Table 1 Summary of findings on longitudinal studies of pharmacotherapy, psychotherapy, electroconvulsive therapy (ECT), and repetitive transcranial magnetic stimulation (rTMS).
From: Brain connectivity in major depressive disorder: a precision component of treatment modalities?
Treatment method | Connectivity measure | Reference | Sample size | Treatment protocol | Analysis method | Main findings |
|---|---|---|---|---|---|---|
Pharmacotherapy | Functional | Ang et al. [14] | 241 MDD, 38 HC | -Sertraline (SSRI) or Bupropion (norepinephrine-dopamine reuptake inhibitor (NDRI)) -8 weeks -Placebo-controlled -Randomized -Double-blind | -Seed-based: NAcc -Baseline MRI | Higher baseline FC between bilateral NAcc and rACC was associated with a positive response to Bupropion but not Sertraline. |
Braund et al. [18] | 226 MDD, 68 HC | -Escitalopram (SSRI), Sertraline (SSRI), or Venlafaxine-XR (SNRI) -8 weeks -Randomized | -Network-based statistic (NBS) with 436 ROIs -MRI before and after treatment | Higher baseline neuroticism was associated with greater FC within and between the SN, FPN, and SMN, which predicted poorer treatment outcomes. FC strength in these networks decreased after treatment. | ||
Chin Fatt et al. [9] | 279 MDD, 38 HC | -Sertraline (SSRI) -8 weeks -Placebo-controlled -Randomized -Double-blind | -Whole-brain parcellation: FPN, DMN, SN, DAN, SN, SMN, hippocampus, thalamus -Baseline MRI | Higher FC within the DMN, and between the DMN and the FPN predicted better outcomes. | ||
Cui et al. [21] | 36 MDD, 61 HC | -Escitalopram (SSRI) -12 weeks | -24 ROIs in the DMN -MRI before and after treatment | Reduced FC in the DMN core (amPFC, PCC) subsystem at baseline was increased after treatment and became comparable with that in the HC. | ||
Pharmacotherapy (cont.) | Functional (cont.) | DeMaster et al. [11] | 34 MDD | -Escitalopram (SSRI) -6 weeks | -EC: Bayesian multi-subject vector autoregressive model -Baseline MRI | Higher FC strength in DMN, FPN, and SN was related to baseline depression severity and response to treatment. |
Fischer et al. [25] | 128 MDD, 58 HC | -Sertraline (SSRI), Venlafaxine-XR (SNRI), or Escitalopram (SSRI) - 8 weeks -Randomized | -Seed-based: NAcc -MRI before and after treatment | Responders had an increase in NAcc-dACC FC, which was associated with an improvement in quality of life, and a decrease in NAcc-inferior parietal lobule (IPL) FC. | ||
Fu et al. [17] | 32 MDD, 25 HC | -Duloxetine (SNRI) -12 weeks | -ICA: DMN -MRI at weeks 0, 1, 8 and 12 | Reduced baseline FC within the orbitofrontal component of the DMN predicted clinical response. There was increased FC in the anterior DMN after treatment. | ||
Goldstein-Piekarski et al. [7] | 75 MDD, 31 HC | -Escitalopram (SSRI), Sertraline (SSRI), or Venlafaxine-XR (SNRI) -8 weeks -Randomized | -Seed-based: PCC -MRI before and after treatment | Remitters were distinguished from non-remitters by intact FC between the PCC and ACC/mPFC, similar to HC, while non-remitters had hypoconnectivity. | ||
Hsu et al. [16] | 22 MDD, 35 HC | -Sertraline (SSRI) -6 weeks | -ICA: DMN, SN, FPN -MRI before and after treatment | Thalamo-PFC FC provided moderate predictive power for the effectiveness of treatment. FC of the medial temporal lobe of DMN, FPN, thalamus, and SN was restored to HC levels after treatment. | ||
Ju et al. [20] | 192 MDD | -SSRIs, SNRIs, and other antidepressants -6 months | -Whole-brain parcellation with 268 ROIs -Baseline MRI | Connectome-based predictive modeling could predict symptom improvement at the 2-week, 1-, 2-, and 3-month time points after treatment. | ||
Pharmacotherapy (cont.) | Functional (cont.) | Ju et al. [24] | 110 MDD, 136 HC | -Antidepressant treatment based on clinical judgment, mostly with Paroxetine (SSRI) -6 months | -Network construction by 368 ROIs -MRI at baseline and every 6 months over 2 years | Only the DMN FC change distinguished non-remitters from the remitters at 6 months, and recurring from stable MDD during the 2-year follow-up. |
