Introduction

Up to 80 % of Parkinson disease (PD) patients develop dementia during the course of their illness [1]. Several baseline factors have been associated with more rapid cognitive decline in PD including advanced age at time of diagnosis, male gender, and education level [2]. The degenerative process may begin years before the first signs of cognitive impairment or dementia become apparent in both Alzheimer’s disease (AD) and PD, and these changes may be reflected in CSF marker profile [3, 4]. Additionally, evidence has shown that lower CSF levels of β-amyloid predict faster cognitive decline among PD patients [5–7]. A study from the Norwegian ParkWest project, a population-based cohort study of newly diagnosed PD patients (mean disease duration of 2.2 years), found that CSF β-amyloid 1-42 (Aβ1-42) was associated with memory decline at the time of patient diagnosis [8]. Later analysis from the same cohort found that low CSF Aβ1-42 at baseline was associated with early dementia associated with PD (median time to dementia: 4.3 years) [9].

The Parkinson’s Progression Markers Initiative (PPMI) is an international, multicenter prospective study following PD cases and controls with baseline CSF sampling and biannual neurological and cognitive evaluations [10]. A previous cross-sectional design study using partial PPMI data showed slight but significantly lower levels of CSF Aβ1-42, α-synuclein (α-syn), total tau (T-tau), and tau phosphorylated at threonine 181 (P-tau181) in PD compared with healthy controls, with Aβ1-42 and P-tau181 remaining lower than healthy control data after controlling for age, gender, and education [11]. We examined whether CSF concentration of Aβ1-42, α-syn, T-tau, and P-tau181 at baseline predicts diagnosis of cognitive impairment at 2-year follow-up among newly diagnosed, drug-naïve PD patients without cognitive impairment.

Methods

Participants

Data were obtained from the PPMI database (www.ppmi-info.org/data) on August 10, 2015 [10]. Participants were included in this PD cohort if they enrolled in the study within 2 years of diagnosis and were not expected to require PD medication within 6 months of their baseline evaluation. Participants were excluded if they had a clinical diagnosis of dementia, were unable to participate in lumbar puncture, or had MRI evidence of another clinically significant neurological disorder. Scans without evidence for dopaminergic deficit (SWEDD) participants and control participants were not included in this study.

Individuals from the PD cohort who had 2 years of follow-up were included in the analysis. At the time of data acquisition, 341 PD participants (of original 406 with baseline data) had completed 2 years of study follow-up. Written informed consent was obtained from all participants, and all PPMI sites received approval from their respective ethics committee on human experimentation prior to study initiation.

Assessments

The annual assessment included six cognitive tests, which can be divided into four cognitive domains: memory (Hopkins Verbal Learning Test-Revised [HVLT-R] Recall, HVLT-R Recognition Discrimination) [12], visuospatial (Judgment of Line Orientation [JOLO]) [13], working memory-executive function (Letter Number Sequencing [LNS], Semantic Fluency) [14, 15], and attention-processing speed (Symbol Digit Modalities Test [SDMT]) [16]. Cognitive impairment was defined as having at least 2 test scores (of six; irrespective of test domain) greater than 1.5 standard deviation below the age- and education-standardized mean score based on published norms in healthy controls [17].

CSF Analysis

CSF samples were collected from all participants enrolled in the study at baseline including Aβ1-42, α-syn, T-tau, and P-tau181. Additional information on how CSF samples were collected and analyzed was previously reported [11].

Statistical Analysis

Data were analyzed using SAS v 9.3. Participants with baseline cognitive impairment by the aforementioned definition were not included in future analyses. Baseline demographics (age, gender, PD duration from diagnosis, education, and 15-item Geriatric Depression Scale [GDS-15]), disease characteristics (Movement Disorder Society-Unified Parkinson’s Disease Rating Scale [MDS-UPDRS] motor score part 3, MDS-UPDRS part 1 Cognitive Impairment self-report at baseline, and MDS-UPDRS part 1 Cognitive Impairment self-report at 2 years), and baseline CSF marker levels and their ratios (α-syn/T-tau, T-tau/Aβ1-42, P-tau181/Aβ1-42, P-tau181/T-tau) were compared between participants with and without cognitive impairment (CI) at 2-year follow-up using either the Student t test or the chi-square test. Non-normal data were compared using the Wilcoxon rank sum test. Multivariable logistic regression was used to determine the association of baseline CSF measures and their ratios to CI at 2 years. The adjusted model included baseline age, gender, disease duration, education, MDS-UPDRS part 3 motor score, and Geriatric Depression Scale score. Significance level for all non-ratio tests was set at p < 0.05 (2-tailed test). Adjustment for multiple comparisons was used for all ratio tests with a Bonferroni-Holm correction.

