Changes in Neural Activities and Neuroplasticity Related to Nonpharmacological Interventions for Major Depressive Disorder: A Systematic Literature Review

Abstract

People with major depressive disorder (MDD) can have impaired neuroplasticity. Antidepressant treatment and some nonpharmacological interventions can lead to changes in neuroplasticity that improve MDD symptoms. However, there are no recent systematic literature reviews (SLRs) on the effect of nonpharmacological interventions for MDD on neuroplasticity. Therefore, we conducted an SLR of articles with primary results published between January 1, 2013, and December 6, 2023, that included adults with depression or MDD (MDD used to refer to both) treated with nonpharmacological products that are U.S. Food and Drug Administration (FDA) cleared and indicated for MDD or are investigative and need FDA review and clearance for use outside of clinical trials. From the 1257 records screened, 101 studies with 4746 participants were included. Electroconvulsive therapy was the most common treatment (used by 46.5% of the studies), followed by repetitive transcranial magnetic stimulation (35.6%). Of the 54 studies that included a healthy control comparison group, 42 (77.8%) found brain differences at baseline between the MDD group and the control group. Most of the studies (95 studies; 94.1%) found statistically significant functional or structural changes in the brain following nonpharmacological treatment for MDD. Of the 74 studies that investigated whether there was a relationship between changes in the brain and improvement in MDD symptoms, 53 (71.6%) found that changes in neuroplasticity corresponded with improvement in depression symptoms. This SLR shows that nonpharmacological interventions for MDD lead to changes in neuroplasticity, which correspond with improvement in MDD symptoms.

Plain Language Summary

Major depressive disorder (MDD) can affect neuroplasticity, which is the brain’s ability to change and adapt to new experiences. We reviewed studies on nondrug treatments for MDD to better understand how these treatments affect neuroplasticity. The most commonly studied treatments included electroconvulsive therapy, transcranial magnetic stimulation, transcutaneous vagus nerve stimulation, and cognitive behavioral therapy. We found that nondrug treatments for MDD can lead to changes in neuroplasticity and help improve MDD symptoms.

Plain Language Summary

Major depressive disorder (MDD) can affect neuroplasticity, which is the brain’s ability to change and adapt to new experiences. We reviewed studies on nondrug treatments for MDD to better understand how these treatments affect neuroplasticity. The most commonly studied treatments included electroconvulsive therapy, transcranial magnetic stimulation, transcutaneous vagus nerve stimulation, and cognitive behavioral therapy. We found that nondrug treatments for MDD can lead to changes in neuroplasticity and help improve MDD symptoms.

Major depressive disorder (MDD) affects 1 in 5 U.S. adults during their lifetime (1), and depressive disorders are a leading contributor of disability worldwide (2,3). Numerous obstacles prevent patients from accessing and adhering to treatment, such as limited availability of providers, financial barriers, stigma about mental illness, partial or nonresponse to antidepressant treatment, and medication side effects (4, 5, 6, 7). Understanding the underlying causes of MDD can improve the treatment landscape. One hypothesis is that MDD develops via impaired neuroplasticity (8,9). Neuroplasticity is the brain’s ability to undergo changes in structure and function in response to stimuli (10). Mechanisms of neuroplasticity include changes in synaptic strength (e.g., through the timing of action potentials, ion channel density, release of neurotransmitters) and structural modifications (e.g., morphological changes in axons, dendritic branches and spines, myelination) (10,11). Neuroplasticity changes related to adverse experiences during brain development can be maladaptive in the long term and increase the risk of developing a mental illness (12,13). Diminished neuroplasticity has been reported in people with MDD, including reduced synaptogenesis, increased glial cell dysfunction, reductions in neurotrophic factors, dendritic atrophy, and altered connectivity between neural networks (14,15).

Treatments for MDD aim to increase neuroplasticity in specifically targeted regions of the brain associated with symptoms. Brain regions implicated in MDD that have altered activity and neuroplasticity include the anterior cingulate cortex (ACC) (decreased activity), prefrontal cortex (PFC) (decreased activity), and amygdala (increased neuroplasticity and activity) (13). Antidepressant treatment (ADT) has been shown to restore some of these differences in neuroplasticity. For example, ADT can decrease hyperactivity of the amygdala and ventral striatum and decrease functional connectivity of the amygdala to the PFC and orbitofrontal cortex (9,13,14). Cognitive behavioral therapy (CBT) has been associated with changes in connectivity between the medial PFC and ACC, which correlates with depression symptom improvement (16,17).

Better understanding of the neuroplasticity differences involved in both the development and treatment of MDD may help guide therapeutic decision making (11). Although a recent review reported on the effects of pharmacological treatment for MDD on neuroplasticity (13), no recent systematic literature review (SLR) exists on the effects of nonpharmacological treatments. We aimed to synthesize the published information about the changes in brain structure and function associated with nonpharmacological treatments for people with MDD.

Methods and Materials

Study Selection

We conducted an SLR of articles with primary results published in English between January 1, 2013, and December 6, 2023, in Ovid MEDLINE In-Process & Other Non-Indexed Citations, Embase, and the Cochrane Library (CDSR, DARE, CENTRAL, CMR, and NHS EED) that report on neuroplasticity outcomes of nonpharmacological treatments for depression in adults, using the search string shown in Table 1.

Table 1.

