Table of Contents for Supplementary Online Content Mechanisms of cognitive-behavioral therapy for obsessive-compulsive disorder involve robust and extensive increases in brain network connectivity Protocol for OCD Treatment with CBT Supplementary Protocol 1. Cognitive-behavioral therapy (CBT) treatment for obsessive- compulsive disorder (OCD) NBS Results at Different Thresholds for OCT post-CBT > pre-CBT Supplementary Figure 1. Networks Stronger post-CBT at Different t-stat thresholds Supplementary Table 1. Network connections and t-stat values corresponding to Supplementary Figure 1 OCD Recruitment Flow Supplementary Figure 2. CONSORT Diagram OCD Psychometric Assessments Supplementary Table 2. Psychometrics for OCD Sample Including YBOCS Item Scores OCD Symptom Dimensions Supplementary Figure 3. OCD Symptom Dimensions OCI-R scores NBS Connectivity Identified by Intrinsic Networks Supplementary Figure 4. Canonical Network Membership of Nodes from the NBS Networks in OCD that Demonstrated Significantly Stronger Connectivity pre- to post-CBT NBS Results at Different Thresholds for Other Comparisons Supplementary Figure 5. Healthy Control 4-Week Comparison At Lower Threshold Supplementary Figure 6. AOCD vs CON pre-CBT Using a Lower Threshold OCD vs Control Connectivity Supplementary Figure 7. Networks with Stronger Connectivity for OCD post-CBT than HC 1
Transcript
Table of Contents for Supplementary Online Content
Mechanisms of cognitive-behavioral therapy for obsessive-compulsive disorder involve robust and extensive increases in brain network connectivity
Protocol for OCD Treatment with CBT
Supplementary Protocol 1. Cognitive-behavioral therapy (CBT) treatment for obsessive-
compulsive disorder (OCD)
NBS Results at Different Thresholds for OCT post-CBT > pre-CBT
Supplementary Figure 1. Networks Stronger post-CBT at Different t-stat thresholds
Supplementary Table 1. Network connections and t-stat values corresponding to
Supplementary Figure 1
OCD Recruitment Flow
Supplementary Figure 2. CONSORT Diagram
OCD Psychometric Assessments
Supplementary Table 2. Psychometrics for OCD Sample Including YBOCS Item Scores
Two licensed therapists with extensive training in CBT for OCD conducted the treatment. Both had received 3 or more years of specialty training in the UCLA OCD Intensive Treatment Program under the supervision of JDF and other senior therapists in addition to 3 or more years of training in outpatient CBT for OCD. Therapy sessions of participants who provided consent (61%) were videotaped, and an independent evaluator (also a trained CBT therapist with 8 years experience treating OCD) rated all sessions for quality assurance, including adherence to the treatment manual and overall quality of the session. Average treatment adherence was rated at 97.7% and average quality of sessions was rated at 9.96 (0 to 10).
The therapists followed the treatment protocol for Exposure and Response Prevention (ERP, a type of CBT for OCD) based on the manual by Kozak & Foa 2. Each OCD participant was treated one-on-one by one of the two study therapists, who were experienced in ERP for OCD. The ERP sessions were 90 minutes each in duration, and were 5 days per week (Monday through Friday) for 4 weeks, for a total of 20 sessions. Participants were told that they could not be more than 10 minutes late to sessions or miss sessions. The study psychiatrist (JDF, or JL as backup) also met with each participant once weekly for 20 minutes. At each of these visits, if the participant was taking a stable dose of a serotonin reuptake inhibitor prior to enrollment, the psychiatrist assessed if he/she maintained the same dose of medication, and assessed for medication adherence. During the last week of treatment, the study therapist and study psychiatrist assisted the participant in referrals to outpatient treatment.
