DIFFERENTIAL EFFECTS OF TRANSCRANIAL PHOTOBIOMODULATION ON FUNCTIONAL NETWORK REORGANIZATION ACROSS CLINICAL POPULATIONS

Transcranial photobiomodulation (t-PBM) is a noninvasive neuromodulation method that enhances cerebral mitochondrial function with emerging evidence as a network modulator in healthy adults [1]. However, its influence on the temporal organization of intrinsic brain states remains unclear. Static connectivity analyses overlook rapid shifts in network configuration that may underlie cognitive and mood states as well as therapeutic response. To address this, we applied Co-Activation Pattern (CAP) [2] analysis to dissect the spatiotemporal stability of brain networks in two distinct clinical populations: Alzheimer’s disease (AD) and Major Depressive Disorder (MDD). Our framework isolated diseasespecific signatures (Analysis 1), session-related effects of t-PBM (Analysis 2), diagnosisspecific modulation (Analysis 3), and dose-related effects within AD (Analysis 4). Data were drawn from two clinical trials: an AD study with a single t-PBM session (medium-dose continuous wave vs. high-dose pulsed wave) and an MDD study (medium-vs. high-dose). Detailed cohort characteristics, participant demographics, study design, including fMRI parameters, are described elsewhere [Iosifescu et al., 2017 Healthcare; Iosifescu et al., 2023 Photonics). Each participant completed pre-, during-, and post-stimulation fMRI scans. During the stimulation condition, participants received t-PBM at either medium- (CW, irradiance ~300 mW/cm2) or high-dose (PW, peak irradiance 900 mW/cm2, average irradiance 300 mW/cm2) while in the MRI scanner. The analyzed sample included the following groups: AD high dose (n = 14), AD medium dose (n = 28), MDD high dose (n = 11), and MDD medium dose (n = 17). Preprocessing included fMRIPrep and XCP-D, with band-pass filtering (0.01–0.1 Hz) and 4S1056Parcels parcellation. CAPs were identified via temporal clustering; occurrence, incidence, and dwell time were quantified. Our analytic strategy was designed to systematically isolate disease signatures from intervention effects: Analysis 1: Baseline CAP differences were tested with Welch’s t-tests. MDD showed higher CAP5 dwell time (t(58.84)=2.39, p=0.030) and occurrence (t(53.91)=2.50, p=0.030). No other CAPs survived FDR; CAP3 and CAP4 differences were nominal only. Analysis 2: Within-group mixed-effects models showed session effects only in MDD medium dose. CAP1 dwell time (F(2,35.5)=7.79, p=0.0047) decreased pre to during (p=0.018) and increased during to post (p=0.0013); CAP1 incidence (F(2,37.0)=4.99, p=0.018) decreased post vs pre (p=0.013) and vs during (p=0.045). CAP3 incidence also differed (F(2,36.0)=5.24, p=0.030) via a during→post reduction (p=0.009). No AD effects. Analysis 3: Between-group effects appeared only in medium dose. CAP2 dwell (F(1,44.9)=8.07, p=0.010) and occurrence (F(1,44.1)=9.38, p_FDR=0.010); CAP3 incidence (F(1,44.7)=6.39, p=0.023) and occurrence (F(1,44.0)=6.82, p=0.023); CAP4 incidence (F(1,45.0)=7.28, p=0.029); CAP5 dwell (F(1,42.9)=12.3, p=0.003) and occurrence (F(1,43.6)=10.3, p=0.004). Analysis 4: No FDR-significant AD dose effects. CAP2 occurrence showed a nominal difference (t(32.03)=2.35, p=0.025; FDR=0.075). Other metrics nonsignificant. MDD showed baseline default-mode CAP instability with greater dwell and occurrence than AD. Only MDD under medium dose showed session-related CAP modulation, mainly in attention and visual networks. Medium-dose stimulation also amplified diagnostic differences across multiple CAPs, whereas high dose and all AD conditions showed minimal effects. Overall, MDD exhibited greater context-sensitive network flexibility, with medium dose most effective for revealing disease-specific dynamics.