Introduction
Major depressive disorder (MDD) is a debilitating psychiatric condition, ranked as the sixth most disabling disorder globally, affecting approximately 350 million people worldwide and significantly contributing to morbidity and mortality (Global Burden of Disease, 2020; Walker, McGee, & Druss, Reference Walker, McGee and Druss2015; World Health Organization, 2017). Its pathophysiology involves alterations in neurotrophic pathways, inflammation, immune responses, endocrine function, the monoaminergic system, and stress-related pathways such as the hypothalamic–pituitary–adrenal axis (Anderson & Maes, Reference Anderson and Maes2014; Berger et al., Reference Berger, Kraeuter, Romanik, Malouf, Amminger and Sarnyai2016; Chocyk et al., Reference Chocyk, Majcher-Maślanka, Dudys, Przyborowska and Wędzony2013; Maffioletti, Minelli, Tardito, & Gennarelli, Reference Maffioletti, Minelli, Tardito and Gennarelli2020). Current pharmacological treatments primarily target these biomarkers, particularly within the monoaminergic system. However, they often fail to fully alleviate symptoms and are associated with side effects, including weight gain, cardiovascular issues, and sexual dysfunctions (Dodd et al., Reference Dodd, Bauer, Carvalho, Eyre, Fava, Kasper and Berk2021). Consequently, non-pharmacological interventions are increasingly recognized for their role in treating depression symptoms.
Physical exercise is an effective non-pharmacological strategy for MDD treatment, symptom alleviation, and as an adjunct to enhance the efficacy of antidepressant medications and other therapies. Several systematic reviews and meta-analyses have demonstrated that physical exercise significantly reduces depression symptoms (Cooney et al., Reference Cooney, Dwan, Greig, Lawlor, Rimer, Waugh and Mead2013; Josefsson, Lindwall, & Archer, Reference Josefsson, Lindwall and Archer2014; Krogh et al., Reference Krogh, Hjorthøj, Speyer, Gluud and Nordentoft2017; Schuch et al., Reference Schuch, Vancampfort, Richards, Rosenbaum, Ward and Stubbs2016). The latest guidelines for MDD management recommend regular physical activity at low to moderate intensity (30–40 minutes per session, three to four times per week, for at least 9 weeks) as a primary treatment for mild depression and an adjunct treatment for moderate cases (Lam et al., Reference Lam, Kennedy, Adams, Bahji, Beaulieu, Bhat and Milev2024; National Institute for Health and Care Excellence, 2022; Zhou, Puder, & Fabiano, Reference Zhou, Puder and Fabiano2024). A meta-review demonstrated that physical activity significantly improves depressive symptoms, with effects comparable to those of antidepressants and psychotherapy, recommending moderate to vigorous aerobic exercise, ideally combined with resistance training, performed two to three times per week for a total of 150 minutes under professional supervision (Stubbs et al., Reference Stubbs, Vancampfort, Hallgren, Firth, Veronese, Solmi and Kahl2018).
Despite robust evidence on the effectiveness, the neurobiological mechanisms underlying exercise’s antidepressant effects are not fully understood. Previous systematic reviews and meta-analyses indicate that physical activity induces sustained responses in hormones, neurotrophins, and inflammatory/immune processes (Feter, Alt, Dias, & Rombaldi, Reference Feter, Alt, Dias and Rombaldi2019; Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014). Exercise is known to modulate depression-related biomarkers, including interleukins (IL) such as IL-6, a key component of the immune and inflammatory systems, and other molecular signatures associated with MDD pathophysiology (Chow et al., Reference Chow, Gerszten, Taylor, Pedersen, van Praag, Trappe and Snyder2022; Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014).
A recent systematic review and meta-analysis investigating the chronic effects of physical activity on biomarkers in MDD patients included 12 randomized controlled trials (RCTs) with 757 participants. The study found that exercise significantly increased circulating levels of brain-derived neurotrophic factor (BDNF) and kynurenine while reducing depression symptoms compared to the control groups (da Cunha, Feter, Alt, & Rombaldi, Reference da Cunha, Feter, Alt and Rombaldi2023). Improvements in circulating BDNF, kynurenine, and IL-6 levels were associated with reductions in depression symptoms (da Cunha et al., Reference da Cunha, Feter, Alt and Rombaldi2023). Another systematic review and meta-analysis examined the effects of antidepressant treatment, with or without physical exercise, on inflammatory biomarkers in MDD patients (Fernandes et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022). The authors identified six trials investigating biomarkers, such as IL-1β, IL-6, serum cortisol, and C-reactive protein (CRP), reporting significant methodological heterogeneity and conflicting results (Fernandes, Scotti-Muzzi, & Soeiro-de-Souza, Reference Fernandes, Scotti-Muzzi and Soeiro-de-Souza2022).
Although evidence highlights the neurobiological effects of physical exercise in MDD, findings remain inconsistent and predominantly focus on inflammatory pathways, with limited data on other biological mechanisms. Comprehensive and updated analyses are necessary to better understand the specific mechanisms underlying the effects of various forms of physical exercise in MDD treatment.
