Differential default mode network changes following progressive aerobic training and free training in internet gaming disorder

This randomized trial of progressive aerobic training (PAT) for internet gaming disorder (IGD) included behavioral and task-fMRI outcomes, whereas intervention-related changes in resting-state brain network topology remain unclear. This secondary neuroimaging analysis examined default mode network (DMN) participation coefficient (PC) as its principal novel outcome. This secondary neuroimaging analysis used data from the registered randomized trial (ChiCTR2400090834). Seventy-two individuals with IGD were randomized to PAT or free training (FT). Behavioral analyses included 63 participants with usable pre- and post-intervention data (PAT, 33; FT, 30), whereas resting-state MRI analyses included 59 participants with usable paired scans (PAT, 29; FT, 30). PAT participants completed 20 progressively intensified treadmill sessions over one month; FT participants engaged in self-selected exercise. DMN PC was calculated across a 10%–30% sparsity range, and the area under the PC-versus-sparsity curve (AUC) was the principal neuroimaging outcome. Behavioral measures (IAT, DSM-5, and QGU) were also analyzed. The principal novel finding was a significant TIME × GROUP interaction in DMN PC AUC (F(1, 57) = 9.270, p =.0035, partial η² = 0.140), driven by a larger increase in the PAT group; the PAT pre-post contrast remained significant after Bonferroni adjustment, whereas the FT contrast did not. The result remained significant after adjustment for change in mean framewise displacement (F(1, 56) = 6.982, p =.0107). Behavioral analyses in 63 participants showed TIME × GROUP interactions for IAT ( p =.010), DSM-5 ( p =.040), and QGU ( p =.010). Exploratory ΔAUC correlations with QGU, DSM-5, and IAT were negative in PAT but nonsignificant in FT. PAT-versus-FT differences for QGU and IAT survived FDR, but not Bonferroni, correction; these associations therefore remain hypothesis-generating. In this secondary neuroimaging analysis of the registered randomized trial, the principal novel finding was a group-by-time difference in sparsity-range-integrated DMN PC AUC, reflecting a larger increase in the PAT condition than in the FT condition. Because PAT and FT differed across multiple intervention dimensions, this result should be interpreted as a difference between the two intervention conditions rather than as evidence of a PAT-specific mechanism. The result was robust to threshold-wise analysis and adjustment for residual head motion. Behavioral outcomes were also examined, while the brain-behavior associations should be interpreted as exploratory. Chinese Clinical Trial Registry, ChiCTR2400090834. Registered on 14 October 2024.