Motor outcome and electrode location in deep brain stimulation in Parkinson's disease

Abstract Objectives To evaluate the efficacy and adverse effects of subthalamic deep brain stimulation (STN‐DBS) in patients with advanced Parkinson's disease (PD) and the possible correlation between electrode location and clinical outcome. Methods We retrospectively reviewed 87 PD‐related STN‐DBS operations at Helsinki University Hospital (HUH) from 2007 to 2014. The changes of Unified Parkinson's Disease Rating Scale (UPDRS) part III score, Hoehn & Yahr stage, antiparkinson medication, and adverse effects were studied. We estimated the active electrode location in three different coordinate systems: direct visual analysis of MRI correlated to brain atlas, location in relation to the nucleus borders and location in relation to the midcommisural point. Results At 6 months after operation, both levodopa equivalent doses (LEDs; 35%, Wilcoxon signed‐rank test = 0.000) and UPDRS part III scores significantly decreased (38%, Wilcoxon signed‐rank test = 0.000). Four patients (5%) suffered from moderate DBS‐related dysarthria. The generator and electrodes had to be removed in one patient due to infection (1%). Electrode coordinates in the three coordinate systems correlated well with each other. On the left side, more ventral location of the active contact was associated with greater LED decrease. Conclusions STN‐DBS improves motor function and enables the reduction in antiparkinson medication with an acceptable adverse effect profile. More ventral location of the active contact may allow stronger LED reduction. Further research on the correlation between contact location, clinical outcome, and LED reduction is warranted.

Although available since 1995 in Finland, only few DBS studies have been published (Erola et al., 2005;Heikkinen et al., 2004). This retrospective study is the first review of DBS outcome in Finland on a larger scale. The objective of this study was to evaluate the outcome of DBS-treated patients with advanced PD at Helsinki University Hospital (HUH) between 2007 and 2014.

