Abstract
Background:
Although earlier studies reported variable speech changes following subthalamic nucleus (STN) deep brain stimulation (DBS) in Parkinson’s disease (PD) patients, the effects of globus pallidus internus (GPi) DBS on speech performance in PD remain largely unknown.
Objective:
We aimed to characterize speech changes following PD GPi-DBS.
Methods:
We retrospectively analyzed clinical and speech outcomes of 25 PD patients treated with bilateral GPi-DBS at a single center. Outcome measures included the Unified Parkinson’s Disease Rating Scale (UPDRS), speech subsystem domains (respiratory, laryngeal, resonance, orofacial, rate, prosody, rhythm, and naturalness), and overall speech intelligibility. Scores at baseline were compared with those at 6 months, 1 year, and the longest clinical follow-up available.
Results:
In the off-medication state, activities of daily living and motor function based on UPDRS II and III significantly improved postoperatively. We observed unique patterns of speech changes in patients with PD following GPi-DBS in the short- (n = 25) and longer-term (n = 8) follow-up periods. Velopharyngeal (resonance), laryngeal components, and prosody worsened after bilateral GPi-DBS (p < 0.015). Speech intelligibility did not worsen after GPi-DBS in the short-term, but there was a trend to deteriorate at long-term follow-up (e.g., one year and beyond). We observed worsening of hypokinetic dysarthria in individual patients. Also, a minority of patients developed stuttering, spastic dysarthria, or ataxic dysarthria.
Conclusion:
Bilateral GPi-DBS worsened several modalities of parkinsonian speech without compromising overall speech intelligibility. GPi-DBS can potentially worsen or induce hypokinetic dysarthria, stuttering, spastic dysarthria, or ataxic dysarthria. GPi-DBS may have different and variable effects on speech function when compared to STN-DBS.
INTRODUCTION
Speech can be negatively affected in patients with Parkinson’s disease (PD) at any stage [1, 2]. The characteristic PD speech pattern is termed hypokinetic dysarthria, which classically includes reduced loudness, reduced prosodic contour or stress, short rushes of speech and/or a fast rate of speech, imprecise articulation, and a breathy or hoarse voice [3]. Various speech production systems may be affected in hypokinetic dysarthria, including respiration, laryngeal, resonance, articulation, and prosody [4]. The degree of impairment within each of these speech systems directly impacts communication effectiveness.
Medication and surgical therapies have variable effects on speech function [4, 5]. Studies have shown mixed results on the impact of dopaminergic treatment on speech and voice in PD [6]. While some have found improvements in consonant articulation and speech prosody with levodopa administration [7, 8], others have found more dysfluent speech [9] or no significant change [10] after dopaminergic therapy. Deep brain stimulation (DBS) in the subthalamic nucleus (STN) and globus pallidus internus (GPi) are effective treatments for medically refractory tremor, motor fluctuations, and troublesome dyskinesia [11, 12]. Currently, most literature has focused on speech changes following STN-DBS [5]. A meta-analysis reported that stimulation-related dysarthria was observed in 9.3% of PD patients treated with STN-DBS [13]. However, the rate of stimulation-related dysarthria may actually be higher in studies where detailed speech evaluations were administered [14–16]. Studies have shown greater deterioration in speech intelligibility at 1 year after STN-DBS than in patients treated with medications alone [15, 17]. Several distinct phenotypic speech/voice disorders have also been described in PD patients treated with STN-DBS, which are uncommon in PD patients without DBS. Some of these changes include strained-strangled vocal quality, spastic dysarthria, or adult-onset neurogenic stuttering [14, 19]. A limitation in comparing prior studies has been the variability in characterizing speech changes in PD, including how investigators refer to repetitive speech disorders in PD (e.g., stuttering or palilalia) [20]. Potential predictive factors for deterioration of speech intelligibility after STN-DBS included lower preoperative on-medication speech intelligibility, longer disease duration, and a relatively medial placement of a left hemisphere STN active DBS electrode contact [17]. Additionally, current spread into the cerebellothalamic and corticobulbar fibers is likely associated with DBS related speech and voice disorders [14, 21].
