Introduction: Despite the prevalence of and negative health consequences associated with perceived loneliness in older adults, few studies have examined interactions among behavioral, psychosocial, and neural mechanisms. Research suggests that physical activity and improvements in perceived social support and stress are related to reductions in loneliness. Yet, the influence of brain structure on these changes is unknown. The present study examined whether change in regional brain volume mediated the effects of changes in social support and stress on change in perceived loneliness after an exercise intervention. We also examined the extent to which baseline brain volumes moderated the relationship between changes in social support, stress, and loneliness. Methods: Participants were 247 older adults (65.4 ± 4.6 years-old) enrolled in a 6-month randomized controlled trial comprised of four exercise conditions: Dance (n = 69), Strength/Stretching/Stability (n = 70), Walk (n = 54), and Walk Plus (n = 54). All groups met for 1 h, three times weekly. Participants completed questionnaires assessing perceived social support, stress, and loneliness at baseline and post-intervention. Regional brain volumes (amygdala, prefrontal cortex [PFC], hippocampus) before and after intervention were measured with automatic segmentation of each participant's T1-weighted structural MRI. Data were analyzed in a latent modeling framework. Results: Perceived social support increased (p = 0.003), while stress (p < 0.001), and loneliness (p = 0.001) decreased over the intervention. Increased social support directly (-0.63, p < 0.01) and indirectly, through decreased stress (-0.10, p = 0.02), predicted decreased loneliness. Changes in amygdala, PFC, and hippocampus volumes were unrelated to change in psychosocial variables (all p ≥ 0.44). However, individuals with larger baseline amygdalae experienced greater decreases in loneliness due to greater reductions in stress (0.35, p = 0.02). Further, individuals with larger baseline PFC volumes experienced greater reductions in stress due to greater increases in social support (-0.47, p = 0.02). No group differences in these pathways were observed. Conclusions: The social support environment and resulting reductions in stress, as opposed to exercise mode, may represent important features of exercise programs for improving older adults' perceived loneliness. As amygdala volume has been linked to anxiety, depression and impaired cognitive control processes in the PFC, moderation findings suggest further investigation in this area is warranted. Trial Registration: ClinicalTrials.gov identifier NCT01472744 (https://clinicaltrials.gov/ct2/show/NCT01472744?term=NCT01472744&rank=1).
Keywords: aging; amygdala; exercise; loneliness; prefrontal cortex; psychological; social support; stress.