Subthalamic nucleus deep brain stimulation connectivity related to apathy outcomes in Parkinson's disease

Abstract

Background

Subthalamic nucleus (STN) deep brain stimulation (DBS) is an effective treatment for Parkinson's disease (PD); however, apathy is a commonly reported side-effect that counteracts quality of life improvements offered by DBS. To date, the structural connectivity of postoperative apathy is largely unknown.

Objective

This study aimed to identify streamlines associated with worsening apathy following STN-DBS in PD.

Methods

Patients with PD who received bilateral STN-DBS (n = 53) were pseudorandomized into “training” (n = 38) and “holdout” (n = 15) groups. In the training group, a normative structural connectivity analysis was conducted using pre- and postoperative apathy outcomes [Frontal Systems Behavior Scale (FrSBe) apathy subscale]. The resulting model was cross-validated and then tested in the holdout group.

Results

An asymmetric profile emerged that was associated with worsening apathy. Right hemisphere streamlines were consistent with the dentatorubrothalamic pathway, connecting to the supplementary motor area. In the left hemisphere, the model included streamlines spanning the anteromedial STN to the anterior cingulate and orbitofrontal areas. The validity of the connectivity model was evaluated using cross-validation paradigms (leave-one-patient-out: R = 0.38, p = 0.023; 5-fold: R = 0.46, p = 0.004; 10-fold: R = 0.33, p = 0.048). The model was used to assess its association with apathy changes in the holdout group (R = 0.71, p = 0.005).

Conclusion

Results suggest that worsening of apathy following STN-DBS in PD involves the stimulation of circuits associated with a range of emotional and motivational processes, potentially disrupting auto-activation and emotional-affective processes in some patients.

Plain Language Title

Understanding Apathy in Parkinson's Disease After Deep Brain Stimulation

Plain Language Summary

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is a well-known treatment for Parkinson's disease, which helps to control movement symptoms. However, a common side effect of this treatment is apathy, which means a loss of motivation and interest in activities. This can reduce the quality of life improvements that DBS provides.

In this study, we aimed to find out which specific brain pathways are linked to worsening apathy after STN-DBS. We included 53 patients with Parkinson's disease who had undergone this treatment. The patients were split into two groups: the “training” group (38 patients) and the “holdout” group (15 patients).

For the training group, we used brain scans taken before and after the DBS procedure to create a 3D model. This model helped us identify the brain pathways associated with increased apathy. We then tested this model on the holdout group to see if it was associated with changes in their apathy levels based on the locations of the DBS electrodes in their brains.

Our findings showed that different pathways in the left and right sides of the brain were involved. On the right side, pathways were consistent with the dentatorubrothalamic tract, which connects to the supplementary motor area. On the left side, pathways connected the anteromedial STN to areas involved in emotional and motivational processes, like the anterior cingulate and orbitofrontal areas.

The model showed a significant association with apathy changes in the holdout group, suggesting that the worsening of apathy after STN-DBS might be due to the stimulation of brain circuits involved in motivation and emotions. While more research is needed to confirm these findings, this model could help doctors better target DBS treatments to avoid worsening apathy in patients with Parkinson's disease.

Introduction

Subthalamic nucleus (STN) deep brain stimulation (DBS) is a well-established adjunctive treatment for Parkinson's disease (PD) that improves quality of life and motor function,^1–3 while also reducing dopaminergic medication dosages and levodopa-induced motor complications such as dyskinesia.^4,5 Despite these desired effects on PD motor symptoms following STN-DBS, negative psychiatric effects, such as impulsivity,^6–8 depression,^8–10 apathy, ¹¹ and anxiety, ¹² can arise following surgery for some patients.

Apathy, a loss of motivation, is commonly reported and the most frequent long-term behavioral side effect following STN-DBS in PD, ⁸ estimated to affect 24–54% of patients.^8,11,13–16 The relationship between changes in apathy following STN-DBS is widely debated with some studies also reporting a decrease in apathy^17,18 while others report no change.^19–22 Apathy is associated with decreased quality of life, potentially counteracting benefits from motor improvement.^23–25 Due to the significant variation in postoperative apathy outcomes and its potential to undermine the benefits of DBS, several studies have investigated why some people have increased postoperative apathy. Factors such as preoperative severity of dyskinesia, ¹³ preoperative apathy score, preoperative episodic memory free recall score, and one-year postoperative motor responsiveness have been associated with postoperative apathy. ²⁵

