The impact of hypoxia exposure on glucose homeostasis in metabolically compromised humans: A systematic review

Humans living at a higher altitude are less prone to suffer from impaired glucose homeostasis and type 2 diabetes mellitus (T2DM), which might at least partly be explained by lower oxygen availability at higher altitudes. The present systematic review aimed to provide an overview of the current literature on the effects of hypoxia exposure on glucose homeostasis in metabolically compromised humans. Several databases were searched up to August 10th, 2020. The search strategy identified 368 unique records. Following assessment for eligibility based on the selection criteria, 16 studies were included in this review. Six studies (2 controlled studies; 4 uncontrolled studies) demonstrated beneficial effects of hypoxia exposure on glucose homeostasis, while 10 studies (8 controlled studies; 2 uncontrolled studies) reported no improvement in glucose homeostasis following hypoxia exposure. Notably, passive hypoxia exposure seemed to improve glucose homeostasis, whereas hypoxic exercise training (2–8 weeks) appeared to have no additional/synergistic effects on glucose homeostasis compared to normoxia exposure. Due to the heterogeneity in study populations and intervention duration (acute studies / 2–8 wks training), it is difficult to indicate which factors may explain conflicting study outcomes. Moreover, these results should be interpreted with some caution, as several studies did not include a control group. Taken together, hypoxia exposure under resting and exercise conditions might provide a novel therapeutic strategy to improve glucose homeostasis in metabolically compromised individuals, but more randomized controlled trials are warranted before strong conclusions on the effects of hypoxia exposure on glucose homeostasis can be drawn.


Introduction
The current obesity epidemic is accompanied by a proportionate increase in the prevalence of several chronic diseases such as cardiovascular diseases and type 2 diabetes mellitus (T2DM) [1]. With the increasing worldwide prevalence of obesity, interventions targeting obesity and its cardiometabolic complications are increasingly warranted. Many of the current interventions rely on lifestyle modifications or pharmacological treatment. Indeed, the majority of studies have demonstrated that adopting a healthy lifestyle can prevent or delay obesity-related complications, including impaired glucose homeostasis as present in patients with T2DM [2][3][4]. Importantly, however, dietary changes and regular exercise have often proven hard to maintain for individuals [5]. Additionally, pharmacological therapies often come with side effects [6], and large inter-individual differences in the response to these interventions are often apparent [7]. Therefore, alternative or complementary strategies are needed to mitigate the development of metabolic impairments and related chronic diseases, especially in people at high risk of developing metabolic complications.
The Nobel Prize in Physiology or Medicine was awarded to William Kaelin, Jr., Sir Peter Ratcliffe, and Gregg Semenza in 2019 for their discoveries of how cells sense and adapt to oxygen availability. Intriguingly, humans living at a higher altitude are less prone to suffer from metabolic impairments, independent of body mass index (BMI) [8]. It is tempting to postulate that the lower oxygen availability at higher altitudes may at least partly contribute to the lower prevalence of impaired glucose homeostasis (i.e. T2DM) in highlanders compared to sea level residents. Notably, many confounding factors such as dietary habits and the level of physical activity may also play a role in the beneficial effects of residing at high altitude on glucose homeostasis.
It is well established that adipose tissue (AT) dysfunction and impairments in skeletal muscle (SM) metabolism are major contributors to the development of impaired glucose homeostasis and cardiometabolic complications in humans, as extensively reviewed elsewhere [9]. Interestingly, accumulating evidence indicates that oxygen partial pressure (pO 2 ) in AT and SM may affect cellular and organ functioning. Indeed, several in vitro experiments using adipocytes and myotubes have demonstrated that pO 2 is causally related to alterations in glucose and lipid metabolism, as recently reviewed [10]. Under hypoxic circumstances, the body is inclined to switch from fat oxidation towards glucose utilization [10]. In line, acute hypoxia exposure (1% versus 21% O 2 ) has been found to increase basal glucose uptake in both human and rodent adipocytes [11][12][13]. Additionally, we have previously demonstrated that prolonged hypoxia exposure (14 days) to physiological hypoxia (5% O 2 ) increased basal glucose uptake in differentiated human multipotent adipose-derived stem cells [14]. Moreover, many studies have shown that pO 2 alters the expression and secretion of pro-inflammatory cytokines in rodents and human adipocytes [10]. Interestingly, we have previously reported that AT pO 2 was higher in obese compared to lean individuals [15], and was inversely associated with whole-body insulin sensitivity, independently of adiposity, in men and women [16]. In accordance with these findings, diet-induced weight loss in overweight/obese humans decreased in situ AT pO 2 , with parallel improvements in insulin sensitivity [17]. Finally, several studies have shown that hypoxia exposure increases glucose uptake in rodent and human myocytes, as reviewed [10]. Taken together, previous results from both in vitro and in vivo studies in rodents and humans indicate that tissue pO 2 is related to glucose homeostasis and/or insulin sensitivity. Therefore, interventions targeting tissue pO 2 may provide a promising therapeutic avenue to improve glucose homeostasis in humans.
The aim of the present systematic review was to provide an overview of the current literature on the effects of hypoxia exposure on glucose homeostasis in metabolically compromised humans.

