A current unmet need in Parkinson’s disease (PD) is a biological marker that predicts disease progression from a baseline measurement more accurately than the currently used pure clinical assessment. ^{Reference Espay, Kalia and Gan-Or1} A link between PD and sphingolipid metabolism and lysosomal pathways has been highlighted by numerous clinical and basic science observations, ^{Reference Robak, Jansen, van Rooij, Uitterlinden, Kraaij and Jankovic2,Reference Nguyen, Wong and Ysselstein3} including glucocerebrosidase (GBA) acting as a genetic risk factor for developing PD ^{Reference Chahine, Qiang and Ashbridge4} along with other lysosomal genes (SMPD1), ^{Reference Alcalay, Mallett and Vanderperre5} and reports of altered levels of glucosylceramide (GlcCer) and sphingomyelin (SM) in the substantia nigra (SN) and plasma of patients with PD. ^{Reference Huebecker, Moloney and van der Spoel6,Reference Galper, Dean and Pickford7} Levels of these sphingolipids in plasma and cerebrospinal fluid (CSF) have been correlated with progression of motor and cognitive features of PD. ^{Reference Huh, Park and Chiang8} If replicated in other studies, GlcCer or SM levels could potentially be used to identify patients at risk of more aggressive disease progression. Additionally, these shared pathways could be targeted by therapies influencing sphingolipid production or accumulation.

Similarly, although not specific for PD, neurofilament light-chain (NfL) in CSF and in serum is one of the most promising biomarkers of disease severity and progression and has been suggested as a potential marker to assess response to disease-modifying treatments. ^{Reference Mollenhauer, Dakna and Kruse9}

Previous studies measuring sphingolipids in PD analyzed disease progression mainly based on clinical outcomes, which is influenced by medication and non-PD-related factors. ^{Reference Huh, Park and Chiang8,Reference Lerche, Schulte and Wurster10} In contrast, dopaminergic denervation of the striatum, measured using dopamine transporter (DAT) 123-I Ioflupane (DaTscan©) single-photon emission computed tomography imaging, is a biological readout and potentially more sensitive marker of disease progression in early-stage patients. ^{Reference Simuni, Siderowf and Lasch11}

Here, we examined whether CSF sphingolipids and serum NfL at baseline were associated with the change over time in putaminal dopaminergic denervation in patients with early-stage PD or with their clinical progression assessed with the global composite outcome (GCO), a multidomain clinical measure of disease severity. As secondary outcomes, we evaluated rate of caudate dopaminergic denervation since it has been recently demonstrated that early caudate dopaminergic denervation is associated with faster cognitive decline in this PPMI cohort, ^{Reference Pasquini, Durcan and Wiblin12} and global cognitive performance measured by the Montreal Cognitive Assessment (MoCA).

Participants. Newly diagnosed, unmedicated patients with PD were included from the Parkinson’s Progression Markers Initiative (PPMI) cohort. The PPMI is an ongoing international, multicenter observational cohort study aimed at identifying serological, genetic, spinal fluid, and imaging biomarkers of PD progression. Participants with CSF biomarkers and longitudinal DaTscan© imaging data were included (see supplementary flow diagram 1).

Clinical data. Clinical data were extracted from all available visits. Nonmotor and motor symptoms and motor signs were measured using the Movement Disorders Society endorsed-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS-parts I-III); activities of daily living using the Schwab and England (ADL); global cognition using the MoCA. Clinical progression of disease was assessed by change in the GCO. The GCO weights nonmotor symptoms, motor symptoms, motor signs, overall activities of daily living, and global cognition, standardized by averaging the z-scores of each component. As previously described, ^{Reference Fereshtehnejad, Zeighami, Dagher and Postuma13} we calculated GCO change using the mean/SD from the baseline scores as reference.