Ovid Search String Used to Conduct This Systematic Literature Review

Search String

(Nonpharmacological OR cognitive behavior therapy OR transcranial magnetic stimulation OR tDCS OR tACS OR vagus nerve stimulation OR electroconvulsive therapy OR DTx OR digital therapeutic OR digital therapy) AND (Major depressive disorder OR depression) AND (Neuroplasticity OR EEG OR MEG OR fMRI OR resting state connectivity OR functional connectivity OR structural connectivity OR connectomics OR neuromodulation)

DTx, digital therapeutics; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; MEG, magnetoencephalography; tACS, transcranial alternating current stimulation; tDCS, transcranial direct current stimulation.

We also screened references from relevant review articles (16,18). Included articles reported on brain structure or function at baseline and posttreatment to detect any treatment-related changes. Studies involving participants with MDD or depression were included. Although MDD and depression were both used in the included studies, for succinctness, we use the term MDD throughout.

There were 2 rounds of screening (titles and abstracts, then full text). Four people conducted the screening, with each article screened by 1 person each round. If a reviewer was uncertain about whether an article should be included, a second reviewer was consulted. We included articles with adults who had depression or MDD as the primary diagnosis that reported quantifiable findings on changes in the brain and that described primary results of studies of nonpharmacological depression treatments. Full inclusion and exclusion criteria are described in Table 2, and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist is shown in Table S1. The review was not registered, and all methods are described in this publication.

Table 2.

PICOS Criteria for Study Selection

Description	Inclusion Criteria	Exclusion Criteria
Population	Adults with depression or MDDa	•Studies on mental health diagnoses other than depression or MDD (e.g., addiction disorders, bipolar depression, eating disorders, premenstrual dysphoric disorder, postpartum depression, schizophrenia, substance abuse) •Studies on participants with co-occurring health diagnoses (e.g., cancer, COVID-19, COPD, epilepsy, chronic pain) •Studies of animal models •Nonadults (<18 years of age) •Studies on late-life depression (depression diagnosed at ≥65 years of age) and studies focusing on older adult populations
Intervention	•Nonpharmacological interventions that were FDA approved or cleared and indicated for depression and MDD •Investigative nonpharmacological interventions for depression and MDD that would need FDA review and approval or clearance to be used outside of clinical trials •Interventions that met these criteria included but were not limited to CBT, DTx, ECT, TMS, VNS	•Psychotherapy other than CBT (e.g., non-CBT psychotherapy), including DTx using other forms of psychotherapy •Pharmacological interventions, including medication, ketamine, esketamine, 3,4-methylenedioxymethamphetamine, psychedelics (i.e., psilocybin, lysergic acid diethylamide, ayahuasca) •Nonpharmacological interventions other than those specifically listed for inclusion, such as exercise, meditation, yoga
Comparators	Not restricted	NA
Outcomes	•Quantifiable findings on changes to neuroplasticity elements at study baseline (before intervention) and outcome (after intervention) •Brain changes (i.e., structure and function) as measured by resting-state connectivity, functional connectivity, structural connectivity, EEG, MEG, or connectomics	•Studies that included only outcomes that are not localized to the brain (e.g., changes in the blood, muscles, hormones) •Studies that reported depression symptoms only •Studies that did not report on brain traits both at baseline and posttreatment (e.g., studies that used modeling or classification but did not quantify/report the change from baseline to posttreatment)
Study Design	Includes primary results	•Reviews and systematic reviews •Meta-analyses
Language Restrictions	English language	Non-English languages
Publication Type	Primary article	Publications such as congress abstracts, editorials, foreign-language articles with no English-language abstract, gray literature, letters, protocol-only publications, reviews, secondary publications, unpublished data (ongoing studies in registries)
Date Restrictions	January 1, 2013, to December 6, 2023	Prior to 2013
Database Restrictions	•Ovid MEDLINE In-Process & Other Non-Indexed Citations •Embase •Cochrane	NA

CBT, cognitive behavioral therapy; COPD, chronic obstructive pulmonary disease; DTx, digital therapeutics; ECT, electroconvulsive therapy; EEG, electroencephalography; FDA, U.S. Food and Drug Administration; MDD, major depressive disorder; MEG, magnetoencephalography; NA, not applicable; PICOS, population, intervention, comparators, outcomes, study design; TMS, transcranial magnetic stimulation; VNS, vagus nerve stimulation.

^aBecause studies vary widely in how a diagnosis of MDD was determined (e.g., multiple guidelines, different scale scores, clinician assessment), definitions of MDD and treatment-resistant depression overlapped, and treatment-resistant depression could not be excluded.

Extraction

The following variables were extracted: the year the study was published; the type of study (e.g., randomized controlled trial [RCT]); primary diagnosis (MDD, depression); disease diagnostic criteria; disease severity score for inclusion (e.g., Hamilton Depression Rating Scale [HAMD] ≥ 21); type of nonpharmacological intervention used (e.g., electroconvulsive therapy [ECT], repetitive transcranial magnetic stimulation [rTMS]); duration of treatment, including number of treatment sessions per week; MDD outcomes measured (e.g., Montgomery–Åsberg Depression Rating Scale [MADRS], Beck Depression Inventory [BDI] score); number of study participants; use of an intervention control (e.g., sham intervention, placebo); participant demographic characteristics (age, sex, race, ethnicity); any brain differences reported at baseline between participants with MDD and healthy control participants; any structural or functional brain differences (in brain regions or cognitive networks) observed posttreatment; any neuroplasticity differences between depression treatment responders and nonresponders; any neurocognitive changes after treatment; whether depression symptoms improved after treatment; and any correlations between neuroplasticity and depression improvement. We appraised study quality using the rubric developed by Hawker et al. (19), which included the following: assigning a score of good, fair, poor, or very poor to the clarity of the abstract, introduction, methods, and results; the sampling strategy; the analytical rigor; the discussion of ethics and bias; and the study’s generalizability and usefulness.

