Depression is a major public health concern and a global priority for research and development [1,2]. Unipolar depressive disorder is predicted to rise from the third leading cause of global burden of disease in 2004 to number one in 2030 [3]. Depression is a prevalent condition with a 12-month prevalence rate of more than 5% in most high-, middle- and low-income countries [4]. It generates substantial loss of quality of life, morbidity, despair and loss of productivity. Depression has been found to increase the risk of all-cause mortality [5].

Alongside the issue of depression is the global concern regarding physical inactivity and metabolic syndrome [6,7]. Physical inactivity is now the fourth leading cause of death worldwide and is thought to have increased in prevalence worldwide due to reducing rates of incidental, transportation-related and occupational PA [6,8]. Physical inactivity is a major causative factor in a large number of major diseases (i.e. coronary heart disease, type 2 diabetes mellitus, breast and colon cancer) and premature mortality [6]. Interestingly, recent data has emerged to suggest that physical inactivity and depression are closely inter-related [9]. Physical inactivity and lower fitness levels may increase the risk of depression and depression may increase the risk of physical inactivity [10-20].

The role of Physical Activity (PA) in the treatment and prevention of psychiatric disorders (e.g. anxiety disorders, bipolar disorder and age-related cognitive dysfunction) has been developing in recent times with a significant number of quality trials emerging [21-25]. Indeed, PA is seen as an efficacious treatment strategy in depression, as well as a preventive strategy, however other studies show no effect [21,26]. PA is defined as “any bodily movement produced by skeletal muscles that requires energy expenditure” while physical exercise is “a subcategory of PA that is planned, structured, repetitive, and purposeful in the sense that the improvement or maintenance of one or more components of physical fitness is the objective” [27]. PA subtypes include aerobic, resistance, flexibility, neuromotor (e.g. involving balance, agility and co-ordination), mind-body (e.g. tai chi, qi gong and yoga; combining neuromotor, flexibility and meditation), exer-gaming (entertaining video games that combine game play with exercise) and mixed [28-35].

The clinical utility of PA appears to be manifold. PA may be used as a stand-alone treatment, as well as an adjunct for mild to moderate depression [31,36,37]. PA may also has a role in the treatment and prevention of other significant cardio-metabolic disorders, often cooccurring with depression, (i.e. cardiovascular disease, obesity and diabetes mellitus) [38-42]. In the treatment of late-life depression, PA is shown to prevent falls in the elderly via enhancing strength, flexibility and balance [43]. Strategies to increase PA levels can occur in a variety of settings, from the traditional clinical setting, to broader public health initiatives such as large-scale informational dissemination, environmental re-design and policy approaches [44]. From a mechanistic perspective, PA may exert it’s effects via a number of positive effects on neurobiological systems,(i.e. increasing production of serotonin (5-HT), Noradrenaline (NA) and Dopamine (DA), increasing Hippocampal (HC) neuroplasticity, reducing neuroinflammation and reduced microglial activation) [21]. Understanding the variety of strategies to increase PA levels, as well as the underlying neurobiological effects of PA is important to progress this field.

This paper aims to provide an up-to-date, critical assessment of the epidemiological, clinical trial and clinical immunological evidence for PA interventions in the treatment and prevention of depression. A review integrating epidemiological, clinical and neuroscience evidence for PA in the treatment and prevention of depression is novel in the literature.

This paper is a narrative review. It has utilized search engines Google Scholar, PubMed and ScienceDirect. Keywords in the searches included various combinations of the following key words (i.e. physical activity, exercise, clinical, public health, neuroscience, population, epidemiology, and depression).

Clinical Implications of Prevention Science in Depression

Prior to investigating the role of PA in the prevention of depression, it is important to first frame the clinical implications of preventive science-based approaches in depression. It is important to identify the types of preventive approaches, the various populations under investigation and the end point of preventive approaches.