This review aims to address these gaps by updating knowledge on the biological mechanisms of exercise in MDD and broadening the scope to encompass a wider range of biological systems. This study broadens the scope of previous analyses by exploring less-studied biological systems beyond the immune-inflammatory and neurotrophic pathways, including oxidative stress, monoaminergic, and stress-response mechanisms. It also incorporates updated evidence from RCTs, offering a more comprehensive and current synthesis of the biological mechanisms of exercise. By analyzing a broader range of RCTs, we aim to contribute to a clearer understanding of the neurobiological effects of physical exercise in adult MDD patients.
Materials and methods
This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow and Moher2021) and was registered on PROSPERO (CRD42024548511). See the literature search in Supplementary Materials.
Eligibility criteria
All retrieved records were imported into Zotero for reference management, and duplicates were removed. Two independent reviewers screened titles and abstracts. Disagreements on inclusion were resolved by a third reviewer. Abstracts lacking sufficient inclusion or exclusion information were assessed in full text.
Studies were included if they met the following criteria: (1) RCTs evaluating physical exercise interventions and their effects on biomarkers of any biological system, with pre- and post-intervention blood markers. Only RCTs with a control group or those comparing two or more interventions were considered. Exercise was defined as a planned, structured, and repetitive intervention aimed at improving or maintaining physical conditioning; (2) conducted in adults aged 18 or older, diagnosed with MDD according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV, DSM-IV-TR, DSM-5, International Classification of Diseases (ICD)-10, and ICD-11 criteria; and (3) published in English.
Exclusion criteria were: (1) studies involving children and adolescents or those focusing on conditions other than MDD; (2) non-RCTs; (3) RCTs on physical exercise in MDD patients without pre- and post-intervention biomarker data.
Data extraction
Data were extracted following PRISMA guidelines and included authors, year of publication, study design, sample size, participant characteristics (age, gender, socioeconomic status), diagnostic criteria, medications, treatment duration, inpatient/outpatient status, exercise intervention details (type, duration, intensity, supervision), depression and symptom measures, biomarkers (pre- and post-intervention), assay type, and manufacturer. The primary outcome was the identification of the biological effects of physical exercise in MDD patients. Secondary outcomes included changes in depression and related symptoms following exercise interventions.
Risk of bias (quality) assessment
The Revised Cochrane Risk of Bias (RoB 2) tool for randomized trials was used to assess methodological quality and internal validity (Sterne et al., Reference Sterne, Savović, Page, Elbers, Blencowe, Boutron and Higgins2019). This tool evaluates bias across five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. The overall study quality was coded as low, of some concerns, or high. Two independent researchers did the study quality assessments, discussing with a third reviewer to resolve inconsistencies.
Results
Characteristics of the included studies
A flowchart of the search process is shown in Figure 1, whereas the main characteristics of the included studies are summarized in Table 1. A total of 2194 subjects were included. The mean age in the exercise intervention groups was 42.3 years, compared to 42.8 years in the control groups. The percentage of females in the intervention and control groups was 68.5% and 65%, respectively.
Figure 1. Flowchart showing the study selection procedure.
Table 1. Summary of the main characteristics of the included randomized controlled trials
Abbreviations: AD: antidepressant; BDNF: brain-derived neurotrophic factor; CBT: cognitive behavior therapy; COMT: catechol-O-methyltransferase; ECT: electroconvulsive therapy; F: female; KKW: kilocalories per kilogram of bodyweight per week; MDD: major depressive disorder.
Information on illness duration could not be systematically reported, as it was often missing or inconsistently described across the included studies. The clinical course and duration of depression varied widely, ranging from single moderate episodes to chronic, recurrent forms of depression. All included studies confirmed the diagnosis of depression using standardized diagnostic instruments.
Main findings Inflammatory and immune systems
Fifteen studies exploring the effects of physical exercise on inflammatory and immune biomarkers in MDD were included.