For motor^26,27 and other non-motor^27–34 features of PD, the location of deep brain stimulation is a key determinant of postoperative changes. Independent of preoperative patient characteristics,^13,25 stimulation location may be a driving factor for postoperative apathy that helps explain the significant variation in postoperative outcomes. Prior studies examining stimulation location in relation to apathy have used a range of spatial and modeling approaches with heterogeneous findings. One study projected stimulation locations along the longitudinal axis of the STN and reported that worsening of apathy was associated with more dorsolateral stimulation, limited to the right hemisphere. ³² Another study, using probabilistic stimulation mapping based on volumes of tissue activated, found that improvement in apathy was associated with more ventral stimulation. ³³ Similarly, an analysis based on Cartesian coordinates of active contact locations demonstrated that more ventral stimulation was associated with postoperative improvement in apathy. ³⁴ These group-level findings are contrasted by a recent case report describing improvement in apathy after repositioning the electrode stimulation more dorsal, ³¹ a hypothesis under current investigation in a subsequent clinical trial. ³⁵ This discrepancy highlights the need for further investigation into the effects of stimulation location on apathy in PD.

Similar to other PD symptoms,^36–38 apathy may be exacerbated in PD patients with stimulation of specific fiber tracts and their associated networks. The variability across optimal stimulation sites reported for apathy after STN-DBS in PD^31–34 may be due to the activation of specific fiber tracts. In other words, distinct electrode locations may stimulate the same tracts while yielding similar outcomes.^39,40

Due to the common prevalence of postoperative apathy and its potential to undermine the quality of life improvements afforded by DBS, it is important to identify stimulation locations associated with increased apathy. This study's objective was to investigate the structural connectivity associated with postoperative changes in apathy in a cohort of PD patients treated with STN-DBS. Gaining a deeper understanding of the mechanisms underlying postoperative symptom changes, coupled with the integration of structural connectivity data into DBS programming, could facilitate more personalized and optimized management of PD symptoms.

Methods

Study cohort

Patients with Parkinson's disease who underwent bilateral STN-DBS treatment at Vanderbilt University Medical Center (VUMC) between February 2013 and April 2019 were included in this retrospective analysis. Most subjects received Medtronic 3389 electrodes (52/53), with one subject receiving Medtronic 3387. These patients were part of a DBS outcomes study approved by the Vanderbilt University Institutional Review Board (IRB#130646) and had both pre-operative and post-operative Frontal Systems Behavior Scale (FrSBe) apathy subscale scores ⁴¹ available (n = 53). All subjects provided written informed consent. Subjects were pseudorandomly assigned to a “training” group (n = 38) or a “holdout” group (n = 15) prior to any connectivity analyses, consistent with prior DBS connectivity studies. ⁴² Assignment was stratified on apathy response category (top, typical, or poor responder based on quartiles of percent change in FrSBe Apathy T-score), sex, and age at surgery to ensure both cohorts contained the full range of clinical outcomes.

Outcome measure

Apathy was assessed using the FrSBe apathy subscale preoperatively and postoperatively. FrSBe assessments were collected as part of routine neuropsychological care at Vanderbilt University Medical Center and were administered by a licensed clinical neuropsychologist with assistance from a trained psychometrist. Postoperative evaluations are routinely scheduled at least 6 months after surgery to allow for recovery and stabilization of DBS programming while limiting confounding from longer-term disease progression. Each of the 14 items is rated from 1 (almost never) to 5 (almost always). Raw scores were converted to standardized norm-referenced T scores adjusted for age and sex, with a mean T score of 50 and a standard deviation of 10. Higher scores indicate worse apathy. Change in apathy was determined by subtracting the preoperative score from the postoperative score, with positive change scores indicating a worsening of apathy.

