The data analyzed in the study were recorded from patients who received microstimulation to determine SSB during unilateral or bilateral STN-DBS surgery as a treatment for PD or dystonia at Beijing Tiantan Hospital between October 2019 and December 2020. Their intraoperative electrophysiological recording data were reviewed so that trajectories with poor neural signal or trajectories in which electrodes deviated significantly from STN, as shown in postoperative reconstruction (see below), were excluded. The study was performed in accordance with the Declaration of Helsinki, and all procedures were approved by the Institutional Review Board of our hospital (permission no. KY2019-097-02). Informed written consents were acquired from all patients. This case series has been reported in line with the PROCESS Guideline (16).

All patients went through a routine neurological assessment conducted by movement disorder specialists, including the Unified Parkinson's Disease Rating Scale (UPDRS) for PD patients, the Unified Dystonia Rating Scale (UDRS), and multiple other scales and examinations, etc., to confirm the diagnosis and indication for surgery (5, 12). The preoperative brain MRI scans of patients were acquired at admission using a 3-Tesla MRI scanner (SIGNA; GE Healthcare, Waukesha, WI, USA), including 3D-T1-weighted (slice thickness: 1 mm, repetition time: 9.4 ms, echo time: 4.3 ms, spacing between slices: 0), axial T2-weighted (slice thickness: 1 mm, repetition time: 7,881 ms, echo time: 104.9 ms, spacing between slices: 0), and coronal T2-weighted (slice thickness: 1 mm, repetition time: 8,947 ms, echo time: 102 ms, spacing between slices: 0) images. The patients received a unilateral (two cases) or bilateral (39 cases) DBS surgery. The procedures of the surgery were described in detail in previous studies (12, 17, 18). Targeting and trajectory planning were conducted on an image-based preoperative planning system (Leksell SurgiPlan 10.1, Elekta Instrument, Sweden) using the direct targeting method as described before (19–21) (Figure 1A, Figures 2A,B). On each side, after incising the scalp and making a burr hole under local anesthesia, a shielded tungsten microelectrode (Neuroprobe, Alpha Omega, Israel; an average impedance of 458 ± 183 kΩ) was inserted into the brain toward the target area driven by a microdrive (Drive Headstage, Alpha Omega, Israel). Electrophysiological signals were recorded to guide the implantation of the DBS leads (Figure 1B) using a Neuro Omega system (Neuro Omega, Alpha Omega, Israel). Microelectrodes were advanced at a step-size of 0.2–0.4 mm once the entrance of the STN was identified, and when the tips of the microelectrodes were near the bottom of the STN, the step-size was adjusted to 0.1–0.2 mm. The gap between the exit of the STN and the putative entrance of SNr was considered SSB. The thickness of the SSB was annotated.

We conducted microstimulation (0.5-s train at 200 Hz in frequency, 300 μA in amplitude, and 60 μs in pulse width) at every stop near the bottom of the STN to investigate the neuronal responses using a similar approach as described in previous literature (15) (Figure 1B). At every stop, the microstimulation was repeated two to four times. The recorded signal was examined to determine whether microstimulation induced the inhibition of neuronal firing. The inhibition period was defined as the time length between the end of the microstimulation train and the first spike afterwards. If the inhibition periods were unidentifiable (lower than 50 ms) in the STN and much longer in SNr (longer than 50 ms, usually longer than 200 ms), the electrophysiological recording in this trajectory was considered concordant with the findings described in previous studies (15, 22); otherwise, it was considered discordant.

When a trajectory with satisfactory signal was obtained, a DBS lead was implanted along the same track for trial stimulation (Figure 1C). The therapeutic windows (i.e., the range between the minimal intensity of stimulation required to obtain meaningful clinical benefits) and the intensity of stimulation at which the first persistent side effect occurred were noted. UPDRS or UDRS assessments were performed by a neurologist. The ultimate decision on the implantation depths of DBS leads in the track was made based upon the consensus among the neurosurgeons, a neurologist, and a neurophysiologist after considering the length of STN, the possible location of SNr, the intraoperative UPDRS/UDRS assessments performed by the neurologist, side effects, and the therapeutic window. The locations of the lead tips were noted for further verification.

After surgery, a standardized postoperative computed tomography (CT) scan was conducted according to existing protocol (12). The postoperative CT was co-registered to the preoperative MRI to localize the electrodes and the tip of the contact array. A semi-automated Matlab toolbox, the Lead DBS toolbox (23), was used to visualize the leads in 3D virtual space (Figure 3). The precision of the reconstruction was verified by overlapping the postoperative CT upon the preoperative MRI (Figures 2C,D). The trajectories whose electrodes were obviously outside STN were excluded. The location of the lead was assessed according to the relative relationship between the tip of the contact array and SSB. The location of the lead was considered satisfactory if the ventral tip of the contact array was within or very close to SSB (i.e., the distance between the tip of contacts and SSB was smaller than 1 mm) as revealed by Lead DBS and the notes of microelectrode recording. The number of contacts within the STN nucleus shown in the reconstructed images was counted. A contact was counted as half if only part of the contact was inside of the STN.

We employed the Fisher exact test for categorical data and the t-test for continuous data to determine statistical significance. All statistical analyses were performed using R software (version 4.0.2) (24). The number of SSB identification was calculated for two groups (trajectories with macrostimulation vs. trajectories without microstimulation). The Fisher exact test was used to test the significance of proportion difference between groups with and without microstimulation. Descriptive statistics including means and standard errors were calculated. Unpaired two-tailed Student's t-test was used to test the significance of the mean difference between groups with identified SSB and without identified SSB.