Korgaonkar et al. [10] | 163 MDD, 62 HC | -Escitalopram (SSRI), Sertraline (SSRI) or Venlafaxine-XR (SNRI) -8 weeks -Randomized | -NBS with 333 ROIs -MRI before and after treatment | Remitters were distinguished from non-remitters by higher FC within the DMN. Hypoconnectivity of non-remitters distinguished them from HC at baseline and increased after treatment. FC of remitters was higher than HC at baseline and also following remission. | ||
Liu et al. [26] | 56 MDD, 111 HC | -Paroxetine (SSRI) -6 months | -NBS -MRI before and after treatment | Impaired FC in SMN, DMN, and DAN was associated with the number of episodes and total illness duration. The disrupted FC in MDD did not change significantly after treatment. | ||
Martens et al. [13] | 34 MDD, 31 HC | -Escitalopram (SSRI) -6 weeks | -ICA: 16 networks of interest -Baseline MRI | Treatment response was associated with increased FC of the right FPN with the posterior DMN, SMN, and somatosensory association cortex (SAC). | ||
van der Wijk et al. [19] | 129 MDD, 99 HC | -Escitalopram (SSRI), and add-on Aripiprazole (atypical antipsychotic) for non-responders -16 weeks | -Seed-based: ACC, PCC, insula, dlPFC -Baseline MRI | Baseline FC of the ACC, PCC, and insula differentiated early-, late-, and non-remitters. | ||
Pharmacotherapy (cont.) | Functional (cont.) | Wu et al. [12] | 81 MDD | -Escitalopram (SSRI) -12 weeks | -Seed-based: 36 seeds from emotion regulation networks -Baseline MRI | Patients with remission and non-remission were separated with an accuracy of 82%. The FC between the left medial SFG, right IFG, and PCu were the features with the highest discrimination ability. |
Xiao et al. [15] | 36 MDD | -Escitalopram (SSRI) -2 weeks | -Seed-based: hippocampus -Baseline MRI | The early improved patients had higher FC between the left hippocampus and left IFG and PCu than the non-improved patients, which was positively correlated with the reduction of depressive symptoms. | ||
Yang et al. [22] | 20 MDD, 25 HC | -Sertraline (SSRI) -8 weeks | -Seed-based: PCC -MRI before and after treatment | FC of the PCC positive and PCC negative networks was lower in MDD. Treatment increased the FC in the PCC positive network, with a correlation with depression scores. | ||
Ye et al. [8] | 66 MDD, 57 HC | -Escitalopram (SSRI) or Venlafaxine (SNRI) -4 weeks | -ICA: DMN -Baseline MRI | The remitted group had higher intra-FC in the right angular gyrus, which was positively correlated with the reduction of depressive symptoms. | ||
Zhang et al. [23] | 59 MDD, 59 HC | -Escitalopram (SSRI), Sertraline (SSRI), or Fluoxetine (SSRI) -12 weeks | -Seed-based: ACC -MRI before and after treatment | Only a subregion of dACC and rACC had increased FC with SN after treatment. | ||
Structural | Davis et al. [32] | 200 MDD, 112 HC | -Escitalopram (SSRI) -8 weeks | -FA, MD, AD, and RD for 40 ROIs -MRI at baseline, 2 and 8 weeks after treatment | Baseline AD in the external capsule, which overlaps the SLF, was associated with treatment response. Further baseline differences of responders compared with non-responders were in the cingulum, sagittal stratum, and corona radiata. No changes in SC were observed after the treatment. | |
Dong et al. [34] | 62 MDD, 118 HC | -SSRIs or SNRIs -6 months | -FA -MRI before and after treatment | Decreased FA in the left insula, left middle occipital gyrus, right thalamus, left pallidum, and left PCu were observed in MDD compared with HC. No significant change in FA was observed after the treatment. | ||
Pharmacotherapy (cont.) | Structural (cont.) | Fan et al. [35] | 59 MDD, 118 HC | -Paroxetine (SSRI) -6 months | -FA -MRI before and after treatment | Decreased SC between the right ATC and the posterior temporal cortex (PTC), and between the left temporal cortex and the auditory cortex were reversed at remission. |
