Mechanisms of Antidepressant Response to Electroconvulsive Therapy Studied With Perfusion Magnetic Resonance Imaging

Accepted Manuscript Mechanisms of antidepressant response to ECT studied with perfusion MRI Amber M. Leaver, PhD, Megha Vasavada, PhD, Shantanu H. Joshi, PhD, Benjamin Wade, PhD, Roger P. Woods, MD, Randall Espinoza, MD MPH, Katherine L. Narr, PhD PII:

S0006-3223(18)31892-4

DOI:

10.1016/j.biopsych.2018.09.021

Reference:

BPS 13651

To appear in:

Biological Psychiatry

Received Date: 13 March 2018 Revised Date:

10 September 2018

Accepted Date: 23 September 2018

Please cite this article as: Leaver A.M., Vasavada M., Joshi S.H., Wade B., Woods R.P., Espinoza R. & Narr K.L., Mechanisms of antidepressant response to ECT studied with perfusion MRI, Biological Psychiatry (2018), doi: https://doi.org/10.1016/j.biopsych.2018.09.021. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Mechanisms of antidepressant response to ECT studied with perfusion MRI Short Title: Perfusion MRI and antidepressant mechanisms of ECT 1,3

1

1

1

Amber M. Leaver PhD *, Megha Vasavada PhD , Shantanu H. Joshi PhD , Benjamin Wade PhD , Roger P. Woods 1,2 2 1,2 MD , Randall Espinoza MD MPH , Katherine L. Narr PhD 1

RI PT

Ahmanson-Lovelace Brain Mapping Center, Department of Neurology, University of California Los Angeles, Los Angeles, CA, 90095 2 Department of Psychiatry and Biobehavioral Sciences, University of California Los Angeles, Los Angeles, CA, 90095 3 Department of Radiology, Northwestern University, Chicago, IL, 60611

*Corresponding Author:

M AN U

Address: 737 N Michigan Ave Suite 1600 Chicago, IL 60611 Phone 312 694 2966 Fax 310 926 5991 Email [email protected]

SC

Amber M. Leaver Ph.D.

Key words: electroconvulsive therapy, depression, seizure, fMRI, cerebral blood flow, hippocampus

EP

TE D

Word count, abstract: 250 Word count, main text: 3,997 Figures: 5 Tables: 2 Supplement: 1 document (Text, 1 Table, 1 Figure)

AC C

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

1

ACCEPTED MANUSCRIPT 1 2 3 4

ABSTRACT

5

severe depression, though how this relates to antidepressant response is less clear. Arterial spin-labeled (ASL) functional

6

MRI (fMRI) tracks absolute changes in cerebral blood flow (CBF) linked with brain function, and offers a potentially

7

powerful tool when observing neurofunctional plasticity with fMRI.

8

RI PT

INTRODUCTION: Converging evidence suggests electroconvulsive therapy (ECT) induces neuroplasticity in patients with

METHODS: Using ASL-fMRI, we measured global and regional CBF associated with clinically prescribed ECT and

10

therapeutic response in patients (n=57, 30 female) before ECT, after two treatments, after completing an ECT treatment

11

“index” (~4 weeks), and after long-term follow-up (6 months). Age- and sex-matched controls were also scanned twice

12

(n=36, 19 female), ~4 weeks apart.

SC

9

M AN U

13

RESULTS: Patients with lower baseline global CBF were more likely to respond to ECT. Regional CBF increased in the

15

right anterior hippocampus in all patients irrespective of clinical outcome, both after 2 treatments and after ECT index.

16

However, hippocampal CBF increases post-index were more pronounced in nonresponders. ECT responders exhibited

17

CBF increases in the dorsomedial thalamus and motor cortex near the vertex ECT electrode, as well as decreased CBF

18

within lateral fronto-parietal regions.

19

TE D

14

CONCLUSIONS: ECT induces functional neuroplasticity in the hippocampus, which could represent functional precursors

21

of ECT-induced increases in hippocampal volume reported previously. However, excessive functional neuroplasticity

22

within the hippocampus may not be conducive to positive clinical outcome. Instead, our results suggest that, although

23

hippocampal plasticity may contribute to antidepressant response in ECT, balanced plasticity in regions relevant to

24

seizure physiology including thalamo-cortical networks may also play a critical role.

AC C

25 26

EP

20

2

ACCEPTED MANUSCRIPT 1 2 3

INTRODUCTION

4

and response times (≤1 month) superior to other currently available treatments (1–3). With ECT, a controlled transient

5

seizure is elicited every 2-3 days over 2-4 weeks with anesthesia and muscle relaxants, and is sometimes followed by

6

maintenance treatments after the ECT index phase (4). Although generalized seizures are elicited, a growing body of

7

evidence suggests that ECT induces structural and functional neuroplasticity in specific brain structures and networks,

8

though how these changes relate to therapeutic response remains less clear (5–11).

RI PT

Electroconvulsive therapy (ECT) is an effective intervention for severe major depression, with response rates (50-80%)

9

An increasing number of neuroimaging studies have addressed the neural effects of ECT, implicating several regions

11

throughout the brain. ECT-related hippocampal volume enlargements are most replicated (10, 11), complementing reports

12

of decreased hippocampal volume in depression (12) and functional plasticity in hippocampal networks with ECT (13, 14).

13

Electroconvulsive shock, the animal model of ECT, is shown to promote neurotrophic effects including hippocampal

14

neurogenesis (15, 16), which may explain neuroplastic effects in humans. However, links between hippocampal plasticity

15

and symptom improvement in ECT are less replicated and a subject of debate (17, 18). Further, ECT-related changes in

16

several other brain regions have been identified, including the ACC (7–9, 19), thalamus (5, 20–22), basal ganglia (14, 23),

17

and amygdala (11, 24). A complete mechanistic understanding of ECT and its success as an antidepressant treatment

18

thus remains elusive.

19

TE D

M AN U

SC

10

A critical issue in ECT research is parsing the nonspecific effects of ECT from those effects related to (and perhaps

21

responsible for) antidepressant response to ECT. Although considered the “gold standard” treatment for intractable major

22

depression, not all patients respond to ECT; for example, just over half (55-65%) experience remission when using right-

23

unilateral ECT with optimal parameters (25, 26). Therefore, brain networks affected by ECT-induced seizures in all

24

patients may differ from, or only partially overlap, networks supporting improved depressive symptoms. Previously

25

reported ECT-related effects may thus not underlie clinical outcome, but instead reflect nonspecific physiological effects of

26

ECT.

AC C

27

EP

20

28

In the current study, we use arterial spin-labeled (ASL) perfusion fMRI to track absolute changes in cerebral blood flow

29

(CBF) at rest (i.e., “functional neuroplasticity”) associated with the lasting effects of ECT itself, as well as antidepressant

30

response to ECT. Application of ASL-fMRI offers a novel, complementary contribution to existing functional neuroimaging

31

literature with certain advantages. Most importantly, ASL-fMRI offers an absolute measure of brain function like positron 3

ACCEPTED MANUSCRIPT emission tomography (PET) and single-photon emission computed tomography (SPECT), but without the use of

2

radioactive tracers and with greater spatial resolution. Since ASL-fMRI offers an absolute metric of blood flow (i.e.,

3

mL/min/100g), it provides an important advantage over more widely used blood-oxygenation-level-dependent (BOLD)

4

fMRI, where relative metrics (i.e., arbitrary units) indirectly estimate brain function by determining: A) changes in BOLD

5

signal between active and rest or baseline conditions, or B) correlations in BOLD timecourses amongst brain regions to

6

infer “functional connectivity.” Thus, ASL-fMRI is particularly well-suited for multi-session longitudinal studies, where CBF

7

changes can be more directly interpreted as changes in brain function over time. To our knowledge, no previous ECT

8

studies have measured CBF changes with ASL-fMRI, and thus ours constitutes a novel contribution to the field.

RI PT

1

SC

9

Here, we measure global and regional CBF with ASL-fMRI in depressed patients followed prospectively while receiving

11

ECT, as well as non-depressed control volunteers not receiving treatment. We report CBF changes associated with ECT

12

and therapeutic response occurring after the initial 2 ECT sessions and after the ECT “index” treatment series. We

13

hypothesized that, while some CBF changes would occur in all patients regardless of ECT outcome, separable brain

14

regions would associate with positive antidepressant response. We also explored the extent to which acute and post-

15

index functional neuroplasticity is maintained 6-months after treatment. Finally, because CBF measured with ASL-fMRI

16

can be related to tissue content (27), we explored relationships, if any, between CBF and gray-matter (GM) plasticity.

MATERIALS AND METHODS Subjects

TE D

17 18 19 20 21 22

M AN U

10

Patients (n=57) and demographically similar non-depressed volunteers (n=36) gave informed written consent to participate in this UCLA IRB-approved study. All patients were diagnosed as experiencing a major depressive episode

24

(MDE; DSM-IV-TR(28)), and were treatment refractory (i.e., did not respond to 2+ prior antidepressant therapies). Patients

25

with comorbid psychiatric or neurological disorders, or concurrent serious illness were excluded. All patients ceased

26

psychotropic medications (antidepressants and benzodiazepines) at least 48-72 hours prior to and for the duration of the

27

index treatment, and had not received neuromodulation treatment within 6 months prior. For healthy volunteers, exclusion

28

criteria included any history of serious illness, neurological disorders, or psychiatric disorders (Mini-International

29

Neuropsychiatric Interview (29)).

AC C

EP

23

30 31

Depressive symptoms were assessed in patients using the Hamilton Depression Inventory (17-item) (30), Montgomery

32

Åsberg Depression Rating Scale (31), and Quick Inventory of Depressive Symptomatology (32). Scores on these

33

inventories were highly intercorrelated; therefore, clinical response was characterized using a composite score calculated 4

ACCEPTED MANUSCRIPT 1

as the mean proportional change between baseline and post-ECT-index scores (averaged across the three scales). ECT

2

responders were identified as having

3

publications including participants overlapping with the current cohort reported ECT-related structural (7, 11, 23, 34),

4

functional (14, 35, 36), and neurochemical (37, 38) changes, and evaluation of cognitive measures (39).

