Neuromodulation of cognition in Parkinson’s disease

Abstract

Neuromodulation is a widely used treatment for motor symptoms of Parkinson’s disease (PD). It can be a highly effective treatment as a result of knowledge of circuit dysfunction associated with motor symptoms in PD. However, the mechanisms underlying cognitive symptoms of PD are less well-known, and the effects of neuromodulation on these symptoms are less consistent. Nonetheless, neuromodulation provides a unique opportunity to modulate motor and cognitive circuits while minimizing off-target side effects. We review the modalities of neuromodulation used in PD and the potential implications for cognitive symptoms. There have been some encouraging findings with both invasive and noninvasive modalities of neuromodulation, and there are promising advances being made in the field of therapeutic neuromodulation. Substantial work is needed to determine which modulation targets are most effective for the different types of cognitive deficits of PD.

1. Introduction

Neuromodulation is the alteration of neural signals in the central, peripheral, or autonomic nervous system through electrical means (Krames et al., 2009). It is currently used for the treatment of several diseases, including movement disorders, epilepsy, psychiatric disorders (primarily depression, anxiety, and obsessive-compulsive disorder), and chronic pain (Krames et al., 2009). Neuromodulation can have advantages over other pharmacological, behavioral, and surgical therapies because it can locally manipulate dysfunctional neural circuits in a highly selective and specific fashion (Shealy et al., 1967). This is distinct from chemical “neuromodulators,” such as hormones, which can dramatically change the state of neuronal networks by complex molecular cascades. Neuromodulation has great potential because it can be performed through invasive or noninvasive techniques, it does not destroy neural tissue, and it is often reversible. It can work by potently stimulating, inhibiting, modifying, or otherwise regulating activity within neural systems. Neuromodulation is also adjustable, meaning it can be fine-tuned according to the patient’s needs.

Here, we focus primarily on the use of neuromodulation in Parkinson’s disease (PD). Neuromodulation is particularly useful when conditions are difficult to treat with medication or when patients do not respond well to medication. The latter commonly occurs when patients take levodopa, the primary medication prescribed for PD. Neuromodulation has been enormously effective for the motor symptoms of PD (Perlmutter and Mink, 2006) for two primary reasons: (1) motor symptoms of PD are easy to measure and immediately apparent, and (2) the functional neuroanatomy of movement in the basal ganglia is relatively well-understood (Albin et al., 1989; Lozano et al., 2010). The ability to see and measure motor symptoms in PD makes it easy to rapidly assess the effect of levodopa or neuromodulation, and this relationship enabled detailed investigation into the neurophysiology of movement (Marsden, 1989). With the advent of the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) primate model of PD, neurophysiologists were able to map the neuroanatomy of movement in primate models (Bergman et al., 1994). MPTP, a neurotoxin used to reproduce many of the clinical and pathological features of PD in monkey models, was instrumental in implicating subcortical basal ganglia structures, such as the subthalamic nucleus (STN) and the globus pallidus pars internus (GPi), in Parkinsonism (Bergman et al., 1990, 1994; Chiueh et al., 1984).

In PD, the most prominent form of neuromodulation is deep brain stimulation (DBS) (Shipton, 2012). Several observations advanced this therapy. Accidental occlusion of the GPi during an unrelated surgery in the 1950s unexpectedly improved a patient’s tremor. Similarly, thalamotomy and pallidotomy were discovered to be effective in alleviating motor symptoms of PD, primarily tremor, rigidity, bradykinesia, as well as dyskinesias (Laitinen et al., 1992). These data advanced lesioning of targets in the basal ganglia and thalamus for treating symptoms of movement disorders (Benabid, 2003; Cooper, 1956). Then, during a typical lesioning surgery it was discovered that high-frequency electrical stimulation, which was used to control electrode placement, also effectively reduced tremor (Benabid, 2003; Benabid et al., 1987; Limousin et al., 1998). High-frequency stimulation of these regions, now known as DBS, was thus introduced as a safe, effective alternative to lesioning. Reversibility and adjustability make DBS significantly more beneficial to patients than the original lesioning method. While there continues to be debate about the mechanism of action of DBS, it is clear that high-frequency electrical oscillations disrupt target pathological activity in brain structures, leading to improvements in bradykinesia and other motor features of PD (Baron et al., 2002; Chiken and Nambu, 2016; Herrington et al., 2016). Large randomized trials documented a consistent and sustained benefit of DBS in these structures in PD (Deuschl et al., 2006; Follett et al., 2010; Krack et al., 2019).

