The effect of total sleep deprivation on autonomic nervous system and cortisol responses to acute stressors in healthy individuals: A systematic review

Objective: To synthesise the literature examining the autonomic nervous system (ANS) and cortisol responses to an acute stressor following total sleep deprivation (TSD) in healthy adult subjects


Introduction
Sleep deprivation affects the stress systems and may be a potential mechanism by which sleep loss leads to poor health (Nollet et al., 2020).Sleep deprivation is commonly regarded as a stressor in itself, because the main stress systems, which are the autonomic nervous system (ANS) and the hypothalamic pituitary adrenal (HPA) axis become activated (Meerlo et al., 2008).Moreover, the intrinsic circadian system, which is affected by episodes of sleep loss, is clearly and reciprocally connected with the stress system (Agorastos et al., 2020a).Both systems are essential to life and control one another's functions to get the organism ready for upcoming periodic challenges (Agorastos et al., 2020a).Thus, a comprehension of the interplay between sleep deprivation and stress is crucial in understanding pathophysiological routes that modulate disease risk.
The activation of the stress system involves immediate ANS physiological response [e.g., increase in heart rate (HR) and blood pressure (BP)] and delayed response, including hormonal responses (e.g., increased cortisol release) (Chrousos, 2009;Russell and Lightman, 2019).Changes in ANS and cortisol responses to stress are clinically important.A greater BP and cortisol reactivity to laboratory induced stress and poor recovery are associated with a higher incidence of hypertension (Chida and Steptoe, 2010;Hamer and Steptoe, 2012;Phillips et al., 2013;Turner et al., 2020).On the other hand, evidence suggests that low cardiovascular reactivity to stress, as well as low cortisol reactivity, may have serious adverse consequences for health and behaviour and are associated with obesity, depression, cardiovascular disease and other negative outcomes (Phillips et al., 2013;Turner et al., 2020).
Two systematic reviews addressed how different aspects of sleep may influence cortisol responses to acute stress (Van Dalfsen and Markus, 2018;Zhao et al., 2021).The results of both reviews indicated that inadequate sleep increased, blunted, or did not affect cortisol responsiveness to acute stressors (Van Dalfsen and Markus, 2018;Zhao et al., 2021).The inconsistent results in the literature may be related to different sleep deprivation protocols, types of stressors, participant characteristics, prior stress experience and different indicators of cortisol used (Zhao et al., 2021).In addition, other studies have shown that sleep loss causes changes or not in cardiovascular stress reactivity (Kato et al., 2000;Mezick et al., 2014) suggesting that the results using ANS markers as an outcome are inconsistent too.
To the best of our knowledge, no systematic reviews have investigated how sleep loss may influence ANS responses to acute stress.In addition, a systematic review with a focus on studies that did standardised experimental sleep manipulation [i.e. total sleep deprivation (TSD)] would help, from clinical and pathophysiological perspectives, to understand better the effects of sleep loss on ANS and cortisol responses to stress.Given that numerous occupations have involved working night shifts and essential services must always be available in a "24-hour society.",studies with a focus on TSD are important and could be useful.This systematic review will provide a comprehensive overview of the current evidence on whether TSD, compared with a normal night of sleep, promotes changes in ANS and cortisol responses to acute stressors in healthy individuals.The objective of this study is to evaluate and synthesise studies that investigated the ANS and cortisol responses to acute stressors after TSD in healthy individuals.Although the relationship between sleep and stress is bidirectional, this review will describe the one-directional influence of TSD on ANS and cortisol responses to an acute stressor.

Methods
This study follows the PRISMA -Preferred Reporting Items for Systematic Reviews and Meta-Analysis -checklist and diagram flow (Page et al., 2021) (Appendix A).We registered the protocol of this systematic review in the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42022293857).

Search strategy
We searched five electronic databases [Medline (via Ovid), Embase (via Ovid), PsycINFO (via Ovid), CINAHL complete and Scopus] for available English language studies up to January 2024.We used broad search terms and MeSH terms related to "sleep deprivation", "acute stress", "ANS" and "cortisol responses".See Appendix B for the full search strategy.

Eligibility criteria
We aimed to identify studies exploring the influence of TSD on ANS markers and cortisol responses after experimental laboratory stress.The inclusion criteria were (1) healthy human subjects; (2) use of a standardised psychosocial (e.g.free speech task, mental arithmetic) or cognitive (e.g.Stroop-task) or physical (e.g.cold pressor task) laboratory challenge; (3) volunteers were total sleep deprived (complete absence of sleep) more than 24 hours; (4) TSD was induced by experimental and controlled methods; (5) clinical studies of all types (randomised, non-randomised, cluster, crossover, etc.); ( 6) assessed at least one variable related with the ANS or cortisol; and (7) only full texts in peerreviewed journals (no industry reports or grey literature were included).The exclusion criteria were: (1) observational studies including crosssectional, case-control, and cohort designs, case reports, opinion articles, commentaries, letters, conference abstracts, reviews, and metaanalyses and (2) Physical exercise as a laboratory stress challenge.We excluded studies investigating physical exercise as stress because there is strong evidence suggesting that acute exercise reduces stress reactivity (Morava et al., 2024).Furthermore, participant fitness/physical activity levels (possible confounder) may influence ANS and cortisol responses to acute physical exercise as a stress task (Mariano et al., 2023).We also excluded animal and in vitro studies.There was no restriction regarding the study location, sample size and date of publication.