Exercise combined with cognitive-behavioral therapy (CBT) significantly increased anti-inflammatory IL-10 levels in patients with MDD (Euteneuer et al., Reference Euteneuer, Dannehl, del Rey, Engler, Schedlowski and Rief2017). However, other studies found no changes in IL-10 levels in MDD patients submitted to exercise in comparison to controls (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). Lower baseline levels of the pro-inflammatory IL-6 predicted treatment non-response in patients with MDD undergoing two exercise doses for 12 weeks (Chad D. Rethorst, South, Rush, Greer, & Trivedi, Reference Rethorst, South, Rush, Greer and Trivedi2017). IL-6 levels significantly reduced after hatha yoga intervention in comparison with health education control in depressed patients (Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021). In addition, yoga and meditation-based lifestyle interventions reduced IL-6 levels in comparison to controls in patients with MDD (Tolahunase, Sagar, Faiq, & Dada, Reference Tolahunase, Sagar, Faiq and Dada2018). In other findings, although elevated IL-6 levels were observed in patients with MDD, this marker was not significantly influenced by exercise intervention (Krogh et al., Reference Krogh, Benros, Jørgensen, Vesterager, Elfving and Nordentoft2014). Also, other results did not show significant changes in IL-6 following exercise in addition to pharmacological treatment in depressed patients in comparison with drug therapy alone (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023) or following exercise intervention in patients with MDD versus controls (Hennings et al., Reference Hennings, Schwarz, Riemer, Stapf, Selberdinger and Rief2013). Other cytokines were also studied. No significant changes in IL-12, IL-8, and IL-1β were observed following exercise intervention in comparison to controls in patients with MDD (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022). Similarly, no significant IL-1β changes were observed following a 15-week exercise intervention in MDD patients (Shi & Miao, Reference Shi and Miao2017). A large panel of cytokines was analyzed by Siddarth and colleagues comparing patients with MDD undergoing Tai Chi Chih (TCC) or health education, with no differences between groups (Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). However, exercise and mind–body practices reduced IL-1β levels in MDD patients (Torelly, Novak, Bristot, Schuch, & Pio de Almeida Fleck, Reference Torelly, Novak, Bristot, Schuch and Pio de Almeida Fleck2022). Also, reductions in hypersomnia were linked to decreases in IL-1β following exercise in patients with MDD, suggesting a relationship between cytokine changes and depressive symptoms after exercise (C. D. Rethorst et al., Reference Rethorst, Greer, Toups, Bernstein, Carmody and Trivedi2015). These findings suggest that specific ILs, particularly IL-6, IL-10, and IL-1β, may be differentially affected by exercise and mind–body interventions, with varying results.
Exercise and mind–body practices showed mixed findings on CRP. Exercise combined with CBT marginally reduced CRP levels in depressed patients, particularly in those with cardiovascular risk (Euteneuer et al., Reference Euteneuer, Dannehl, del Rey, Engler, Schedlowski and Rief2017). Reduced CRP levels were observed in elderly MDD patients undergoing TCC alongside escitalopram, supporting the notion that physical and mind–body practices may reduce inflammation in subgroups of depressed individuals (Lavretsky et al., Reference Lavretsky, Alstein, Olmstead, Ercoli, Riparetti-Brown, St. Cyr and Irwin2011). In contrast, no significant changes in CRP levels were found in depressed individuals following hatha yoga intervention (Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021), and no differences in CRP levels were found between aerobic and stretching exercise programs (Krogh et al., Reference Krogh, Videbech, Thomsen, Gluud and Nordentoft2012). Elevated CRP levels were observed in MDD patients, but this marker was not significantly influenced by exercise (Krogh, Benros, et al., Reference Krogh, Benros, Jørgensen, Vesterager, Elfving and Nordentoft2014).
Tumor necrosis factor-alpha (TNF-α) responses were similarly inconsistent. Baseline TNF-α levels showed a correlation with treatment response, with higher baseline levels predicting greater reductions in depressive symptoms during exercise (C. D. Rethorst et al., Reference Rethorst, Toups, Greer, Nakonezny, Carmody, Grannemann and Trivedi2013), suggesting that TNF-α may predict treatment efficacy. An acute decrease in TNF-α levels after a single session of mixed aerobic and anaerobic exercise was reported by Torelly et al. (Reference Torelly, Novak, Bristot, Schuch and Pio de Almeida Fleck2022). On the contrary, other authors found no significant changes in TNF-α levels following hatha yoga (Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021) or TCC (Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023), and no differences were observed between aerobic exercise and controls for this biomarker (Imboden et al., Reference Imboden, Gerber, Beck, Eckert, Lejri, Pühse and Hatzinger2021). These findings underline the complexity of TNF-α as a biomarker of exercise in MDD, with its responsiveness varying depending on the type and duration of the intervention.
Neurotrophic system
Twelve studies investigated the effects of exercise on BDNF, a neurotrophic factor linked to mood regulation and neuroplasticity, in exercise interventions for MDD. Aerobic exercise for 6 weeks significantly increased serum BDNF concentrations in patients with MDD (Kerling et al., Reference Kerling, Kück, Tegtbur, Grams, Weber-Spickschen, Hanke and Kahl2017). Mind–body practices and physical exercise increased BDNF levels in MDD patients, with a time-dependent effect (Torelly et al., Reference Torelly, Novak, Bristot, Schuch and Pio de Almeida Fleck2022). Similarly, other results showed increased BDNF levels following exercise, although these changes did not directly correlate with improvements in depressive symptoms (Szuhany & Otto, Reference Szuhany and Otto2020). Higher BDNF levels predicted remission, while lower levels of this marker predicted non-response in MDD patients randomized for two exercise doses over 12 weeks (Chad D. Rethorst et al., Reference Rethorst, South, Rush, Greer and Trivedi2017). Conversely, BDNF reductions were associated with improved hypersomnia in non-remitted MDD patients (C. D. Rethorst et al., Reference Rethorst, Greer, Toups, Bernstein, Carmody and Trivedi2015). Other studies showed mixed results regarding the relationship between exercise and BDNF levels in MDD. An increase in serum BDNF levels was observed during a 6-week inpatient treatment program for depression, with no significant differences between the aerobic exercise and stretching groups (Imboden et al., Reference Imboden, Gerber, Beck, Eckert, Lejri, Pühse and Hatzinger2021). Similarly, no significant changes were found in BDNF levels in MDD patients following a 3-month exercise intervention compared to controls, despite improved aerobic capacity (Krogh et al., Reference Krogh, Rostrup, Thomsen, Elfving, Videbech and Nordentoft2014). Other findings showed no significant changes in BDNF levels with exercise, despite improvements in depressive symptoms, highlighting the variability in BDNF responses to physical activity (Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014).
