Introduction
Idiopathic Parkinson’s disease (PD) is associated with speech-related difficulties such as low speech intensity, abnormal voice quality, and speech that is monoloud and monopitch impacting prosody of speech.Reference Duffy1 The indirect pathway of the basal ganglia-thalamocortical circuit is targeted for deep brain stimulation (DBS) in PD patients, as this pathway is related to increased inhibition of movements in PD pathology.Reference Tewari, Jog and Jog2 Although deep brain stimulation of the subthalamic nucleus (STN-DBS) is an effective treatment for most major symptoms of PD,Reference Benabid, Chabardes, Mitrofanis and Pollak3–Reference Pinto, Gentil and Krack6 effects on speech have been inconsistent across studies.Reference Pinto, Gentil and Krack6–Reference Xie, Zhang and Zheng21
STN-DBS Programming
The DBS pulse generator enables programming of the voltage or amplitude of the electrical signal (volts/V), the pulse width or the duration of the electrical pulse (microseconds/μs), and the frequency or number of electrical pulses transmitted per second (hertz/Hz).Reference Isaias, Tagliati, Tarsy, Vitek, Starr and Okun22 Given that the value of these three stimulation parameters (voltage, pulse width, and frequency) can be adjusted independently and set to various combinations, there is the potential to identify optimized settings for speech. Standard therapeutic STN-DBS settings generally use 130–180 Hz, 2.5–3.5 V, and 60–90 μs pulse-width stimulation.Reference Groiss, Wojtecki, Südmeyer and Schnitzler23 Although success of optimization for most major symptoms of the disease is high, preliminary studies suggest that standard STN-DBS stimulation parameters may not be optimized for the treatment of speech symptoms.Reference Pinto, Gentil and Krack6–Reference Klostermann, Ehlen and Vesper9 Still, few studies have examined manipulations of each parameter setting on speech.
Previous studies highlight the inconsistencies with regard to impact of STN-DBS on voice quality, speech intensity, and prosody; for example, some studies have found improvements in these speech symptoms (voice quality, speech intensity, prosody), worsening of symptoms, and others observing no change.Reference Pinto, Gentil and Krack6–Reference Xie, Zhang and Zheng21 It is important to note, however, that previous study conditions included examination of one parameter setting in isolation, examined standard clinical setting combinations with DBS “on” compared to “off,” or did not report specific DBS settings. These studies provide a starting point for programming STN-DBS for speech improvements; however, an important gap in the literature remains. The three stimulation parameters (voltage, pulse width, and frequency) each require programming, and it remains unclear whether different combinations of stimulation settings may result in improved speech outcomes. It is important to examine all potential features of the device. This is particularly true of pulse width and voltage, as relatively few studies have focused on examination of these two parameters. This is also important prior to fully understanding the mechanisms related to speech disorder in PD.
To our knowledge, no previous study has performed a systematic evaluation of a wide range of STN-DBS stimulation parameters on acoustic and perceptual measures of speech in PD.
Purpose
This STN-DBS study involves a systematic manipulation of different amplitude, frequency, and pulse-width settings on speech production in PD. The purpose of the current study was to provide optimized STN-DBS settings for speech intensity, voice quality, and speech prosody in PD.
Methods
Participants
The current study was part of a larger investigation of other motor responses (gait disturbances, bradykinesia, and tremor) to STN-DBS. PD patients’ eligibility for STN-DBS surgery included (1) diagnosis of PD with debilitating motor symptoms and (2) severe motor fluctuations including dyskinesia during “on” phases and disabling “off” periods. Exclusion criteria included dementia or severe cognitive impairment as assessed by the Mini-Mental State Examination (cutoff score of 23/30). Exclusionary criterion for participation in the current study included lack of English proficiency. Ten individuals with PD (females; n = 4) who were receiving bilateral STN-DBS treatment served as participants. Twelve participants were recruited, with two participants forced to drop out of the study due to inability to tolerate all of the setting changes and inability to attend all of the scheduled study visits, respectively. Mean age was 63.9 years (range 52–69 years), and mean years since diagnosis was 10.6 years (range 6–17 years). All participants were diagnosed with PD based on Movement Disorder Society (MDS) criteria and recommended to have STN-DBS surgery by neurologist (MD) based on standard clinical treatment protocol for PD. Participant demographics including Unified Parkinson’s Disease Rating Scale (UPDRS), Montreal Cognitive Assessment (MOCA), and levodopa equivalent dose (LED) scores are provided in Table 1. PD participants continued with their pharmacological treatment following STN-DBS surgery, with standard medication titration as assessed by the neurologist to avoid withdrawal symptoms and dyskinesia.