Grieve et al. [27] | -74 MDD to calculate FA measures -83 MDD for replication | -Escitalopram (SSRI), Sertraline (SSRI), or Venlafaxine-XR (SNRI) -8 weeks | -FA of stria terminalis and cingulate bundle | The ratio of the FA of the stria terminalis over the cingulate identified non-remitters with an accuracy of 83-88%, with greater specificity for Escitalopram and Sertraline. | ||
Korgaonkar et al. [29] | 157 MDD | -Escitalopram (SSRI), Sertraline (SSRI) or Venlafaxine-XR (SNRI) -8 weeks -Randomized | -FA for 46 tracts -Baseline MRI | FA in the left cingulum bundle, right SFOF, and right SLF identified 15% of the non-remitters with 84% accuracy. | ||
Korgaonkar et al. [28] | 74 MDD, 34 HC | -Escitalopram (SSRI), Sertraline (SSRI) or Venlafaxine-XR (SNRI) -8 weeks -Randomized | -FA for cingulate, fornix, stria terminalis and uncinate fasciculus -Baseline MRI | Altered SC of the cingulate and stria terminalis predicted remission with 62% accuracy. Prediction improved to 74% when age was added to the model. | ||
Pillai et al. [31] | 144 MDD | -Sertraline (SSRI) -Placebo-controlled | -FA between raphe nucleus, amygdala, and hippocampus -Baseline MRI | FA was lower in remitters than in non-remitters in raphe nucleus-amygdala tracts, which correlated with depressive scores. | ||
Vieira et al. [33] | 20 MDD | -Paroxetine (SSRI) -6-12 weeks | -FA, MD, AD, RD -Baseline MRI | There was an increase in FA and a decrease in RD in forceps minor and SLF in responders compared to non-responders at baseline. | ||
Pharmacotherapy (cont.) | Structural (cont.) | Zhou et al. [30] | 40 MDD | -SSRIs -8 weeks | -FA -Baseline MRI | Reductions in FA were found in the hippocampus in treatment-resistant depression compared to MDD. |
Pharmacotherapy or psychotherapy | Functional | Dunlop et al. [38] | 122 MDD | -CBT (16 sessions) or pharmacotherapy (Escitalopram or Duloxetine, double-blind) -Randomized -12 weeks | -Seed-based: sgACC -Baseline MRI | FC of the left vlPFC, the dorsal midbrain, and the left vmPFC with the sgACC was differentially associated with remission to CBT and pharmacotherapy. Positive FC was associated with remission with CBT and treatment failure with pharmacotherapy, whereas negative FC was associated with remission to pharmacotherapy and treatment failure with CBT. |
Psychotherapy | Functional | Crowther et al. [36] | 23 MDD, 20 HC | -Behavioral activation treatment -12 sessions | -Seed-based: AI, dACC, superior parietal lobule (SPL), dlPFC, PCu, mPFC -Baseline MRI | Clinical response was predicted by baseline FC of the right AI with the right MTG, and the left intraparietal sulcus with the OFC. |
Späti et al. [37] | 21 MDD, 35 HC | -CBT -22 weeks | -Seed-based: medial frontal gyrus (MFG) -MRI at baseline | Increased FC of PFC with the dACC was related to higher levels of adaptive rumination and better response to treatment. | ||
Structural | Wang et al. [39] | 21 MDD, 22 HC | -Guided imagery psychotherapy -4 weeks | -FA -MRI before and after treatment | MDD had increased FA in the right thalamus compared to HC. SC changes did not recover after treatment, but a novel region of increased FA was found in the supplementary motor area. | |
ECT | Functional | Leaver et al. [46] | 33 MDD, 33 HC | -Right unilateral or a combination of right unilateral and bifrontal ECT -11.8(±3.3) sessions -2–4 weeks | -ICA: DMN, SN, thalamus/basal ganglia network -MRI at baseline and after treatment | VS-ventral DMN hyperconnectivity was reduced, while VS-anterior DMN hypoconnectivity only modestly improved after treatment. FC between the SN and dmPFC was reduced in MDD, but did not change after treatment. |
ECT (cont.) | Functional (cont.) | Mo et al. [44] | 28 MDD, 20 HC | -Modified bifrontal ECT | -Seed-based: Left angular gyrus -MRI at baseline | FC of the left angular gyrus with bilateral inferior temporal gyrus, bilateral MFG, left SFG, left MTG, left PCu, left PCC, and right angular gyrus was strengthened after treatment. |
