Discriminating risk and resilience endophenotypes from lifetime illness effects in familial major depressive disorder

Abstract

Importance: The neural systems that confer risk or vulnerability for developing familial depression, and those that protect against or confer resilience to becoming ill, can be disentangled from the effects of prior illness by comparing brain imaging measures in previously ill and never ill persons who have either a high or low familial risk for depression.

Objective: To distinguish risk and resilience endophenotypes for major depression from the effects of prior lifetime illness.

Design, setting, and participants: We used functional magnetic resonance imaging to measure and compare brain function during performance of an attentional, self-regulatory task across a large sample of multigenerational families ascertained specifically to be at either high or low risk for developing major depression. Study procedures were performed in a university setting. A total of 143 community participants were followed up prospectively for more than 20 years in a university setting. The sample was enriched with persons who were at higher or lower familial risk for developing depression based on being biological offspring of either a clinical sample of persons with major depression or a community control sample of persons with no discernible lifetime illness.

Main outcomes and measures: Task-related change in blood oxygen level-dependent functional magnetic resonance imaging signal.

Results: A risk endophenotype included greater activation of cortical attention circuits. A resilience endophenotype included greater activation of the dorsal anterior cingulate cortex. The effects of prior lifetime illness were common to both risk groups and included greater deactivation of default-mode circuits.

Conclusions and relevance: These findings identify neural systems that increase risk for depression, those that protect from illness, and those that endure following illness onset, and they suggest circuits to target for developing novel preventive and therapeutic interventions.

Figure 1. Behavioral Performance on the Simon Task

Reaction times (A) and accuracy scores (B) from each trial across all 10 runs of the Simon task were entered as dependent variables in separate repeated-measures, linear mixed models in SAS (SAS Institute Inc), with risk group (high risk [HR] or low risk [LR]), stimulus congruence (incongruent or congruent), age, sex, and run number (0–10) included as independent variables. The error bars represent standard errors. A, The HR group responded significantly faster, on average, on the incongruent but not congruent trials than did the LR group, but this faster performance when resolving the cognitive interference was driven by members of the HR group who had never had lifetime major depressive disorder or anxiety disorder (Table 2), which suggests that better performance was a resilience effect. B, The HR group, on average, was more accurate than the LR group on incongruent trials, but these group differences were driven by HR participants who had a lifetime history of illness (Table 2).

Figure 2. Hypothesis Testing for Interference-Related Activity

A, Representative 2-dimensional axial slices are shown that demonstrate the effects of interest. The risk effects map identifying brain features for risk was constructed by comparing interference maps for all high-risk (HR) participants with interference maps for all low-risk (LR) participants, while covarying for age, sex, interference effects, prior illness, generation (second or third generation), and genealogical index of familiality and then masking out of this map all voxels where we detected a significant risk group × illness interaction. The resilience effects map was constructed by comparing the interference map for the HR_Healthy group with the map for the LR_Healthy group. We then masked out of this map the voxels where we detected a significant main effect of risk, leaving only resilience (or protective) effects. Maps constructed when including persons with major depressive disorder alone or anxiety disorder alone were very similar (eAppendix 3 and eFigure 3 in the Supplement). The lifetime illness effects map for the main effect of illness was constructed by comparing the interference maps (while covarying for age, sex, interference scores, and risk group) for all persons who had lifetime illness (the HR_Ill and LR_Ill groups combined) with the interference maps for all persons without lifetime illness (the HR_Healthy and LR_Healthy groups combined) and then masking out of this map the voxels where we detected a significant risk group × illness interaction (eFigures 1 and 2, eTable 1, and eAppendix 2 in the Supplement), leaving illness effects that were common to both risk groups. B, Surface renderings of the lateral and medial surfaces of the brain generated using Caret 5.62 are shown, with the locations of each of the risk effects (yellow), resilience effects (green), and illness effects (red) color-coded. ACC indicates anterior cingulate cortex; dACC, dorsal anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; Ins, insula; IPL, inferior parietal lobe; LPFC, lateral prefrontal cortex; MFG, middle frontal gyrus; MTG, middle temporal gyrus; PCC, posterior cingulate cortex; PCu, precuneus; pgACC, pregenual anterior cingulate cortex; PH, parahippocampus; SFG, superior frontal gyrus; SPL, superior parietal lobe; STG, superior temporal gyrus; and vACC, ventral anterior cingulate cortex.

Figure 3. Hypothesis Testing for Error-Related Activity

Maps were constructed as described in Figure 2 for interference-related activity, except the contrast maps on which they were based were those for error-related activity (incongruent incorrect vs incongruent correct trials). All maps were covaried for age, sex, interference scores, generation (second or third generation), and genealogical index of familiality. We did not detect significant risk effects or illness effects in the low-risk (LR) group for error-related activity, and therefore we do not show those maps. Resilience was associated with greater activation (yellow) of ventral and pregenual anterior cingulate cortex and insular cortex in the “HR not ill” group compared with the “LR not ill” group. Lifetime illness effects included reduced activation (blue) of ventral and pregenual portions of the anterior cingulate cortex in the “HR ill” group compared with the “HR not ill” group. HR indicates high risk; Ins, insula; pgACC, pregenual anterior cingulate cortex; and vACC, ventral anterior cingulate cortex.

Figure 4. Interference Effects for Participants Older Than 25 Years

Effects were defined as in Figure 2. Findings are unchanged from those identified in the entire cohort. ACC indicates anterior cingulate cortex; dACC, dorsal anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; Ins, insula; IPL, inferior parietal lobe; LPFC, lateral prefrontal cortex; MFG, middle frontal gyrus; MTG, middle temporal gyrus; PCC, posterior cingulate cortex; PCu, precuneus; pgACC, pregenual anterior cingulate cortex; PH, parahippocampus; SFG, superior frontal gyrus; SPL, superior parietal lobe; STG, superior temporal gyrus; and vACC, ventral anterior cingulate cortex.

Figure 5. Association of Activations With Importance of Religion

The first column shown the comparison of interference-related activations in participants for whom religion was of high importance (n = 23) vs participants for whom religion was of low importance (n = 74), while controlling for age, sex, interference scores, and risk group. The second column shows the risk endophenotype in the selected slices, whereas the third column shows the conjunction of activations in the first and second columns, in effect identifying activations associated with the importance of religion that are located within the risk endophenotype. The fourth column shows the resilience endophenotype in the selected slices, and the fifth column shows the conjunction of the first and fourth columns, identifying activations associated with the importance of religion that are located within the resilience endophenotype (high importance of religion was associated with a more prominent deactivation of the superior frontal gyrus in 2 adjacent slices). A similar conjunction of the first column with the effects of lifetime illness was empty and therefore is not shown. The high importance of religion is associated primarily with reduced activation within regions that constitute the risk endophenotype. dACC, indicates dorsal anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; LPFC, lateral prefrontal cortex; PCC, posterior cingulate cortex; PCu, precuneus; pgACC, pregenual anterior cingulate cortex; PH, parahippocampus; SFG, superior frontal gyrus; SPL, superior parietal lobe; and STG, superior temporal gyrus.