Analysis

We calculated the percentage of studies by the following: studies that used each type of nonpharmacological intervention; found brain differences between participants with depression and healthy control participants; found a change in depression symptomology from baseline to postintervention; reported neurocognitive differences posttreatment compared with baseline; identified brain differences between depression treatment responders and nonresponders; described changes in the brain posttreatment compared with baseline; and found that neuroplasticity correlated with improvement in depression symptoms. We also calculated the sum of participants (which included the total number of healthy individuals and those with depression or MDD) across all studies and the number of female and male participants.

Results

Studies Included

The search yielded 1327 records; an additional 15 records were identified from other sources (Figure 1). After removing duplicates, we screened 1257 records. During abstract screening, we excluded 545 records, mostly records that were not published articles (239/545, 43.9%) or when study participants had characteristics outside the scope of this review (136/545, 25.0%) (Figure 1 and Table S2). We screened the full text of the remaining 712 articles and excluded 611 articles. The most common reasons for exclusion were that the study outcomes (242/611, 39.6%) or characteristics of study participants (232/611, 38.0%) were outside the scope of this review (Figure 1 and Table S2). The remaining 101 studies were included in this SLR (17,20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119); a summary of their main characteristics is provided in Table S3.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart.

Study and Participant Characteristics

The 101 studies included 4746 participants, with an average age of 39.0 years (Table 3). Most studies were published in 2019 or later, specified a primary diagnosis of MDD, and were appraised as medium quality (Table 3). Few were RCTs (10.9%) or reported participant race or ethnicity data (4.0%) (Table 3). The studies featured 9 types of nonpharmacological interventions for depression; the most common were ECT (46.5%) and rTMS (35.6%) (Figure 2A). Brief descriptions of the mechanisms of action (MoAs) for each included intervention are provided in Table 4.

Table 3.

Study and Participant Characteristics

Characteristic	% or Mean (Mean Range)
Studies
Published 2013–2018	39.6%
Published 2019–2023	60.4%
RCTs	10.9%
Appraised as High Quality	22.8%
Appraised as Medium Quality	69.3%
Appraised as Low Quality	7.9%
Included an Intervention Control	24.8%
Included a Comparison Group of Healthy Control Participants	53.5%
Primary Diagnosis MDD	82.2%
Primary Diagnosis Depression	17.8%
Reported Race or Ethnicity	4.0%
Participants, N = 4746a
Mean Age, Yearsb	39.0 (24.1–60.4)
Female/Male	57.0%/43.0%

MDD, major depressive disorder; RCT, randomized controlled trial.

^aSome research groups published multiple articles on the same participants. Data from a total of 367 participants were deemed to have been included in multiple studies in this systematic literature review. When tallying the total number, percent female, and average age of participants, we included those participants only once.

^bOf the 19 studies that reported an age range for participants, the lowest age included in any study was 18 years, and the highest age included was 73 years.

Figure 2.

Synthesis of study outcomes. The studies in this SLR featured a variety of different nonpharmacological interventions for depression (some studies featured >1 intervention, so the percentages reported add to >100) (A). Some of the studies identified brain differences between healthy control participants and participants with depression at baseline (B). As part of the inclusion criteria, all studies reported on whether there were changes in brain structure or function posttreatment compared with baseline, and most found significant differences (C). Most studies reported improvement in depression symptoms after nonpharmacological intervention (D). Many studies also examined the relationship between brain changes and improvement in depression symptoms (E). Some studies investigated whether there were brain differences between participants who responded to depression treatment and participants who did not (F). CBT, cognitive behavioral therapy; ECT, electroconvulsive therapy; EFMT, Emotional Faces Memory Task; fMRI, functional magnetic resonance imaging; MST, magnetic seizure therapy; rTMS, repetitive transcranial magnetic stimulation; SLR, systematic literature review; tACS, transcranial alternating current stimulation; tDCS, transcranial direct current stimulation; tVNS, transcutaneous vagus nerve stimulation.

Table 4.

Characteristics and Neural Mechanisms of Action Related to Nonpharmacological MDD Treatment Included in This SLR