In stark contrast, the ability to treat cognitive impairments in PD—or in any disease—is far less clear. Cognition itself is a vast concept encompassing working memory, language, attention, reasoning, decision making, and many other high-level operations (see chapters “Cognitive control in Parkinson’s disease” by Cavanagh et al., as well as “Neuropsychology of Parkinson’s disease” by Toovey and Anderson, in this volume) (Allen, 2017). Within PD, cognitive deficits early in the disease are relatively well circumscribed to executive dysfunction involving working memory, timing, processing speed, and attention (Aarsland et al., 2009; Aldridge et al., 2018; Kehagia et al., 2013; Parker et al., 2013). Cognitive symptoms can range from mild cognitive impairment to dementia with progressive deficits (Goldman and Sieg, 2020; Zhang et al., 2020). Such symptoms can be debilitating and are a predictor of loss of independence (i.e., nursing home placement) and death.

The effectiveness of neuromodulation for cognitive impairments in PD is limited by the need to develop suitable assays of cognitive function. None of the cognitive domains that are affected in PD can be measured as rapidly as bradykinesia, and some require complex assays that must be normalized across education and life experience to be interpreted. A further challenge is that as the disease progresses, cognitive function might also change and require different assessment instruments. Despite these challenges, further research into these assays and how performance on these assays evolves over the disease may help elucidate functional neuroanatomy of cognitive dysfunction in PD and the effective use of neuromodulation. Our group has used interval timing for this purpose, which requires participants to estimate an interval of several seconds with a motor response. Interval timing recruits working memory for temporal rules, attention to the passage of time, and decision making about when to respond (Kelley et al., 2018; Parker et al., 2013; Singh et al., 2021). There may be many other cognitive paradigms that could be used to assay cognitive functions in PD. To be useful, they would need to be simple, reliable, suitable for cognitively impaired PD patients, and compatible in intraoperative settings. In addition, these cognitive paradigms would need to quickly reflect changes in a patient’s cognitive function in response to neuromodulation or medication. A suite of robust and easily administered assays might help further characterize the spectrum of cognitive dysfunction in PD, both early in the disease and as it evolves. This suite may prove useful for developing neuromodulation targeted at cognitive symptoms in neurodegenerative disease.

2. Types and targets of neuromodulation used for PD

2.1. Invasive neuromodulation

There are several different methods of neuromodulation, including invasive and noninvasive techniques (Fig. 1). Here we describe some of these methods and the evidence (or lack thereof) of their effectiveness toward treating cognitive symptoms in PD.

FIG. 1.

Types of neuromodulation that might be useful for cognition in Parkinson’s disease.

2.1.1. Deep brain stimulation

DBS remains the most frequently used method for neuromodulation in PD today. As of 2018, over 100,000 PD patients worldwide have been treated with DBS (Brittain and Cagnan, 2018). DBS involves the surgical implantation of electrodes specifically targeting areas of the basal ganglia such as the GPi and the STN (McDonald, 2016). Motor symptoms of PD result, in part, from dysfunction in recursive neural networks within the basal ganglia. One beneficial implication of this is that multiple areas of the basal ganglia can be targeted with DBS. Briefly, high-frequency electrical impulses (typically 60–185Hz) delivered directly to these subcortical regions disrupt pathological neural activity across the motor circuit, suppressing elevated beta (13–30Hz) activity and alleviating certain symptoms of PD, most often rigidity, bradykinesia, and tremor.