Pang et al. [41] | 33 MDD | -Modified bifrontal ECT -7.49 (±1.92) sessions | -Seed-based: mPFC, PCu, angular gyrus, parahippocampus, MTG, vlPFC, dlPFC, precentral gyrus, cerebellum -MRI before and after treatment | Baseline FC within the DMN, and between the DMN and FPN could predict clinical improvement. FC within the DMN and between DMN and FPN increased after treatment, and the changed FC between the dmPFC and vlPFC was negatively correlated with clinical improvement. | ||
van Waarde et al. [40] | 45 MDD | -Right unilateral or bilateral ECT -18.7(±7.1) sessions | -ICA -Baseline MRI | Networks centered in the dmPFC and the ACC had a sensitivity of 80–84% and a specificity of 75–85% for predicting recovery. | ||
Wang et al. [45] | 23 MDD, 25 HC | -Modified bifrontal ECT -three times a week -2–3 weeks | -Seed-Based amygdala -GCA: the amygdala and fusiform face area (FFA) -MRI before and after treatment | There was increased FC between left amygdala and left FFA, and EC from FFA to the amygdala after treatment. | ||
Wei et al. [43] | 26 MDD, 20 HC | -Modified bifrontal ECT -3 times a week until remission | -Whole-brain connectivity matrix -MRI before and after treatment | FC strength of the left angular gyrus increased after treatment. | ||
Zhang et al. [47] | 46 MDD, 33 HC | -Modified bifrontal ECT -6–12 sessions | -Seed-based: dmPFC, dlPFC, OFC -MRI at baseline and after treatment | There was a correlation between the changed dmPFC-dlPFC FC after treatment and the change in anhedonia. | ||
ECT (cont.) | Structural | Lyden et al. [49] | 20 MDD, 28 HC | -Right unilateral, or right unilateral and bitemporal ECT | -FA, RD, AD, MD -MRI before the first ECT, after the second ECT, and after completion of the index ECT series | There was increased FA in the anterior cingulum, forceps minor, and left SLF after treatment. There were decreases in RD and MD in overlapping regions and the ATR. Changed SC in pathways connecting frontal and limbic areas was related to therapeutic response. |
Repple et al. [48] | 98 MDD, 52 HC | -ECT -13.86(±3.53 sessions) | -FA, MD, RD, AD -MRI before and after treatment | Baseline FA was positively and MD negatively correlated with symptoms after ECT, mainly in CC, internal capsule, and corona radiata. MD increased after ECT in the right uncinate fasciculus, posterior limb of the internal capsule, ILF, and IFOF. | ||
rTMS | Functional | Baeken et al. [50] | 20 MDD | -20 Hz rTMS -20 sessions -To the left dlPFC -Sham-controlled | -Seed-based: sgACC -MRI before and after treatment | Responders had stronger negative FC between the sgACC and left dmPFC. After successful treatment, there was stronger positive FC in these regions. |
Baeken et al. [59] | 44 MDD, 44 HC | -Accelerated intermittent theta burst stimulation (aiTBS) -To the left dlPFC -20 sessions -Sham-controlled | -Seed-based: sgACC -MRI before and after treatment | A positive sgACC FC with the mOFC could distinguish responders from non-responders at baseline. Treatment increased sgACC–mOFC FC, which was associated with a decrease in feelings of hopelessness. | ||
Chen et al. [64] | 20 MDD, 20 HC | -10 Hz rTMS -To the left dlPFC -25 sessions -4 weeks -Sham-controlled | -Seed-based: amygdala -MRI before and after treatment | There was an increase in FC in the left insula, right IFG, right SFG, and right IPL; and a reduction in FC in the PCu after treatment. Change in FC between the left insula and left amygdala was positively correlated with the change in depressive scores. | ||
rTMS (cont.) | Functional (cont.) | Du et al. [60] | 22 MDD | -High frequency (HF) rTMS -To the left dlPFC -10 sessions -2 weeks | Seed-based: Left dlPFC stimulation target, left NAcc -Baseline MRI | The early improvers had increased negative FC between the stimulated dlPFC and left NAcc compared to non-improvers. The stimulated dlPFC–NAcc FC negatively correlated with improved depressive symptoms. |