When conceptualizing approaches in prevention science, the most commonly used models are those of the Institute of Medicine and the World Health Organization’s (WHO) framework of levels of prevention (i.e. primary, secondary and tertiary prevention) [45,46]. A report from the Institute of Medicine suggests prevention may be directed toward the whole population (universal prevention), high-risk groups (selective prevention), or those with subsyndromal symptoms (indicated prevention) [45]. The WHO’s prevention framework suggests primary prevention involves strategies aimed at preventing the development of disease; secondary prevention involves strategies to diagnose and treat existent disease in early stages before significant morbidity occurs; and, tertiary prevention involves strategies to reduce negative impact of existent disease by restoring function and reducing disease-related complications [46]. Table 1 outlines these conceptual frameworks of preventive science, as well as clinical examples for the field of PA in the treatment and prevention of depression.

Evidence for a Role of Physical Activity in the Treatment and Prevention of Depression

There are a number of levels of evidence to be considered when investigating the role of physical activity in the prevention of depression. This level of evidence can be conceptualized as public health, clinical and basic science evidence. In Table 2 we outline these levels of evidence. In Table 3 we outline the salient studies in these various fields.

Table 1: Clinical implications of preventive science frameworks in depression.

Table 2: Approaches to examining the role of physical activity in the treatmentand prevention of depression.

Table 3: Selected studies examining the role of physical activity in the prevention and treatment of depression.

A number of prospective epidemiological studies have examined the relationship between levels of PA and subsequent depressive symptom development. Physical inactivity and lower fitness levels may increase the risk of depression, however other studies have found no association [10-18,47]. Interestingly, depression may increase the risk of physical inactivity [10,19,20]. An analysis of the 9,309 adult participant, Whitehall II prospective cohort study found the association between PA and depression appears to be bidirectional [10]. This analysis was conducted across 8 years, over 3 time points and utilized a selfadministered questionnaire to determine PA levels. Participants were 35 to 55 years at baseline, in 1985. A cross-sectional inverse association between PA and depressive symptoms was found at baseline (OR 0.72). In cumulative analyses, regular PA across all three data waves, but not irregular PA, was associated with reduced likelihood of depressive symptoms at follow up (OR 0.71). In a converse analysis, participants with anxiety and depressive symptoms at baseline had higher odds of not meeting the recommended levels of PA at follow up (OR 1.79). A study by Dishman et al. examined longitudinal change in Cardiorespiratory Fitness (CRF) and incident depression complaints made to a physician [12]. This study assessed CRF at four clinic visits between 1971 and 2006 in 7936 men and 1261 women, aged 20-85 years. Subjects did not have depression at baseline. During subsequent visits there were 446 incident cases of depression in men and 153 cases in women. It was found that each 1-minute decline in treadmill endurance (i.e. decline in CRF of 1 half-MET) between ages 51 and 55 years in men and 53 and 56 years in women increased the odds of incident depression by approximately 2% and 9.5%, respectively. Therefore, maintenance of CRF during late middle age, when decline in CRF usually accelerates, helps protect against the onset of depression. Another study by Strawbridge et al. has compared the effects of higher levels of PA on prevalent and incident depression [15]. This study examined 1947 community-dwelling adults from the Alameda County Study aged 50-94 years at baseline in 1994 with 5 years of follow-up. Greater PA was protective for both prevalent depression (OR 0.90) and incident depression (OR 0.83) over 5 years. In a study of younger adults, Suija et al. examined a study population consisting of 5497 males and females of 31 years of age [18]. CRF was measured by 4-min step test and muscle fitness by tests of maximal isometric handgrip and trunk extension. They found lower levels of isometric endurance capacity of trunk extensor muscles were associated with high levels of depressive symptoms in both sexes. In males poor handgrip strength was associated with increased levels of depressive symptoms. There was no association between CRF and depressive symptoms.

