Amin Saberi - Institute of Medical Sciences and Technologies, Shahid Beheshti University, Tehran, Iran
Esmaeil Mohammadi - Institute of Medical Sciences and Technologies, Shahid Beheshti University, Tehran, Iran
Claudia R. Eickhoff - Institute of Neuroscience and Medicine, Brain & Behaviour (INM-۱/INM-۷), Research Centre Jülich, Jülich, Germany,Institute of Clinical Neuroscience and Medical Psychology, Heinrich Heine University, Düsseldorf, Germany
Mojtaba Zarei - Institute of Medical Sciences and Technologies, Shahid Beheshti University, Tehran, Iran
Simon B. Eickhoff - Institute of Neuroscience and Medicine, Brain & Behaviour (INM-۱/INM-۷), Research Centre Jülich, Jülich, Germany
Masoud Tahmasian - Institute of Medical Sciences and Technologies, Shahid Beheshti University, Tehran, Iran

Late life depression (LLD) is turning into a health concern with the aging of the population. Its unique presentation with somatic symptoms and cognitive dysfunction highlights the need to study LLD as a distinct disease process from depression in younger adults. Numerous functional and structural neuroimaging studies have investigated the neural correlates of LLD, but have reported inconsistent results. In this study, we aimed to identify locations in the brain that were most consistently altered across the existing studies.Method We searched PubMed, Embase, and Web of Knowledge for relevant studies published until July 2019 and retrieved 2657 potential studies. All whole-brain voxel-based morphometry (VBM), positron emission tomography (PET), and functional magnetic resonance imaging (fMRI) studies comparing patients with LLD and healthy controls (HC) were included. Two authors independently extracted the data and the peak coordinates of the significant foci. Coordinate-based meta-analysis was performed using the revised activation likelihood estimation (ALE) method according to the current best-practice guidelines.Results We identified 23 eligible studies (eight VBM, one PET, five resting-state fMRI, and ten task-based fMRI) including 516 LLD patients and 490 HCs. No significant converging structural and functional brain abnormalities between LLD and HC (27 experiments) was observed after controlling for multiple comparisons using a stringent cluster-level family-wise error correction (pcFWE = 0.680). Additionally, the meta-analysis on LLD > HC with 13 experiments (pcFWE = 0.069) and LLD < HC with 14 experiments (pcFWE = 0.357) revealed no significant convergent abnormalities.Conclusions The lack of convergence across individual studies might be related to the differences in experimental designs and statistical approaches, heterogeneity of clinical populations, and small sample sizes of the individual studies. Currently, there is not enough evidence available to pinpoint a specific underlying neural pathology associated with LLD.