Because nuclei within the basal ganglia integrate information from motor and cognitive networks, DBS influences neural activity throughout the network by altering oscillations in the target region (David et al., 2020). One mechanistic theory is that the result of high-frequency stimulation is an ‘informational lesion’ in the indirect pathway that is traditionally associated with inhibiting movements via striato-pallido-subthalamic connectivity. Inhibiting the indirect pathway thus results in “faster” movement, counteracting the bradykinesia of PD (Benabid, 2003; Benabid et al., 2005; Benazzouz and Hallett, 2000). There are several other additional mechanistic principles that may be at work over varying time courses (for review see the following: Ashkan et al., 2017; McIntyre et al., 2004; McIntyre and Anderson, 2016); evaluating the evidence for and further refining these principles will prove key to extending DBS to other targets and indications.

Although highly effective for motor symptoms, DBS has notable risks and limitations. DBS can yield side effects such as dysarthria, anxiety, impulsivity, prickling sensations in the hands and feet, and in some cases, weight gain (Brittain and Cagnan, 2018). Other associated risks center around the surgical implantation of DBS, such as bleeding in the brain, infection at the surgical site, breathing and heart problems, and even stroke (McDonald, 2016). DBS may also not be effective in all patients, particularly those who show no response to dopaminergic medication (Brittain and Cagnan, 2018; Pollak et al., 2002). Further, placebo effects are known to be large in clinical trials involving PD, particularly in studies of invasive treatment techniques (Albin, 2002). Related to DBS, the limited use of sham surgery controls may lead to bias in clinical trials and should be considered when evaluating the results.

Although DBS has been shown to improve health-related quality of life (Ferrara et al., 2010), there is evidence that DBS can negatively affect non-motor symptoms of PD (see Brittain and Cagnan, 2018 for review). In some situations, clinical stimulation parameters have been shown to acutely worsen cognition, though Mehanna et al. (2017) found in their critical review that the negative impacts of DBS on cognition were rare and subtle, although individuals with more severe cognitive impairments were excluded. On a longer, longitudinal timescale, You et al. (2020) found mixed effects of 1 year of high-frequency STN DBS on cognitive performance. They found that verbal fluency and performance on an implicit learning task decreased in individuals with DBS, suggesting a negative long-term impact of DBS. Interestingly, visuospatial ability seemed to improve after 1 year with DBS, and there was no effect on memory. These findings suggest that long-term effects of DBS are domain-specific. Barboza E Barbosa and Fichman (2019) reviewed the current data on cognitive functioning in patients with DBS and found conflicting results across 27 articles. They reviewed global cognitive functioning, memory, executive function, perception, attention, language, and visuospatial skills. Some articles revealed declines in cognition, particularly in executive functioning, which included verbal fluency, working memory, planning, and cognitive flexibility. Most studies, however, revealed no effect of DBS on other aspects of cognition. Notably, while Williams et al. (2011) found impairments in nonverbal memory, oral information processing speed, and language after 2 years with DBS, they also found that the incidence of progression to dementia after DBS was no different than in medically-controlled PD patients after the same time period. Together, these studies reveal mixed findings for both acute and long-term effects of high-frequency STN DBS in PD patients, with the most reliable declines being in verbal fluency and executive functioning. These differences may be more prominent for STN DBS compared to the GPi (Okun et al., 2009). However, these deficits may be transient (Heo et al., 2008).

In controlled laboratory settings, deficits are consistently documented. Frank et al. (2007) showed that high-frequency STN DBS affected the normal reaction to slow down when making a decision under high-conflict conditions. Cavanagh et al. (2011) found that high-frequency stimulation of the STN was associated with increased impulsivity and poorer performance, although reaction times were faster. These data suggest that the STN is responsible for slowing down responses under high-conflict situations, and that high-frequency stimulation of the STN results in more impulsive decisions, speeding reaction time. These data lead to a view that in addition to motor systems, the STN may play a role in dynamically adjusting control over response inhibition (Chen et al., 2020; Jahanshahi et al., 2015; Narayanan et al., 2020). Cognitive side effects of STN DBS are of great interest because the STN receives massively convergent input from a wide array of cognitive cortical areas, including “hyperdirect” input, or monosynaptic projections from the cortex directly to the STN, bypassing other basal ganglia pathways. Thus, there is the exciting possibility that if cognitive functions can be disrupted by DBS, then with different stimulation parameters, they might be enhanced.