Study selection
We imported the references into Covidence (Covidence, Melbourne, VIC) for screening.After removing duplicates, two authors (RM and MB) independently assessed titles, abstracts and full texts for inclusion.Any disagreements were resolved through discussion or by a third author (ESA) if arbitration was necessary.The process of study selection is shown in the PRISMA diagram (Fig. 1).

Data extraction
The following information from each paper was extracted independently by RM and MB.The information extracted included characteristics of the study setting (where not specified, the setting was assumed to be the first author's affiliation), where the participants were recruited (where not specified, we assumed that participants were recruited from the general population), study design, sex difference, sample size, age, objectives of the study, outcomes analysed and results.Also, we extracted the type of stress protocols used and the time that it was applied, place of sleep or TSD, length of TSD and the summary of the outcomes.Finally, we extracted possible confounding variables from each study (Appendix C).

Risk of bias assessment
Two independent assessors determined the risk of bias (MB and ESA), and the discrepancies were resolved through a consensus.We used the two risk of bias assessment tools as we included different study types in this systematic review.The Cochrane Risk of Bias 2.0 (RoB 2) was used for randomised controlled trials and an official adaptation for crossover studies (Sterne et al., 2019).The Risk of Bias in Non-randomized Studies (ROBINS-I) was used for non-randomised trials (Sterne et al., 2016).Depending on the study type, RoB 2 classifies studies into "low risk", "some concerns" and "high risk" of bias, while the ROBINS-I classifies the studies into "low risk", moderate risk", "serious risk", critical risk" of bias or "no information".

Study selection
We identified a total of 7127 records through database searches and 3538 records remained after removing duplicates.Of those, 3495 were removed during title and abstract screening, with a further 27 studies excluded during full text screening.Sixteen studies were eligible for inclusion in this review.Primary reasons for exclusion based on full text screening included: Conference abstract (n = 20), absence of a control/ comparative group (n = 2), no cortisol analyses in the control group (n = 1), no laboratory stress in the control group (n = 1), no standardised stress protocol (n = 1), sleep deprivation after exposure to stress (n = 1) and use of interventions to manipulate the stress responses (n = 1).See Fig. 1 and Appendix D.

Studies design and setting
Table 1 presents the features of the 16 eligible studies.Of those, four were non-randomised experimental trials, seven were crossover studies and five were randomised controlled trials.The crossover studies included TSD and a control night (at least 7 hours of opportunity to sleep), of which the order was randomly/counterbalanced assigned.The studies were set in The United States of America (n = 8), Europe (n = 4), Singapore (n = 1), Turkey (n = 1), Lithuania (n = 1), and Canada (n = 1).

Objectives
All studies investigated the ANS and/or cortisol responses to an acute stressor after a TSD condition.Bottenheft et al. (2023) investigated the combined effects of TSD and acute social stress on some markers of SNA and cortisol as a secondary outcome.One study tested the effect of sex (Yang et al., 2012) and two studies tested the effect of age (Robillard et al., 2011;Schwarz et al., 2018) on ANS or cortisol responses to stress after TSD.

TSD protocol
The summary information about the TSD protocol is described in Table 2.The range of hours of TSD before acute stress was from ~24-86 hours.In all studies, participants were sleep-deprived in a laboratory.In the sleep control (SC) condition, the participants slept at home in three studies (Åkerstedt and Fröberg, 1979;Schwarz et al., 2018;Bottenheft et al., 2023) and slept in the laboratory in 13 studies.

