Significance of neurodegeneration and neuroplasticity serum biomarkers in Parkinson’s disease patients treated with subthalamic stimulation

Abstract

The ability of serum biomarkers to predict the prognosis and response to deep-brain stimulation (DBS) therapy in Parkinson’s disease (PD) patients is promising. Here, we showed that NfL differed between healthy individuals and PD patients and that changes in NfL, GFAP, and BDNF occurred only transiently after DBS surgery. Therefore, subthalamic stimulation does not promote neurodegeneration, and these biomarkers do not serve as clinical improvement endpoints in PD DBS patients.

Introduction

Preclinical research has further bolstered the notion of Parkinson’s disease (PD) modification following subthalamic nucleus deep-brain stimulation (STN-DBS) through neural mechanisms such as increased survival of dopaminergic nigral neurons, increased brain-derived neurotrophic factor (BDNF) production, increased synaptic remodeling, dampened neuroinflammation, and decreased glutamate excitotoxicity^1,2. However, while these promising reports have pointed to DBS-mediated neuroprotective mechanisms in PD animal models, clinical studies have shown limited evidence supporting the disease-modifying effect of DBS in humans^3–5. Moreover, although patients typically respond well to DBS therapy⁶, clinical and neuroimaging studies have also revealed dopaminergic neuron loss comparable to that in non-DBS patients, continued progression of symptoms, and STN-DBS-induced side effects^3,7,8.

To gain additional insights into the neural processes underpinning DBS treatment in PD, recent attention has focused on the study of protein fluid- and blood-based biomarkers, which have been progressively incorporated into clinical routines and trials, providing optimal analytical sensitivity to track disease progression in several neurological conditions^9–11. However, the few studies in humans that have investigated these biomarkers in DBS-treated PD patients have not detected signals of potential disease modifications¹². In this context, it is crucial to identify affordable and reliable biomarkers of neurodegeneration and neuroplasticity that can assist in clarifying the underlying mechanisms associated with STN-DBS use, discerning whether they predominantly arise from natural disease progression, the effects of DBS surgery itself, or the direct influence of STN stimulation.

The main objective of this study was to investigate the impact of DBS surgery and continuous STN stimulation on peripheral blood-based biomarkers associated with neuroaxonal damage, astrocyte reactivity, and neuroplasticity, as well as their relationship with patient clinical status, to further identify the neurodegenerative and neuroprotective signatures associated with STN-DBS in PD patients. To achieve this aim, we quantified the serum protein levels of neurofilament light chain (sNfL), glial fibrillary acidic protein (sGFAP), and sBDNF in PD patients. Initially, we longitudinally assessed the acute effects of DBS electrode implantation and short-term STN stimulation in PD patients who newly underwent STN-DBS treatment (“new” DBS, PD-nDBS) at baseline and at the 1-week, 1-month, and 1-year follow-ups. Moreover, to further understand the impact of continuous high-frequency stimulation of the STN on these biomarkers, we compared PD patients undergoing chronic STN stimulation (PD-chrDBS) with a symptomatology-matched PD patient group receiving the best medical treatment (PD-BMT) and an additional age-, sex-, and education-matched healthy control (HC) cohort (Fig. 1). Assessments of the clinical status of PD patients were also performed.

Fig. 1. Schematic overview of the study design, detailing the timeline and frequency of measurements for the four participant cohorts.

The top left panel highlights the longitudinal design for the PD-nDBS cohort, and the top right panel shows the cross-sectional characteristics of the PD-chrDBS, PD-BMT, and HC groups. Specifically, the PD-nDBS group included PD patients who underwent STN-DBS surgery and who were longitudinally assessed at four-time points: before electrode implantation (baseline) and at three subsequent intervals postoperatively (1 week, 1 month, and 1 year). In this group, blood was collected at all four-time points, while clinical and neurocognitive assessments were performed only at baseline and at the 1-year follow-up. In contrast, PD patients receiving short- and midterm DBS treatment (PD-chrDBS), PD patients receiving the best pharmacological treatment (PD-BMT), and HCs were compared at a single time point in terms of serum biomarker levels as well as clinical and neurocognitive status. The total number of participants per cohort group is shown. DBS deep brain stimulation, HC healthy control, PD-BMT PD patients receiving the best medical treatment, PD-chrDBS PD patients undergoing chronic STN stimulation, PD-nDBS PD patients newly receiving DBS treatment, sBDNF serum brain-derived neurotrophic factor, sGFAP serum glial fibrillary acidic protein, sNfL serum neurofilament light chain.

