3D Kinematics Quantifies Gait Response to Levodopa earlier and to a more Comprehensive Extent than the MDS-Unified Parkinson's Disease Rating Scale in Patients with Motor Complications

Abstract

Background: Quantitative 3D movement analysis using inertial measurement units (IMUs) allows for a more detailed characterization of motor patterns than clinical assessment alone. It is essential to discriminate between gait features that are responsive or unresponsive to current therapies to better understand the underlying pathophysiological basis and identify potential therapeutic strategies.

Objectives: This study aims to characterize the responsiveness and temporal evolution of different gait subcomponents in Parkinson's disease (PD) patients in their OFF and various ON states following levodopa administration, utilizing both wearable sensors and the gold-standard MDS-UPDRS motor part III.

Methods: Seventeen PD patients were assessed while wearing a full-body set of 15 IMUs in their OFF state and at 20-minute intervals following the administration of a supra-threshold levodopa dose. Gait was reconstructed using a biomechanical model of the human body to quantify how each feature was modulated. Comparisons with non-PD control subjects were conducted in parallel.

Results: Significant motor changes were observed in both the upper and lower limbs according to the MDS-UPDRS III, 40 minutes after levodopa intake. IMU-assisted 3D kinematics detected significant motor alterations as early as 20 minutes after levodopa administration, particularly in upper limbs metrics. Although all "pace-domain" gait features showed significant improvement in the Best-ON state, most rhythmicity, asymmetry, and variability features did not.

Conclusion: IMUs are capable of detecting motor alterations earlier and in a more comprehensive manner than the MDS-UPDRS III. The upper limbs respond more rapidly to levodopa, possibly reflecting distinct thresholds to levodopa across striatal regions.

Keywords: Parkinson's disease; gait; kinematic analysis; levodopa challenge test; wearable devices.

Figure 1

Body sensor placement (A, head, arms, wrists, back, legs, and feet), data collection pipeline (B) and step‐by‐step representation of the SWS test (C). Swarm‐violin plots of clinical variables under study over time after the administration of levodopa (D–J) and overall sub‐scores between the OFF and Best ON states (K–O). Each group depicts a different time‐point after the administration of a levodopa dose corresponding to 150% of the usual morning antiparkinsonian medication dose. From baseline up to 80 minutes post‐administration. Each clinical variable only significantly differs from baseline from 40 min onwards. For metrics assessed on both sides of the body (D–H) only the scores from the worst side were included in this analysis. Refer to the respective table for the complete statistical analysis thereof using the Friedman test with Conover posthoc correction for multiple comparisons or the Wilcoxon signed‐rank test. Axial score (items 3.1, 3.10, 3.11, 3.12). Short MDS‐UPDRS III (items 3.1, 3.3, 3.4, 3.8, 3.10, 3.11, 3.12, 3.15, 3.17). Tremor score (items 3.15–3.18). Rigidity Score (item 3.3). Bradykinesia (items 3.4–3.8). *P‐value <0.05 (vs left‐most, OFF condition).

Figure 2

Swarm‐violin plots of the kinematic variables under study first presenting significant changes at 20 (A) 40 (B) 60 (C) and 80 (D) minutes post LD administration. Each group depicts a different time‐point after the administration of a levodopa dose corresponding to 150% of the usual morning antiparkinsonian medication dose. From baseline up to 80 minutes post‐administration. Data collected during free walking with the patients wearing 15 inertial measurement units used consisting of a tri‐axial accelerometer, a gyroscope and a magnetometer as detailed in the methods section. For variables evaluated in both sides of the body, the worst side was considered for analysis. Refer to the respective table for the complete statistical analysis thereof using the Friedman test with Conover posthoc correction for multiple comparisons. COM: Center of mass; ROM: Range of movement; Vel: Velocity. *P‐value <0.05 (vs left‐most, OFF condition).

Figure 3

Swarm‐violin plots of the kinematic variables under study upon levodopa administration for non‐PD controls, and PD subjects in their OFF and Best ON states. Data collected during free walking with the patients wearing 15 inertial measurement units used consisting of a tri‐axial accelerometer, a gyroscope and a magnetometer as detailed in the methods section. For variables evaluated in both sides of the body, the worst side was considered for analysis. For variables evaluated in both sides of the body, the worst side was considered for analysis. Refer to the respective table for the complete statistical analysis thereof using the Kruskal–Wallis test with Dunn's posthoc correction for multiple comparisons. COM: Center of mass; ROM: Range of movement; Vel: Velocity. *P‐value <0.05 (vs left‐most, OFF condition).

Figure 4

Local weighted regression fit of absolute percentage change of over time for the MDS‐UPDRS III and example kinematic variables from 20 to 80 min post levodopa administration corresponding to 150% of the usual morning antiparkinsonian medication dose. Data collected during free walking with the patients wearing 15 inertial measurement units used consisting of a tri‐axial accelerometer, a gyroscope, and a magnetometer as detailed in the methods section. Summary smoothed fitting, not intended for modeling purposes.