A number of primary research, meta-analyses and systematic reviews have examined the efficacy of various subtypes of PA in depression [26,34,29-32,48-54]. Two recent Cochrane Database metaanalyses examining PA effects in depression treatment suggests mixed and resistance PA are more effective than aerobic exercise [26,55]. The most recent meta-analysis by Cooney et al. examined 3 studies with mixed PA, 28 with aerobic PA and 4 with resistance type, with 1352 participants in total [55]. The standardized mean differences (SMD) for aerobic PA indicated a moderate clinical effect (SMD -0.55, 95% CI -0.77 to -0.34), whilst the SMDs for both mixed PA(SMD -0.85, 95% CI -1.85 to 0.15) and resistance PA(SMD -1.03, 95% CI -1.52 to -0.53) indicated large effect sizes, but confidence intervals were wide. When all 35 studies (1356 participants) comparing exercise with no treatment or a control intervention are examined, the pooled SMD at the end of treatment was -0.62 (95% CI -0.81 to -0.42), indicating a moderate clinical effect. There was moderate heterogeneity (I²=63%). Importantly, however, when only 6 methodologically robust trials with 464 participants are included (i.e. with adequate allocation concealment, intention-to-treat analysis and blinded outcome assessment) the pooled SMD was not significant (-0.18, 95% CI -0.47 to 0.11). An RCT by Blumenthal et al. compared the efficacy of aerobic exercise, an antidepressant (sertraline) or combination for 16 weeks [56]. This study involved older subjects with MDD and found no significant different in efficacy between PA and sertraline. Sertraline produced the fastest initial response. A 6-month follow up analysis of this study by Babyak et al. found the exercise group had lower relapse rates as compared to the antidepressant group [57]. Other randomized-controlled trials suggest there is no difference between aerobic vs. stretching, aerobic vs. strength vs. relaxation training on depressive symptoms, however these studies are limited by diagnostic heterogeneity and variability in PA program design [52,53,58]. A meta-analysis of 4 trials and 253 participants examined the effectiveness of tai chi in reducing depressive symptoms among older adults. All four studies compared tai chi to waiting list control group. The pooled SMD was -0.27 (95% CI -0.52 to -0.02, P=0.03). Further research was recommended by the study author with larger sample sizes, more clarity on trial design, intervention and longer-term follow up. A recent RCT found tai chi was effective in augmenting escitalopram in the treatment of geriatric depression [31]. A 12-week pilot study has examined the short-term efficacy of exer-games (entertaining video games that combine game play with exercise) for sub-syndromal depression in 19 older adults [35]. The subjects utilized exer-gaming for three 35-minute sessions a week and a significant improvement in depressive symptoms was found (measured by the Quick Inventory of Depressive Symptoms – Clinician Rated Version). The abovementioned evidence suggests there are some antidepressive effects for various subtypes of PA, however, methodological heterogeneity makes these studies difficult to compare and there are no studies comparing aerobic, resistance and mind-body PA in a randomized, controlled setting.

Rethorst and Trivedi have recently released guidelines for the treatment of MDD, based on a review of available literature [54]. These guidelines suggest for patients with MDD, aerobic or resistance training should be considered, 3-4 exercise sessions/week, for 45 - 60 minutes, for at least 10 weeks and with an intensity of 50-60% of maximal heart rate. Of course, these guidelines should be modified according to individual patient-related factors.

Clinical neuroscience evidence in this field can be examined from a number of neurobiological perspectives [21,24,59,60]. Systems include neurotransmission, neuroplasticity, neuroimmunology, the Hypothalamopituitary Adrenal (HPA) axis, oxidative stress and mitochondrial physiology [21,24,59,60]. The effects of PA will be outlined below, with a specific focus on immune effects.

