Subthalamic nucleus physiology is correlated with deep brain stimulation motor and non-motor outcomes

Abstract

Subthalamic nucleus deep brain stimulation is commonly indicated for symptomatic relief of idiopathic Parkinson's disease. Despite the known improvement in motor scores, affective, cognitive, voice and speech functions might deteriorate following this procedure. Recent studies have correlated motor outcomes with intraoperative microelectrode recordings. However, there are no microelectrode recording-based tools with predictive values relating to long-term outcomes of integrative motor and non-motor symptoms. We conducted a retrospective analysis of the outcomes of patients with idiopathic Parkinson's disease who had subthalamic nucleus deep brain stimulation at Tel Aviv Sourasky Medical Centre (Tel Aviv, Israel) during 2015-2016. Forty-eight patients (19 women, 29 men; mean age, 58 ± 8 years) who were implanted with a subthalamic nucleus deep brain stimulation device underwent pre- and postsurgical assessments of motor, neuropsychological, voice and speech symptoms. Significant improvements in all motor symptoms (except axial signs) and levodopa equivalent daily dose were noted in all patients. Mild improvements were observed in more posterior-related neuropsychological functions (verbal memory, visual memory and organization) while mild deterioration was observed in frontal functions (personality changes, executive functioning and verbal fluency). The concomitant decline in speech intelligibility was mild and only partial, probably in accordance with the neuropsychological verbal fluency results. Acoustic characteristics were the least affected and remained within normal values. Dimensionality reduction of motor, neuropsychological and voice scores rendered six principal components that reflect the main clinical aspects: the tremor-dominant versus the rigidity-bradykinesia-dominant motor symptoms, frontal versus posterior neuropsychological deficits and acoustic characteristics versus speech intelligibility abnormalities. Microelectrode recordings of subthalamic nucleus spiking activity were analysed off-line and correlated with the original scores and with the principal component results. Based on 198 microelectrode recording trajectories, we suggest an intraoperative subthalamic nucleus deep brain stimulation score, which is a simple sum of three microelectrode recording properties: normalized neuronal activity, the subthalamic nucleus width and the relative proportion of the subthalamic nucleus dorsolateral oscillatory region. A threshold subthalamic nucleus deep brain stimulation score >2.5 (preferentially composed of normalized root mean square >1.5, subthalamic nucleus width >3 mm and a dorsolateral oscillatory region/subthalamic nucleus width ratio >1/3) predicts better motor and non-motor long-term outcomes. The algorithm presented here optimizes intraoperative decision-making of deep brain stimulation contact localization based on microelectrode recording with the aim of improving long-term (>1 year) motor, neuropsychological and voice symptoms.

Keywords: Parkinson’s disease; deep brain stimulation; microelectrode recording; subthalamic nucleus; symptoms.

Graphical Abstract

Figure 1

Change in motor symptoms, ADL and medication postsurgery. (A) All motor symptoms significantly improved (limb bradykinesia P = 0.0001; rigidity P = 0.006; tremor (P = 0.02), except axial signs, which persistently declined (P = 0.03). (B) UPDRS-total, UPDRS-III and UPDRS-IV significantly improved (P < 0.05, P = 0.0016 and P = 0.001, respectively) in contrast with UPDRS-I and UPDRS-II. (C) Both S&E score for ADL and LEDD significantly improved (P = 0.016 and P = 0.046, respectively). (D) A net increased usage of both anti-depressants and anti-psychotics was noted following STN-DBS. *P < 0.05; **P < 0.0001. Exact P-values are noted in the figure legend. DBS, deep brain stimulation; LEDD, levodopa equivalent daily dose; STN, subthalamic nucleus; UPDRS, unified Parkinson’s disease rating scale.

Figure 2

Change in neuropsychological outcomes postsurgery. (A) All frontal functions of FLOPS declined but were not statistically significantly. (B) Attention, attention switch and cognitive flexibility declined (WCST P = 0.0008; TMT-A P = 0.0072). (C) Among the verbal scores, the most significant decline was in the phonemic (frontal function—FAS) verbal fluency score (P < 0.0001) and less in the posterior function (ANIMALS) and verbal memory (RAVLT). In contrast, mixed functions such as working memory and posterior functions that relate to verbal memory (RAVLT Scores) (C) and visual memory (ROCF scores) (D) have improved to different degrees. *P < 0.05; **P < 0.0001. Exact P-values are noted in the figure legend. FAS, phonemic verbal fluency; FLOPS, Frontal Lobe Personality Scale; Digit BWD, Digit Span Backward; Digit FWD, Digit Span Forward; Digit SS, combined Forward/Backward Digit Span Score; LTM, long-term memory; RAVLT, Rey Auditory Verbal Learning Test; ROCF, Rey–Osterrieth Complex Figure Test; STM, short-term memory; TMT-A, Trail Making Test Part A; TMT-B, Trail Making Test Part B; WCST, Wisconsin Card Sorting Test.