Table 1: Participant demographics
a LED was calculated according to Tomlinson et al.Reference Tomlinson, Stowe, Patel, Rick, Gray and Clarke24
b UPDRS Part III.
Standard Protocol Approvals, Registrations, and Patient Consents
This study was approved by the Human Subjects Research Ethics Board (HSREB) (Western University Ethics (WUE) No. 103928). Written informed consent was provided by all participants.
Surgical Procedures
Pre-operative magnetic resonance imaging (MRI) and computed tomography were used to determine the best STN location for target stimulation and entry site. Patients were anesthetized with local anesthesia during implantation and a stereotactic frame was used for burr hole drilling and implantation procedures. Five 60 µm diameter tungsten microelectrodes (impedance of 0.5–1.0 mΩ at 1 kHz; FHC Inc., Bowdoinham, ME) were used to identify STN boundaries.Reference Benabid, Chabardes, Mitrofanis and Pollak3 Microelectrode recordings began 10 mm above the surgical target and extended 4–5 mm below the target. The microelectrode track that produced the most beneficial motor response to stimulation (alleviation of tremor/rigidity) and fewest side effects (oculomotor, speech) was determined by the neurosurgeons. A chronic therapeutic lead (Model 3389, Medtronic, Minneapolis, MN, USA; 1.5 mm contact length, 0.5 spacing, 1.27 mm diameter) was then permanently implanted into the selected track. The same process was completed bilaterally with fluoroscopy confirmation completed during and after implantation. The Implantable Pulse Generator (Activa PC) for both sides was then implanted subcutaneously into the subclavicular area under general anesthesia. Both monopolar (pulse generator used as the positive contact point) and bipolar settings (contact points for both cathode and anode) were used depending on optimal symptom alleviation as determined by the neurologist during the programming sessions. Post-operative electrode contact location data are presented in Table 2.
Table 2: Localization of active contacts. The position of the active contact was classified as “within the STN,” “outside the STN,” or “at the interface”
Electrode contact localization was performed using the Lead-DBS toolbox in Matlab (Mathworks, Inc., Natick, MA).Reference Horn and Kühn25 The post-operative MRI was linearly co-registered with the pre-operative MRI. The co-registered acquisitions were non-linearly normalized into MNI space using the DARTEL method implemented using SPM12.Reference Ashburner26 Coordinates of the active contacts were acquired and each contact was classified as either “within” or “outside” the STN, or “at the interface” between the STN and overlying structures.Reference Claire, Ranoux, Guehl, Burbaud and Cuny27 Two authors (AA and GG) independently assessed the anatomical position of each contact in relation to the STN in the coronal, sagittal, and transverse planes. Inter-rater reliability for contact coordinates within AC-PC stereotactic space ranged from 98.3% to 99.9%. ANOVA revealed no significant effect on any of the speech measures based on anatomical contact location.
