Cortical signatures of sleep are altered following effective deep brain stimulation for depression

Deep brain stimulation (DBS) of the subcallosal cingulate cortex (SCC) is an experimental therapy for treatment-resistant depression (TRD). Chronic SCC DBS leads to long-term changes in the electrophysiological dynamics measured from local field potential (LFP) during wakefulness, but it is unclear how it impacts sleep-related brain activity. This is a crucial gap in knowledge, given the link between depression and sleep disturbances, and an emerging interest in the interaction between DBS, sleep, and circadian rhythms. We therefore sought to characterize changes in electrophysiological markers of sleep associated with DBS treatment for depression. We analyzed key electrophysiological signatures of sleep—slow-wave activity (SWA, 0.5–4.5 Hz) and sleep spindles—in LFPs recorded from the SCC of 9 patients who responded to DBS for TRD. This allowed us to compare the electrophysiological changes before and after 24 weeks of therapeutically effective SCC DBS. SWA power was highly correlated between hemispheres, consistent with a global sleep state. Furthermore, SWA occurred earlier in the night after chronic DBS and had a more prominent peak. While we found no evidence for changes to slow-wave power or stability, we found an increase in the density of sleep spindles. Our results represent a first-of-its-kind report on long-term electrophysiological markers of sleep recorded from the SCC in patients with TRD, and provides evidence of earlier NREM sleep and increased sleep spindle activity following clinically effective DBS treatment. Future work is needed to establish the causal relationship between long-term DBS and the neural mechanisms underlying sleep.


Figure S2 :
Figure S2: P001 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S3 :
Figure S3: P002 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S4 :
Figure S4: P003 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S5 :
Figure S5: P004 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S6 :
Figure S6: P005 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S7 :
Figure S7: P006 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S8 :
Figure S8: P007 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S9 :
Figure S9: P008 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.

Figure S10 :
Figure S10: P009 SWA data overview A, B: SCC LFP sample times in the Pre-DBS (A) and Post-DBS (B) phase.C, D: Correlation of left and right SCC SWA power Pre-DBS (C) and Post-DBS (D).E, F: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), plotted around the 24h diurnal cycle.The coloured line represents a smoothing spline fit to the data; the dashed line represents the time of the maximum night-time (18:00-10:00) peak of the SWA fit.G: SWA power (normalised to median) of all LFPs for this participant (mean across valid hemispheres), with a mean±SD fit line (solid black line and dashed black lines) superimposed on 1000 fit lines (thin, orange) obtained through random re-sampling of Post-DBS data according to the sample times Pre-DBS (see Methods for details).Also indicated is the median fit peak estimate (black vertical line) superimposed on 1000 fit peak estimate lines (thin orange vertical lines).H: Distribution of fit peak time estimates obtained through the 1000 random resamples of the Post-DBS phase, with the median peak time estimate indicated with the orange vertical line and the fit peak time from Phase B indicated with the blue vertical line.