A multicenter longitudinal study of cholinergic subgroups in Parkinson disease

Abstract

Parkinson disease (PD) is a heterogeneous syndrome. There is a need for biology-driven subtyping to inform specific therapeutic strategies. In a two-center study with de novo and established PD cohorts, we use vesicular acetylcholine transporter ligand [¹⁸F]FEOBV brain PET to assess cholinergic systems changes in early to moderate PD. Principal component analysis (PCA) is applied to data from 245 PD subjects to define cholinergic subgroups at baseline. Three PD subgroups are identified: hypercholinergic (regional upregulation; 29%), mixed (regional upregulation and regional deficits; 40.8%) and hypocholinergic (regional deficits only; 30.2%). Evidence of upregulation is observed in the subcortical-anterior cortical regions, whereas cholinergic downregulation is found in posterior cortical regions. Cholinergic upregulation and downregulation exhibit distinct associations with clinical symptoms. Longitudinal analysis (2-3 year interval) in 128 PD subjects reveals differential progressions by subgroup. This subtyping approach expands understanding of cholinergic progression in PD and may inform identification of new therapeutic targets.

Fig. 1. Schematic representation of cholinergic subgroup biology-based classification algorithm.

Step 1 involves the Z-scoring of raw patient data (DVR, distribution volume ratio) relative to controls, with adjustment for effects of age and sex among controls (white: normative level, red: higher than normative, blue: lower than normative). Step 2 involves running a principal component analysis (PCA) on control and patient data, to obtain principal component scores. Step 3 involves using principal component scores as obtained among control participants to define normative component score cutoffs for hypo-cholinergic (5th percentile) and hyper-cholinergic (95th percentile) label assignment. Step 4 involves assigning normative labels to each patient’s cholinergic system principal components based on normative cutoffs obtained from control participants in the previous step. Step 5 involves using the pattern of principal component labels obtained for each patient to assign them into a cholinergic subgroup with a rule-based classification heuristic. Created in BioRender. Roytman, S. (2025) https://BioRender.com/jk21glb. DVR distribution volume ratio, PCA principal component analysis.

Fig. 2. Main voxel-based SPM analysis comparing each of the baseline cholinergic subgroups against control participants with independent samples t tests.

Statistical t-contrast images were thresholded with false discovery rate (FDR) threshold of P < 0.05. Positive contrast values (red color scale) indicate statistically significant upregulation of [¹⁸F]FEOBV uptake in subgroups relative to control subjects, with negative values indicating statistically significant deficits (blue color scale).

Fig. 3. Main voxel-based SPM analysis comparing each subgroup within-subject at follow-up visits relative to baseline visits with paired samples t tests (adjusted for days between visits, age, and scanner).

Statistical contrast images (positive and negative) were thresholded with false discovery rate (FDR) correction at P < 0.05. Negative (blue) contrast values indicate statistically significant downregulation of [¹⁸F]FEOBV uptake in a given region relative to values observed at baseline.