Parkinson's disease (PD) is a multifactorial neurodegenerative disorder that affects multiple neurotransmitter systems, including the cholinergic system. Cholinergic denervation varies between patients and is not readily inferred from clinical presentation alone. This study evaluated whether commonly available MRI methods—structural MRI (sMRI) and resting-state functional MRI (fMRI)—are sensitive to cholinergic degeneration related to PD and to cognitive performance, and compared MRI findings with previously reported [18F]FEOBV PET results, which are considered a gold standard for cholinergic imaging.
Participants included 34 PD patients and 10 healthy controls who underwent structural T1-weighted MRI. A subset comprising 14 PD patients and 9 controls also completed resting-state fMRI. From sMRI images, bilateral volumes of the nucleus basalis of Meynert (NBM) were extracted as a structural marker of cholinergic integrity. From fMRI data, functional connectivity between the NBM and cortical regions (NBM-FC) was computed.
To manage the high dimensionality of NBM-FC maps, principal component analysis (PCA) was applied. Stepwise logistic regression models were used to test the ability of NBM-FC components to distinguish PD patients from controls; a separate logistic regression assessed NBM volume for the same classification task. Relations between MRI measures and cognitive performance across several domains were investigated using (stepwise) linear regression. Robustness of the classification and correlation results was evaluated using leave-one-out cross validation (LOOCV) and bootstrapping.
Classification performance favored the functional connectivity measure. NBM-FC discriminated PD patients from controls with an area under the receiver operating characteristic curve (AUC) of 0.84 (95% CI: 0.62–1). NBM volume also performed better than chance but with lower accuracy: AUC 0.75 (95% CI: 0.57–0.93).
In domain-specific analyses, higher NBM-FC was associated with better performance in the attentional domain (r = 0.63; p = .015). Separately, larger NBM volume correlated with better global cognition (r = 0.55; p = .001). These relationships suggest differential links between functional and structural markers of the NBM and distinct cognitive outcomes.
Bootstrapping analyses indicated that only the left-hemisphere NBM volume showed a stable relationship with PD diagnosis and with global cognition among PD patients. For NBM-FC, lower connectivity in specific cortical areas—including the fusiform gyrus, supramarginal gyrus and dorsolateral prefrontal cortex—was associated with PD diagnosis. However, bootstrapping did not reveal a stable NBM-FC pattern associated with attention across resamples.
The authors compared their MRI-derived findings against previously reported results using [18F]FEOBV PET tracer uptake. MRI results were noted to be slightly inferior to the [18F]FEOBV PET outcomes described in the reference chapter, though exact PET performance metrics are reported elsewhere. The current MRI measures nonetheless demonstrated meaningful discrimination and cognitive associations.
Model validation used LOOCV and bootstrapping to probe stability. LOOCV supported the reported discriminative performance of NBM-FC and NBM volume. Bootstrapping identified that left NBM volume associations were the most stable finding, whereas NBM-FC associations—particularly for attentional function—lacked consistent stability across resamples.
This preprint indicates that resting-state fMRI–derived NBM-FC may distinguish PD patients from controls with reasonable accuracy and that structural NBM volume correlates with global cognition. The pattern of lower NBM-FC in regions such as the fusiform gyrus, supramarginal gyrus and dorsolateral prefrontal cortex suggests regionally specific network alterations linked to cholinergic dysfunction in PD.
Although both MRI measures were somewhat less sensitive than previously reported [18F]FEOBV PET results, the authors highlight that MRI is more widely available and less costly than PET. MRI-based markers could therefore offer a pragmatic alternative in research settings or in clinical centers without access to cholinergic PET tracers.
The authors recommend exploring MRI measures in longitudinal designs to determine whether MRI can predict or monitor response to cholinergic-targeted treatments. Such work would test MRI utility as a biomarker for treatment selection and monitoring.
This work is a preprint and has not undergone peer review; the authors explicitly note this status. Sample sizes, particularly for the fMRI subset (14 patients, 9 controls), are modest, which may limit generalizability and statistical power. The source reports that the MRI findings were inferior to [18F]FEOBV PET results, but detailed PET metrics are referenced elsewhere rather than provided in this report. Bootstrapping revealed limited stability for some functional connectivity findings, indicating the need for replication in larger cohorts.
Future research recommended by the authors includes longitudinal studies to assess MRI as a predictor and monitor for cholinergic treatment effects in PD, and further validation against PET-based cholinergic measures.
sMRI and resting-state fMRI measures targeting the NBM detect cholinergic-related differences between PD patients and controls and relate to cognitive performance. NBM-FC showed higher discrimination (AUC 0.84) than NBM volume (AUC 0.75), but both were slightly inferior to previously reported [18F]FEOBV PET results. Given cost and availability advantages, MRI could serve as a pragmatic alternative for assessing cholinergic degeneration in PD, warranting longitudinal validation for use as a biomarker of treatment response.