Baseline BDNF moderated response to exercise in MDD patients partially responding to antidepressants, with higher baseline levels linked to faster symptom improvement, even though overall BDNF levels did not correlate with depression improvement (Toups et al., Reference Toups, Greer, Kurian, Grannemann, Carmody, Huebinger and Trivedi2011). Significant BDNF increases and symptom improvements were observed with electroconvulsive therapy (ECT), aerobic exercise, and combined treatment in MDD patients, with the highest remission rates found in the ECT combined with exercise (Salehi et al., Reference Salehi, Hosseini, Haghighi, Jahangard, Bajoghli, Gerber and Brand2016). Significant correlations between depressive symptoms decrease and increased BDNF levels were found in MDD patients undergoing yoga- and meditation-based lifestyle intervention compared to the controls (Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018).
Studies on other neurotrophic factors, such as vascular endothelial growth factor (VEGF) and insulin-like growth factor 1 (IGF-1), provided inconsistent results. No significant effects of aerobic exercise were found on VEGF or IGF-1 levels in MDD patients, despite memory and depressive symptoms improvements (Krogh, Rostrup, et al., Reference Krogh, Rostrup, Thomsen, Elfving, Videbech and Nordentoft2014). No significant VEGF changes were found in depressed adults undergoing TCC versus controls (Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). These findings suggest that VEGF and IGF-1 may not be primary targets for exercise interventions in MDD, although they could play a role in cognitive improvements associated with exercise. Other factors, including epidermal growth factor (EGF), fibroblast growth factor-2, and transforming growth factor-α, were also unaffected by TCC in MDD patients (Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023).
Stress-response system
Eight studies on the effects of exercise on the stress-response system in MDD were included. Cortisol, a key stress-response marker, is widely studied in this context. Six weeks of aerobic exercise did not significantly alter cortisol reactivity to a laboratory stress task in MDD patients, many of whom displayed blunted baseline cortisol responses, indicating a limited impact of exercise on hypothalamic–pituitary–adrenal (HPA) axis reactivity (Gerber et al., Reference Gerber, Imboden, Beck, Brand, Colledge, Eckert and Hatzinger2020). Similarly, overall cortisol reductions during antidepressant treatment showed no significant differences between patients engaging in physical exercise and those who did not, revealing limited additional effects of exercise on cortisol regulation in MDD (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022). Adding yoga to quetiapine or escitalopram treatment did not significantly enhance HPA axis downregulation beyond medication alone (Sarubin et al., Reference Sarubin, Nothdurfter, Schüle, Lieb, Uhr, Born and Baghai2014), and exercise combined with pharmacotherapy did not alter cortisol levels compared to pharmacotherapy alone (Carneiro et al., Reference Carneiro, Mota, Vieira-Coelho, Alves, Fonseca and Vasconcelos-Raposo2017). Krogh and colleagues found no effects of strength or aerobic training on resting cortisol levels or cortisol responses to acute exercise stress in MDD patients versus controls (Krogh, Nordentoft, Mohammad-Nezhad, & Westrin, Reference Krogh, Nordentoft, Mohammad-Nezhad and Westrin2010). However, yoga and meditation significantly reduced cortisol levels in the intervention group in comparison with controls (Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018).
Research on the cortisol awakening response (CAR) presented mixed results. A decreased CAR over time was observed in moderate-to-severe MDD patients during inpatient treatment, regardless of whether aerobic exercise or stretching was used, indicating that the decline may be treatment-related rather than exercise-specific (Imboden et al., Reference Imboden, Gerber, Beck, Eckert, Lejri, Pühse and Hatzinger2021). In contrast, Krogh et al. found no significant associations between cortisol, adrenocorticotropic hormone, or copeptin in depressed patients at rest or after acute exercise, indicating that these stress markers do not consistently respond to exercise interventions. Additionally, copeptin levels showed no significant differences between depressed patients and healthy controls, further indicating it may not play a critical role in exercise responses (Krogh et al., Reference Krogh, Gøtze, Jørgensen, Kristensen, Kistorp and Nordentoft2013).
Monoaminergic system
The relationship between physical exercise, MDD, and monoaminergic biomarkers has also been studied, although findings remain inconsistent. Our review included four studies, with one genetic study and others focusing on candidate protein biomarkers in peripheral blood.
A genetic study explored serotonin-related gene polymorphisms, specifically the serotonin transporter-linked polymorphic region (5-HTTLPR) within the serotonin transporter gene (SLC6A4) and MTHFR 677C > T, in MDD patients. Yoga significantly increased remission odds compared to pharmacotherapy, regardless of genotype. Interestingly, childhood adversity interacted with these polymorphisms to reduce treatment response in the pharmacotherapy group but not in the yoga arm (Tolahunase, Sagar, & Dada, Reference Tolahunase, Sagar and Dada2018).
