Article Figures & Data
Figures
- Figure 1.
Overview of analysis procedure. The analysis for patients and controls is used as an example, but steps were the same for analyses with controls only and patients only. a, The fMRI data from resting state, affective go/no-go, and anhedonia (or monetary incentive delay) tasks were preprocessed using the OPPNI pipeline (Churchill et al., 2015, 2017), parcellated into 326 regions and concatenated for each participant and session. b, FC was calculated as the correlation between the time courses of each region pair, resulting in a region-by-region FC matrix. c, For the whole-brain analyses, the upper triangle (gray) of the FC matrix from each participant and session was correlated with that of every other participant and session to create a similarity matrix. d, For the region-specific analyses, a similarity matrix was constructed for each ROI by correlating each ROI's row in the correlation matrix with that of the same row of other participants and sessions. e, The contribution of each source of variance was estimated by calculating the average similarity over different configurations of the similarity matrix. From left to right, the presented patterns display the configuration for the calculation of the common effect (similar FC across all participants and sessions), the session effect (similar FC across participants within sessions), the sex effect (similar FC among female and among male participants), the MD effect (similar FC among patients with an MD diagnosis and among controls), the MD × session interaction (similar FC among patients and controls within sessions), the MD × sex interaction (similar FC among female patients, male patients, female controls, and male controls), and the individual effect (similar FC within individuals across sessions). f, From the average similarity for each effect, we calculated the normalized relative effect magnitude by first subtracting the baseline for each effect (indicated by the dashed red lines) and then dividing by the total relative similarity. For the region-specific analyses, we visualized the normalized relative effect magnitudes but did not perform statistical comparisons. g, For the whole-brain analyses, we also calculated the average similarity for each effect per individual by taking the average of the patterns for the three rows representing each participant. The example shows this for the MD effect. For each participant, indicated by different color outlines, we calculated the average similarity of that participant to all other participants and sessions within the same group (i.e., by taking the average across black squares within the three rows representing the three recording sessions of that participant). This way, we calculated each of the effects shown in part e of the figure at an individual level as well. These participant-specific average similarity values were then entered into dependent samples t tests to examine differences in effect magnitudes. Each capital letters (A–F) represent individuals and the lowercase letters (x–z) the different effects (i.e., Ax is the MD effect magnitude for participant 1, Az is the effect magnitude for a different effect for that same participant). FC, functional connectivity; ROI, region of interest; MD, major depressive disorder.
- Figure 2.
Contributions of different sources of variance in FC in A, controls only, B, patients and controls, and C, patients only. a, Similarity matrix displaying the Fisher transformed correlations between the whole-brain FC of different participants and sessions. The matrices are organized by group (sex and MD diagnosis or response), individual, and session (baseline, Week 2, and Week 8) as illustrated in Figure 1. b, Average similarity for each effect. Figure 1e illustrates how each average was calculated. The red dashed lines indicate the baseline for each effect, which was subtracted to calculate the normalized relative effect magnitudes. c, Normalized relative effect magnitude for each effect.
- Figure 3.
Correlation between parcel-level connectivity calculated using split-half sampling of data for control participants, where half of the data (∼45 min) was compared with increasingly longer segments of data from the other half. Colored lines represent individual participants, while the black dotted line represents the average across participants.
- Figure 4.
Normalized relative effect magnitude in each region of the brain for each effect (common, MD, sex, MD × sex, and individual) in patients and controls. The MD, sex, and MD × sex effects are presented on a narrower color scale so the localization of these effects can be distinguished.
- Figure 5.
Normalized relative effect magnitude in each region of the brain for each effect (common, response, sex, response × sex, and individual) in patients only. The response, sex, and response × sex effects are presented on a narrower color scale so the localization of these effects can be distinguished.
- Figure 6.