Several randomized clinical trials have compared STN- and GPi-DBS in patients with PD. Similar improvements in primary outcomes (motor and functional disability) have been reported, and there were no significant differences in adverse effects on speech in either target [22, 23]. However, in these trials, no formal speech evaluations, based on standardized speech assessment batteries, were performed. To date, much less is known about the effects of GPi-DBS on PD speech. In the present study, we evaluated speech changes following bilateral GPi-DBS in PD patients and we utilized standardized speech assessment batteries.
METHODS
Subjects
We collected retrospective patient data from the institutional review board (IRB)-approved database at the University of Florida (UF) (IRB201901807). The inclusion criteria were as follows: 1) diagnosis of PD based on the UK Parkinson’s Disease Society brain bank criteria [24]; 2) bilateral GPi-DBS performed at UF between January 2007 and August 2018; 3) no prior history of stereotactic brain surgery; 4) preoperative speech assessment data with short-term postoperative evaluation at 6 months after GPi implantation (allowed time interval, 3–9 months); and 5) pre- and postoperative motor assessments using the Unified Parkinson’s Disease Rating Scale (UPDRS). Individuals who underwent bilateral GPi-DBS surgeries with an interval between surgeries of 9 months or more were excluded to minimize the impact of disease progression [25].
Surgical procedure
Surgical procedures were performed as previously described [25]. Briefly, preoperative imaging was used to determine the best stereotactic coordinates for lead placement within the GPi. Target refinement was individualized through the identification of white and gray matter, and the selection of a safe trajectory. Using microelectrode recordings and macrostimulation testing, DBS electrodes (model 3387; Medtronic, Minneapolis, MN, USA) were implanted under local anesthesia. Surgeries were typically staged, whereby the first and contralateral leads were implanted on different dates, as this is the standard of care at the institution. The latest neurostimulators available at the time of surgery (Activa PC/SC, Soletra, or Kinetra; Medtronic) were placed about 4 weeks after electrode implantation and activated during the first clinical visit for DBS programming. For the first 6 months, the patients were evaluated monthly to optimize DBS settings and medications. This procedure was followed by visits every 3 months during the first year, followed by annual visits thereafter. Patients had additional visits if clinically indicated.
Clinical and speech evaluations
Baseline demographics included age, gender, age of onset, and age at DBS implantation (Table 1). The clinical assessment included UPDRS II and III; speech sub-scores (items 5 and 18) were also extracted. At baseline and post-DBS follow-up visits, UPDRS-III was completed in the on- and off-medication states. We collected postoperative short- and long-term motor assessments (1 year or more from bilateral GPi-DBS), along with speech evaluations. In the off-medication state, patients were evaluated after overnight withdrawal of dopaminergic medications. In the on-medication state, patients were evaluated under their optimized medication regimen. The levodopa equivalent daily dose (LEDD) was calculated as previously described [26].
Baseline clinical characteristics of the cohort
Clinical characteristics are represented as mean (standard deviation, SD). LEDD, levodopa equivalent daily dose; DBS, deep brain stimulation; UPDRS, Unified Parkinson’s Disease Rating Scale; SD, standard deviation.
Speech evaluations were completed in the on-state of patients’ optimized medications preoperatively, and with both medication and GPi-DBS in the on-state postoperatively. Speech evaluation measures were developed based on a previously described speech evaluation tasks for dysarthria [27, 28]. Trained speech-language pathologists with more than 2-years of experience evaluating motor speech disorders completed all of the speech evaluations. The speech evaluations consisted of several tasks targeting seven speech subsystems. Outcome data consisted of maximum phonation duration on the vowel “ah” (in seconds), speech intelligibility (%), and ratings across the seven speech domains. Perceptual severity ratings were made on a Likert scale, ranging from 0 (no dysfunction) to 7 (anarthric) (Supplementary Table 1) [28]. Motor speech diagnosis (dysarthria type) and severity were also recorded. Speech evaluation measures were recorded at baseline, 6 months, 1 year, and at the last follow-up.