Biomarkers. GlcCer and SM were quantified at baseline from CSF samples by liquid chromatography-tandem mass spectrometry (LC–MS/MS) in human CSF (SANOFI, PPMI Project ID: 135). NfL at baseline was measured in serum samples from the PPMI sample collection on the Simoa singleplex NF light assay (University Medical Center Goettingen and Paracelsus Elena-Klinik, Kassel, Germany; PPMI Project ID: 144) (see Supplementary material for detailed storage time).

Presynaptic dopamine transporter (DAT) scan. Image acquisition and post-processing are described in the imaging technical operations manual. Putamen and caudate specific binding ratio (SBR) ipsilateral to the more affected side of the body based on the limb-related items of the MDS-UPDRS-part III were used instead of the contralateral side based on a higher sensitivity to detect change on the least affected (ipsilateral) compared to contralateral putamen. ^{Reference Simuni, Siderowf and Lasch11} DatScan data were collected from available analyzed LONI repository of PPMI, and we used baseline/screening visits and longitudinal follow-up visits at 12, 24, and 48 months (visit http://ppmi-info.org for the technical manual, file PPMI_DatScan_SPECT_Image_Processing_SBR_Calculation_Methods_20130823).

Statistical analysis. The associations between CSF sphingolipids and serum NfL at baseline and the primary and secondary outcomes were estimated using linear mixed models with covariates of time (per year), the log-transformed biomarker, and an interaction term between time and the log-transformed biomarker, adjusted for age, sex, and baseline GCO score. Due to non-normality of residuals in the DaTscan values, they were log-transformed. A two-sided p-value<0.05 was considered significant. All analyses were completed using R version 4.1.1.

Standard protocol approvals, registrations, and patient consents. The study was approved by the institutional review board at each PPMI site. Written informed consent for research was obtained from all participants in the study.

A total of 191 patients with a median follow-up of 2.1 years were analyzed (see table 1 for additional features). Primary outcomes: Unadjusted SBR in the ipsilateral putamen declined by a mean of 7.25% per year. After adjusting for age, gender, and baseline GCO score, no relationships were found between change in ipsilateral putamen SBR and baseline CSF GlcCer (time*log(CSF GlcCer) estimate = 1.00; [95%-CI:0.98, 1.01]; p = 0.62), CSF SM (time*log(CSF SM) estimate = 0.99; [95%-CI:0.98, 1.01]; p = 0.27), GlcCer/SM ratio (time*log(GlcCer/SM ratio) estimate = 1.00; [95%-CI:0.99, 1.02]; p=0.50), and serum NfL (time*log(NfL) estimate = 0.99; [95%-CI:0.97, 1.01]; p = 0.26; supplementary table 1). After adjusting for age, sex, and baseline GCO-score, a higher ratio of CSF GlcCer/SM (time*log(GlcCer/SM ratio) estimate = 1.01; [95%-CI:1.00, 1.02]; p=0.07) and higher serum NfL (time*log(NfL) estimate = 1.02; [95%-CI:1.00, 1.04]; p = 0.08) associated with a greater increase in the GCO-score per year. Secondary outcomes: We found no association between CSF sphingolipids or serum NfL and either MoCA score or ipsilateral caudate SBR (supplementary table 2).

Table 1: Demographic and clinical information, sphingolipid and NfL measurements at baseline

CSF = cerebrospinal fluid; GlcCer = glucosyl ceramide; MDS-UPDRS = movement disorders society endorsed unified Parkinson disease rating scale; MoCA = Montreal cognitive assessment; NfL = neurofilament light chain; SM = sphingomyelin.

The main results of this study are the lack of association of the baseline CSF-sphingolipid biomarkers or serum NfL with dopaminergic imaging markers of disease progression in patients with early-stage iPD over a median follow-up of 2.1 years. However, there was a suggestion (nonstatistically significant) that a higher level of serum NfL (p = 0.08) and higher CSF GlcCer/SM ratio (p = 0.07) would associate with greater clinical progression (GCO score) over time. Therefore, it is possible that either CSF GlcCer/SM ratio or serum NfL could predict the change in GCO score with a larger number of subjects and more homogeneous cohort with respect to symptomatic treatment. This is supported by a relatively higher magnitude of change in GCO score during the first year of follow-up (2.7%), where most of the patients were still untreated for symptoms, compared to smaller changes by the 2^nd and 4^th year (Fig. 1), where a great proportion of the patients were under dopaminergic replacement treatment. The change in the GCO score in the first year represent a true change in underlying disease compared with the subsequent time points. Additionally, inclusion of longitudinal measurements of these biomarkers might be needed to demonstrate significant association with the magnitude of change of GCO.

