Introduction

There are several common circumstances in which treatments outside the traditional Western Allopathic medical tradition can be useful for patients with treatment-resistant depression (TRD). While many patients benefit only partially from maximized medication and psychotherapy and desire additional evidence-based recommendations, a holistic approach has other advantages. First, many patients request alternative and complementary treatments with a goal of avoiding or reducing pharmacological therapies and their side effects. Second, behavioral changes that fall outside the commonly used psychotherapeutic approaches can be addressed via complementary approaches. Finally, patients may come from cultural or personal perspectives that value a mind-body approach which when used adjunctively with traditional treatment can improve engagement and adherence and strengthen the therapeutic alliance. Fortunately, many interventions have sufficient clinical evidence of efficacy and underlying scientific and mechanistic foundation to provide a strong menu of recommendations for mental health professionals. Here we will review diet, exercise, mindfulness, and closely related interventions, including some that cross domains.

Diet and nutrition

Although the idea the diet is an important determinant of health, including mental health, has persisted for thousands of years, there is little high-quality contemporary evidence defining “healthy diet” with respect to specific mental disorders. In general, diets are described by ratios of macronutrients and their subtypes (e.g., saturated vs unsaturated fats), by food type (e.g., legumes, dairy, vegetables), by micronutrient and/or fiber content (e.g., low sodium), or a combination of these ( ). The concept of “health” in turn is defined based on long-term weight, cardiovascular, and metabolic outcomes ( ). Thus, expert consensus defines a “healthy diet” as one with carbohydrate > protein > fat caloric macronutrient content, consisting largely of minimally processed plant-derived foods to ensure sufficient fiber, vitamins, and micronutrients ( ).

Most studies focused on diet with respect to major depressive disorder (MDD) are epidemiological, with some clinical trials of diet education interventions in clinical or population samples using self-reported assessments. Mediterranean diets and those conforming to Food and Drug Administration or European health agency guidelines are the most commonly studied. Overall, these data consistently support that healthier diets, grossly defined, are associated with better mood. At the population level, multiple metrics of diet health are associated with lower current and longitudinal depression symptoms ( ). Similarly, trials of diet education/planning interventions in various clinical populations find small decreases in depression symptoms ( ). One small, low quality clinical trial has been specifically designed to reduce depression in a sample with investigator validated MDD; it found that a Mediterranean diet support program was superior to “social support” control ( ; ). However, three trials in obese depressed adults suggest that diet interventions designed to improve metabolic health and lower weight also improve depression symptoms ( ; ; ). Unfortunately, there are no large, high-quality trials of direct diet interventions and none which recruited subjects who were identified as treatment resistant.

Despite the lack of TRD-specific data, the close relationship between mood and metabolic outcomes means that clinicians can confidently recommend the same diet recommended to the general population to patients. Additional recommendations may apply for those with TRD ( ). Patients should be screened and treated appropriately for eating disorders, particularly binge eating disorder which has a high rate of comorbidity with depression ( ; ). The impact of commonly used adjunctive medications that can increase appetite and weight, as well as directly induce metabolic changes should be considered and minimized ( ). Good quality evidence supports the use of metformin or topiramate for metabolic risk reduction in patients on atypical antipsychotics regardless of indication ( ).

Probiotics

Extensive evidence from animal models links the gut microbiome to MDD. Experimental inductions of depression-like behaviors induce changes in microbiome composition, and vice versa ( ). Furthermore, comparisons of depressed human subjects to healthy controls consistently find differences in microbiome composition; although the nature of the differences is variable, a possible theme may be a reduction in taxa producing short chain fatty acids and other metabolically active or antiinflammatory compounds ( ). Lack of consensus on the nature of microbial changes in MDD has not impeded the investigation of probiotic supplements in clinical trials. Overall, there is a small reduction in depression symptoms across treatments ( ; ), with those including more rigorously defined clinical samples showing a slightly larger effect ( ; ). Only two nonrandomized studies recruited patients with rigorously defined TRD ( ; ). This remains an area in which more work is needed to make clinical recommendations, particularly as the risks of probiotic supplements are poorly understood; cases of bacterial over-growth, including small bowel colonization, and infection in immunocompromised persons, have been reported ( ).

Mindfulness

Broadly, mindfulness describes the practice of “bare attention” a common English translation of the Buddhist term sati , and in this sense is a subtype of traditional Buddhist meditation originating in China and best known in the western world via Japanese Zen Buddhism ( ). Modern mindfulness practice aims to foster mental calm, improve awareness of physical and mental states, and develop self-compassion ( ). Clinically, mindfulness programs have been designed to improve tolerance of negative affect, reduce symptoms of depression, and facilitate healthy regulation of emotion and behavior ( ). Meditation can be practiced by patients independently, with support from a large variety of written, audio, and digital guides, or within a psychotherapeutic frame. Alternatively, manualized therapies incorporating meditation—“mindfulness-based interventions” (MBIs)—integrate secular mindfulness exercises (meditation) with group psychoeducation and psychotherapeutic techniques ( ).

