Reduction in voice tremor under thalamic stimulation

Deep brain stimulation (DBS) of the thalamus has been shown to reduce disabling essential tremor (ET) of the hands.^1-5 However, the effect of DBS on tremor in other body sites has not been well described. Voice symptoms are of particular interest given their relatively frequent occurrence in patients with ET (15 to 25%⁶). Although vocal tremor usually does not distort speech significantly, it can prove distracting and is known to provoke a negative stereotype in listeners.⁷ Reduction in vocal tremor has been reported for ET patients following ventralis lateralis thalamotomies.⁸ However, no investigations of DBS have included systematic evaluations of voice. This project is a first effort to provide such observations.

Methods. Subjects. All ET patients seen for DBS follow-up over a 6-month period at the University of Kansas Medical Center were reviewed for possible inclusion in the study. Seven individuals were identified who had clinical evidence of consistent voice symptoms with DBS"off" (5 males and 2 females). Each had undergone the surgery for management of hand tremor as part of a multicenter trial.⁵ The site of stimulation was the ventral intermediate nucleus of the thalamus. Equipment included a Medtronic model 3382 DBS lead and an ITREL II Implantable Pulse Generator System (IPG, Medtronic, Minneapolis, MN). Surgical criteria, implantation technique, and subsequent IPG programming have been described previously.^2,5 Two of the seven patients in this study, both males, had received bilateral implants in staged procedures; the remaining five had unilateral implants (left side). All participants reported right-hand dominance. Each described less severe hand tremor under DBS, with self-ratings recorded as 1 to 3 points lower on a 5-point scale (table). Associated chronic or transitory side effects with stimulation were characteristic of those cited previously (numbness/tingling).^2,5 One individual also experienced a persistent, mild dysarthria (slurred speech) subsequent to his second implant.

Voice evaluations were conducted from 1 to 32 months postimplantation(mean, 18 months). Patient age at the time of assessment ranged from 65 to 80 years. Onset of vocal tremor reportedly emerged after age 45 for all participants and always followed emergence of the hand tremor. One of the seven patients was unaware of any voice tremor. Two individuals (one male and one female) used medication as part of their management of hand tremor(initiated pre-DBS). Medical histories were negative for other factors potentially affecting voice. No patient demonstrated voice problems other than tremor.

Assessments. Voice measures included severity ratings, generated both by the patients and by speech clinicians, and acoustic analyses. The patients' self-ratings were based on a 0- to 4-point scale (4 = severe) and reflected their general impressions of severity with and without DBS in daily activities. All other voice measures relied on audio recordings of the patients producing a series of speech tasks, including three productions of the vowel "ah" (5-second duration each) under three conditions: stimulators on, stimulators off, and a repeat of stimulators on. A minimum of 1 minute was interspersed between conditions. Condition order was the same across subjects. The repeated vowel productions and replication of one condition (stimulator on) were included to ensure that any gains that were identified exceeded natural variation in tremor severity. Recordings were made with a discrete head cassette deck (Nakamichi Corp., Tokyo, Japan) and a head-mounted microphone.

Both the clinical ratings and the acoustic analyses were based on the vowel samples because of their sensitivity in reflecting tremor. The clinical ratings (using the same 0- to 4-point scale) were provided independently by two certified speech-language pathologists and a graduate student in speech pathology in a common listening session. The judges were instructed to rate each vowel, focusing on the center segment of the sample. The specific DBS condition was not identified per sample. Order of conditions was randomized across subjects. All samples were given identical scale values by at least two of the judges. Any discrepancies from unanimous agreement were within 1 scale point. Composite values per vowel and across the three vowels per condition were based on the majority rating of the three judges.

The acoustic analyses (using CSpeech software, Paul Menkovic, Madison, WI) were completed by two of the speech judges who jointly corroborated all measurement points. Values were taken from the center 2-second segment of each of the three vowels per condition. Measures included both rate of tremor(number of cycles per second) and mean relative amplitude of tremor (average across cycles of the difference between the maximal voltage and the minimal voltage, divided by the maximal voltage). To ensure data were representative of typical performance, the average per condition was calculated excluding the vowel with the most dissimilar amplitude value. One patient's sample could not be measured because there were no clear peaks/troughs identifying tremor cycles, although they could be heard on tape.

Results. The table summarizes the effect of DBS on voice for each patient according to the four forms of measurement: patient and clinician ratings of tremor severity, and acoustic analyses of tremor rate and amplitude. The changes reported for the clinical ratings and the acoustic values are those that exceeded the natural variation recorded in the two DBS-on conditions. (Acoustic changes within 1 cycle/sec or 10% relative amplitude are likely due to sampling error). The patients' ratings of hand tremor with DBS off and on are also included for comparative purposes.

