People with type 2 diabetes mellitus (T2DM) are at increased risk of cerebral small vessel disease and cardiac structural changes such as left ventricular hypertrophy (LVH), both of which may contribute to cognitive decline and dementia. The Diabetes & Dementia (D2) cohort study aimed to describe 2-year trajectories of brain volume and cognition in people with T2DM and to determine whether baseline LVH causally influences brain atrophy and cognitive decline.
D2 is a multicentre observational cohort performed in Melbourne, Australia. Recruitment occurred between 20 May 2016 and 20 March 2020. Participants were aged over 50 years and were recruited from two hospital outpatient clinics, three private clinics, and study advertisements. Exclusion criteria included pre-existing cognitive impairment, life-limiting medical illness, and severe chronic renal impairment. From 2,378 screened individuals, 702 were eligible, 196 consented, 150 completed baseline assessments, and 123 completed 2-year follow-up with complete MRI, transthoracic echocardiography (TTE), and cognitive data, yielding 17.4% attrition.
Participants attended study visits at baseline and at 2 years that included brain MRI for volumetric measures, transthoracic echocardiography to determine the presence of LVH, and a standardized battery of cognitive tests. The exposure variable for analyses was LVH assessed on baseline TTE. Baseline cognitive status was recorded and used in secondary analyses of trajectories.
The pre-specified primary outcomes were change in total brain volume (TBV) over 2 years and cognitive decline, defined as a z-score change ≤ –1 in any cognitive domain. Regression analyses examined associations between baseline variables and outcomes. To assess causal effects of LVH on outcomes, the authors applied an inverse probability of treatment weighting approach to standardize for confounding covariates; participants with non-positivity on age and baseline TBV were excluded from the causal analysis.
At baseline, LVH was associated with several clinical and demographic factors: female sex, older age, lower educational attainment, lower mood scores, hypertension, obesity, and beta-blocker use. Those with LVH had smaller TBV at baseline compared with those without LVH.
Participants who entered the study with baseline cognitive impairment experienced greater brain atrophy over the 2-year follow-up. Independent predictors of cognitive decline included lower educational attainment, presence of hypertension, and lower baseline cognitive scores.
After applying inverse probability of treatment weighting and excluding participants for non-positivity, the causal analysis sample comprised 62 participants without LVH (20 women; mean [SD] age 66.9 [5.9] years) and 31 participants with LVH (17 women; mean [SD] age 67.4 [5.4] years). In this analysis, baseline LVH was associated with a reduction in TBV change over 2 years: standardized mean difference 6.3 cm3 (95% CI 0.1 to 12.5), P = .048. In other words, LVH was associated with less brain volume loss in this adjusted causal model. The analysis showed no effect of LVH on the pre-specified cognitive decline endpoint over 2 years.
In this cohort of older adults with T2DM, baseline cognitive impairment predicted greater brain atrophy over 2 years, and lower education, hypertension, and lower baseline cognition predicted cognitive decline. Using a causal inference approach, baseline LVH was associated with less TBV change over 2 years, while it had no measurable effect on cognitive decline in this timeframe. The authors interpret these findings to suggest that guideline-directed LVH therapies (for example, beta-blockers) may confer both cardioprotective and possible neuroprotective effects.
The trial was registered (ACTRN12616000546459) and ethics approval was obtained (Human Research Ethics Committee of Austin Health). The authors declared no competing interests. Data, the imaging analysis pipeline, and related study documents are available on reasonable request and deidentified participant data are planned to be shared from January 1, 2027, subject to Steering Committee approval.
Notes on interpretation: all results and conclusions here reflect the reported analyses from the D2 preprint. The causal inference findings applied a weighted observational approach and excluded participants for non-positivity on some covariates; further confirmation in independent cohorts or randomized studies would be required to establish treatment recommendations.