The two major MBIs are mindfulness-based stress reduction (MBSR) developed for patients with chronic, painful, or debilitating medical illness and comorbid psychiatric symptoms ( ), and mindfulness-based cognitive therapy (MBCT), developed for relapse prevention in MDD ( ). MBCT has been studied specifically for TRD in several trials, while MBSR has been applied more specifically to treatment resistance in those with chronic medical illness ( ). Other interventions may use concepts of mindfulness without meditation practice (e.g., Acceptance and Commitment Therapy); here we focus on therapies featuring the practice of regular mediation.

Overall, MBIs are superior to treatment-as-usual, and nonactive or active nonspecific controls and at least as effective as other evidence-based therapies; when used as augmentation MBIs have a larger effect size, suggesting that they may be more effective in TRD patients compared to the general MDD population ( ; ; ). MBCT, unsurprisingly, may be more effective for relapse prevention than acute treatment ( ) but some studies in TRD have shown superiority to active control conditions ( ; ). MBSR appears to be effective for depression symptoms in a wide array of clinical conditions, with a small effect size, when compared to inactive control, and is more effective for anxiety and somatic symptoms (e.g., pain, fatigue) than for core symptoms of depression ( ; ). Limited data suggest similar efficacy for a primary diagnosis of MDD without comorbidity ( ). Independent practice of mindfulness exercises (e.g., body scan, breathing or sitting meditation) using audio, or video guide materials is common, and more feasible than MBIs, though it appears to have less benefit ( ). Adherence may mitigate this gap in efficacy; however, there are little data from samples with well validated MDD or in which independent mindfulness is used as augmentation. Similarly, studies with active controls show little or no benefit to independent practice ( ).

Overall, it appears likely that patients with TRD can benefit from mindfulness interventions, but may require the higher level of structure and active support of an MBI to maintain engaged practice and obtain symptomatic improvement. MBCT has more applicability and evidence for use in TRD, but patients dealing with pain or loss of function from medical illness are good candidates for treatment augmentation with MBSR. While sparse cases of induction of psychotic or manic symptoms following intense meditation have been reported, mindfulness is considered low risk, and can be safely recommended for most patients.

Mindful movement

Mindful movement practices combine physical activity with mindful breathing and meditation. Although commonly used for wellness, these practices have been adapted less rigorously to treatment of mental illness, and have less data examining their use as treatment compared to MBIs ( ; ). Yoga combines physical postures (asanas), breath control, and mindful attention, and has the largest evidence base for MDD. Modern practice may involve significant athletic, even aerobic, activity but this style of practice dates only to the mid-20th century, while older forms emphasize meditation ( ). Although there are methodological limitations to published studies ( ), yoga appears effective for depression symptoms compared to inactive control conditions and roughly equivalent to other exercise or mindfulness ( ; ). Data are decidedly more mixed for well-defined samples with MDD ( ; ; ). Two studies comparing samples with some degree of treatment resistance to active control were both negative ( ; ). Some data suggest that in clinical samples with more severe symptoms lower adherence may contribute to negative findings ( ; ). Additionally, the rate of yoga associated adverse events may be higher than for either aerobic exercise or meditation ( ).

With roots stretching back centuries in traditional Chinese Medicine, Tai Chi and Qi Gong share components of body posture, fluid movement, and synchronization with breathing ( ). Both practices show efficacy for symptoms of depression compared to inactive control conditions, and light activity (e.g., walking) though studies are of low quality ( ; ). One study of moderate quality examined Tai Chi vs health education augmentation of SSRI in older depressed adults and found it superior for mood improvement ( ), but there are no data specifically pertaining to TRD. Tai Chi appears to be low risk ( ).

Both the quality of evidence and the support for efficacy of mindful movement are less strong than for MBIs, or for exercise (discussed later), in particular, there is a lack of support for use of mindful movement in patients with TRD. However, in patients who express a preference for mindful movement, they can be recommended as adjuncts, keeping in mind that patients who have low fitness/flexibility should use caution in undertaking a yoga practice due to its relatively high rate of minor musculoskeletal injuries. As with other mindful interventions, support structures which enhance adherence, such as supervision and group education may improve the benefits associated with mindful movement.

Exercise

There is significant evidence substantiating the efficacy of exercise as a treatment for MDD with moderate effect size, similar to other evidence supported treatments ( ; ). Larger effects are found in studies including only inactive controls and smaller effects in studies with more rigorous trial designs ( ; ; ); however, the overall study quality is close to that of pharmacologic trials ( ). Generally, studies examining exercise as a combination or adjunctive treatment with antidepressant medication find similar results ( ). There are lack of data specifically examining well-defined TRD, with only two very small trials ( ; ).