Illustrations of the similarity matrix and matrix configurations used to calculate the averages for each effect in the exploratory analyses, taking the patients and controls sample as the example. a, Structure of the similarity matrix exemplified for eight participants. There is now a separate row for each participant, session, and task (instead of just for each participant and session in the main analyses). b, Matrix configuration for each of the effects that were also examined in the main analyses (Fig. 1e). The effects were calculated by averaging over the black squares in the matrix. From left to right and top to bottom, the patterns represent the common effect (similar FC across all participants and sessions), the session effect (similar FC across participants within sessions), the sex effect (similar FC among female and among male participants), the MD effect (similar FC among patients with an MD diagnosis and among controls), the MD × session interaction (similar FC among patients and controls within sessions), the MD × sex interaction (similar FC among female patients, male patients, female controls, and male controls), and the individual effect (similar FC within individuals across sessions). c, Matrix configurations of the effects only examined in the exploratory analyses. The individual × time interaction (similar FC within individual and session) is illustrated on the left, while the individual × task interaction (similar FC within individual and task) is illustrated on the right.
- Figure 7.
Contributions of different sources of variance in FC as examined in our exploratory analyses in A, controls only, B, patients and controls, and C, patients only. a, Similarity matrix displaying the Fisher transformed correlations between the whole-brain FC of different participants and sessions. The matrices are organized by group (sex and MD diagnosis or response), individual, task (resting state, affective go/no-go, and anhedonia), and session (baseline, Week 2, and Week 8) as illustrated in Figure 6. b, Average similarity for each effect. See Figure 6 for the pattern across which each average was calculated. The red dashed lines indicate the baseline for each effect, which were subtracted to calculate the normalized relative effect magnitudes. c, Normalized relative effect magnitude for each effect.
- Figure 8.
Normalized relative effect magnitude in each region of the brain for selected effects from the exploratory analyses (common, individual, individual × task, and individual × session) in A, patients and controls and B, patients only.
Tables
- Table 1.
Demographic characteristics of female and male patients with MD and controls (mean ± SEM), and their statistical differences
Patients with MD (N = 68) Controls (N = 39) Statistical difference Female (N = 42) Male (N = 26) Female (N = 21) Male (N = 18) Age (years) 34.36 ± 1.81 (18–59) 38.46 ± 2.46 (18–59) 33.00 ± 2.59 (18–60) 34.89 ± 3.02 (21–57) MD: F(1,103) = 1.00, p = 0.32; sex: F(1,103) = 1.48, p = 0.23; interaction: F(1,103) = 0.65, p = 0.65 Education (years) 17.00 ± 0.26 (14–20) 17.38 ± 0.38 (14–20) 18.71 ± 0.38 (14–22) 18.22 ± 0.53 (13–22) MD: F(1,103) = 11.34, p = 0.001; sex: F(1,103) = 0.02, p = 0.89; interaction: F(1,103) = 1.34, p = 0.25 Race/ethnicitya Black (2); East Asian (4); Jewish (1); Latin American/Hispanic (2); South Asian (1); Southeast Asian (2); White (34); Other (2) Latin American/Hispanic (1); South Asian (2); Southeast Asian (1); White (22); Other (1) East Asian (5); South Asian (3); Southeast Asian (2); White (12) Latin American/Hispanic (1); South Asian (3); White (15) NA Site CAM (4); MCU (7); QNS (2); UBC (20); UCA (9) MCU (6); QNS (5); TGH (1); UBC (7); UCA (7) CAM (2); MCU (3); QNS (6); TGH (2); UBC (4); UCA (4) CAM (3); MCU (3); QNS (1); TGH (1); UBC (2); UCA (8) NA Handedness Left (2); right (38); ambidextrous (2) Left (5); right (21) Left (2); right (19) Left (5); right (10); ambidextrous (3) NA Current marital status Never married (22); separated (1); married (9); divorced (4); domestic partnership (5); widowed (1) Never married (14); separated (2); married (7); divorced (2); domestic partnership (1) Never married (12); married (5); domestic partnership (4) Never married (11); married (5); divorced (1); domestic partnership (1) NA MD, major depressive disorder; CAM, Centre for Addiction and Mental Health; MCU, McMaster University; UBC, University of British Columbia; TGH, Toronto General Hospital; QNS, Queen's University; UCA, University of Calgary, NA, not applicable.
↵a Participants were asked which of the presented race/ethnicity categories they most closely identified with. Categories were based on Canadian census questionnaires (Statistics Canada, 2006).
