Associations between cerebrospinal fluids total phosphatidylcholines, neurodegeneration, cognitive decline and risk of mild cognitive impairment in the Mayo Clinic Study of Aging

Abstract

It is unclear whether cerebrospinal fluid (CSF) phosphatidylcholines (PC) are associated with neuroimaging measures of amyloid deposition and neurodegeneration (glucose metabolism, cortical thickness, hippocampal volume), cognitive decline, or risk of mild cognitive impairment (MCI) among cognitively unimpaired (CU) older adults. This study investigated the associations of 19 individual CSF PC concentrations and their total sum with cross-sectional and longitudinal measures of amyloid deposition and neurodegeneration, global and domain-specific cognitive z-scores, and risk of MCI among 655 CU participants, mean age of 71 years, enrolled in the Mayo Clinic Study of Aging. Neither the CSF total PC concentration nor individual CSF PCs were cross-sectionally or longitudinally associated with neuroimaging measures, cognition, or risk of MCI.

Introduction

Brain phosphatidylcholines (PCs) are major components of cell membranes; changes in brain cell membranes precede dementia. In addition, CSF PCs play important roles in brain structure and function. For example, PCs are sources of omega-3 fatty acids, such as docosahexaenoic acid (DHA), that are important regulators of brain glucose uptake and can influence cognition. A growing number of studies have assessed the association between plasma PCs and neurodegeneration, cognitive decline, and risk or diagnosis of Alzheimer’s disease (AD) dementia (Li, et al., 2019; Li, et al., 2016a; Mapstone, et al., 2014) Collectively, these studies suggest that plasma PCs are cross-sectionally associated with cognition but are not longitudinally predictive or prognostic of cognitive change or risk of AD dementia. Cerebrospinal fluid (CSF) PCs are more representative of brain PCs than plasma. Previous studies of CSF PCs in the context of brain structure and function were small, cross-sectional, and case-control (Koal, et al., 2015; Kosicek, et al., 2012; Mulder, et al., 1998; Torretta, et al., 2018; Walter, et al., 2004). The objective of this study was to investigate whether CSF PCs, including individual PCs and total PCs, were associated with cross-sectional and longitudinal measures of amyloid deposition and neurodegeneration, global and domain-specific cognitive z-scores, and risk of mild cognitive impairment (MCI) among cognitively unimpaired (CU) participants in the Mayo Clinic Study of Aging (MCSA).

Methods

The MCSA is a population-based epidemiological cognitive aging study of Olmsted County, MN residents (Roberts, et al., 2008). Beginning in 2004, the MCSA enrolled residents aged 70–89 years in a random age- and sex-stratified design. In 2012, residents aged 50 to 69 years were also enrolled using the same sampling design. Participant assessment, including both cognitive and neuroimaging (amyloid-PET, FDG-PET, hippocampal volume) measures, was reported previously (Li, et al., 2019). The present study included 655 participants who were adjudicated as being CU, had available CSF PCs concentrations measured, and cross-sectional and longitudinal cognitive function assessed with an extensive neuropsychological battery that included 9 tests over 4 domains. Among the 655 participants, 574 had brain MRI and 513 had amyloid-PET and FDG-PET. The study was approved by the Mayo Clinic and the Olmsted Medical Center Institutional Review Boards. All participants provided written informed consent.

Fasting lumbar punctures were performed early in the morning in the lateral decubitus position from the L3 and L4 intravertebral space using a 20 or 22 gauge Quincke needle. CSF was divided into 0.5cc aliquots and stored at −80°C. Samples had not undergone a freeze-thaw cycle prior to being pulled for the PC assays. CSF PCs were only measured at one point in time using liquid chromatography electrospray ionization tandem mass spectrometry (LC/ESI/MS/MS) using an AB Sciex 6500 mass spectrometer with an ESI probe and interfaced with an UHPLC system in the positive multiple-reaction monitoring (MRM) mode (see Supplementary Table 1 and Supplementary material).