Outlines of the content of the therapy by session number are as follows:
Sessions 1 and/or 2:• Rapport building• Patient history• Ascertainment of level of knowledge about OCD• Facilitated discussion about the impact of OCD on their lives• Description of OCD as a neurobehavioral disorder• Demonstration of use of monitoring forms of symptoms, daily schedule, and
structure• Homework assignment: monitoring, or listing of obsessions and compulsions,
and/or reading• Reading material provided
Sessions 2 and/or 3:• Homework review• Identifying obsessions and compulsions• Distinguishing obsessions and compulsions from other problems• Description of the cycle of obsessions and compulsions• Rationale and description of ERP• Visual presentation and explanation of the course of OCD with and without
response prevention• Explanation of subjective units of distress (SUDS) graph• Creation of hierarchy• Homework assignment: self-monitoring of obsessions and compulsions, reading
Sessions 4/5 through 18/19:
3
• Review of significant events since last session• Homework review
Cognitive-behavioral therapy (CBT) treatment for obsessive-compulsive disorder (OCD) – cont.
• Cognitive restructuring: a) to manage anxiety and resist compulsions if necessary; b) for reappraisal after exposure to consolidate learning; c) discussion of when it is not appropriate to use cognitive restructuring techniques (during exposures)
• Exposures exercises in session (in vivo or imaginal)• Homework assignment: specific exposure exercises, self monitoring using SUDS
Sessions 19/20:• Homework review• Assessment of progress: review progress on hierarchy, discuss improvements
overall with obsession thoughts and compulsive behaviors, improvements in overall functioning, improvements in overall anxiety and mood
• Current symptomatology and course of treatment: discuss and reinforce participant’s positive changes made during treatment (symptom reduction and better functioning)
• Inquiry of what participant learned from treatment• Discuss relapse prevention: future plans for continuing treatment with outpatient
CBT therapist; recognizing and dealing with symptoms as they arise; soliciting help from family or friends when necessary; increasing activities (e.g. work, school, relationships, hobbies); goals for 1 months, 3 months, 6 months, and 12 months in the future.
• Address termination • Discussion of importance of follow-up treatment
Independent evaluators
Independent evaluators not involved in treatment or assessments administered psychometric instruments. Independent evaluators were PhD-level psychotherapists specializing in OCD with several years experience with CBT as performed in the trial and supervised weekly by JDF. Estimated reliability between evaluators was high (Intraclass Correlation Coefficient of 0.74, 95% confidence interval: -0.87,1.00).
4
NBS Results at Different Thresholds for OCD post-CBT > pre-CBT
Supplementary Figure 1. Networks Stronger in OCD post-CBT at Different t-stat ThresholdsA. Connection and T-statistic results for NBS networks thresholded at t=5.5. B. Connection and T-statistic results for NBS networks thresholded at t=6.5. Not shown, for NBS networks thresholded at t=4.0, the result was a single network with 1167 connections and 155 nodes. See following Supplementary Table 1 for connections and t-stat values.
Supplementary Figure 2. CONSORT Diagram: Study Design and CONSORT Flow Diagram for OCD Patients and Healthy Controls
Abbreviations: OCD indicates meeting DSM-IV criteria for OCD; HC, healthy comparison group; TREAT, OCD arm that received treatment and no waitlist; WAIT, OCD arm that received waitlist first and then treatment; CBT, cognitive behavioral therapy; MRI, magnetic resonance imaging
Flow of OCD Patients and Healthy Controls Through the Study of Connectivity Changes before and after CBT treatment.
