Motor speech patterns in Huntington disease

Abstract

Objective

Dysarthric speech of persons with Huntington disease (HD) is typically described as hyperkinetic; however, studies suggest that dysarthria can vary and resemble patterns in other neurologic conditions. To test the hypothesis that distinct motor speech subgroups can be identified within a larger cohort of patients with HD, we performed a cluster analysis on speech perceptual characteristics of patient audio recordings.

Methods

Audio recordings of 48 patients with mild to moderate dysarthria due to HD were presented to 6 trained raters. Raters provided scores for various speech features (e.g., voice, articulation, prosody) of audio recordings using the classic Mayo Clinic dysarthria rating scale. Scores were submitted to an unsupervised k-means cluster analysis to determine the most salient speech features of subgroups based on motor speech patterns.

Results

Four unique subgroups emerged from the cohort of patients with HD. Subgroup 1 was characterized by an abnormally fast speaking rate among other unique speech features, whereas subgroups 2 and 3 were defined by an abnormally slow speaking rate. Salient speech features for subgroup 2 overlapped with subgroup 3; however, the severity of dysarthria differed. Subgroup 4 was characterized by mild deviations of speech features with typical speech rate. Length of CAG repeats, Unified Huntington’s Disease Rating Scale total motor score, and percent intelligibility were significantly different for pairwise comparisons of subgroups.

Conclusion

This study supports the existence of distinct presentations of dysarthria in patients with HD, which may be due to divergent pathologic processes. The findings are discussed in relation to previous literature and clinical implications.

Dysarthric speech is a common clinical symptom of Huntington disease (HD), present in 93% of individuals,¹ yet few studies carefully characterize speech patterns in patients. With increasing duration of motor symptoms, dysarthria severity progresses and patients have increasing needs for speech–language pathology to treat deficiencies in communication and speech. Motor manifest symptoms in HD are typically hyperkinetic, and include myoclonus, chorea, and ballismus, where chorea is the most frequently encountered symptom. Not surprisingly, the motor speech disturbances of patients with HD are commonly referred to as hyperkinetic dysarthria. This categorization describes speech with variable rate, prolonged intervals, inappropriate silences, reduced pitch variability, irregular and imprecise articulation, phonatory deviations, and sudden forced inspiration or expiration.^2,3

The earliest description of hyperkinetic dysarthria was based on 2 separate cohorts of patients, with no clinical descriptors of diagnosis or symptom other than chorea or dystonia.³ Empirically, however, motor speech patterns in HD are diverse. For instance, hyperkinetic speech patterns of patients with HD are not always distinct from other neurologic conditions.⁴ Other studies have attempted to reduce the heterogeneity of their cohort with HD by examining speech profiles of a choreatic vs bradykinetic subgroup based on the patients’ motor profiles.⁵ Thus, a comprehensive characterization of speech patterns in HD is needed. In the present study, we systematically investigated the speech perceptual characteristics of a large cohort of individuals with HD to identify potential motor speech subgroups. Such insights will improve diagnostic precision, guiding management of dysarthric symptoms for this population.

Methods

Standard protocol approvals, registrations, and patient consents

This study was approved by the institutional review board at Vanderbilt University. All speech recordings were accessed retrospectively. Raters were consented for study participation and compensated for their time.