In the PD-nDBS group, the different time-point assessments, including preoperative, lead implantation, and STN stimulation, enabled a more detailed examination of different phases after STN-DBS, distinguishing between the effects of the DBS surgical procedure and those related to STN stimulation at the 1-year follow-up. Initially, following electrode placement but preceding stimulation initiation, patient Movement Disorder Society Unified Parkinson’s Disease Rating Scale—part III (MDS-UPDRS-III) scores decreased by 39% between the drug-off/DBS-off stage at 48 hours after lead implantation and the drug-off baseline period before surgery (p = 0.001). This temporary motor improvement may be attributed to the well-established phenomenon known as the “microlesion effect”, which describes the immediate reduction in PD symptoms, even with medication suspended and DBS switched off, resulting from the insertion of an inactive electrode into the STN¹³. These outcomes were verified by quantifying the dynamic changes in serum biomarkers (Fig. 2a). During the interval from the preoperative period to 1 week after DBS surgery, a transient statistically significant increase in sGFAP (p = 0.019) was noted, along with slight increases in sNfL and (nonsignificant) decreases in sBDNF levels. Subsequent evaluation at the 1-month follow-up revealed a continued increase in sNfL when it reached its peak level, which significantly differed from the preoperative values (p ≤ 0.001), whereas sGFAP and sBDNF levels seemed to return to levels comparable to the baseline values. The initial increase in sGFAP levels indicates an acute neuroinflammatory astrocyte response to surgical trauma and electrode insertion¹⁴, whereas the transient decrease in BDNF may be due to downregulation linked to early increases in proinflammatory mediators¹⁵ and the promotion of apoptotic mechanisms¹⁶. Although our data suggest a decrease in BDNF expression and signaling postsurgery, the interaction between GFAP and BDNF in this context remains to be fully elucidated. Likewise, the delayed increase in sNfL levels points to neuronal injury, linking STN-DBS surgery to neuroaxonal damage¹⁷. This temporal dynamic of sNfL and sGFAP has been previously documented in analogous studies on cerebrospinal fluid (CSF)¹⁸, as well as in stroke¹⁹ and traumatic brain injury conditions²⁰. Although there is evidence linking slow CSF flow and localized damage to the delayed sNfL increase¹⁸, the dynamics of sGFAP release appear to be influenced by other brain processes, suggesting that the varying presence of these surgery DBS-induced proteins in the bloodstream could be affected by different intrinsic release and clearance kinetics. Together, these results indicate that astrocyte reactivity and neuroinflammation play significant roles in the initial weeks following STN-DBS surgery, before peripheral blood-based signs of neuronal damage emerge. At the 1-year follow-up, compared with patients in the preoperative period, PD-nDBS patients in the postoperative period showed significant improvements in clinical outcomes, controlled motor symptoms, improved activities of daily living, and reduced reliance on dopaminergic medication, which is consistent with solid clinical evidence²¹. In contrast, there was no considerable variation in the serum biomarker levels between the preoperative and 1-year postoperative time points (Table 1). However, the sNfL and sGFAP levels decreased considerably compared with the notable increase they exhibited during the early postoperative period (p = 0.001 and p = 0.006, respectively), which indicates the possible stabilization of neuroinflammatory and neural damage processes. Conversely, sBDNF levels at the 1-year follow-up were considerably greater than sBDNF levels at 1 week following electrode implantation (p = 0.021), although they remained similar to baseline levels, indicating transient and resolved neural damage after surgery rather than potential DBS therapy-induced synaptic plasticity mechanisms (Fig. 2a). These findings are in line with those of earlier studies showing that, at least temporarily, brain tissue injury is not promoted by the presence of electrodes or continuous high-frequency bilateral STN stimulation over a short-term period^17,18. Interestingly, the levels of protein biomarkers in CSF have been shown to remain normal despite long STN-DBS treatment, PD duration, and severe disability¹². Although several studies indicate a potential association between blood biomarkers and both motor and nonmotor outcomes in PD^22–24, motor clinical improvement was not clearly associated with any decrease or increase in serum biomarker levels at any of the time points assessed, so changes in neurodegenerative or neuroplasticity biomarkers did not seem to influence susceptibility to STN-DBS clinical response, at least at the 1-year follow-up.

