Brain MRI marker tracks tissue chemistry, not Parkinson’s severity

A widely used MRI measure of brain fluid clearance may be telling researchers less about Parkinson’s disease than they hoped, and more about the tissue it samples. In a cross-sectional study of 76 people with idiopathic Parkinson’s disease, the diffusion tensor image analysis along the perivascular space index, known as DTI-ALPS, did not track clinical severity. Instead, it covaried with the magnetic susceptibility of one small thalamic region and with age, and its behaviour was driven mainly by the denominator of the ratio rather than the perivascular numerator it is named for. The findings appear in Fluids and Barriers of the CNS.

The DTI-ALPS index is popular because it offers a non-invasive window onto the glymphatic system, the network that exchanges fluid along channels surrounding blood vessels in the brain. Its region of interest sits in white matter that carries thalamocortical projections, close to susceptibility-rich deep grey nuclei. That anatomical neighbourhood raises an uncomfortable question: how much of the index reflects fluid movement along perivascular spaces, and how much simply reflects the composition of the tissue being measured?

To find out, the researchers scanned all 76 patients on a 3.0 Tesla scanner with a multimodal protocol and collected 16 clinical scales. Diffusion imaging was single-shell, with a b-value of 2000 s/mm² across 60 directions. From these data they derived the DTI-ALPS index itself, its four directional diffusivity components, bi-tensor free water measures, and total and source-separated susceptibility in 84 deep grey-matter and thalamic regions. Correlations were age-adjusted Spearman partial correlations, with Benjamini–Hochberg false-discovery-rate control applied within each imaging marker rather than across the whole battery. Basal-ganglia free water served as a pre-specified positive control.

The screen produced one striking hit. In the exploratory 84-region survey, the index covaried with total susceptibility of the left centralis lateralis, an intralaminar thalamic nucleus, with a Spearman coefficient of −0.42, a false-discovery-rate-adjusted q of 0.007 and a family-wise permutation p of 0.010. The association was regional rather than global: the correlation with whole-brain susceptibility was essentially zero at ρ = 0.01, and the regional effect persisted when candidate regions were excluded from the calculation, reaching ρ = −0.48 after adjustment. The finding also held up statistically against confounds including atrophy, white-matter free water and visible perivascular spaces.

A regression model sharpened the picture. Age and intralaminar susceptibility, not motor severity, predicted the ALPS index, with the model explaining 40 percent of the variance between patients. The effect was left-lateralised at the ratio level, with the right hemisphere showing a weaker coefficient of −0.25. But the component analysis revealed something more fundamental underneath that asymmetry.

Breaking the index into its four directional diffusivity components showed that the association was denominator-driven. It was carried predominantly by Dyy, the transverse diffusivity in the projection region, which correlated positively with susceptibility in the left hemisphere at ρ = +0.38 (q = 0.004) and at ρ = +0.46 for the bilateral mean, a pattern consistent across hemispheres that accounted for roughly 90 percent of the between-subject variance. The numerator, the term meant to capture perivascular diffusion, was not the driver. In other words, the index moved because the tissue beneath it changed, not because the fluid channel it was designed to probe did.

The clinical side of the ledger came up empty. No association between the ALPS index and any clinical scale survived statistical correction. By contrast, the four free-water compartments yielded nine associations that did, underscoring that the negative result reflects the index’s specificity rather than a lack of statistical power in the design. The study was powered to detect moderate associations, defined as absolute Spearman coefficients of at least 0.32, but not small ones.

The authors are careful about what these cross-sectional associations can and cannot show. They do not establish causation, and the localisation to the left centralis lateralis is exploratory rather than confirmatory. Still, the practical implication is direct: where the composition of deep nuclei is altered, as it can be in Parkinson’s disease, a tissue measure should accompany any DTI-ALPS reading. Interpreting the index as a specific measure of clearance warrants caution.

For a field that has adopted DTI-ALPS rapidly as a proxy for glymphatic function in neurodegenerative disease, the study adds to a growing body of work questioning the measure’s biological specificity. Related cross-sectional and meta-analytic studies in Parkinson’s and Alzheimer’s disease have raised similar concerns. What this new analysis contributes is a mechanistic account of why the index behaves the way it does: its susceptibility-rich anatomical neighbourhood leaks into the measurement, and the ratio’s denominator does most of the leaking. A tissue marker alongside the index may be the minimum requirement for interpreting future DTI-ALPS findings responsibly.

Subject of Research: Biological specificity of the DTI-ALPS index as a glymphatic function proxy in Parkinson's disease

Article Title: The DTI-ALPS index covaries with regional thalamic tissue susceptibility rather than clinical severity in Parkinson’s disease: a cross-sectional multimodal MRI study

Article References: Ren, Z., Wang, R., Shi, D., Wang, G., Huang, H., Li, S., Jia, Y., Yan, G., & Ren, K. (2026). The DTI-ALPS index covaries with regional thalamic tissue susceptibility rather than clinical severity in Parkinson’s disease: a cross-sectional multimodal MRI study. Fluids and Barriers of the CNS. https://doi.org/10.1186/s12987-026-00879-2

Image Credits: AI Generated

DOI: 10.1186/s12987-026-00879-2

Keywords: DTI-ALPS index, Parkinson's disease, glymphatic system, thalamic susceptibility, diffusion MRI, free water imaging, deep grey matter, clinical severity, multimodal MRI, cross-sectional study, perivascular spaces, magnetic susceptibility