Eshel et al. [65] | 36 MDD, 28 HC | -10 Hz rTMS -To the left dlPFC -20 sessions -Randomized -Sham-controlled -Double-blind | -Seed-based: Left dlPFC stimulation target -MRI before and after treatment | There was an impaired inhibitory effect of the dlPFC on the amygdala in MDD compared to HC. rTMS increased dlPFC global FC and induced a negative dlPFC-amygdala FC. | ||
Ge et al. [58] | 32 MDD, 24 HC | -10 Hz rTMS or iTBS -To the left dlPFC -20-30 sessions | -Seed-based: sgACC, rACC -MRI before and 12 weeks after the treatment | Treatment response was associated with lower FC of sgACC to right dlPFC, and higher FC of rACC to left lateral parietal cortex at baseline. The hyperconnectivity between sgACC and the visual cortex was normalized to a level comparable to that of HC after treatment. | ||
Iwabuchi et al. [62] | 27 MDD | -10 Hz rTMS or iTBS -16 sessions -4 weeks -To the left dlPFC | -Seed-based: right AI -GCA: right AI and left dlPFC stimulation target -ICA: DMN, CEN, SN. -MRI before and after treatment | Baseline fronto-insular EC and SN FC were positively correlated with early (1 month) response to treatment but not sustained response (3 months). FC measures did not change significantly after treatment. | ||
Kang et al. [61] | 24 MDD | -10 Hz rTMS -To the left dlPFC -10 sessions -2 weeks -Sham-controlled | Seed-based: dlPFC -MRI before and after treatment | Reduced dlPFC-left caudate FC predicted clinical improvement. There was a reduction of the FC between these regions after treatment compared to sham. | ||
Hopman et al. [57] | 70 MDD | -HF rTMS -To the left dlPFC -20 sessions -4 weeks | -Seed-based: Left dlPFC, sgACC -Baseline MRI | dlPFC-sgACC FC was not associated with treatment outcome. Long-term non-responders had poorer FC between the sgACC-frontal pole, SPL, and occipital cortex, and dlPFC-central opercular cortex. | ||
Taylor et al. [63] | 32 MDD | -10 Hz rTMS -To the left dlPFC -20 sessions -Sham-controlled | -Seed-based: sgACC, amygdala, PCC, dlPFC stimulation site -MRI before and after treatment | Baseline FC between PCC and AI was lower in responders compared to non-responders. There was no significant effect of rTMS over sham on the depressive scores or on FC. sgACC FC to AN, DMN, and FPN decreased in responders, but not in non-responders. | ||
rTMS (cont.) | Functional (cont.) | Weigand et al. [52] | 41 MDD | -rTMS -To the left dlPFC -28.5(±3.4) sessions -4–7 weeks -Sham-controlled | Seed-based: left dlPFC stimulation target and sgACC -Baseline MRI | Treatment efficacy was predicted by left dlPFC stimulation sites that were more anterolateral and more negatively correlated with the sgACC. |
Structural | Klooster et al. [67] | 40 MDD | -aiTBS -To the left dlPFC -5 sessions -4 days -Sham-controlled | - FA, MD, tract density, tract volume, and number of tracts: between left dlPFC and ACC -Baseline MRI | SC between the patient-specific left dlPFC stimulation site and the PCC had the predictive potential for clinical response. | |
Ning et al. [66] | 21 MDD | -10 Hz rTMS -Left dlPFC -36 sessions | - FA, RD, AD: left PFC -MRI before and after treatment | Baseline SC of the dACC and vlPFC were correlated with changes in depressive scores. FA was increased, and RD was decreased in amPFC. SC changes in the vlPFC were correlated with treatment response. | ||
rTMS (cont.) | Functional and Structural | Chen et al. [69] | 33 MDD | -rTMS -To the left dlPFC -10 sessions | -Seed-based: PFC, motor, somatosensory, parietal, and temporal cortices, thalamus -FA: thalamocortical tracts -Baseline MRI | FC and SC between the thalamus (mediodorsal nucleus) and PFC (dlPFC, vlPFC, OFC) predicted treatment efficacy. |
Fu et al. [68] | 27 MDD | -rTMS -To the left dlPFC -10 sessions -2 weeks | Seed-based: left dlPFC and bilateral insula -FA: tracts between left dlPFC and bilateral insula -Baseline MRI | FC and SC between left dlPFC and insula had a positive correlation with clinical improvement. |