5 6 7 8

Study Visits and ECT

9

(Supplemental Methods); 20 were transitioned to bilateral electrode placement per clinical determination (Table 1).

10

Patients completed four MRI scans: 1) within 24 hours before first ECT session (baseline), 2) immediately before their

11

third ECT appointment (~4 days after baseline), 3) after their clinically determined ECT index series (~4 weeks after

12

baseline), and 4) approximately 6 months after ECT index. Non-depressed volunteers completed two MRIs approximately

13

4 weeks apart.

14 15 16 17

Image Acquisition and Preprocessing

18

Methods). ASL images were first corrected for motion (FSL; FMRIB, Functional MRI of the Brain), and then CBF was

19

quantified using the simple subtraction method in ASLtoolbox (27). CBF images were registered to T1-weighted

20

anatomical scans and MNI templates including interpolation to 2x2x2 mm resolution using SPM8 (Wellcome Trust Centre

21

for Neuroimaging) and smoothed (6mm FWHM Gaussian kernel, FSL). Images were then averaged to yield a single CBF

22

image per session for subsequent analysis. Voxelwise gray matter volume (GMV) was calculated in SPM8 during

23

standard MNI normalization processes. Upon visual inspection, 7 baseline (3 patients, 4 controls) and 13 follow-up (10

24

patients, 3 controls) ASL scans were identified to have poor quality (i.e., susceptibility artifacts, slice artifacts, and/or

25

global CBF <20 mL/100g/min) and were not analyzed further (Table 1). After preprocessing and normalization, ASL

26

images were masked using SPM8’s GM template (>20% probability GM classification), and to ensure a common field of

27

view (ASL pseudo-BOLD value > 100 (27) for all images). Because the ASL-fMRI field-of-view did not include the entire

28

cerebellum in all subjects, this mask did not include the cerebellum. Global CBF was calculated by averaging voxelwise

29

CBF within this mask for each ASL-CBF image. GMV was similarly calculated by averaging voxelwise GMV within the

30

mask for each GMV image.

31 32 33 34

Statistical Analyses

RI PT

≥50% reduction in composite depression scores after ECT index (33). Prior

M AN U

SC

Patients volunteered for this research study before initiating a clinically prescribed course of right-unilateral ECT

3

AC C

EP

TE D

ASL functional and T1-weight anatomical images were acquired using a 3T Siemens Allegra scanner (Supplemental

All statistical analyses were completed in R (https://www.r-project.org). See Supplemental Methods for additional details. 5

ACCEPTED MANUSCRIPT 1 Global CBF was analyzed with linear mixed-effects models, targeting a main effect of time (a fixed categorical factor

3

including all four timepoints) and ECT response (fixed factor) in depressed patients. Nuisance factors included age, ECT

4

lead placement (percentage right-unilateral), total number of treatments (fixed factors), and subject (random factor).

5

Pairwise comparisons were applied post hoc to assess significant change between timepoints in patients, as well as

6

between patients and controls (nuisance factors: age, global GMV).

7

RI PT

2

Regional (voxelwise) CBF was analyzed with two omnibus linear mixed-effects models, which specifically targeted

9

changes hypothesized to be most neurobiologically relevant to ECT and therapeutic response. These were: (1) changes

10

that occurred acutely (pre-treatment vs. after two ECT treatments), and (2) changes that occurred after treatment (pre-

11

treatment vs. after ECT index). Here, time was the fixed factor of interest, and nuisance factors included age, change in

12

depression score after ECT index, ECT lead placement, total number of treatments (fixed factors), and subject (random

13

factor). These models examined effects of ECT on brain function irrespective of antidepressant response; thus, change in

14

depression score post-index was a nuisance variable in this analysis. In these voxelwise analyses, a cluster-level

15

correction of p(corr) < 0.05 was applied using random field theory (40), with voxelwise threshold p < 0.01.

M AN U

SC

8

16

In clusters identified by voxelwise tests, post-hoc region of interest (ROI) analyses assessed whether the results of the

18

omnibus tests were influenced by tissue content (Supplemental Methods). These analyses used the same linear mixed

19

effects models described above, but included mean GMV (calculated using SPM8 as described above) for each ROI as

20

an additional fixed nuisance factor. Pairwise post-hoc ROI tests compared CBF between relevant timepoints; each

21

timepoint was compared to baseline and its immediate precursor using the same linear mixed effects models to yield five

22

distinct tests per ROI. The Bonferroni method corrected for the number of tests performed at each ROI. Post-hoc ROI

23

tests also compared GMV across timepoints, also using the same statistical models and Bonferroni-correction, but

24

including mean CBF for each ROI as a nuisance factor.

AC C

EP

TE D

17

25 26

Additional ROI analyses were applied to CBF data in healthy controls, for each region defined by the previously described

27

omnibus analyses. First, linear mixed-effects models measured possible CBF change in each region in healthy controls

28

(baseline vs. after 4 weeks), with age and GMV as fixed nuisance factors and subject as a random nuisance factor.

29

Second, CBF was compared between healthy controls and depressed patients, with age and GMV as fixed nuisance

30

factors. 6

ACCEPTED MANUSCRIPT 1 Finally, exploratory analyses attempted to identify regional (voxelwise) CBF changes associated with antidepressant

3

response. Here, we separated depressed patients into responders and nonresponders to ECT as described in the

4

Subjects section. In these analyses, we examined changes between baseline and post-index CBF, using the same

5

voxelwise and post-hoc ROI statistical tests described above.

6 7 8 9 10 11

RI PT

2

RESULTS Demographic and clinical variables

Depression scores improved significantly after ECT index (p < 0.00001 for all), with 22 of 42 of patients exhibiting at least a 50% reduction in depression scores and 16 patients meeting criteria for remission on any symptom inventory (Table 1;

13

see also Table S1). Patients and healthy controls did not differ in age or sex (p > 0.05 for both).

14 15 16 17

Global CBF

18

patients (p > 0.05, Figure 1). However, there was an overall effect of response, such that CBF was elevated in patients

19

who did not respond to ECT (p = 0.02x10-14).

20 21

In pairwise comparisons conducted post hoc separately for responders and nonresponders (Figure 1A), global CBF was

22

elevated in nonresponders at baseline compared to both responders (p = 0.004) and healthy controls (p = 0.027), while

23

baseline global CBF was not different between responders and controls (p = 0.541). Global CBF was also elevated in

24

nonresponders after ECT index (vs. controls p = 0.027; vs. responders p = 0.06) and at the 6-month follow-up visit at the

25

trend level (vs. controls p = 0.056 vs. responders p = 0.177). However, global CBF did not differ between responders and

26

nonresponders after 2 treatments (p = 0.181), indicating an acute increase in global CBF occurred in responders at this

27

timepoint (p = 0.003).

28 29

In analyses addressing the relationship between baseline global CBF and ECT response, change in depression scores

30

after ECT index correlated with baseline global CBF (i.e., patients with lower baseline CBF were more likely to improve,

31

Figure 1B).

32 33 34 35

Regional CBF: acute changes after 2 ECT sessions

36

identified in the dorsal caudate after two ECT treatments (Figure 2). This cluster did not appear to overlap the amygdala

37

(e.g., Figure 2A, bottom left panel), though previous studies demonstrated ECT effects in this region (24, 41, 42). These 7

M AN U

SC

12

AC C

EP

TE D

Global CBF did not change over time in patients or controls, nor was there a significant response-by-time interaction in

In all depressed patients, acute CBF increases were noted in the right anterior hippocampus, and CBF decreases were

ACCEPTED MANUSCRIPT acute effects persisted in post hoc ROI analyses when controlling for GMV. Additional pairwise comparisons between

2

other timepoints indicated that CBF remained significantly elevated in the right anterior hippocampus after ECT index and

3

6 months post-index compared to baseline measurements (Figure 2B). Notably, regional GMV in this anterior

4

hippocampal region increased after ECT index as indicated previously for this dataset (41) and others (13, 24, 43, 44);

5

however GMV did not significantly increase acutely (after 2 sessions Figure 2C, left panel). Post-hoc analyses also

6

indicated that acute CBF decreases in bilateral dorsal caudate returned to baseline levels after ECT index, and this

7

transient CBF decrease was accompanied by a transient increase in GMV. Coordinates for these and all other analyses

8

are displayed in Table 2.

RI PT

1

Regional CBF: changes after ECT index

13

posteriorly than the site of acute change described above. Post-hoc ROI analyses indicated CBF increases occurred both

14

acutely and post-index when controlling for GM content (Figure 3B), with corresponding GM increases occurring after

15

ECT index that were maintained for 6 months after index (Figure 3C).

SC

9 10 11 12

M AN U

After ECT index, CBF increased in the anterior portion of the right hippocampus (Figure 3A), extending slightly more

16

CBF increases were also observed in bilateral ventral basal ganglia, overlapping the ventral striatum and pallidum (Figure

18

3A). Here, post-hoc tests indicated that post-index CBF increases persisted when controlling for GMV (Figure 3B). GMV

19

did not change significantly in these regions over time (Figure 3C).

20 21 22 23

Exploring regional CBF with respect to antidepressant outcome

24

in the bilateral thalamus and left somatomotor cortex after ECT index, while decreases were observed in lateral parietal

25

and inferior frontal cortices, as well as the precuneus (Figure 4A). These effects persisted in post-hoc analyses

26

controlling for regional GMV. CBF did not change significantly in these regions in ECT-nonresponders or controls.

27

Notably, CBF in bilateral thalamus was lower in ECT-responders prior to treatment when compared to both

28

nonresponders and controls.