| MATERIAL S AND ME THODS
The medical history of 103 patients with PD who underwent STN-DBS at HUH between 2007 and 2014 was reviewed. In 87 patients, all the necessary data and sufficient imaging were available, and the assessment of the active electrode contact could be made. The clinical inclusion criteria for DBS operation were idiopathic PD with suboptimal response to conventional medication, one or more of the following symptoms: daily ON/OFF fluctuations, severe dyskinesia or drug-resistant rest tremor. Positive response to levodopa was required during challenge test. Contraindications were dementia, existing psychosis or severe depression, clinical or radiological suspicion of atypical parkinsonism and/or a significant brain atrophy or vascular changes seen in brain MRI.
To assess the clinical outcome of DBS, UPDRS part III (UPDRS-III) scores at baseline in medication OFF state was chosen for evaluating the baseline severity of Parkinson's disease and at 6 months in medication OFF and DBS ON state to evaluate the efficacy of DBS stimulation. Secondary outcomes were changes of H&Y stage and dopaminergic medication calculated in levodopa equivalent dose, LED, as suggested by Thomlinson et al. (2010). LEDs were obtained at baseline and at 6 months. H&Y staging was derived from the medical records during the pre-DBS screening and at 6 months. The stimulator settings (voltage, pulse width, polarity, and frequencies) at 6 months were used for analysis. Complications including intracerebral hemorrhage (ICH), infection and dysarthria were reviewed. An infection was defined as requiring treatment with antibiotics and/ or revision.
The DBS operation including targeting and imaging was conducted according to common clinical practice and the choice of the operating surgeon. Each patient had a preoperative brain MRI without a frame before the operation day and a CT scan with an attached Leksell stereotactic frame on the morning of DBS operation. In most cases, the targeting was performed by two surgeons individually and the final coordinates were concluded by comparing the two coordinate sets. The DBS operation was performed under light sedation and local anesthesia. Intraoperatively, the accuracy of implantation was confirmed with X-ray fluoroscopy in AP and lateral directions.
Immediately after fluoroscopy, all four contacts were tested intraoperatively for side effects and clinical benefit while the patient was awake.
Analyses of electrode location were performed from MRI and CT scans acquired as part of clinical routine and no extraimaging was made for this study. Image analysis was performed with Agfa Healthcare N.V.'s Impax (version 6.5.5.1608, Belgium) and Brainlab iPlan (version 3.0.5, Germany) stereotactic software. Postoperative CT scans (on the 1. or 2. postoperative day) were reviewed for complications and to determine the amount of intracranial air. The amount of midbrain midline shift compared to skull midline was measured from preoperative and postoperative CT. Postoperative CT controls were fused to preoperative MRI scans using whole brain MRI and CT with the target area visually inspected for sufficiently accurate fusion, which was realigned in AC-PC orientation. Six months' DBS programming record entry was used to find the active electrical contact (defined as the negative electrode) while the implanted pulse generator, implantable pulse generator (IPG) was the positive. It was recorded whether the stimulation was bipolar or monopolar. If there were two active negative contacts, the target point was defined as the middle point between these two contacts. To define the MCP, the anterior and posterior commissure were identified preferentially from 1.5 T T2 image and the anatomical analysis of the target area was performed preferentially from 1.5 T susceptibility weighed images (SWI, SWAN; Vertinsky et al., 2009). In case of suboptimal image quality of these pre-DBS images, we used previous 3 T MRI scans obtained during screening. The MRI images were reconstructed in axial, sagittal and coronal planes as defined by anterior and posterior commissure before electrode location analysis.
Images of MRI and CT fusion and of method of analysis of active electrode location can be found in Supplementary Materials.
The location of the active electrode was determined in three coordinate systems: direct visual analysis of the MRI scans which was correlated by the researcher to the Mai atlas, location in relation to the NR borders (anterior, lateral and superior), similarly as described by others, and location in relation to MCP (Houshmand, Cummings, Chou, & Patil, 2014;Mai, Paxinos, & Voss, 2008;Rabie, Verhagen Metman, & Slavin, 2016;Slavin, Thulborn, Wess, & Nersesyan, 2006). The target coordinates were expressed in relation to midcommissural point (MCP) in anterior commissure-posterior commissure (AC-PC) coordinates and by a method based on direct MRI visualization of the subthalamic nucleus (STN) and/or the nucleus ruber (NR; Rabie et al., 2016).
The location was recorded in all three dimensions as X-, Y-, and Z-coordinates representing mediolateral, antero-posterior, and dorsoventral directions. The location of the median coordinates of electrodes was compared between patients with less than 30% and those with 30% or more reduction in LED between baseline and at 6 months.
Statistical analysis was performed with IBM SPSS Statistics (version 22.0.0, Armonk, NY, USA). Data are presented as median (interquartile range, IQR). Data analysis was carried out using Mann-Whitney U-test when appropriate. Correlations were calculated with Spearman correlation. The study was approved by the HUH Medical Ethics Committee.

| ADVER S E E VENTS
One intracranial hemorrhage was reported (incidence of 1%), with good recovery. Four ventricle punctures were reported during the operation without clinical significance. Skin infections were the most common infections, either of the IPG (3%) or of the trepanation wound (10%). Fourteen (16%) skin infections were treated with antibiotics, yet only in one patient (1%) the IPG and the electrodes had to be removed due to severe infection. Twenty-two patients (25%) reported dysarthria related to DBS, which was confirmed by turning DBS stimulation off at DBS programming session. With four patients, the UPDRS-III subscore number 18 (speech) remained the same at the baseline and at 6 months' follow-up. Only with four patients (5%), the score increased by 2 points which were regarded as a significant change. With fourteen patients, the score increased by one point.
One patient committed suicide within 6 postoperative months.
One postoperative depression was noted and five patients (6%) suffered from a transient confusional state in the first postoperative weeks. No statistically significant difference was seen between electrode contact positions in those patients with neuropsychological adverse effects compared to those without (Mann-Whitney U-test = 0.06-0.98).