DBS electrode localization

All participants underwent preoperative structural MRI (T1-weighted and T2-weighted) and postoperative CT imaging as part of standard clinical care. ⁴³ T1-weighted images were acquired with TR/TE ≈ 7.9/3.65 ms and 1.0 mm isotropic spatial resolution. T2-weighted images were acquired with TR/TE = 3000/80 ms, an in-plane resolution of approximately 0.47 × 0.47 mm, and a slice thickness of 2.0 mm. Postoperative CT scans were obtained at 120 kVp with a 512 × 512 acquisition matrix, in-plane resolution of approximately 0.43–0.59 mm, and slice thickness of approximately 0.63–1.5 mm. Electrodes were localized using the advanced processing pipeline in Lead-DBS v3.0. ⁴⁴ Pre-operative T1 and T2 MRIs and post-operative CTs were linearly co-registered using advanced normalization tools (ANTs). ⁴⁵ Co-registrations were inspected and refined if needed, and the Lead-DBS brain shift correction was applied. Pre-operative volumes were normalized to ICBM 2009b NLIN asymmetric (“MNI”) space applying the ANTs SyN Diffeomorphic Mapping ⁴⁶ with the “effective: low variance default + subcortical refinement” preset. PaCER ⁴⁷ was used to automatically pre-reconstruct DBS electrodes and manual refinement was used as needed. Lead-Group ⁴⁸ was used to perform group visualizations and analyses. Atlas segmentations included in this study were derived from the DISTAL Minimal ⁴⁹ and DBS Tractography atlases. ⁵⁰

Electric field modeling

Volume of tissue modulated around electrodes was estimated using electric field vector magnitudes (i.e., E-fields). DBS programming settings at the time of the post-operative assessment were used to calculate E-fields via an adaptation of the SimBio/FieldTrip pipeline ⁵¹ implemented in Lead-DBS.

DBS network mapping

Streamline connectivity was assessed using a connectome modified from a DBS tractography atlas ⁵⁰ that featured additional connections from cortex to STN and from STN to substantia nigra,^26,52 seeding from active contacts. Streamlines were considered connected if the mean E-field magnitude they traversed was >50.0 V/m. Streamlines connected to more than 25% of E-fields were included to identify streamlines with substantial stimulation engagement while excluding spurious connections. Sensitivity analyses at nearby thresholds (20%, 30%) yielded comparable model performance, consistent with prior work. ⁴⁴

Statistical analyses

The connectivity profile for apathy was developed with the training group, including streamlines Pearson-correlated with change in apathy outcome. The model was cross-validated within the training group using Pearson correlation (leave-one-patient-out, 5-fold, and 10-fold) and then used to estimate apathy score variance in the holdout group using Spearman rank correlation, due to limited sample size. Mann-Whitney U tests were used to compare active contact coordinates between hemispheres. Variability in stimulation parameters between the training and holdout groups was assessed using Mann-Whitney U tests for patient-level stimulation amplitude and Fisher's exact tests for the distributions of stimulation frequency and pulse width. Amplitude was defined as the mean of the left- and right-sided settings and was analyzed separately for voltage-based and current-based stimulation.

Subject characteristics of the training and holdout groups were descriptively reported using means ± standard deviation (SD), including ranges for age, disease duration, and change in apathy (Table 1). Between-group differences were assessed using Mann-Whitney U tests for continuous variables and Fisher's exact tests for categorical variables. In line with clinical practice, meaningful worsening of apathy was defined based on a pre- to postoperative change in the FrSBe apathy subscale score greater than 10 (equivalent to one standard deviation in FrSBe's norm-referenced T scores). Clinically significant apathy was defined by a FrSBe apathy subscale score of 65 or higher. ⁴¹ The levodopa equivalent daily dose (LEDD) was calculated as previously described. ⁵³ Relationships between postoperative changes in medication (LEDD) and motor improvement (UPDRS-III) versus changes in apathy were assessed using Spearman's correlation. The LEDD vs. change in apathy correlation included all subjects from both groups (n = 53); however, due to missing postoperative UPDRS-III scores, only 41 subjects were included in UPDRS-III vs. apathy change correlation. Changes in apathy scores across subjects who continued, discontinued, and did not use dopamine agonists were compared using a Kruskal-Wallis test. Cholinesterase inhibitor use was evaluated descriptively. All statistical analyses involving subject characteristics were conducted using Stata 18.0 (StataCorp LP, College Station, TX).

Table 1.

Subject characteristics of training and holdout groups.