TE D

17

AC C

EP

In exploratory analyses targeting regional CBF separately in patients who responded to ECT, CBF increases were noted

29 30

By contrast, ECT nonresponders exhibited CBF increases after ECT index in a large cluster overlapping the right anterior

31

hippocampus and ventral striatum, along with CBF decreases in the posterior cingulate cortex and inferior precuneus

32

(Figure 4B). These effects were robust to tissue-content in post-hoc tests, and did not change in responders or controls

33

over time. 8

ACCEPTED MANUSCRIPT 1 2 3 4 5

DISCUSSION

6

depression. In our study, we targeted ECT, arguably the most effective fast-acting treatment for severe depression, using

7

ASL-fMRI, a powerful and novel quantitative functional neuroimaging technique. Our data demonstrate that ECT is

8

associated with increased CBF in a subregion of the right anterior hippocampus in all patients regardless of

9

antidepressant outcome. Notably, hippocampal CBF increased acutely after only two treatments, and appeared to

10

precede associated GM increases reported previously (24, 41, 43, 44). Furthermore, robust hippocampal and striatal

11

plasticity in our study was associated with poor clinical response to ECT, while successful antidepressant response was

12

linked to functional changes in other structures like the dorsomedial thalamus and cortical regions. Taken together, our

13

results suggest that a balance between functional plasticity in brain regions located near the right-temporal ECT electrode

14

(hippocampus, striatum) and plasticity in regions near the vertex electrode and associated with seizure generalization

15

(dorsal thalamus and cortex) is key to successful clinical outcome in right-unilateral ECT (Figure 5). Our results also

16

indicate that excessive hippocampal plasticity may not be a biomarker of positive ECT outcome.

17 18 19 20 21

Global vs. Regional Neurofunctional Plasticity in ECT

22

where seizure activity occurs throughout the brain. A “successful” ECT session is typically defined as inducing a

23

generalized seizure, evidenced by highly coordinated seizure activity at all/most recording sites during multi-channel EEG

24

and motor symptoms (45). However, there is an abundance of evidence from animal models (46, 47), patients with

25

epilepsy (48), and ECT (5, 22, 49) demonstrating that neuronal activity measured during generalized seizures is not

26

homogenous throughout the brain (20). Correspondingly, global CBF remained relatively static in our study, while robust

27

regional CBF change occurred in circumscribed areas in all patients and according to antidepressant response (discussed

28

further below).

RI PT

SC

M AN U

EP

TE D

In ECT, alternating electrical current is briefly applied via electrodes placed on the head to elicit generalized seizures,

AC C

29

Understanding the mechanisms of antidepressant response is paramount to developing more effective treatments for

30

Global CBF was, however, associated with ECT outcome. Pre-treatment global CBF was lower in ECT responders than

31

both nonresponders and healthy controls, and a modest linear correlation between pre-treatment global CBF and

32

symptom change was apparent. Global CBF also increased transiently after the initial two ECT treatments in ECT

33

responders. This latter effect supports a previous PET study reporting increased global CBF during ECT-induced

34

seizures, though relationships with antidepressant response were not reported (50). However, the nature of this potential 9

ACCEPTED MANUSCRIPT link between lower levels of pre-treatment global brain activity and successful antidepressant response is unclear. Lower

2

global function at baseline could prevent neuronal hyperexcitability during treatment (which can be detrimental to cell

3

health), or could allow more capacity for neurons to respond when exposed to ECT-related electric fields and/or seizure

4

activity. Future animal studies will be better able to probe this putative relationship between baseline CBF and

5

antidepressant response at a cellular level. Clinical studies could also explore whether inducing a state of lower global

6

brain function prior to initiating ECT treatment would increase the probability of positive antidepressant outcome. This

7

modest effect aside, overall our data support that the effects of ECT are not global, but instead occur in specific brain

8

networks relevant to depression and seizure physiology.

RI PT

1

Hippocampal plasticity and ECT outcome

14

surrounding regions after treatment, including increases in a variety of neuroimaging markers of gray matter in humans

15

(24, 41, 43, 44, 51), increased cellular plasticity including neurogenesis in animal models (15, 16, 52), and several reports

16

of associated neurofunctional plasticity (13, 14, 19, 38). Our data corroborate these findings using a novel technique; we

17

report that brain activity measured with ASL-fMRI increased in right anterior hippocampus in all volunteers after right-

18

unilateral ECT (effects were bilateral at more lenient thresholds). Notably, CBF increased after only two treatments,

19

apparently preceding changes in gray-matter volume in this region. These neurofunctional changes could occur alongside

20

microstructural changes not resolved by MRI (15, 16, 52), perhaps also related to inflammatory processes (and

21

associated neurotrophic effects) (53). Furthermore, when analyzing data from ECT nonresponders separately from

22

responders, hippocampal CBF change appeared more pronounced in nonresponders. Previous human neuroimaging

23

studies have either reported modest (41) or no (24, 43, 44) correlations (including in meta-analyses (54, 55) and a mega-

24

analysis (56)) between improved depressive symptoms and GMV increases in hippocampus and/or surrounding cortical

25

tissue. Thus, the nature of the link between hippocampal plasticity and antidepressant response to ECT remains tenuous,

26

though our results suggest that more may not be beneficial when it comes to neurofunctional plasticity in the hippocampus

27

and other ventral structures like the striatum. Hippocampal plasticity may also be linked to possible ECT-related memory

28

complaints (57), though this area is understudied (39, 58).

SC

9 10 11 12 13

AC C

EP

TE D

M AN U

The most consistent finding reported across the ECT literature is increased structural plasticity of the hippocampi and

29 30

Perhaps the most parsimonious explanation of ECT-related hippocampal plasticity is the long-term consequences of

31

seizure activity due to its proximity to the temporal electrode(s). The anteromedial temporal lobes are highly susceptible to

32

seizures as indicated by the epilepsy literature (59). SPECT studies have shown increased activity in anteromedial 10

ACCEPTED MANUSCRIPT temporal regions at the beginning of the ECT session, which is often interpreted as reflecting seizure initiation (60). After

2

seizure initiation, however, brain-activity changes occur elsewhere during ECT-induced seizures, typically including

3

increased thalamic and brainstem activity coupled with decreased cortical activity (50, 60–62) consistent with patterns of

4

long-term changes we report in ECT responders (discussed below). Given the complexity of current models of both

5

seizure neurophysiology and depressive neuropathophysiology, the hippocampus is unlikely to be the sole (or perhaps

6

even the most) critical component of the mechanisms of antidepressant response to ECT.

RI PT

1

7 8 9 10 11

Relevance of dorsal thalamo-cortical regions to ECT physiology

12

and healthy controls. These thalamic CBF levels increased, or “normalized,” acutely after two ECT treatments in ECT

13

responders and remained stable through the 6-month follow-up MRI. The thalamus has been linked to the propagation of

14

generalized seizures in animal models (63, 64), and the thalamus and/or thalamo-cortical networks are thought to play an

15

important role in the generalization of seizures during ECT (60, 65), which may be critical to its clinical success (66).

16

Acute increases in thalamic activity in responders were accompanied by decreased cortical activity in lateral frontal and

17

parietal areas in our study, similar to patterns reported in SPECT studies tracking real-time changes in brain activity

18

during generalized seizures induced during ECT (thalamic and brainstem increases, cortical decreases) (50, 60–62).

19

Clearly, generalized seizures occur in both responders and nonresponders to ECT; this is closely monitored during each

20

treatment (45). However, our data suggest that perhaps the lasting neuroplastic effects of seizure generalization (i.e.,

21

long-term plasticity in thalamo-cortical networks occurring during ECT index) are more pronounced in ECT responders.

SC

M AN U

TE D

EP

22

Patients who responded to ECT in our study had lower thalamic CBF before treatment compared to both nonresponders

ECT responders also exhibited increased CBF in somatomotor cortices contralateral to the position of the vertex

24

electrode, which is just right of midline in right unilateral ECT (67). Our group and others have reported links between ECT

25

and motor/supplementary-motor regions (35, 68–70), which could be relevant to avolitional and amotivational symptoms

26

often associated with depression (though complex goal-oriented and motor-planning behavior is typically linked to more

27

rostrofrontal structures (71)). Animal studies also show that lateral somatomotor cortex is susceptible to “kindling” (i.e., the

28

initiation of seizure activity with electrical stimulation) (72, 73). Given the proximity of our reported effect in lateral

29

somatomotor cortex to the vertex electrode, it is also possible that CBF increases in this region reflect the lasting

30

neuroplastic effects of motor-cortex seizure activity initiated under or near the vertex electrode, and/or the effects of

31

alternating current applied at the vertex electrode during treatment.

AC C

23

32 33 11

ACCEPTED MANUSCRIPT 1 2 3

Conclusions and Limitations

4

the right-temporal (hippocampus, ventral striatum) and vertex (somatomotor cortex) electrodes, as well as regions linked

5

to seizure generalization (thalamus, cortex), may be critical to positive outcome. By contrast, “too much” plasticity in

6

regions near the right-temporal electrode (hippocampus, ventral striatum) may not be conducive to positive outcome.

7

Thus, although the distribution of electrical current applied at each electrode may be comparable during ECT (74, 75) and

8

all patients experience generalized seizures (45), the regional distribution of ECT-induced seizure activity and/or its lasting

9

functional effects may differ according to antidepressant response (Figure 5).