9
OCD Psychometric Assessments
Psychometric pre-CBT post-CBT
YBOCS Total Score 24.5 (4.7) 14.8 (5.2)***
YBOCS Item subscores:
Obsessions subscores
1. Time occupied with obsessions 2.6 (0.9) 1.8 (0.8)***
2. Interference 2.4 (0.7) 1.5 (0.7)***
3. Distress 2.4 (0.7) 1.7 (0.7)***
4. Resistance 2.0 (0.8) 1.2 (0.9)***
5. Control 2.5 (0.7) 1.7 (0.8)***
Compulsions subscores
6. Time occupied with compulsions 2.6 (0.8) 1.6 (0.7)***
7. Interference 2.3 (0.7) 1.1 (0.8)***
8. Distress 2.7 (0.6) 1.7 (0.7)***
9. Resistance 2.2 (0.9) 1.1 (0.7)***
10. Control 2.8 (0.6) 1.5 (0.7)***
HAMA 12.4 (5.4) 8.4 (5.1)***
MADRS 15.3 (9.5) 10.8 (8.9)***
GAS 57.6 (8.5) 69.6 (13.3) ***
WASI 108.2 (9.1) --
Supplementary Table 2. Group-Mean (SD) Psychometrics for OCD Sample Including YBOCS Item Scores
Abbreviations: SD, standard deviation; OCD, obsessive-compulsive disorder; YBOCS, Yale-Brown Obsessive-Compulsive Scale; CBT, cognitive-behavioral therapy; HAMA, Hamilton Anxiety Rating Scale; MADRS, Montgomery-Åshberg Depression Rating Scale; WASI, Wechsler Abbreviated Scale of Intelligence (administered only pre-CBT). ***P<0.001 for pre-CBT versus post-CBT, paired t-test.
Supplementary Figure 3. OCD Symptom Dimensions OCI-R scores OCD participants showed a relatively even distribution of symptoms across domains, with washing, doubting, and ordering showing the highest averages. All dimensions showed statistically significant improvement (p<0.01), except hoarding which showed a trend for improvement (p=0.10), 2-sample t-tests.
11
NBS Connectivity Identified by Intrinsic Networks
Supplementary Figure 4. Canonical Network Membership of Nodes from the NBS Networks in OCD that Demonstrated Significantly Stronger Connectivity pre- to post-CBT. Depicted are
12
the eight networks identified with the network based statistic (NBS) that significantly increased in functional connectivity strength pre- to post-CBT in the OCD group, with the nodes color-coded corresponding to the 6 previously identified functional connectivity networks for the node set we used: cingulo-opercular, fronto-parietal, default mode, sensorimotor, occipital, and cerebellum.1 In all 8 networks connectivity strength increased across two or more canonical networks.
Abbreviations: L, left; R, right; L.Crus I, L. cerebellum crus I; L.frPole, L. frontal pole; L.ICC, L. intracalcarine cortex; L.Insula, L. posterior insula; L.Lingual, L. lingual gyrus; L.LOC, L. superior lateral occipital; L.parOper, L. parietal operculum; L.PCC, L. posterior cingulate; L.pCun, L. precuneus; L.Put, L. putamen; L.spLobule, L. superior parietal lobule; L.Thal, L. thalamus, caudate; L.V, L. cerebellum V; L.VI, L. cerebellum VI; R.Crus I, R. cerebellum crus I; R.DAC, R. dorsal anterior cingulate; R.frPole, R. frontal pole; R.ICC, R. intracalcarine cortex; R.IFG, R. inferior frontal gyrus, precentral gyrus; R.MFG, R. middle frontal gyrus; R.pCun, R. precuneus; R.STG, R. superior temporal gyrus; R.Thal, R. thalamus.
13
NBS Results at Different Thresholds for Other Comparisons
Supplementary Figure 5. Healthy Control 4-Week Comparison At Lower Statistical Threshold
Only at a very low statistical threshold were there network differences in healthy controls from baseline to rescan after 4-weeks, using a t-statistic of 0.1 and P value=0.5. Even at that, an effect was observed in only a single network, between right caudate and left thalamus/caudate
14
NBS Results at Different Thresholds for Other Comparisons – continued
Supplementary Figure 6. OCD vs healthy controls pre-CBT using a Lower Statistical Threshold
At a low statistical threshold there were significant differences between OCD and HC pre-CBT. For HC>OCD, using FDR, with 10,000 permutations and a significance of p=0.1, a single network is revealed between L IPL and L precentral gyrus, (node 63 to 137). A regressor for medication status was used. We did not find a network for OCD > HC at this liberal significance threshold.
