Article Figures & Data
Figures
- Figure 1.
Economic task and STN LFP recordings. A, Letters, wins (W) and loss (L) probability (P) and value (€), and expected values (EV) associated with low-risk (LR) and high-risk (HR) options. B, Risk avoidance in GDPs (n = 6) and NGDPs (n = 6). Bars represent median confidence interval. Marker indicates significant difference (p = 0.013). C, Window of interest (3.3 s preceding options presentation). Top, sequence of visual stimuli in the window of interest. Bottom, processing of recorded LFP: raw data from right and left leads, and average LFP. D, 3D reconstruction of STN location of the STN (blue structures) and of the Medtronic 3389 DBS leads (red cylinders) with 0–3 contacts in one example GDP subject (#6). CA and CP indicate the anterior and posterior commissure.
- Figure 2.
2D reconstruction of recording coordinates. Location of STN (pale blue lines) and of Medtronic 3389 DBS contact 1 in GDP #6 (already shown in Fig. 1D). R, right; L, left; CA, anterior commissure; CP, posterior commissure; PM, middle point between CA and CP line; CO, MRI coronal view; SW, MRI sagittal view; AX, MRI axial (transverse) view; Sup, superior; Inf, inferior; A, anterior; P, posterior.
- Figure 3.
Risk avoidance bias precedes options presentation. A, Comparison of reaction times in GDP (n = 6) and NGDP (n = 6). From left to right: over all conditions, low-risk choice in conflictual (C-LR) and equivalent choice (EC-LR) trials, high-risk choice in equivalent choice (EC-HR) and conflictual (C-HR) trials. B, Relative reaction times (RT) for the different conditions pooling all sessions of each patient group. The dashed line indicates the median RT. C, Modulation of risk aversion due to previous outcome (PO) for GDPs (purple) and NGDPs (green). D, χ2 discrepancy between observed responses and expected responses in an equivalent memoryless Bernoulli process for GDPs (n = 6) and NGDPs (n = 6). Bars and error bars represent median and median confidence interval, respectively, for all panels. Horizontal line inserts highlight significant differences reporting p value (see Results for details) for all panels.
- Figure 4.
Spectral modulations of subthalamic nucleus local field potential and future decisions. A, Average power spectrum of STN LFP preceding options presentations over all trials for patients with gambling disorder (GDPs) and without (NGDPs). Dashed lines indicate beta band (12–30 Hz). B, Average power spectrum of STN LFP preceding options presentations over all low-risk (LR) and high-risk (HR) tasks for patients with gambling disorder (GDPs). C, Same as B for patients without gambling disorder (NGDPs). D, Mutual information between future choice and average power of the different bands for GDPs. White bars and error bars, respectively, represent median and 75th percentile of bootstrap information (see Methods). E, Same as D for NGDPs.
- Figure 5.
Subjectwise correlation between low-frequency STN LFP preceding options presentation and risk avoidance in GDPs. A–C, Comparison of the average value of GDPs LFP in the three intervals for the two conditions. Title reports significance of Wilcoxon rank test. D–F, Modulation from average risk avoidance associated with trials in which LFP averaged over the three different intervals belongs to the four percentiles. Markers and bars indicate medians and range over GDP. Circle indicates significant difference from average value (p < 0.05, Wilcoxon rank test).
- Figure 6.
Groupwise correlation between low-frequency STN LFP preceding options presentation and risk avoidance in GDPs. A, Grand average of z-scored LFP preceding task in which GDPs opted for HR (red) or LR (blue) option. Areas indicate median value confidence. Horizontal dotted line represents z = 0. Vertical dashed lines indicate the interval in which the outcome of previous choice, the accumulated money, and the blank screen are displayed. See Fig. 1C for details. B, Average value of GDPs LFP in the three intervals for the two conditions. Markers indicate p < 0.05 significant difference. C, Same as B, considering only trials in which the outcome of the previous choice was a loss. D, Same as B, considering only trials in which the outcome of the previous choice was a win. E, Mutual information between LFP levels in the different intervals and future choice (black bars). White bars and associated error indicate average and 75th percentile bootstrap information over 200 permutations (see Methods). Marker indicates p < 0.05 significance of information (bootstrap test with Bonferroni correction).
- Figure 7.
Subjectwise correlation between low-frequency STN LFP preceding options presentation and risk avoidance in NGDPs. A–C, Comparison of the average value of NGDPs LFP in the three intervals for the two conditions. Title reports significance of Wilcoxon rank test. D–F, Modulation from average risk avoidance associated with trials in which LFP averaged over the three different intervals belongs to the four percentiles. Markers and bars indicate medians and range over NGDP.
- Figure 8.
Groupwise correlation between low-frequency STN LFP preceding options presentation and risk avoidance in NGDPs. A, Grand average of z-scored LFP preceding task in which NGDPs opted for HR (red) or LR (blue) option. Areas indicate median value confidence. Horizontal dotted line represents z = 0. Vertical dashed lines indicate the interval in which the outcome of previous choice, the accumulated money, and the blank screen are displayed. See Fig. 1C for details. B, Average value of NGDPs LFP in the three intervals for the two conditions. C, Mutual information between LFP levels in the different intervals and future choice (black bars). White bars and associated error indicate average and 75th percentile bootstrap information over 200 permutations (see Methods).