SC

10

RI PT

Taken together, our results suggest that balanced neurofunctional plasticity across putative sites of seizure initiation near

Our results require further empirical validation in independent samples and using complementary model systems and

12

techniques. We interpreted our results as reflecting the long-term effects of seizure activity; however, future longitudinal

13

studies monitoring brain activity during ECT-induced seizures (e.g., high-field EEG, SPECT, high-resolution PET-MRI) will

14

be able to directly assess these relationships, and their consequences for antidepressant response. Such studies could

15

also address relationships between pre-treatment brain “states” and patterns of seizure activity elicited during ECT with

16

respect to outcome. Several additional limitations could be addressed by future research (see also Supplemental

17

Discussion). Our sample was intentionally heterogeneous to boost generalizability and translational value (76), yet multi-

18

site studies with larger sample sizes are needed to parse the potential contributions of diagnosis, co-morbidities, ECT

19

stimulation parameters, cognitive outcomes (39, 57), antidepressant treatment history, and other factors on ECT-induced

20

neuroplasticity. The cerebellum is linked to seizure termination in both humans (61) and animal models (77, 78), and yet

21

was excluded from our analyses because it was not fully captured by our field-of-view. New ASL sequences with

22

improved spatio-temporal resolution will be helpful in measuring the effects of ECT on the cerebellum and smaller

23

structures like the amygdala. Finally, our interpretation is speculative (Figure 5); however, the field is need of testable,

24

mechanistic models beyond the continued characterization of ECT. We propose this interpretation with the goal of

25

informing much-needed future research.

TE D

EP

AC C

26

M AN U

11

12

ACCEPTED MANUSCRIPT 1 2 3

ACKNOWLEDGEMENTS

4

MH102743 to Dr. Narr, the Muriel Harris Chair in Geriatric Psychiatry to Dr. Espinoza, as well as the Brian and Behavior

5

Research Foundation, a NARSAD Young Investigator award to Dr. Leaver.

EP

TE D

M AN U

SC

All authors report no biomedical financial interests or conflicts of interest.

RI PT

CONFLICT OF INTEREST

AC C

6 7 8 9 10

This work was supported by the NIH, including R01 MH092301 and U01 MH110008 to Drs. Narr and Espinoza and K24

13

ACCEPTED MANUSCRIPT

4 5 6 7

REFERENCES 1. Dierckx B, Heijnen WT, van den Broek WW, Birkenhäger TK (2012): Efficacy of electroconvulsive therapy in bipolar versus unipolar major depression: a meta-analysis. Bipolar Disord. 14: 146–150. 2. Fink M (2014): What was learned: studies by the consortium for research in ECT (CORE) 1997-2011. Acta Psychiatr Scand. 129: 417–26.

RI PT

1 2 3

3. Husain MM, Rush AJ, Fink M, Knapp R, Petrides G, Rummans T, et al. (2004): Speed of response and remission in

8

major depressive disorder with acute electroconvulsive therapy (ECT): a Consortium for Research in ECT

9

(CORE) report. J Clin Psychiatry. 65: 485–91.

4. Lisanby SH (2007): Electroconvulsive therapy for depression. N Engl J Med. 357: 1939–45.

11

5. Blumenfeld H, Westerveld M, Ostroff RB, Vanderhill SD, Freeman J, Necochea A, et al. (2003): Selective frontal, parietal, and temporal networks in generalized seizures. NeuroImage. 19: 1556–1566.

M AN U

12

SC

10

13

6. Nobler MS (1994): Regional Cerebral Blood Flow in Mood Disorders, III. Arch Gen Psychiatry. 51: 884.

14

7. Lyden H, Espinoza RT, Pirnia T, Clark K, Joshi SH, Leaver AM, et al. (2014): Electroconvulsive therapy mediates

15 16

neuroplasticity of white matter microstructure in major depression. Transl Psychiatry. 4: e380. 8. Abbott CC, Lemke NT, Gopal S, Thoma RJ, Bustillo J, Calhoun VD, Turner JA (2013): Electroconvulsive therapy response in major depressive disorder: a pilot functional network connectivity resting state FMRI investigation.

18

Front Psychiatry. 4: 10.

19

TE D

17

9. van Waarde JA, Scholte HS, van Oudheusden LJB, Verwey B, Denys D, van Wingen GA (2015): A functional MRI marker may predict the outcome of electroconvulsive therapy in severe and treatment-resistant depression. Mol

21

Psychiatry. 20: 609–614.

22

EP

20

10. Tendolkar I, van Beek M, van Oostrom I, Mulder M, Janzing J, Voshaar RO, van Eijndhoven P (2013): Electroconvulsive therapy increases hippocampal and amygdala volume in therapy refractory depression: a

24

longitudinal pilot study. Psychiatry Res. 214: 197–203.

25

AC C

23

11. Joshi SH, Espinoza RT, Pirnia T, Shi J, Wang Y, Ayers B, et al. (2015): Structural plasticity of the hippocampus and

26

amygdala induced by electroconvulsive therapy in major depression. Biol Psychiatry. 79: 282–92.

27

12. Sheline YI, Sanghavi M, Mintun MA, Gado MH (1999): Depression Duration But Not Age Predicts Hippocampal

28 29 30

Volume Loss in Medically Healthy Women with Recurrent Major Depression. J Neurosci. 19: 5034–5043. 13. Abbott CC, Jones T, Lemke NT, Gallegos P, McClintock SM, Mayer AR, et al. (2014): Hippocampal structural and functional changes associated with electroconvulsive therapy response. Transl Psychiatry. 4: e483.

14

ACCEPTED MANUSCRIPT

2 3 4 5 6

14. Leaver AM, Espinoza R, Pirnia T, Joshi SH, Woods RP, Narr KL (2016): Modulation of Intrinsic Brain Activity by Electroconvulsive Therapy in Major Depression. Biol Psychiatry Cogn Neurosci Neuroimaging. 1: 77–86. 15. Chen F, Madsen TM, Wegener G, Nyengaard JR (2009): Repeated electroconvulsive seizures increase the total number of synapses in adult male rat hippocampus. Eur Neuropsychopharmacol. 19: 329–338. 16. Madsen TM, Treschow A, Bengzon J, Bolwig TG, Lindvall O, Tingström A (2000): Increased neurogenesis in a model of electroconvulsive therapy. Biol Psychiatry. 47: 1043–1049.

RI PT

1

7

17. Eisch AJ, Petrik D (2012): Depression and hippocampal neurogenesis: a road to remission? Science. 338: 72–5.

8

18. Hanson ND, Owens MJ, Nemeroff CB (2011): Depression, Antidepressants, and Neurogenesis: A Critical

11 12 13

SC

10

Reappraisal. Neuropsychopharmacology. 36: 2589–2602.

19. Argyelan M, Lencz T, Kaliora S, Sarpal DK, Weissman N, Kingsley PB, et al. (2016): Subgenual cingulate cortical activity predicts the efficacy of electroconvulsive therapy. Transl Psychiatry. 6: e789.

M AN U

9

20. McNally KA, Blumenfeld H (2004): Focal network involvement in generalized seizures: new insights from electroconvulsive therapy. Epilepsy Behav. 5: 3–12.

14

21. Bolwig TG (2014): Neuroimaging and electroconvulsive therapy: a review. J ECT. 30: 138–42.

15

22. Enev M, McNally KA, Varghese G, Zubal IG, Ostroff RB, Blumenfeld H (2007): Imaging onset and propagation of

17

ECT-induced seizures. Epilepsia. 48: 238–44.

23. Wade BSC, Joshi SH, Njau S, Leaver AM, Vasavada M, Woods RP, et al. (2016): Effect of Electroconvulsive Therapy

TE D

16

18

on Striatal Morphometry in Major Depressive Disorder. Neuropsychopharmacology. 41: 2481–2491.

19

24. Dukart J, Regen F, Kherif F, Colla M, Bajbouj M, Heuser I, et al. (2014): Electroconvulsive therapy-induced brain

22 23

EP

21

plasticity determines therapeutic outcome in mood disorders. Proc Natl Acad Sci U S A. 111: 1156–61. 25. Kellner CH, Knapp R, Husain MM, Rasmussen K, Sampson S, Cullum M, et al. (2010): Bifrontal, bitemporal and right unilateral electrode placement in ECT: randomised trial. Br J Psychiatry J Ment Sci. 196: 226–34.

AC C

20

26. Sackeim HA, Prudic J, Devanand DP, Nobler MS, Lisanby SH, Peyser S, et al. (2000): A prospective, randomized,

24

double-blind comparison of bilateral and right unilateral electroconvulsive therapy at different stimulus intensities.

25

Arch Gen Psychiatry. 57: 425.

26 27 28 29

27. Wang Z, Aguirre GK, Rao H, Wang J, Fernández-Seara MA, Childress AR, Detre JA (2008): Empirical optimization of ASL data analysis using an ASL data processing toolbox: ASLtbx. Magn Reson Imaging. 26: 261–269. 28. American Psychiatric Association (2000): Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition: DSMIV-TR®. American Psychiatric Association.

15

ACCEPTED MANUSCRIPT 1

29. Sheehan D V, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, et al. (1998): The Mini-International

2

Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric

3

interview for DSM-IV and ICD-10. J Clin Psychiatry. 59 Suppl 2: 22-33;quiz 34-57. 30. Hamilton M (1980): Rating depressive patients. J Clin Psychiatry. 41: 21–24.

5

31. Montgomery SA, Asberg M (1979): A new depression scale designed to be sensitive to change. Br J Psychiatry J

6 7

RI PT

4

Ment Sci. 134: 382–389.

32. Rush AJ, Trivedi MH, Ibrahim HM, Carmody TJ, Arnow B, Klein DN, et al. (2003): The 16-Item Quick Inventory of Depressive Symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): a psychometric

9

evaluation in patients with chronic major depression. Biol Psychiatry. 54: 573–583.

10

SC

8

33. Nierenberg AA, DeCecco LM (2001): Definitions of antidepressant treatment response, remission, nonresponse, partial response, and other relevant outcomes: a focus on treatment-resistant depression. J Clin Psychiatry. 62

12

Suppl 16: 5–9.

M AN U

11

13

34. Pirnia T, Joshi SH, Leaver AM, Vasavada M, Njau S, Woods RP, et al. (2016): Electroconvulsive therapy and

14

structural neuroplasticity in neocortical, limbic and paralimbic cortex. Transl Psychiatry. 6: e832.

