Resting-State Functional Connectivity and Anxiety Predict Relapse in Remitted Late-Life Depression

Abstract

<h3>Introduction</h3> Late-life depression (LLD) refers to depression occurring in older age and is associated with a multitude of psychosocial and biological factors, including conditions related to pathological aging, including cardiovascular and cerebrovascular diseases, and dementia. Relevant psychosocial factors comprise personality disorders, neuroticism, trauma and stress. Depending on the relative contribution of conditions related to pathological aging versus psychosocial factors, some patients may not respond to antidepressants and others may be prone to relapse even when they achieved initial remission on an antidepressant. The neural mechanisms related to recurrence of LLD are not clear. About 50% of individuals who have recovered from a first episode of depression will suffer from at least one additional episode. Therefore, identifying patients prone to relapse is important for developing tailored treatment regimens, preventing suicide, and reducing economic burdens and other morbidities. To understand the neural mechanisms related to relapse of depression, and identify neuroimaging markers to predict individuals who tend to relapse, we conducted a secondary data analysis on clinical and fMRI data collected from our NBOLD project. We hypothesized that, compared with individuals who did not relapse during a two-year clinical follow-up period, those who relapsed would show significantly altered resting-state activity patterns in core neural networks related to LLD including the default mode, executive control and salience networks. <h3>Methods</h3> The sample included 102 adults with LLD who were free of medication for at least two weeks before study enrollment. Depression severity was measured by MADRS and was assessed by a study psychiatrist at each study visit, which occurred frequently while the subject remained depressed and quarterly when remission was achieved. Among the enrolled participants, 52 (age range of 60 – 89) achieved depression remission after antidepressant treatment (MADRS< 8 for at least 2 months). We defined relapse as having a MADRS > 15 for at least two weeks. Individuals were classified as non-relapsed if they maintained a MADRS <13 after remission during the two follow-up years. Among the 52 LLD patients, 27 of them relapsed and 25 did not. We also measured trait and state anxiety and neuroticism using State-Trait Anxiety Inventory (STAI), NEO Personality Inventory (NEO PI). The neuroimaging data was collected from 3T Siemens Skyra scanner. The imaging parameters for T1-weighted images were: TR/TE=2200/2.8ms, flip angle=13^o, matrix = 256 × 256 × 169, voxelsize=1 × 1 × 1mm³. The 7-min resting-state fMRI data was collected using EPI sequence with FOV=240mm, flip angle=90^o, TR/TE=2000/31ms, matrix=64 × 64 × 34, voxelsize=3.8 × 3.8 × 3.75mm³. The resting-state fMRI data was preprocessed using the default protocol settings in Conn Toolbox version17f. The preprocessed data was entered to DPABI software for calculating the Amplitude of Low-Frequency Fluctuations (ALFF). Functional connectivity (FC) was done using Conn Toolbox version17f. Two sample t-tests were conducted to compare resting-state activity, ALFF, and FC to examine differences in ALFF and FC between relapsed and non-relapsed LLD groups (age and gender were controlled). Significance was set at p<0.05 with FDR correction for multiple comparisons. <h3>Results</h3> There were no significant differences between relapsed and non-relapsed groups in age, gender, state anxiety or neuroticism. Those who relapsed during the two-year follow-up had less severe depressive symptoms at baseline (mean (SD) MADRS=18.40(4.06) than those non-depressed group (mean (SD) MADRS=20.81(6.23) with p=0.001), whereas the relapsed group had higher STAI-trait anxiety than the non-relapsed group (p=0.04). Two-sample t-tests on FC showed that the relapsed group had a lower FC between the right parahippocampus and the superior and inferior parietal cortex. The relapsed group also had a reduced FC between the right globus pallidum and the right superior frontal gyrus. In addition, the relapsed group had significantly greater ALFF in the left amygdala and globus pallidum compared with the non-relapsed group; however, the significance was lost after controlling for depression severity and STAI-trait anxiety. <h3>Conclusions</h3> This is one of few studies that have examined clinical and neural characters in relapsed LLD. Lower FC between the right parahippocampus and the superior and inferior parietal cortex suggests a potential reduction in attention for memory. Low FC between the right globus pallidum and the right superior frontal gyrus may indicate reduced executive function in relapsed individuals. Importantly, the relapsed group also showed lower depressive symptoms yet higher trait anxiety. These results suggest that depression comorbid with anxiety might be an essential factor of relapse tendency. Future studies will examine whether the relapsed group also has lower executive function.