Decreased Cerebrospinal Fluid Aβ₄₂ in Patients with Idiopathic Parkinson’s Disease and White Matter Lesions

Cerebral small vessel disease (SVD) as well as Parkinson’s disease (PD) are common disorders with increasing prevalence with aging [1, 2]. Sporadic, non-amyloidal, vascular risk factor-related SVD is associated with pathological findings of distal arteriolosclerosis, fibrinoid necrosis and lipohyalinosis [3]. Small vessels cannot be visualized in vivo, and the consequences of their changes in brain parenchyma including white matter lesions (WMLs), lacunar infarcts, perivascular spaces and microbleeds, have been adopted as markers of SVD [4]. WMLs are present in over 50% of healthy elderly individuals [5], and are correlated with decline in cognitive and motor performance [3]. Hypertension [6] and diabetes [7] are the strongest, modifiable risk factors associated with WMLs, but studies on the effect blood pressure (BP) lowering treatment on cognitive outcome have given mixed results [8]. In PD, WMLs have mostly been studied in regard to cognition, and have shown that high burden and progression of WMLs are associated with increased risk for development and progression of impairment of memory and executive function [9, 10]. However, in an earlier cross sectional study in drug-naive PD patients no similar correlation was found [11]. In a diffusion tensor imaging study in early PD, increased white matter diffusivity correlated with cognitive task performance, and was suggested to possibly precede grey matter volume loss [12]. Also, comorbid WMLs have been reported to be a greater determinant of axial motor symptoms in PD than the degree of nigrostriatal dopaminergic denervation [13]. The role of vascular risk factors in PD risk is not clear, and conflicting results have been published on its association with cholesterol [14–16], hypertension [17, 18], and diabetes [19].

Cerebrospinal fluid (CSF) amyloid-β_1–42 (Aβ₄₂), a proxy biomarker for Alzheimer’s dementia (AD), has recently been studied in older individuals free of stroke and dementia, and shown to be associated with WMLs volume, as did CSF neurofilament light levels, a marker of axonal damage [20]. In PD, low CSF Aβ₄₂ has been associated with postural instability/gait difficulties motor phenotype [21] and with increased risk for early onset of cognitive impairment [22, 23]. In a recent study on CSF Aβ₄₂ and WMLs in PD, low levels of CSF Aβ₄₂ were associated with moderate-to-severe WMLs and with progression to dementia at 18 months [24]. To date, there are no other studies investigating the inter-correlation of CSF Aβ₄₂ and WMLs in PD.

The aim of this study is to investigate the frequency and grade of WMLs in PD patients, their association with clinical evaluation of motor and non-motor symptoms, and with CSF markers of neurodegeneration.

Our study is a cohort of patients with parkinsonism recruited consecutively in the Movement Disorders Clinic of Karolinska University Hospital Huddinge since 2011. Clinical evaluation is performed at baseline and every 5 years as part of the routine visit, and includes self-evaluation questionnaires, including Non-Motor Symptoms Questionnaire (NMSQuest) [25], and Hospital Anxiety and Depression scale (HADS) [26]. Motor symptoms are evaluated with the Unified Parkinson’s Disease Scale (UPDRS) part 3 in the ON medication state. Non-motor and motor symptoms impact in daily activities are evaluated with UPDRS part 1 and part 2, respectively. UPDRS part 4 is used to measure motor complications [27]. Montreal Cognitive Assessment (MoCA) is used for the cognitive evaluation [28]. Lumbar puncture and magnetic resonance imaging (MRI) or computed tomography of the brain are performed during the diagnostic evaluation.

In the present study, we identified 100 patients with Idiopathic PD that had undergone brain MRI with FLAIR sequences available for investigation by an experienced neuroradiologist (SK) who graded WMLs according to Fazekas scale [29]. Thirty patients had undergone two brain MRIs at different time-points. Vascular risk factors including smoking, hypertension, diabetes type 2, atrial fibrillation, heart insufficiency and hypercholesterolemia were considered present when mentioned in the patient’s medical records, and/or patients were on ongoing treatment. Clinical evaluation including UPDRS, MoCA, NMSQuest, HADS were performed at one time-point. During the last two years, we have replaced UPDRS with the Movement Disorder Society (MDS)-UPDRS [30]. For that reason, we report MDS-UPDRS parts 2 and 3, after conversion of 65 values measured with UPDRS with a previously calibrated formula [31].