Figure 1: Longitudinal data represented in boxplots for DaTscan, GCO score, and MoCA.

In secondary analyses, the changes in MoCA and ipsilateral caudate SBR over time were not associated with CSF sphingolipid biomarkers or serum NfL. Our results do not replicate the associations found in a previous analysis of the same cohort ^{Reference Huh, Park and Chiang8} that found that patients in the top quartile of GlcCer/SM ratio at baseline showed a faster decline in MoCA during longitudinal follow-up compared with those in the lowest quartile. The authors did not adjust for disease severity at baseline which may account for the difference, if CSF GlcCer/SM ratio at baseline is associated with more aggressive initial manifestations of the disease. In our analyses, neither sphingolipids nor NfL predicted the change in ipsilateral caudate SBR.

We found that both serum NfL and CSF GlcSer/SM ratio showed association with GCO change over time, but this did not reach statistical significance. In keeping with this, a single-center study that followed 118 patients with PD with longitudinal UPDRS-III showed that higher plasma NfL at baseline predicted faster increase in motor scores and faster decrease in memory scores, but MoCA was not used for these analyses. Additional findings of another study ^{Reference Ye, Locascio and Goodheart14} generated from 376 patients of the same cohort showed an association between higher baseline NfL levels and greater changes in motor scores and DaT imaging over time, but no association with MoCA scores. Although patient selection was different to our study, these discrepancies highlight the substantial variability of results within the same cohort, particularly when combining different subsets of data and contrasting them against different outcomes. It suggests that there is significant interindividual variability in the relationships between the biomarkers studied and motor and cognitive outcomes, possibly reflecting heterogeneity of the underlying biology of PD. Better understanding the determinants of the relationships will be critical for patient stratification for research and clinical trials. Of note, our results also point to the usefulness of the GCO score as an outcome of early-PD progression and patient stratification as previously demonstrated. ^{Reference De Pablo-Fernández, Lees, Holton and Warner15}

Our results have implications for the potential use of these biomarkers as outcomes of disease progression in clinical trials aiming to enroll patients with early PD. As new pharmacological targets for disease modification pathways are defined and tested in patients with clinical diagnosis confirmed by DaTscan, a careful appraisal is needed on evidence related to predicting which patients have a higher biological vulnerability and are prone to a faster decline. Our results stress the need for replication and consistency of evidence across cohorts and analyses within cohorts. Furthermore, it is also possible that PD is sufficiently biologically heterogeneous that CSF sphingolipid biomarkers might be more useful to predict progression in selected subsets of patients that would need to be defined to allow proper enrichment of the study population for a future clinical trial of a potential disease modifying therapy targeting these cellular pathways.

Strengths of our analyses include accounting for baseline disease severity, the choice of a biological outcome (DaTscan) that is less influenced by the effect of dopaminergic replacement therapies and that has been shown to correlate well with part III and total scores of MDS-UPDRS in early stages of the disease, ^{Reference Pasquini, Durcan and Wiblin12} as well as the inclusion of a newer measure of global (motor + nonmotor) disease progression.

Our results do not support an association between baseline CSF sphingolipid measures or serum NfL and striatal DaTscan© SBR, or MoCA scores. However, a potential trend was observed between both CSF GlcCer/SM ratio and serum NfL at baseline and GCO score change over time. The GCO may be a more sensitive marker of clinical progression than MoCA or striatal DaTscan changes alone, particularly in early PD, combining measures of motor, cognitive, and autonomic symptoms. These results support incorporating composite clinical outcomes such as GCO-score in future cohorts.