Intervention	Brief Description	Mechanism of Action	Number of Studies Using This Intervention Included in This SLRa
Neuromodulatory Interventions
ECT (142)	A procedure that induces tonic-clonic seizures in anesthetized patients; uses an electric current administered to the brain through electrodes placed on the scalp	How ECT works is poorly understood. Current research is examining the effects of the electrical field and the seizures separately to better understand individual roles in outcomes. Brain regions thought to be associated with positive outcomes include the hippocampus, ACC, temporal cortex, and PFC. ECT is also thought to affect the default mode network.	47b
rTMS (143,144)	A procedure that uses electromagnetic pulses to induce an electric current in targeted regions of the brain. Unlike ECT, rTMS does not cause seizures and does not require anesthesia.	rTMS for MDD is most frequently specifically targeted to stimulate the left dorsolateral PFC. Research suggests that rTMS modulates connectivity in the default mode network, between the left dorsolateral PFC and the insula and between the salience network and the hippocampus.	36c
tVNS/taVNS (145,146)	A treatment that applies electrical stimulation to the VN using electrodes applied superficially to the cervical nerve (tVNS) or specific parts of the ear (taVNS)	The nucleus tractus solitarii receives input from the VN, and it is hypothesized that tVNS affects brain regions connected to the nucleus tractus solitarii, including the amygdala, hippocampus, ACC, and insula. tVNS has also been shown to affect resting-state functional connectivity of the default mode network, decreasing connectivity with the anterior insula and parahippocampus and increasing it with the orbital and medial PFC, ACC, and precuneus.	9d
MST (142,147)	A procedure that induces tonic-clonic seizures in anesthetized patients; uses electromagnetic induction to create an electrical field that is weaker and more targeted than in ECT	MST is still under clinical investigation, and how it works is not yet well understood. A preliminary study found that MST resulted in increased functional connectivity between the subgenual ACC and the parietal cortex, which was correlated with depression symptom improvement.	3
tDCS (148,149)	A treatment that uses a weak electric current applied to the brain through scalp electrodes placed over either the left or right dorsolateral PFC	Although tDCS is considered a nonfocal treatment, variations in electrode placements have shown modification in electrical current distribution. It is hypothesized to stimulate and inhibit the left and right dorsolateral prefrontal cortices, respectively. tDCS is also thought to affect resting-state connectivity in the default mode and frontoparietal networks.	3
tACS (150,151)	A treatment running a low-amplitude alternating electric current through the brain between electrodes placed on the scalp, usually over structures in the frontal lobe	The electric field created by the alternating current in tACS is set at the target parameters and alters endogenous brain oscillations to have a similar frequency. tACS has been shown to alter connectivity in cortical regions and affect cortical excitability, especially in the left frontal regions.	1
Non-Neuromodulatory Interventions
CBT (16,152)	A type of psychotherapy that is designed to help patients understand how their own thoughts and behaviors influence how they perceive their symptoms. It also teaches coping skills that patients can use to modify their thoughts and behaviors. CBT can be conducted for an individual or in group sessions and can be delivered face to face and using technological tools (e.g., iCBT).	The neural mechanism of action of CBT is not well understood. Research indicates that CBT likely modulates regions of the brain associated with emotional experience and cognitive control, including the amygdala, hippocampus, PFC, and ACC; CBT may also affect the reward circuit and default mode network.	9e
fMRI Neurofeedback (153)	A technique that targets specific regions of interest in the brain. fMRI is used to collect feedback, which is converted to a visual format and shared in real time with the patient, often as they perform specific tasks. Patients then observe and attempt to modify their neural response.	fMRI neurofeedback is targeted to specific regions of the brain associated with MDD, such as the amygdala and medial PFC, and treatment effects are specific to those regions. There is some evidence that feedback in a region of interest in the left hemisphere is more effective than in the right hemisphere. Additionally, the treatment can affect connectivity networks associated with the targeted region; for example, if the amygdala is targeted, effects can be seen in the default mode network.	2
EFMT (48,154)	A modified n-back working memory task that asks the participant to recall the emotional expression from a series of faces, conducted using a computer or smartphone	A preliminary study indicates an increase in effective connectivity from the right dorsolateral PFC to the amygdala and a decrease in effective connectivity from the bilateral dorsal ACC to the amygdala.	2

ACC, anterior cingulate cortex; CBT, cognitive behavioral therapy; cTBS, continuous TBS; ECT, electroconvulsive therapy; EEG, electroencephalography; EFMT, Emotional Faces Memory Task; fMRI, functional magnetic resonance imaging; iCBT, internet CBT; iTBS, intermittent TBS; MDD, major depressive disorder; MST, magnetic seizure therapy; PFC, prefrontal cortex; rTMS, repetitive TMS; SLR, systematic literature review; tACS, transcranial alternating current stimulation; taVNS, transcutaneous auricular VNS; TBS, theta burst stimulation; tDCS, transcranial direct current stimulation; TMS, transcranial magnetic stimulation; tVNS, transcutaneous VNS; VNS, vagus nerve stimulation.

^aVariations and modifications of treatment types were grouped together as listed per treatment.

^bThe ECT category includes 2 studies that used modified ECT.

^cThe rTMS category includes studies that used iTBS (n = 6), cTBS (n = 1), paired associative stimulation (n = 2), TMS-EEG (n = 1), and TMS (n = 1).

^dThe tVNS category includes studies that used taVNS (n = 2) and VNS (n = 1).

^eThe CBT category includes studies that used iCBT (n = 2) and group CBT (n = 2).

Baseline Differences Between Participants With MDD and Healthy Control Participants

Of the 54 studies with a healthy control comparison group, 77.8% (42/54) found baseline differences in the brains of participants with MDD compared with healthy control participants (Figure 2B). Some examples of findings from individual articles include that, compared with healthy control participants and before treatment initiation, participants with MDD had significantly lower functional connectivity between the right nucleus accumbens and right dorsomedial PFC (113), higher functional connectivity between the default mode network and executive control network (26) and between the left amygdala and bilateral precuneus (30), and lower gray matter volume in the right cingulate gyrus (81).

Brain Changes Posttreatment Compared With Baseline

Most studies (95/101, 94.1%) found statistically significant changes in the brain posttreatment compared with baseline (Figure 2C). Seventeen studies reported changes in cognitive networks, most commonly the default mode network, with changes in connectivity (structural or functional) reported by 15 studies that used a variety of nonpharmacological treatments (9 used ECT, 3 used rTMS, 2 used CBT or functional magnetic resonance imaging [fMRI] neurofeedback, 1 used intermittent theta burst stimulation [iTBS]) (Table 5). Ten studies reported structural or functional connectivity changes in the control network (5 used ECT, 2 used rTMS, 2 used CBT or fMRI neurofeedback, 1 used ECT or magnetic seizure therapy [MST]), and 7 studies reported changes in the salience network (3 used ECT, 1 used rTMS, 1 used ECT or MST, 1 used CBT or fMRI neurofeedback, 1 used iTBS). Connectivity changes were also reported in the somatomotor (2 used CBT or fMRI neurofeedback, 1 used ECT), visual (1 used ECT), and attention networks (1 used ECT).