Acute stress protocol
A range of stress protocols was used (Table 2).One study used Stroop Task (Åkerstedt and Fröberg, 1979), three studies the Cold Pressor Test (CPT) or Cold Air Test (Fiorica et al., 1968;Caine-Bish et al., 2005;Costa et al., 2010), four studies used a Standardise Trial Social Stress Test (TSST) (Minkel et al., 2014;Vargas and Lopez-Duran, 2017;Schwarz et al., 2018;Bottenheft et al., 2023), one study used psychosocial stress (Liu et al., 2015), one study used a Maastricht Acute Stress Test (MAST) (Hansen et al., 2021), one study used acute heat stress (Cernych et al., 2021) and one study used the orthostatic challenge as a stress (Robillard et al., 2011).In addition, four studies used a combination of protocols including one study using the Stroop Task and speech task (Franzen et al., 2011), one study using the CPT and mental stress (MS) task (Yang et al., 2012), one study used the CPT, the MS, the maximal forearm ischemic response and the handgrip test (Kato et al., 2000) and, finally, one study used Valsalva manoeuvre, CPT, the MS and the handgrip test (Bozer et al., 2021).
Regarding the time of day (Table 2), eight studies applied the stress protocol after midday (PM), while six studies applied the stress protocol before midday (AM), and in one study the acute stressor was applied between AM and PM (Vargas and Lopez-Duran, 2017).Additionally, the stressor was used at two times in the study of Åkerstedt and Fröberg (1979), the first time between 11:00 AM and 02:00 PM, and the second time between 11:00 PM and 02:00 AM.

Effects of age and sex on the interaction between TSD and stress response
Ten studies had a mixed sample (male and females), while five studies included only males and one did not report sex differences.Only Yang et al. ( 2012) evaluated sex differences and found no sex differences in the MAP, HR, FVC and FBF reactivities to stress following either sleep condition.
Regarding age status, two studies evaluated the influence of age on stress responses after TSD.Robillard et al. (2011) found no significant interaction between age, stress and sleep condition for SBP, DBP, or HR.Schwarz et al. (2018) found no difference in the TSST reactivity or recovery after TSD for any of the outcome measures and age did not moderate the effect of TSD.

Confounders
Appendix C shows some of the possible confounder effects mentioned by the authors, including smoking status, sleep assessment before/during the protocol, alcohol, abstention from physical exercise, physical activity levels, caffeine, menstrual cycle, diet, use of devices during sleep deprivation, lightning during sleep conditions, transmeridian travel and shift/night work.Other confounder effects were    One half of the studies mentioned smoking information of subjects (including non-smokers).Ten studies reported sleep assessment before the experimental protocol and six during the protocol.Eleven studies reported that the subjects abstained from alcohol before protocol.Twelve studies reported information about physical activity (abstention before protocol, abstention/permission during protocol) and three studies reported that subjects were physically active.Almost all studies reported consumption of caffeine (abstention before protocol or consumption during TSD) (n=13).
Only three studies that recruited female subjects (three out of 11) reported their menstrual cycle.All studies reported information about the time of the day of stress was applied and the moment of outcomes was measured.Nine studies did not report specifically the position of measurements.Only two studies did not mention the dietary intake of subjects and we did not find information about devices (e.g., video games, telephone use, books, television) during TSD in nine studies.Moreover, six studies reported information of lighting during the protocol and five studies reported preceding transmeridian travel or shift/ night work as ineligible participants.

Risk of bias assessment
Figs. 2 and 3 were created using the Robvis tool (McGuinness and Higgins, 2021) and provide information about bias judgments among each bias domain and the overall risk of bias.The risk of bias in randomised controlled trials and crossover studies is presented in Fig. 2. Overall, we classified ten studies as "some concerns'' in the RoB 2 assessment (Costa et al., 2010;Robillard et al., 2011;Franzen et al., 2011;Yang et al., 2012;Minkel et al., 2014;Liu et al., 2015;Vargas and Lopez-Duran, 2017;Schwarz et al., 2018;Cernych et al., 2021;Bottenheft et al., 2023).We assigned two studies as "high" risk of bias (Kato et al., 2000;Caine-Bish et al., 2005).
This may be attributed to the fact that all studies were classified as "some concerns" in the "randomisation process" (no information about the allocation sequence was concealed) and "deviations from intended intervention" (participants and/or assessors were aware of the intervention).Concerning the "period and carryover effects" item, two crossover studies were assigned as "high" risk of bias because of an insufficient wash-out period (we consider at least 7 days) between SC and TSD protocols, four days in the study by Kato et al. (2000) and only 48 hours in the protocol of Caine-Bish et al. (2005).Despite this, a higher risk of bias in RCTs was not associated with overestimated results.
The summary of the risk of bias in non-randomised trials is presented in Fig. 3.In the overall ROBINS-I, we classified one study as "moderate" risk of bias (Hansen et al., 2021) and in three studies as "serious" risk (Fiorica et al., 1968;Åkerstedt and Fröberg, 1979;Bozer et al., 2021).
The "confounding" item (see Appendix C) was responsible for the "serious" risk of bias in these studies as most confounder effects information was not even mentioned.In addition, only Hansen et al. (2021) reported statistically appropriate methods to control for some confounders, although they did not control for all possible confounders and were assigned as "moderate" risk of bias.Moreover, the study of Hansen et al. ( 2021) was assigned as "moderate" risk of bias in the "missing data" item, because they reported that seven subjects (>5 % of the sample) participated in the study but were not included in the formal analysis.Furthermore, concerning the "measurement of outcomes" item, we assigned one study as "moderate" risk of bias because the method was not the most appropriate, as also reported by the authors through discussion (Bozer et al., 2021).A higher risk of bias also does not appear to be associated with overreported results in non-randomised studies.

