We included 64 drug-naive PD patients (mean age 58.2 ± 12.3 years) and 47 healthy controls (HC) (mean age 60.4 ± 9.2 years) (Table 1). All participants completed at least 9 years of education. Patients with PD were diagnosed according to the Movement Disorder Society clinical diagnostic criteria for PD (Postuma et al. 2015) and investigated before the introduction of dopaminergic therapy. In all patients enrolled in this study, a clear beneficial response to dopaminergic therapy was observed at follow-up.

The healthy control group was acquired by advertising to the general public. The exclusion criteria for HC were neurological or psychiatric disorders, the use of psychoactive substances, concurrent oncological or other major somatic diseases, the presence of REM sleep behavioral disorder. For both groups, participants with major hearing and vision problems were excluded and only non-demented, cognitively normal participants with MoCA scores of 24 or higher were included. The cut-off value of 24 refers to the results of a normative study for the Czech population (Kopecek et al. 2017). All included participants had a full score of 3 points in MoCA Serial 7 Subtraction Task, i.e., they didn’t show any impairment in the cognitive test that was also used as a competitive task in dual-task walking.

All participants underwent a detailed medical interview, and a neurological and neuropsychological examination (the battery included five cognitive domains, as recommended by the Movement Disorder Society: attention and working memory; executive functions; language; delayed recall, visuospatial abilities (Litvan et al. 2012). The Movement Disorder Society Unified Parkinson’s Disease Rating Scale, motor subscale (MDS-UPDRS part III) (Goetz et al. 2008) and the Montreal Cognitive Assessment (MoCA) were used to assess motor and cognitive performance (Hobson 2015). All participants gave informed consent. The study received approval from the Ethics Committee of the General University Hospital in Prague and has been performed in accordance with the ethical standards established in the 1964 Declaration of Helsinki.

All participants completed the expanded Timed Up & Go Test (TUG) (Wall et al. 2000): get up from a chair, walk 10 m at the preferred walking speed, turn around, walk back and sit down again. TUG was performed twice. For data measurements, a 5.15 m long and 0.9 m wide pressure walkway (Platinum model GAITRite®, CIR System Inc.) was placed 2.43 m from the chair in the middle of the straight gait walkway. Participants were instructed to walk at a normal pace under two different settings: (i) in the single-task (ST) condition and (ii) in the dual-task (DT) condition while performing serial subtraction, counting down from 100 by sevens. During straight walk, individual gait cycles were detected and analysed, and the gait speed, stride length and cadence were investigated (Zampieri et al. 2010).

Dual-task cost (DTC) was calculated for all monitored gait parameters (speed, stride length, and cadence), to evaluate the relative impact of the additional cognitive task on gait performance:

The DTC parameter includes information about how the additional cognitive load in DT influences gait performance and thus combines gait and cognitive assessment which was further used to divide patients into subgroups. A principal component analysis (PCA) was performed that included all DTC gait parameters in the HC data set. Based on the gait evaluation and comparison with the control group, patients with PD were divided into two subgroups with normal (nDTC) and abnormal DTC (iDTC). The criterion for inclusion in the nDTC group was to have the DTC first PCA component above the 10th percentile of HC, while the iDTC group consisted of individuals with values below the 10th percentile.

A 3T MRI scanner (Siemens Skyra 3T, Siemens Healthcare, Erlangen, Germany) with a 32-channel head coil was used to perform the examination.

Morphometry analysis was performed on T1-weighted 3D magnetization-prepared rapid acquisition with gradient echo (MPRAGE) images in the axial plane with the following acquisition parameters: repetition time (TR), 2,200 ms; echo time (TE), 2.4 ms; inversion time (TI) 900 ms; flip angle (FA) 8°; field of view (FOV) 230 × 197 × 176 mm; spatial resolution 1 mm3 isotropic.

The pre-processing and segmentation of T1 weighted images were performed with the Computational Anatomy Toolbox software (CAT12), version 12.8.2 (Gaser et al. 2022) implemented in the statistical parametric mapping software (SPM, version 7771) (Friston 2007) in Matlab (The MathWorks Inc. 2022). By reviewing one slice of every brain, a visual quality check was performed to find obvious artefacts in the scans and inaccurately orientated images. The homogeneity of the data was checked by applying the CAT batch of data quality. The segmentation quality for every single image was accepted at a minimum of C + in all quality parameters. The modulated, normalized grey matter segments were smoothed using a Gaussian kernel with an 8 mm3 full width at half maximum to perform the voxel-based morphometry (VBM).

Differences in gait parameters between groups were performed by using the general linear model with age and sex as covariates. VBM analysis was performed using a multiple regression model with the covariate DTC (speed or stride length or cadence), total intracranial volume (TIV), age, and sex. The statistical map for the correlation analysis was thresholded at cluster level at the statistical level p < 0.05 corrected by family-wise error (FWE).

The subgroup of HC-iDTC was excluded from all statistical analyses because of the low number of participants.