15
OCD vs Control Connectivity
Supplementary Figure 7. Networks with stronger connectivity for OCD post-CBT than healthy controls corresponding to Supplementary Table 3
Abbreviations: L.Angular, L Angular gyrus; L.COC, L Central opercular cortex; L.LOC, L Superior lateral occipital; L.PPCG, L Precentral and postcentral gyrus; L.V, L Cerebellum V; R.ICC, R Intracalcarine; R.PreCG, R Precentral gyrus; R.PPCO, R Precentral and postcentral gyrus, central opercular; R.SPL, R Superior parietal lobule and postcentral gyrus; R.VI, R Cerebellum VI; SMA, Bilateral supplementary motor cortex; V.VI, Bilateral Vermis VI.
(Left panel) Semi-opaque volumetric rendering of MNI atlas brain in left-sagittal (left column), axial (middle column) and coronal (right column) views showing 4 networks (top-to-bottom rows; 12 connections total) identified by NBS analysis as having significantly (t-threshold=4, P<0.05) stronger group-mean functional connectivity in post-CBTOCD participants than in control participants scanned after Week4. (For anatomic sites and t-statistics see Supplementary Table 3).
(Right panel) All 12 connections plotted together in axial view. There were no significant between-group differences in connectivity prior to CBT, and no networks with weaker connectivity in post-CBT OCD versus Week4 controls.
16
3
2
1
Supplementary Table 3. Network connections with stronger connectivity for OCD post-CBT than healthy controls
Right Caudate 14 6 7 Right Intracalcarine 9 -76 14 4.24
Left Superior lateral occipital -36 -69 40 Right Intracalcarine 9 -76 14 4.47
Left Thalamus -12 -3 13 Right Intracalcarine 9 -76 14 4.01
Network 2
Bilat Paracingulate, ACC 0 15 45 Right
Superior parietal lobule, postcentral gyrus
34 -39 65 4.56
Left Central opercular cortex -48 6 1 Right
Superior parietal lobule, postcentral gyrus
34 -39 65 4.31
Network 3
Left Angular gyrus -41 -47 29 Right Cerebellum VI 21 -64 -22 4.08
Left Angular gyrus -41 -47 29 Bilat Vermis VI 1 -66 -24 4.52
Bilat Supplementary motor cortex 0 -1 52 Bilat Vermis VI 1 -66 -24 4.80
Network 4
Left Cerebellum V -6 -60 -15 Right Precentral gyrus 46 -8 24 4.50
Left Cerebellum V -6 -60 -15 Right
Precentral gyrus, postcentral gyrus, central opercular cortex
58 -3 17 4.37
Left Cerebellum V -6 -60 -15 Left Precentral gyrus, postcentral gyrus -54 -9 23 4.42
Supplementary Table 3. Network connections with stronger connectivity for OCD post-CBT than healthy controls corresponding to Supplementary Figure 7Abbreviations: OCD, obsessive-compulsive disorder; HC, healthy comparison; BOLD, blood-oxygenation-level dependent effect; fMRI, functional magnetic resonance imaging; NBS, network based statistic; Bilat, bilateral.
Between OCD and HC, NBS compared connectivity strength (time-series correlations between nodes) of resting-state BOLD fMRI at all pairs of 160 functional nodes for post-CBT versus HC. For 4 networks, comprising 12 connections, connectivity was significantly stronger for OCD (t-threshold=4, P=0.05); no node pairs had significantly stronger connectivity for the HC group compared to the OCD group.