Tables
Preoperative therapy Patient Age at implant (years) Gender Disease duration at implant (years) LEDD
(mg)DA in LEDD
(mg)Preoperative UPDRS III score MMSE SOGS BIS Y1-STAI Y2-STAI Cohort 1: GDPs 1 49 M 8 1400 0 73.9 26.6 7 18 29 41 2 41 M 15 550 0 73.5 24.6 9 7 20 34 3 52 M 4 200 0 50.0 27.1 10 24 35 28 4 60 F 6 940 240 61.3 27.5 5 26 55 53 5 78 F 10 1200 280 56.2 30 7 19 46 35 6 48 M 10 900 0 71.1 27.6 15 22 48 36 Mean (SD) 53.0 (10.0) 8.8 (3.8) 865.0 (435.0) 86.6 (134.5) 64.3 (10.0) 27.2 (1.7) 8.8 (3.5) 19.3 (6.7) 38.8 (13.1) 37.8 (8.5) Cohort 2: NGDPs 1 67 F 14 2275 420 38.2 28.49 0 26 41 46 2 63 F 9 2575 70 62.9 26.27 0 22 22 38 3 60 F 14 1550 350 70.3 28.27 1 25 52 54 4 64 F 11 910 0 61.5 29.27 0 22 33 39 5 47 M 10 820 240 58.0 25.89 0 14 38 31 6 61 F 11 1100 210 60.7 NA 0 NA 39 47 Mean (SD) 60.0 (6.0) 11.5 (2.1) 1152.1 (691.5) 215 (160) 58.6 (10.8) 27.6 (1.5) 0.2 (0.4) 21.8 (4.7) 37.5 (9.8) 42.5 (8.1) p-value p > 0.05 p > 0.05 p > 0.05 p > 0.05 p > 0.05 p > 0.05 p > 0.05 p ≤ 0.05 p > 0.05 p > 0.05 p > 0.05 BIS, Barratt Impulsiveness Scale; DA, dopamine agonist; LEED, Levodopa Equivalent daily dose; MMSE, Mini-Mental State Examination; NA, not available; SOGS, South Oaks Gambling Screen; UPDRS III, Unified Disease Rating – scale motor score; Y1–Y2 STAI, State–Trait Anxiety Inventory. Patients with gambling disorders (GDs) and without (NGDs) were compared with a paired t test for age and disease duration at implant, preoperative therapy, preoperative UPDRS motor score, preoperative MMSE, SOGS, BIS, and Y2 STAI, stereotactic coordinates; Fisher’s exact test for gender. Differences were considered significant if p < 0.05. Data were expressed as mean ± SD.
Contact 1 stereotactic coordinates Patient Right (x, y, z) Left (x, y, z) Cohort 1: GDPs 1 12.4, –4.7, –0.3 11.3, –6.1, –4.7 2 11.2, –3.4, –3,1 11.4, 6.6, –6.7 3 13.0, 1.3, –3.0 10.3, –0.3, –3.3 4 11.6, –3.0, –1.5 10.3, –4.1, –1.6 5 10.9, –1.4, –3.1 11.9, –2.9, –5.5 6 13.1, –1.3, –3.6 9.9, 0.3, –6.9 Mean (SD) 12.0 (0.9), –2 (2), –2.4 (1.2) 10.8 (0.8), –1.1 (4.4), –4.8 (2) Cohort 2: NGDPs 1 9.8, 4.2, 0.75 11.8, –6.2, –1.8 2 11.7, –3.4, –4.2 11.8, –3.2, –4.0 3 9.6, –1.9, –5.1 10.3, –4.4, –5.0 4 10.7, –1.2, –2.6 11.3, –3.1, –2.1 5 12.4, –2.8, –4.3 12.1, –4.0, –3.7 6 13.6, –4.5, –3.8 11.3, –3.0, –4.0 Mean (SD) 11.3 (1.6), –1.6 (3), –3.2 (1.2) 11.4 (0.6), –3.9 (1.2), –3.4 (1.2) p-value p > 0.05 p > 0.05 Stereotactic coordinates of subthalamic leads’ contact 1 were rendered according to the anterior commissural–posterior commissural (AC–PC) line and the mid-commissural point (MCP) between AC-PC line; x = mm lateral from AC-PC line; y = mm anterior (+) or posterior (–) from MCP; z = mm depth according to AC–PC line (–, if ventral; +, if dorsal). Differences were considered significant if p < 0.05. Data were expressed as mean ± SD.
In this issue
Jump to section
- Article
- Abstract
- Significance Statement
- Introduction
- Materials and Methods
- Results
- Decision bias precedes options presentation
- Bandwise STN LFP spectral content correlates with patient condition, but not with risk avoidance
- STN low-frequency fluctuations correlate with risk avoidance in GDPs, but not in NGDPs
- Discussion
- Acknowledgments
- Footnotes
- References
- Synthesis
- Figures & Data
- Info & Metrics
- eLetters