Study Procedures
Each participant was seen for three baseline and five treatment visits. Baseline visits consisted of the following: (1) pre-operative visit, (2) 1-week post-operative with stimulator “off,” and (3) 2-week post-operative with stimulator turned “on” with high-frequency (130 Hz), mid-pulse-width (90 μs), and low-voltage (1.5 V) settings used. The 5 treatment visits consisted of monthly visits to the laboratory across 6 months during which 24-stimulation parameter setting variations were examined. Settings included permutations of three voltage (low, medium, and high), three frequency (low, medium, and high), and three pulse-width (low, medium, and high) settings. Each visit consisted of four sessions during which four stimulation setting combinations were evaluated. These included the participant’s standard clinical stimulation settings (programmed by neurologist) and three randomly determined experimental setting combinations. Participants were given between 30 and 60 min to adjust to each experimental setting before testing was resumed. Participants received each experimental stimulation setting combination once throughout the entire study. With the exception of the pre-operative session (Visit 1), participants were tested “off” medication. A speech protocol consisting of 5 speech tasks was administered during each of the 20 sessions (4 sessions at each of the 5 visits). Speech was recorded using a unidirectional condenser headset microphone (DPA 4060, 6 cm from the mouth) attached to a portable digital audio recorder (M-Audio Microtrack 2). The following two tasks were analyzed: (1) prolonged vowel “ah” and (2) sentence production of “She Saw Patty Buy Two Poppies.” Measures of speech intensity (dB), voice quality (jitter % or cycle-to-cycle frequency variation and shimmer % or cycle-to-cycle amplitude variation values, harmonics–noise/(H/N) ratio which is a reflection of the amount of additive noise in the voice signal), and speech prosody acoustics (semitone standard deviation (STSD) with higher relative values in this measure indicating improved prosodic variation) were obtained using the software program Praat.Reference Boersma and Weenink28 Average speech intensity was obtained with the root mean squared (RMS) intensity contour method from both a 2-s mid-section of the prolonged vowel and from the entire sentence production with long (+500 ms) unvoiced segments or pauses selectively removed. Voice quality measures were obtained from a 2-s mid-section of the prolonged vowel and STSD from the duration of the sentence production. STSD was calculated from the mean and standard deviation of fundamental frequency (F 0) in hertz using the following equation: STSD = 12/0.301 × log [(Hz mean + Hz SD/2)/(Hz mean − Hz SD/2)]. Listener ratings (5 speech language pathology graduate students) of voice quality and prosody (visual analog rating scales) were obtained from the vowel and sentence audio-recordings, respectively, for each STN-DBS condition. Listeners heard and evaluated each sample twice and average ratings were used for the analysis. The Pearson correlation coefficients obtained for intra-rater and inter-rater reliability were 0.86 and 0.83, respectively.
Analysis
All data were tested for normality using the Shapiro–Wilk normality test. The tests were not significant, and therefore, normality was assumed for all subsequent analyses.
Non-experimental Comparisons
Average data across participants were used for all of the following analyses. Visit 1 (pre-operative) was compared to Visit 2 (post-operative, DBS “off”) to examine any potential micro-lesion effects from the surgical procedure itself.Reference Jech, Mueller and Urgosik29 Visit 1 was also compared to Visit 3 (low-voltage DBS) to examine potential effects of low-voltage setting stimulation for speech when there is no other observable motor benefit. Visit 4 was compared to the final clinical setting to determine the effect of the stimulator settings from initial programming to final clinical programming.
Selection of Optimal Settings for Speech
Settings for each of the three electrical parameters (frequency, voltage, and pulse width) were binned into three categories: low, mid, and high. This binning procedure was selected to assess the relative contribution of each setting in order to simplify potential clinical recommendations. Parameter setting bins are reported in Table 3. The effects on speech at Visits 3–7 (experimental sessions) compared to Visit 8 (standard clinical setting, 6-month post-operative; selected by the neurologist to address cardinal motor symptoms) were examined to determine the effect of STN-DBS parameter permutations and optimization potential for speech. The optimal setting for each speech measure (jitter, H/N, shimmer, STSD, intensity, perceptual ratings of voice quality, and prosody) was first compared to the standard clinical DBS settings. The median values for the standard clinical settings were 130 Hz, 120 µs, and 3.4 V. The standard clinical settings were also compared with the optimal speech settings for each participant (not all participants were able to tolerate each parameter setting tested; as such, if a participant found an experimental setting uncomfortable, the session was terminated and replaced with the following randomly selected setting. Thus, not all participants underwent the same number of optimal setting combinations). Optimal speech settings were identified by first examining each parameter setting in isolation. The optimal frequency of stimulation was determined by comparing results (paired t-tests) from each speech measure across participants for low, mid, and high frequency to the standard clinical setting. The same analysis was completed for pulse width and voltage. All possible combinations of these optimal settings were then analyzed to account for the combined parameters typically adjusted in STN-DBS. These were referred to as the optimal combined settings. Paired, two-tailed t-tests were used to determine whether each of these optimal combined settings led to significant improvements in each of the speech measures compared to the standard clinical setting.