Syntheses of results
The synthesis of results is summarised in Tables 2 and 3. We presented the results divided into four main sections: (1) Baseline, (2) Acute stress, (3) Recovery and (4) Acute stress -Time of the day.

Baseline
Most of the studies (n = 10) reported that the baseline levels of, at least, one marker of ANS or cortisol were affected by the TSD protocol (p < 0.05).TSD increased BP in all studies (n = 5) that compare baseline levels between TSD and SC conditions, while only one study out of six found that TSD decreased baseline HR (Fiorica et al., 1968).Regarding cortisol, three studies found that TSD increased baseline levels and three studies reported no differences in baseline levels of cortisol between TSD and SC groups.

Acute stress
The majority of the studies (n = 14) found that acute stress challenge changed significantly (p < 0.05) baseline levels of ANS markers or cortisol response in both TSD and SC groups.Generally, the stress protocol promotes an increase in the investigated outcome although the type of stress can influence the direction of the stress response.For example, Robillard et al. (2011) found a decrease in the SBP after an orthostatic challenge and Cernych et al. ( 2021) found a decrease in BP after heat stress.Although Kato et al. (2000) and Bozer et al. (2021) reported an increase in ANS markers after stress, no statistical analyses were reported.
Regarding differences between TSD and SC groups, 10 out of 16 studies reported changes in at least one marker of ANS or cortisol responses to acute stressors (Table 3).Overall, 64 results were extracted from 16 eligible studies.Of those, TSD changed ANS or cortisol responses to acute stressors in 17 reported results (27 %) (p < 0.05) while 47 reported results were found to be not significantly (p > 0.05) affected by TSD.Franzen et al. (2011) found that SBP was higher in the TSD condition compared with the SC condition during the stress task (effect size d = 0.73; p = 0.009).Yang et al. (2012) found higher HR reactivity to acute stress in sleep deprived conditions (p = 0.001).Minkel et al. (2014) found that sleep deprivation was associated with an amplified cortisol response in five min (effect size d = 0.89; p = 0.003), 20 min (effect size d = 1.15; p = 0.003) and 40 min (effect size d = 0.46; p = 0.003) after the stressor.In addition, Liu et al. (2015) showed that participants overall presented an increased skin conductance reactivity following the psychosocial stress and that skin conductance in TSD participants continued to rise with increasing stress (p = 0.04).Cernych et al. (2021), using acute heat stress (sauna) as a stressor, found that SBP, DBP and MAP were significantly higher in TSD conditions during sauna on day 2 compared with sauna on day 1 (effect size η2 > 0.16, p < 0.05), suggesting that TSD increases cardiovascular stress responses (Cernych et al., 2021).Bozer et al. (2021) used four types of acute stress (Valsalva manoeuvre, CPT, the MS and the handgrip test) and observed that percent change values of SBP (p < 0.01) and DBP (p < 0.05) were higher after the handgrip test in TSD compared with SC condition.In addition, Bozer et al. (2021) found that DBP increased (p < 0.02) during SC and this increase was intensified on the TSD day after CPT.

TSD blunted ANS and cortisol stress responses.
In general, four studies found that TSD blunted ANS stress response.There was a significant difference in percent change values for SBP (p < 0.05) after the Valsalva manoeuvre test (SBP increased during SC but not during TSD condition) (Bozer et al., 2021).Cernych et al. (2021) found no sauna effect (sauna day 1 vs. sauna day 2) on HR in the SC trial, but the TSD trial led to lower overall HR to sauna exposure on day 2 compared with day 1 (effect size η2 > 0.19, p < 0.05).After TSD, a significantly higher HRV (effect size η2 > 0.24, p < 0.05,) was observed on day 2 compared with sauna exposure on day 1 (Cernych et al., 2021).Robillard et al. (2011) found that TSD attenuated the SBP orthostatic response compared with SC (p < 0.05).Fiorica et al. (1968) found a similar HR response between SC and TSD, except for the fourth stress exposure (82 h of TSD) when no change from control was observed only in the TSD group.
Regarding cortisol stress responses, Vargas and Lopez-Duran (2017) showed that participants in the sleep deprivation condition had a blunted cortisol response to the stressor compared to their non sleep deprived peers after controlling for covariates (p = 0.03).Also, Hansen et al. (2021) found a blunted cortisol response to the stressor (effect size d = 0.84, p = 0.039) compared to their non-sleep deprived.

