Neurofeedback in Clinical High Risk

Identification of brain abnormalities in first episode schizophrenia (SZ) patients has provided evidence that disease-related abnormalities develop prior to psychosis and helped shift attention to the identification of a prepsychotic,clinical high risk (CHR) for psychosis prodromal period 11,12 A recent meta-analysis showed that CHR individuals have an elevated risk for developing psychosis of 11% after 6 months, 15% after 12 months, 20% after 24 months, and 23% after 36 months8. Furthermore, our work and others demonstrate that CHR individuals show deficits in both brain function and cognition similar to those in SZ 1,8. These include default mode network (DMN) hyper-connectivity, (between medial prefrontal cortex (MPFC) and posterior cingulate cortex (PCC)) and reduced MPFC-dorsolateral prefrontal cortex (DLPFC: part of a central executive network (CEN)) anticorrelation in CHR, similar to findings reported in SZ. These results not only point to a way in which abnormal brain function contributes to SZ development, but also motivate the search for approaches to alter the disease trajectory. Thus, in the R61 portion of the current application, we will examine the efficacy of mindfulness meditation augmented, real-time-fMRI-neurofeedback (rt-NFB) in reducing DMN hyperconnectivity and increasing MPFC-DLPFC anticorrelations, post-NFB, similar to what we observed in chronic SZ, and assess glutamate (Glu) and gammaaminobutyric acid (GABA) levels in MPFC, PCC, and DLPFC. In the R33, we will test the rt-NFB efficacy in a larger sample, examine the relationship between rt-NFB mediated neural network changes and behavioral measures (reductions in symptoms and improvement in working memory (WM)/attentional function). We will also examine cellular level changes via magnetic resonance spectroscopy (MRS) measures of Glu and GABA to arrive at a more comprehensive understanding of what cellular phenomena underlie rt-NFB. This application builds on three sources of evidence: 1. Fogarty grant (PI: Stone, co-PI: Niznikiewicz, Whitfield-Gabrieli) showing MPFC-PCC hyperconnectivity and its correlations with SIPS scores in CHR (see preliminary data (PD)); 2. R21 results (PI: Niznikiewicz) showing that rt-NFB can reduce BOLD activation in the superior temporal gyrus (STG) 13, and hyperconnectivity in DMN 14 in SZ, which in turn were associated with a reduction in auditory hallucinations (AH); and 3. R61/33 (mPI: Niznikiewicz, Whitfield-Gabrieli) ) that confirmed that rt-NFB can modulate DMN connectivity and STG activation in SZ with AH. Based on our PD and the existing literature, in R61, we propose that: 1) rt-NFB directed at DMN will lead to MPFC-PCC hyper-connectivity reductions and an increase in MPFC-DLPFC anticorrelations. For R33 we propose to 1). confirm these findings in a larger independent sample 2) show that improvements in MPFCPCC hyper-connectivity, and MPFC-DLPFC anticorrelations will lead to improvements in clinical symptoms and cognition respectively; and 3) changes in resting state (rs) connectivity will be related to changes in Glu and GABA levels, post-rt-NFB, as assessed by co-localized cellular and connectivity changes pre- vs post-rt-NFB.

R61 PHASE: Overview: We will demonstrate brain target engagement, i.e., the MPFC-PCC rsconnectivity reduction. We will test 40 CHR randomized to 1) one session of rt- NFB directed at DMN (real-rt-NFB; N=20) 2) sham-rt-NFB (N=20) (rt-NFB derived from the motor cortex activation). We will test 20 HC as a benchmark comparison group. The MFC-PCC rs-connectivity reduction in the real rt-NFB group post-rt-NFB, will be the R61 GO criterion. We will also collect MRS data from the MPFC, PCC, and DLPFC.

Aim 1: Brain target engagement using rt-NFB. Based on PD, we predict DMN rs-connectivity reduction in the real rt-NFB group receiving rt-NFB from the DMN-CEN, but not in the sham rt-NFB group. (R61 GO criteria).

Aim 2: Measure Glu and GABA (exploratory) in MPFC, PCC, and DLPFC voxels of interest (VOI). R33 PHASE: Overview: In CHR, we will use a randomized trial, with partial cross-over design: CHR will be randomized to real rt-NFB (N=30), or one sham rt-NFB (N=30) session, followed by real rt-NFB session.; 30 HCs will be scanned once. MRS in the two VOIs will be recorded before and after each rt-NFB session (either real or sham). We will replicate the R61 findings and relate them to clinical and cognitive data. The R33 GO criteria (for future studies): replicating R61 results, significant clinical symptom reductions related to MPFC-PCC connectivity reduction and cognitive improvement related to MPFC-DLPFC anticorrelations increase, post-rt-NFB.

Aim 1: Replicate the R61 Aim 1: We predict that DMN-directed rt-NFB will result in MPFC-PCC rs-connectivity reduction, and MPFC-DLPFC increased anticorrelation in the real rt-NFB group only.

Aim 2: Demonstrate association between DMN rs-connectivity and clinical changes and MPFC-DLPFC anticorrelations and cognitive changes, post-rtfMRI. We predict: associations between 1) reductions of MPFC-PCC connectivity and reductions of SIPS scores, and 2) MPFC-DLPFC anticorrelations increases and WM/attentional improvements in the real rt-NFB group only (see PD).

Aim 3: Co-localization of MRS and fMRI data: Measure Glu and GABA in two selected, individually defined VOIs (based on rs-data) before and after rt-NFB in CHR and HC at baseline.

Aim 4: (exploratory) Correlate Glu/GABA (see Aim 3 for VOIs definitions) with rs-connectivity and clinical and cognitive data, post-rt-NFB. Based on our PD, we predict 1) positive associations between MPFC GABA levels and MPFC-DLPFC anti-correlations, and WM/attention improvements. 2) Glu/GABA ratios associations with MPFC-PCC connectivity reductions, and SIPS scores reductions.

• GABA in MPFC [Time frame: one week]