| D ISCUSS I ON
The coordinates acquired with different methods correlated significantly with each other. Correlations were under 0.9 in all cases providing support for previous publications that different methods of acquiring coordinates have different validity (Slavin et al., 2006). The ever-improving quality of MRI scans is increasing the capability to directly visualize the STN as shown by DBS operation under general anesthesia based in intraoperative MRI (Matias, Frizon, Nagel, Lobel, & Machado, 2018). However, NR is better visualized in MRI than STN and has been found in some studies to more reliable than left side, that is the more ventral active contact led to greater LED decrease. This novel observation requires more detailed research to be confirmed. Castrioto et al. (2011)  Parkinson's disease specialists on optimal target for STN stimulation.
In this survey, Parkinson's disease specialists were able to point out their preferred position for an active contact in STN on brain atlas.
Some experts preferred the dorsolateral STN and subthalamic area, yet the survey concluded that there is no homogenous perception of the optimal anatomical target and the optimal target needs further specification (Hamel et al., 2017). In Garcia-Garcia's (2016) study (2016), the optimal stimulation site for highest antiparkinsonian advantage was in rostral and most lateral parts of the motor region of STN and at the interface of this region and its adjacent areas (zona incerta and thalamic fasciculus). In Herzog's study with 14 patients, the dorsolateral border of STN or active electrode within STN was most effective when considering motor improvement (Herzog et al., 2004). The possibility of a beneficial effect of a more ventral electrode in our study was found only on the left side where electrodes were also more medial, which correlates with stronger connections to premotor areas, which might explain the finding in part (Romanelli et al., 2005). In addition, STN is located more ventrally in the medial parts. The more ventrally located electrodes might also lead to combined STN and substantia nigra stimulation which has been also studied to provide additional improvement for axial symptoms (Romanelli, Esposito, Schaal, & Heit, 2005;Weiss et al., 2013).
There are some limitations in this study. Firstly, this is a retrospective analysis that restricts interpretation of the results.
Secondly, electrode locations analysis is based on postoperative CT scans taken on first or second postoperative day. Kim et al. (2010) demonstrated a significant difference in the electrode positions between the postoperative CT and CT at 6 months. This observation causes a limitation in our study, as the electrode location was determined from postoperative CT scan. However, the observed shift in midline was negligible and the amount of intracranial air was small. Additionally, presentation of median electrode locations between different study groups and the statistical testing of electrode locations may not represent optimal measure of individual treatment outcomes. However, it may be beneficial that DBS studies provide some information of the electrode location. If a suboptimal benefit of STN stimulation is noted postoperatively, an imaging study may reveal exact location of the electrode.
In our study, UPDRS part III scores decreased significantly (40%) during first 6 postoperative months. This is comparable to earlier studies (Deuschl et al., 2006;Weaver et al., 2009). Herzog and Schüpbach have reported up to 51%-53% decrease in UPDRS part III scores (Herzog et al., 2003;Schüpbach et al., 2013). Our results showed a significant LED decrease at 6 months that is in accordance with the previous studies (Herzog et al., 2003;Schüpbach et al., 2013;Weaver et al., 2009). We found less frequent neuropsychiatric adverse effects than previous studies (Herzog et al., 2003), yet the neuropsychiatric side effects occurred during the first postoperative months as described earlier. However, a routine neuropsychological examination at 6 months' follow-up was not conducted and therefore mild neuropsychological problems might have been unnoticed.
There was not any evidence suggesting that those with neuropsychiatric problems had more ventral active electrodes than those with no side effects in our study. This is contrary to observations by Welter et al. (2014) in which the ventral contact location and the form of the disease (younger age, shorter disease duration and higher levodopa responsiveness) related to stimulation-induced hypomania. This may be due to the methodological differences in studies. Reported rates of ICH were comparable to previous studies Weaver et al., 2009).
In this study, a statistically significant correlation between the active contact location and LED reduction was noted. Further research on the correlation between the active contact location, clinical outcome, and LED reduction is warranted.

ACK N OWLED G M ENTS
There are no acknowledgments for this article.