Table 5.

Cognitive Networks and Brain Regions Where Changes Were Detected After Nonpharmacological Treatment Compared With Baseline

Cognitive Networka (Anatomical Name)	Processes With Network Involvement	Number of Studies That Found Changes in the Network After Treatmentb	Brain Regions in the Networka,c	Brain Regions With Changes After Treatment Reported in Studies Included in This SLR (Number of Studies and Treatments Used)d
Default Mode (Medial Frontoparietal)	Labeled as “default” to represent areas of the brain that are more active when not focused on any specific cognitive task	15	Medial PFC; posterior cingulate cortex; posterior extent of the inferior parietal lobule; inferior frontal gyrus; middle temporal gyrus; superior temporal sulcus; parahippocampal cortex; hippocampus; superior/middle frontal gyrus; ventral frontal cortex, anterior temporal lobes; temporoparietal junction; orbitofrontal cortex; areas dorsal and ventral to the posterior cingulate, the precuneus, and retrosplenial cortex, respectively; angular gyrusb	Hippocampus (7 ECT, 3 rTMS, 1 CBT, 1 VNS); posterior cingulate cortex (7 ECT, 1 tVNS, 1 rTMS, 1 tDCS, 1 iCBT or MACe); middle frontal gyrus (8 ECT, 1 rTMS, 1 iTBS, 1 taVNS); angular gyrus (10 ECT, 1 taVNS); superior frontal gyrus (7 ECT, 2 rTMS, 1 iTBS); orbitofrontal cortex (3 ECT, 1 MECT, 3 rTMS); dorsomedial PFC (5 ECT, 2 tVNS); middle temporal gyrus (6 ECT, 1 tVNS, 1 rTMS); medial PFC (3 ECT, 1 tVNS, 2 group CBT, 1 tDCS); inferior frontal gyrus (4 ECT)
Control (Lateral Frontoparietal)	Executive control (e.g., goal oriented cognition, working memory)	10c	Lateral PFC, middle frontal gyrus, rostral PFC, dorsolateral PFC, anterior inferior parietal lobule, intraparietal sulcus, midcingulate gyrus, dorsal precuneus, posterior inferior temporal lobe, dorsomedial thalamus, head of the caudate, ACCb, precentral gyrusb	ACC (7 ECT, 7 rTMS, 3 tVNS, 1 tDCS, 1 iTBS, 2 group CBT, 1 iCBT or MACe, 1 MECT); dorsolateral PFC (9 rTMS, 4 ECT, 2 TMS-EEG, 1 MST, 1 tDCS, 1 MST or ECTe, 1 tVNS); precentral gyrus (4 ECT, 1 MECT, 1 tVNS, 1 CBT); caudate (2 ECT, 1 rTMS, 1 taVNS, 1 group CBT); inferior parietal lobule (3 ECT, 2 rTMS)
Salience (Midcingulo-Insular)	Identification of salient information	7	Bilateral anterior insula, anterior midcingulate cortex, inferior parietal cortex, right temporal parietal junction, lateral PFC, substantia nigra/ventral tegmental area, periaqueductal gray, amygdala, hypothalamus, parabrachial nucleus, basal ventromedial nucleus of the thalamus	Amygdala (8 ECT, 4 rTMS, 1 tVNS, 1 iTBS); insula (4 ECT, 1 tVNS, 2 rTMS, 1 CBT or fMRI neurofeedbacke)
Somatomotor (Pericentral)	Somatosensory and motor processes	3	Motor cortex, somatomotor cortex, juxtapositional lobule (supplementary motor area), auditory cortex of the superior temporal gyrus	Superior temporal gyrus (7 ECT, 2 rTMS)
Visual (Occipital)	Visual processing	1	Occipital lobe, striate cortex, extrastriate cortex, lateral geniculate nucleus of the thalamus	None
Attention (Dorsal Frontoparietal)	Visuospatial attention	1	Superior parietal lobule, intraparietal sulcus, middle temporal complex, ventral premotor cortex, right-lateralized dorsolateral PFC, superior colliculus, supramarginal gyrus, precuneusc	Precuneus (8 ECT, 2 taVNS, 1 rTMS, 1 tVNS, 1 CBT); superior parietal lobule (2 ECT, 1 rTMS); supramarginal gyrus (3 ECT, 1 rTMS, 1 taVNS, 1 MECT)
Unclassifiedf	NA	NA	NA	Postcentral gyrus (4 ECT, 2 rTMS, 1 tVNS, 1 group CBT, 1 taVNS); cerebellum (6 ECT, 1 tVNS, 1 MECT, 1 taVNS); lingual gyrus (3 ECT, 2 rTMS, 1 tVNS, 1 group CBT, 1 taVNS); fusiform gyrus (5 ECT, 1 MECT, 1 rTMS); inferior temporal gyrus (4 ECT, 1 rTMS); parahippocampal gyrus (7 ECT, 1 rTMS); posterior cingulate gyrus (1 ECT, 1 MECT, 1 CBT, 1 taVNS); ventrolateral PFC (3 ECT, 1 tVNS); nucleus accumbens (2 tVNS, 2 rTMS, 1 iCBT or MACe, 1 ECT); putamen (1 iCBT or MACf, 1 CBT, 1 group CBT, 1 CBT or fMRI neurofeedbacke, 1 ECT, 1 rTMS); medial temporal lobe (2 ECT, 1 CBT)