16 17 18 19

35. Leaver AM, Wade B, Vasavada M, Hellemann G, Joshi SH, Espinoza R, Narr KL (2018): Fronto-Temporal Connectivity Predicts ECT Outcome in Major Depression. Front Psychiatry. 9. doi: 10.3389/fpsyt.2018.00092. 36. Leaver AM, Espinoza R, Joshi SH, Vasavada M, Njau S, Woods RP, Narr KL (2016): Desynchronization and Plasticity

TE D

15

of Striato-frontal Connectivity in Major Depressive Disorder. Cereb Cortex N Y N 1991. 26: 4337–4346. 37. Njau S, Joshi SH, Espinoza R, Leaver AM, Vasavada M, Marquina A, et al. (2017): Neurochemical correlates of rapid treatment response to electroconvulsive therapy in patients with major depression. J Psychiatry Neurosci JPN.

21

42: 6–16.

23 24 25 26 27 28 29

38. Njau S, Joshi SH, Leaver AM, Vasavada M, Fleet J, Espinoza R, Narr KL (2016): Variations in myo-inositol in fronto-

AC C

22

EP

20

limbic regions and clinical response to electroconvulsive therapy in major depression. J Psychiatr Res. 80: 45–51. 39. Vasavada MM, Leaver AM, Njau S, Joshi SH, Ercoli L, Hellemann G, et al. (2017): Short- and Long-term Cognitive Outcomes in Patients With Major Depression Treated With Electroconvulsive Therapy. J ECT. 33: 278–285. 40. Worsley KJ, Evans AC, Marrett S, Neelin P (1992): A three-dimensional statistical analysis for CBF activation studies in human brain. J Cereb Blood Flow Metab. 12: 900–18. 41. Joshi SH, Espinoza RT, Pirnia T, Shi J, Wang Y, Ayers B, et al. (2015): Structural plasticity of the hippocampus and amygdala induced by electroconvulsive therapy in major depression. Biol Psychiatry. 79: 282–92.

16

ACCEPTED MANUSCRIPT 1

42. Redlich R, Bürger C, Dohm K, Grotegerd D, Opel N, Zaremba D, et al. (2017): Effects of electroconvulsive therapy on

2

amygdala function in major depression - a longitudinal functional magnetic resonance imaging study. Psychol

3

Med. 47: 2166–2176.

4

43. Redlich R, Opel N, Grotegerd D, Dohm K, Zaremba D, Bürger C, et al. (2016): Prediction of Individual Response to Electroconvulsive Therapy via Machine Learning on Structural Magnetic Resonance Imaging Data. JAMA

6

Psychiatry. 73: 557.

7

RI PT

5

44. Tendolkar I, Beek M, Oostrom I, Mulder M, Janzing J, Voshaar RO, Eijndhoven P (2013): Electroconvulsive therapy increases hippocampal and amygdala volume in therapy refractory depression: a longitudinal pilot study.

9

Psychiatry Res. 214: 197–203.

12 13 14 15 16 17

Press.

M AN U

11

45. Kellner CH, Pritchett JT, Beale MD, Coffey CE (1997): Handbook of ECT. Washington, D.C.: American Psychiatric

46. Nersesyan H, Hyder F, Rothman DL, Blumenfeld H (2004): Dynamic fMRI and EEG recordings during spike-wave seizures and generalized tonic-clonic seizures in WAG/Rij rats. J Cereb Blood Flow Metab. 24: 589–599. 47. Blumenfeld H, McCormick DA (2000): Corticothalamic Inputs Control the Pattern of Activity Generated in Thalamocortical Networks. J Neurosci. 20: 5153–5162.

48. Blumenfeld H, Varghese GI, Purcaro MJ, Motelow JE, Enev M, McNally KA, et al. (2009): Cortical and subcortical networks in human secondarily generalized tonic-clonic seizures. Brain J Neurol. 132: 999–1012.

TE D

10

SC

8

18

49. Miller JW (2010): Are Generalized Tonic–Clonic Seizures Really “Generalized”? Epilepsy Curr. 10: 80–81.

19

50. Takano H, Motohashi N, Uema T, Ogawa K, Ohnishi T, Nishikawa M, Matsuda H (2011): Differences in cerebral blood flow between missed and generalized seizures with electroconvulsive therapy: A positron emission tomographic

21

study. Epilepsy Res. 97: 225–228.

51. Cano M, Martínez-Zalacaín I, Bernabéu-Sanz Á, Contreras-Rodríguez O, Hernández-Ribas R, Via E, et al. (2017):

AC C

22

EP

20

23

Brain volumetric and metabolic correlates of electroconvulsive therapy for treatment-resistant depression: a

24

longitudinal neuroimaging study. Transl Psychiatry. 7: e1023.

25 26 27 28 29 30

52. Perera TD, Coplan JD, Lisanby SH, Lipira CM, Arif M, Carpio C, et al. (2007): Antidepressant-induced neurogenesis in the hippocampus of adult nonhuman primates. J Neurosci Off J Soc Neurosci. 27: 4894–4901. 53. van Buel EM, Patas K, Peters M, Bosker FJ, Eisel ULM, Klein HC (2015): Immune and neurotrophin stimulation by electroconvulsive therapy: is some inflammation needed after all? Transl Psychiatry. 5: e609. 54. Wilkinson ST, Sanacora G, Bloch MH (2017): Hippocampal Volume Changes Following Electroconvulsive Therapy: A Systematic Review and Meta-analysis. Biol Psychiatry Cogn Neurosci Neuroimaging. 2: 327–335. 17

ACCEPTED MANUSCRIPT 1

55. Takamiya A, Chung JK, Liang K, Graff-Guerrero A, Mimura M, Kishimoto T (2018): Effect of electroconvulsive therapy

2

on hippocampal and amygdala volumes: systematic review and meta-analysis. Br J Psychiatry. 212: 19–26.

3

56. Oltedal et al. (In Press): Volume of the Human Hippocampus and Clinical Response Following Electroconvulsive

6 7 8 9

57. Lisanby SH, Maddox JH, Prudic J, Devanand DP, Sackeim HA (2000): The Effects of Electroconvulsive Therapy on

RI PT

5

Therapy. Biol Psychiatry. . doi: 10.1016/j.biopsych.2018.05.017.

Memory of Autobiographical and Public Events. Arch Gen Psychiatry. 57: 581–590.

58. Szabo K, Hirsch JG, Krause M, Ende G, Henn FA, Sartorius A, Gass A (2007): Diffusion weighted MRI in the early phase after electroconvulsive therapy. Neurol Res. 29: 256–259.

59. Margerison JH, Corsellis JA (1966): Epilepsy and the temporal lobes. A clinical, electroencephalographic and

SC

4

neuropathological study of the brain in epilepsy, with particular reference to the temporal lobes. Brain J Neurol.

11

89: 499–530.

12 13 14 15 16

M AN U

10

60. Enev M, McNally KA, Varghese G, Zubal IG, Ostroff RB, Blumenfeld H (2007): Imaging onset and propagation of ECT-induced seizures. Epilepsia. 48: 238–244.

61. Blumenfeld H, Varghese GI, Purcaro MJ, Motelow JE, Enev M, McNally KA, et al. (2009): Cortical and subcortical networks in human secondarily generalized tonic–clonic seizures. Brain. 132: 999–1012. 62. Takano H, Motohashi N, Uema T, Ogawa K, Ohnishi T, Nishikawa M, et al. (2007): Changes in regional cerebral blood flow during acute electroconvulsive therapy in patients with depression: Positron emission tomographic study. Br

18

J Psychiatry. 190: 63–68.

21 22 23 24 25 26

815–846.

EP

20

63. McCormick DA, Contreras D (2001): On The Cellular and Network Bases of Epileptic Seizures. Annu Rev Physiol. 63:

64. Bertram EH, Zhang D, Williamson JM (2008): Multiple roles of midline dorsal thalamic nuclei in induction and spread of limbic seizures. Epilepsia. 49: 256–268.

AC C

19

TE D

17

65. Bolwig TG (2011): How does electroconvulsive therapy work? Theories on its mechanism. Can J Psychiatry Rev Can Psychiatr. 56: 13–18.

66. Jan Shah A, Wadoo O, Latoo J (2013): Electroconvulsive Therapy (ECT): Important parameters which influence its effectiveness. Br J Med Pract. 6: 31–36.

27

67. d’Elia G (1970): Unilateral electroconvulsive therapy. Acta Psychiatr Scand Suppl. 215: 1–98.

28

68. Waarde JA, Scholte HS, Oudheusden LJB, Verwey B, Denys D, Wingen GA (2015): A functional MRI marker may

29

predict the outcome of electroconvulsive therapy in severe and treatment-resistant depression. Mol Psychiatry.

30

20: 609–614. 18

ACCEPTED MANUSCRIPT 1

69. Jiang R, Abbott CC, Jiang T, Du Y, Espinoza R, Narr KL, et al. (2017): SMRI Biomarkers Predict Electroconvulsive

2

Treatment Outcomes: Accuracy with Independent Data Sets. Neuropsychopharmacology. . doi:

3

10.1038/npp.2017.165.

4

70. Wade BSC, Sui J, Hellemann G, Leaver AM, Espinoza RT, Woods RP, et al. (2017): Inter and intra-hemispheric structural imaging markers predict depression relapse after electroconvulsive therapy: a multisite study. Transl

6

Psychiatry. 7: 1270.

9 10 11 12 13

Mov Sci. 26: 590–616.

72. French J, Gernandt B, Livingston R (1955): Regional differences in seizure susceptibility in monkey cortex. Trans Am

SC

8

71. Grafton ST, de C. Hamilton AF (2007): Evidence for a distributed hierarchy of action representation in the brain. Hum

Neurol Assoc. 194–196.

73. Fukushima J, Kohsaka S, Fukushima K, Kato M (1987): Motor Cortical Kindling in Cats: A Comparison of Adult Cats and Kittens. Epilepsia. 28: 651–657.

M AN U

7

RI PT

5

74. Bai S, Loo C, Al Abed A, Dokos S (2012): A computational model of direct brain excitation induced by

14

electroconvulsive therapy: comparison among three conventional electrode placements. Brain Stimulat. 5: 408–

15

421.

