All patients fulfilled the Queen Square Brain Bank Criteria for diagnosis of PD and experienced significant motor complications despite maximal medical therapy. Prior to surgery, each patient was evaluated for suitability for DBS by a specialized Movement Disorder Group comprised of neurologists, neurosurgeons, radiologist, clinical psychologist, a nurse specialist and various allied health members. The assessment included clinical UPDRS with levodopa challenge test, neuroimaging, neurocognitive tests and psychiatric consultation. Patients whom met the below criterion were considered for GA-based DBS procedure, due to the likely intolerance towards a LA-based procedure. Our criteria included, (1) significant anxiety or depression as per assessment of the clinical psychology team and (2) known underlying psychiatric disorders that requires active psychiatric follow up although not considered a contraindication for DBS implantation. In those whom met the above criteria, GA-based DBS procedure was recommended. For those who did not meet the above criterion, both LA-based and GA-based DBS implantation were the options as the patient preferred. The clinical outcome was assessed at one year post-operatively. The study protocol was approved by CUHK-PWH ethics committee (IRB Reference Number: 2022.479). Informed consent was obtained from all patients.

Magnetic resonance imaging (MRI) 1.5 T or 3 T of the brain was acquired before the operation, including T1 three-dimension (3D), T2 axial, coronal and sagittal 2 mm fine cut, 3D susceptibility weighted imaging (SWI) and diffusion tractography imaging (slice thickness 2 mm, Acquisition matrix 112 × 112, 32 direction). The software for target and trajectory planning were iPlan or Elements (BrainLab, Feldkirche, Germany). The target was marked at the centre of STN corresponding to the anterior border of red nucleus at T2 axial image. The preliminary X/Y/Z coordinates were 12–13 mm lateral, 2–3 mm posterior and 3–4 mm inferior to the mid-commissural point respectively. The entry point was 2 to 4 mm from midline and at 1–2 cm anterior the coronal suture. Fine adjustment of the trajectory was made to avoid nearby sulcus and vessels. On T2 coronal view, the target was adjusted to 2.5 to 3 mm medial to the medial edge of the pyramidal tract using tractography data as reference. In patients whom received surgery after 2020, targeting was also guided by the auto-segmented STN by the Elements software. On the day of operation, a computed tomography (CT) of the brain (helical scan, 0.625 per slice,) was performed with the stereotactic frame (Leksell G frame, or Leksell Vantage frame since 2000) fixed to the head. The CT was then fused to the MRI plan and the frame coordinates were generated by the planning software.

All patients were off dopaminergic medication for at least 12 h before surgery. Bilateral DBS lead and pulse generator implantation surgery was performed under GA with tracheal intubation. All patients were given total intravenous anesthesia with propofol and remifentanil. Bispectral index (BIS) monitoring was used as guidance for depth of anaesthesia. Propofol concentrations were targeted between 2–6mcg/ml using the pharmacokinetic Marsh model. Remifentanil dose ranged from 0.01–0.5mcg/kg/min intraoperatively. Muscle relaxants was stopped after an initial induction dose to facilitate intubation of the patient and transport of patient for CT. The patient was positioned in supine with the head in neutral position. Needle EMG was set up to the upper limbs of the deltoid and forearm extensor muscle groups (Figure 1a). A burr hole was made and a microelectrode was inserted for microelectrode recording (MER). After MER, monopolar macrostimulation was performed through the macro contact of the electrode. The site of macrostimulation was at the location with best MER signal and the site was documented by its distance from the target along the tract. EMG signal response was detected from the upper limbs muscle groups, including deltoid and forearm extensors contralateral to the side of stimulation (Figures 1b,c). The stimulation began at 0.5 mA/60 uS/130 Hz. The intensity of stimulation was gradually increased by incrementations of 0.5 mA until the EMG signal was activated defined as the amplitude doubling or greater than the base line (Figures 1b,c). This was documented as the intra-operative EMG-based side effect threshold. During each step of stimulation mA increment, we also looked for signs of facial asymmetry, eyes deviation or pupillary change. Bispectral index (BIS) was maintained between 40 and 60 at all times during MER and macrostimulation. Targeted propofol concentration was titrated down, as guided by BIS, to optimize MER/EMG recordings. If the EMG-based side effect threshold was less than 3 mA, alternative tract (s) away from the pyramidal tract were explored. After the appropriate trajectory was selected, the microelectrode was removed and the DBS lead (Medronic 3389 or Medronic directional lead or Vercise Cartesia directional lead, Boston scientific) was implanted. The procedure was performed on the right side first followed by the left side. The pulse generator was implanted during the same operation.