Of 100 patients, 50 underwent lumbar puncture and CSF levels of Aβ₄₂, tau and phospho-tau were measured at Karolinska University Laboratory with a solid-phase enzyme immunoassay (Innotest^®, Fujirebio). Cut-off value for Aβ₄₂ is >550 ng/L, for phospho-tau 60 and 80 ng/L for ages 20–59 and >59 years, respectively, and for tau 250, 300, and 400 ng/L for ages <18, 18–44, and >44 years.

Data on age, sex, presence of vascular risk factors and Hoehn and Yahr stage were complete, four patients had missing data on plasma glucose values and MoCA score, seven patients had missing data on UPDRS part 2, six on UPDRS part 3, and 12 patients had missing data on NMSQuest and HADS scores.

The Local Ethics Committee has approved the study (2016/19-31/12), and all patients have given written informed consent for participation.

Of 100 patients included in the analysis, 86 had WMLs, 58 punctate, and 28 early confluent/confluent. Age and frequency of history of hypertension and smoking were increasing with increased WMLs severity (Table 1). Vascular risk factors were present in 43 patients; 18 (64%) of those with early confluent/confluent WMLs, 25 (43%) with punctate WMLs and 3 (21%) with no WMLs had > = 1 vascular risk factors (not shown). MDS-UPDRS part 3 and Hoehn and Yahr stage were increasing and MoCA score was decreasing with increasing WML severity (Table 1). A moderate correlation between WML severity and age (rho = 0.53; p < 0.0001), and weak correlations with MDS-UPDRS part 3 (rho = 0.25; p = 0.01), Hoehn and Yahr (rho = 0.22; p = 0.03), and MoCA (rho = –0.31; p = 0.002) were observed. Frontal localization was more common in patients with punctate WMLs, whereas fronto-parietal and periventricular lesions were more common in those with early confluent/confluent lesions (Table 1).

Low MoCA-score was observed in none (0%), 10 (18%) and seven (26%) of patients with no, punctate and early confluent/confluent WMLs, respectively (p = 0.04; Table 1). Localization of early confluent/confluent WMLs did not differ significantly between patients with low vs. high MoCA; periventricular lesions were somewhat more frequent in those with low MoCA score (Supplementary Table 1).

Ordered logistic regression analysis was performed in the entire cohort and only age was an independent predictor of WMLs severity (OR 1.1; 95% CI 1.04–1.16; p < 0.0001), in the multivariate model adjusted for hypertension, Hoehn and Yahr stage, MDS-UPDRS part 3, and MoCA score.

Of 50 patients with CSF measurements, 5 (10%) had no WMLs, 30 (60%) had punctate WMLs, and 15 (30%) had early confluent/confluent WMLs. CSF Aβ₄₂ levels were lower in patients with early confluent/confluent vs. those with no and punctate WMLs (Table 1). This difference was observed also when investigated separately in deep white matter and in periventricular regions (Supplementary Table 2). Similar difference was also observed both in high and low-MoCA subgroups (Supplementary Table 3). CSF tau and phospho-tau levels did not differ overall (Table 1), neither in subgroups with low and high MoCA, nor in subgroups with WMLs in different regions (Supplementary Tables 2 and 3).

Logistic regression analysis was applied in the group of patients with available CSF analyses, and age (OR 1.13; 95% CI 1.007–1.28; p = 0.04) and Aβ₄₂ (OR 0.62; 95% for every 100 ng/L change; 95% CI 0.44–0.88; p = 0.007) were significant predictors of WML severity (no/punctate vs. early confluent/confluent) in the multivariate model adjusted for hypertension. ROC analysis of the model showed excellent prediction accuracy (Area under ROC 0.9; 95% CI 0.8–1; Fig. 1).

Of 30 patients with repeated MRIs, 14 (47%) showed progression of WMLs, of whom 10 had punctate, and four had early confluent/confluent WMLs in the initial examination. Median time between MRIs did not differ between those with and without progression (68 months, IQR 55.4, and 62.5 months, IQR 43.3 respectively, p = 0.9; not shown).

In our study, WMLs were very common in patients with PD, and were associated with increasing age and lower CSF Aβ₄₂ levels.