TSD does not change ANS and cortisol stress responses.
Overall, seven studies did not find any significant effects (p > 0.05) of TSD in ANS or cortisol stress responses for any of the outcomes evaluated.In addition, another five studies found no changes in, at least, one marker of ANS or cortisol stress responses.Åkerstedt and Fröberg (1979) and Costa et al. (2010) found no significant difference in catecholamine concentrations after acute stress between TSD and SC conditions.Costa et al. (2010) found no significant difference in HR during cold exposure between groups after 34 and 58 hours of TSD.Costa et al. (2010) found no significant difference in cortisol after acute stress between TSD and SC conditions while Robillard et al. (2011) found that TSD did not affect the DBP and HR orthostatic response compared with SC (p < 0.05).Franzen et al. (2011) found no significant difference in HR and DBP in the TSD condition compared with the SC condition during the stress task.Kato et al. (2000) found that TSD did not affect MAP, HR and MSNA responses to any stressful stimuli compared with SC condition.Caine-Bish et al. (2005) investigated the effects of 12℃ cold exposure for 180 minutes in TSD and SC conditions and there was no significant effect between groups (including interaction time x groups) for adrenaline, noradrenaline and cortisol.Yang et al. (2012) used markers of sympathetic activity (FVC, FBF and MSNA) and MAP responses to MS or CPT and did not find any significant difference between SC and TSD conditions.Minkel et al. (2014) found no significant effect of TSD at baseline or in response to the stressor.Schwarz et al. (2018) found that after TSD or a normal night's sleep, cortisol and sAA significantly increased in response to the TSST similarly and HRV significantly decreased in response to the TSST similarly in both groups.Bozer et al. (2021) found that HR responses to stress were not significantly different among the SC, TSD, and sleep recovery conditions for all stress tests.Bottenheft et al. (2023) found no interaction between TSD and TSST for cortisol, HR and EDA.

Recovery
Only four studies reported the recovery values in the SC group separately (Costa et al., 2010;Schwarz et al., 2018;Cernych et al., 2021;Hansen et al., 2021).In 12 studies, the ANS and cortisol response differences between recovery and stress were not reported (n = 6) or were not evaluated (n = 6).Regarding cortisol, three studies found that TSD increased baseline levels and three studies found that plasma cortisol concentration returned to baseline levels after 1 h into recovery, while plasma norepinephrine concentration remained significantly higher than baseline levels 1 h into recovery (P < 0.05).The other three studies reported that cortisol (n = 2), BP (n = 1), HRV (n = 2), HR (n = 1) and sAA (n = 1) returned to baseline levels after stress.
Five out of eight studies reported no difference between SC and TSD conditions during the recovery session.Yang et al. (2012) and Minkel et al. (2014) found, respectively, that HR and cortisol were elevated during the recovery period in the TSD group.Also, Cernych et al. (2021) found higher values of HRV and BP during the recovery stage (R4) from heat stress in TSD conditions during the sauna on day 2 compared with the sauna on day 1.

Acute stress -time of the day
Five out of eight studies that applied the AM stress protocol, found at least one significant difference in ANS responses.Two studies that applied the stress protocol during AM did not find differences in cortisol responses between SC and TSD conditions (Costa et al., 2010;Bottenheft et al., 2023).Concerning studies that applied the PM stress protocols, four out of six found a significant difference in cortisol (n = 2) or ANS responses (n = 2).Minkel et al. (2014) found a significant increase, while Hansen et al. (2021) found a significant blunted cortisol response.In addition, Cernych et al. (2021) found a significant difference in ANS responses applying the stressor in the early evening.Moreover, Åkerstedt and Fröberg (1979), did not find any differences in catecholamine reactivity, applying the stressor between 11:00 PM and 02:00 PM or 11:00 PM and 02:00 AM.Finally, Vargas and Lopez-Duran (2017) found a significant blunted response in cortisol in the late morning/early afternoon.

Overview
Sleep deprivation and acute stress commonly occur in today's society, however evidence on this topic is still limited.We included 16 studies with 581 participants.Overall, results show that TSD leads to exaggerated or blunted ANS markers or cortisol responses to laboratory stressors providing evidence for the "bidirectional multi-system reactivity hypothesis" proposal by (Turner et al., 2020).It appears that intermediate (between being too big and too small) or "Goldilocks" stress responses are healthy adaptive stress responses (Turner et al., 2020).In addition, the eligible studies with a lower risk of bias (some concerns) found inconsistent results although 80% of the studies considered as a high or serious risk of bias, did not find any significant difference in ANS or cortisol responses to an acute stressor between TSD and SC conditions.
Regarding the insignificant outcomes, TSD did not change ANS or cortisol responses to acute stressors in 73 % of the total reported results.A simple calculation of results could offer a helpful summary, but it could also present an unrealistically basic evaluation of the topic.Although there is no strong compelling evidence for or against TSD on ANS and cortisol responses to acute stressors in healthy individuals, the effects of TSD on stress responses may be better understood by an in depth assessment of the combinations of results between and among studies.In this context, several factors may have contributed to the inconsistent results found in the present review, such as the length of TSD, different types of stressors, which marker of ANS was used and how cortisol was assessed.