Regions with changes detected by ≥3 studies are reported in the table (17,21, 22, 23, 24, 25, 26, 27, 28,31,34, 35, 36,38, 39, 40, 41, 42, 43, 44, 45,47,49, 50, 51, 52, 53, 54, 55,57, 58, 59, 60, 61, 62, 63, 64,66, 67, 68, 69,71, 72, 73, 74,77, 78, 79, 80, 81, 82, 83,85,87, 88, 89, 90,92,93,100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 95, 96, 97, 98, 99,118,119).

ACC, anterior cingulate cortex; CBT, cognitive behavior therapy; ECT, electroconvulsive therapy; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; iCBT, internet CBT; iTBS, intermittent theta burst stimulation; MAC, monitored attention control; MECT, modified ECT; MST, magnetic seizure therapy; NA, not applicable; PFC, prefrontal cortex; rTMS, repetitive transcranial magnetic stimulation; SLR, systematic literature review; taVNS, transcutaneous auricular VNS; tDCS, transcranial direct current stimulation; tVNS, transcutaneous VNS; VNS, vagus nerve stimulation.

^aNetworks as described by Uddin et al. 2019 (120).

^bAdditional networks falling outside this framework that were each mentioned once by studies in the SLR were the affective network (30,89), auditory network (26), emotion regulation network (109), functional network (69), left frontoparietal network (26), medial visual network (26), occipital pole network (26), and structural covariance network (64).

^cRegion not mentioned in Uddin et al. (120) but mentioned in Williams (155).

^dMultiple names are used for the control network. For this SLR, we classified the central executive network, executive control network, and frontoparietal network as control networks.

^eIn some studies, participants could be randomly assigned to 2 different treatments, but the study did not report which treatment led to the effect reported here.

^fMany brain regions are not currently classified under the 6-network taxonomy proposed by Uddin et al. (120).

Eighty-six studies reported changes in specific brain regions. Brain regions were presented at varying degrees of granularity in the studies (e.g., frontal lobe vs. PFC vs. dorsolateral PFC). To contextualize our findings, we summarized the brain regions based on their associated cognitive networks, using the 6-network taxonomy proposed by Uddin et al. (120) (Table 5). The 2 brain regions most commonly found to experience changes after nonpharmacological treatment were both in the control network: the ACC and the dorsolateral PFC. Alterations in the ACC were reported in 23 studies (7 used rTMS, 7 used ECT, 3 used transcutaneous vagus nerve stimulation [tVNS], 2 used group CBT, 1 used internet CBT or monitored attention control, 1 used transcranial direct current stimulation [tDCS], 1 used modified ECT, 1 used iTBS), and 19 studies reported alterations in the dorsolateral PFC (9 used rTMS, 4 used ECT, 2 used TMS–electroencephalography (EEG), 1 used MST, 1 used MST or ECT, 1 used tVNS, 1 used tDCS) (Table 5). Changes in many regions of the default mode network were also reported, including the hippocampus (12 studies); posterior cingulate cortex, middle frontal gyrus, and angular gyrus (11 studies each); and superior frontal gyrus (10 studies) (Table 5).

Changes in MDD Symptomology and Neurocognition After Nonpharmacological Treatment

Many studies used multiple tools for assessing MDD symptomology; the HAMD (81/101, 80.2%) and MADRS (16/101, 15.8%) were used most often. Of the 84 studies that investigated whether there was a change in MDD symptomology after nonpharmacological treatment, 94.0% (79/84) found a significant improvement in symptoms (Figure 2D). An investigation of whether the treatment was associated with any change in neurocognition was conducted in 15.8% (16/101) of the studies, primarily those that used ECT, which is known to induce memory impairment (100). The impact of ECT on neurocognition is an important treatment consideration but is outside the scope of this SLR. However, the recent review and meta-analysis by Guo et al. (121) reports on how ECT affects neurocognition.

Correlation Between Brain Changes and Improvement in MDD

Of the 74 studies that investigated whether there was a relationship between brain changes and MDD improvement, 71.6% (53/74) found that these changes corresponded with symptom improvement. Some studies found that increases in connectivity and activity in certain brain regions were associated with symptom improvement, including the following: increased functional connectivity between the subgenual ACC and right middle temporal pole correlated with HAMD score improvement after ECT (63); increased functional connectivity between the left amygdala and left default mode network correlated with MADRS score improvement after a form of rTMS called Stanford Neuromodulation Therapy (25); increased gray matter volume in the left medial temporal lobe correlated with HAMD score improvement after ECT (28); and increased functional connectivity between the left amygdala and left insula correlated with HAMD score improvement after rTMS (30). Some researchers have identified that connectivity and activity decreases in other brain regions were associated with symptom improvement. For example, decreased functional connectivity between the ACC and medial PFC correlated with BDI score improvement after group CBT (17); decreased effective connectivity between the dorsolateral PFC and right angular gyrus correlated with HAMD score improvement after ECT (100); decreased connectivity between the medial hypothalamus and rostral ACC correlated with HAMD score improvement after tVNS (93); and a reduction in prefrontal alpha power correlated with HAMD score improvement after MST (47).

