Session Time: 12:30pm-2:00pm
Location: Exhibit Hall located in Hall B, Level 2
Objective: To test for an association between LRRK2 mutation and pathological changes in voice.
Background: Voice impairment, characterized by reduced volume, breathiness, roughness and exaggerated vocal tremor, is a common symptom of PD. LRRK2 mutations are associated with increased risk for PD. Although several studies have demonstrated differences in motor and non-motor symptoms between idiopathic PD (iPD) and LRRK2-associated PD, the relationship between LRRK2 mutation and voice impairment has not yet been explored. Subtle changes in voice could be an early motor sign useful in detection of prodromal PD.
Methods: Sustained vowel phonations (‘aaah’) were obtained cross-sectionally from individuals with LRRK2-associated PD (n=8); iPD (n=17); non-manifesting carriers (n=19); related controls (first degree relatives) (n=24) and unrelated controls (n=23). To quantify subtle voice changes we extracted a wide range of multiple dysphonia measures and calculated the sensitivity and specificity to distinguish between groups using two independent methods: 1. Random forests: a statistical machine learning technique commonly used to separate generic data into several classes. 2. Chance predictions: randomized determinations. This method tests the null hypothesis that the discrimination results obtained using random forests are no better than chance predictions of group membership.
Results: The multiple dysphonia measures had a sensitivity of 81.4% (SD 26.5) and specificity of 78.0% (SD 25.9) in discriminating LRRK2-associated PD from iPD using random forests, whereas chance predictions resulted in a sensitivity of 52.02% (SD 45.40) and specificity of 53.32% (SD 45.66). The sensitivity in discriminating non-manifesting carriers from unrelated controls was 70.97% (SD 26.51) and specificity 75.35% (SD 25.67) using random forests and 50.50% (SD 34.51) and 50.58% (SD 34.47) respectively using chance predictions. Finally, the sensitivity and specificity to discriminate between non-manifesting carriers and unrelated controls was 68.46% (SD 26.46) and 72.38% (SD 25.94) respectively using random forests and 49.58% (SD 34.49) and 49.47% (SD 34.39) using chance predictions.
Conclusions: Voice impairment in LRRK2-associated PD may differ to that in iPD. This warrants further investigation. Furthermore, longitudinal studies will determine the potential for voice impairment in non-manifesting carriers to be predictive of transition to manifesting PD.
To cite this abstract in AMA style:S. Arora, N.P. Visanji, T.A. Mestre, T. Ghate, A.E. Lang, M. Little, C. Marras. Investigating voice as a biomarker of LRRK2-associated Parkinson’s disease (PD) [abstract]. Mov Disord. 2016; 31 (suppl 2). http://www.mdsabstracts.org/abstract/investigating-voice-as-a-biomarker-of-lrrk2-associated-parkinsons-disease-pd/. Accessed November 22, 2017.
« Back to 20th International Congress
MDS Abstracts - http://www.mdsabstracts.org/abstract/investigating-voice-as-a-biomarker-of-lrrk2-associated-parkinsons-disease-pd/