Distinct Dimensions of Depression Are Associated With Different Brain-Related Biomarkers

Management of depression, a potentially modifiable risk factor for dementia, is of significant importance in elderly with type 2 diabetes (T2D), specifically because of their higher risk for depression and for dementia. We recently published in Diabetes Care evidence showing that in older adults (aged ≥65 years) with T2D participating in the Israel Diabetes and Cognitive Decline (IDCD) study, different dimensions of depression were associated with distinct patterns of cognitive outcomes. Specifically, apathy, but not other depression dimensions, was associated with faster decline in executive functions (1).

Here, we wish to strengthen the construct validity of different depression-related dimensions by demonstrating their relationships with different neuropathologies, namely, total gray matter (GM) volume, white matter hyperintensities (WMH), and amyloid burden. Of the 1,338 IDCD participants, a random sample of 209 had a brain MRI, 42 of whom also had an amyloid positron emission tomography scan. Analysis of depression and its dimensions is based on the 15-item version of the Geriatric Depression Scale (GDS-15). The following dimensions of depression were defined as previously described (1,2): dysphoric mood, withdrawal-apathy-vigor (WAV) or apathy, anxiety, hopelessness, and subjective memory complaint.

GM and WMH volumes were examined with a high-resolution 3-Tesla brain MRI (detailed in Livny et al. [3]). Amyloid-β burden was defined as mean standardized uptake value ratios of [¹⁸F]flutemetamol in the frontal, parietal, and temporal cortices compared with the cerebellum as the reference region. Given the likely nonlinear relationship between GDS-15 and brain measures, we used two-step linear regression models. The initial linear regression was used to partial out the contribution, of demographic factors (age, sex, education), time gaps between the brain imaging and GDS-15 assessment dates, and total brain volume, to a particular imaging outcome of interest. We then correlated the residuals with the GDS-15, domains and total, using the Spearman correlation. Results are presented as a two-dimensional heat map showing numerical values in the Spearman correlation matrix by colors. Statistical significance was set at P < 0.05.

Participants’ mean age was 70.55 (SD 4.25) years and 38.5% were female, with mean 13.94 (3.53) years of education. At baseline, mean Mini-Mental State Examination (MMSE) and total GDS-15 scores were 28.20 (1.86) and 1.93 (2.39), respectively. Higher total GDS-15 score was associated with lower GM volume (r = −0.14; P = 0.037) and with higher volume of WMH (r = 0.20; P = 0.003) but not with brain amyloid-β burden (r = −0.13; P = 0.37). Among specific depression dimensions, apathy (WAV) (r = 0.22; P = 0.001) and hopelessness (r = 0.15; P = 0.033) were associated with higher WMH burden, whereas the subjective memory complaints dimension was associated with lower total GM volume (r = −0.15; P = 0.026). Depression dimensions were not associated with amyloid burden (all P values >0.1) (Fig. 1). With further adjustment for metabolic and cardiovascular risk factors (i.e., systolic and diastolic blood pressure, total cholesterol, duration of T2D and mean hemoglobin A_1c), the results for apathy and subjective memory complaint remained significant (r = 0.211, P = 0.003, and r = −0.144, P = 0.042, respectively), but the relationship of the GDS-15 hopelessness dimension with WMH volume was attenuated (r = 0.120; P = 0.094).

In cognitively normal older adults with T2D, higher depression scores, even below the cutoff defined as clinically significant, were associated with lower GM and higher WMH volumes. Importantly, specific depression dimensions varied in their brain correlates. Apathy was associated with higher WMH burden, whereas subjective memory complaints were associated with brain atrophy. There was not enough evidence to suggest associations between depression dimensions and the extent of amyloid-β burden, consistent with previous findings in non-T2D samples, on the role of vascular, rather than Alzheimer disease, pathology in depression (4). This explanation is supported by the consistent role of WMH, markers of small vessel disease, in late-life depression (4) and by the increased risk for small vessel disease in T2D (5). Together with our previous findings (1), our results emphasize the need for prevention of cerebrovascular pathology in T2D patients who are at risk for depression and cognitive decline and suggest apathy as a potential marker to identify those who may be candidates for more vigorous depression and dementia-prevention treatments. The magnitude of the associations observed in IDCD participants, who are devoid of significant affective psychopathology, is relatively small but may become more robust with increasing depression severity. Our results suggest distinct pathology underlying different depression dimensions, pointing to the need for more targeted interventions. A limitation of the study pertains to the extrapolation of the depression dimensions from GDS-30 to GDS-15. Yet, the apathy dimension, where the main results of the current study were found, is the same for both GDS-30 and GDS-15 and is quite consistent over different studies (2).