Subject characteristics	Training, n = 38	Holdout, n = 15	P-value
Sex, male, n (%)	30 (79%)	10 (67%)	0.480
Age, years
Mean ± SD	63.2 ± 9.4	67.9 ± 4.7	0.122
Range	40.6 to 75.9	60.5 to 77.2
Disease duration at time of surgery, years
Mean ± SD	7.9 ± 5.0	8.5 ± 3.9	0.304
Range	0.65 to 21.1	0.69 to 15.4
Medication use, mean ± SD (mg)
Preoperative LEDD	1273 ± 668	1370 ± 903	0.896
Postoperative LEDD	951 ± 667	1029 ± 749	0.872
Stimulation parameters
Amplitude, mean ± SD	2.0 ± 0.6 Va	2.0 ± 0.4 Vb	0.882
	2.1 ± 0.4 mAc	1.8 ± 0.4 mAd	0.300
Frequency, 130 Hz bilateral, n (%)	35 (92%)	13 (87%)	0.614
Frequency, other, n (%)*	3 (8%)	2 (13%)
Pulse width, 60 μs bilateral, n (%)	22 (58%)	7 (47%)	0.716
Pulse width, 90 μs bilateral, n (%)	9 (24%)	4 (27%)
Pulse width, asymmetric, n (%)∼	7 (18%)	4 (27%)
Depression and anxiety, mean ± SD
Preoperative BDI score	12.7 ± 8.5	13.1 ± 7.3	0.691
Preoperative BAI score	9.1 ± 6.0e	11.6 ± 7.7f	0.434
Cognitive function, mean ± SD
Preoperative MMSE score	28.3 ± 1.4g	28.1 ± 1.3h	0.821
Preoperative DRS-2 score†	12.9 ± 1.9i	13.1 ± 2.0j	0.860
UPDRS-III, mean ± SD
Preoperative score	21.2 ± 11.2	21.2 ± 12.2	0.934
Postoperative score	19.3 ± 9.8k	19.8 ± 10.0l	0.643
FrSBe apathy, mean ± SD
Preoperative score	65.6 ± 17.4	59.6 ± 7.3	0.472
Postoperative score	70.1 ± 19.4	59.3 ± 16.8	0.105
Change	4.6 ± 14.2	−0.27 ± 14.2	0.466
Range	−20 to 37	−27 to 21

Sample sizes varied by variable. Stimulation parameter sample sizes reflect programming method: voltage-programmed (a: n=29, b: n=12) and current-programmed (c: n=9, d: n=3). Missing BAI scores (e: n=21, f: n=8). Preoperative assessment varied by instrument: MMSE (g: n=17, h: n=7) and DRS-2 (i: n=14, j: n=8); 7 training subjects did not have either assessment. Missing postoperative UPDRS-III scores (k: n=28, l: n=13).

*Non-130 Hz frequencies included 100 Hz (n=1), 180 Hz (n=2), and asymmetric settings (n=2).

~Asymmetric pulse width combinations included 60/90 μs (n=9), 60/120 μs (n=1), and 80/90 μs (n=1).

†DRS-2 scores are demographically corrected scaled scores (M = 10, SD = 3).

P-values were calculated using Mann-Whitney U tests for continuous variables and Fisher’s exact tests for categorical variables.

Abbreviations: LEDD = levodopa equivalent daily doses; FrSBe = Frontal Systems Behavior Scale; UPDRS-III = Unified Parkinson Disease Rating Scale Part III; BDI = Beck Depression Inventory; BAI = Beck Anxiety Inventory; MMSE = Mini-Mental State Examination; DRS-2 = Dementia Rating Scale-2.

Results

Clinical data

Apathy assessments were conducted on average 2.4 ± 2.2 months before surgery and 8.5 ± 1.8 months after surgery. Subjects from both groups had a mean age of 64.5 ± 8.6 years, were 75% male, with an average disease duration of 8.0 ± 4.7 years at the time of surgery. For the full cohort, mean stimulation amplitude was 2.0 ± 0.5 V (n = 41) and 2.0 ± 0.4 mA (n = 12). Frequency was 130 Hz bilaterally in 91% of subjects (48/53; range: 60–180 Hz). Pulse width was 60 μs bilaterally in 55% (29/53), 90 μs bilaterally in 25% (13/53), and asymmetric in 21% (11/53) of subjects. Stimulation amplitude did not differ between training and holdout groups among subjects with voltage-based settings (p = 0.882) and among those with current-based settings (p = 0.300). Frequency (p = 0.614) and pulse width (p = 0.716) distributions also did not differ. Subject characteristics, available baseline neuropsychiatric and cognitive measures, and changes in clinical measures are summarized by group in Table 1. Subject characteristics were similar between groups. Mean age was numerically higher in the holdout cohort (67.9 ± 4.7 vs. 63.2 ± 9.4 years), but this difference did not reach statistical significance (p = 0.122).