Figure 3

Speech intelligibility outcomes postsurgery. All speech and voice deteriorations were mild. (A, B) Among the speech intelligibility scores, VAS-5, VAS-6 and six-point scale significantly decreased (P < 0.001, P = 0.01 and P = 0.035, respectively). (C, D) Among the acoustic characteristics, only (D) GRBAS scores were most sensitive and significantly increased (GRBAS breathiness P = 0.005; asthenia P = 0.04; strain P = 0.06; total P = 0.03; loudness variation P = 0.03). Univariate analysis (MEANS procedure) was applied for continuous variables and chi-squared test (FREQ procedure) for categorical variables (n = 48 for all analyses). *P < 0.05. Exact P-values are noted in the figure legend. GRBAS, grade, roughness, breathiness, asthenia and strain; MPT, mean phonation time; NHR, noise to harmonic ratio, VAS, Visual Analogue Scale; VHI, Voice Handicap Index.

Figure 4

The occurrence of β-oscillations across the cohort. (A) NRMS of neuronal firing rate by estimated distance from the centre of the STN (0) based on presurgical MRI planification. MER exploration began 20 mm dorsal to this point. (B) A spectral analysis of the neuronal activity at 0–250 Hz by estimated distance from the centre of the STN. In each of the 198 trajectories, five MER properties of the STN were tracked: (i) NRMS of the neuronal firing rate; (ii) STN width (defined by the increased NRMS); (iii) DLOR width; (iv) DLOR/STN ratio; and (v) β-oscillation occurrence across the cohort. x-axes in both upper (A) and lower (B) figures describe the distance from the estimated centre of the STN (0) based on presurgical MRI planification. MER exploration began 20 mm dorsal to this point. (B) A spectral analysis of the neuronal activity was done at 0–250 Hz.

Figure 5

MER properties of the STN. (A) Mean NRMS-peak by STN side (overall mean was 2.44 ± 1.5). (B) Mean STN and DLOR widths by STN side. (C) β-Oscillations within the DLOR. In 71.3% of trajectories, a β-oscillatory increment was found in the dorsolateral STN, and 58% had a concomitant decreased γ-oscillatory activity. (D) Localization of the active contact in relation to the DLOR. Eighty-seven per cent of clinically active contacts were within the DLOR. DLOR, dorso-lateral oscillatory region; MER, microelectrode recordings; NRMS, normalized root mean square; STN, subthalamic nucleus.

Figure 6

MERs have distinctive predictive values on motor and neuropsychological scores compared to voice/speech. All the symptoms (axial, acoustic characteristics, speech intelligibility, tremor, rigidity and limb bradykinesia, frontal and posterior neuropsychological symptoms) are aligned along the x-axes. R² on the y-axes is presented regarding each of the MER properties (NRMS, STN width and β-oscillations). Univariate analyses (N = 48) show that NRMS and STN width have higher correlations compared with β-oscillations. However, these correlations differ regarding the different symptoms. Both NRMS and STN width highly (and similarly) correlate with motor (except axial signs) and neuropsychological scores compared with acoustics and speech. β-Oscillations correlate only with motor (except axial signs) and neuropsychological functions. Axial symptoms are the only ones not to significantly correlate with any of the MER properties. MER, microelectrode recordings; NRMS, normalized root mean square; STN, subthalamic nucleus.

Figure 7

A PCA-based simplified and thorough clinical scoring protocol. (A) The actual clinical protocol (of >100 scores) takes more than 4.5 h and therefore requires two sessions to avoid ON/OFF transitions, which might distort results. (B) PCA of these motor, neuropsychological and acoustic/speech scores rendered six PCs, i.e. PC 1–PC 6. (C) Cronbach’s α (higher than 0.7, except for factor 6) renders a reliable representation of each of these six PCs, which simplifies the protocol into a single session of not >1.5 h. (D) The explained variance is presented for each of the symptoms’ sub-groups (motor, neuropsychological and acoustic/speech) with error bars representing the differences between pre- and post-DBS calculations. (E) Cronbach’s α values of the selected tests are presented, showing similar values before and after STN-DBS (N = 117). DBS, deep brain stimulation; FAS, phonemic verbal fluency; FLOPS, Frontal Lobe Personality Scale; GRBAS, grade, roughness, breathiness, asthenia and strain; H&Y, Hoehn and Yahr; LTM, long-term memory; PC, principal component; RAVLT, Rey Auditory Verbal Learning Test; ROCF, Rey–Osterrieth Complex Figure Test; S&E, Schwab and England; STM, short-term memory; UPDRS, unified Parkinson’s disease rating scale; VHI, Voice Handicap Index.

Figure 8

Predictive values of STN-DBS-MER score on outcomes. (A) Each sub-score (NRMS, STN width and β-oscillation/STN) is graded between 1 and 3. The STN-DBS score is a simple linear summation of each of the three sub-scores (range 3–9). The mean STN-DBS score was 5.5 ± 2.8. (B–D) Each of the three MER sub-scores shows a significant correlation with all symptom sub-groups, except axial signs, and a non-significant correlation with acoustic characteristics. (E) The STN-DBS score, which integrates these MER sub-scores, maintains these patterns. Higher STN-DBS scores indicate better motor outcomes (except axial signs) and fewer neuropsychological and speech adverse effects (except acoustic characteristics). (F) Correlation between STN-DBS score and symptom groups. (G) Correlation between STN-DBS score and PCs. STN-DBS score similarly correlates with the symptom groups and the PCs. R² selection methods (R-SQUARE) were used for predicting clinical scores by STN-DBS scores (N = 48 for all). DBS, deep brain stimulation; MER, microelectrode recordings; NRMS, normalized root mean square; PC, principal component; STN, subthalamic nucleus.