The majority of trials have incorporated aerobic exercise, which is easier to standardize across subjects in units of kilocalories expended in a given study period, while a minority prescribe resistance training or a combination. It appears likely that these types of exercise are similar in efficacy for symptom reduction ( ; ). Similarly, differences in dosing (time/week) and intensity (light, moderate, vigorous) seem to have small or no difference in symptom reduction, although this may be due to lower adherence to higher prescribed doses. Findings of no difference in trials with active controls such as walking or stretching ( ) suggest that some of the efficacy of exercise comes from shared elements of behavioral activation. There are lack of data on moderators/mediators of outcome other than adherence, which is often low, and likely explains findings that supervision from a professional trainer, group or other structured programs may have superior efficacy ( ; ). In general, enjoyment of exercise rather than self-efficacy (i.e., the feeling that one is engaging in a “should do” activity) is associated with long term engagement in exercise, and suggests the importance of patient preference in choosing among the various options for type, dosing, and intensity ( ). Surprisingly, few studies have rigorously collected and reported adverse event data, and risks vary substantially by type of exercise. All of these factors point toward recommendations of exercise with supervision from professionals on an individual or group basis in which patients can engage in activities that fit with their personal preferences and physical capabilities.

Acupuncture

Acupuncture is a centuries-old component of traditional Chinese medicine in which specified body points are stimulated, traditionally using fine needles, to manipulate a body force called Qi ( ). It has been applied broadly to a variety of clinical conditions but the evidence base is limited by a failure to harmonize the concept of Qi with modern physiology and by low to very low quality of clinical trials ( ). Additionally, MDD is nosologically inconsistent with traditional Chinese medicine, so there is no consensus on which points should be targeted ( ). Among many positive results, studies which use high quality design with adequate blinding of sham procedures find no effect ( ). Of 13 studies acupuncture augmentation of antidepressant treatment, only two reported any blinding, and the overall effect size was negligible ( ). Although most of the interventions included here share similar challenges in trial design and execution, high-level interpretation of acupuncture data suggests effects are likely due to placebo effect ( ). Acupuncture cannot be recommended for depression or for TRD as a primary or secondary intervention. However, risks associated with acupuncture are low, and it may be helpful for patients who desire to seek it out for treatment.

Common mechanisms of action of lifestyle modifications

All of these lifestyle interventions share a foundational assumption that the health of the body and mind are linked, so it is not surprising that they share proposed and probable mechanisms of action based on the interaction of somatic systems with the nervous system. These interactions are mediated primarily by humoral signaling molecules and the autonomic nervous system ( ; ; ; ); the brain processes somatic information via the insula-centered interoceptive system ( ), and the hypothalamic system, important as a hub for metabolic control as well as the stress response ( ). These systems are interlinked and modulate other brain circuits critical for depression, such as the limbic system, and have been implicated repeatedly in the neurobiology of MDD ( ; ).

Chronic increases in proinflammatory cytokines are now a well-established finding in MDD ( ), and likely form a central hub in the network of somatic mechanisms of action of diet, exercise, and mindfulness. Chronic inflammation is associated with illness (e.g., autoimmune disease), obesity, and glucocorticoid resistance, all of which increase the risk of depression ( ; ; ). The rate of overweight and obesity in depression is about twice that of the general population, and adiposity appears to be the single largest predictor of inflammation in depressed adults ( ). Impairments in cortisol signaling and interoceptive autonomic nervous system efferent activity also increase peripheral inflammation by reducing feedback inhibition on the immune system, may contribute to further adiposity by impairing hypothalamic metabolic control and accurate sensation of hunger and satiety, respectively ( ; ; ). Inflammation alters functional connectivity to reduce hedonic drive and increase feels of social exclusion ( ; ). Local CNS inflammation impairs neuroplasticity and neurogenesis by decreasing the production of neurotrophins and increasing that of neurotoxic kynurenine metabolites ( ).

While data are not fully consistent, there is a probable clinical association between treatment resistance and chronic inflammation ( ; ), as well as some evidence that antiinflammatory therapies may successfully treat TRD ( ). While it is probable that obesity associated inflammation largely explains the consistency of the long-term association between metabolically healthy diets and mood, measures of diet associated inflammation are often better predictors of depression symptoms ( ; ; ; ; ; ). Exercise broadly decreases indices of inflammation, and some data support particular efficacy in those who have elevated cytokines at study entry ( ; ). Exercise also directly increases neurotrophin production and reduces the production of neurotoxic kynurenine metabolites, counteracting the impact of inflammation in the brain ( ). Mindfulness and mindful movement decrease peripheral inflammation, though perhaps less robustly than exercise, and may also decrease cortisol resistance ( ; ; ).

At the brain circuit level, lifestyle interventions may exert antidepressant effects by improving insula dependent interoception. Accurate interoception is related to higher resting functional connectivity between insular regions and between the insula and other parts of the salience network ( ), and the ability to functionally disconnect and suppress insula activity during somatic attention tasks and those requiring attention to outside stimuli, respectively ( ; ). Depressed subjects have lower interoceptive accuracy, increased resting activity and functional connectivity of the anterior insula compared to healthy subjects, and show impairments in their ability to modify insula activity and connectivity in tasks requiring shifting attention ( ). Several studies have found that excess anterior insula activity correlates with treatment resistance to antidepressants or TMS ( ; ; ). Although there are lack of data in depressed samples, mindfulness training improves interoceptive accuracy ( ) and may normalize insula activity and integration into the salience network ( ; ).