Temporal Investigations of the Changes in Presynaptic Inhibition Associated With Subthalamic Nucleus-Deep-Brain Stimulation

The spinal locomotor network may play a role in the symptomatology of Parkinson’s disease (PD).1, 2, 3 Presynaptic inhibition (PSI) is a critical spinal inhibitory mechanism for modulating muscle coordination that works by adjusting both supraspinal motor commands and sensory feedback at the spinal level. It is easy to evaluate this mechanism by investigating the H-reflex,4, 5, 6 which has probably contributed to the role of PSI in the symptomatology and pathophysiology of various neurological diseases including fibromyalgia, polyneuropathy, and stroke being investigated by various researchers.1, 6, 7 There is evidence that PSI is disturbed in patients with PD, and although there are controversies, several studies have supported an association between PD severity and PSI.1, 8 Some authors have also detected alterations in this pathway following medical therapies and physical training programs,1, 5 which also suggest that this pathway is involved in the therapeutic mechanisms.

Recent interest regarding the role of this pathway in PD symptomatology has turned attention to the therapeutic mechanisms underlying subthalamic nucleus deep-brain stimulation (STN-DBS).9, 10, 11, 12, 13 Our previous H-reflex investigations12 did not produce strong evidence for a dynamic change in this pathway associated with STN-DBS stimulation. In light of the inconsistencies with the related literature data,9, 10, 11 we revealed some limitations including the small numbers of patients, lack of clinical correlates, and the short evaluation time after the cessation of stimulation.12 Therefore, in the present study we aimed to comprehensively reveal the role of the PSI mechanism. In a larger group of patients, we investigated the possible alterations in the pathway by repeating electrophysiological investigations over a longer interval after the stimulation is turned off. Moreover, by repeating the clinical evaluations concurrently, we aimed to identify how our findings were associated with clinical correlates.

We enrolled 20 PD subjects who had received STN-DBS therapy and been admitted to the Movement Disorders Polyclinics of Diskapi Yildirim Beyazit Training and Research Hospital for routine polyclinic evaluations between January 2022 and March 2022 (IRB No. 2021_KAEK_189_2021.09.09_05). A signed informed-consent form was obtained from all of the patients. Demographic and clinical data including the disease duration, disease subtype, symptom-onset side, drug usage, and duration of DBS treatments were obtained. DBS parameters including the amplitude, pulse width, and frequency were recorded bilaterally in all of the individuals. Exclusion criteria were the presence of a neurological disease other than PD, musculoskeletal disease, or a major systemic disease (e.g., chronic renal failure, congestive heart failure, or chronic obstructive pulmonary disease) that could affect mobilization. In addition to the bilateral tibial H-reflex investigations, a nerve conduction study was performed in all participants to exclude coexisting polyneuropathies that could affect the H-reflex results; this resulted in the exclusion of two subjects due to evidence of polyneuropathy. The H-reflex examinations were performed during the medication-off (at least 12 hours after the last levodopa dose), stimulation-on period. The investigations were performed bilaterally, and all of them were repeated 10 and 20 minutes after the simulation was turned off. Simultaneous evaluations based on the Movement-Disorder-Society-sponsored revision of the Unified Parkinson’s Disease Rating Scale part III (MDS-UPDRS-III) were performed in all of the patients at baseline in the medication-off, stimulation-on period. The MDS-UPDRS-III evaluations were also repeated 20 minutes after the simulation was turned off (simultaneously with the final H-reflex measurements). The axial symptoms subscores of the MDS-UPDRS-III (items 3.9–3.14) were also recorded in all of the subjects.

Ultimately we enrolled 18 patients aged 58.7±9.3 years (mean±standard deviation), including 10 females. Eight of the patients were diagnosed with the tremor-dominant subtype while the other 10 were diagnosed with the akinetic-rigid subtype. The duration of the disease was 15.3±6.7 years, and the mean duration of DBS treatment was 3.5 years (range=1–13 years). The levodopa equivalent dose was 697.2±305.2 mg. The median MDS-UPDRS-III score during the stimulation-on state was 25 (range=12–64). However, during the stimulation-off state, the MDS-UPDRS-III score deteriorated to 63.1±18.1, revealing the clinical efficacy of DBS in the general group. In 72% of the patients the MDS-UPDRS-III score decreased by more than half during the stimulation-on period in comparison with the stimulation-off period. Fifty percent of the patients showed a decrease of more than 60% in MDS-UPDRS-III score during the stimulation-on period (Table 1).

Table 1
Demographic and clinical data of the patients (n=18)

• Click for larger image
• Click for full table
• Download as Excel file

The comparative analyses of the DBS parameters (amplitude, frequency, and pulse width) between patients with and without a marked clinical response to stimulation did not identify a difference in any parameter (Table 2). The comparative analyses of the repeated measurements made according to the stimulation status revealed significant differences only in the left H/M ratio (p=0.042). In post-hoc tests, the left H/M ratio was higher during the stimulation-off periods than at baseline (Table 3). However, in the subgroup of patients with a decrease in MDS-UPDRS-III score of more than 60% (n=9), repeated analyses showed no difference in the left H/M ratio, while there was a difference in the left H-reflex latency (p=0.012). It was particularly noteworthy that none of these parameters differed between 10 and 20 minutes after the stimulation was turned off (Table 4). The analyses in the subgroup of patients with akinetic-rigid PD revealed alterations in the H/M ratio, whereas no difference was found in any parameters in the tremor-dominant subgroup (Table 5).