Brain Differences Between Treatment Responders and Nonresponders

Fourteen studies compared treatment responders with nonresponders; of these, 85.7% (12/14) found neuroplasticity differences between responders and nonresponders (Figure 2E). For example, functional connectivity between the subgenual ACC and medial orbitofrontal cortex increased after iTBS only in treatment responders (23).

Subgroup Assessment of Brain Changes and MDD Symptomology Posttreatment

Most included studies used neuromodulatory treatments (91/101). When posttreatment results were compared to baseline of the studies that used neuromodulatory treatments, 95.6% (87/91) found significant changes in the brain, 93.5% (72/77) found significant improvement in MDD symptomology, and 72.3% (47/65) found a correlation between MDD symptomology and brain changes. When posttreatment results were compared to the baseline of the studies that used non-neuromodulatory treatments, 80.0% (8/10) found significant changes in the brain, 100.0% (7/7) found significant improvement in MDD symptomology, and 66.7% (6/9) found a correlation between MDD symptomology and brain changes.

When considering how changes in neuroplasticity were assessed based on brain state, 72.3% (73/101) assessed participants in a resting state, and 13.9% (14/101) used task-based assessments; few studies (n = 3) used both states. Of the remaining 17 studies, 13 examined structural changes, and 4 assessed brain state while treatment was ongoing.

Comparing posttreatment results with baseline of the studies that assessed brain changes while in resting state, 94.5% (69/73) found significant changes in the brain, 93.5% (58/62) found significant improvement in MDD symptomology, and 73.2% (41/56) found a correlation between MDD symptomology and brain changes. Comparing posttreatment results with baseline of the studies that assessed brain changes in a task-based state, 78.6% (11/14) found significant changes in the brain, 100.0% (11/11) found significant improvement in MDD symptomology, and 66.7% (6/9) found a correlation between MDD symptomology and brain changes. There were not enough studies that used a task-based state to assess results based on the testing paradigm.

Three studies assessed changes in neuroplasticity in both resting and task-based states (48,70,115). One study used rTMS and found that after treatment, there were statistically significant brain changes in both resting and task states, improved MDD symptomology, and correlations between task-based brain changes and symptom improvement (115). Another study used the Emotional Faces Memory Task (EFMT) and found improved MDD symptomology posttreatment; however, the brain changes observed for either brain state were not statistically significant. There was a trend-level association between task-based changes in effective connectivity from the dorsal ACC to the amygdala and improvement in MDD symptomology (48). The final study assessed tDCS with the EFMT and found improved MDD symptomology after treatment. However, brain changes were not statistically significant; this study did not assess correlations between brain changes and symptomology improvement (70).

We also looked at how neuroplasticity was measured. Most studies assessed neuroplasticity changes through fMRI (55/101, 54.5%) or EEG (21/101, 20.8%). Other mechanisms of assessment included structural MRI (n = 13); diffusion tensor imaging (n = 5); motor evoked potential amplitude (n = 4); near-infrared spectroscopy and magnetoencephalography (both n = 2); and functional near-infrared spectroscopy, magnetic resonance spectroscopy, positron emission tomography, and single-photon emission computed tomography (all n = 1). Five studies used multiple methods to assess changes in neuroplasticity.

Comparing posttreatment results with baseline of the studies that used fMRI, 94.5% (52/55) found significant changes in the brain, 97.9% (46/47) found significant improvement in MDD symptomology, and 74.4% (32/43) found a correlation between MDD symptomology and brain changes. Comparing posttreatment results with baseline of the studies that used EEG treatments, 95.2% (20/21) found significant changes in the brain, 88.2% (15/17) found significant improvement in MDD symptomology, and 76.9% (10/13) found a correlation between MDD symptomology and brain changes.

Discussion

Overview

Both pharmacological and nonpharmacological treatments for MDD have been shown to elicit structural and functional changes in the brain (11). Neuromodulatory nonpharmacological tools aimed at improving neuroplasticity by exciting or inhibiting neurons, in particular brain circuits, include ECT, MST, and TMS (including iTBS, a type of TMS approved by the U.S. Food and Drug Administration in 2018 for MDD) (11). Additionally, changes in neuroplasticity induced by non-neuromodulatory interventions, such as CBT and fMRI neurofeedback, have been correlated with depression symptom improvement (11,122). To synthesize the published data on how nonpharmacological treatments for MDD change brain structure and function, we conducted an SLR of 101 studies of 9 nonpharmacological interventions. Most of the studies found statistically significant changes in the brain following nonpharmacological treatment for MDD, and 53 of the 74 studies that investigated the relationship between these brain changes and improvement in MDD symptoms found a correlation.

To more directly compare outcomes across studies with many different types of treatment, we also considered subgroups of studies based on the treatment MoA, the brain state used for assessment, and the way neuroplasticity was measured. Most studies used neuromodulatory treatments, examined the brain in resting state, and measured neuroplasticity using fMRI. Most studies found statistically significant changes in the brain and significant improvement in MDD symptomology after treatment, regardless of the treatment or measurement type. Correlations between brain changes and changes in MDD symptomology were lower but proportionally similar for these 2 subgroups.

The biggest difference in subgroups was seen when comparing resting and task-based assessments. While most studies that took resting-state measurements followed the overall pattern of results, a smaller proportion of studies that used a task-based state for assessments found statistically significant changes in the brain posttreatment and a correlation between those changes and change in MDD symptomology. As only 14 studies used task-based assessments, it is difficult to draw strong conclusions, but this may indicate that the treatment effects on neuroplasticity changes engaged other regions or circuits that were not being measured or that treatment effects may take longer to appear for task-based measurements. As task-based measurements can in part indicate how a person is functioning in their daily life, this indicates the need for a better understanding of how treatments for MDD affect the brain when performing different functions.