16

75. Lee WH, Deng Z-D, Kim T-S, Laine AF, Lisanby SH, Peterchev AV (2012): Regional electric field induced by electroconvulsive therapy in a realistic finite element head model: Influence of white matter anisotropic

18

conductivity. NeuroImage. 59: 2110–2123.

20

76. Insel T, Cuthbert B, Garvey M, Heinssen R, Pine DS, Quinn K, et al. (2010): Research domain criteria (RDoC): toward a new classification framework for research on mental disorders. Am J Psychiatry. 167: 748–51.

EP

19

TE D

17

77. Krook-Magnuson E, Szabo GG, Armstrong C, Oijala M, Soltesz I (2014): Cerebellar Directed Optogenetic Intervention

22

Inhibits Spontaneous Hippocampal Seizures in a Mouse Model of Temporal Lobe Epilepsy. eNeuro. 1. doi:

23

10.1523/ENEURO.0005-14.2014.

24 25 26 27

AC C

21

78. Salgado-Benftez A, Briones R, Fernandez-Guardiola A (1982): Purkinje Cell Responses to a Cerebral PenicillinInduced Epileptogenic Focus in the Cat. Epilepsia. 23: 597–606.

19

ACCEPTED MANUSCRIPT 1 2 3

FIGURE CAPTIONS

4

black solid line) and nonresponders (NR, gray dashed lines) to ECT, measured before treatment (“0”), after 2

5

treatments or approximately four days later (“+4d”), after ECT index approximately four weeks after baseline

6

(“+4w”), and 6 months after finishing ECT treatment (“+6m”). Global CBF is also plotted for healthy control

7

volunteers (HC, open gray circles), measured at baseline and approximately 4 weeks later (HC did not receive

8

ECT). Error bars reflect standard error. B. Baseline global CBF is plotted for each patient against change in

9

depression scores after ECT treatment (where -1 = 100% reduction/improvement). Regression line highlights a

SC

RI PT

Figure 1. Global CBF and antidepressant response to ECT. A. Mean global CBF is plotted for responders (R,

modest negative correlation between these values, p = 0.02.

11 12

Figure 2. Acute regional CBF change after two treatments. A. Basal brain function measured with CBF

13

increased in right anterior hippocampus after 2 ECT treatments (orange), and decreased in bilateral dorsal

14

basal ganglia (caudate, blue voxels). B & C. Mean regional CBF (corrected for global CBF) and mean regional

15

GM volume (GMV, corrected for global GM) are plotted for the significant results shown in A at top (B) and

16

bottom (C) panels, respectively. Data for patients is plotted in black lines; open circles reflect data from healthy

17

controls. The results of post-hoc pairwise comparisons amongst timepoints of interest in patients are indicated

18

below each graph (Bonferroni corrected, p(corr) < 0.05).

19 20

Figure 3. Regional CBF change after ECT treatment (index series). A. Basal brain function measured with

21

CBF increased in three regions after ECT treatment (orange), including the right anterior hippocampus and

22

bilateral ventral basal ganglia. B & C. Mean regional CBF (corrected for global CBF) and mean regional GM

23

volume (GMV, corrected for global GM) are plotted for three regions surviving cluster-level thresholding at top

24

(B) and bottom (C) panels, respectively. Black lines reflect means for patients, open circles plot means for

25

healthy controls, and standard error bars are plotted for both groups. The results of post-hoc pairwise

26

comparisons amongst time-points of interest are indicated below each graph (p(corr) < 0.05, Bonferroni

27

correction). Note that analyses of CBF shown in B statistically control for cluster GMV, and vice versa for panel

28

C.

AC C

EP

TE D

M AN U

10

29 20

ACCEPTED MANUSCRIPT Figure 4. Exploratory analyses yielded different patterns of regional CBF change after ECT treatment (index

2

series) in Responders and Nonresponders. A. In ECT responders, basal brain function measured with CBF

3

increased in the thalamus, brainstem, and left somatomotor cortex after ECT treatment (orange), and

4

decreased the regions of the right lateral frontal and parietal cortex (blue). Mean CBF values (corrected for

5

global CBF) are plotted for each region below, with solid black lines indicating responders, gray dashed lines

6

indicating nonresponders, and open circles reflecting data from healthy controls. Error bars mark standard

7

error. Statistically significant (p(corr) < 0.05, Bonferroni correction) changes over time are plotted for

8

responders in solid black lines (R, top); nonresponders and healthy controls did not change over time (n.s.). B.

9

Regions of significant change in CBF after ECT index are shown for nonresponders, including increased CBF

SC

RI PT

1

(orange) in right hippocampus and bilateral ventral basal ganglia and decreased CBF in posterior cingulate

11

cortex (blue). At right, mean CBF (corrected for global values) are plotted for responders in black,

12

nonresponders in gray dashed lines, and for healthy controls in open circles. Again, error bars reflect standard

13

error. Significant pairwise tests for these regions are indicated below each plot for nonresponders in gray

14

dashed lines (p(corr) < 0.05, Bonferroni correction); responders and healthy controls did not change over time

15

(n.s.).

16 17

Figure 5. Results Summary: Balanced Long-Term Neuroplastic Effects in ECT. We propose that a balance in

18

functional neuroplasticity in brain regions associated with the three well-known stages of seizure progression –

19

initiation (green), generalization (blue), and termination (red) – may be an indicator of positive antidepressant

20

response to ECT. This schematic summarizes the results of our study in ECT responders, who exhibited CBF

21

changes in right hippocampus and somato-motor cortex (green) and thalamo-cortical regions (blue). The

22

cerebellum (red) was not measured (as indicated by the question mark), but has been linked to seizure

23

termination in previous studies and may also play a role in ECT.

TE D

EP

AC C

24

M AN U

10

25 26

21

ACCEPTED MANUSCRIPT 1 2

TABLES Table 1. Demographic and clinical information Depressed Sample Size Age, mean (SD)

Controls

n = 57

n = 36

41.42 (12.98)

39.06 (12.29)

30/27

19/17

Sex, females/males Diagnosis, unipolar/bipolar

47/10

ECT lead placement, only-RUL/other

37/20

ECT outcome (any scale), responder/nonresponder

25/21

ECT outcome (composite), responder/nonresponder

19/27

Age at 1st diagnosed depressive episode, mean (SD)

25.30 (11.46) Post-2tx

Post-Index/4wk

Post-6mo

HAM-17, mean (SD)

23.88 (5.94)

18.08 (6.93)a

13.26 (7.46) a,b

11.52 (8.22)a

MADRS, mean (SD)

37.64 (8.49)

29.0 (10.97) a

QIDS-SR, mean (SD)

19.75 (4.12)

15.84 (5.99)

a

Corrected Sample Size (after attrition and MRI QC)

n = 57

n = 51

SC

Baseline

19.05 (11.26) a,b 11.26 (5.77) n = 43

a,b

Baselin e

Post-4wk

n = 32

n = 33

16.81 (12.76)a

10.48 (6.30) a n = 27

M AN U

Study Visits

RI PT

Clinical Information

Results of paired t-tests for depression scores are indicated as follows: a significant difference between baseline and follow-up, p < 0.0001, b Significant difference from previous visit, p < 0.005; Baseline depression scores were missing for a single volunteer due to clerical error, and this volunteer was not included in any calculations or analyses requiring depression scores.

AC C

EP

TE D

3

22

ACCEPTED MANUSCRIPT Table 2. Regional CBF change with ECT MNI Coordinates (Center of Gravity)

Baseline vs. +2tx

Right hippocampus

31.2

-13.1

-24.8

370

0.0328

Caudate

-2.7

-9.3

19.5

797

0.000188

Left putamen & accumbens Right putamen & accumbens

-15.1

-21.1

550

0.00785

3.3

-2.9

622

0.00363

21.1

2.8

-2.9

477

0.0177

Left thalamus

-10.9

-14.5

-3.0

1133

2.50E-06

Left somatomotor cortex

-22.7

-10.2

60.4

879

3.68E-05

Left occipital cortex

-13.7

-90.5

0.6

321

0.0438

Right angular gyrus

49.4

-50.2

24.0

2260

1.09E-10

Right frontal operculum

43.6

13.0

3.1

457

0.00622

3.3

-50.2

66.7

344

0.031

Precuneus Right hippocampus & accumbens

29.7

-5.1

-13.8

1931

1.79E-07

Posterior cingulate cortex

-7.3

-58.3

33.3

843

0.000767

M AN U

Nonresponders Only, Baseline vs. post-ECT (+4wks)

34.3 -19.3

SC

Responders Only, Baseline vs. post-ECT (+4wks)

Z

RI PT

Right hippocampus

Y

p, RFT corrected

Anatomical Location

Baseline vs. post-ECT (+4wks)

X

Volume (mm3)

Model Effect

AC C

EP

TE D

1

23

ACCEPTED MANUSCRIPT

A.

B.

60

90

Global CBF

45 40

60 50 40 30 20 10

0 +4d

+4w

+6m

SC

Global CBF

70

50

35

RI PT

80 55

0.1

-0.2

-0.5

-0.8

AC C

EP

TE D

M AN U

Change in Depression Score

-1.1

ACCEPTED MANUSCRIPT A. Acute CBF Change After 2 ECT Tx

B.

1.1

Right Hippocampus

0.75

1.05

Caudate

0.7

Mean CBF

1 0.95

0.65

0.9

0.6

0.85 0.55

0.75

x = 28

0 +4d

+4w

T Tests

x = 15

C.

1.67

Right Hippocampus

1.66

1.61

+6m

+4w

+6m

Caudate

0.49 0.48

1.59

T Tests

+4w

0.5

1.62

1.6

TE D EP AC C

0.53

0 +4d

0.51

1.63

1.58

0.5

SC

Mean GMV

1.64

M AN U

y = -2

+6m

0.52

1.65

z = -20

RI PT

0.8

0 +4d

+4w

+6m

0.47

0 +4d

ACCEPTED MANUSCRIPT A. CBF Change After ECT Index

y = -20

y=6

z = -8

z = -20

z=0

B.