While high-frequency stimulation targets motor symptoms and either doesn’t affect or worsens cognition, low-frequency stimulation (<100Hz) has been used effectively to improve some non-motor symptoms, such as gait (60Hz: Brozova et al., 2009; 60Hz: Moreau et al., 2008; 60Hz: Xie et al., 2015), swallowing (60Hz: Xie et al., 2015), and speech (for review see Baizabal-Carvallo and Alonso-Juarez, 2016; David et al., 2020; Sidiropoulos et al., 2013). Specifically related to cognition, our group found in a small study that 4-Hz stimulation of the STN normalized interval timing performance (Kelley et al., 2018). In this study, PD patients displayed abnormal interval timing performance with DBS off and DBS on at the clinically-programmed high-frequency stimulation setting. On the other hand, low-frequency stimulation during the task resulted in performance that was more similar to that of normal, healthy older adults. Further, midfrontal delta (1–4Hz) power was increased during 4-Hz STN stimulation. Similarly, Scangos et al. (2018) found that low-frequency stimulation (5Hz), compared to DBS off or on at 130-Hz stimulation, improved performance on a Stroop task, which assays ability to overcome cognitive interference. These findings suggest that delta/theta-frequency stimulation can improve STN and cortex communication in a way that supports cognition, while high-frequency stimulation seems to disrupt this communication. Of note, multiple studies have shown no effect of low-frequency stimulation on cognitive function (for review see Baizabal-Carvallo and Alonso-Juarez, 2016; David et al., 2020). Advances in DBS for motor symptoms in PD benefitted from decades of sustained research in animal models, in parallel with human research on post-stroke and surgical lesions, prior to well-controlled randomized large-scale trials (Deuschl et al., 2006; Follett et al., 2010). The reported conflicting results indicate that studies of DBS for cognition are at an early stage and mechanistic work needs to be done. One challenge for this effort is the lack of strong animal models that have relevance for complex cognitive functions in human PD.

2.1.2. Targets of DBS used in PD

The primary targets of DBS for treating the motor symptoms of PD are STN and GPi (DeLong and Wichmann, 2015). Other targets may be equally or more effective for targeting cognitive symptoms (Fig. 2). Stimulating the pedunculopontine nucleus (PPN) has shown benefits in working memory, delayed recall, and executive function (Alessandro et al., 2010; for review see Baizabal-Carvallo and Alonso-Juarez, 2016; Costa et al., 2010). Costa et al. (2010) found that response times on a working memory task were faster during 25-Hz stimulation of the PPN compared to no stimulation. These findings suggest that stimulating the PPN may improve the speed of processing. However, the PPN has been shown to be associated with alertness and attention; thus, the beneficial effects observed could be due to increased attention. It should also be noted that this study employed a very small sample size of five PD patients, limiting its generalizability.

FIG. 2.

Potential targets for neuromodulation that might improve cognition in Parkinson’s disease.

Another target of DBS that has been evaluated for its effects on cognition is the nucleus basalis of Meynert (NBM), which is located inferior to the GPi and has been implicated in cognitive and behavioral functions. The NBM has also been shown to deteriorate significantly in PD dementia (PDD; see chapter “Cholinergic systems, attentional-motor integration, and cognitive control in Parkinson’s disease” by Albin et al., in this volume) (Whitehouse et al., 1983). Two case studies examined the effects of NBM DBS on higher order motor control and cognitive functions (Barnikol et al., 2010; Freund et al., 2011). Barnikol et al. (2010) showed evidence in one patient that 20-Hz NBM stimulation, in addition to 130-Hz STN stimulation, resulted in improvement in apraxia (inability to execute a learned movement), suggesting NBM was involved in higher order motor control. More relevant to cognition, Freund et al. (2011) found improvement in memory, visuospatial organization, and executive function of a 71-year-old man during simultaneous stimulation of the STN (130Hz) and NBM (20Hz). In a slightly larger study, Gratwicke et al. (2018) found that although NBM DBS was tolerated by all six patients studied, there were no significant changes in the main cognitive outcome measures after 6 weeks of 20-Hz NBM stimulation compared to 6 weeks with no stimulation. There was, however, an improvement in visual hallucinations in the active stimulation condition, compared to when stimulation was off.