Fig. 2. Graphical representation of serum biomarker levels across study groups.

Panel a displays longitudinal measurements of sNfL, sGFAP, and sBDNF before and after STN-DBS surgery in the PD-nDBS cohort. Panel b illustrates sNfL differences between HCs and PD patients without grouping discrimination. Panel c shows blood-based biomarker levels for PD-chrDBS, PD-BMT, and HCs. In panel c, the colors purple (sNfL), green (sGFAP), and blue (sBDNF) represent the serum biomarkers, with the intensity of the colors differentiating the study groups into high (PD-chrDBS), medium (PD-BMT), and low (HC) intensities. The y-axis represents biomarker levels expressed in pg/mL. Asterisks indicate statistically significant values (≤ 0.05). In the boxplots, the central box displays the interquartile range (IQR), with the box hinges representing the first (Q1) and third (Q3) quartiles. The line inside the box marks the median. The whiskers follow Tukey’s method and extend from the hinges to the highest and lowest values within 1.5 times the IQRs from Q1 and Q3.

Table 1.

Demographic and clinical features of the study cohorts

	PD-nDBS - Baseline -	PD-nDBS - 1-year -	p value	PD-chrDBS	PD-BMT	HC	p value
n	15			15	17	17
Age (years)	55.7 ± 11.1			55.9 ± 7.2	58.9 ± 8.6	59.4 ± 8.2	0.415
Sex	F = 7/M = 8			F = 5/M = 10	F = 8/M = 9	F = 6/M = 11	0.682
Disease duration (years)	11.6 ± 7.5			13.5 ± 7.4	8.7 ± 3.6		0.022
DBS duration (years)				2.3 ± 2.3
LEDD (mg) (Levodopa equivalent daily dose)	1266.0 ± 505.4	697.1 ± 413.7	0.000	902.3 ± 441.2	917.6 ± 288.0		0.910
MDS-UPDRS III
On (drug)	16.7 ± 9.5	20.2 ± 8.6∆	0.421	30.9 ± 14.0∆	27.6 ± 10.7		0.458
Off (drug)	53.5 ± 12.2	27.3 ± 17.1♦	0.001	53.5 ± 13.6 ∇
MDS-UPDRS II	16.7 ± 6.4	10.1 ± 4.8	0.000
MDS-UPDRS IV	9.6 ± 2.6	2.7 ± 1.3	0.000
PDQ-39 (Parkinson’s Disease Questionnaire)	33.0 ± 17.9	26.3 ± 11.2	0.233	30.2 ± 16.3	22.7 ± 15.6	7.4 ± 7.9	0.000 a
Education (years)	10.3 ± 5.1			11.8 ± 3.1	10.4 ± 3.4	15.5 ± 5.0	0.035 b
MMP (Mini-Mental Parkinson)	27.2 ± 2.3	25.6 ± 3.6	0.383	25.9 ± 4.0	28.4 ± 2.5	28.9 ± 2.4	0.025 c
FAB (Frontal Assessment Battery)	14.7 ± 2.5	14.8 ± 2.6	0.730	15.3 ± 2.6	16.4 ± 3.5	17.7 ± 0.5	0.002 d
BDI-II (Beck’s Depression Inventory)	13.3 ± 5.6	11.6 ± 6.9	0.513	12.4 ± 8.5	14.0 ± 9.8	9.0 ± 6.2	0.145
HAM-A (Hamilton Anxiety Rating Scale)	14.8 ± 10.4	8.8 ± 5.0	0.207	11.0 ± 9.6	8.2 ± 8.8	7.9 ± 8.1	0.456
SAS (Starkstein Apathy Scale)	11.7 ± 5.3	11.7 ± 4.5	0.718	10.6 ± 5.4	10.8 ± 6.0	10.7 ± 5.7	0.908
sNFL (pg/mL)	13.8 ± 8.6	18.3 ± 12.8	0.256	14.7 ± 11.5	11.8 ± 6.1	8.2 ± 4.4	0.078 †
sGFAP (pg/mL)	110.3 ± 67.8	121.1 ± 96.5	0.934	94.0 ± 50.0	105.0 ± 68.5	82.5 ± 64.7	0.597
sBDNF (ng/mL)	16.5 ± 4.7	22.7 ± 10.2	0.095	22.8 ± 11.2	19.4 ± 8.1	16.9 ± 6.7	0.360