Search strategy
Studies that investigated the effects of hypoxia exposure on glycemic control and/or insulin sensitivity were retrieved from the PubMed database and Web of Science.
Additionally, the following search string was entered into the Web of Science database: TITLE: (("hypox*" OR "hyperox*" OR "low oxygen" OR "oxygen deficiency" OR "decreased oxygen")) AND ALL FIELDS: (("glyc* control" OR "insulin sensitivity" OR "Blood Glucose" [Mesh] OR "Glycemic Index"[Mesh] OR "glucose levels" OR "glycated hemoglobin" OR "glyc* variability" OR "hypoglyc*" OR "hyperglyc*" OR "HOMA index") AND ("Metabolic Syndrome"[Mesh] OR "Insulin Resistance"[Mesh] OR "prediabetes" OR "impaired glucose tolerance" OR "impaired fasting glucose" OR "prediabetic" OR "obese" OR "obesity" OR "overweight")) NOT ALL FIELDS: (("cancer AND apnea AND osas)). A filter was activated so that only original articles were included. Searches were not limited by article publication date. The literature search was performed on the 10 th of August 2020. From the identified articles, the titles and abstracts were assessed and, if considered relevant for the present systematic review, the full text of the article was examined in detail.

Selection criteria
Eligible studies included humans with impaired glucose homeostasis. Thus, patients with overweight/obesity, the metabolic syndrome and/or patients with T2DM were included. Only studies with a clear description of the intervention (and control condition, if applicable), as well as parameters related to glucose metabolism as study outcomes were included. Only original articles written in English were included.

Literature search
A flowchart to illustrate the inclusion process is shown in Fig. 1. The literature search identified 368 unique records. Titles and/or abstracts were screened, which resulted in the exclusion of 315 studies. 6 additional records were identified through searching reference lists. The remaining 53 full-text articles were retrieved and assessed for eligibility based on the selection criteria. From these full-text articles, a total of 37 articles were excluded. Reasons for exclusion were related to the study population, study design and outcome parameters. Thus, a total of 16 studies were included in the present systematic review. From these studies, 10 studies were randomized controlled trials (RCTs), whereas the other 6 studies did not include a control group.

The effects of hypoxia exposure on glucose homeostasis
As mentioned earlier, previous in vitro and in vivo studies in rodents have indicated that hypoxia exposure impacts glucose homeostasis, as reviewed recently [10]. Based on these studies, it could be argued that hypoxia exposure may provide a potential therapeutic strategy to combat disturbances in glucose metabolism. However, only a limited number of in vivo studies have investigated the effects of