Carneiro and colleagues found that combining exercise with pharmacotherapy reduced the activity of soluble catechol-O-methyltransferase, and enzyme involved in catecholamine metabolism and relevant to mood regulation (Carneiro et al., Reference Carneiro, Fonseca, Serrão, Mota, Vasconcelos-Raposo and Vieira-Coelho2016). However, other authors reported no significant changes in plasma levels of monoamine neurotransmitters, such as serotonin, dopamine, noradrenaline, or adrenaline, when aerobic exercise was added to pharmacotherapy in depressed women (Carneiro et al., Reference Carneiro, Mota, Vieira-Coelho, Alves, Fonseca and Vasconcelos-Raposo2017), highlighting the complexity of exercise’s effects on monoamines in MDD.
Regarding serotonin metabolism, short-term graded exercise did not alter serotonergic markers such as tryptophan, kynurenine, or 5-hydroxyindoleacetic acid in patients with MDD or somatoform syndrome (Hennings et al., Reference Hennings, Schwarz, Riemer, Stapf, Selberdinger and Rief2013). Exercise did not significantly affect serotonergic parameters, especially in short durations or when used as an isolated intervention.
For noradrenaline, Dabidy Roshan and colleagues found increased 24-hour urinary MHPG sulfate, a noradrenaline metabolite, after intermittent water-walking exercise, correlating with symptom reduction (Dabidy Roshan, Pourasghar, & Mohammadian, Reference Dabidy Roshan, Pourasghar and Mohammadian2011). These findings suggest the noradrenergic system may mediate mood improvements linked to exercise, although the mechanisms remain unclear.
Other systems
Other biological systems, including the endocrine system (two studies), oxidative stress markers (two studies), and β-endorphins (one study), have also been examined.
The impact of exercise on the endocrine system in MDD shows mixed results. Altered growth hormone (GH) and cortisol responses to acute exercise were observed in MDD patients compared to controls, but no significant changes in resting hormone levels occurred over time (Krogh et al., Reference Krogh, Nordentoft, Mohammad-Nezhad and Westrin2010). Conversely, aerobic exercise improved metabolic markers, including lower fasting glucose levels and reduced waist circumference in MDD patients, although without significant antidepressant effects (Krogh et al., Reference Krogh, Videbech, Thomsen, Gluud and Nordentoft2012).
Exercise’s ability to reduce oxidative stress was supported by findings of reduced thiobarbituric acid-reactive substances, a marker of oxidative stress, in severely depressed inpatients following exercise (Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014), supporting that exercise can reduce oxidative damage. Yoga- and meditation-based interventions also decreased oxidative stress markers in MDD patients, highlighting the benefits of these practices (Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018).
Regarding β-endorphins, exercise combined with antidepressant therapy increased serum β-endorphin levels, correlating with greater reductions in depression severity in the intervention versus control groups (Wang & Li, Reference Wang and Li2022). This suggests that exercise-induced β-endorphin release may enhance clinical outcomes.
The main findings of the included studies are summarized in Table 2, which provides an overview of the effects of physical exercise on various biological systems in patients with MDD.
Table 2. Summary of the main findings
Abbreviations: 5-HTTLPR: serotonin-transporter-linked promoter region; ACTH: adrenocorticotropic hormone; AUC: area under the curve; BDNF: brain derived neurotrophic factor; BMI: body mass index; CAR: cortisol awakening response; CBT: cognitive behavior therapy; COMT: catechol-O-methyltransferase; CRP: C-reactive protein; DEX/CRH: dexamethasone/corticotropin releasing hormone; ECT: electroconvulsive therapy; EGF: epidermal growth factor; FGF: fibroblast growth factor; GH: growth hormone; HPA: hypothalamic pituitary adrenal axis; IGF: insulin-like growth factor; IL: interleukin; KKW: kilocalories per kilogram of bodyweight per week; MDD: major depressive disorder; MTHFR: methylenetetrahydrofolate reductase; SE: standard error; TAU: treatment as usual; TCC: Tai Chi Chih; TBARS: thiobarbituric acid-reactive substances; TNF-α: tumor necrosis factor-alpha; VEGF: vascular endothelial growth factor.
Biomarkers as potential predictors of treatment response
Higher baseline levels of TNF-α were associated with greater reductions in depressive symptoms during exercise treatment in patients with MDD (C. D. Rethorst et al., Reference Rethorst, Toups, Greer, Nakonezny, Carmody, Grannemann and Trivedi2013). Similarly, baseline levels of IL-6 were predictive of treatment response, with lower IL-6 levels identifying non-responders to a 12-week exercise program (Chad D. Rethorst et al., Reference Rethorst, South, Rush, Greer and Trivedi2017). In the neurotrophic domain, higher baseline concentrations of BDNF were associated with a faster symptom improvement in patients partially responding to antidepressants who received adjunctive exercise (Toups et al., Reference Toups, Greer, Kurian, Grannemann, Carmody, Huebinger and Trivedi2011). Additionally, BDNF levels at baseline predicted remission following a structured aerobic exercise program, with higher levels correlating with positive treatment outcomes (Chad D. Rethorst et al., Reference Rethorst, South, Rush, Greer and Trivedi2017). Genetic variations also showed relevance in treatment stratification. Polymorphisms in 5-HTTLPR and MTHFR 677C > T interacted with a history of childhood adversity to predict reduced responsiveness to pharmacological interventions but were not associated with treatment outcomes in the yoga intervention group (Tolahunase, Sagar, & Dada, Reference Tolahunase, Sagar and Dada2018).