The values are presented as the means ± SDs, except for sex and n. A p-value ≤ 0.05 was considered to indicate statistical significance. Bold numbers indicate statistically significant values. † Indicates a statistical trend. ^∆drug-on/DBS-on; ^♦drug-off/DBS-on; ^∇drug-off/DBS-off. ^a HC vs. PD-chrDBS (p = 0.000) & HC vs. PD-BMT (p = 0.010); ^b HC vs. PD-BMT (p = 0.031); ^c HC vs. PD-chrDBS (p = 0.033); ^d HC vs. PD-chrDBS (p = 0.002) & HC vs. PD-BMT (p = 0.028).

Despite the induction of neuroinflammation and neuronal damage following lead implantation, biomarker levels returned to baseline levels within 1 year of follow-up, with no definitive evidence of neuroprotection or disease modification. These findings, however, do not clarify the extent to which the observed beneficial or adverse clinical outcomes of chronic STN-DBS usage might be directly attributable to the impact of ongoing STN stimulation as opposed to the natural course of PD. Consequently, PD-chrDBS patients were compared to clinically similar PD-BMT patients receiving only pharmacological treatment, as well as to HCs (Table 1). Clinically, PD-chrDBS and PD-BMT patients exhibited comparable clinical profiles and were distinguished solely by a longer disease duration in the PD-chrDBS group (p = 0.022), an unsurprising finding consistent with the natural progression of the disease and the use of advanced therapy such as DBS. With respect to biomarkers, sNfL but not sGFAP or sBNDF was significantly increased in PD patients compared with HCs (p = 0.033; Fig. 2b), with no differences between the PD-chrDBS and PD-BMT subgroups (Fig. 2c). However, when adjusting for age, disease duration, or cognitive status, no discernible distinctions among different PD therapeutic intervention groups were detected for any of the serum-based biomarkers analyzed (Fig. 2c), which is consistent, in part, with previous findings from other studies¹⁷. Despite some contradicting evidence from early-stage PD and CSF samples²⁵, these observations suggest that NfL is a potential neurodegeneration biomarker in PD, distinguishing patients with advanced disease stages from HCs, with NfL levels increasing with disease progression and worsening symptoms²⁶. Remarkably, sNfL elevation may indicate underlying disease pathology rather than treatment-specific effects, as suggested by the lack of differentiation between the PD-chrDBS and PD-BMT cohorts. Although STN-DBS may modulate certain disease features, its impact on serum biomarkers seems to be influenced by baseline PD characteristics, emphasizing that DBS does not promote neurodegenerative processes, even though its neuroprotective effects remain to be verified.

The small sample size, absence of a treatment-naïve PD group, lack of association with clinical outcomes, and nonspecificity of the biomarkers are some of the drawbacks of the current study that should be addressed in future research. Nonetheless, the potential impact was enhanced by successfully evaluating three blood biomarkers in multiple PD patient groups.