Aside from these few studies, there have been few other studies on the effects of NBM DBS in PD. However, in the context of Alzheimer’s Disease (AD), Kuhn et al. (2015) found that NBM DBS may slightly improve or stabilize cognitive symptoms. Despite these findings, NBM DBS is not currently a widely used treatment for cognitive symptoms in PD, as stronger evidence would be needed to warrant regular usage.

Much of the other work on the effects of DBS on cognition has been done in patient populations other than PD (for review see Bick and Eskandar, 2016). There are some studies of DBS targets within the Papez circuit, which is a neural circuit connecting the hippocampus and fornix with the thalamus, most well-known for its involvement in emotional regulation and memory. While one clinical trial of patients with Alzheimer’s disease did find that fornix stimulation corrected pathological alterations in cerebral glucose metabolism, the evidence was mixed in support of improvements in cognition (Laxton et al., 2010), and this result hasn’t been subsequently replicated. Other studies of Papez circuit stimulation have primarily focused on improvements in declarative memory, which is not a primary symptom of PD. Thus, this particular brain circuit has not been tested as a DBS target in PD. Miller et al. (2015) found that, in four patients with epilepsy, burst stimulation (stimulating at 200Hz with a bursting pattern) of the fornix resulted in robust improvement in visual memory. On the other hand, Lozano et al. (2016) used different cognitive outcomes, and they found no overall beneficial effects of fornix stimulation on cognitive function, as measured by the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (Kueper et al., 2018), as well as verbal learning.

2.2. Noninvasive neuromodulation

Neuromodulation for PD can also be accomplished using noninvasive techniques, such as transcranial stimulation. Unlike DBS, which is invasive and delivers a focal form of stimulation within neural tissue, transcranial stimulation targets cortical regions through external stimulation of the brain. Given lower risk of infections and fewer extended hospital stays, noninvasive approaches may be preferrable. However, transcranial stimulation does not result in the same benefits for motor symptoms as those provided by invasive approaches, such as DBS (Brittain and Cagnan, 2018). It remains to be seen whether noninvasive methods will be effective enough to treat cognitive symptoms and be used clinically.

2.2.1. Transcranial magnetic stimulation and its targets used in PD

One type of transcranial stimulation that has shown promise is transcranial magnetic stimulation (TMS), which establishes a magnetic field through circular coils placed near cortical regions of interest. This magnetic field induces an electrical field, depolarizing neurons and ultimately moderating cortical excitability (Klomjai et al., 2015). While TMS is currently being studied, it has the potential to normalize pathologically-decreased output from the basal ganglia through high-frequency cortically excitatory stimulation. Indeed, repetitive TMS (rTMS) over the primary motor cortex has been shown to effectively improve cardinal symptoms of PD (Elahi et al., 2009).