The effects of hypoxia exposure on circulating glucose and insulin concentrations
Several but not all studies have shown beneficial effects of hypoxia exposure on glucose homeostasis (Table 1). Duennwald et al. [18] have found that merely one hour of passive intermittent hypoxia (F i O 2 13%) exposure significantly decreased fasting plasma glucose levels in individuals with overweight, while normoxia exposure did not significantly change glucose concentrations. Notably, the difference between the effects of hypoxia compared with normoxia exposure was statistically significant [18]. Additionally, Serebrovska et al. [19] found that three weeks of moderate passive intermittent hypoxia exposure (F i O 2 12%, 40 min/day, 3x/week) improved fasting and postprandial glucose concentrations in healthy individuals as well in as people with impaired glucose homeostasis. In line, Lecoultre et al. [20] reported significantly lower fasting glucose levels in obese individuals after ten consecutive nights of passive moderate hypoxia exposure (~ 10 h/night at F i O 2 ~ 15% O 2 ). Importantly, however, the latter two studies did not include a control group [19,20]. Furthermore, Mackenzie et al. [21] have shown that plasma insulin levels during a 4-h intravenous glucose tolerance test were significantly decreased after one 60-min bout of hypoxic exercise (F i O 2 14.6%) at 90% of the lactate threshold compared to normoxic exercise at the same relative intensity in T2DM patients. In a follow-up study, Mackenzie et al. [22] demonstrated that exercise for 60 min at 90% of the lactate threshold in hypoxic conditions (F i O 2 ~ 15% O 2 ) decreased fasting glucose and insulin levels in T2DM patients, while these effects were not present following only 20 min of hypoxic exercise. Notably, it is not clear whether these beneficial effects were due to hypoxia exposure or stem purely from the exercise bout, as a control group was not included in this study [22]. In addition, Marlatt et al. [23] found a significant reduction in 2-h glucose levels during an OGTT in metabolically compromised individuals following 14 nights of exposure to moderate hypoxia (sleeping at home in a hypoxic tent maintained at 15% O 2 ), although fasting glucose levels remained unchanged. Also in the latter study, no control group was included [23]. Finally, De Groote et al. [24] demonstrated that six weeks of hypoxic training (F i O 2 15%, 3 × 1 h/wk) but not normoxic training significantly decreased plasma insulin levels during an oral glucose tolerance test and improved glucose tolerance in obese adolescents. Yet, the differences between the effects of hypoxic and normoxic training on postprandial glucose and insulin concentrations did not reach statistical significance in the latter study [24].
Taken together, these studies suggest that passive hypoxia exposure (i.e. hypoxia exposure under resting conditions) [18][19][20] as well as hypoxic exposure during an acute bout of exercise [18,22] or exercise training program [24] may improve glucose homeostasis in metabolically compromised individuals.
In contrast, several other studies that investigated the effects of hypoxic exercise on glucose homeostasis did not find significant changes in parameters related to glucose homeostasis (Table 1). Wiesner et al. [25] have shown that a four-week hypoxic training program (F i O 2 ~ 13% O 2 , 60 min/ day, 3 days/week) at a heart rate corresponding to 65% of maximum oxygen consumption significantly decreased fasting insulin levels compared to baseline in metabolically compromised individuals. Importantly, however, an identical training program under normoxic circumstances yielded similar effects, suggesting that these improvements are likely due to exercise per se rather than hypoxia exposure. Lippl et al. [26] found that seven days of hypobaric hypoxia exposure (altitude of 2,650 m) did not alter metabolic parameters, including fasting glucose and insulin levels, compared with normobaric normoxia exposure (altitude of 530 m) in metabolically compromised individuals. Somewhat surprising, however, a significant decrease in HbA1c levels following one week of hypoxic exposure was found in the latter study, which was not yet seen after seven days of normoxic exposure [26]. Similarly, Morishima et al. [27,28] demonstrated that hypoxia exposure did not significantly impact fasting and postprandial glucose concentrations in overweight men, despite a pronounced increase in postprandial carbohydrate oxidation under hypoxic resting and exercise conditions compared to normoxia exposure [28]. Neither hypoxia exposure (F i O 2 15%) under resting conditions [28], nor hypoxic exercise (12 × 1 h training sessions over 2-4 wks) [27] altered fasting blood glucose and serum insulin concentrations. Furthermore, Gutwenger et al. [29] have shown that fasting plasma glucose and insulin levels were not significantly affected in response to a two-week hiking vacation (3 h/day, 4 days/week) at high (1,900 m) versus low (300 m) altitude in patients with the metabolic syndrome. Likewise, Chacaroun et al. [30] were not able to show any improvements in plasma insulin levels and HOMA2-IR after an eight-week hypoxic exercise intervention with target SpO 2 of 80% during the exercise sessions in overweight and obese individuals. Finally, Klug et al. [31] found that a six-week hypoxic (F i O 2 15%) exercise intervention in men with the metabolic syndrome did not significantly change blood glucose and insulin levels after an oral glucose load. In line, no differences were found in glucose tolerance and HbA1c between normobaric hypoxic and normobaric normoxic training conditions [31].
To summarize, 6 studies (2 controlled studies; 4 uncontrolled studies) demonstrated beneficial effects of hypoxia exposure on glucose homeostasis, while 10 studies (8 controlled studies; 2 uncontrolled studies) reported no improvements in glucose homeostasis following hypoxia exposure. Notably, passive hypoxia exposure seemed to improve glucose homeostasis, while hypoxic exercise training (2-8 weeks) appeared to have no additional/synergistic effects on glucose homeostasis compared to normoxia exposure.