In summary, our findings confirm that DBS surgery temporarily causes neuronal damage and triggers the regulation of proinflammatory astrocyte activation, possibly by influencing BDNF signaling. Nevertheless, the normalization of biomarkers after 1 year of STN stimulation, along with the absence of differences based on PD treatment, confirms that STN-DBS does not promote neurodegeneration, thereby leading to further questions concerning the presumed neuroprotective effect of STN-DBS. Interestingly, serum biomarkers may help in tracking disease severity and progression, although inconsistencies between changes and clinical improvement challenge the validity of biomarkers thus far. Further research is necessary to clarify how serum biomarker kinetics, clinical outcomes, and DBS treatment interact, enabling tailored therapeutic approaches in PD.

Methods

Participants

The study was conducted under a protocol approved by the Andalusian Biomedical Research Ethics Committee (Refs. Acta03/2018 and 2169-N-19), in agreement with the guidelines of the Declaration of Helsinki. During the informed consent process, participants received a comprehensive explanation of the different study stages and experimental procedures, and they provided written consent to participate. This investigation was performed exclusively for research purposes, and the participants received no financial compensation for their participation. After obtaining formal written consent from the participants, all blood samples from each participant were transferred to and stored at the facilities of the Biobank of the Andalusian Public Health System at the Puerta del Mar University Hospital.

A total of 64 participants were recruited from the Movement Disease Unit, Neurology Service, Puerta del Mar University Hospital, and the Movement Disorders Unit, Department of Neurology, Virgen de las Nieves University Hospital. All individuals enrolled in this study underwent clinical, neurological, and neurocognitive examinations in conjunction with the collection of blood samples. The study involved four different cohorts: i) PD-nDBS, 15 PD patients assessed longitudinally both before DBS surgery and at different postoperative time points up to the 1-year follow-up; ii) PD-chrDBS, 15 PD patients receiving chronic DBS treatment between 6 months and 5 years; iii) PD-BMT, 17 PD patients receiving the best pharmacological medical treatment; and iv) HCs, 17 age-, sex-, and education-matched healthy individuals. The cohorts of patients included in this study were selected with specific criteria to address different aspects of PD and STN-DBS therapy (Fig. 1). Specifically, the PD-nDBS group comprised PD patients who met the criteria for DBS treatment as defined by the Core Assessment Program for Surgical Interventional Therapies in Parkinson’s Disease²⁷. These patients were longitudinally assessed at four different time points: i) at baseline, within 1 month before bilateral electrode placement; ii) at 1 week after surgery; iii) at 1 month after STN-DBS surgery; and iv) at the 1-year follow-up after receiving continuous STN-DBS treatment. The PD-chrDBS cohort comprised PD patients who underwent short- and midterm bilateral STN-DBS therapy, with treatment durations ranging from 6 months to 5 years. These patients had maintained stable stimulator settings that effectively managed clinically manifested motor symptoms for approximately 3 months before their enrollment in the study. The PD-BMT group consisted of individuals diagnosed with PD who were receiving the best pharmacological medical treatment. Additional inclusion criteria for the PD-BMT cohort included patients who scored between II and III on the Hoehn and Yahr scale in the on-medication state, who had a disease duration of at least 5 years, and who demonstrated clinical and therapeutic stability throughout the 2-month period immediately preceding the recruitment phase. All aforementioned patient cohorts were diagnosed with PD according to the Movement Disorder Society (MDS) clinical diagnostic criteria for PD²⁸. Finally, an HC group of nonneurologically impaired participants who were matched to PD patients in terms of age, sex, and education level was also recruited. All participants in the HC group denied having active known or treated neurological or psychiatric conditions, a family history of idiopathic PD, or being under the influence of any medications that were considered to impact neurological functioning. To maintain homogeneity within the study population and minimize potential confounding variables, certain exclusion criteria were also established. Specifically, participants were excluded from the study if they met any of the following criteria: (i) severe or moderate cognitive impairment comparable to dementia, as revealed by a Mini-Mental Parkinson (MMP) score < 24; (ii) any incapacitating psychiatric condition; (iii) a history of drug or alcohol abuse or impulse control disorder; (iv) the presence of a serious systemic disease, other neurologic disorders, or medical conditions; (v) atypical parkinsonism or neurological comorbidities; (vi) recent traumatic events, acute fever, or neuroinflammation; and (vii) DBS surgical complications, such as hemorrhage, infection, perielectrode edema, breaking DBS leads or wires, electrode impedance, and hardware-related issues. In the PD-nDBS group, two of the 15 recruited patients were not able to be examined at the one-year follow-up because of surgical complications. The first patient’s DBS treatment was interrupted due to lead wire breakage, and the second patient developed an infection that led to the removal of their DBS system.