Table 3: Parameter setting binning
Corrections for multiple comparisons in the current study were not completed due to concerns related to Type 2 errors. We present data as preliminary and exploratory, and replications in the future should involve statistical procedures for multiple comparison corrections and larger sample sizes.
Total Electrical Energy Delivered
Total Electrical Energy Delivered (TEED) delivered to the STN was calculated as TEED1 s= [(voltage2 × frequency × pulse width)/impedance] × 1 s.Reference Koss, Alterman, Tagliati and Shils30 Pearson correlation analysis was used to examine the relationship between TEED and each of the speech measures.
Results
Ten individuals with PD (mean age, 63.9 years; range, 52–69 years) who received bilateral STN-DBS participated in the current study. Mean years since diagnosis was 10.6 years (range, 6–17 years). Participants presented with no or only mild dysarthria symptoms at baseline and no other speech-related complaints.
Non-experimental Comparisons
The post-operative DBS off condition (Visit 2) was found to have a significantly reduced STSD (2.79 ± 1.61) compared to the pre-operative condition (Visit 1) (3.51 ± 1.84) (t(9) = 2.77, p = 0.022). The post-operative DBS off condition was also observed to have a reduced vowel intensity (67.97 ± 3.49) compared to the pre-operative vowel intensity (70.58 ± 4.51) (t(9) = 2.553, p = 0.031). This suggests that either there was a detrimental surgical lesion effect on speech prosody and speech intensity or there was a pharmacological impact on these speech symptoms, or a combination of both was observed. The initial clinician-based DBS programming (Visit 4) was found to have significantly higher STSD (4.38 ± 2.07) compared to the Visit 2 (post-operative, DBS off) STSD measure (2.90 ± 1.67) (t(8) = 2.58, p = 0.032), suggesting that although either micro-lesion effects from surgery or lack of PD medication had a negative impact on speech prosody, this was resolved through low-level stimulation. Similarly, increased sentence intensity at Visit 4 (initial programming session) (66.62 ± 2.29) was observed compared to Visit 2 (post-operative, DBS off) (64.61 ± 2.51) and at Visit 3 (post-operative, DBS on, minimal stimulation) (63.75 ± 2.48) (t(8) = −2.941, p = 0.019; t(8) = −5.25, p = 0.001, respectively). Interestingly, we found reduced STSD at Visit 8 (final clinical setting) (2.41 ± 1.45) compared to Visit 4 (initial programming) (4.38 ± 2.07) (t(8) = 2.31, p = 0.050), suggesting a decline in speech prosody by the end of STN-DBS programming. We did not find any other differences in the non-experimental comparisons. These comparisons suggest that vowel intensity and the prosodic measure of pitch variability may be sensitive measures of speech change following STN-DBS surgery. In addition, these comparisons highlight the minimal change in speech intensity, voice quality, and speech prosody from pre-surgical levels to post-surgical clinical settings. This is in contrast to the numerous significant effects found for the optimal experimental conditions explored below.
Optimal Speech Settings
The best speech score for each of the measures (jitter, H/N, shimmer, STSD, intensity, perceptual ratings of voice quality, and prosody) was compared to the standard clinical DBS settings to explore the potential for STN-DBS optimization. Paired t-test analysis revealed significant improvement with the best speech score from experimental conditions compared to the standard clinical setting for all of the speech measures (p < 0.05) (refer to Table 4).
Table 4: Means for each speech outcome measure under standard clinical settings and the best speech scores. p values obtained from paired t-tests for each speech outcome comparing the standard clinical setting and the best speech score
Frequency of STN-DBS Stimulation
Mid frequency led to improved H/N (19.98 ± 3.52) compared to the standard clinical setting (18.77 ± 4.85); however, this only approached significance (t(9) = 1.99, p = 0.078). Low (3.70 ± 0.92) and mid frequency (3.33 ± 0.58) led to improved STSD compared to the standard clinical setting (2.37 ± 1.37) (t(9) = 2.70, p = 0.024; t(9) = 2.47, p = 0.036, respectively). Low (66.30 ± 2.22) and mid frequency (67.48 ± 2.49) also led to improved sentence intensity compared to the standard clinical setting (64.55 ± 3.00) (t(9) = 2.51, p = 0.033; t(9) = 4.11, p = 0.03, respectively).