Quality of studies (RoB analyses)
The RoB evaluation for the included studies showed that 29 studies had an intermediate RoB, whereas one study had a high risk. In the analysis of each of the five domains, four studies had low risk, and 26 studies had some concerns in bias arising from the randomization process. For bias due to deviations from intended interventions, one study had a low risk, and 29 studies had some concerns. Bias due to missing outcome data was low in 29 studies, with one study showing some concerns. In bias related to outcome measurement, 26 studies had low risk, and four studies had some concerns. Finally, for bias in the selection of reported results, 20 studies had low risk, and 10 studies had some concerns. The main findings are summarized in Supplementary Figure 1.
Discussion
In our systematic review, we contribute to the literature on the biological effects of physical exercise in MDD patients, highlighting various biological systems likely involved in exercise’s effects on MDD. The included studies showed mixed results.
For the inflammatory and immune systems, exercise combined with CBT increased IL-10 levels (Euteneuer et al., Reference Euteneuer, Dannehl, del Rey, Engler, Schedlowski and Rief2017), highlighting the potential of exercise to modulate immune responses, although other studies found no significant changes (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). IL-6 reductions were associated with certain interventions (Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021; Chad D. Rethorst et al., Reference Rethorst, South, Rush, Greer and Trivedi2017; Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018), with other studies showing variable findings (Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022; Krogh, Benros, et al., Reference Krogh, Benros, Jørgensen, Vesterager, Elfving and Nordentoft2014; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). IL-1β responses were variable (C. D. Rethorst et al., Reference Rethorst, Greer, Toups, Bernstein, Carmody and Trivedi2015; Shi & Miao, Reference Shi and Miao2017; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023; Torelly et al., Reference Torelly, Novak, Bristot, Schuch and Pio de Almeida Fleck2022), suggesting that cytokine modulation depends on the type of intervention and patient characteristics. In general, exercise and mind–body practices reduced IL-1β levels in MDD patients, pointing to an anti-inflammatory effect of these interventions. CRP exhibited limited responsiveness to exercise (Krogh, Benros, et al., Reference Krogh, Benros, Jørgensen, Vesterager, Elfving and Nordentoft2014; Lavretsky et al., Reference Lavretsky, Alstein, Olmstead, Ercoli, Riparetti-Brown, St. Cyr and Irwin2011; Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021), with reductions in cardiovascular risk subgroups, suggesting a modest anti-inflammatory effect of the intervention (Euteneuer et al., Reference Euteneuer, Dannehl, del Rey, Engler, Schedlowski and Rief2017). These findings indicate that while CRP might be modulated by certain exercise interventions, its sensitivity as a biomarker in MDD treatment may be influenced by patient characteristics such as age or cardiovascular risk. TNF-α levels predicted treatment response (Rethorst et al., Reference Rethorst, Toups, Greer, Nakonezny, Carmody, Grannemann and Trivedi2013), although the findings were inconsistent (Imboden et al., Reference Imboden, Gerber, Beck, Eckert, Lejri, Pühse and Hatzinger2021; Nugent et al., Reference Nugent, Brick, Armey, Tyrka, Ridout and Uebelacker2021; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). Overall, findings related to the inflammatory and immune systems demonstrate that exercise and mind–body practices can influence inflammatory and immune biomarkers in patients with MDD. While ILs, CRP, and TNF-α are modulated by these interventions, their effects are not consistent across studies, highlighting the need for further research.
In the neurotrophic system, exercise increased BDNF levels, supporting physical activity as an effective adjunct therapy (Kerling et al., Reference Kerling, Kück, Tegtbur, Grams, Weber-Spickschen, Hanke and Kahl2017; Salehi et al., Reference Salehi, Hosseini, Haghighi, Jahangard, Bajoghli, Gerber and Brand2016; Szuhany & Otto, Reference Szuhany and Otto2020; Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018; Torelly et al., Reference Torelly, Novak, Bristot, Schuch and Pio de Almeida Fleck2022), although its link to symptom improvement is unclear (Imboden et al., Reference Imboden, Gerber, Beck, Eckert, Lejri, Pühse and Hatzinger2021; Krogh, Rostrup, et al., Reference Krogh, Rostrup, Thomsen, Elfving, Videbech and Nordentoft2014; Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014). The findings suggest that while exercise influences BDNF, its role in mediating clinical benefits may be complex. Other neurotrophic markers, including VEGF and IGF-1, were less consistently affected by exercise (Krogh, Rostrup, et al., Reference Krogh, Rostrup, Thomsen, Elfving, Videbech and Nordentoft2014; Siddarth et al., Reference Siddarth, Abikenari, Grzenda, Cappelletti, Oughli, Liu and Lavretsky2023). In conclusion, BDNF is a key neurotrophic factor in depression treatment, but its role in mediating clinical improvements through exercise remains unclear. VEGF and IGF-1 may play less direct roles in MDD recovery, requiring further exploration. The neurotrophic system’s influence on MDD appears to be complex, depending on baseline characteristics and intervention types.