Linear mixed effect models were used to examine the cross-sectional and longitudinal associations between the 19 CSF PC concentrations (modeled in z-scores) or their sum (total PC concentration), and each cognitive outcome. The univariable models included terms for baseline PC concentration (indicating the cross-sectional association between each PC and outcome), time (indicating change in the outcome over the follow-up), and the interaction between each PC and time (indicating the longitudinal association between each baseline PC concentration and change in outcome). We specified a random intercept and random slope, and used an unstructured covariance matrix. Multivariable models were run in the same manner and adjusted for age, sex, education, medical comorbidity, dyslipidemia, statin use, and APOE. Cox proportional hazards model were used to assess the association between CSF PCs and risk of MCI using age as the timescale. A Bonferroni corrected p-value < 0.0026 (0.05/19) was considered statistically significant. All analyses were performed using SAS (SAS Institute, Cary, NC).

Results

This study included 655 CU participants (43.2% female) with mean (SD) age of 71.0 (9.9) years, education of 14.7 (2.6) years, and follow-up of 4.4 (2.3) years (median[range] = 3.9[0.0, 10.1]; Supplementary Table 2). Among the 655 participants, 574 had brain magnetic resonance imaging (MRI) with a mean (SD) follow up 2.7 (2.4) years. Higher CSF total PC levels were cross-sectionally associated with worse global cognition, language, and attention z-scores and with lower hippocampal volume z-score in univariable analyses (Supplementary Figure 1). However, these results were attenuated and no longer significant in the multivariable analysis. Higher levels of four individual CSF PCs (i.e., PC C36:3, PC C38:0, PC C38:5, PC C16:0_22:6) were associated with lower attention z-scores (Supplementary Table 3). In addition, higher levels of PC C36:5 was cross-sectionally associated with lower hippocampal volume in univariable analysis (Supplementary Table 3). However, all models were attenuated and no longer significant in multivariable analyses (Supplementary Tables 3 and 4). Longitudinally, neither the CSF total PC concentration nor individual CSF PCs were associated with change in global or domain-specific z-scores or any of the neuroimaging measures in univariable or multivariable analyses (Supplementary Tables 5 and 6). In additional analyses, CSF PCs were not associated with risk of MCI (Supplementary Tables 7).

Discussion

Most research examining the relationship between PCs and imaging and cognitive outcomes have focused on plasma PCs. Collectively, these studies suggest that plasma PCs are cross-sectionally associated with cognition but are not longitudinally predictive or prognostic of cognitive change or risk of AD dementia (Li, et al., 2019; Li, et al., 2016a; Li, et al., 2016b; Mapstone, et al., 2014). Because CSF PCs are more representative of brain PCs than plasma, we hypothesized that we would be more likely to observe an association between CSF PCs and neuroimaging and cognitive outcomes. However, findings from this study do not support the utility of CSF PCs as a prognostic marker of cognitive decline.

Despite the negative findings with CSF PCs, this study does not invalidate the studies reporting that plasma PCs are associated with risk of AD dementia. Blood PC levels have been associated with a host of other conditions and diseases in the literature including insulin resistance, diabetes, and cardiovascular disease. Because all of these diseases are associated with an increased risk of Alzheimer’s disease and other dementias, plasma PCs are likely indirect risk factors. Additional research focused on better understanding of the mechanistic and mediation pathways are needed.

The study strengths include the population-based study design, large sample size, and longitudinal analyses of cognitive and neuroimaging measures. Study limitations warrant consideration. First, CSF PC concentrations were measured at one time point so it is unclear whether changes in CSF PCs are related to changes in cognition or neuroimaging. Second, only CU participants were included and it is possible that CSF PCs may be a better prognostic marker at later stages of the disease process, i.e., transition from MCI to dementia. Finally, whether these study findings apply to other race and ethnic groups beyond Caucasian remains unknown. In summary, this study suggests that CSF PCs are not prognostic biomarkers of neurodegeneration, cognitive decline or risk of MCI among older CU individuals.

Highlights.

• We examined associations of 19 CSF PCs with neuroimaging measures and cognition.

• CSF PCs were not cross-sectionally or longitudinally associated with neuroimaging

• CSF PCs were not cross-sectionally or longitudinally associated with cognition.