TSD protocols
TSD is the most widely used laboratory-based sleep manipulation approach.Many people who regularly work 'shifts' are exposed to it (e.g., truck drivers, healthcare professionals, anesthesiologists, and flight attendants) (Reynolds and Banks, 2010).All studies in this review, except Costa et al. (2010), evaluated ANS or cortisol responses following acute stress and a period of between ~24-36 hours of TSD.In addition, Fiorica et al. (1968) analysed catecholamine responses to acute stress after 58 and 82 hours of TSD and Åkerstedt and Fröberg (1979) after 40-43 and 49-52 hours.There is disagreement within the literature regarding ANS and cortisol responses to an acute stressor after ~24-36 hours of TSD which suggests that there is not a clear number of hours for TSD to promote changes in acute stress responses.
All of the included studies that found significant differences in ANS or cortisol responses following acute stress used a single night of TSD.Although Fiorica et al. (1968) found no change in HR during could exposure only in the TSD group, the authors suggested that 84-86 hours of TSD does not induce the expected increase in adrenal and sympathetic activity because physiological functions operating under less cerebral control are more resistant to the effects of sleep deprivation.In this context, we believe that TSD protocols higher than 36 hours may cause short term adaptations to the stress system leading to no changes in ANS or cortisol responses.

Acute stress protocols
A variety of stress tasks were performed to elicit acute stress at different times during the day inducing multiple responses.For example, stress protocols that involve physical demands, like the CPT, seem to be especially effective in eliciting ANS responses (Bali and Jaggi, 2015).On the other hand, stressors that are uncontrollable and represent a threat to social evaluation, like the TSST, are more likely to generate stress responses related to the HPA axis (Bali and Jaggi, 2015;Skoluda et al., 2015).Furthermore, differences in methodological variation of protocol for the same stress task can lead to differences in the magnitude of reactivity (Goodman et al., 2017).
All of the stress protocols used, independent of the time of the day it was applied, significantly increased at least one marker of ANS or cortisol in the SC group, suggesting that these protocols were effective.However, the results pertaining to TSD group responses varied considerably.For example, no significant differences between TSD and SC were found in studies that used cold (Fiorica et al., 1968;Caine-Bish et al., 2005;Costa et al., 2010), TSST (Vargas and Lopez-Duran, 2017;Bottenheft et al., 2023) or Stroop task (Åkerstedt and Fröberg, 1979).On the other hand, most of the studies that used multiple stressors found significant differences between TSD and SC.These results showed that applying multiple stressors had a higher chance of finding changes after TSD while the studies that only applied cold as a stressor had a lower chance of finding changes in stress responses.

ANS responses to an acute stressor after TSD
HR was the most used ANS outcome in the studies included in our review while HRV was evaluated only in two studies.Cernych et al. (2021) demonstrated lower values of HR and fasting HRV recovery, due to the enhanced vagal-mediated autonomic control.Yang et al. (2012) showed a significant increase in HR after MS and CPT in sleep deprived participants accompanied by a reduced HR recovery from stress.Both HR reactivity and HR recovery from acute stress may predict disease status in the future (Light et al., 1992;Heponiemi et al., 2007).Despite that, five out of seven studies did not report significant HR differences between TSD and SC conditions (Fiorica et al., 1968;Kato et al., 2000;Franzen et al., 2011;Bozer et al., 2021;Bottenheft et al., 2023).Bozer et al. (2021) hypothesised that distinct brainstem centres exist for BP and HR, and sleep loss might change vascular responses, increasing BP, without affecting HR.
Different results of BP in sleep-deprived individuals following acute stressors were found in the included studies and SBP was the outcome more sensitive to TSD.Considering increased values of SBP (Franzen et al., 2011;Bozer et al., 2021;Cernych et al., 2021) DBP (Bozer et al., 2021;Cernych et al., 2021) or MAP (Cernych et al., 2021), a systematic review reported that TSD could promote changes in BP or activate compensatory pathways that favour its increase, both through sympathetic activation and parasympathetic inhibition (Sá Gomes E Farias et al., 2022).However, lowered SBP values (Robillard et al., 2011;Bozer et al., 2021) would indicate a dual effect of TSD, since the study by Bozer et al. (2021) found divergent results for BP stress dependent.
Several studies have failed to show an effect of TSD followed by an acute stress on markers of autonomic function.Cold exposure, used in most studies that investigated catecholamines (Fiorica et al., 1968;Caine-Bish et al., 2005;Costa et al., 2010), is recognised to increase plasma catecholamine due to sympathetic activity stimulation (Castellani et al., 2002).However, TSD may not influence catecholamine reactivity according to reviewed studies.Although sAA proved to be a potential indicator of ANS dysregulation (Klaus et al., 2019), Minkel et al. (2014) and Schwarz et al. (2018) did not find a significant difference after the TSST to sleep-deprived subjects because to observe a modulation of sAA, a massive and/or chronic sleep impairment (several weeks) would be required (Thieux et al., 2024).Moreover, very limited evidence with divergent results to EDA (Liu et al., 2015;Bottenheft et al., 2023) and no differences to FVC, FBF (Yang et al., 2012) and MSNA (Kato et al., 2000;Yang et al., 2012) were reported.