A main target for TMS that has been evaluated in the context of cognition in PD is the dorsolateral prefrontal cortex (DLPFC). This structure plays a key role in a broad range of cognitive functions, exerting top-down control over brain networks serving high-level cognitive function (Fuster, 2015). Boggio et al. (2005) showed that 2 weeks of ~15-Hz rTMS targeting the DLPFC improved performance on a Stroop task relative to baseline; of note, similar improvements were seen in a group that received fluoxetine for 2 weeks. This study was consistent with a prior report that TMS improved clinical cognitive metrics, as measured by the mini-mental state exam (MMSE) and activities of daily living (ADL); in this study, TMS was more effective than fluoxetine (Fregni et al., 2004). Neither study included a true control arm, raising the possibility that there may be placebo effects, which can be particularly challenging to account for with this method, since the TMS apparatus makes an audible sound and can induce scalp sensations (Duecker and Sack, 2015). A systematic review (Dinkelbach et al., 2017) suggests that across several studies, in hundreds of PD patients, TMS and transcranial direct current stimulation (tDCS) targeting DLPFC improved performance of several executive functions, as assessed by the Trail Making Test Part A and B, the Wais Adult Intelligence Scale, and the Wisconsin Card Sorting Test. Anatomical analyses were of great interest, as improvements were most prominent when stimulation targeted the right and left DLPFC, with little effect when other frontal areas such as supplementary or primary motor areas were targeted. Interestingly, there were also larger effect sizes for measures of mood such as depression, and it was unclear whether depression served as a mediating variable. Given that DLPFC TMS can powerfully improve depression (George et al., 2000) and that depression can be a symptom of PD, determining this relationship is key for discerning the scope and efficacy of neuromodulation in PD patients. Careful, well-done sham stimulation studies are critical to future efforts for all neuromodulation studies.

2.2.2. Transcranial direct current stimulation and its targets used in PD

Additional transcranial methods have been studied in PD due to their noninvasive nature. tDCS applies either a positive (cathodal) or negative (anodal) direct current (measured in milliamps (mA)) to cortical areas through electrodes placed on the head. Differential effects can be seen based on the type of stimulation, the strength of stimulation, and the area being stimulated. Through external stimulation, tDCS modulates GABAergic and glutamatergic concentrations by depolarizing or hyperpolarizing brain regions based on cathodal or anodal stimulation. Anodal application of tDCS hyperpolarizes brain regions and stimulation targeting the motor cortex has been effective for the attenuation of motor symptoms in PD (Benninger et al., 2010; for review see Lefaucheur et al., 2017).

Regarding the effect of tDCS on cognition, studies have shown mixed findings (for review see Biundo et al., 2017). For instance, Boggio et al. (2006) tested whether working memory could be impacted by anodal tDCS over the left DLPFC. In their sample of 18 individuals with PD, they found that working memory accuracy improved after 15min of tDCS, though this was only true for 2mA stimulation, not 1mA stimulation. Notably, tDCS over the left DLPFC was more effective for working memory than stimulation over primary motor cortex. Similarly, Doruk et al. (2014) found that 10 consecutive days of tDCS over either the left or right DLPFC resulted in improved executive function performance, as assessed by the Trail Making Test Part B, at the 1-month follow-up after the start of treatment.

Targeting a different brain region, Ishikuro et al. (2018) found that 5 days of anodal tDCS treatment over the frontal polar area (FPA; also known as the anterior dorsomedial prefrontal cortex) in nine patients with PD resulted in faster performance on the Trail Making Test Part A. The treatment involved 15min of 1mA stimulation daily. Notably, the study also found improvements in motor function, which had not been observed in studies of tDCS over the DLPFC. Thus, the authors posit that the FPA may be an alternative site of stimulation to treat both motor and cognitive symptoms of PD. This finding may be further supported by evidence from Lau et al. (2019), who found that a single session of tDCS over the left DLPFC resulted in no improvements in visual working memory or inhibitory control in PD patients. It is possible that a single stimulation session is not always sufficient for cognitive improvements, whereas the accumulating effect of multiple days of treatment may be more likely to result in sustained improvements.

Although tDCS has potential as a treatment of PD, inconsistencies in the findings of studies using this modality renders it experimental (Marson et al., 2021). While the overall impact remains to be seen, tDCS may be used in the future to treat PD (Brittain and Cagnan, 2018). Notably, see Biundo et al. (2017) for a review of cognitive improvements observed after repeated anodal stimulation of the left DLPFC, especially when stimulation was combined with cognitive training. In some cases, greater improvement has been seen when stimulation was paired with cognitive training. This evidence could guide future attempts at treating PD.