The primary analysis was comprised of differences in speech measures from baseline to 6 months following GPi implantation. Additional post hoc analyses were conducted to assess speech changes at longer follow up intervals (at 1 year and at the last clinical follow-up).
Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics 26 (Armonk, NY: IBM Corp). The normal distribution of data was tested with the Shapiro-Wilk test. Categorical variables were reported as percentages and number of cases, and quantitative variables as mean and standard deviation (SD). We compared baseline characteristics between the patients who completed the regular visits and those who missed long-term speech data using independent t-tests, or Mann-Whitney U tests, as appropriate. The statistical significance was set to a threshold of p value <0.05. The clinical and speech outcome measures at each time-point were compared with baseline status using chi-square tests, paired t-tests, or Wilcoxon signed-rank tests, as appropriate, with Holm-Bonferroni correction for multiple comparisons, which adjusts p values for each hypothesis with a range of significance thresholds (0.017–0.050).
RESULTS
Study population
Twenty-five patients who met the inclusion criteria were analyzed in this retrospective study. The clinical characteristics of the cohort are shown in Table 1. Based on the patients’ dates of DBS implantation, six patients did not reach long-term follow-up (≥1 year) at the time of our analysis. One patient was lost to follow-up (no reason found in the chart). We had to exclude long-term data of 3 patients whose speech evaluations were done in the off-medication state (logistical/scheduling issues). Additionally, some patients simply did not undergo annual speech evaluations despite ongoing DBS programming (n = 7). Regardless, when comparing the clinical characteristics of patients who completed regular visits to date and those who did not, we found no significant differences in their clinical backgrounds (Table 1). For 8 patients who had long-term speech evaluation follow-up beyond 1 year, the mean follow-up duration was 29 months (SD 9.7, range 14–47).
UPDRS outcomes after GPi-DBS
Patients experienced improvements in total UPDRS II and III scores in the off-medication/on-stimulation states at 6 months and 1 year, compared to baseline (pre-DBS) (Table 2). There were no significant differences in total UPDRS III scores in the on-medication states at any of the time points. There were also no significant changes in speech sub-items of the UPDRS following GPi-DBS.
UPDRS outcomes and LEDD changes
*Denotes significance after adjusting familywise error rates (FWER) for multiple comparisons using the Holm-Bonferroni method. UPDRS outcomes and LEDD are represented as mean (standard deviation, SD). BL, baseline; UPDRS, Unified Parkinson’s Disease Rating Scale; LEDD, levodopa equivalent daily dose; DBS, deep brain stimulation; SD, standard deviation. N/A, unable to perform statistical analyses due to insufficient sample size. “Medication on” evaluations were completed based on the dopaminergic regimen at the time of the specified evaluation.
Medication and DBS programming parameters
There were no significant changes in LEDD postoperatively (Table 2). At a group level, the DBS programming parameters remained similar at short- and longer-term follow-up (Supplementary Table 2). Supplementary Table 3 shows individual patient data of lead tip localizations, in relation to LEDD and speech subtypes over time.
Short-term effects of GPi-DBS on speech pattern
No significant difference was found in overall intelligibility between baseline (98.5±3.0%) and 6 months after GPi-DBS (99.1±1.7%) (p = 0.47, Table 3). Regarding specific speech subsystem domains, we found significant worsening in resonance and prosody at 6 months following bilateral GPi implantation compared with baseline (all p < 0.015). Stress component reached p < 0.05, but was not statistically significant after correction for multiple comparisons. We did not observe significant changes following GPi-DBS at the group-level in vocal tremor, loudness, pitch, rate, rhythm, or neurogenic stuttering. However, at the individual patient level, a few patients did show worsening in some of these domains, and two patients developed neurogenic stuttering at 6-months after surgery.