Materials

Patients with HD were assessed as part of clinical visits at the Vanderbilt University Medical Center Huntington’s Disease Center of Excellence. All completed a standardized speech assessment. Two reading tasks were of interest: the Sentence Intelligibility Test (SIT)⁶ and Rainbow Passage.⁷ Speech assessments were recorded using a Tascam (Santa Fe Springs, CA) digital recorder (DR-100MKII) and a lapel microphone (Audio-Technica [Tokyo, Japan] AT899) with a microphone-to-mouth distance of approximately 6 inches. For the purpose of this study, speech recordings of 48 patients were selected retrospectively based on the following inclusion criteria: (1) a diagnosis of HD by a board-certified neurologist, (2) genetic testing confirming a diagnosis of HD, (3) no diagnosis of any comorbid neurologic disorders that would affect speech (e.g., stroke, traumatic brain injury), (4) native English speaker, (5) perceived dysarthria by study authors (S.K.D., A.M., and M.d.R.), and (6) SIT score between 80% and 100%. The SIT score range was selected to reduce the effect of dysarthria severity as a driving factor in the cluster analysis and only 1 patient recording was excluded from the study due to low intelligibility. In addition to the recorded reading tasks, the following additional information was collected: Unified Huntington’s Disease Rating Scale (UHDRS)⁸ motor assessment (to determine total maximal chorea, total maximal dystonia, and overall scores), the number of CAG repeats (genetic mutation responsible for HD pathology), CAP score [CAG age product used to demonstrate the effect of age and CAG length; ],^9,10 concomitant medications, and demographic information. Medications were examined due to previous findings supporting limb motor changes for patients with psychiatric disorders treated with antipsychotics^11,12 and motor speech changes for individuals with HD treated with antipsychotics.¹ While little information exists to determine the effect of vesicular monoamine transporter-2 (VMAT-2) inhibitors on motor speech, we included this as a variable due to its effect on presynaptic dopamine release. A summary of the obtained patient information is provided in table 1.

Table 1.

Demographic and disease characteristics for all patients

Raters

The speech recordings of patients with HD were provided to 6 consented raters. All passed a hearing screening and were second-year graduate students in speech–language pathology, native English speakers, and blinded to the patients’ diagnosis and study hypotheses. Each rater completed a 16-week course in motor speech disorders and an additional 2-hour training module, specific to this study, during which they were familiarized with the classic Mayo Clinic dysarthria rating scale.^3,13 This dysarthria assessment scale requires the rater to indicate the degree to which specific speech characteristics are deviant (impaired) from normal speech by providing a score ranging from 1 (normal) to 7 (severely impaired). The scale consists of 38 items that belonged to 7 categories: pitch characteristics (items 1–4), loudness (items 5–9), vocal quality (items 10–18), respiration (items 19–21), prosody (items 22–31), articulation (items 32–36), and general impression (items 37–38). The general impression category includes speech intelligibility (understandability of speech) and bizarreness (the amount that speech calls attention to itself because of unusualness or peculiarity).³

The training session followed similar procedures to those described by Darley et al.³ Each speech sample was played multiple times and during each round, raters scored only the items belonging to one category (e.g., items 1–4 belonging to “pitch characteristics”). Three patient recordings of the SIT were used for training purposes. The same 3 patients were included in later study tasks but using recordings of the Rainbow Passage.⁷ Raters received feedback on their scores by comparing their responses to those agreed upon by study authors (S.K.D., A.M., and M.d.R.). We considered the training session successful when at least 80% of the scores fell within 1 point of the responses agreed upon by study authors for 2 patient recordings. All raters passed the training based on this criterion.

Experimental task

Each rater completed the dysarthria assessment scale for all 48 recordings of patients with HD across 3 to 6 listening sessions, each lasting up to 90 minutes, including breaks to minimize fatigue. Scoring procedures were consistent with those implemented during the training session; however, the presented speech samples were now the readings of the Rainbow Passage.⁷ Raters were instructed to score the items of one category at a time and to replay the audio sample as often as needed. Once all 38 items were scored for one patient with HD, raters proceeded to the next patient.

Statistical analysis

For each of the 38 items on the dysarthria rating scale, a mean score was calculated for each patient with HD based on the raw scores provided by the 6 raters, and total sample mean scores of each item were calculated. These were ranked from highest to lowest, where the highest score was considered the most severely impaired speech characteristic, while the item with the lowest score indicated the least impaired speech characteristic. Mean scores derived from the cohort of patients with HD in the current study were compared descriptively to those examined in Darley et al.^3,13 to determine similarities or differences in salient characteristics.

When the mean score of an item (calculated based on the scores of the 6 raters) was less than 2.0 on the 7-point scale in all 48 patients with HD included in the study, the item was disregarded from further analysis because the ratings indicated that this speech characteristic was perceived as normal in all patients of our cohort.³ Based on this rule, 6 items were disregarded: pitch breaks, hyponasality, nasal emission, grunt at end of expiration, increase of rate in segments, and repeated phonemes.