The effects of hypoxia exposure on insulin sensitivity
Although conflicting findings on the effects of hypoxia exposure on circulating glucose and insulin concentrations have been reported (Table 1), results of studies that have examined the effects of hypoxia exposure using more sophisticated measures of insulin sensitivity and glucose tolerance such as the hyperinsulinemic-euglycemic clamp and the intravenous glucose tolerance test, respectively, suggest that hypoxia may improve insulin sensitivity.
Lecoultre et al. [20] showed that ten consecutive nights of moderate hypoxia (F i O 2 15%) exposure improved whole-body insulin sensitivity, assessed with the goldstandard hyperinsulinemic-euglycemic clamp. Strikingly, the improvements in whole-body insulin sensitivity were more pronounced in individuals with the lowest baseline insulin sensitivity, underlining the potential of hypoxia exposure as a therapy for people with (severe) insulin resistance. Yet, the latter study did not include a control group [20]. Furthermore, Mackenzie et al. [21] demonstrated improvements in glucose tolerance, determined using an intravenous glucose tolerance test, in response to acute hypoxia (F i O 2 15%) compared to normoxia exposure in patients with T2DM. Conversely, a study performed by Chobanyan-Jürgens et al. [32] did not show any significant additive beneficial effects of exercise under hypoxic circumstances (F i O 2 15%, 3 × 30-40 min/week for 8 weeks) compared to normoxic circumstances on insulin sensitivity, assessed by a hyperinsulinemic-euglycemic clamp, in older sedentary overweight individuals.
In addition, studies that used surrogate markers of insulin sensitivity found improved [21,22,24,25,32,33] or unchanged [27][28][29][30][31] insulin sensitivity following a hypoxia exercise program. Importantly, several of the studies that did find improved insulin sensitivity following hypoxia exposure did not include a control group or the improvement in the active arm (hypoxia exposure) was not different from the control condition (normoxia exposure) [24,25,32].
Thus, both acute and more prolonged exposure to hypoxia with or without the addition of exercise may have beneficial effects on insulin sensitivity in metabolically compromised individuals, but conflicting findings have been reported.