Speech outcomes following bilateral GPi-DBS
Speech outcomes are represented as mean (standard deviation, SD). *Denotes significance after adjusting familywise error rates (FWER) for multiple comparisons using the Holm-Bonferroni method. †Denotes p < 0.05, but not significant after adjusting for multiple comparisons using Holm-Bonferroni method. BL, baseline; DBS, deep brain stimulation; GPi, globus pallidus internus.
Qualitatively, the majority of patients had mild to moderate hypokinetic dysarthria at baseline (n = 19, 76%); five patients (20%) had mixed hypokinetic-hyperkinetic dysarthria (Table 4). When hyperkinetic features were present, they were attributed to on-medication dyskinesias, or dystonia (all features that contributed to the patients being considered for DBS).
In the short-term, six patients experienced worsening of hypokinetic dysarthria (i.e., from mild to mild-moderate, n = 3; mild to moderate, n = 1; mild to moderate-severe, n = 1; mild-moderate to moderate, n = 1). Four patients had slight improvement in their hypokinetic dysarthria (i.e., from moderate to mild-moderate, n = 2; moderate to mild, n = 1; mild to very mild, n = 1). The majority of the cohort had no changes in their hypokinetic dysarthria (Table 4). One patient had worsening of mixed hypokinetic-hyperkinetic dysarthria. Notably, two patients developed additional dysarthria types: namely, mixed hypokinetic-spastic and mixed hypokinetic-ataxic dysarthria (Table 4). Subjectively, 60% of patients (n = 15) had speech complaints in the short-term postoperatively, most commonly complaining of “softer voice” (i.e., hypophonia). A few complained of a “gravelly” voice, associated with throat-clearing. Ten patients had either no speech complaints, or reported subjective improvements in volume and enunciation. However, at the individual patient level, patients’ subjective complaints of worsening speech did not match up with speech domain changes, such as intelligibility.
Long-term effects of GPi-DBS on speech pattern
The scores for speech intelligibility remained stable up to 1 year but showed a tendency to deteriorate at the last follow-up (p = 0.032, not statistically significant after adjusting for multiple comparisons). Likewise, naturalness showed a tendency to deteriorate in the long-term without a statistical significance (p = 0.032 at 1 year, and p = 0.063 at the last follow-up). We found significant worsening in the laryngeal mechanism at 1 year (p < 0.01). The laryngeal mechanism was observed to be more hoarse, breathy, and in some cases strained. The prosodic contour continued to flatten. While we did not find any parameters at last follow-up that reached statistical significance after adjusting p for multiple comparisons, respiratory mechanism had p < 0.05. At the last follow-up, prosody showed a tendency toward statistical significance (p = 0.063).
Qualitatively, two patients had slight worsening of hypokinetic dysarthria at 1 year, compared to baseline and short-term (i.e., mild to moderate, n = 1; mild to mild-moderate, n = 1). At last follow-up, two patients continued to have worsening of hypokinetic dysarthria, compared to baseline and short-term (i.e., mild to moderate, n = 1; moderate to moderate-severe, n = 1) (Table 4). At 1 year, two patients’ hypokinetic dysarthria changed phenotypes to hypokinetic-spastic type and hypokinetic-hyperkinetic type.
Dysarthria type before and following bilateral GPi-DBS
*At short-term follow-up, 2 of 5 patients with baseline hypokinetic-hyperkinetic dysarthria developed hypokinetic-spastic type and hypokinetic dysarthria. †At short-term follow-up, 1 patient with baseline hypokinetic dysarthria developed mixed hypokinetic-ataxia dysarthria. BL, baseline; DBS, deep brain stimulation; GPi, globus pallidus internus; mod, moderate.
DISCUSSION
We evaluated speech function in patients with PD before and following bilateral GPi-DBS and we observed unique patterns of speech changes in the short- and longer-term follow-up periods. Velopharyngeal (resonance), laryngeal components, and prosody significantly changed after bilateral GPi-DBS. Specifically, resonance consistently worsened and transformed into more hypernasal speech within 6 months following GPi-DBS. The laryngeal mechanism was observed to be more hoarse, breathy, and occasionally strained (likely indicative of a spastic component). The prosodic contour continued to flatten, likely from reduced pitch and loudness variation across utterances. Intelligibility did not worsen after GPi-DBS in the short-term, but there was a trend to deteriorate at long-term follow-up (e.g., 1 year and beyond).