Pulse Width of STN-DBS Stimulation
Mid pulse width led to improved H/N (20.28 ± 3.37) compared to the standard clinical setting (18.77 ± 4.85); however, this only approached significance (t(9) = 2.1, p = 0.065). Low pulse width led to improved STSD (3.35 ± 0.89) compared to the standard clinical setting (2.37 ± 1.37) (t(9) = 2.40, p = 0.040), and improvement with mid pulse width (3.64 ± 1.14) approached significance (t(9) = 2.08, p = 0.067). Low (67.12 ± 1.96) and mid pulse width (66.39 ± 3.01) led to improved sentence intensity compared to the standard clinical setting (64.55 ± 3.00) (t(9) = 3.53, p = 0.006; t(9) = 2.26, p = 0.050, respectively).
Voltage of STN-DBS Stimulation
Mid voltage led to improved STSD (3.47 ± 0.56) compared to the standard clinical setting (2.37 ± 1.37) (t(9) = 2.71, p = 0.024). Low (67.04 ± 2.12) and mid voltage (67.11 ± 2.43) led to improved sentence intensity compared to the standard clinical setting (64.55 ± 3.00) (t(9) = 3.38, p = 0.008; t(9) = 3.19, p = 0.011, respectively), with high-voltage improvements (66.17 ± 2.31) approaching significance (t(9) = 2.13, p = 0.062).
We did not find any other significant comparisons in other speech measures. The results suggest that, in general, low–mid frequency, low–mid pulse width, and mid–high voltage were associated with improvements in the speech measures, compared to the standard clinical settings.
All possible combinations (eight combinations; refer to Table 5.) of these optimal settings were then tested, and paired t-tests revealed that three specific combinations of settings were associated with improved speech scores compared to the standard clinical settings (3 = low frequency, mid pulse width, mid voltage; 4 = low frequency, mid pulse width, high voltage; 5 = mid frequency, low pulse width, mid voltage; refer to Table 6). Harmonics–noise (21.52 ± 2.65) was associated with improved scores with optimal setting no. 3 (low frequency, mid pulse width, and mid voltage) compared to the standard clinical setting (18.33 ± 4.93) (H/N, t(8) = 2.550, p = 0.034), with shimmer (4.32 ± 1.16) and perceptual ratings of pitch variability (4.6 ± 0.94) comparisons to the standard clinical settings (6.55 ± 3.97; 3.01 ± 1.62, respectively) approaching significance for this setting (p = 0.06; p = 0.08, respectively). Improved speech intensity (vowel, 72.80 ± 3.44; sentence, 67.40 ± 2.45) was associated with optimal setting no. 4 (low frequency, mid pulse width, and high voltage) compared to the standard clinical setting (vowel, 67.61 ± 4.37, t(6) = 4.984, p = 0.002; sentence intensity, 64.11 ± 3.30, t(6) = 2.584, p =0.042), with improved shimmer (3.74 ± .91) and harmonics–noise (20.5 ± 2.86) compared to the standard clinical setting (6.12 ± 3.99; 17.21 ± 4.75, respectively) approaching significance (p = 0.06). Improved STSD (3.04 ± 1.20) and sentence intensity (67.9 ± 2.51) were associated with improvements with optimal setting no. 5 (mid frequency, low pulse width, and mid voltage) compared to the standard clinical setting (2.37 ± 1.37; 64.55 ± 3.0, respectively) (STSD, t(9) = 3.06, p = 0.014; sentence intensity, t(9) = 6.471, p = 0.000), with harmonics–noise (20.35 ± 4.33) compared to the standard clinical setting (18.77 ± 4.85) approaching significance (p = 0.054). Significant comparisons were not found for jitter, shimmer, or perceptual ratings of voice quality measures. It is important to note that the mean standard clinical setting was equivalent to optimal setting no. 1 (low frequency, low pulse width, mid voltage); however, this setting was not related to specific improvements in any of the speech measures evaluated.
