Statistical analysis

Statistical analysis was performed using SPSS 19.0 Software (SPSS Inc., Chicago, IL, USA). A threshold of p < 0.05 (2-tailed) was applied. One-way ANOVA and Chi square tests were conducted to compare baseline characteristics of the participants between groups. There were no significant differences in age and gender between the fibromyalgia and matched healthy control groups.

Functional BOLD data were preprocessed using SPM 12 (Statistical Parametric Mapping. Welcome Department of Cognitive Neurology, London, UK; implemented by MATLAB R3012b, Math Works, Inc., Natick, MA, USA). During the preprocessing, images were realigned, segmented, and co-registered to each subject’s high-resolution T1 scan, which was used to normalize to the standard Montreal Neurological Institute (MNI) template. Images were also smoothed using an 8 mm full-width at half-maximum (FWHM) Gaussian kernel and filtered with a frequency window of 0.008–0.09 Hz. In addition to these steps, we employed segmentation of gray matter, white matter, and cerebrospinal fluid (CSF) areas for the removal of temporal confounding factors (white matter and CSF) (Whitfield-Gabrieli & Nieto-Castanon, 2012). Data were then submitted to motion correction using the artifact detection toolbox (http://www.nitrc.org/projects/artifact_detect/). For each subject, we treated images as outliers if the composite movement from a preceding image exceeded 0.5 mm or if the global mean intensity was greater than 3 standard deviations from the mean image intensity for the entire resting scan. Outliers were included as regressors in the first-level general linear model along with other six regular motion parameters (Redcay et al., 2013).

Resting state functional connectivity analysis was conducted using the CONN toolbox v15.g (Whitfield-Gabrieli & Nieto-Castanon, 2012) (http://www.nitrc.org/projects/conn). We used an a priori DLPFC seed (peak coordinate: ± 36, 27, 29, with 5 mm radius), which has been used in previous studies (J. Hwang et al., 2015; Sheline et al., 2010; Tao, Chen, Egorova, et al., 2017). Functional connectivity measures were computed between the seed and every other voxel in the brain. First-level correlation maps were produced by extracting the residual BOLD time course from the DLPFC and by computing Pearson’s correlation coefficients between that time course and the time courses of all other voxels in the brain. Correlation coefficients were transformed into Fisher’s ‘Z’ scores to increases normality and allow for improved second-level general linear model analyses.

The baseline DLPFC rsFC of fibromyalgia patients and healthy control subjects were compared using a two-sample t-test. The Tai Chi practice effect (post-practice minus pre-practice) on fibromyalgia patients was compared using a paired t-test. Additionally, we also compared the DLPFC rsFC of fibromyalgia patients after practicing Tai Chi to healthy controls using a two-sample t-test. Age, gender, and BDI scores were included as covariates. A threshold of voxel-wise p < 0.005 (uncorrected) and cluster-level p < 0.05 (family-wise error correction) were applied for data analyses. Given the important role of the periaqueductal grey in pain modulation, we defined the PAG as a region of interest and used small volume correction to correct the p value at a level of p < 0.05. Similar to previous studies (Eippert et al., 2009; Kong et al., 2013b), correction was based on peak coordinates (x, y, z: 1, −25, −12) with a 4 mm radius obtained from a previous PAG meta-analysis (Linnman, Moulton, Barmettler, Becerra, & Borsook, 2012).