Discussion
The findings of the studies included in the present systematic review suggest that hypoxia exposure has beneficial or neutral effects on glucose homeostasis in metabolically compromised humans. More specific, passive hypoxia exposure seemed to improve glucose homeostasis, whereas hypoxic exercise training (2-8 weeks) appeared to have no additional/synergistic effects on glucose homeostasis compared to normoxia exposure. Due to the heterogeneity in study populations in these studies with respect to age, sex, metabolic status (i.e. no T2DM/T2DM) and ethnicity, as well as differences in severity and duration (acute studies / 2-8 wks training) of hypoxia exposure, it is difficult to indicate which factors may explain the different study outcomes. Importantly, as several studies that reported beneficial effects on glucose homeostasis did not include a control group, these results should be interpreted with caution. Clearly, more well-controlled studies are needed to be able to disentangle which parameters related to study population, hypoxia exposure protocol and/or intervention duration determine intervention outcomes.
Human studies have demonstrated a great variety in responses to hypoxia exposure on glucose homeostasis. One could argue that exercise under mild hypoxic conditions might induce more pronounced adaptations in glucose homeostasis as compared to hypoxia exposure under resting conditions due to a stronger hypoxic stimulus under exercise conditions [34]. Even more importantly, the adaptations underlying the effects of hypoxia may vary based on the severity and duration of hypoxia exposure (duration of each hypoxic episode and total exposure duration) as well as the pattern (intermittent/continuous) of exposure [10]. Indeed, it has been suggested that mild hypoxia exposure with low cycle numbers (3-15 cycles/day) may improve several metabolic parameters, whereas severe hypoxia, with highfrequent episodes may result in pathological adaptations [35]. For example, severe (intermittent) hypoxia exposure may increase sympathetic nervous system activity, blood pressure, inflammation, cholesterol levels, the risk of atherosclerosis and right ventricular hypertrophy, and impair cognitive function, as extensively reviewed [35]. Since severe hypoxia exposure may induce Acute Mountain Sickness symptoms and other pathogenic effects, it is important to closely monitor adverse events when conducting hypoxia exposure intervention studies. Importantly, none of the studies included in the present systematic review reported more Adverse Events related to the intervention during hypoxia exposure compared to normoxia exposure. The results of the studies included in the present systematic review seem conflicting and are difficult to compare due to the heterogeneity of study designs. In addition, several studies did not include a control group, which clearly hampers robust conclusions on the effects of hypoxia exposure on glucose homeostasis and insulin sensitivity.
At the cellular level, findings on the effects of hypoxia exposure on glucose homeostasis are less conflicting. A potential mechanism by which hypoxia causes a decrease in blood glucose levels is by inducing a switch from aerobic to anaerobic metabolism, mediated by the hypoxiainducible factor (HIF)-1 system [10,36]. Indeed, both hypoxia exposure and contraction (in vitro electrical pulse stimulation) have been demonstrated to improve insulin action and glucose metabolism in myotubes via activations of the HIF-1α pathway [37]. In addition, hypoxia and exercise induced an increase in the AMP/ATP ratio in soleus muscle of lean rats, thereby increasing insulin-independent glucose uptake through AMP-activated protein kinase activity [38]. Moreover, hypoxia increased the intracellular free Ca 2+ levels and subsequently calmodulin-dependent protein kinase, ultimately enhancing insulin-independent glucose uptake [39]. Interestingly, synergistic effects of hypoxia exposure and contraction have been demonstrated regarding stimulation of glucose transport in rat hindlimb muscle [40]. Furthermore, many experiments have demonstrated that changes in oxygen levels impact the functionality of (pre)adipocytes and immune cells, leading to alterations in adipose tissue inflammation, lipid and glucose metabolism, as extensively reviewed elsewhere [10].
In conclusion, the results of the studies included in the present systematic review suggest that hypoxia exposure, either under resting conditions or during exercise, might provide a novel, non-pharmacological therapeutic strategy to improve glucose homeostasis in metabolically compromised individuals. More specific, passive hypoxia exposure seemed to improve glucose homeostasis, whereas hypoxic exercise training (2-8 weeks) appeared to have no additional/synergistic effects on glucose homeostasis compared to normoxia exposure. Importantly, however, more well-controlled RCTs with detailed metabolic phenotyping (i.e. measurement of tissue-specific insulin sensitivity) are warranted before robust conclusions on the effects of hypoxia exposure on insulin sensitivity and glucose homeostasis can be drawn. In addition, it is important to investigate whether the metabolic effects of hypoxia exposure are age-specific or sex-specific, and depend on the severity, mode (passive hypoxia exposure or hypoxia exposure during exercise) and duration of hypoxia exposure, as well as medication use. Finally, a better understanding of the mechanisms underlying the putative effects of hypoxia on glucose metabolism is needed, since this will contribute to the development and optimization of strategies to prevent and treat impairments in glucose homeostasis and related chronic diseases.