- Table 2.
Clinical and demographic characteristics of female and male responders and nonresponders to antidepressant pharmacotherapy (mean ± SEM), and their statistical differences
Responders (N = 33) Nonresponders (N = 35) Statistical difference Female (N = 19) Male (N = 14) Female (N = 23) Male (N = 12) MADRS at baseline 28.16 ± 1.00 28.93 ± 1.07 30.39 ± 1.32 30.00 ± 1.33 Response: F(1,64) = 1.67, p = 0.20; sex: F(1,64) = 0.02, p = 0.88; interaction: F(1,64) = 0.21, p = 0.65 MADRS at Week 2 18.11 ± 1.73 18.57 ± 1.73 24.48 ± 1.27 23.50 ± 1.78 Response: F(1,64) = 11.61, p = 0.001; sex: F(1,64) = 0.02, p = 0.88; interaction: F(1,64) = 0.19, p = 0.66 MADRS at Week 8 8.21 ± 1.15 7.07 ± 1.49 20.61 ± 1.09 22.92 ± 1.35 Response: F(1,64) = 119.95, p < 0.001; sex: F(1,64) = 0.21, p = 0.65; Interaction: F(1,64) = 1.79, p = 0.19 Age (years) 34 ± 2.49 (19–55) 38.36 ± 3.44 (18–59) 34.65 ± 2.63 (18–59) 38.58 ± 3.68 (20–59) Response: F(1,64) = 0.02, p = 0.89; sex: F(1,64) = 1.84, p = 0.18; interaction: F(1,64) < 0.01, p = 0.94 Education (years) 17.21 ± 0.43 (14–20) 17.29 ± 0.61 (14–20) 16.83 ± 0.32 (14–19) 17.50 ± 0.45 (16–20) Response: F(1,64) = 0.04, p = 0.85; sex: F(1,64) = 0.68, p = 0.41; interaction: F(1,64) = 0.43, p = 0.51 Race/ethnicitya Black (1); East Asian (3); South Asian (1); White (15); Other Latin American/Hispanic (1); South Asian (1); White (12); Other (1) Black (1); East Asian (1); Jewish (1); Latin American/Hispanic (2); Southeast Asian (2); White (19); Other (1) South Asian (1); Southeast Asian (1); White (10) NA Site CAM (2); MCU (2); UBC (9); UCA (6) MCU (3); QNS (3); TGH (1); UBC (4); UCA (3) CAM (2); MCU (5); QNS (2); UBC (11); UCA (3) MCU (3); QNS (2); UBC (3); UCA (4) NA Handedness Left (2); right (16); ambidextrous (1) Left (3); right (11) Right (22); ambidextrous (1) Left (2); right (10) NA Current marital status Never married (11); married (5); divorced (2); domestic partnership (1) Never married (8); separated (1); married (3); divorced (1); domestic partnership (1) Never married (11); separated (1); married (4); divorced (2); domestic partnership (4); widowed (1) Never married (6); separated (1); married (4); divorced (1) NA MADRS, Montgomery–Åsberg Depression Rating Scale; MD, major depressive disorder; CAM, Centre for Addiction and Mental Health; MCU, McMaster University; UBC, University of British Columbia; TGH, Toronto General Hospital; QNS, Queen's University; UCA, University of Calgary, NA, not applicable.
↵a Participants were asked which of the presented race/ethnicity categories they most closely identified with. Categories were based on Canadian census questionnaires (Statistics Canada, 2006).
- Table 3.