PA appears to exert significant effects on the immune system in depression. A clinical trial by Kohut et al. randomized 87 healthy older adults (64 to 87 years) to either aerobic (CARDIO) or strength and flexibility (FLEX) training 3 days/week, 45 min/day for 10 months [61]. A subgroup of subjects treated with non-selective β1 and β2 adrenergic antagonists were included to evaluate the potential role of β-adrenergic receptor adaptations as mediators of PA-induced change in inflammation. The study found CARDIO treatment resulted in significant reductions in serum CRP, IL-6 and IL-18 compared to FLEX treatment, whereas TNF-α declined in both groups. However, both groups had similar improvements with depressive symptoms (measured by the Geriatric Depression Scale (GDS)). This may be explained by emerging literature suggesting both beneficial and detrimental effects of pro-inflammatory cytokines, depending on their environmental milieu (outlined extensively in another review [62]). We have previously suggested that these beneficial and detrimental effects of PICs and other immune factors are in a state of balance, and that differing PA subtypes may exert variable effects on this balance [21]. For example, TNF-α has been found to exert detrimental and antineuroplastic effects via TNF-R1/p55, whereas TNF-R2/p75 has been found to exert beneficial and pro-neuroplastic effects [62-64]. A recent study by Rethorst et al. investigated the extent to which inflammatory markers can be used to predict response to exercise treatment after an incomplete response to a Selective Serotonin Reuptake Inhibitor (SSRI) [65]. This study randomized 73 participants aged 18 to 70 years to 12 weeks of either 4 or 16 kilocalories per Kilogram of Body Weight per Week (KWW), via aerobic PA. The study examined serum IFN-γ, IL-1β, IL-6, TNF-α and made a number of interesting findings. High baseline TNF-α (>5.493 pg/ml) was associated with a greater reduction in depressive symptoms (measured by Inventory for Depressive Symptomatology Clinical (IDS-C)). There was also a significant correlation between reductions of IL-1β and depressive symptoms in the 16 KWW group, but not the 4 KWW group. Otherwise, there was no significant change in cytokine levels following the 12 week PA intervention, and a non-significant association between PA dose and change in cytokine levels. A cross-sectional study by the same authors, Rethorst et al., examined the relationship between IL-6 and depressive symptoms by participation in moderate-intensity PA in a sample of 97 primary care patients aged > 40 years [66]. The patients had a Center for Epidemiological Studies Depression (CES-D) scale of > 15 and PA was determined by a questionnaire. In this study, there was no correlation between IL-6 and depressive symptoms and no effect for PA on IL-6 levels; however, the association between IL-6 and depressive symptoms was moderated by PA. In physically inactive subjects, higher depressive symptoms were associated with higher IL-6 levels. An RCT by Irwin et al. investigated the effects of Tai Chi Chih(TCC) on inflammatory markers (IL-6, CRP, sIL-1ra, sIL-6, sICAM, IL-18), as compared to control (Health Education (HE)) [67]. This study involved 83 healthy older adults (aged 59-86) in a 16 week intervention with 9 week follow up. This study found subjects with high IL-6 at entry (>2.46 pg/ml) had TCC-induced reductions in IL-6 to levels comparable to those found in TCC and HE subgroups who have low levels of IL-6 at baseline, whereas IL-6 levels in HE remained higher than the TCC and HE subgroups with low baseline IL-6. Reductions in depressive symptoms in the two groups correlated with decreases of IL-6. TCC did not affect other inflammatory markers.

A recent study may assist in correlating epidemiological and neuroscience findings in this area. A prospective cohort study based on participants from the Whitehall II study has found regular PA is associated with lower markers of inflammation over 10 years of followup [68]. This is important given the clinical and pre-clinical findings that depression may be precipitated by a pro-inflammatory state [69,70]. The study by Hamer et al. utilized 4289 men and women (mean age 49.2 years) and attained self-reported PA and inflammatory markers (hsCRP and IL-6) at baseline (1991) and follow up (2002) [68]. Higher PA levels over 10 years were associated with lower CRP and IL-6 levels as compared to lower PA levels. This study suggests that high-adherence to PA guidelines may be important in preventing the pro-inflammatory state seen with physical inactivity, as well as depression [68].

PA is found to enhance 5-HT, HA and DA signaling [21,24,59,60]. PA has been found to induce ROS formation leading to an up-regulation in endogenous antioxidant defenses, increased anti-oxidant/oxidative damage-repairing enzyme activity, increased resistance to oxidative stress and lowered levels of oxidative damage [21,24,59,60]. Long-term PA is shown to improve HPA axis responsiveness [21,24,59,60]. For example, PA has recently been found to reduce rumination (chronic stress) related increases in cortisol [71]. HPA axis hyperactivity has been repeatedly shown in depressed subjects, and is thought to be due to reduce glucocorticoid receptor functioning leading to glucocorticoid insensitivity [39,72]. PA is associated with enhanced adult HC neurogenesis and increased activity-dependent synaptic plasticity [21,24,59,60]. A recent review by Voelcker-Rehage and Niemann compares the clinical effect of ‘metabolic exercise’ (i.e. cardiovascular and resistance training) and co-ordinative PA (i.e. motor fitness, coordination and flexibility) on cognitive function in aging and markers of neuroplasticity [73]. The authors present evidence arguing pure metabolic PA has greater effects on brain volume and functional activity, particularly in the prefrontal and Hippocampal (HC) areas, as compared to stretching, toning or relaxation interventions (for primary evidence see); they contribute this effect to differing metabolic demands [73-76]. How this relates to studies in depression is unclear.