Forty-one patients (16 males and 25 females; 31 PD and 10 dystonia cases in the cohort, with an average age of 58.2 ± 10.9 years) who underwent STN-DBS surgery (two unilateral and 39 bilateral) were included in the study. The average length of the disease was 8.5 ± 5.3 years. In total, 80 sessions of neural data that were recorded from microelectrodes implanted through different trajectories on these patients were analyzed. Typical STN signal was recorded in 50 trajectories (62.5%) (Figure 4A). In the rest of the 30 trajectories (37.5%), the signal was atypical inside the nucleus. In 56 trajectories (70.0%), putative SNr signal was identified at 1 mm above to 5 mm below the target, featured by a higher firing rate and a lower background compared to STN (Figure 4B). In the other 24 trajectories (30.0%), SNr signal was either not reached out for because a satisfactory coverage of STN was achieved or not identified during the microelectrode recording.

Of the 39 patients who underwent bilateral surgery, 35 of them received unilateral and five received bilateral microstimulation. In total, 155 trials of microstimulation were applied in 45 trajectories (56.3%), while in the other 35 trajectories (43.8%) microstimulation was not conducted. No patient reported any discomfort with microstimulation, and no abnormal scenario (like seizure) was induced. In 38/45 trajectories with microstimulation (84.4%), STN signal could not be inhibited by microstimulation, while SNr presented a long inhibition period following microstimulation (566 ± 217 ms, Figure 5), which was in line with the previous study (15). By contrast, in three trajectories (6.7%), mild inhibitory responses were induced near the bottom of the STN (162 ± 77 ms), and a putative SNr signal could not be inhibited in the other four trajectories (8.9%).

In our study, SSB was identified in 48/80 trajectories (60.0%), either purely by recognizing the exit of STN and the entrance of SNr in microelectrode recording or with the aid from microstimulation to test the inhibitory responses of STN and SNr. The average length of SSB was 0.87 ± 0.32 mm. No difference was detected between the length of SSB in trajectories with and without microstimulation (P = 0.070, Figure 6A). The identification rate of SSB was further compared between the tracks with and without microstimulation. Using microstimulation, SSB was identified in 33/45 trajectories (73.3%), which is significantly higher than those without microstimulation where SSB was found in only 15/35 trajectories (42.9%, P = 0.011, Figure 6B).

In 43/80 trajectories (53.8%), the distal tips of the contacts were placed within or very close to SSB, which is considered satisfactory, while in the other 37/80 trajectories, the distal tips of the contacts were at least 1 mm away from SSB (46.2%), which was considered unsatisfactory. In all trajectories, the mean number of contacts within STN was 2.4 ± 0.9. To determine whether the recognition of SSB might benefit the lead placement, we compared the relative location of the electrode tip and the number of contacts within STN between the trajectories whose SSB was identified and whose SSB was not identified. The percentage of a satisfying electrode tip location was 70.8% in trajectories with SSB identified and 28.1% in trajectories whose SSB was not identified (P = 0.0002, Figure 7A). On the other hand, the contact number within the STN in trajectories with the SSB identified was 2.6 ± 0.9, which was significantly higher than those trajectories whose SSB was not identified (2.0 ± 0.9, P = 0.005, Figure 7B).

To assess the clinical values of microstimulation in terms of therapeutic benefits to the patients, we compared the alterations in UPDRS/UDRS scores assessed during the trial stimulation and before the surgery (ΔUPDRS/ΔUDRS) as well as the therapeutic windows between the tracks with and without microstimulation. The therapeutic windows were 2.27 ± 0.12 V in the tracks with microstimulation (Stim tracks) and 1.93 ± 0.13 V in the tracks without microstimulation (Non-Stim tracks, P = 0.056, Figure 8A). The ΔUPDRS/ΔUDRS scores were 15.13 ± 1.24 in the Stim tracks, which were significantly higher than those of the Non-Stim tracks (10.69 ± 1.14, P = 0.010, Figure 8B). Besides this, we also compared the ΔUPDRS/ΔUDRS scores and the therapeutic windows between the tracks in which the SSB was identified vs. the ones that were not. The therapeutic windows were 2.31 ± 0.11 V in the SSB-identified tracks (SSB tracks) and 1.85 ± 0.14 V in the SSB-unidentified tracks (Non-SSB tracks). The ΔUPDRS/ΔUDRS scores were 15.46 ± 1.23 in the SSB tracks and 9.78 ± 0.97 in Non-SSB tracks. Both the therapeutic windows and the ΔUPDRS/ΔUDRS scores were higher in the SSB tracks than in the Non-SSB tracks (P = 0.014 and 0.001, respectively, Figures 8C,D). These results show that microstimulation is promoting better therapeutic effectiveness.

The study that we presented here is primarily limited by the selection of trajectories. We only picked tracks that electrophysiologically displayed a typical STN signal and radiologically traveled through a good portion of the STN (shown by reconstruction). Another limitation of our study lies in the reconstruction method that we employed. Lead DBS software projects the DBS leads to standardized brain and atlas after normalization. Although studies have verified the accuracy of Lead DBS (23, 26), a tiny discordance between the computed nuclei and the real brain may exist. Therefore, a future study with a better design is needed to investigate the efficacy of microstimulation for guiding lead implantation.