Cortisol responses to an acute stressor after TSD
Vargas and Lopez-Duran (2017) demonstrated that TSD blunted cortisol reactivity using a short period of cortisol sampling time (≅ 6 min).Schwarz et al. (2018) utilising a longer sampling duration (≅ 17 min) and comparable cortisol analysis techniques (intercept; slope), found no differences in the cortisol response pattern between groups.These results suggest that a shortened period of the cortisol sample would be the best way to detect cortisol reactivity and recovery from stress (Zhao et al., 2021).Minkel et al. (2014) and Hansen et al. (2021) used a similar cortisol sample time to Schwarz et al. (2018) but a different statistical approach (repeated measures analysis of variance with sleep as the between-subject factor and cortisol sampling time as a within subject factor) and the results were contradictory.While Minkel et al. (2014) found that TSD increased (large effect size) cortisol response to stress, Hansen et al. (2021) reported that TSD blunted (large effect size) cortisol responses to stress suggesting that TSD has a bidirectional effect on cortisol response to stress.Caine-Bish et al. (2005) and Costa et al. (2010) found no differences in cortisol responses to acute stressors using cold exposure as stress and blood cortisol measurements.It is possible that TSD and cold exposure combined neutralised each other's unique reactions (Caine-Bish et al., 2005) and, also, there is additional stress involved in collecting blood samples from the vein, which could affect the outcome (Bozovic et al., 2013).Bottenheft et al. (2023) suggested that methodological elements such as the use of a single jury member during the TSST, acclimating to the laboratory environment, and, considering the other kind of social stress test conducted on the first day, specifically acclimating to a social stress test may have affected the outcomes.The cortisol response to acute stress is dependent on a variety of factors such as the type of stress induction, time of stress induction, time of cortisol measurement and cortisol analysis method (Zhao et al., 2021).4.6.Time of day that stress protocol was applied, outcomes were measured, and other possible confounders Throughout the 24-hour day, cortisol is released with the highest levels occurring in the early hours following regular wake time and levels drop throughout the day, reaching a nadir in the late evening (Agorastos and Chrousos, 2022).Similarly to cortisol, markers of ANS show strong circadian alterations with a peak of sympathetic activity and a nadir of parasympathetic activity in the morning hours (Agorastos and Chrousos, 2022).Thus, acute physical and psychological stress both cause the greatest stress response in the afternoon, when the ANS and HPA axis are less active.In the present study, acute stress exposure was observed to significantly alter baseline levels of ANS markers or cortisol response in almost all of the studies (n = 14).Furthermore, we cannot assume that there was a clear effect of the time of day the stress was applied, or the outcome was measured in the interaction between TSD and acute stress responses as well based on our results.
We extracted other possible confounder effects from the included studies and curiously the study that controlled most (Schwarz et al., 2018) and the study that controlled less (Åkerstedt and Fröberg, 1979) of the confounder factors found no effects of TSD on stress response.Information about confounder effects and appropriate methods to control these effects was only considered in the risk of bias assessment in non-randomised studies because RoB 2 considers that the randomisation process removes almost all of them.However, some confounder effects can not be controlled by the randomisation process, such as the position of measurements, and the time of day that the stress was applied and the outcome was measured, and this could be a potential limitation in using the Cochrane tool for RCTs studies.

The interplay between ANS and HPA axis
There is some reciprocity between the HPA axis and ANS activity at various neuroendocrine levels (Agorastos and Chrousos, 2022).The ANS has a role in the physiological functions of the HPA axis, while cortisol signalling may also play a significant role in controlling ANS activity (Thayer and Sternberg, 2006;Pulopulos et al., 2018;Agorastos and Chrousos, 2022).Agorastos et al. (2019) assessed the interaction between the ANS and HPA axis measuring simultaneously HPA axis and ANS activity in healthy subjects (Agorastos et al., 2019).The authors found that HPA axis stimulation is associated with reduced HRV, suggesting that the parasympathetic nervous system plays a role in regulating both the cardiovascular reactivity to stress and the vulnerability to stress-related illnesses (Agorastos et al., 2019).Although five studies investigated markers of ANS and cortisol outcomes (Caine-Bish et al., 2005;Costa et al., 2010;Minkel et al., 2014;Schwarz et al., 2018;Bottenheft et al., 2023), no studies investigated how TSD may influence the interactions between ANS and the HPA axis or if there were any associations between any ANS marker and cortisol responses to acute stress.