Effect of Nonpharmacological Treatments on the Brain

The most commonly used measures of neuroplasticity in the studies included in this SLR were EEG and fMRI, both of which can only measure neuroplasticity indirectly through related mechanisms but may also capture activity that is unrelated to neuroplasticity (12). The difficulty in elucidating the exact neurophysiological effects of treatment is supported by research on ADT. A meta-analysis of the effects of ADT on the brain found that the dorsolateral PFC and subgenual ACC were important target regions based on imaging analysis; however, target neurotransmitters were not concentrated in these areas, showing that more research is needed to understand the relationship between target regions and the MoA of individual treatments (123). A study that compared ADT to psychotherapy found convergent effects on the salience network; however, the underlying neural changes seen when comparing ADT and psychotherapy did not overlap (124). More research is needed on all treatment types to better understand the precise mechanisms of neuroplasticity effects. In this SLR, we used baseline and healthy control groups as a proxy to identify areas of treatment-related changes in neuroplasticity.

The ACC and dorsolateral PFC have been implicated in MDD (11,125), and they both featured prominently in the results of this SLR. Decreased volume and activity of the ACC can be seen with MDD (13). We found 6 studies with ACC differences at baseline between healthy control participants and participants with MDD, such as higher amplitude of low-frequency fluctuation (ALFF) values (45) and altered functional connectivity with other regions (17,23,104,115,116). Twenty-three studies using 7 treatment types (ECT, rTMS, tVNS, tDCS, group CBT, and internet CBT or monitored attention control) found changes in the ACC after treatment. Alterations in the dorsolateral PFC, such as synaptic loss and lower volume, may be associated with the development of MDD (11). Two studies in this SLR reported baseline differences in the dorsolateral PFC between participants with MDD and healthy control participants, such as hypoactivity and lower ALFF levels (42,45). Six different nonpharmacological interventions (rTMS, ECT, MST, tVNS, tDCS, and TMS-EEG) were reported to alter the dorsolateral PFC in this SLR. It was reported in multiple articles that baseline differences between healthy control participants and participants with MDD were ameliorated after nonpharmacological treatment, such as by increasing functional connectivity between the dorsolateral PFC and the right ventral anterior insula after ECT (100).

The Need for Research on Emerging Nonpharmacological Interventions

While most of the studies in this SLR used ECT and rTMS, several studies covered newer treatments such as MST and EFMT. MST has recently emerged as a treatment with comparable efficacy to ECT but fewer cognitive side effects (126,127). Only 3 studies in this SLR used MST, but all showed that it is an effective treatment for MDD (22,47,87) and a promising treatment to investigate further. Two studies used the EFMT (48,70), an n-back working memory task with faces showing different emotions, thought to improve cognitive control over emotional information processing in MDD (48). Both studies found a significant decrease in MDD symptoms posttreatment, and one found a nonsignificant trend associating MDD symptom change with the amygdala’s functional connectivity with the dorsolateral PFC and dorsal ACC. Both studies were small (≤20 participants), and more research is needed. Another emerging class of nonpharmacological treatment is digital therapeutics (DTx)—health software that delivers clinically validated medical interventions to treat or alleviate medical conditions (128,129). DTx can be used remotely and may have fewer side effects than ADTs (130,131). None of the articles that we screened for this SLR tracked brain changes with the use of DTx for MDD, highlighting a crucial gap in the literature.

Limitations and Strengths

This study has several limitations. With a large number of nonpharmacological treatments for MDD available (132), it was not feasible to include them all. However, other treatments have been covered in recent systematic reviews, such as exercise (133) and yoga (134). Among the studies included here, we found that study design and reporting were inconsistent and not always consistent with best practices, which affect the strength of our findings. Most of the study designs were not rigorous and neglected to include an intervention control group, making it difficult to understand the cause of changes in neuroplasticity. An additional consideration is that research has shown that placebo treatments can lead to changes in neuroplasticity in MDD (135); however, evaluating a placebo effect in this SLR was beyond our scope.

Most studies did not consider social determinants of health or report race or ethnicity data. Even after the 2017 requirement to submit race and ethnicity data to the clinical trial database registry, less than two-thirds of publications include these data (136,137). The studies also did not consider sex differences in the brain, although it is well established that such differences exist both in the neurobiology and in the response to nonpharmacological treatment for MDD (138, 139, 140).

Another consideration for study design was how MDD was defined and diagnosed. The included studies were inconsistent; in addition to multiple available guidelines for determining a diagnosis, some studies used a scale score (e.g., HAMD), clinician assessment, participant report, or did not report how they defined MDD. Therefore, overlaps between definitions of MDD and treatment-resistant depression occurred, and we were not able to exclude the latter. Additionally, it was difficult to exclude late-life depression. Although we excluded references that explicitly focused on late-life depression or older-age populations, many studies had a wide age range with an upper age limit >65 years or did not include age in their reported inclusion and exclusion criteria. As the etiology and presentation of MDD can change with age, this might have affected our findings (141).

Despite these limitations, this SLR has many strengths. It synthesized information from a relatively large number of articles and participants, showing that nonpharmacological interventions change regions of the brain known to be altered in MDD and that these changes correlate with improvements in depression symptoms. This result underscores the importance of nonpharmacological interventions as therapeutic options for patients with MDD.