Leader and Follower Roles in Adapted Tango Reveal Complementary Neural Routes to Motor Improvement in Parkinson’s Disease

Parkinson’s disease (PD) disrupts internally generated (IG) movement through impaired striato-thalamo-cortical circuitry, while externally guided (EG) movement may recruit relatively preserved cerebello-thalamo-cortical pathways. We tested whether IG and EG movement strategies differentially shape clinical outcomes and resting-state functional connectivity by using partnered adapted tango as a naturalistic IG/EG movement manipulation. People with mild-to-moderate PD were randomized to 12 weeks of adapted tango as Leaders, adapted tango as Followers, or a non-dance health education Control condition. During therapeutic adaptango classes, Leaders self-initiated movement direction, timing, and amplitude, whereas Followers used auditory, tactile, and proprioceptive cues from the partner to guide their movement. We identified significant clinical improvement in both dance groups, with Leaders and Followers showing reduced overall disease severity and motor impairment, while Controls did not show a corresponding significant pattern. When Leaders and Followers were pooled, the Dance group also showed overall clinical and motor improvement, in addition to enhanced lower-limb motor outcomes. Connectivity findings suggested partly distinct neural routes to benefit. Leaders showed relative network stability and shifted directionally toward lower connectivity values resembling those of healthy older adults on cerebellar-sensorimotor connections that are typically strengthened in PD. Followers showed increased connectivity across sensorimotor and caudate-motor circuits, consistent with cue-supported action updating. Controls showed overlapping connectivity increases in some motor-network regions of interest, but without comparable clinical gains, suggesting that increased connectivity alone may not be adaptive. These findings indicate that adapted tango can improve motor outcomes in PD through complementary IG and EG movement-weighted mechanisms, with Leader training potentially supporting neural normalization and Follower training engaging distributed cue-to-action networks.