Heterogeneous cognitive and neuroimaging profiles in older adults with type 2 diabetes: the modifying effect of coronary artery disease

Background Cognitive impairment in older adults with type 2 diabetes mellitus (T2DM) is increasingly recognized, and coronary artery disease (CAD) is also linked to cognitive decline. However, their combined impact remains underexplored. This exploratory study characterized neurocognitive profiles and brain MRI features associated with comorbid CAD in older T2DM patients with cognitive impairment. Methods This cross-sectional study screened 1,217 older adults with T2DM; 523 with cognitive impairment (MoCA < 26) were included. All underwent brain MRI and were stratified by CAD status. Between-group comparisons were performed to examine differences in cognitive and neuroimaging measures, with false discovery rate correction applied for multiple comparisons. Multiple linear regression was used to assess the independent association between CAD and cognitive function. Logistic regression was further performed to evaluate the association between CAD and binary neuroimaging outcomes. Finally, two-step cluster analysis was performed to identify distinct cognitive and brain imaging profiles among these participants. Results Patients with T2DM and CAD exhibited higher frequencies of cerebrovascular markers: lacunar infarcts (74.65% vs. 57.10%, p < 0.001), strategic infarcts (32.86% vs. 19.35%, p = 0.001), and white matter hyperintensities (66.67% vs. 36.77%, p < 0.001), but lower medial temporal lobe atrophy (30.52% vs. 44.52%, p = 0.001). After false discovery rate correction, these differences remained significant. The CAD group scored lower on MoCA total, visuospatial/executive function, attention, and orientation. CAD independently predicted lower global cognition ( B = −0.74, 95% CI: −1.42 to −0.06, p = 0.03). Logistic regression further confirmed that CAD was independently associated with higher odds of lacunar infarcts, strategic infarcts, and white matter hyperintensities (all p < 0.05), but not with medial temporal lobe atrophy ( p = 0.518). Cluster analysis revealed distinct profiles: among non-CAD patients, three subgroups (cerebrovascular-predominant, mixed, atrophy-predominant); among CAD patients, two subgroups (both with significant lacunar infarcts, differentiated by atrophy prominence and delayed recall). Conclusion In older T2DM patients with cognitive impairment, comorbid CAD was independently associated with a vascular-dominant phenotype. Cluster analysis suggested CAD may modify phenotypic heterogeneity, though these exploratory findings require validation. They provide a hypothesis-generating framework for integrating cardiovascular risk management into cognitive treatment.