17
Supplementary Table 4. r and P-values for correlations between increase in connection strength and improvement in YBOCS resistance to obsession scores (YBOCS item #9) and control over obsession scores (YBOCS item #10)
YBOCS item #9 YBOCS item #10
Hemi Region Hemi Region r P-value r P-valueNetwork 1
R Dorsal anteriorcingulate L Lingual gyrus 0.03 0.86 0.29 0.06
Network 2L Frontal pole L Cerebellum Crus I 0.35 0.02 0.11 0.48
Network 3R Caudate L Cerebellum VI 0.12 0.46 0.17 0.27R Caudate R Intracalcarine cortex 0.18 0.25 0.11 0.50
Network 4L Putamen R Cerebellum Crus I 0.14 0.38 0.10 0.54L Putamen R Precuneus 0.17 0.29 0.34 0.02L Intracalcarine cortex R Precuneus 0.06 0.69 0.22 0.15R Precuneus L Thalamus, caudate 0.09 0.55 0.34 0.02L Posterior cingulate L Thalamus, caudate 0.19 0.23 0.29 0.06
Network 5R Middle frontal gyrus L Posterior insula 0.02 0.88 0.02 0.88
L Superior lateral occipital L Posterior insula 0.03 0.86 0.13 0.39
Network 6R Middle frontal gyrus L Cerebellum V 0.11 0.47 0.00 0.98
R Inferior frontal gyrus, precentral gyrus L Cerebellum V 0.26 0.09 0.21 0.18
R Frontal pole L Cerebellum V 0.03 0.86 0.08 0.59
R Superior temporal gyrus L Cerebellum V 0.11 0.50 0.09 0.55
R Superior temporal gyrus L Superior parietal lobule 0.10 0.52 0.05 0.76
Network 7R Frontal pole L Parietal operculum 0.51 <0.001* 0.09 0.58
Network 8L Precuneus R Thalamus 0.20 0.19 0.15 0.35
Supplementary Table 4. P-values for correlations between increase in connection strength and improvement in YBOCS resistance to obsession scores (YBOCS item #9) and control over obsession scores (YBOCS item #10). Within the OCD sample, NBS compared connectivity strength (time-series correlations between nodes) of resting-state BOLD fMRI at all pairs of 160 functional nodes for pre- versus post-CBT. For 8 networks, comprising 18 connections, connectivity was significantly stronger after CBT (t-threshold=4, P=0.01); no node pairs had significantly weaker connectivity after CBT. The R and P-values shown here are for correlations between connectivity changes at the edges shown and YBOCS subscores indexing resistance to compulsions (Item #9) and control over compulsions (Item #10). Coordinates for regions are shown in Table 2 of the Manuscript. * Significant correlation for Item #9, as shown in Figure 3A of manuscript.
We performed exploratory analyses to test whether pretreatment connectivity predicts improvements in anxiety (HAMA) or depression (MADRS) symptoms. Low pre-CBT connectivity strength between left precuneus and right thalamus was associated with improvement (greater decrease) in HAMA scores (r = 0.46, p=0.002; nodes centered at (-2, -75, 32) and (11, -12, 6)). Higher pretreatment connectivity between this node pair may presage superior response of anxiety symptoms to CBT. Regression results for MADRS did not pass correction for multiple comparisons.
19
Nodal connectivity analysis
We employed a node-to-node analysis using the Matlab REST toolbox (http://www.rest.restfmri.net) to
further investigate functional connectivity in CSTC nodes previously associated with OCD diagnoses
and symptomatology, because the widely-used node set identified by Dosenbach1, which we used for
the data-driven NBS analysis, did not include all the specific nodes required for this analysis.
HypothesesWe hypothesized that OCD would have stronger connectivity than healthy controls for two node pairs:
left caudate to left orbital frontal cortex and right caudate to right orbital frontal cortex 3,4,5. In addition,
we performed exploratory analysis for all other combinations of the 7 nodes.
MethodsSeed selection We used meta-analytic data from the Neurosynth database (http://www.neurosynth.org/) to define
seeds that have been implicated in neuroimaging studies of OCD. We identified 7 regions from the
Neurosynth mask generated using the forward inference and the search term “Obsessive Compulsive”
in the Neurosynth database and created 5mm radius spherical seeds for probing nodal functional
connectivity: left caudate nucleus (-14,12,4), left frontal orbital cortex (-40,26,-8), left paracingulate (-
4,22,38), left thalamus (-10,-18,10), right caudate (12,12,0), right frontal orbital cortex, (5,34,24,-6), right
paracingulate (6,22,38).