Table 2
Results of comparative analyses of DBS parameters between patients with and without marked reductions in MDS-UPDRS-III scores during the stimulation-on period

• Click for larger image
• Click for full table
• Download as Excel file

Table 3
Results of ANOVA tests of the H-reflex parameters according to the stimulation state in the overall group (n=18)

• Click for larger image
• Click for full table
• Download as Excel file

Table 4
Results from subgroup analyses of patients showing ≥60% reductions in MDS-UPDRS-III scores during the stimulation-on period (n=9)

• Click for larger image
• Click for full table
• Download as Excel file

Table 5
Probability values from ANOVA tests of the H-reflex parameters according to the disease subtype

• Click for larger image
• Click for full table
• Download as Excel file

The comparative analyses after categorization according to the symptom-onset side did not reveal any difference in the H-reflex parameters on the most-affected side between the distinct stimulation phases. However, analyses on the contralateral side (less-affected side) revealed differences in both the H-reflex amplitude and H/M ratio. Remarkably, there were significant differences in these parameters at both 10 and 20 minutes relative to baseline (Table 6). Finally, the comparative analyses between patients with and without significant responses in axial symptoms revealed that the left H-reflex latency was significantly longer during the stimulation-on period in patients with marked improvement in axial symptoms (left H-reflex latency differed between baseline and stimulation-off 10-minute evaluations [p=0.033] and between baseline and stimulation-off 20-minute evaluations [p=0.010]).

Table 6
Results of ANOVA tests of the H-reflex parameters on the less-affected side according to the stimulation state

• Click for larger image
• Click for full table
• Download as Excel file

The initial analyses performed in this study did not reveal convincing alterations in the H-reflex parameters associated with the stimulation status. Previous studies9, 10, 11, 13 involving small numbers of patients have produced results that contrast with our findings. Pötter et al.13 found that STN-DBS restored the abnormally reduced autogenic inhibition in 10 PD patients undergoing DBS surgery, and the H-reflex changes during stimulation were significantly correlated with clinical improvements in gait and posture. Raoul et al.11 investigated PSI in nine PD patients receiving chronic bilateral STN-DBS, and found that PSI disappeared when the DBS was turned off. Those authors found similar findings in both the upper (flexor carpi radialis) and lower (soleus) extremities. Parmentier et al.10 also showed significant recovery in the audiospinal reflex in 14 PD subjects receiving STN-DBS during the stimulation-on state. Andrews et al.9 investigated the potential usefulness of the H-reflex as an adjunct marker for determining the optimal electrode placement during surgery, and found that H-reflex recovery normalized in 21 of 26 nuclei during macrostimulation, which correlated with on-table motor improvement. Our results are not consistent with these data reported in the small amount of available literature. However, in our previous study evaluating eight PD subjects receiving STN-DBS, the measurements made 5 minutes after the stimulation was turned off also produced no significant difference.12 In that study we revealed that the interval between the two evaluation statuses (stimulation off and stimulation on) was too short to identify possible stimulation-related changes. We also reported that the duration after the initiation of DBS therapy was longer (median 4.5 years) than in other studies,9, 13 revealing marked alterations in H-reflex responses in association with DBS stimulation. In our current study, the median time after the initiation of the DBS therapy was 3.5 years (range=1–13 years), which was still longer than in the other related studies.9, 13 However, we included a large number of patients (n=18) to increase the power of the analyses. Moreover, to increase the accuracy of detecting stimulation-related alterations in our electrophysiological measurements, we increased the period (at 10 minutes) after the stimulation was turned off and performed a second evaluation to understand the possible alterations according to the temporal course (at 20 minutes).