Within the stress-response system, cortisol showed limited sensitivity to exercise interventions (Carneiro et al., Reference Carneiro, Mota, Vieira-Coelho, Alves, Fonseca and Vasconcelos-Raposo2017; Fernandes, Siqueira, et al., Reference Fernandes, Siqueira, Vieira, Moreno and Soeiro-de-Souza2022; Gerber et al., Reference Gerber, Imboden, Beck, Brand, Colledge, Eckert and Hatzinger2020; Krogh et al., Reference Krogh, Nordentoft, Mohammad-Nezhad and Westrin2010); however, mind–body practices, such as yoga, reduced cortisol levels (Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018). The effects of physical exercise on the stress-response system in MDD appear to be potential but limited, as most studies reported no significant changes in cortisol levels or reactivity, with only mind–body practices like yoga and meditation showing some promise in reducing cortisol, suggesting modest benefits of exercise on HPA axis regulation.
The monoaminergic system findings indicated that gene polymorphisms, particularly 5-HTTLPR and MTHFR 677C > T, interacted with childhood adversity to modulate treatment response (Tolahunase, Sagar, & Dada, Reference Tolahunase, Sagar and Dada2018). Exercise also influenced catecholamine metabolism (Carneiro et al., Reference Carneiro, Fonseca, Serrão, Mota, Vasconcelos-Raposo and Vieira-Coelho2016), but effects on serotonin and other monoamines were inconsistent (Carneiro et al., Reference Carneiro, Mota, Vieira-Coelho, Alves, Fonseca and Vasconcelos-Raposo2017; Hennings et al., Reference Hennings, Schwarz, Riemer, Stapf, Selberdinger and Rief2013). The effects of physical exercise on monoaminergic biomarkers in MDD appear to be promising but limited, with inconsistent findings across studies. While some evidence suggests modest benefits on catecholamine metabolism and noradrenergic activity, other studies report no significant changes in serotonin-related markers, highlighting the need for further research to clarify these complex and variable responses.
In other systems, exercise influenced metabolic markers (Krogh et al., Reference Krogh, Videbech, Thomsen, Gluud and Nordentoft2012); however, it did not consistently affect hormones like GH (Krogh et al., Reference Krogh, Nordentoft, Mohammad-Nezhad and Westrin2010). The results indicate that, although not translating into a significant antidepressant effect, exercise can have beneficial effects on metabolic aspects of the endocrine system. Exercise, yoga, and meditation-based interventions also reduced oxidative stress (Schuch et al., Reference Schuch, Vasconcelos-Moreno, Borowsky, Zimmermann, Wollenhaupt-Aguiar, Ferrari and de Almeida Fleck2014; Tolahunase, Sagar, Faiq, et al., Reference Tolahunase, Sagar, Faiq and Dada2018), indicating the possible benefits of these practices. In summary, while endocrine responses to exercise remain inconsistent, metabolic improvements, such as reductions in oxidative stress and potential increases in β-endorphin levels (Wang & Li, Reference Wang and Li2022), are more consistently observed and may support better treatment outcomes. Further research is needed to clarify the conditions under which exercise most effectively modulates these biological markers.
Although there are systematic reviews and meta-analyses in this domain (da Cunha et al., Reference da Cunha, Feter, Alt and Rombaldi2023; Fernandes, Scotti-Muzzi, & Soeiro-de-Souza, Reference Fernandes, Scotti-Muzzi and Soeiro-de-Souza2022), our review is distinct in several aspects. While prior reviews focused on established biological systems involved in MDD and physical exercise, such as the inflammatory/immune and neurotrophic systems, our study employed a broader approach by considering additional biological systems. This strategy allows us to examine the biological effects of physical exercise from multiple perspectives, uncovering interactions and mechanisms not sufficiently reviewed in the literature. Indeed, our review expands on previous analyses by examining less-explored biological systems beyond immune-inflammatory and neurotrophic pathways, including oxidative stress, monoaminergic systems, and stress-response mechanisms. An additional strength of our review is the updated search, including studies published after 2020, providing a more current perspective on this evolving field. It further integrates updated evidence from RCTs, providing a more comprehensive and contemporary understanding of the biological mechanisms underlying exercise in MDD.
Analyzing the included studies based on the RoB assessment revealed issues that need to be addressed. The RoB evaluation showed that most studies had an intermediate risk, with one classified as high risk. Concerns were prevalent in domains associated with the randomization process, deviations from intended interventions, and the selection of reported results, where many studies exhibited a moderate RoB. These findings emphasize the need for methodological improvements in these areas to enhance the reliability of randomized trial outcomes.