Only one patient showed improvement in all measures used (A.L.). Three others reflected gains in at least two of the four measures. One was judged improved by self-report and by tremor amplitude values (J.G.), and another by clinician rating and amplitude data (S.L.). The third patient (D.L.) reflected improvement both by self-report and clinician rating, although no acoustic documentation occurred. A fifth patient judged himself as improved, but no other measure supported the change.

Degree of improvement varied per subject, ranging from a 1- to 3- point change on the severity scale and from a 24 to 60% difference in relative amplitude. Only one individual showed substantial reduction in rate of tremor. Improvement in voice was restricted to those who demonstrated the more severe symptoms (per clinicians' rating or relative amplitude) in the DBS-off condition. Degree of voice change did not appear to parallel degree of reduction in hand tremor.

The individual showing the greatest and most consistent change across measures (A.L.) was one of the two patients with bilateral implantation. Separate right/left trials indicated that tremor reduction was almost solely attributable to the original left-side implant. The other patient who underwent the bilateral procedure (K.E.) showed no consistent gain. However, this second individual also had limited tremor without DBS (clinician rating of 1 point), which precluded any significant shifts in severity.

Discussion. Overall, four of the seven individuals showed documentable reduction in voice tremor under DBS in at least two of the measures used. Corroboration across measures was considered important given potential instability in any one of these values independently. The association of voice improvement with greater initial tremor severity (per clinical ratings or relative amplitude) is unexplained but may just reflect increased ease in measurement. However, the lack of relationship with degree of improvement in hand tremor suggests a discrete effect for the two symptoms. The bilateral nature of voice control might explain the dissimilarity (although the source of the tremor may not be equally distributed bilaterally). If this were the case, then bilateral implants might be predicted to have the greater impact on voice. Unfortunately our data were too limited to resolve this question.

These findings of reduced voice tremor under DBS have not been reported previously. However, Goldman et al.⁸ reported a voice effect with thalamotomy. Five of their seven ET patients with vocal tremor were improved following surgery, with three showing total resolution of voice symptoms. Although incidence of reduced tremor was grossly comparable in the current study (four of seven patients), no one experienced a similar elimination of tremor with DBS. This could represent a true difference in the outcomes between the two studies. However, Goldman et al.⁸ did not fully describe their measures or provide specific values, precluding detailed comparison.

Any discrepancy with this related study is not unexpected given procedural differences, including the treatment applied. Discrepancies among our own measures per patient are less easily explained. Some of the apparent inconsistencies might have been resolved had we accounted even more for natural variability in tremor (e.g., by replicating all conditions). Reliability of measures also could be questioned, although agreement among at least two of the examiners for the clinical and acoustic values should have minimized this potential confounder. Restricting the measures to identical samples (i.e., all vowels and the exact same segment of the vowel) might have helped, too. Specific measures also may have contributed to these mismatches in evidence of tremor reduction. For example, data for rate of tremor may have been misleading for several samples. Distinguishing tremor cycles in acoustic displays often can be difficult given "noise" in the signal. Choice of the starting point for the measured segment also can influence values. Consequently, some differences in rate of tremor could be explained easily by sampling difficulties and differences.

Some data discrepancies may have been specific to particular patients. Both C.F.'s and J.G.'s self-ratings of improvement were not corroborated clinically (see table). However, gains were noted for these two individuals in one, but not both, of the clinical ratings for the replicated samples. These patients might have based their self-ratings on this occasional improvement. It could be, too, that J.G.'s amplitude reduction (the lowest of the three reported) was too limited to provoke a change in the clinical rating, particularly in light of the increased rate of tremor. S.L.'s failure to note change, although documented in two other measures, was more surprising. However, she had commented that both her hand and voice tremor were worsening, which may have made her less sensitive to gains with DBS. The lack of acoustic documentation for D.L. is unexplained although the previously noted limitations in acoustic analyses could have had some influence.

Despite these qualifiers, findings per patient are comparable across at least two of the four forms of measurement used, strengthening the interpretation that reduction in voice tremor occurred as a function of DBS for the majority of our sample. Associated medications were not seen as a contributor to the measured effects because they were constant across conditions and applied to only two individuals, one of whom profited from DBS and one of whom did not. We also considered whether voice improvement had any relationship to site of the implants or to stimulator settings. Neither appeared to be a factor, but our sample is small, which may mask certain correlates.

Given that the changes in voice with DBS typically were not dramatic per patient, improvement in this feature is not likely to serve as the sole incentive for intervention of this type. However, similar outcomes could have relevance for individuals with significant voice tremor. Anecdotal reports from our patients suggested that these voice gains were a distinct bonus to this procedure.

Acknowledgment

Thanks are extended to S. Blazak for assistance with the data acquisition and analysis.