Table 5: All possible combinations of identified optimal settings for speech
Table 6: Means (and standard deviation) for each speech outcome measure and the best three optimal setting combinations
* Indicates significant difference with the standard clinical setting <0.05;
** Indicates significant difference with the standard clinical setting <0.000.
Total Electrical Energy Delivered
Pearson correlation analysis of the relationship between TEED and voice quality measures revealed weak correlations (r = 0.162, p < 0.05; r = 0.186, p < 0.01; r = −0.277, p < 0.01) for jitter, shimmer, and H/N, respectively. Additionally, weak negative correlations were found between TEED and the speech measures of vowel intensity (r = −0.175, p < 0.05), and sentence intensity (r = −0.157, p < 0.05).
Discussion
Limitations
Several limitations of the current study are identified. The small sample size of 10 individuals and the large amount of individual variability across STN-DBS settings warrant caution in clinical application of the suggested parameter settings. Second, transient speech effects of the stimulation settings are possible and long-term investigation of the stability of observed speech outcomes is suggested. Furthermore, the current study did not include limb or other axial symptom comparisons. This limitation impacts our understanding of the overall impact of optimization for speech and careful consideration of patient symptomology is recommended during programming. Finally, many patients in the current study received bipolar stimulation; only few receiving monopolar stimulation (refer to Table 2) were able to tolerate the treatment settings. It is important to consider the type of stimulation when programming and caution is recommended as the current study findings relate primarily to a bipolar mode of stimulation. Future research is recommended in the area of monopolar stimulation related to different parameter settings.
Summary
The current study explored performance on a number of speech measures across a range of stimulation parameter manipulations. In the current study, we found that standard clinical settings are not optimized for speech. Results from 10 individuals with PD show that improvements in mean speech intensity, jitter, shimmer, H/N ratio, STSD, and perceptual ratings of voice quality and pitch variability were associated with the following stimulation parameters: lower frequency (60–130 Hz), lower pulse width (60–150 μs), and higher voltage (3–4.5 V). These parameter combinations appear to be associated with improved speech outcomes compared to standard clinical settings. The mean values for the standard clinical settings were 117 Hz (low), 95 µs (low), and 3.4 Volts (mid). Despite the approximation of the standard clinical settings to the optimal settings identified in the current study, the specific combination of low frequency, low pulse width, and mid voltage did not result in improved speech measures in our participants. Therefore, although these results highlight the potential optimization of STN-DBS for speech, they also highlight the complex interaction of the parameter settings when programming the device. A starting point from which exploration of STN-DBS clinical programming may begin includes addressing improved intensity of speech with higher voltage, lower frequency, and mid pulse width. Prosodic concerns may be addressed by exploring mid voltage with mid frequency and lower pulse width. Voice quality concerns could be explored using mid voltage with low frequency and mid pulse width.
Due to variability between patients, speech measures, and low sample size in the current study, these optimization recommendations may not be appropriate for all patients but rather provide a starting point or guideline from which to program DBS devices. Individuals with PD may present with one or multiple pre-morbid speech-related symptoms. Existing concerns may be identified in collaboration with speech-language pathology professional. These concerns may be addressed depending on the relative impact of each speech symptom on the patient’s daily life and communication ability with family and friends. Clinical judgment may be used to determine the optimal setting for each participant; however, to date, there is limited literature support for exploration of all three STN-DBS settings from which neurologists can begin to treat speech symptoms.