While the mean FrSBe apathy change score after DBS was 3.2 ± 14.2 (SD), there was considerable variability. Change scores ranged from a 27-point improvement (−2.7 SD) to a 37-point worsening (+3.7 SD). This variance in apathy changes after STN-DBS is consistent with prior reports.^11,16 Notably, 55% of subjects (29/53) experienced a worsening in apathy measures, while 40% (21/53) saw an improvement post-surgery. Among those whose apathy worsened, 66% experienced a meaningful worsening of apathy, and 34% developed clinically significant apathy.

Connectivity related to worsening apathy following STN-DBS

A structural connectivity analysis revealed an asymmetric profile associated with apathy following STN-DBS, where only streamlines associated with worsened apathy were identified (Figure 1A). In the right hemisphere, the model included streamlines consistent with the dentatorubrothalamic tract (DRTT), particularly those connecting to the supplementary motor area (SMA) (Figure 1B). Conversely, in the left hemisphere, the analysis identified streamlines linking the medial orbitofrontal cortex (mOFC) and the subgenual anterior cingulate cortex (sgACC) to the anteromedial STN via the nucleus accumbens (NAc) (Figure 1C). Active contact coordinates did not differ significantly between hemispheres in MNI space, with all absolute coordinate differences < 0.5 mm. Mean X-coordinates were 11.27 ± 1.30 mm on the right and −11.51 ± 1.52 mm on the left (p = 0.509). Mean Y-coordinates were −13.12 ± 1.75 mm on the right and −13.56 ± 1.44 mm on the left (p = 0.075). Mean Z-coordinates were −5.94 ± 2.23 mm on the right and −5.77 ± 2.47 mm on the left (p = 0.867).

Figure 1.

Streamlines associated with worsening apathy following STN-DBS for Parkinson's disease. Streamlines (light blue) associated with worsening apathy were identified through a fiber filtering analysis in the training group (n = 38). Atlas segmentations, derived from the DISTAL Minimal ⁴⁹ and DBS Tractography ⁵⁰ atlases, are overlaid on a 100 μm, 7 T brain scan in Montreal Neurological Institute (MNI) 152 space. ⁵⁴ The subthalamic nucleus (STN) is segmented into motor (yellow), associative (blue), and limbic (purple) regions. (A) Coronal view from y = −27.5 mm, including both left and right hemisphere streamlines traversing the STN. (B) Right hemisphere (RH) streamlines, depicted from x = −8.0 mm sagittal slice, include streamlines consistent with the dentatorubrothalamic tract (DRTT; red). (C) Left hemisphere (LH) streamlines, depicted from x = −13.0 mm sagittal slice, traverse the nucleus accumbens (NAc; green).

In the training group, validity of the connectivity model was evaluated using three cross-validation paradigms: leave-one-patient-out (R = 0.38, p = 0.023), 5-fold (R = 0.46, p = 0.004), and 10-fold (R = 0.33, p = 0.048) (Figure 2). The robustness of the model was further demonstrated by its significant association with changes in apathy in the holdout group (R = 0.71, p = 0.005) (Figure 3).

Figure 2.

Leave-one-patient-out cross-validation of the apathy connectivity model derived from the training group. This iterative analysis involved excluding one subject at a time and estimating the left-out subject's change in FrSBe apathy score based on the connectivity profile derived from the remaining subjects.

Figure 3.

Correlation between changes in FrSBe apathy scores and modulation of the apathy connectivity model (fiber R-scores) in the holdout group. The apathy connectivity model developed from the training group was independent of data from the holdout group.