The remaining 32 items were submitted to an unsupervised k-means clustering approach to identify whether subgroups of speakers existed within the sample of patients with HD. This analysis was completed without prior knowledge of the number of clusters within the sample. Based on the adjusted R² and Akaike information criterion methods, 4 clusters were deemed to be most appropriate. A multinomial least absolute shrinkage and selection operator (multi-class LASSO) and 500 bootstraps were performed to determine which items of the dysarthria scale were most important for the identification of cluster membership. Item mean scores were then used to rank symptoms within each cluster from highest to lowest (most to least impaired).

Finally, to identify clinical variables potentially associated with subgroups, between-group analyses of variance and subsequent Tukey pairwise comparisons were completed for the continuous clinical variables such as age, UHDRS motor assessment (to determine total maximal chorea, total maximal dystonia, and overall scores), number of CAG repeats, CAP score, percent intelligibility, and disease duration in years. Subgroup differences for the categorical variables such as sex and medication status were determined by computing individual logistic regression with subsequent pairwise comparisons using Tukey honestly significant difference. Medication status was categorized as “not medicated” and “medicated” for specific drug types known to affect hyperkinetic movements (e.g., atypical antipsychotics and VMAT-2 inhibitors).

Data availability

Anonymized data will be shared by the authors on request from any qualified investigator.

Results

Rater reliability

Analysis of interrater agreement revealed an intraclass correlation coefficient (ICC) average measures value of 0.91. Analysis of intrarater agreement revealed an ICC average measures value of 0.907 for rater 1, 0.85 for rater 2, 0.944 for rater 3, 0.896 for rater 4, 0.881 for rater 5, and 0.974 for rater 6. These results indicate that ratings within and across listeners were relatively consistent.

Descriptive speech patterns in HD

Mean scores across all 38 items of the dysarthria rating scale for the entire patient sample are provided in figure 1 and ranked means of items are shown in table 2. We report mean scores of 2.0 and above as salient features of dysarthric speech in patients with HD (reflecting substantial deviations from normal speech) while mean scores between 1.5 and 2.0 are considered additional abnormalities (reflecting only subtle deviations from normal speech) on the 7-point scale. Across all patients, the most common category of impairment was prosody, as indicated by high scores on items that addressed speech rate, presence of inappropriate silences, and presence of excess and equal stress. Other items that received high scores included monopitch, monoloudness, harsh voice, imprecise consonants, and irregular articulatory breakdowns.

Figure 1. Speech perceptual characteristics for entire cohort.

Mean scores, showing amount of deviation from normal, of each item on the Mayo Clinic 7-point dysarthria rating scale of all 48 patients with Huntington disease (HD). Deviations in rate include both patients with slow and patients with fast speaking rate. Items listed above the dashed line have a mean rating of 2.0 or greater, which indicates that these items are salient features (substantial deviation from normal speech) of dysarthric speech for the particular subgroup. Items between the dashed line and the dotted line have a mean rating between 1.50 and 2.0, which indicates only subtle deviations from normal speech. Error bars indicate the SD. Item labels include the following: 1 = pitch level, 2 = pitch breaks, 3 = monopitch, 4 = voice tremor, 5 = monoloudness, 6 = excess loudness variation, 7 = loudness decay, 8 = alternating loudness, 9 = loudness (overall), 10 = harsh voice, 11 = hoarse (wet) voice, 12 = breathy voice (continuous), 13 = breathy voice (transient), 14 = strained–strangled voice, 15 = voice stoppages, 16 = hypernasality, 17 = hyponasality, 18 = nasal emission, 19 = forced inspiration–expiration, 20 = audible inspiration, 21 = grunt at end of expiration, 22 = rate, 23 = phrases short, 24 = increase of rate in segments, 25 = increase of rate overall, 26 = reduced stress, 27 = variable rate, 28 = intervals prolonged, 29 = inappropriate silences, 30 = short rushes of speech, 31 = excess and equal stress, 32 = imprecise consonants, 33 = prolonged phonemes, 34 = repeated phonemes, 35 = irregular articulatory breakdown, 36 = distorted vowels, 37 = intelligibility (overall), 38 = bizarreness (overall).

