Adults with type 2 diabetes had lower scores on the MoCA cognitive test than healthy controls in a comparison involving 49 patients and 43 controls; DST and CDT scores did not differ significantly. The scans showed lower blood flow in both hippocampi, both anterior thalami and the right amygdala, along with weaker resting neurovascular coupling in the left hippocampus and left amygdala.
The pattern varied by measure and region. Right anterior thalamic volume was smaller in the diabetes group, while ReHo, a measure of how similarly nearby brain areas fluctuate, was lower in the right anterior thalamus and left amygdala. Two other functional measures, fALFF and DC, showed no significant group differences.
How the researchers measured it
Participants completed MoCA, DST and CDT cognitive tests and underwent 3T MRI that combined high-resolution T1-weighted imaging, 3D-ASL and resting-state fMRI. The researchers calculated regional volume, cerebral blood flow, fALFF, ReHo, DC and resting-state NVC. Here, NVC meant the correlation between blood flow and fALFF across the small image units, or voxels, within each region. In ordinary terms, it measured how closely local blood supply and spontaneous MRI-signal fluctuations varied together at rest.
Blood sugar links inside the diabetes group
Among the participants with diabetes, MoCA scores were negatively associated with fasting blood glucose (r = -0.405, P = 0.005) and positively associated with right thalamic volume (r = 0.311, P = 0.035). HbA1c and fasting blood glucose were each negatively associated with blood flow in the left anterior thalamus (r = -0.344, P = 0.019, and r = -0.367, P = 0.012, respectively).
By contrast, no significant group differences were detected in the reported volume, blood-flow, ReHo or NVC measures for the caudate, putamen, globus pallidus or nucleus accumbens. The null result is limited to the structures and measures examined and does not establish that the basal ganglia are biologically unaffected.
Why the findings need caution
For the imaging comparisons, the researchers used adjusted statistical models that accounted for demographic factors and, for functional measures, head motion. They applied Benjamini-Hochberg false discovery rate correction and treated corrected results below 0.05 as significant; the within-diabetes correlations used the same covariates.
These results came from a cross-sectional comparison, meaning the participants were assessed at one point rather than followed over time. The design can show associations, but it cannot show that type 2 diabetes caused the MRI differences or cognitive scores, or establish which change came first. The sample was limited, and the researchers did not perform subgroup analyses by disease duration or medication use.
Technical limits also matter. The study noted that the ASL and fMRI resolution could introduce partial-volume effects, in which measurements from neighboring tissues blur together. Because NVC was defined as a correlation, it describes how two MRI-derived signals were related in these data; it does not by itself establish cause and effect.
Funding and disclosures
The work was funded by the Scientific Research Project of Guangdong Provincial Bureau of Traditional Chinese Medicine and the Zhanjiang Science and Technology Project. One author, identified as WL, was employed by GE Healthcare; the remaining authors declared no potentially conflicting commercial or financial relationships. The authors said raw data supporting the conclusions would be made available without undue reservation.
Paper data and sources
Original title: Neurovascular uncoupling in type 2 diabetes mellitus: multimodal MRI reveals the association between hippocampal and subcortical structures dysfunction and cognitive decline
Authors: Chen W, Li M, Liu W et al.
Journal/Repository: Frontiers in neurology
Status: Peer-reviewed
First online: 2026-10-19
DOI: Not available
Original paper · Full text