Subject characteristics vs apathy changes

Since prior reports have proposed postoperative medication and/or motor changes as determinants of changes in apathy measures, we also evaluated available data for medication change (LEDD) and motor improvement (UPDRS-III) versus changes in apathy. There was no correlation between changes in LEDD and changes in apathy scores (R = −0.11, p = 0.433) or between changes in UPDRS-III scores and changes in apathy scores (R = 0.14, p = 0.381). Additionally, since dopamine agonists and cholinesterase inhibitors are some of the most commonly prescribed pharmacological treatments for apathy in PD,^55,56 we evaluated their influence on changes in apathy scores. Sixty-eight percent (36/53) of subjects were prescribed dopamine agonists during the study period. Apathy change scores did not significantly differ across subjects who continued postoperatively (n = 24), discontinued postoperatively (n = 12), or did not use dopamine agonists (n = 17; ${\chi }^2$ (2) = 1.55, p = 0.461). Four participants were initiated on cholinesterase inhibitors postoperatively, and their apathy change scores fell within 1 SD of the sample mean. Medication changes do not appear to account for the observed apathy outcomes.

Discussion

This study identified a structural connectivity profile associated with worsening apathy following STN-DBS in Parkinson's disease. The profile exhibited asymmetry involving distinct cortical streamlines in both hemispheres. In the right hemisphere, streamlines included those consistent with the dentatorubrothalamic tract (DRTT), specifically connecting the STN and supplementary motor area (SMA). In the left hemisphere, streamlines spanned between the medial orbitofrontal cortex (mOFC) and subgenual anterior cingulate cortex (sgACC) to the anteromedial STN via the nucleus accumbens (NAc). Significant correlations were observed between changes in FrSBe apathy subscale scores and stimulation of these specific streamlines, both in cross-validation analyses and in the holdout group not used to generate the connectivity profile. Furthermore, variance in apathy outcomes was independent of post-operative changes in medications and motor symptoms.

To date, only one other study has investigated the structural connectivity associated with apathy following STN-DBS, which involved patient-specific network analyses of 6 patients with worsening apathy and 16 patients whose apathy did not worsen. ⁵⁷ That study found that the occurrence of apathy was linked to active contacts in regions with a higher density of projections to associative cortical areas and a lower density of motor projections in both hemispheres. Our findings are generally consistent with these results and further expand upon them by identifying specific streamlines and associative regions related to worsening apathy, particularly the mOFC and sgACC. While the left hemisphere streamlines in our study are consistent with those findings, the right hemisphere of the connectivity profile identified here diverges via its inclusion of STN-SMA streamlines associated with worsening apathy.

The lateralization of apathy is consistent with previous functional connectivity studies.^14,32 Thobois et al. found that apathy following STN-DBS involves dysfunction in projections to the OFC, cingulate cortices, and the dorsolateral prefrontal cortex (DLPFC). ¹⁴ Their study also highlights the critical role of the left ventral striatum in the development of apathy, which complements our findings of left hemisphere streamlines passing through the nucleus accumbens and projecting to the left mOFC and sgACC. Conversely, a recent magnetoencephalography study associated worsening apathy post-surgery with decreased alpha1 band functional connectivity in the bilateral DLPFC, but not in the mOFC or ACC. ³² Notably, they also observed that more dorsolateral stimulation was linked to worsening apathy in the right hemisphere exclusively.^32,58 These left-right distinctions in apathy occurrence post-STN-DBS support the model identified in our study, which includes two distinct hemispheric streamline bundles. The differences in involved structures may be attributed to the distinction between structural and functional connectivity, with the DLPFC potentially influenced indirectly by DBS through downstream effects. ³²

An intriguing finding of this study was the inclusion of streamlines consistent with the right DRTT connecting to the SMA. This inclusion was unexpected, as the DRTT is typically associated with increased tremor suppression in DBS for essential tremor.^59,60 Given these results, one might hypothesize that patients undergoing DBS for essential tremor would exhibit higher levels of apathy; however, studies investigating apathy following DBS for essential tremor are currently lacking. Moreover, stimulating SMA tracts is generally associated with improvement in hypokinetic symptoms following STN-DBS in PD^61–63 and has not been previously linked to worsening apathy. Nonetheless, SMA activity is closely linked to motivational and initiation processes, and disruptions in this activity (e.g., decreased connectivity between the ACC and SMA) are associated with increased apathy.^56,64–67 These findings are hypothesis-generating and warrant further investigation through additional studies.

The connectivity profile identified in this study includes many of the streamlines associated with optimal clinical improvement in STN-DBS for obsessive-compulsive disorder (OCD). ⁶⁸ The OFC, ACC, and NAc are involved in both the worsening of apathy in PD and OCD symptom improvements following DBS.^68,69 Given the roles these pathways play in reward processing and motivation, it is not surprising that their stimulation affects both apathy and OCD. While further investigation is needed, this potential connection supports personalized DBS and tractography-based targeting approaches to optimize quality of life outcomes.