Table 2.

Comparison of current findings to previous descriptions of hyperkinetic dysarthria

Previously reported salient speech perceptual characteristics for patients with hyperkinetic dysarthria due to chorea are shown in table 2.³ We replicate previous findings for the presence of monopitch, imprecise consonants, inappropriate silences, harsh voice, prolonged phonemes, reduced stress, and strained–strangled voice. In contrast to previous work, distorted vowels and variable rate received less severe ratings, and monoloudness, irregular articulatory breakdown, and excess and equal stress received higher impairment ratings. We found low scores (indicating no impairment) for the presence of prolonged intervals, excess loudness variation, short phrases, and hypernasality (see table 2 for specific deviations in item rankings between the 2 studies). Finally, we extend previous work by emphasizing 2 groups based on speech rate: 8 patients with HD had an abnormally fast rate (2.0 or greater) and 23 patients demonstrated abnormally slow rate (−2.0 or lower). The original study exclusively noted variable rate for patients with chorea.

Expected speech patterns for patients with hyperkinetic dysarthria due to dystonia are shown in table 2.³ We replicate findings of imprecise consonants, harsh voice, irregular articulatory breakdown, monopitch, monoloudness, prolonged phonemes, and reduced stress. In contrast, our raters scored distortion of vowels and strained–strangled voice lower, and slow rate and inappropriate silences as more severe. Four items (short phrases, prolonged intervals, excess loudness variation, and voice stoppages) perceived as impaired for patients with dystonia in the original study were found to be normal in the current study. Neither description of hyperkinetic dysarthria due to chorea nor due to dystonia from the previous study included the presence of audible inspiration among salient features.³ In the current study, audible inspiration was identified as a subtle deviation.

Speech subgroups

Cluster analysis revealed 4 subgroups of patients with HD (table 3, figure 2). Multinomial LASSO and 500 bootstraps with percent chosen >80% identified 4 speech perceptual items as important when sorting patients into these subgroups: (1) monoloudness, (2) rate, (3) intelligibility, and (4) bizarreness. Within each subgroup, items with mean scores of 2.0 or above were considered salient features of that subgroup, but items with scores of 1.5 and greater are displayed in table 3.

Table 3.

Mean ratings by subgroup

Figure 2. Speech perceptual characteristics by subgroup.

(A–D) Mean scores, showing amount of deviation from normal, of each item on the Mayo Clinic 7-point dysarthria rating scale by subgroups resulting from the cluster analysis. The deviations in speaking rate are noted (i.e., noting whether speaking rate is fast, slow, or normal for each subgroup). Deviations in rate include both patients with slow and patients with fast speaking rate. Items listed above the dashed line have a mean rating of 2.0 or greater, which indicates that these items are salient features (substantial deviation from normal speech) of dysarthric speech for the particular subgroup. Items between the dashed line and the dotted line have a mean rating between 1.50 and 2.0, which indicates only subtle deviations from normal speech. Error bars indicate the SD.

Subgroup 1 (fast rate subgroup–mild dysarthria) included 9 patients. The speech features in this group were monopitch, monoloud, harsh voice, abnormally fast rate, reduced stress, inappropriate silences, imprecise consonants, and irregular articulatory breakdown. When compared to the other subgroups, subgroup 1 is unique in having an abnormally fast speaking rate and reduced stress. Additional less prominent speech characteristics included distortion of vowels, excess and equal stress, audible inspiration, strained–strangled voice, voice tremor, variable rate, and short rushes of speech.

Subgroup 2 (slow rate subgroup–moderate dysarthria) included 9 patients with HD. Speech features of this subgroup were monopitch, monoloud, imprecise articulation, irregular articulatory breakdown, abnormally slow rate, inappropriate silences, distorted vowels, prolonged phonemes, harsh voice, strained–strangled voice, audible inspiration, and excess and equal stress. Articulation and prosody characteristics were rated more severe than other categories (e.g., pitch, loudness, respiration, vocal quality). Subgroup 2 had abnormally slow rate and unique deviations of vocal quality, articulation, and respiration. Additional, less prominent, speech characteristics included prolonged intervals, reduced stress, alternating loudness, breathy voice, hoarse (wet) voice, phrases short, excess loudness variation, and loudness decay. Similar to subgroup 1, variable speaking rate was also noted as a subtle deviation from normal.