The rise in burden of depression, physical inactivity and metabolic syndrome make the case for developing innovative, PA-related strategies in the prevention of depression compelling [1,2,6,7]. Developing an integrated understanding of PA-related strategies from public health, clinical and neuroscience perspectives is important given the complex nature of this issue. Figure 1 provides a graphical representation of the multiple facets involved in preventive science approaches using PA as a preventive and treatment strategy for depression.

PA has shown positive effects in the treatment and prevention of depression and this data comes from epidemiological, clinical, psychological and neuroscience perspectives. A review of the epidemiological literature suggests higher levels of PA and Cardio- Respiratory Fitness (CRF) are associated with lower levels of depressive symptoms, lower incidence of depression and lower inflammation levels [10-18]. Clinical trials suggest that aerobic, resistance and mind-body PA types may be therapeutic in the treatment of sub-syndromal and major depression, as well as relapse prevention [26,34,29-32,48-53].

Figure 1: Schema for further exploring the role of physical activity in the treatment and prevention of depression The figure outlines the importance of considering physical activity subtypes, as well as the type of research and preventive approach.

Emerging evidence suggests there may be unique biological profiles of PA subtypes, and a greater understanding of these profiles may assist in increasing PA efficacy by personalizing treatment approaches. From an immune perspective, the RCT by Kohut et al. suggests aerobic treatment resulted in significant reductions in serum CRP, IL-6 and IL-18 compared to strength and flexibility treatment, whereas TNF-α declined in both groups [61]. Another study compared resistance and aerobic exercise in sedentary adults and found resistance PA produced a greater reduction in CRP than aerobic PA, 32.8% vs. 16.1%, respectively [77]. Although these first studies are suggestive of distinct PA subtype specific immunological changes, it is difficult to draw firm conclusions on the specific immunomodulatory profiles for these PA subtypes at this stage of the research given the small number of comparative studies and methodological heterogeneity between studies (e.g. study population age and illness severity. Biological factors which may underlie immunomodulatory and neuroplastic profiles of PA subtypes include specific effects on adipose tissue, muscle, blood vessels, vagal tone and the brain [77-80].

Immune markers may be moderators of the treatment efficacy of various treatment approaches. Aerobic PA is shown to be more efficacious with subjects (partial responders to SSRIs) who have a higher baseline TNF-α (>5.493 pg/ml) [65]. This finding may suggest TNF-α as a moderator between SSRI and exercise treatment, and TNF-α levels could be used to recommend exercise rather than medication as part of a personalized treatment algorithm [65]. Eller et al. found high baseline TNF-α associated with non-response to an SSRI, and the Hannestad et al. meta-analysis also supports this association [81,82]. The RCT by Irwin et al.found subjects which high IL-6 at entry (>2.46 pg/mL) have greater reductions in IL-6 and depressive symptoms than control [67]. The magnitude of this reduction was similar to aerobic PA [83].

It is important to acknowledge the psycho-social benefits of PA in the treatment and prevention of depression. Studies have shown exercise regimens have a distraction effect (from negative thoughts and ruminations), provide a sense of mastery via the learning of new skills, and hence enhance self-efficacy and self-esteem [84-86]. Additionally, exercise regimens in a group setting may have a beneficial effect via training social skill deficits [26]. The social training aspects of PA may be a confounder when assessing the biological effects of PA. Social isolation stress is repeatedly shown to enhance inflammation in clinical and pre-clinical models [87].

The investigation of the preventive effects of PA on depression is a rapidly developing, important field. This paper summarizes the most recent evidence from public health, clinical and immunological studies which taken together support a role for PA in the treatment and prevention of depression. This paper proposes an integrated approach to this field in order to enhance the efficacy of preventive measures in the future.

All authors declare that there are no conflicts of interest

There are no competing interests

HE conceived of the review and drafted the manuscript. KS and BB made substantial contributions to review design, analysis, interpretation and critically revision of data and the manuscript. All authors read and approved the final manuscripts.

KS supported by an Australian Research Council Future Fellowship (FT991524).The presented work is supported by the National Health and Medical Research Council Australia (APP 1043771 to BTB). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.