equivocal and requires further investigation concerning insulin sensitivity, lipid profile, and molecular markers in skeletal muscle. Overall, we contend that acute exercise or short-term exercise intervention has therapeutic potential, at least during short-term periods of sleep loss, to mitigate the adverse effects of insufficient sleep on fasting markers of glucose homeostasis.

Authors’ contributions

BE and VSF contributed to the study’s conception and design, literature search, study selection, data extraction, risk of bias assessment, data analysis, and manuscript drafting; LM made substantial contributions to the literature search, risk of bias assessment, and data extraction, while ADOH contributed to the literature search and study selection. All authors have read and approved the final version of the manuscript, and agree with the order of presentation of the authors.

Competing interests

The authors declare that they have no competing interests.