Spectral Dynamics of Resting State fMRI Within the Ventral Tegmental Area and Dorsal Raphe Nuclei in Medication-Free Major Depressive Disorder in Young Adults

Abstract

Background: Dorsal raphe nucleus (DRN) and ventral tegmental area (VTA) are major brainstem monamine nuclei consisting of serotonin and dopamine neurons respectively. Animal studies show that firing patterns in both nuclei are altered when animals exhibit depression like behaviors. Functional MRI studies in humans have shown reduced VTA activation and DRN connectivity in depression. This study for the first time aims at investigating the functional integrity of local neuronal firing concurrently in both the VTA and DRN in vivo in humans using spectral analysis of resting state low frequency fluctuation fMRI. Method: A total of 97 medication-free subjects-67 medication-free young patients (ages 18-30) with major depressive disorder and 30 closely matched healthy controls were included in the study to detect aberrant dynamics in DRN and VTA. For the investigation of altered localized dynamics we conducted power spectral analysis and above this spectral cross correlation between the two groups. Complementary to this, spectral dependence of permutation entropy, an information theoretical measure, was compared between groups. Results: Patients displayed significant spectral slowing in VTA vs. controls (p = 0.035, corrected). In DRN, spectral slowing was less pronounced, but the amount of slowing significantly correlated with 17-item Hamilton Depression Rating scores of depression severity (p = 0.038). Signal complexity as assessed via permutation entropy showed spectral alterations inline with the results on spectral slowing. Conclusion: Our results indicate that altered functional dynamics of VTA and DRN in depression can be detected from regional fMRI signal. On this basis, impact of antidepressant treatment and treatment response can be assessed using these markers in future studies.

Keywords: depression; dopamine; dorsal raphe nucleus; resting state; resting state fMRI; serotonin; ventral tegmental nucleus.

Figure 1

Analysis scheme. BOLD time series extracted from DRN and VTA were orthogonalized toward each other for each subject. Subsequently, the orthogonalized time courses were (i) converted to spectra, and (ii) were analyzed for information content.

Figure 2

(A) Regions of interest for time course extraction and (B) mean and standard error of mean of spectra for patients (blue) and healthy controls (green). Clipping of spectra in subsequent analyses is indicated by hatching on both ends of the spectra. (C) Spectral group differences are indicated as P-values from Wilcoxon rank sum tests the red lines indicate P = 0.05, uncorrected for multiple comparisons.

Figure 3

Spectral slowing. (A) Mean and standard error of mean of correlation coefficients RFZ (Fisher Z-transformed) of individual patients (blue) and controls (green) to the group of healthy controls (see Methods) vs. shift in frequency. Significant correlations at Pc < 0.05, corrected for multiple comparisons (Bonferroni-factor = 17) are indicated by an asterisk. (B) HamD score per patient vs. the frequency shift creating the highest (optimal) average correlation of patient spectra to spectra of healthy controls (Dfopt). Median, 75% confidence intervals and outliers are given for all realized shifts. Least square lines to the individual data are added as guide to the eye. Significance of the correlation (corrected for nuisance variables), is indicated within the panels.

Figure 4

Signal complexity: PeEn vs. a correlate of frequency ~ 1/n in (A) healthy controls, and (B) patients as mean and standard error of mean. Significant within-group differences between the PeEn of the two ROIs at P < 0.05, uncorrected for multiple comparisons (Wilcoxon signed rank test) are indicated by asterisks. Red asterisks indicate significant interactions of group × ROI at the same significance level.