One potential factor is the long-term tissue changes due to lead implantation. While the acute microlesion in the STN benefits motor performance, the lead could also impact neural function in the tissue it passes through to impair cognitive performance. On a cellular level, animal and human studies have shown long-term tissue changes due to implantation, including sustained gliosis around the implant (19). This long-term gliosis can alter the neuronal function in the vicinity of the implant, the connectivity within associated circuits, and the volume and metabolic activity of connected regions (19). These tissue changes can impact many different brain areas. At our surgical centers, the lead typically passes through the middle frontal gyrus, frontal white matter, reticular thalamus, internal capsule, thalamic fasciculus and lenticular fasciculus, zona incerta, and STN. Several studies have hypothesized that the electrode trajectory through frontal regions could negatively affect cognitive functions (20, 21). Similarly, studies have shown that the location of the electrode cortical entry point is related to post-surgical verbal fluency decline (22, 23). Changes to frontal activity could impact saccade latency and reach RT directly or through its connections with the parietal cortex, which is associated with attention and sensory-motor transformations (24, 25). Additionally, the lead implantation could intersect and damage white matter tracts that carry motor and oculomotor information (26), although the exact tracts that are damaged are still being explored (27). For example, the internal capsule contains projections leading to the superior colliculus that can affect saccade latency (28) and the longitudinal fasciculus contains projections associated with spatial coordination and attention, which can impact both saccade and reach performance (29). Finally, damage to the zona incerta could also impact saccade latency because it sends projections to the superior colliculus and receives input from frontal and parietal eye fields (30). We have previously shown that turning on STN-DBS decreases saccade latency compared to OFF-DBS in this same task (16). However, STN-DBS does not significantly improve reach RT (16), suggesting that long-term tissue changes could still affect performance even with stimulation.

Another potential factor is the electrophysiological neural change after withdrawal of chronic STN-DBS. For instance, increased alpha band power has been shown after 1-h withdrawal of chronic STN-DBS at 3 years compared to initial programming (31). Alpha power has been related to attention (31) and increased alpha power could contribute to or reflect greater attentional deficits, leading to increased saccade latency and reach RT. There is also some evidence that chronic stimulation could induce changes in brain volume (32, 33) and functional network organization (34), but no conclusions can yet be drawn (35). Network changes due to chronic STN-DBS could explain the difference between our findings and those of Antoniades et al. (15). While Antoniades et al. (15) reported a return to pre-surgery saccade latency values within 3–4 weeks before stimulation was initiated, we show a substantial prolongation of latency at 6–10 months after withdrawal from chronic STN-DBS.

Another potential factor is the reduction in medication post-surgery due to STN-DBS. One hallmark benefit of STN-DBS is that daily medication dosage can be reduced (36). We demonstrated that LEDD was significantly reduced post-surgery, such that medication dosage was reduced, on average, by 69% in our sample. This reduction indicates that our participants had a beneficial and typical response to STN-DBS itself.

If the reduction of medication was an important underlying factor, we would expect that reducing medication would result in a decrease in latency. This is because we have previously reported that medication prolongs saccade latency compared to being OFF medication (16, 37). However, we observe the opposite relationship: a post-surgical reduction in medication accompanied by a prolongation of saccade latency. Thus, it is unlikely that the post-surgical reduction in medication dosage accounts for our findings.

Finally, another potential factor is disease progression. In PD, disease progression is typically characterized by a change in UPDRS Part III score. While disease progression is not linear and changes with disease stage, a 2.4 point increase is expected per year on average early in the disease (38). In our sample, change in MDS-UPDRS Part III score was not a significant covariate and, when included in our linear mixed model, had no effect on the significance of the pre- to post-surgical differences in saccade latency or reach RT. This suggests that disease progression did not significantly affect our results.

The effect of disease progression on visually-guided saccades or reaching has not been studied. However, a preliminary study reported that disease progression over 12 months did not change antisaccade latency in people with PD without STN-DBS (39). Therefore, it is unlikely that the observed increase in latency and RT is solely due to disease progression considering the magnitude of change.

We found that there was no change in OFF treatment MDS-UPDRS Part III or ON treatment MoCA scores post- vs. pre-surgery, demonstrating that our saccade and reach findings cannot be explained by a global deterioration of motor or cognitive performance. Although it should be noted that since MoCA scores were collected ON treatment, it is possible that deterioration of general cognitive performance could be detected OFF treatment and this should be considered in future studies. Previous reports on UPDRS Part III similarly demonstrated that there was no change between pre- and 1–48 months post-surgery (7, 9, 40). In addition, recent studies have shown that there was no change in MoCA score between pre- and 6–12 months post-surgery (41, 42). Thus, our data agrees with previous literature that there are limited changes in average motor severity or general cognitive function in the months post-surgery. This signifies the importance of using sensitive and quantitative behavioral measures, such as saccade latency and reach RT, to elucidate the behavioral changes after surgery.

Previous studies reported that bilateral STN-DBS decreased saccade latency in a saccade only task (43–45) and decreased reach RT in a simple RT task (45–47) compared to OFF stimulation. Similarly, our previous study using the same visually-guided reach task found that bilateral STN-DBS significantly decreased saccade latency, but only non-significantly decreased reach RT compared to OFF stimulation (16). This is important to note because STN-DBS reduced latency and RT toward healthy control levels. Therefore, while our current findings show an OFF treatment post-surgical detriment to saccade latency and reach RT, turning on STN-DBS will likely counteract this detriment. Although it should be considered that long-term tissue changes could still affect performance even with stimulation, as bilateral STN-DBS did not significantly improve reach RT in our previous study (16).

As individuals with STN-DBS are typically being stimulated, the detriment we present here should not impact daily function. However, our results indicate that there are significant behavioral changes due to implantation itself, withdrawal of chronic stimulation, or both, suggesting that long-term neurological changes are occurring. These long-term changes are important to understand in the instance that a patient discontinues STN-DBS temporarily or permanently, or that STN-DBS efficacy declines. Understanding the OFF treatment behavioral changes also highlights which circuits are being affected by implantation and stimulation.

There are several limitations of this preliminary study. First, we had a small sample size with data from only 9 individuals. However, previous studies have used a similar sample size to demonstrate the difference between pre- and post-surgery (10, 15, 48). Second, there is a high variability between participants in task performance. However, the majority of individuals have increased saccade latency (8 out of 9) and reach RT (7 out of 9) post-surgery, so the direction of change was relatively consistent in our sample. Third, there is high variability in our participant demographics and clinical scores, but this is representative of the natural variability of those individuals who opt for STN-DBS. While there was a high variability in individual change in MDS-UPDRS Part III, this variability has been reflected in previous studies who have compared UPDRS Part III pre- and months post-surgery (7, 9, 40). We also confirmed that every participant had a beneficial response to bilateral STN-DBS, as shown by a decrease in MDS-UPDRS Part III score when ON stimulation (Table 1). Each of these limitations emphasizes the need for these findings to be replicated in a larger sample.