The association of multiple cortical regions, including the mOFC, sgACC, and SMA, with apathy in the present study is likely attributable to the multidimensional nature of apathy. Apathy can be induced by disruptions in three distinct subtypes of processing: emotional-affective, cognitive, and auto-activation.^56,70 Specifically, the dysfunction of networks involving the mOFC is likely indicative of emotional-affective apathy, while those involving the SMA are likely representative of auto-activation apathy. ⁵⁶ The sgACC is implicated in both emotional-affective and auto-activation apathy. ⁵⁶ In contrast, cognitive apathy is most commonly associated with dysfunction in the DLPFC and related structures.^56,70

The primary measure utilized in this study was the FrSBe apathy subscale. While this scale is recognized for its validity and reliability, ⁷¹ it assesses apathy as a single syndromic dimension. ⁷² This approach contrasts with the prevailing view in the research community that apathy is a multidimensional syndrome. ⁷² Consequently, this represents a limitation of the present study, as symptoms indicative of cognitive apathy may not have been fully captured by the FrSBe apathy subscale, especially given that the FrSBe includes a separate subscale for cognitive executive dysfunction. In contrast, dedicated apathy scales typically evaluate multiple distinct domains. For example, the Apathy Evaluation Scale (AES), a more commonly used measure of apathy, assesses it as a psychological dimension encompassing simultaneous deficits in the behavioral, cognitive, and emotional components of goal-directed behavior. ⁷³ Given this limitation, it is probable that the present model does not fully capture all aspects of apathy. Other streamlines and cortical structures, particularly those implicated in the cognitive apathy subdomain (e.g., the DLPFC), may also be associated with apathy following STN-DBS.

Several additional limitations of this study warrant discussion. Firstly, a normative connectome was utilized due to the unavailability of patient-specific tractography data. Although patient-specific connectivity might elucidate slightly more variance than group-averaged streamlines, both modalities have been demonstrated to produce highly similar results.^74,75 Consequently, the use of a normative connectome introduced biases towards nonsignificance, stemming from inherent inaccuracies when aggregating patient scans into a common space. Additional uncertainty was introduced during the reconstruction of electrode placements and the co-registration of patient scans with varying resolutions and types. To mitigate these biases, a purpose-built pipeline was employed, featuring adjustments for brain shift, multispectral normalizations using a validated segmentation framework,^51,76 and an automated electrode localization algorithm (PaCER) validated with phantom models. ⁴⁷ Furthermore, the absence of longitudinal data in the present study limits the ability to draw conclusions about long-term applicability. Additionally, the cohort was predominantly composed of biological male participants, which restricts sex-based analyses and limits the generalizability of our findings. Lastly, the dataset used in the present study was derived from a single institution and was limited in sample size, decreasing the generalizability of the results. Baseline neuropsychiatric and cognitive characterization was also incomplete, as depressive, anxiety, and cognitive measures were not uniformly available across the clinical cohort. Future research should validate this connectivity profile in independent cohorts with larger sample sizes and individualized connectomic data, and incorporate standardized neuropsychiatric and cognitive phenotyping. Additionally, evaluating model performance across multidomain apathy instruments, such as the Lille Apathy Rating Scale, and longitudinal assessments could help clarify the heterogeneity observed across studies.

In conclusion, while STN-DBS is a safe and efficacious treatment for PD, apathy worsening remains a common side effect that alters quality of life for many but not all patients. This study identified that stimulation of asymmetric streamlines connecting to the left mOFC and sgACC, as well as the right SMA, correlates with worsening apathy following STN-DBS. These findings underscore the broader network-wide implications of STN-DBS and suggest potential disruptions in auto-activation and emotional-affective processes in some patients. Future work in larger, independent cohorts using individualized connectomic data, longitudinal apathy measures, and more comprehensive neuropsychiatric and cognitive characterization will be necessary to further evaluate these exploratory findings. Within this context, the identified connectivity profile offers insights that may help inform refinement of clinical targeting strategies for STN-DBS. Moreover, these results support continued investigation of tractography-based positioning and programming to optimize DBS outcomes and personalize treatment.