2.2.3. Transcranial alternating current stimulation and its targets used in PD

Like tDCS, transcranial alternating current stimulation (tACS) amends pathophysiological brain states by applying alternating current through scalp electrodes. tACS yields neurochemical changes through alternating currents that can interfere with or enhance “normal” brain waves depending on the frequency of stimulation (Brittain et al., 2013). This method is highly effective at altering brain waves and oscillatory activity associated with motor symptoms in PD. Brittain et al. (2013) have demonstrated that this method can help control and suppress tremors in PD through stimulation of the motor cortex. Because oscillations may play a key role in cognition as well as motor fluctuations, tACS may be uniquely powerful for cognitive functions of PD (Kim et al., 2021). While this method has not yet been studied for its effects on cognition in PD, there are multiple studies demonstrating beneficial effects of tACS on cognitive functions in healthy populations (Hopfinger et al., 2017: attention; Hoy et al., 2015: working memory; Nomura et al., 2019: episodic memory). These studies delivered current oscillating at a standard frequency of either 40 or 60Hz. However, frequency bands vary across individuals according to age and disease state, which suggests that individualized stimulation parameters may more effectively influence cognitive outcomes. In one relevant study, Zaehle et al. (2010) found that individualized delivery of alpha tACS over the occipital cortex of 10 healthy adults elevated alpha power in parieto-central electrodes: stimulation parameters were individualized by tuning the stimulation frequency to each participant’s baseline alpha frequency peak. Although this study did not directly test the effects of stimulation on cognitive function, this evidence suggests that individualized stimulation parameters may be a more effective approach than applying standard frequency stimulation. As such, individualized tACS should be tested in the future for its effects on power in different frequency bands and associated cognitive functions in PD.

Notably, one recent study (Kim et al., 2021) directly compared the effects of tACS and tDCS, both targeting the DLPFC, on cognitive functions in older individuals with Mild Cognitive Impairment without PD. They found that 30min of tACS improved performance on both the Stroop Task and the Trail Making Test Part B. These improvements were not seen after 30min of sham stimulation or tDCS. Based on these findings, tACS is a strong candidate for modulating cognitive function in PD.

Transcranial modalities of stimulation such as TMS, tDCS, and tACS allow multifocal targeting, which could enable multifaceted treatments of target regions associated with both cognitive and behavioral impairments in PD. Unfortunately, transcranial methods are often associated with discomfort and pain from stimulation, though these effects are often counteracted using topical anesthetics (Brittain and Cagnan, 2018). It is also important to note that the results of transcranial stimulation are not always consistent. The variety of targeting methods used across studies may be one reason for such inconsistent results. Recent research has been aimed at finding better targeting strategies at the individual subject level (Oathes et al., 2021). These strategies could reveal more specific brain regions that could be targeted to improve cognitive deficits, and similar approaches could be used to treat PD patients at the individual subject level. As such, further research is needed in order to effectively treat the symptoms of PD using noninvasive brain stimulation.

2.2.4. Focused ultrasound and its targets used in PD

DBS, TMS, tACS, and tDCS offer temporary forms of pathophysiological alterations that may yield long-term changes with consistent, repeated stimulation. In cases where permanent lesions may be more effective for treating cardinal symptoms of PD, imaging-guided focused ultrasound (FUS) may be used instead. FUS is noninvasive and uses ultrasonic waves to target tissues in the brain for ablation. The targets for FUS are dependent on patient symptoms, but often include STN, GPi, and ventral intermediate nucleus of the thalamus (Moosa et al., 2019). Although FDA-approved for treatment of PD and mostly used for tremor, FUS is relatively novel and thus is only recommended in select cases where symptoms cannot be improved by other forms of treatment (Magara et al., 2014). Imaging-guided FUS allows a permanent means of modulating the brain regions associated with Parkinson’s symptomatology (Ito et al., 2018). It has less risk of infection than DBS and requires less postoperative recovery than DBS (Moosa et al., 2019). However, DBS allows greater control of the stimulation and frequency needed to counteract motor symptoms of PD such as tremors. A recent systematic review found that there has been limited evaluation of nonmotor neurocognitive outcomes of FUS in PD (Lennon and Hassan, 2021). The few studies that have evaluated nonmotor outcomes have generally found little cognitive impairment following the procedure. Sperling et al. (2018) found improvements in memory and executive functioning, but noted that these may have been due to practice effects. The authors also found declines in performance on the Stroop task, underscoring the complex nature of nonmotor outcomes related to a permanent treatment such as FUS.