STN-DBS surgery

The clinical planning, surgical procedures, and targeting details for surgical placement of DBS electrodes bilaterally into the STN were applied following standard guidelines²⁹. Patients underwent both stereotactic frame-based (CRW frame; Integra Radionics, MA, USA) and frameless (Nexframe, Medtronic, MN, USA) surgeries. Briefly, the electrode trajectories were preoperatively traced via individual stereotactic magnetic resonance imaging (MRI) based on the intended coordinates for the STN via Medtronic StealthStation navigation systems (Medtronic). During surgery, intraoperative techniques such as microelectrode array recordings, macrostimulation, and assessment of the clinical effects of stimulation were used to adjust DBS electrode placement at the STN and to confirm the lack of stimulation-induced side effects. Moreover, during stereotactic frame surgery, the lead position was confirmed intraoperatively via a computed tomography (CT) scan (O-Arm, Medtronic). The definitive location of the DBS lead was confirmed through the fusion of postoperative CT and preoperative MRI scans. For all participants, a conventional quadripolar macroelectrode (model 3389, Medtronic) was used. These electrodes had four cylindrical platinum‒iridium surfaces with a diameter of 1.27 mm, contact length of 1.5 mm, and 0.5 mm spacing between contacts.

Clinical and neurocognitive assessments

Demographic and clinical data were collected by neurologists specializing in movement disorders who were certified through the MDS-UPDRS training program. These assessments included information about disease duration (recorded as years since the first manifestation of classical parkinsonism motor symptoms), levodopa-equivalent daily dose (LEDD), motor status, and the presence of disease-related complications. The motor symptoms of PD patients were evaluated via part III of the MDS-UPDRS. Motor parameters were scored in both the off and on states for all groups of PD patients. In the off state, patients stopped taking dopaminergic drugs overnight, and only the PD-chrDBS group also had DBS stimulation turned off during this period. In contrast, the term on state denotes the period following a patient’s usual dose of levodopa, and specifically, within the PD-chrDBS group, this designation also implies the simultaneous activation of DBS systems. Furthermore, motor experiences in daily living and the severity and impact of motor complications were quantified via parts II and IV of the MDS-UPDRS, respectively. The impact of PD on patients’ daily lives was evaluated through the Parkinson’s Disease Questionnaire (PDQ-39). Higher scores on the aforementioned tests reflect more severe symptoms or impairment. The LEDD, expressed in mg, was calculated following the standard conversion³⁰. All the participants involved in this study were also assessed to quantify their cognitive and psychiatric status by experienced neuropsychologists specializing in the evaluation of PD patients. Global cognitive performance was assessed by the MMP. In addition, executive functions of the frontal lobes were specifically screened with the Frontal Assessment Battery (FAB). A psychiatric assessment was applied, including the Beck Depression Inventory-Second Edition (BDI-II) for the evaluation of depressive symptomatology severity, the Hamilton Anxiety Rating Scale (HAM-A) for measuring the severity of anxiety symptoms, and the Starkstein Apathy Scale (SAS) for screening and quantifying the severity of apathetic symptoms. The administration of neurocognitive tests as well as the scoring and correction of data were performed according to the specific international recommendation for each test and following Spanish adaptation guidelines. All PD patients underwent neurocognitive assessments in the on-drug state, whereas PD-chrDBS patients underwent testing both in the on-stimulation and on-drug states. Information on education level and years of schooling was also collected.