Functional connectivity analysisFunctional connectivity (FC) analysis was performed with Resting-State fMRI Data Analysis Toolkit
were used for seed-to-seed connectivity analysis. After 0.01–0.8 Hz band pass filtering, the time series
of voxels within each seed region was averaged as the seed reference time course. For each subject,
FC of each seed reference time course with other remaining 6 seeds were calculated separately to
obtain correlation coefficient (FC values) resulting a matrix of 7X7 seed-to-seed FC values for each
subject. Data of all connectivity matrices (HC: n = 24; OCD: n = 43) was then used in SPSS (IBM SPSS
Statistics for Windows, Version 24.0, Armonk, NY) for statistical analyses.
Nodal connectivity resultsAs it was reported in the main text, analysis of our two hypotheses did not reveal any significant result,
and only a trend for decreased connectivity between left caudate to left orbital frontal cortex after CBT
(p=0.052) was detected which did not survive multiple comparisons correction (Bonferroni corrected p-
20
value = 0.025). For the 19 exploratory node pairs, no connections survived FDR correction for
differences in strength between 7 nodes (control vs OCD pre-CBT or pre vs post CBT treatment for
OCD).
Nodal connectivity conclusionsThe lack of significant differences between controls and OCD pre-CBT is in agreement with our NBS
results in which we found no differences in functional connectivity between groups at baseline.
Comparing pre-CBT vs post-CBT nodal connectivity in the OCD group did not reveal any connectivity
changes in CTSC nodes. These results support the idea connectivity-strengthening processes may
reflect effects of CBT independent of normalization of pathophysiological circuits.
Supplementary Figure 9. Seeds (5mm radius spheres) derived from Neurosynth database, using “Obsessive Compulsive” forward inference.
21
NBS Functional Connectivity Videos
Supplementary Video 1. Video of Networks with Stronger Connectivity for OCD post-CBT vs pre-CBT
Video showing 3-D rendition of the 8 networks with stronger connectivity after CBT treatment. This video shows the connections described in Supplementary Figure 1 and Supplementary Table 1.
Supplementary Video 2. Video of Networks with Stronger Connectivity for OCD post-CBT than HC
Video showing 3-D rendition of the 4 networks with stronger connectivity after CBT treatment for the comparison of OCD vs control. This video shows the connections described in Supplementary Figure 7 and Supplementary Table 3.
References
1. Dosenbach NU, et al. Prediction of individual brain maturity using fMRI. Science (New York, NY) 329, 1358-1361 (2010).
2. Kozak MJ, Foa EB. Mastery of obsessive-compulsive disorder: a cognitive-behavioral approach client workbook. Oxford University Press (1997).
3. Hou JM, Zhao M, Zhang W, Song LH, Wu WJ, Wang J, et al. Resting-state functional connectivity abnormalities in patients with obsessive-compulsive disorder and their healthy first-degree relatives. Journal of Psychiatry and Neuroscience: 39(5), 304-311 (2014).
4. B.J. Harrison, C. Soriano-Mas, J. Pujol, H. Ortiz, M. Lopez-Sola, et al. Altered corticostriatal functional connectivity in obsessive-compulsive disorder Arch. Gen. Psychiatry, 66, 1189–1200 (2009).
5. Sakai Y, Narumoto J, Nishida S, Nakamae T, Yamada, K, et al. Corticostriatal functional connectivity in non-medicated patients with obsessive-compulsive disorder. Eur. Psychiatry, 26, 463–469 (2011).
![arXiv:1805.08298v2 [cs.CV] 24 Nov 2018epxing/papers/2018/Li_etal_nips18.pdfradiologists for the chest x-ray image on the left column. The right column contains three reports generated](https://static.documents.pub/doc/80x56/5e8fcaf26dd187783230ac64/arxiv180508298v2-cscv-24-nov-epxingpapers2018lietalnips18pdf-radiologists.jpg)