A critical conclusion from our study was that later analyses according to the symptom-onset side revealed that the parameters associated with PSI deteriorated significantly on the less-affected side when the stimulation was turned off, whereas no difference was found in the parameters on the symptom-onset side. Moreover, the differences in these values in comparisons with their baseline values were sustained both at 10 and 20 minutes after the stimulation was turned off, which supported the association between the stimulation and changes in PSI. No previous study has evaluated PSI specifically according to the affected side (most- and less-affected sides), and so our findings need to be evaluated carefully. A crucial hypothesis regarding the recovery effect of STN-DBS in PSI is that high-frequency stimulation of the STN releases the pedunculopontine nucleus (PPN) from increased GABAergic inhibition via pallidotegmental pathways.13 The PPN has been shown to have reciprocal connections with the spinal cord in both anatomical and experimental studies,19, 20 and it is connected to the cervical and thoracic spinal cord via noncholinergic direct projections.19, 20 Remarkably, the PPN is also known to have outputs on the caudal pontine reticular formation that is implied in PSI.21 A crucial view is that with the progression of the disease, in addition to the involvement of the neocortex, the PD pathology spreads to some specific brainstem nuclei such as the PPN that may be asymmetrical according to the affected side.22 Among these localizations, disturbance in the PPN is in particular thought to be responsible for axial symptoms (via the above-mentioned connections), including severe gait and postural impairments that are not ameliorated by levodopa or STN stimulation.23 Although there are inconsistencies, clinical trials of DBS of the PPN have allowed convincing conclusions to be drawn regarding the impact of this region on gait, postural symptoms, freezing of gait (FOG), and falls.24, 25, 26 Our group consisted of patients with a longer disease duration than in the previous studies showing an association between DBS and PSI.9, 10, 11 The duration of the disease was 15.33±6.71 years in our group, whereas it was 7 years in the study of Raoul et al.,11 9.8 years in another study evaluating the intraoperative measurements,9 and 12.9 years in the research by Parmentier et al.10 that included 14 PD patients receiving STN-DBS. It was particularly noteworthy that the mean MDS-UPDRS-III score was 63.2, representing a markedly advanced stage of the disease. Our initial analyses including the whole group, the subgroup responding markedly to the stimulation (>60% decrease in MDS-UPDRS-III scores), and those on the symptom-onset side did not reveal differences, whereas the analyses of the less-affected side yielded encouraging evidence of the association between the stimulation and PSI.

Taking the above-mentioned findings together, we hypothesize that the involvement of the PPN and the descending tracts from PD pathology on the more-affected side is responsible for the lack of an effect of stimulation on PSI, whereas the alteration in the PSI parameters on the less-affected side might be associated with the intact PPN function. Our hypothesis is supported by the comparisons of the H-reflex parameters between patients showing improvement in axial symptoms with stimulation and those not showing such improvements also revealing alterations in the H-reflex latency in only the stimulation-responsive group. Clarifying these hypotheses in future studies may allow critical conclusions to be drawn. First, the demonstration of PSI recovery in electrophysiological investigations may provide important clues regarding the anatomical and functional disturbances caused by PD pathology in individual patients. Second, the investigation results may allow alternative targets for DBS to be determined. Remarkably, the benefit of stimulating the PPN or spinal cord on PD gait function has already been demonstrated in a small number of patients, both singularly25, 27, 28 and when combined with the administration of STN-DBS therapy.29, 30 Furthermore, some authors have reported the benefit of PPN stimulation as salvage therapy in patients suffering from newly emerging gait disturbance in long-term follow-ups after STN-DBS.31, 32 However, the current evidence is only weak, and so additional studies are required to identify the optimal candidate patients and for these alternative targets for DBS to be approved for the treatment of movement disorders in routine clinical practice.2

In light of these data, our study results may be critical. For example, in those patients whose H-reflex parameters do not respond to STN-DBS, targeting additional areas such as the PPN or spinal cord might be rational. On the other hand, the alterations in the PSI parameters with stimulation suggest normal functioning of the PPN and the descending tracts; therefore, focusing on the optimal adjustment of DBS programming may be logical in those patients, such as switching to low-frequency stimulation or interleaving the stimulation. However, future studies including larger numbers of patients in distinct phases of the disease are certainly warranted to clarify these hypotheses. Utilizing advanced functional neuroimaging methods to reveal the extension of the PD pathology or alternative investigations of rat models designed to identify the possible pathological involvement of those regions of the PPN and descending tracts concurrently might contribute substantially to these discussions.

The main limitation of this study could be the long duration and heterogeneity of the DBS therapy. The mechanisms underlying the efficacy of DBS differ according to the time after therapy initiation. For example, neuromodulation plays a critical role during the early period of the stimulation, whereas plasticity and anatomical organization are involved in the later phases of the therapy.33 Therefore, our results demonstrating nonsignificant alterations of the H-reflex parameters (compared with previous studies9, 13) in association with the stimulation status may be related to the irreversible mechanisms that are particularly involved in the late phase of the DBS therapy. Another limitation may be related to the specific method used to evaluate the H-reflex. In other studies of alterations in this pathway in the whole PD group,9, 13 the authors performed a method of conditioning the H-reflex that was more sensitive than the standard evaluation method used in our study. In addition, we did not conduct the H-reflex investigations in the upper extremities, which might have aided the understand if these alterations are associated with general motor manifestations of PD or may be specific to the locomotor function, sparing the upper extremities. Finally, the investigation of a more-detailed clinical scale focusing on axial symptoms (e.g., FOG scale or the Berg Balance Test) might also contribute crucially to the above discussions.

In conclusion, we have produced evidence of the recovery of PSI on the less-affected side of PD subjects associated with STN-DBS. However, the response rate to the stimulation was not found to influence the changes in PSI. We hypothesize that the involvement of this spinal pathway and its contribution to the mechanisms underlying the effects of DBS differ between individuals based on the severity of the disease and which brainstem regions and descending tracts are involved. Clarifying these hypotheses in future studies may improve our understanding of the mechanisms underlying DBS efficacy and provide critical evidence for use in clinical practice.