Bidirectional multi-system reactivity
The findings of our study support the theory of "bidirectional multisystem reactivity" (Turner et al., 2020) by demonstrating that TSD generates exaggerated or blunted ANS indicators or cortisol responses to laboratory stresses suggesting that future health and illness outcomes are correlated with atypical stress reaction.For example, Agorastos et al. (2020) found in people with major depressive disorder an atypical responsivity (i.e., exaggerated or blunted) of ANS which was influenced according to their depressive history status (i.e., first vs.recurrent episode) (Agorastos et al., 2020b).According to Martí et al. (1994), similar effects have been demonstrated with the HPA axis activity in an animal model of sustained stress exposure (Martí et al., 1994).In this context, we believe that the frequency of previous acute episodes of TSD that the volunteers have been exposed to as well as their sleep patterns and habits may influence the magnitude (exaggerated or blunted) of the stress responses.

Strengths and limitations
This is the first systematic review to summarise the ANS and cortisol responses to acute stressors in individuals in a total sleep-deprived condition.We included various stressor types (physical, psychological and combination) that may have previously been overlooked.In addition, only studies using TSD protocols were considered in this review, whereas other studies have (maybe inappropriately) combined different types of sleep deprivation protocols (total and partial sleep deprivation).Finally, our study is the first systematic review that summarised the effects of TSD on ANS and cortisol reactivity as well as the first review that explored how sleep loss may influence ANS responses to acute stress.
This systematic review has several limitations.We only considered studies on healthy populations, however clinical populations (e.g., chronic pain) or individuals with sleep disorders or chronic stress, may respond differently to TSD.Although cortisol is the main peripheral effector of the HPA axis and the most widely used HPA axis measure in the literature (Chrousos, 2009;Spencer and Deak, 2017), other HPA axis biomarkers (i.e.DHEA, ACTH, CRF, copeptin) were not included in this review.Also, the lack of possible meta-analyses and the quality of the studies found in this systematic review could be considered a limitation.Finally, only studies in English were included in this systematic review which may have meant we missed relevant studies published in other languages.

Clinical implications and future directions
Although chronic sleep restriction is highly prevalent in our society, studies investigating the effects of TSD are important as several professions such as firefighters, airline pilots, healthcare workers, among others, experience TSD and stress in their routine.Considering that TSD seems inevitable, especially in the professions mentioned above, people should be aware that TSD may modify the nature of physiological responses to challenges encountered in everyday life, changing your ability to deal with stressful situations and influencing health outcomes.
Future studies should investigate the interaction between sleep parameters and markers of ANS and cortisol stress responses, as well as the impact of non-pharmacological therapy such as physical exercise in mitigating the effects of sleep deprivation on ANS and cortisol responses to an acute stressor.Also, future reviews should focus on other HPA axis biomarkers as they can be associated with the physiological and pathophysiological mechanisms involving sleep and stress responses.Studies should consider the impact of TSD in ANS and cortisol stress responses in clinical populations (e.g., chronic pain) or individuals with sleep disorders or chronic stress.New RCT studies should be conducted and report adequate concealment of participant allocation, any deviation from the intended intervention, adequate wash-out time (crossover studies) and blind investigators and statisticians to whether subjects had undergone a night of sleep or sleep deprivation.Finally, studies measuring ANS markers and cortisol simultaneously and exploring the associations between ANS and cortisol responses to acute stressors could reveal some new insights into how TSD may influence the interactions between ANS and the HPA axis and their response to a stressor event.

Conclusion
In conclusion, the literature review on markers of ANS and cortisol responses to acute stress after TSD in healthy individuals reveals a scarcity of consistent evidence.However, there is enough evidence suggesting that TSD induces either, blunted or increased ANS or cortisol responses to laboratory stresses supporting the "bidirectional multisystem reactivity hypothesis."It is important to note that a comprehensive understanding of this phenomenon still lacks robust evidence, and further research is needed to clarify these relationships.

Fig. 2 .
Fig. 2. Summary of risk of bias with RoB 2 tool -a) randomised controlled trials b) crossover studies.

Fig. 3 .
Fig. 3. Summary of risk of bias with ROBINS-I tool.

Table 1
Characteristics of the included studies.

Table 2
Summary of protocols.

Table 3
ANS and cortisol differences between TSD and sleep control (SC) conditions after acute stress and during recovery..3.2.1.TSD increases ANS and cortisol stress responses.Six out of 16 studies found that TSD increases, at least, one marker of ANS or cortisol response to stress.