Muscle synergies in Parkinson's disease before and after the deep brain stimulation of the bilateral subthalamic nucleus

Abstract

The aim of this study is to quantitatively assess motor control changes in Parkinson's disease (PD) patients after bilateral deep brain stimulation of the subthalamic nucleus (STN-DBS), based on a novel muscle synergy evaluation approach. A group of 20 PD patients evaluated at baseline (before surgery, T₀), at 3 months (T₁), and at 12 months (T₂) after STN-DBS surgery, as well as a group of 20 age-matched healthy control subjects, underwent an instrumented gait analysis, including surface electromyography recordings from 12 muscles. A smaller number of muscle synergies was found in PD patients (4 muscle synergies, at each time point) compared to control subjects (5 muscle synergies). The neuromuscular robustness of PD patients-that at T₀ was smaller with respect to controls (PD T₀: 69.3 ± 2.2% vs. Controls: 77.6 ± 1.8%, p = 0.004)-increased at T₁ (75.8 ± 1.8%), becoming not different from that of controls at T₂ (77.5 ± 1.9%). The muscle synergies analysis may offer clinicians new knowledge on the neuromuscular structure underlying PD motor types of behavior and how they can improve after electroceutical STN-DBS therapy.

Figure 1

Comparison of muscle synergies extracted during walking from (A) PD patients before DBS (T₀), (B) PD patients 3 months after DBS (T₁), (C) PD patients 12 months after DBS (T₂), and (D) controls. For each synergy k, the weight vector $W_k$ (bars showing the muscles’ contribution levels) and the corresponding activation coefficient ${C(t)}_k$ (time profile of the neural command) are displayed with the same color (average and standard deviation across the population are displayed). Vertical dotted lines represent the transition between the stance and the swing phase of the gait cycle. Muscle abbreviations: VM  vastus medialis, TFL  tensor fasciae latae, GMD  gluteus medius, MH  medial hamstring, LD  longissimus dorsii (for PD patients: LD_M  ≡ LD of the more affected side, LD_L ≡ LD of the less affected side; for controls: LD_D ≡ LD of the dominant side, LD_ND ≡ LD of the non-dominant side), TA  tibialis anterior, LGS  lateral gastrocnemius, PL  peroneus longus, SOL  soleus, RF  rectus femoris, and LH  lateral hamstring.

Figure 2

Panel (A) Raincloud plots of the neuromuscular robustness computed by means of the cross variance accounted For (${\mathit{Cross}}_{\mathit{VAF}}$) from PD patients before DBS (T₀, in red), 3 months after DBS (T₁, in orange), 12 months after DBS (T₂, in yellow), and controls (in blue). For each group, the raincloud plot shows, sequentially, the data distribution (split-half violin plot), a standard visualization of central tendency with a boxplot (representing minimum, 25th percentile, median, 75th percentile, and maximum), and raw jittered data points of each specific individual (scatter plot). Panel (B) Circular bar diagram of the relative changes in muscle synergy robustness at 3 months after DBS with respect to baseline (T₁–T₀), and at 12 months after DBS with respect to baseline (T₂–T₀), for each PD patient. Green bars represent PD patients that improved the robustness of their motor control after DBS neurosurgery, yellow bars PD patients that did not change their robustness, and red bars PD patients that worsened their robustness after DBS neurosurgery. Single (*) and double (**) asterisks represent statistically significant differences with p-values lower than 0.05 and 0.01, respectively.

Figure 3

Agreement between UPDRS-III changes (x-axis) and neuromuscular robustness changes (y-axis). The relative changes at 12 months after DBS with respect to baseline are represented (T₂–T₀). The quadrants showing concordance are highlighted in green.

Figure 4

Experimental design and data recordings. (A) Sensor placement for a healthy representative subject of the sample population. SEMG active probes are placed over twelve muscles of the dominant (or more affected) lower limb and the trunk (bilaterally). Foot-switches are placed beneath the heel, the first, and the fifth metatarsal heads to detect gait phases. Electro-goniometers are placed on the lateral aspect of the knee joint, bilaterally. (B) Examples of sEMG (upper panel) and foot-floor contact (lower panel) signals acquired through the system STEP32 (Medical Technology, Turin, Italy). (C) Schematic representation of the walking path. Subject walked barefoot back and forth along a 9-m straight path, for approximately 5 min at a self-selected pace.