It is likely that each stimulation modality needs to be optimized for its target brain region and specific application. Neuromodulation holds tremendous promise for patient-specific and highly personalized therapies in PD.

3. New advances in neuromodulation

As noted above and throughout this book, cognition in PD is enormously complex. Furthermore, unlike motor symptoms of PD, we are unaware of any lesion or stroke literature that describes improved cognitive function in PD patients, implying that gross disruption in a brain area, as occurs in DBS or FUS, may not be effective for the complex cognitive symptoms of PD. Indeed, it may be that augmenting or enhancing brain function is required for this constellation of PD symptomatology. While this seems like a challenging proposition, there are ample preclinical examples of targeted intervention compensating for cognitive dysfunction in PD models (Emmons et al., 2019; Kim et al., 2017; Kim and Narayanan, 2019) as well as other models (Hao et al., 2015; Weible et al., 2017; Xia et al., 2017). Furthermore, several studies, including those reviewed above, noted improvements in cognition in PD with neuromodulation (Doruk et al., 2014; Kelley et al., 2018; Scangos et al., 2018). These studies suggest that neuromodulation has the potential to improve cognition in PD; however, maximizing its effects may require highly precise and tuned approaches that involve sophisticated new methods.

One potentially transformative technology is closed-loop adaptive brain stimulation. This approach uses ongoing behavior or neural activity to reconfigure brain stimulation parameters—ideally in real-time—to deliver maximally effective brain stimulation for a given indication (Grosenick et al., 2015). Closed-loop adaptive DBS, guided by the beta rhythm 13–30Hz, has been marginally effective for motor symptoms of PD (Little et al., 2013; Little et al., 2016). For cognitive function, different frequency or brain region targets may be particularly effective. For instance, adaptive theta tACS improved working memory performance in cognitively normal adults (Reinhart and Nguyen, 2019). In this study, alternating current was adjusted to patient-specific theta rhythms. Future applications may try such individualized and personal approaches in PD with tACS or other approaches.

Another powerful method to predict potential targets is to use connectomics. This approach analyzes the effective connectivity between brain regions using brain imaging techniques, such as resting-state fMRI or diffusion MRI, and correlates this connectivity with a disease state or outcome. These values can also be compared to a normalized connectome. Horn and colleagues were able to use connectivity to predict side effects and outcomes in STN DBS (Horn et al., 2017). A similar big-data approach based on connectomics, with a focus on cognitive symptoms of PD, may be fruitful for identifying the most effective or underappreciated targets for invasive or noninvasive brain stimulation in PD.

4. Conclusion

In this chapter, we have presented the prospect and challenge of treating cognitive symptoms in PD using neuromodulation. We reviewed modalities of neuromodulation, spanning invasive brain stimulation and noninvasive techniques, and we discussed several potential targets. Finally, we discussed future therapies. Cognitive dysfunction in PD likely involves profound disruption of specific circuits, and neuromodulation provides a unique opportunity to modulate these circuits, while minimizing off-target side effects. Determining the specifics of which modulation targets are most effective for which aspects of PD will require decades of sustained inquiry, but is likely to have far-reaching benefits not only for PD patients, but also for many other patient populations. A final challenge is that neuromodulation targets fixed circuitry, and this circuitry can change over time, particularly in the face of neurodegeneration in PD. However, because much is known about circuit dysfunction in PD relative to other brain diseases, and because PD has a long history of neuromodulation, functional neurosurgery, and invasive interventions, this disease is likely to remain at the forefront of neuromodulation research and advances.