0.95

0.95

0.9

0.9

0.85

0.85

0.8

0.8

0.75

0.75

1.1

Right Ventral Basal Ganglia

1.05

0.95

0.9

0.85

+4w

+6m

Right Hippocampus

1.23

0 +4d

+4w

+6m

0.87

Left Ventral Basal Ganglia

0.86

1.22

0.75

1.04

0 +4d

+4w

+6m

Right Ventral Basal Ganglia

1.02

0.85

1

0.84

EP

1.21 1.2

0.98

0.83

1.19

0.82

1.18

0.96

0.81

0.94

0.8

T Tests

AC C

1.17 1.16

0.8

TE D

0 +4d

1.24

Mean GMV

Left Ventral Basal Ganglia

1

T Tests

Mean CBF

C.

1

M AN U

Right Hippocampus

1

SC

RI PT

x = 28

0 +4d

+4w

+6m

0.79

0 +4d

+4w

[all n.s.]

+6m

0.92

0 +4d

+4w

[all n.s.]

+6m

ACCEPTED MANUSCRIPT A. Baseline vs. Post-Index in ECT Responders

1.15

Thalamus

0.95

1.15

z=5

Inferior Frontal Cx

0.85

0.95

Lateral Parietal Cx

1.05

1.05

0.95

0.95

0.85

0.85

0.75

0.75

0.8

0.85

0.75 0.7

0.75

0.65

0.65

0.6

0.55

0.55

R NR

1.15

[all n.s.]

[all n.s.]

[all n.s.]

Mean CBF

T Tests

TE D

z = -8

AC C

EP

z = -20

z = 30

Precuneus

0.95 0.85 0.75 0.65 0.55

Right Hippocampus + Ventral Basal Ganglia

1.15

0.95

1.1

0.9

1.05

0.85

1

0.8

0.95

0.75

0.9

0.7

0.85

R NR

[all n.s.]

z = 63

1.05

M AN U 1

x = 28

1.15

[all n.s.]

B. Baseline vs. Post-Index in ECT Nonresponders

z = -6

z = 40

SC

Mean CBF

Somatomotor Cx

z = -8

0.9

1.05

T Tests

y = -14

RI PT

z = -48

[all n.s.]

Posterior Cingulate

[all n.s.]

RI PT

ACCEPTED MANUSCRIPT

M AN U

Thalamo-Cortical Networks

SC

Somato-Motor Cortex

Temporal Lobe Regions

AC C

EP

TE D

Cerebellum(?)

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

Mechanisms of Antidepressant Response to Electroconvulsive Therapy Studied With Perfusion Magnetic Resonance Imaging Supplement

RI PT

Supplemental Methods

ECT

In this naturalistic MRI study, patient volunteers underwent a clinically prescribed course of ECT (5000Q MECTA Corp) at the UCLA Resnick Neuropsychiatric Hospital. Treatment was administered by ECT

SC

attending physicians using standard protocols in an inpatient or outpatient setting. In brief, patients began with right-unilateral lead placement (ultra brief pulse; 5-6x seizure threshold) (1), but transitioned to bitemporal lead placement (2x seizure threshold) if indicated clinically (Table 1). ECT was administered

M AN U

three times weekly for an average of 10.9 total treatments (SD=3.8) using short-acting anesthesia and a muscle relaxant. Image Acquisition

Using a 3T Siemens Allegra scanner, continuous ASL images were acquired: 60 volumes (30 label, 30 control), 4x4x7.5 mm3 resolution, 18 axial slices, repetition time 4000 ms, echo time 16 ms, label time 2100 ms, post label delay 1000 ms, and 95% duty cycle. During ASL sequences, subjects were resting with eyes

Statistical Analyses

TE D

closed. A T1-weighted anatomical scan (MPRAGE) was also collected at each session (2–4).

All statistical analyses were completed in R (https://www.r-project.org). Linear mixed-effects models were

EP

implemented using the lme4 package (5). The oro.nifti package was used for voxelwise analyses (6). Estimating p-values for the contribution of individual model factors (e.g., time in our model) in linear mixedeffects models (LMMs) is a matter of debate amongst statisticians (5, 7, 8). In our study, we followed the

AC C

recommendations of the lme4 R package developers (5) and used likelihood ratio tests to compare two nested LMMs, where the factor of interest (i.e., time) is removed from the “null” model. The values resulting from likelihood ratio tests are asymptotically normal; therefore, chi-squared distributions are used to derive p-values for the contribution of the factor of interest (in our case, time) to the model fit (5). Although we report p-values in the main text and Figures, we also include the difference in Akaike Information Criterion (AIC) for target and null models in Figure S1 for reference. With regard to model terms, we constructed a single target statistical model a priori for the sake of simplicity. Model optimization (e.g., by testing different algorithms, nuisance factors, etc.) would be problematic for voxelwise analyses, which involve many thousands of univariate tests – and thus the potential for many

1

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

different optimal model fits. Future multi-site and/or database studies such as the recently established Global ECT MRI Research Collaboration (GEMRIC) (9) will be better able to address these issues, but their empirical treatment is outside the scope of the current manuscript. We chose nuisance model terms likely to have the strongest effects on CBF and neurobiology of

RI PT

antidepressant response to ECT. Age was chosen because it is strongly associated with CBF (10, 11). ECT lead placement and number of treatments both affect ECT “dose” including the amount and distribution of applied current and associated seizure activity, and thus could also affect the lasting neurobiological consequences of ECT. Additionally, both lead placement and number of treatments were associated with clinical decisions made during the course of treatment in relation to symptom change in our sample

SC

(Supplemental Table), and thus should be controlled when examining the neurobiology of clinical response. Biological sex is clearly relevant to depression, but its links to CBF, as well as ECT- and depression-related neurobiology, are unclear and understudied; therefore, biological sex was not included in our analyses.

M AN U

Future studies with larger sample sizes will be better able to address the role of sex and other relevant factors.

Parsing the contribution of tissue content (i.e., partial volume effects) to CBF quantified with ASL-fMRI is not necessarily straightforward. CBF quantification itself makes assumptions regarding tissue content because perfusion profiles differ for gray and white matter, and tissue content can vary from voxel to voxel even within a gray-matter mask (as used in the current study). To address this issue, we used standard

TE D

CBF quantification methods assuming a single tissue distribution (12), and addressed the potential contribution of GMV in post-hoc statistical analyses.

In recent years, neuroimaging researchers have begun to re-examine approaches to statistical analyses,

EP

specifically with regard to voxelwise tests. One concern is balancing the potential for both Type I and Type II error given the large number of tests performed and the sometimes subtle effects studied. Both the criticism of standard methods and the development of new methods requires extensive tests of various MRI

AC C

modalities, preprocessing parameters, and statistical models (13–15). The thresholds we have chosen for our study are within standard limits (16, 17), though alternatives should be considered in the future. Relatedly, there has been discussion in the field regarding the utility of p-values vs. other statistical metrics addressing significance of model parameters. Thus, for completeness, we also include statistical maps of the difference in Akaike information criterion (AIC) values for the target and null models (i.e., with and without time as a model term, respectively) in the Figure S1. Note that all AIC difference values in regions identified as statistically significant (i.e., p(corr) < 0.05 in Figures 2 and 3) are greater than 2 (18).

2

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

Supplemental Discussion There are several additional potential limitations that warrant discussion, though their thorough empirical treatment is outside the scope of the current study. Study limitations are also discussed in the final section of (and throughout) the main text, as well as the Supplemental Methods above. Additional issues are

RI PT

addressed below.

ECT lead placement was not balanced in this naturalistic study – all patients received right-unilateral (RUL) treatment and the majority received only RUL ECT. Perhaps correspondingly, our hippocampal findings were right-lateralized, though effects were bilateral when more permissive thresholds were considered

SC

(e.g., when dropping cluster-correction). Notably, depressed patients exhibit reduced hippocampal volume bilaterally (19), which could explain the relative increased efficacy of bilateral vs. RUL ECT. Indeed, a recent large-scale analysis of multi-site ECT data indicated that ECT lead placement (bilateral versus RUL)

M AN U

affected the extent of volume change in the left, but not the right hippocampus (20). Multi-site studies like these will be better powered to address whether the laterality of functional neuroplasticity in the hippocampus related to ECT lead placement, pathophysiology underlying depression, or a combination of both.

All depressed volunteers were tapered off antidepressant medications prior to our study and beginning ECT. However, long-term use of various antidepressant medications and therapies is a potential

TE D

confounder in all studies of treatment-resistant patients, including the current study. Variable treatment histories may introduce additional variability that prevented us from identifying smaller effects of ECT and antidepressant response to ECT. Maintenance treatment was not controlled in the 6-month period after ECT, which also likely increased variability at this fourth timepoint. This issue may prevent our study and

EP

others from assessing the “pure” effects of ECT and antidepressant response to ECT. However, given that ECT is typically used in treatment-refractory patients, one may also argue that it is best studied in these most severe cases. A future study including treatment-refractory patients undergoing “treatment as usual”

AC C

may be an interesting way of examining the effects of long-term medication use on the longitudinal studies like this one.