It is important to acknowledge the complex interaction between stimulation parameters, positioning of electrodes within the anatomical structures, and the contact positioning of each electrode for optimal clinical benefit.Reference di Biase and Fasano31 Active electrode contacts were localized, and we are able to confirm targets being within, outside, or on the border of the STN. Previous researchers highlight methodological concerns related to electrode contact localization procedures and although we did not find differences in speech outcome measures related to active electrode contact location, future research is required to understand the relationship between anatomical location of contacts and its impact on clinical benefit.Reference Claire, Ranoux, Guehl, Burbaud and Cuny27
Some studies report involvement of the dentatorubrothalamic tract (DRTt) in dysarthria following STN-DBS surgery.Reference Tommasi, Krack and Fraix32, Reference Fenoy, McHenry and SchiessReference Fenoy, McHenry and Schiess33 The DRTt efferent projections connect the dentate nucleus in the cerebellum (deep nuclei), the red nucleus, and the thalamus.Reference Carpenter34 Fenoy et al.Reference Fenoy, McHenry and Schiess33 suggest that reduced perceptual ratings of voice quality and prosody found in their study were related to current diffusion to the DRTt. Other studies cite the impact of current diffusion to corticobulbar fibers as the most probable cause of slurred speech and intensity worsening, as this can lead to contractions of facial, laryngeal, or respiratory muscles.Reference Pinto, Gentil and Krack6, Reference Kumar, Johnson and Marks35 The current findings are not discordant with this hypothesis. Increased frequency of stimulation, which results in the increased spread of neuronal stimulation, indeed led to worsened speech intensity, voice quality, and prosodic control. Increased stimulation settings (above that which is optimal for cardinal PD symptoms) have been associated with reduced speech intensity in another study as well.Reference Pinto, Gentil and Krack6
Positive correlations were found for TEED with jitter and shimmer. This suggests that with decreased electrical current delivered to the STN, irrespective of the specific stimulator parameter settings, there was a decrease in the amount of abnormal vibration patterns of the voice in individuals with PD (increased vocal stability). Further, negative correlational relationships were found between TEED and H/N and speech intensity. Therefore with decreased electrical current, there was a pattern of increased speech intensity and improved distinctiveness of the speech sound harmonics. This is consistent with previous work suggestive of speech deterioration with increased TEED as evaluated using the UPDRS speech rating.Reference Skodda, Gronheit, Schlegel, Sudmeyer, Schnitzler and Wojtecki17 This group of researchers found that higher UPDRS speech scores (worsened speech) were associated with increased TEED to the right STN compared to the left STN.Reference Skodda, Gronheit, Schlegel, Sudmeyer, Schnitzler and Wojtecki17 These results are inconsistent with previous work on variable TEED (TEED that is not held constant) for cardinal symptom alleviation. Previous research suggests that increased TEED leads to improvements in most motor symptoms such as bradykinesia, rigidity, and tremor.Reference Moro, Esselink and Xie36
It is unclear whether pre-morbid speech indices are predictive of STN-DBS outcome. Klostermann et al.Reference Sidiropoulos, Walsh, Meaney, Poon, Fallis and Moro8 included subjects with worse pre-operative voice quality and found resistance to change following stimulation. This is in contrast to Gentil et al.Reference Gentil, Pinto and Pollak12 who found improved voice quality in their participants with relatively stable voice quality with stimulation “off.” Few studies have examined pre-morbid conditions in relation to stimulator settings. This is an important consideration for future research as participants in the current study were identified as having no or mild dysarthria symptoms.
The results of the current study were obtained from standard measures of speech intensity, voice quality, and prosody. Future studies should include other indices of voice quality and intensity valid for PD-related dysarthria such as tremor, subharmonic component, voice irregularity-related analyses, intensity declination, and intensity variability.
In the current study, optimized speech settings did not align with the standard clinical settings used to address the major motor symptoms of PD. Thus, the authors suggest careful clinical judgment when determining STN-DBS programming based on relative importance of addressing major motor symptoms and speech symptoms. Considerations may include different pre-programmed setting combinations depending on the patient’s daily needs, for example, spending time sitting while conversing at home or walking in the grocery store.
Acknowledgements
This research was supported by funding awarded to GG from the Canadian Institutes of Health Research.
Author Contributions
AA designed and conceptualized study; analyzed and interpreted data; and drafted manuscript for intellectual content. SA designed and conceptualized study; analyzed and interpreted data; and revised manuscript for intellectual content. CM revised manuscript for intellectual content. TK revised manuscript for intellectual content. GG designed and conceptualized study; analyzed data; and revised manuscript for intellectual content. MD designed and conceptualized study. MJ designed and conceptualized study.
Disclosures
GG reports a grant from Canadian Institute for Health Research, during the conduct of the study. SA received salary support from Western University and received research grants from Parkinson Canada, Parkinson Society of Southwestern Ontario, and Merz Pharma. AA was a graduate student at Western University and received a graduate student award from Parkinson Canada. TK was a graduate student at Western University and received a graduate student award from the Parkinson Society of Southwestern Ontario. MJ has received grants from CIHR, MITACS, OCE, Merz, Allergan, Abbvie and Boston Scientific. MJ has also received speaker fees from Merz, Allergan, Abbvie and UCB Pharma. MD and CM have nothing to disclose.