Subgroup 3 (slow rate subgroup–mild dysarthria) included 16 patients with HD. This speech pattern was characterized as monoloud, monopitch, with excess and equal stress, abnormally slow rate, inappropriate silences, imprecise consonants, harsh voice, and irregular articulatory breakdown. Patients in subgroup 3 had an abnormally slow speaking rate. Additional, less prominent, speech characteristics included prolonged phonemes, audible inspiration, prolonged intervals, variable rate, distorted vowels, and reduced stress.

Subgroup 4 (normal rate subgroup–mild dysarthria) included 14 patients with HD. The most noted perceptual characteristic in this group was monoloud. However, additional, subtle deviations were noted including monopitch, harsh voice, audible inspiration, imprecise consonants, excess and equal stress, strained–strangled voice, inappropriate silences, and irregular articulatory breakdown. Unlike the other subgroups, speaking rate was perceived as normal for subgroup 4. This subgroup also had lower scores for intelligibility and bizarreness compared to other subgroups, indicating a lower level of speech impairment severity.

Clinical differences between speech subgroups

Clinical variables for each subgroup are shown in table 1. Between subgroups, age, CAG repeat length, UHDRS motor score, and disease duration varied greatly. All patients had confirmed diagnosis through genetic testing but CAG repeat values were unavailable for 8 patients (3 in subgroup 2, 4 in subgroup 3, and 1 in subgroup 4) who were subsequently left out of analyses for CAG repeat value and CAP score. Pairwise subgroup comparisons for clinical variables are shown in table 4. Subgroups did not differ significantly in age, sex, years of disease duration, total maximal chorea, total maximal dystonia, CAP score, or medication state (medicated or not medicated). Percent intelligibility was significantly lower, indicating less intelligible speech, for subgroup 1 compared to subgroup 3 and subgroup 4. Percent intelligibility was also significantly lower, indicating less intelligible speech, for subgroup 2 compared to subgroup 3 and subgroup 4. The number of CAG repeats was significantly greater in subgroup 1 compared to subgroup 4. Pairwise comparisons of UHDRS motor score revealed significant differences when comparing subgroups 2 and 3 and subgroups 2 and 4 with subgroup 2, presenting with the greatest level of motor impairment.

Table 4.

Subgroup pairwise comparisons for clinical variables by subgroup

Discussion

The identification of distinct speech motor subgroups yields important implications for assessments of dysarthria, characterization of disease severity, and clinical management of patients with HD. Study results replicated, in part, previously observed speech patterns of hyperkinetic dysarthria due to HD,³ but further extend the descriptions regarding dysarthric subgroups. For instance, we find that speech patterns overlap with those seen in patients with dystonia (e.g., imprecise and irregular articulation, slow rate). The only speech feature evident in our patients, but not previously identified in patients with dystonia or chorea, is audible inspirations. It is possible that this speech feature is indeed the result of choreatic movements affecting the coordination of phonatory and respiratory muscles in some individuals with HD.

A major discrepancy between previous and current findings is the listeners' perception of speech rate. We found that when patients with HD demonstrated a deviant speech rate, they had either an abnormally fast or an abnormally slow speaking rate. Previous studies describe a variable speaking rate (i.e., ranging within one patient anywhere from abnormally slow to abnormally fast), particularly in patients with chorea. Further, patients with dystonia were previously described as having a mildly slowed rate.³ Others have described the speech of patients with dystonia as abnormally slow with irregular and sustained rhythm and the speech of patients with chorea as abnormally fast in rate and irregular in rhythm.² It is possible that the discrepancies between current and previous findings relate to the presence of divergent pathologic processes. Empirically, it is known that motor manifestation can be heterogeneous in HD, with some patients presenting with abnormalities such as dystonia, rigidity, parkinsonism, or ataxic symptoms,^14–16 including the described Westphal variant of HD (akinetic-rigid).¹⁷ There may be pathologic correlates of this heterogeneity, such as differing patterns of cerebral atrophy.^18–21 Here, we discuss these findings, and the clinical implications of this work.