Results

Participant Characteristics

Two-year follow-up data was available for 341 participants out of 406 assessed at baseline. There were no significant differences in baseline characteristics or biomarker profiles between those with 2 years of follow-up and those without 2 years of follow-up (Supplemental Table 1). Fifty-five participants had CI at baseline (16.1 %) and were not included in future analyses. Those with CI at baseline had greater motor impairment reflected by higher MDS-UPDRS motor scores (24.6 vs. 20.3, p < 0.01), had higher scores on depression rating scale (5.7 vs. 5.2, p = 0.04), and had a higher proportion of self-reported cognitive decline at baseline (43.6 vs. 23.4 %, p < 0.01) (Supplemental Table 2).

Of the 286 remaining participants without CI at baseline, 37 (12.9 %) developed CI by year 2. Demographics and disease characteristics were similar between those with and without CI at 2 years (Table 1). Of note, in this cohort baseline self-reported cognitive changes on the MDS-UPDRS were not associated with later development of CI; however, at year 2 evaluation, those with CI were more likely to report cognitive changes during that evaluation (Table 1).

Table 1 Comparison of demographics and Parkinson’s disease characteristics between participants with and without cognitive impairment at 2 years
Full size table

CSF Markers as Predictor of CI

Participants with CI at 2 years had significantly lower mean baseline CSF Aβ1-42 levels than those who did not (343.8 vs. 380.4 pg/mL, p < 0.01). No significant differences in CSF level were noted for α-syn, T-tau, or P-tau181. Significant differences were not noted in the ratio of CSF T-tau/Aβ1-42, T-tau/α-syn, CSF P-tau1181/Aβ1-42, or P-tau181/T-tau (Table 1).

Logistic regression was used to test whether baseline CSF markers could predict CI at 2 years, adjusting for baseline demographic characteristics (Table 2). After adjustment for age, education, disease duration, gender, MDS-UPDRS motor score, and Geriatric Depression Score, CSF Aβ1-42 still predicted CI at 2 years. α-syn, T-tau, and P-tau181 were not predictive of CI at 2 years in the univariate or multivariable models. CSF ratios of α-syn/T-tau, T-tau/Aβ1-42, P-tau-181/Aβ1-42, and P-tau-181/total tau were also not associated with CI at 2 years in multivariable models.

Table 2 The association between cognitive impairment (outcome) and CSF markers (predictors) in univariate and multivariable regression models [OR for decrease in CSF marker levels]
Full size table

Discussion

In the present study of a medication-naïve PD cohort with 2 years follow-up, we found that lower baseline CSF Aβ1-42 was associated with higher odds of having CI after 2 years of follow-up after adjusting for baseline demographic characteristics. To our knowledge, this is the first study to demonstrate an association between lower baseline CSF Aβ1-42 levels in medication-naïve newly diagnosed PD cases without cognitive impairment and later development of CI as defined by neuropsychological testing. This finding is consistent with findings from the Norwegian ParkWest study, which reported an association between lower levels of CSF Aβ1-42 and cognitive deficits in the domain of memory [8], and later development of PD dementia [9]. It is also consistent with studies of more advanced PD cohorts associating lower CSF Aβ1-42 with development of dementia [4, 7, 18].

That lower CSF Aβ1-42 among medication-naïve early PD cases predicts CI 2 years later argues that the neuropathological process that leads to cognitive impairment in these cases may have started before or in parallel to motor symptom onset. However, the neuropathological correlates of lower CSF Aβ1-42 are not clear. In AD, it is hypothesized that low CSF Aβ1-42 may reflect higher levels of aggregation of Aβ protein in the cortex, signifying a more advanced neurodegenerative process [19–21]. Whether those with low levels of CSF Aβ1-42 may represent a subset of PD with coincident AD remains to be seen and may be elucidated with further follow-up (including postmortem neuropathological evaluation) [22, 23].

Strengths of the present study include finding a robust relationship between CSF markers and CI in a multicenter longitudinal study including a large sample size of participants with newly diagnosed PD without CI. Limitations include a definition of CI dependent only on neuropsychological test scores without assessment of participant’s functional impairment [24], which was not available at time of participant enrollment. Taken together, this study provides evidence that CSF Aβ1-42 may play a role in prediction of early cognitive impairment among persons with newly diagnosed PD with mild motor symptoms. Early identification of markers of cognitive impairment is important for determining which patients may benefit most from therapies once they become available. Further longitudinal analysis following changes in CSF Aβ1-42 as they relate to cognition can be the next step in exploring the role of CSF Aβ1-42 in predicting cognitive impairment and dementia.