Results of paired samples t tests to compare magnitude of effects in controls only, patients and controls, and patients only
Sample Comparison t-value DF SD Uncorrected p-value Cohen's d Controls only Common > session 9.32 38 0.004 <0.001* −1.49 Common < sex −7.85 38 0.006 <0.001* 1.26 Session < sex −10.48 38 0.008 <0.001* 1.68 Sex < individual −15.27 38 0.149 <0.001* 2.44 Patients and controls Common < MD −6.52 106 0.006 <0.001* 0.63 Common > session 6.72 106 0.002 <0.001* −0.65 Common < sex −7.35 106 0.006 <0.001* 0.71 MD > session 8.20 106 0.007 <0.001* −0.79 MD = sex −0.62 106 0.009 0.530 0.06 Session < sex −8.69 106 0.007 <0.001* 0.84 MD > MD × session interaction 11.89 106 0.003 <0.001* −1.15 Sex < MD × sex interaction −7.44 106 0.008 <0.001* 0.72 MD × session interaction < MD × sex interaction −12.11 106 0.009 <0.001* 1.17 MD × sex interaction < individual −25.40 106 0.145 <0.001* 2.46 Patients only Common < Response −3.95 67 0.01 <0.001* 0.48 Common > session 9.72 67 0.00 <0.001* −1.18 Common < sex −4.88 67 0.01 <0.001* 0.59 Response > session 6.68 67 0.01 <0.001* −0.81 Response = sex −1.41 67 0.01 0.179 0.17 Session < sex −6.41 67 0.01 <0.001* 0.78 Response > response × session interaction 13.09 67 0.00 <0.001* −1.59 Sex < response × sex interaction −4.71 67 0.02 <0.001* 0.57 Response × session interaction < response × sex interaction −7.45 67 0.02 <0.001* 0.90 Response × sex interaction < individual −20.18 67 0.14 <0.001* 2.45 ↵* Significant effects after FDR correction.
- Table 4.
Results of paired samples t tests to compare magnitude of effects in our exploratory analyses in controls only, patients and controls, and patients only
Sample Comparison t-value DF SD Uncorrected p-value Cohen's d Controls only Common > session 6.75 38 0.001 <0.001* −1.08 Common < sex −5.95 38 0.006 <0.001* 0.95 Session < sex −6.67 38 0.006 <0.001* 1.07 Sex < individual −16.66 38 0.071 <0.001* 2.67 Individual = Individual × session interaction 0.02 38 0.033 0.98 0.00 Individual < Individual × task interaction −12.46 38 0.040 <0.001* 1.99 Individual × session interaction < Individual × task interaction −7.22 38 0.069 <0.001* 1.16 Patients and controls Common < MD −5.54 106 0.004 <0.001* 0.54 Common = session 0.01 106 0.001 0.99 0.00 Common < sex −7.55 106 0.004 <0.001* 0.73 MD > session 5.38 106 0.004 <0.001* −0.52 MD = sex −1.80 106 0.006 0.072 0.17 Session < sex −7.24 106 0.005 <0.001* 0.70 MD > MD × session interaction 4.16 106 0.001 <0.001* −0.40 Sex < MD × sex interaction −6.58 106 0.006 <0.001* 0.64 MD × session interaction < MD × sex interaction −9.39 106 0.006 <0.001* 0.91 MD × sex interaction < individual −25.64 106 0.076 <0.001* 2.48 Individual = Individual × session interaction −2.01 106 0.037 0.051 0.19 Individual < Individual × task interaction −15.76 106 0.050 <0.001* 1.52 Individual × session interaction < Individual × task interaction −8.76 106 0.082 <0.001* 0.85 Patients only Common < Response −6.04 67 0.004 <0.001* 0.73 Common = session 0.64 67 0.001 0.53 −0.08 Common < sex −5.43 67 0.006 <0.001* 0.66 Response > session 6.05 67 0.004 <0.001* −0.73 Response = sex −1.90 67 0.007 0.058 0.23 Session < sex −5.28 67 0.007 <0.001* 0.64 Response > Response × session interaction 3.73 67 0.001 0.002* −0.45 Sex < Response × sex interaction −4.18 67 0.012 0.001* 0.51 Response × session interaction < Response × sex interaction −5.08 67 0.014 <0.001* 0.62 Response × sex interaction < individual −19.07 67 0.079 <0.001* 2.31 Individual < Individual × session interaction −2.43 67 0.038 0.016* 0.29 Individual < Individual × task interaction −11.12 67 0.056 <0.001* 1.35 Individual × session interaction < Individual × task interaction −5.92 67 0.089 <0.001* 0.72 ↵* Significant effects after FDR correction.
Analysis scripts
Download Analysis scripts, ZIP file.
In this issue