The cluster analysis revealed 4 distinct speech profiles within our cohort of patients with HD. Differences in bizarreness and intelligibility ratings among patients explained, in part, the formation of subgroups and indicate that patients with HD differed in the severity of their speech symptoms. However, distinct differences in speaking rate (abnormally fast, abnormally slow, normal) as well as the presence or absence of perceived monoloudness also contributed to the formation of the 4 subgroups. Rate and loudness modulation appear to be critical features in the identification of distinct motor speech subgroups for patients with HD.

Ranked speech characteristics for subgroup 1 (fast rate subgroup–mild dysarthria) are fairly consistent with previous descriptions of hyperkinetic dysarthria due to chorea.^2,3 However, without knowing the underlying etiology of the dysarthria, speech features of subgroup 1 may also suggest the presence of hypokinetic dysarthria, particularly due to the presentation of cardinal features of hypokinetic dysarthria such as abnormally fast speaking rate, monoloudness, monopitch, and reduced stress.³ Rate has been described as variable for individuals with hyperkinetic and hypokinetic dysarthria, with hypokinetic dysarthria tending to have a slightly faster rate of speech.³ The only salient feature of hypokinetic dysarthria that subgroup 1 does not demonstrate is continuous breathiness. The overlap of speech patterns of subgroup 1 and hypokinetic dysarthria is consistent with findings of a previous study on patients with HD.⁵ In this study, patients with HD were grouped by their motor profile (i.e., bradykinetic or choreatic). The speech of the patients with the bradykinetic motor profile was associated with speech patterns of hypokinetic dysarthria due to Parkinson disease.⁵ Furthermore, in a perceptual study that used a free classification paradigm to assess similarities of speech features across patients with dysarthria due to various etiologies, one patient with HD was also grouped with individuals with Parkinson disease based upon dysarthria characteristics.⁴

It is possible that the individuals in subgroup 1 may present with chorea and additional movement abnormalities (e.g., rigidity, bradykinesia). Alternatively, medications that reduce hyperkinetic movement (e.g., atypical antipsychotics and VMAT-2 inhibitors) may contribute to bradykinetic and hypokinetic speech. In fact, all 9 patients of subgroup 1 were prescribed one or more of these medications. Although medication status was not significantly different across the 4 subgroups, additional factors, such as dose and duration of exposure, may account for the distinct motor speech patterns of these patients. Support for this notion is provided by reports of changes in limb motor function (i.e., hand writing) that varied systematically with the dosage of atypical antipsychotics in patients with psychiatric disorders,^11,12 and articulation, pitch, and loudness differences in patients with HD treated with antipsychotics.¹

Subgroups 2 (slow rate subgroup–moderate dysarthria) and 3 (slow rate subgroup–mild dysarthria) were associated with the slow subgroups of speakers. Although subgroups 2 and 3 exhibited similar speech patterns, listener scores indicated that the severity of these perceived impairments differed among these subgroups, especially the severity of ratings for intelligibility and bizarreness. It is therefore likely that patients with HD in subgroup 2 may have a more advanced stage of speech deterioration than patients in subgroup 3. This notion is further supported by pairwise comparisons showing significantly reduced speech intelligibility and UHDRS motor performance for subgroup 2 when compared to subgroup 3. The speech patterns of the slow subgroup greatly overlapped with those most commonly observed in hyperkinetic dysarthria of patients with dystonia.^2,3 Subgroups 2 and 3 present with some degree of dystonia based upon mean total maximal dystonia scores, although pairwise comparisons revealed no significant differences from other subgroups. Similar to the slow rate in these subgroups, a recent study recognized reduced speaking rate as a feature of dysarthria in HD.¹ Previous studies have also identified an association between accruing cognitive deficits and slowed speaking rate for patients with multiple sclerosis.²² Although the current study did not systematically assess cognitive symptoms, it is possible that the slowed speaking rate observed in subgroups 2 and 3 is associated with cognitive decline.