WMLs were present in 85% of the patients in our PD cohort, which is in accordance with the reported frequency (82.5%) in a previous study on a Caucasian population of similar age [36], and with population-based studies reporting frequencies between 50% and 98%, increasing with age [37]. Age was a strong, independent predictor of WML severity, in our report. We also observed higher frequency of hypertension, lower MoCA, and higher MDS-UPDRS motor scores in patients with higher WML severity, however no independent association was confirmed, presumably due to lack of power. In a previous longitudinal, community-based study, WML progression was associated with increased BP and BP fluctuations [38]. Also, studies on motor and cognitive performance have previously reported an independent association with WML severity and progression [13, 39].

In regard to WML localization, periventricular lesions were twice more common in patients with cognitive impairment (MoCA < 21), which is in line with previous studies that have shown periventricular WMLs to be associated with dementia, and deep white matter lesions with depression [40]. In contrast to the study of Compta et al. [24] where parieto-occipital WMLs correlated with the presence of and progression to dementia, we did not find differences in the lobar localization of WMLs. Difference in the WML-scoring system applied may explain this discrepancy. Also, in our study almost all patients had frontal, deep white matter lesions, either alone (more common in those with punctate WMLs) or in combination with other lobar regions and periventricular white matter (more prevalent in those with advanced lesions). These findings may suggest that WMLs first appear at the frontal deep white matter and spread deeper and more posteriorly with advancing disease, partly in line with a population-based study reporting that WMLs evolve from frontal, periventricular regions to deep white matter and more posterior areas [41].

In the subgroup of patients with CSF measurements, we found a strong association between low CSF Aβ₄₂ levels and early confluent/confluent WMLs overall, and in periventricular and deep white matter regions separately, as well as in both low- and high-MoCA subgroups. On the contrary, in the study of Compta [24], CSF Aβ₄₂ was lower only in PD patients with moderate/severe WMLs with parieto-occipital localization, but not in other regions, neither in the subgroup of patients with PD dementia. Differences in variable definitions, as well as the small number of patients in both studies impede the interpretation of differences in subgroup-level. In our study, CSF Aβ₄₂ level contributed significantly in the multivariate model for prediction between no/punctate vs. early confluent/confluent WMLs. Previous studies in healthy elderly with subjective memory complains [42], and on subjects with cognition ranging from normal to AD including a subgroup with PD [43], have shown an independent association between WMLs and different isoforms of CSF Aβ levels. That repeatedly consistent observation may reflect Aβ deposition in the vessel walls (mostly Aβ₃₈ and Aβ₄₀) and in senile plaques (mostly Aβ₄₂) [44], that in turn may lead to ischemic WMLs due to vessel-wall changes and obstruction. That, together with decreased amyloid perivascular drainage may further accelerate amyloid deposition. Also, WMLs may be associated with decreased production of Aβ in the CSF due to reduced axonal transport of the precursor protein in the ischemic regions and subsequent decreased substrate levels available to the α- and β-secretase activity, as suggested in a stroke study [45]. In line with Compta et al. [24], we found no significant difference of CSF tau levels between groups of different WML severity. In AD, a possible interaction of tau with vascular pathology and WMLs was suggested to contribute to disease progression [46], which is also strengthened by imaging and neuropathological studies [47–49].

In a small subgroup of patients, we found that more than half of those with WML progression had punctate WMLs initially. Previous studies have shown that individuals with no or punctate WMLs have very low tendency for progression, whereas those with early and confluent lesions progress rapidly [50], which has also been confirmed in patients with AD, PD and Lewy body dementia with comorbid WMLs [51]. Comparisons with our results are however complicated by the small number of patients and unequal follow-up time.

Our study has limitations. Information bias may be present; however, all medical records are collected digitally in a patient management system common for the majority of health care providers in Stockholm. Also, patients’ cognitive status ranged from normal to dementia, and disease duration and severity varied, and subgroup analyses were limited by the small number of patients. Finally, the study design does not allow conclusions on possible causative or synergistic relation between WMLs and Aβ₄₂. However, our study is based on a well-described, representative cohort, that confirms previously reported findings and adds further evidence on a possible role of contributing pathologies, beyond that of synuclein, in PD.

Future studies on the temporal evolution of WMLs in association with CSF levels of Aβ isoforms and amyloid neuroimaging in brain parenchyma and vessels would shed further light on the interplay between WMLs and amyloid pathology in PD.

Financial support for this work has been received from the ALF program of Stockholm County Council.

Per Svenningsson is a Wallenberg Clinical Scholar.

The supplementary material is available in the electronic version of this article: https://doi.org/10.3233/JPD-181486.