To the best of our knowledge, the present study is the first to detail the differential effects of pallidal DBS on speech and voice patterns in PD. There are limited available studies on speech changes in patients with dystonia following GPi-DBS. The reported effects of STN-DBS on speech are multifactorial and variable, with different patterns of dysarthria. Various hypothesized mechanisms include spread of current into the cerebellothalamic and corticobulbar fibers, or resulting from disease progression [5, 21]. Compared to GPi-DBS, typical reported speech changes after STN-DBS include a decline in articulatory precision, prosody, phonation, respiration, rate, resonance, and most commonly, intelligibility [17]. Some of these components have been described as post-STN DBS clusters of speech/voice phenotypes [14]. In our study, we observed deterioration in prosody, resonance, and laryngeal mechanism all following bilateral GPi-DBS.
Every PD patient in our cohort had pure hypokinetic dysarthria or mixed hypokinetic-hyperkinetic dysarthria prior to his/her DBS placement. The pathophysiology of hyperkinetic dysarthria is highly complex and involves multiple factors (e.g., scaling and maintaining movement amplitude and effort, or sensory and temporal processing) [29]. Following GPi-DBS, the majority of the patients continued to exhibit hypokinetic dysarthric features. Interestingly, six patients exhibited slight worsening of the hypokinetic dysarthria at 6 months following DBS, which was possibly stimulation-induced rather than disease progression. In patients with PD or dystonia, more postero-ventral GPi stimulation has been shown to induce hypokinetic/parkinsonian motor and also speech related features [30–32]. There were three patients in our cohort who developed additional dysarthria features following DBS (n = 2 at 6-months, and n = 1 at 1 year). Specifically, two patients developed spastic features, and one developed ataxic features. We suspect that in these cases, there was spread of stimulation into the corticobulbar fibers within the internal capsule that may have accounted for the spastic component, and spread of stimulation to the cerebellar fibers accounting for the ataxic component. Lead tip localization is, unfortunately, not the most informative. Instead, the correlation between lead positions and specific speech disorders should be analyzed using modern methods, such as the volume of tissue activated (VTA) analysis [33]. Thus further imaging, VTA and connectivity analyses will be needed (and are underway) to better understand the mechanisms underpinning speech issues in PD GPi-DBS cohorts.
Importantly, GPi-DBS did not significantly worsen all speech subsystems. GPi-DBS may improve some components of speech, including loudness or vocal tremor. This improvement may possibly be due to a motor benefit on hypokinesia, tremor, and rigidity in the speech-related systems [34, 35]. Neurogenic stuttering is frequently observed in advanced PD patients, presumably because of the dysfunction in the basal ganglia–thalamo–cortical speech network [20]. GPi-DBS can possibly worsen or induce stuttering by disruptions of the speech network. We observed this finding at the individual patient level (n = 2). There are similar reports of worsening of stuttering after STN-DBS, and a few reports of improved stuttering in patients with a childhood history of this disorder [14, 36].
Our study had several important limitations. First, the clinical data were analyzed retrospectively. Second, the relatively small number of patients for long-term assessments limited the statistical power. Additionally, there could have been potential biases in the long-term data, although we could not identify significant differences in the baseline characteristics between the patients who completed regular visits and those who missed their long-term speech assessments. A larger prospective longitudinal study will be required to validate our results and to assess long-term outcomes. Another area of potential bias was that the speech evaluations, although completed by trained speech pathologists, remained a subjective perceptual assessment. A future study would benefit from additional objective acoustic speech analyses [18]. In our study, we did not perform on- and off-stimulation speech comparisons. Therefore, we could not clearly differentiate the effects of stimulation and disease progression. However, we interestingly did observe speech changes even in the short-term (i.e., 6 months), and these types of changes would be unusual with disease progression. Future studies would also benefit from evaluating the functional impact of speech domain changes on patients’ voice/speech-related quality of life, using validated scales such as the Voice Handicap Index [14, 16]. As indicated in patients’ subjective speech complaints (pre- vs. postoperatively), they appear unreliable and do not correlate with changes in speech subsystem domains. Finally, we did not report on unilateral GPi-DBS in this study, and we did not establish a clear association with lead location, programming parameters, fiber tract connectivity, and the overall speech outcome. Lead tip localizations are not the most informative, and we are now preparing for the VTA analyses as a subsequent study.