Serum biomarker measurements

Nurse clinicians from the Department of Neurology at Puerta del Mar University Hospital collected blood samples from each participant via venipuncture. All blood samples were collected into clot-activating serum separator tubes, allowed to clot at room temperature for 30 minutes, and subsequently subjected to centrifugation at 1500 × g for 10 minutes to separate the serum from the whole blood. The resulting serum was divided into aliquots and appropriately stored at −80 °C until analysis. Blood was drawn from each participant for subsequent quantification of the serum NfL, GFAP, and BDNF protein levels. All measurements were assessed via single-molecule array (Simoa) technology with an SR-X instrument (Quanterix Corporation, MA, USA). The detailed instructions can be found in the Simoa Homebrew Assay Development Guide. Serum levels of NfL and GFAP were quantified with the neurology 2-plex B assay (Quanterix Corporation), whereas BDNF levels were determined via a BDNF Discovery Kit (Quanterix Corporation). The data were analyzed in accordance with the manufacturer’s instructions and recommendations for assessing blood biomarkers by experienced biologists. The interassay coefficient of variation was less than 17.5% for NfL, 10.5% for GFAP, and 25% for BDNF. The intra-assay coefficients of variation were calculated to be approximately <11%, <13.5% and <11% for NfL, GFAP, and BDNF, respectively. Blood samples were collected at the four aforementioned time points for PD-nDBS patients, whereas for the remaining cohorts, they were collected only once. Blood samples were always collected during the on-state for each of the three PD groups.

Statistical analysis

Statistical analyses were performed with SPSS v.24 (IBM, NY, USA) and customized SPSS syntax routines. The normality of the data distributions for all outcome variables was evaluated via the Shapiro‒Wilk test, which determines which parametric and nonparametric statistical methods should be used for each measure. One-way analysis of variance (ANOVA), a paired t test, an independent-sample t test, the Pearson χ² test, the Wilcoxon signed-rank test, the Mann–Whitney U test, Spearman’s rank correlation coefficient, the Kruskal‒Wallis test and the Friedman test with Dunn’s post hoc test were used as appropriate. To compare the serum biomarker levels among different PD therapy intervention groups, we conducted an analysis of covariance (ANCOVA) for each biomarker, with group as a fixed factor and age, disease duration, and cognitive status, as defined by the MMP score, as covariates. Partial correlation coefficients, adjusted for the covariates age, disease duration and cognitive status, were calculated between serum biomarker levels and the main clinical outcomes, and the predictive value of serum biomarkers for monitoring motor improvement after STN-DBS treatment among PD patients was also explored. All tests were 2-tailed with a significance level set at p ≤ 0.05.

Author contributions

J.J.G.-R. conceived, designed, and directed the project. J.J.G.-R., F.S., and F.C.-C. performed the analysis, drafted the manuscript, and designed the figures. R.R.-L. and R.E.-R. performed clinical evaluations of patients and conducted patient testing during and after the surgery. R.R.-L., R.E.-R., F.E.-S., P.M.-G., F.L.S.-F., and F.L.-S. recruited the patients. A.J.C.-G., E.L.-S., and P.M.-G. conducted the neuropsychological assessments and analyzed the neuropsychological data. J.R.-D. performed the DBS surgeries. F.C.-C. and L.G.-J. performed the acquisition, storage, and analysis of blood samples to obtain serum biomarker levels. J.J.G.-R., F.S., F.C.-C., R.R.-L., and F.E.-S. contributed to the interpretation of the results and edited and revised the manuscript. All the authors have read and approved the final manuscript. F.S. and F.C.-C. share first authorship.

Data availability

The anonymized dataset that supports the findings of this study is available from the corresponding author upon reasonable request.

Competing interests

RR-L and RE-R have received speaker fees from Teva, AbbVie, Zambon, BIAL, and Italfarmaco. FES has received payment or honoraria for lectures, presentations, speaker bureaus and educational events from AbbVie, Bial, Boston Scientific, Esteve, Medtronic, Stada and Zambon. RR-L, JR-D, and RE-R received travel support and training from Medtronic, but this support was not related to this study. The remaining authors state that no commercial or financial relationships that might be considered potential conflicts of interest existed during the research.