Regarding the characteristics of the included studies, all used a candidate gene approach, evaluating a few biomarkers, mainly biochemical ones. While biochemical markers are relevant, the exclusive focus on them is a limitation. Internal and external factors can influence biochemical markers, and focusing solely on these markers may simplify the biological complexity of exercise effects on MDD. Expanding the approach to include genetic, epigenetic, and gene expression studies provide a more comprehensive understanding of the biological impacts of physical exercise on MDD.
Additionally, most included studies concentrated on biomarkers related to the inflammatory, immune, and neurotrophic systems. Although this focus aligns with the known pathophysiology of MDD and the mechanisms through which exercise exerts its benefits, it may limit our understanding of the full spectrum of biological changes that may occur with physical exercise in MDD. Future research should consider a wider array of biomarkers and employ broader approaches, such as omics technologies, to evaluate other systems and biomarkers.
The small sample sizes in many included studies represent another limitation. Small samples reduce the statistical power to detect effects and, combined with the heterogeneity in study designs, complicate the interpretation and generalizability of findings. For example, the control groups varied greatly, with some studies using non-MDD participants or participants engaging in different intensities of activity. This variability in controls influences comparisons between groups and weakens the interpretability of the findings. Specifically, the methodological variability across the included studies likely contributed to the inconsistencies observed in the results. Differences in exercise protocols (e.g., type, intensity, duration), sample characteristics (e.g., age, sex, comorbidities), and timepoints and methods of biomarker collection may have introduced heterogeneity in outcomes. Additionally, many studies lacked standardized outcome measures and often treated biological results as secondary rather than primary outcomes, limiting the rigor of biomarker assessments. Variability in the control group design and insufficient blinding or allocation concealment in several trials, as revealed by the RoB assessment, limit direct comparisons across studies. These methodological differences underscore the need for more standardized protocols and higher-quality RCTs to improve the reproducibility and interpretation of biological findings.
Furthermore, many studies do not effectively report effect sizes, making it challenging to evaluate the real impact of significant findings. Without clear information on effect sizes, it becomes difficult to assess the clinical relevance of the biological changes, limiting the understanding of the practical implications of the results. This gap limits a comprehensive understanding of the strength and consistency of exercise’s effects on biomarkers in MDD, essential for clinical translation.
There is also variability in the characteristics of study populations among the included studies. Differences in age, ethnicity, comorbidities, and other confounding variables add limitations to the interpretation of results. Moreover, the outcomes evaluated across studies are heterogeneous, and often, the biological effects of exercise were secondary outcomes rather than primary outcomes.
In addition to the limitations inherent to the included studies, such as small sample sizes, methodological variability in intervention and control groups, exercise protocols, biomarker assessment methods, and inconsistent reporting of statistical results and effect sizes, our review has certain limitations. One concern is publication and results selection bias, as studies with significant or favorable findings are more likely to be published, overrepresenting positive outcomes. Additionally, the absence of a meta-analysis limits the quantitative synthesis of findings, restricting a comprehensive evaluation of overall effect sizes and their relevance to clinical practice.
In conclusion, our systematic review highlights the complexity of the biological effects of physical exercise in MDD patients, revealing significant heterogeneity across biological systems. While exercise influences biomarkers in the inflammatory, neurotrophic, stress-response, and monoaminergic systems, the results are inconsistent and depend on the intervention type, patient characteristics, and the biomarkers studied. By exploring a broader range of biological systems, our review identifies potential new mechanisms underlying the therapeutic effects of exercise. However, limitations such as small sample sizes, varied study designs, inconsistent reporting of results, and a predominant focus on biochemical markers reduce the generalizability of the findings. Future research should prioritize larger, more homogeneous RCTs with standardized methods, a broader range of biomarkers, and advanced omics approaches, such as genomics, proteomics, and metabolomics, to clarify the biological mechanisms of exercise. Addressing these considerations would improve the understanding of the pathways involved in the beneficial effects of exercise on MDD. Overall, the findings highlight the complexity of physical exercise’s influence on biomarkers in MDD, warranting further research to clarify its therapeutic potential.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291725100743.
Data availability statement
Data from the current study can be made available upon request to the corresponding author.
Acknowledgments
The authors would like to thank Francesca Malandrone and Nicoletta Colombi for their valuable contribution to this work.
Author contribution
Conceptualization: A.M. Data curation: R.C.S., M.M., and G.P. Formal analysis: R.C.S., M.M., G.P., and A.M. Project administration and supervision: A.M. Visualization: M.B. and M.G. Writing-original draft: R.C.S. and A.M. Writing-review and editing: S.C., A.M., and B.B.
Funding statement
This work was supported by the Italian Ministry of Health under Grant (Ricerca Corrente 2024). The post-doc position of Dr. Rosana Carvalho Silva was partly funded by the Psychiatric Hospital ‘Villa Santa Chiara’, Verona, Italy. The assistant research position of Giulia Perusi is funded by ERA-PerMed PROMPT project (IT-MoH ERP-2020-23671059).
Competing interests
The authors declare no conflicts of interest.
Open access