Conclusion
We observed that in PD patients, GPi-DBS worsened several speech modalities without compromising overall speech intelligibility. GPi-DBS can potentially worsen or induce hypokinetic dysarthria and in some cases even stuttering. We observed spastic or ataxic speech changes following GPi-DBS presumably as a result of the current spread into surrounding structures. The findings from the current and earlier studies suggest that GPi-DBS may have different and variable effects on speech function when compared to STN-DBS in the literature. However, comparative speech studies between GPi- and STN-DBS are needed. More investigation will be required to evaluate the differential pathophysiological mechanisms of stimulation-related dysarthria and to establish better DBS treatment strategies for optimizing motor and speech outcomes.
CONFLICT OF INTEREST
SYC was supported by Smallwood Foundation during fellowship. SYC, TT and NEH declare no competing interests pertaining to this manuscript. SYC and NEH receive salary support from the University of Florida. KWH receives grant support from NIH (grant #5R01HD091658-03) and salary support from the University of Florida. AWS reports grants from the NIH and has received grant support from Benign Essential Blepharospasm Research foundation, Dystonia coalition, Dystonia Medical Research foundation, National Organization for Rare Disorders and grant support from NIH (KL2 and K23 NS092957-01A1). AWS reports receiving honoraria from Acadia, Cavion, Elsevier and MJFF in the past; participates as a co-I for several NIH, foundation, and industry sponsored trials over the years but has not received honoraria. LA works as a consultant and participates in advisory boards for Boston Scientific and Medtronic, and has received honoraria for these services. LA has no conflicts of interest pertaining the scope of this manuscript to be declared. KDF reports grants from NIH, and other funding from Donnellan/Einstein/Merz Chair; grants and non-financial support from Medtronic, grants from St Jude, Functional Neuromodulation, and Boston Scientific, and grants and other funding from Neuropace. Additionally, KDF has a patent US 8295935 B2 issued for a DBS cranial lead fixation device. MSO serves as a consultant for the Parkinson’s Foundation, and has received research grants from NIH, Parkinson’s Foundation, the Michael J. Fox Foundation, the Parkinson Alliance, Smallwood Foundation, the Bachmann-Strauss Foundation, the Tourette Syndrome Association, and the UF Foundation. MSO’s DBS research is supported by: NIH R01 NR014852 and R01NS096008. MSO is PI of the NIH R25NS108939 Training Grant. MSO has received royalties for publications with Demos, Manson, Amazon, Smashwords, Books4Patients, Perseus, Robert Rose, Oxford and Cambridge (movement disorders books). MSO is an associate editor for New England Journal of Medicine Journal Watch Neurology. MSO has participated in CME and educational activities on movement disorders sponsored by the Academy for Healthcare Learning, PeerView, Prime, QuantiaMD, WebMD/Medscape, Medicus, MedNet, Einstein, MedNet, Henry Stewart, American Academy of Neurology, Movement Disorders Society and by Vanderbilt University. The institution and not MSO receives grants from Medtronic, Abbvie, Boston Scientific, Abbott and Allergan and the PI has no financial interest in these grants. MSO has participated as a site PI and/or co-I for several NIH, foundation, and industry sponsored trials over the years but has not received honoraria. Research projects at the University of Florida receive device and drug donations. ARZ serves as a consultant for the National Parkinson Foundation and has received research consulting honoraria from Medtronic, Boston scientific, CNS ratings and Bracket.
Ethical standards
All procedures were performed in accordance with the institutional ethics committee and the Declaration of Helsinki.