There are several brain regions that have been previously linked to depression and ECT that were not identified in the current study. One of these is the cerebellum, which was omitted from analysis due to restrictions of the ASL-fMRI field of view. Another is the amygdala, which has been repeatedly implicated in depression and ECT (21–23). Our results do not necessarily imply that these structures are not relevant to depression or ECT; rather, technical limitations like limited spatial resolution and/or the modality chosen are more likely explanations. For example, neuro-functional changes in the amygdala after ECT may be

3

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

stimulus-evoked (23) and may not effect “resting” baseline function. Future studies using faster imaging sequences and/or cross-modal analyses will give a more comprehensive view of ECT. Much of our Discussion in the main text emphasizes the role of seizure physiology, specifically hypothesizing that the regional distribution of the long-term effects of ECT-related seizure activity differs

RI PT

according to antidepressant response. However, other relevant neurobiological processes warrant discussion. The current study primarily addresses neurofunctional change measured via CBF, yet these are very likely to occur alongside molecular, microstructural, and anatomical changes not measured by this study. Inflammation is a good example. Inflammation is thought to play a role in depression (24–26), and is also modulated by ECT (27). Inflammatory processes are also associated with increased blood flow, though

SC

the relationship between systemic or neural inflammation and CBF measured with ASL is unclear. Certainly, inflammation is associated with neurotrophic factors (27) that could play a role in ECT-related

AC C

EP

TE D

and other factors would be informative.

M AN U

neuroplasticity. Multimodal analyses synthesizing ECT’s effects on serum inflammatory markers, MRI data,

4

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

Table S1. Demographic and clinical information for Responders and Nonresponders Responders n = 17 43.53 (13.17) 7/10

Nonresponders n = 26 39.50 (13.97) 16/10

13/4 27.24 (12.31) 15/2 10.35 (2.62)

21/5 23.12 (12.14) 11/15a 12.62 (3.57)a

RI PT

Sample Size Age, mean (SD) Sex, females/males Clinical Information Diagnosis, unipolar/bipolar Age at 1st diagnosed depressive episode, mean (SD) ECT lead placement, only-RUL/other Number of ECT Index Treatments

AC C

EP

TE D

M AN U

SC

Baseline Study Visit HAM-17, mean (SD) 27.06 (5.60) 22.27 (4.98)a MADRS, mean (SD) 43.71 (7.40) 33.69 (5.73)a QIDS-SR, mean (SD) 21.71 (3.51) 18.73 (3.99)a Corrected Sample Size (after attrition and MRI QC) 17 26 Post-2tx Study Visit HAM-17, mean (SD) 19.88 (6.20)b,c 16.5 (7.55)a,b,c MADRS, mean (SD) 32.76 (7.78)b,c 25.20 (11.83)b,c b,c QIDS-SR, mean (SD) 16.35 (4.65) 14.60 (6.67)b,c Corrected Sample Size (after attrition and MRI QC) 17 23 Post-Index Study Visit HAM-17, mean (SD) 7.12 (3.53)b,c 17.50 (6.23)a,b,c b,c MADRS, mean (SD) 8.59 (5.30) 26.35 (8.04)a,b b,c QIDS-SR, mean (SD) 6.76 (4.02) 14.23 (4.81)a,b,c Corrected Sample Size (after attrition and MRI QC) 17 26 Post-6mo Study Visit HAM-17, mean (SD) 12.80 (8.23)b,c 10.71 (6.91)b,c b,c MADRS, mean (SD) 18.20 (13.41) 16.00 (11.70)b,c b,c QIDS-SR, mean (SD) 11.13 (5.57) 9.65 (6.11)b,c Corrected Sample Size (after attrition and MRI QC) 13 14 Results of chi-squared and t-tests are indicated as follows: a Significant difference between Responders and Nonresponders, p < 0.05, b significant difference between baseline and follow-up (within group), p < 0.005, c Significant difference from previous visit (within group), p < 0.02. All other comparisons were not significant.

5

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

M AN U

SC

RI PT

A. Acute CBF Change After 2 ECT Tx, AIC(target) – AIC(null)

AC C

EP

TE D

B. CBF Change After ECT Index, AIC(target) – AIC(null)

Figure S1. Difference in Akaike Information Criterion (AIC) values are displayed for voxelwise analyses of CBF change (A) between baseline and after 2 ECT sessions and (B) baseline and after ECT index. Maps are thresholded using the same criteria as described in the main text (p(corr) < 0.05), but here each voxel’s color indicates the corresponding AIC difference value (key at bottom right in each panel). Difference in AIC between target at null models were above 2 in all voxels shown, with ranges 2.72-13.45 for acute change (panel A) and 2.88-13.71 for post-index change (panel B).

6

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

Supplemental References 1. d’Elia G (1970): Unilateral electroconvulsive therapy. Acta psychiatrica Scandinavica Supplementum. 215: 1–98.

RI PT

2. Pirnia T, Joshi SH, Leaver AM, Vasavada M, Njau S, Woods RP, et al. (2016): Electroconvulsive therapy and structural neuroplasticity in neocortical, limbic and paralimbic cortex. Translational Psychiatry. 6: e832. 3. Joshi SH, Espinoza RT, Pirnia T, Shi J, Wang Y, Ayers B, et al. (2015): Structural plasticity of the hippocampus and amygdala induced by electroconvulsive therapy in major depression. Biological Psychiatry. 79: 282–92.

SC

4. Wade BSC, Joshi SH, Njau S, Leaver AM, Vasavada M, Woods RP, et al. (2016): Effect of Electroconvulsive Therapy on Striatal Morphometry in Major Depressive Disorder. Neuropsychopharmacology. 41: 2481–2491. 5. Bates, Douglas, Mächler, Matin, Bolker, Ben, Walker, Steve (2015): Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. 67. doi: 10.18637/jss.v067.i01.

M AN U

6. Whitcher B, Schmid VJ, Thornton A (2011): Working with the DICOM and NIfTI Data Standards in R. Journal of Statistical Software. 44. doi: 10.18637/jss.v044.i06. 7. Kenward MG, Roger JH (1997): Small Sample Inference for Fixed Effects from Restricted Maximum Likelihood. Biometrics. 53: 983–997. 8. Satterthwaite FE (1946): An Approximate Distribution of Estimates of Variance Components. Biometrics Bulletin. 2: 110–114. 9. Oltedal L, Bartsch H, Sørhaug OJE, Kessler U, Abbott C, Dols A, et al. (2017): The Global ECT-MRI Research Collaboration (GEMRIC): Establishing a multi-site investigation of the neural mechanisms underlying response to electroconvulsive therapy. Neuroimage Clin. 14: 422–432.

TE D

10. Moeller JR, Ishikawa T, Dhawan V, Spetsieris P, Mandel F, Alexander GE, et al. (1996): The metabolic topography of normal aging. J Cereb Blood Flow Metab. 16: 385–398. 11. Zhang N, Gordon ML, Ma Y, Chi B, Gomar JJ, Peng S, et al. (2018): The Age-Related Perfusion Pattern Measured With Arterial Spin Labeling MRI in Healthy Subjects. Front Aging Neurosci. 10. doi: 10.3389/fnagi.2018.00214.

EP

12. Wang Z, Aguirre GK, Rao H, Wang J, Fernández-Seara MA, Childress AR, Detre JA (2008): Empirical optimization of ASL data analysis using an ASL data processing toolbox: ASLtbx. Magnetic Resonance Imaging. 26: 261–269.

AC C

13. Eklund A, Nichols TE, Knutsson H (2016): Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proc Natl Acad Sci USA. 113: 7900–7905. 14. Li H, Nickerson LD, Nichols TE, Gao J-H (2017): Comparison of a non-stationary voxelation-corrected cluster-size test with TFCE for group-Level MRI inference. Hum Brain Mapp. 38: 1269–1280. 15. Mueller K, Lepsien J, Möller HE, Lohmann G (2017): Commentary: Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Front Hum Neurosci. 11: 345. 16. Poldrack RA, Baker CI, Durnez J, Gorgolewski KJ, Matthews PM, Munafò MR, et al. (2017): Scanning the horizon: towards transparent and reproducible neuroimaging research. Nat Rev Neurosci. 18: 115–126. 17. Worsley KJ, Evans AC, Marrett S, Neelin P (1992): A three-dimensional statistical analysis for CBF activation studies in human brain. Journal of Cerebral Blood Flow and Metabolism. 12: 900–18. 18. Burnham KP, Anderson DR (2004): Multimodel Inference: Understanding AIC and BIC in Model Selection. Sociological Methods & Research. 33: 261–304.

7

ACCEPTED MANUSCRIPT Leaver et al.

Supplement

19. Sheline YI, Sanghavi M, Mintun MA, Gado MH (1999): Depression Duration But Not Age Predicts Hippocampal Volume Loss in Medically Healthy Women with Recurrent Major Depression. J Neurosci. 19: 5034–5043. 20. Oltedal et al. (In Press): Volume of the Human Hippocampus and Clinical Response Following Electroconvulsive Therapy. Biological Psychiatry. . doi: 10.1016/j.biopsych.2018.05.017.

RI PT

21. Dukart J, Regen F, Kherif F, Colla M, Bajbouj M, Heuser I, et al. (2014): Electroconvulsive therapyinduced brain plasticity determines therapeutic outcome in mood disorders. Proceedings of the National Academy of Sciences of the United States of America. 111: 1156–61. 22. Joshi SH, Espinoza RT, Pirnia T, Shi J, Wang Y, Ayers B, et al. (2015): Structural plasticity of the hippocampus and amygdala induced by electroconvulsive therapy in major depression. Biological Psychiatry. 79: 282–92.

SC

23. Redlich R, Bürger C, Dohm K, Grotegerd D, Opel N, Zaremba D, et al. (2017): Effects of electroconvulsive therapy on amygdala function in major depression - a longitudinal functional magnetic resonance imaging study. Psychol Med. 47: 2166–2176. 24. Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, Lanctôt KL (2010): A meta-analysis of cytokines in major depression. Biol Psychiatry. 67: 446–457.

M AN U

25. Duman RS, Aghajanian GK, Sanacora G, Krystal JH (2016): Synaptic plasticity and depression: new insights from stress and rapid-acting antidepressants. Nat Med. 22: 238–249. 26. Miller AH, Maletic V, Raison CL (2009): Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biol Psychiatry. 65: 732–741.

AC C

EP

TE D

27. van Buel EM, Patas K, Peters M, Bosker FJ, Eisel ULM, Klein HC (2015): Immune and neurotrophin stimulation by electroconvulsive therapy: is some inflammation needed after all? Transl Psychiatry. 5: e609.

8