Different patterns of local field potentials from limbic DBS targets in patients with major depressive and obsessive compulsive disorder

Abstract

The role of distinct limbic areas in emotion regulation has been largely inferred from neuroimaging studies. Recently, the opportunity for intracranial recordings from limbic areas has arisen in patients undergoing deep brain stimulation (DBS) for neuropsychiatric disorders including major depressive disorder (MDD) and obsessive compulsive disorder (OCD). Here we test the hypothesis that distinct temporal patterns of local field potential (LFP) activity in the human limbic system reflect disease state and symptom severity in MDD and OCD patients. To this end, we recorded LFPs via implanted DBS electrodes from the bed nucleus of stria terminalis (BNST area) in 12 patients (5 OCD, 7 MDD) and from the subgenual cingulate cortex in 7 MDD patients (CG25 area). We found a distinct pattern of oscillatory activity with significantly higher α-power in MDD compared with OCD in the BNST area (broad α-band 8-14 Hz; P<0.01) and a similar level of α-activity in the CG25 area as in the BNST area in MDD patients. The mean α-power correlated with severity of depressive symptoms as assessed by the Beck depression inventory in MDD (n=14, r=0.55, P=0.042) but not with severity of obsessive compulsive symptoms in OCD. Here we show larger α-band activity in MDD patients compared with OCD recorded from intracranial DBS targets. Our results suggest that α-activity in the limbic system may be a signature of symptom severity in MDD and may serve as a potential state biomarker for closed loop DBS in MDD.

Figure 1

Postoperative magnetic resonance imaging (MRI). Sagittal (a) and coronal (b) slice of a postoperative T2-weighted structural MRI with electrodes at CG25 target area (case 1).

Figure 2

Grand mean of normalised power spectra for the major depressive disorder (MDD)-CG25 (a) and MDD-bed nucleus of stria terminalis (BNST), and obsessive compulsive disorder (OCD)-BNST patient groups (b) plotted for the frequency range of 2–40 Hz. Shaded areas of power spectra indicate 95% confidence limits for each patient group and target area. Grey shaded area in panel b shows the significant difference between OCD-BNST (red line) and MDD-BNST (blue line) patient groups (8–14 Hz). The grey horizontal dotted line indicates the false discovery rate (FDR)-corrected significance threshold of P = 0.0074, blue dots indicate the P-value per frequency bin (only uncorrected P-values below P = 0.05 shown). Some of the individual power spectra in both BNST patient groups contained peaks in the β-frequency band (15–35 Hz; 37 out of 78 contact pairs), but β-band activity did not differ between patient groups. No peaks were present above 40 Hz. Power spectra in the figures are interpolated for the purpose of visualisation.

Figure 3

Individual examples of relative power spectra for each group visualised over the recording time. Note the consistent α-power elevation in the major depressive disorder (MDD) patients (a), CG25, patient 7; (b), bed nucleus of stria terminalis (BNST), patient 10. In obsessive compulsive disorder (OCD) patients the α-power peak was smaller and less consistent over time (c), BNST, patient 17. Time–frequency representations in the figure were interpolated for the purpose of visualisation. Colour bar indicates relative spectral power (a.u.).

Figure 4

Statistical analysis of averaged α-power (grand means shown in panel a) revealed significant differences between major depressive disorder (MDD) and obsessive compulsive disorder (OCD) groups (*P<0.05; **P = 0.01), but no effect for laterality or sex. (b) Averaged α-power significantly correlated with the individual Beck depression inventory (BDI) for MDD patients (n = 14 MDD, Spearman’s ρ = 0.55, P = 0.042).