At one month after the operation, a CT brain was performed and fused with pre-operative MRI plan. The DBS lead and contacts were marked manually based on the CT image, or recognized by the planning software (BrainLab Element). During the same follow-up session, the first DBS programming was performed by a nurse specialist with the patient at medication Off for 12 h status. Monopolar stimulation of each DBS lead contact was tested individually with 0.5 mA incrementations in frequency 130 Hz and pulse width 60uS. The effect and PTSE threshold were documented. Contralateral motor symptoms were regarded as pyramidal side effects, including bilateral contralateral eye deviation, and/or contralateral facial or limb muscle spasm. The clinical side effect threshold generated during programming session was defined as the generated side effect threshold in mA of a contact closest to the intraoperative macrostimulation site. Finally, the contact with best therapeutic window was chosen as the active contact for chronic stimulation. Patients were then seen at regular intervals for optimization of stimulation parameters and for medication titration guided by the neurologist of the team. Clinical outcomes were assessed at one year post-operatively including UPDRS, QOL (PD39), LED and neuropsychology assessment.

All statistical analyses was performed on the Statistical Package for the Social Sciences [Windows version 26.0; SPSS Inc, Chicago (IL), US]. Paired student t-test and linear Regression was used to assess the relationship between the intra-operative EMG threshold vs. post-operative clinical threshold. A p-value of 0.05 was used for statistical significance.

There were 44 consecutive patients with Parkinson’s disease received DBS STN under GA between January 2018 to February 2023 (18 males, 26 females, mean age 60.7, range 42–73 years old) (Table 1).

All 44 cases (88 sides) demonstrated activations of EMG of the upper limbs with progressively increased voltage of contralateral stimulation. As illustrated in Figure 2, the mean voltage of intra-operative EMG-based PTSE threshold was 4.3 mA (SD 1.2 mA, range 2.0–8 mA). The mean voltage of post-operative programming PTSE threshold was 3.7 mA (SD 0.8 mA, range 2.0–6.5 mA), with a mean difference of 0.6 mA (SD 1.1 mA, range −=2.0–4 mA). Post-operatively, the majority of the side effects involved contralateral facial spasm, slurring of speech, or gaze deviation towards the contralateral side. During the operation, 7 image-based tracts were repositioned amongst the 88 tracts (8%) due to a low EMG-based PTSE threshold. The direction of change was mainly medially and/or anteriorly along the oblique axis of the STN on axial view (24).

Among the 88 pairs of EMG-based PTSE vs. post-operative programming PTSE thresholds, the EMG-based PTSE threshold was equal or higher than that obtained by programming in 67 pairs (67%). Within these 67 pairs, 60 (90%) were found within a range of 0–2 mA, and 7 (10%) were found within a range of 2.5–4 mA. In the remaining 21 pairs (24%), the programming PTSE threshold was higher than EMG-based PTSE threshold, all within 1 mA except 2 pairs with 1.5 mA difference and 1 pair with a 2 mA difference.

We noted that all EMG-based PTSE threshold had a value of >3 mA in all 88 pairs with the exception of 3 pairs. The lowest value noted was 2.0 mA, whilst the highest value was found to be 8 mA. Within this specific case whereby the EMG-based PTSE threshold was found to be low, we accepted this intra-operative result in view of the excellent MER signal obtained intra-operatively. Despite the low threshold encountered intra-operatively, we hoped to mitigate such findings with bipolar stimulation post-operatively. During post-operative programming, the PTSE threshold was found to be 3 mA. At the one year follow up, the stimulation parameters were left side 2.8 mA (the side with EMG-based PTSE threshold 2.5 mA), right side 3.3 mA, pulse width 60uS and frequency 100 Hz for both sides. The clinical outcomes of this case at one year were UPDRS part III improvement of 43% with DBS on only, and 74% with both medication and DBS on. The LED reduction rate was 59%.

Figure 3 illustrates a simple linear regression plot between the intra-operative EMG-based PTSE threshold vs. the post-operative programming PTSE threshold. The regression output shows that intra-operative EMG-based PTSE threshold was a statistically significant predictor variable for the post-operative programming PTSE threshold (p value <0.001). The regression model plot also demonstrates a linear relationship between the intra-operative CST and the post-operative PTSE threshold (R = 0.42). However, the R-squared value was 0.17, indicating the presence of trend between the two variables, although with significant variability or potential cofounders.

Post-operatively, no cases were found to have an abnormally low PTSE threshold which interfered with subsequent DBS programming. All patients completed a formal clinical assessment at the one year mark after the operation. The UPDRS part III score showed a mean improvement of 54% (SD 12.7%) which was similar to results of 52.7% (SD 12.7%) improvement by the pre-operative dopamine challenge test. The mean levodopa equivalent dose (LED) reduction was 59.5% (SD 23.5%). The mean stimulation parameter at one year were 2.84 mA and 2.70 mA respectively on right and left contacts (SD 0.6 and 0.7 mA), pulse width 60µS (SD 0) and frequency 115 Hz (SD 31 Hz). All results are summarized in Table 1.

A case illustration was showed in Figure 4. Intra-operatively, macrostimulation of the original planned target induced contralateral limb EMG activation at 2.5 mA, which would be too low for an adequate therapeutic window for post-operative programming. Therefore, a 2nd track was explored 1 mm anterior and medially away from the 1st tract with its position remaining in the mediolateral centre of the STN. On repeated test stimulation, the EMG-based PTSE threshold was increased to 3.5 mA. During post-operative programming, the PTSE threshold was noted to be 3.5 mA with a clinical observation of slow gaze movement. The therapeutic window of stimulation was between 2.5–3 mA.