Subgroup 4 (normal rate subgroup–mild dysarthria) had minimal perceivable speech deficits according to listeners' ratings. Some patients in subgroup 4 were rated as normal. Therefore, the average ratings of patients in this subgroup may represent speech perceptual characteristics that are indicative of the earliest signs of dysarthria (e.g., monoloudness, monopitch, harsh voice). The harsh voice quality is predominantly associated with laryngeal muscle dysfunction. Pathologic changes in voice quality (dysphonia) have been reported previously for patients with HD and are thought to reflect the progressive loss of control of vocal fold adduction.²³ Interestingly, dysphonia is also commonly mentioned as one of the earliest signs of dysarthria in patients with other degenerative diseases such as Parkinson disease,²⁴ amyotrophic lateral sclerosis,²⁵ and multiple sclerosis.²⁶ It may be that the laryngeal subsystem of speech production may be more sensitive to motor speech impairments at earlier stages in degenerative neurologic diseases. The findings of the speech perceptual characteristics of subgroup 4 are therefore significant because they may indicate very mild dysarthric symptoms or, at the very least, changes in communication performance that may be addressed initially or monitored for progression with speech therapy.

Awareness of the heterogeneity and potential subgroups of speakers with HD is important to guide medical professionals, such as speech–language pathologists, in accurate evaluation or identification of dysarthria secondary to HD. Differing subgroups of speakers would warrant differences in speech–language pathology recommendations for behavioral strategies to compensate for impairments. For example, the subgroups in this current study had differing deficits in speaking rate. A patient with fast speaking rate may benefit from rate control approaches²⁷ such as pacing strategies²⁸ whereas this may not be as appropriate for patients with reduced speaking rate. It is important to treat patients' symptoms as they present rather than providing identical treatment for individuals with a shared diagnosis.

Although this is currently the largest investigation of dysarthric speech in patients with HD, these findings should be replicated in a separate cohort. Also, we limited speech perceptual ratings to an assessment of a reading task. While paragraph reading is considered the gold standard for speech assessments, it is possible that some speech abnormalities may not be detectable during such tasks. For example, irregular articulatory breakdowns may more likely occur during syllable repetition tasks than during paragraph reading.²⁹ Related to our findings of speaking rate differences, there is emerging evidence that cognitive decline may contribute to a slowed speech rate of patients with dysarthria.³⁰ We did not assess whether cognitive performance differed among subgroups. A follow-up study should determine to what extent cognitive status (i.e., performance on UHDRS cognitive tasks) contributes to an abnormal speaking rate (too fast, too slow) and further investigate other potential factors not systematically studied in this retrospective study (i.e., awareness of deficits, type and dosage of medications, UHDRS functional assessments, detailed motor testing, neuroimaging) in a larger sample of patients. Finally, longitudinal studies examining dysarthria over time are needed to confirm the presence of the subgroups and track clinical and pathologic correlates across disease progression.

Glossary

HD

Huntington disease

ICC

intraclass correlation coefficient

LASSO

least absolute shrinkage and selection operator

SIT

Sentence Intelligibility Test

UHDRS

Unified Huntington’s Disease Rating Scale

VMAT-2

vesicular monoamine transporter-2

Appendix. Authors

Study funding

This study was funded by Vanderbilt University Medical Center’s Department of Hearing and Speech Sciences and Department of Neurology.

Disclosure

S. Diehl, A. Mefferd, and Y. Lin report no disclosures relevant to the manuscript. J. Sellers has received research support from Teva Neuroscience and the Vanderbilt Institute for Clinical and Translational Research. She has also received personal fees for speaker bureau participation from Teva Neuroscience. K. McDonell reports no disclosures relevant to the manuscript. M. de Riesthal has received grant support from the National Institutes of Health. D. Claassen has received grant support from the National Institutes of Health (National Institute of Neurological Disorders and Stroke), Michael J. Fox Foundation, as well as AbbVie, Bristol-Myers Squibb, C2N, CHDI, Eli Lilly, Teva, Vaccinex, and Wave Pharmaceuticals. D.O.C. has received personal fees from AbbVie, Acadia, Huntington Study Group, Lundbeck, Neurocrine, and Teva Neuroscience, outside of submitted work. Go to Neurology.org/N for full disclosures.